Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mapping interorganisational collaboration within biomedicine : collaboration in infection and immunity… Lander, Bryn 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_fall_lander_bryn.pdf [ 3.27MB ]
Metadata
JSON: 24-1.0074093.json
JSON-LD: 24-1.0074093-ld.json
RDF/XML (Pretty): 24-1.0074093-rdf.xml
RDF/JSON: 24-1.0074093-rdf.json
Turtle: 24-1.0074093-turtle.txt
N-Triples: 24-1.0074093-rdf-ntriples.txt
Original Record: 24-1.0074093-source.json
Full Text
24-1.0074093-fulltext.txt
Citation
24-1.0074093.ris

Full Text

   MAPPING INTERORGANISATIONAL COLLABORATION WITHIN BIOMEDICINE: COLLABORATION IN INFECTION AND IMMUNITY RESEARCH  by  BRYN LANDER Bsc, The University of Toronto, Canada, 2003 Msc, The University of Sussex, England, 2005  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate and Postdoctoral Studies (Interdisciplinary Studies) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) August 2013   ?Bryn Lander, 2013   iii   Abstract  Background: Collaborations between and within sectors facilitate research and development by transferring knowledge among individuals; but it is often unclear who is involved, with whom they are collaborating, and why they are collaborating. I studied the collaborations of Vancouver-based infection and immunity researchers both with local and non-local collaborators, combining innovation systems with economic geography, neo-institutional theory, Bourdieu?s concept of fields, and a social network perspective. My objectives were to determine how different types of proximity affect collaboration, investigate what motivates collaboration, and to explore how institutions affect collaboration.  Methods: I used a mixed methods approach that drew on infection and immunity co-authorships, interviews with infection and immunity researchers, and policy documents. I quantitatively analysed co-authorship trends to explore the impact of institutional and geographic proximity on global co-authorship patterns of Vancouver-based infection and immunity researchers through sociograms, proximity variables, and a quasi-Poisson random effects regression. I investigated collaboration rates between and within sectors through relational contingency table and ANOVA network analysis. I mapped the major organisations and regulative institutions involved in Vancouver?s local infection and immunity network by combining interviews, policy documents, and co-authorship data. Based on interviews, I examined how sectoral and organisational institutions and capital influenced action.  Results: I found that Vancouver?s infection and immunity network was dominated by the non-commercial sector, particularly universities. The private sector presence was weak. Geographic and institutional proximity increased the proclivity to co-author papers. Hospitals and universities co-authored more papers together than statistically expected. Vancouver-based infection and immunity researchers collaborated to gain capital to further goals, a process shaped by institutions.  Conclusion: This study has important implications for science and innovation theory as well as science policy. For both, my primary contribution is to further the understanding that interactions between non-commercial actors play in knowledge iv  translation and innovation, a role that is often underemphasized in both theory and policy.     v   Preface This statement is to certify that the work in this thesis was conceived, conducted and written by Bryn Lander. Ethics approval for the quantitative analysis was not required because it was covered by the publicly available data clause (Item 7.10.3) of the University of British Columbia?s Policy #89: Research and other studies involving human subjects. The qualitative component of the research described in this dissertation was approved by the University of British Columbia?s Behavioural Research Board: UBC BREB Number: H10-02527. Bryn Lander is entirely responsible for the work in this dissertation. A version of Chapter 3 has been published: Lander, B. (2013). Sectoral collaboration in biomedical research and development. Scientometrics, 94, 343?357.  vi   Table of contents Abstract ............................................................................................................. iii Preface ............................................................................................................. v Table of contents ..............................................................................................vi List of tables ...................................................................................................... x List of figures ....................................................................................................xi Acknowledgements .......................................................................................... xii 1 Chapter 1: Introduction, literature review, and methodology ................... 1 1.2 Introduction ............................................................................................... 1 1.3 Related theoretical concepts .................................................................... 4 1.3.1 Systems level approaches ................................................................. 4 1.3.1.1 Innovation systems ...................................................................... 4 1.3.1.2 Organisational fields .................................................................... 5 1.3.1.3 Bourdieu?s social field .................................................................. 6 1.3.1.4 Actor network theory .................................................................... 6 1.3.1.5 Systems compared ...................................................................... 7 1.3.2 System boundaries .......................................................................... 10 1.3.3 Proximity .......................................................................................... 11 1.3.4 Institutions ........................................................................................ 14 1.3.5 Capital .............................................................................................. 16 1.3.6 Social network analysis and theory .................................................. 18 1.4 Research questions ................................................................................ 22 1.5 Study focus ............................................................................................. 22 1.5.1 Vancouver as a case study .............................................................. 24 1.5.2 Organisations and institutions .......................................................... 26 1.6 Overview of design and methodology ..................................................... 27 1.6.1 Reflexivity ......................................................................................... 31 1.7 Structure of the dissertation .................................................................... 32 2 Chapter 2: Proximity at a distance in I2 research ..................................... 33 2.1 Introduction ............................................................................................. 33 vii  2.2 Proximity ................................................................................................. 36 2.3 Data and methods .................................................................................. 40 2.3.1 Developing an I2 database ............................................................... 40 2.3.2 Variables and analytic techniques .................................................... 42 2.4 Results .................................................................................................... 48 2.4.1 The geographic and institutional structure of collaboration .............. 48 2.4.2 The role of proximity in collaboration ................................................ 52 2.5 Discussion .............................................................................................. 55 2.6 Conclusion .............................................................................................. 59 3 Chapter 3: Sectoral collaboration in biomedical research and development ..................................................................................................... 62 3.1 Introduction ............................................................................................. 62 3.2 Innovation theory, organisations, and networks ...................................... 64 3.3 Methods .................................................................................................. 67 3.4 Sectors? involvement in Vancouver?s I2 network ..................................... 69 3.5 Co-authorship within and between sectors ............................................. 73 3.6 Discussion .............................................................................................. 76 3.7 Conclusion .............................................................................................. 79 4 Chapter 4: Collaboration and institutions in Vancouver?s local I2 organisational field ........................................................................................... 81 4.1 Introduction ............................................................................................. 81 4.2 Institutions, organisational fields, and collaboration ................................ 83 4.2.1 Institutions ........................................................................................ 83 4.2.2 Organisational fields ......................................................................... 85 4.3 Methods .................................................................................................. 87 4.3.1 Quantitative approach ...................................................................... 88 4.3.2 Qualitative approach ........................................................................ 89 4.3.2.1 Interview sampling ..................................................................... 89 4.3.2.2 Interviews .................................................................................. 92 4.3.2.3 Organisational and institutional documents ............................... 94 viii  4.3.2.4 Qualitative analysis .................................................................... 95 4.4 Vancouver?s I2 organisational field .......................................................... 98 4.4.1 Public presence within the field ...................................................... 102 4.4.2 Private presence within the field ..................................................... 107 4.4.3 Boundary spanning within the field ................................................. 110 4.5 Discussion and conclusion ................................................................... 114 5 Chapter 5: The microfoundations of collaborations within Vancouver?s I2 field .................................................................................................................. 117 5.1 Introduction ........................................................................................... 117 5.2 Capital and collaborations: What I have is what you need.................... 123 5.3 Sectors and capital ............................................................................... 127 5.3.1 Healthcare ...................................................................................... 127 5.3.2 Universities ..................................................................................... 128 5.3.3 Firms .............................................................................................. 131 5.4 Collaboration and the regulative institutional pillar ................................ 133 5.5 Sectors and the normative and cultural-cognitive institutional pillars .... 136 5.6 Discussion ............................................................................................ 142 5.7 Conclusion ............................................................................................ 146 6 Chapter 6: Discussion and conclusion ................................................... 149 6.1 Introduction ........................................................................................... 149 6.2 Summary of main findings .................................................................... 150 6.3 Cross cutting themes ............................................................................ 152 6.3.1 Proximity ........................................................................................ 152 6.3.2 Scarcity and prestige ...................................................................... 154 6.3.3 Vancouver?s I2 organisational field ................................................. 156 6.4 Importance of work ............................................................................... 157 6.5 Shortcomings in work ........................................................................... 160 6.6 Areas for future research ...................................................................... 163 6.7 Conclusion ............................................................................................ 165 7 References ................................................................................................. 170 ix  8 Appendix A: Creating an I2 database ...................................................... 191 A.1 Coding and cleaning the I2 database .................................................... 194 A.2 Adding data to the database ................................................................ 196 A.3 References ........................................................................................... 197 9 Appendix B: Summary of participants interviewed ............................... 208 10 Appendix C: Sample interview guide and related sociograms ............. 212 11 Appendix D: Coding list used in analysis ............................................... 217 12 Appendix E: Acronyms............................................................................. 222      x   List of tables Table 1: Three pillars of institutions ............................................................................... 16 Table 2: Variables for quasi-Poisson regression analysis ............................................. 44 Table 3: Authors with multiple affiliations ...................................................................... 48 Table 4: Affiliations by region ........................................................................................ 51 Table 5: Affiliations of Vancouver's co-authors by region and sector ............................ 52 Table 6: Quasi-Poisson random effects incidence rate ratios ....................................... 54 Table 7: Sectoral activity ............................................................................................... 73 Table 8: Observed rates of co-authorship within and between sectors weighted by sector ............................................................................................................................ 74 Table 9: Observed over expected rates of co-authorships within and between sectors 75 Table 10: Mean tie densities of co-authorships within and between sectors ................. 75 Table 11: Sectoral affiliations of sampling dataset, ideal participants, and participants 91 Table 12: Vancouver authors and organisations by sector ............................................ 98 Table 13: Co-authorships for researchers affiliated with Vancouver organisations ..... 100 Table 14: Summary characteristics of participants ...................................................... 120 Table 15: The three pillars in I2 organisational field ..................................................... 143 Table 16: Specialist journals used in creating dataset................................................. 198 Table 17: Generalist journals used in creating dataset................................................ 202 Table 18: Keywords used in creating the database ..................................................... 206   xi   List of figures  Figure 1: Likelihood of collaboration based on geographic and institutional proximity .. 14 Figure 2: Closure and structural holes ........................................................................... 21 Figure 3: Co-authorships outside of North America ...................................................... 49 Figure 4: Co-authorships in Canada and USA .............................................................. 50 Figure 5: Vancouver's global I2 by organisation ............................................................ 71 Figure 6: The core of Vancouver's global I2 network ..................................................... 72 Figure 7: Interview sampling ......................................................................................... 90 Figure 8: Organisations in Vancouver's local I2 network.............................................. 102 Figure 9: Participants interviewed and their co-authors ............................................... 119 Figure 10: Creation of an I2 database .......................................................................... 192 Figure 11: Two-mode egonet of participants interviewed and co-authors connected by papers ......................................................................................................................... 215 Figure 12: Two-mode sociogram of organisations affiliated with participant and co-authors connected by papers ...................................................................................... 216   xii   Acknowledgements First and foremost I would like to thank Christian. I know that partners often provide the emotional and mental support necessary for dissertations to get written, but Christian went beyond that. Without him this dissertation literally would not have been possible. He gave up his previous life, moving to Vancouver so that I could pursue my PhD. He provided pro bono programming work, creating the bilibiometric database that forms the basis of this PhD. (In the process forcing me to learn MySql because Microsoft Access is inferior and changing all my variable names to American spellings.) He figured out how I could use his computer to run my regressions. I thank him for helping me achieve my goal in any way he could. I would also like to thank my supervisor Dr. Steve Morgan. Steve will be the first to admit that our academic interests do not fully align, but he remained dedicated in supporting me in my work. He provided pragmatic and intellectual advice that enabled me write?and perhaps most importantly to finish?my dissertation. Steve gave me a home at CHSPR. I?m not sure if the people at CHSPR realise how much it meant to me to have a desk in their office. Interdisciplinary studies can be a lonely business and CHSPR became my community. Without them, writing this dissertation would have been a lot harder, if not impossible. I?d like to thank all of CHSPR, particularly its ?drug team.? In addition to essential intellectual feedback and support, CHSPR gave me somewhere to work and people to talk to.  I would like to thank my thesis committee members, Dr. Janet Atkinson-Grosjean, Dr. Adam Holbrook, and Dr. David Tindall for their invaluable input and assistance. Their thoughtful guidance, support, and most of all, precious time, has meant so much to me.  I should also extend my gratitude to those who have financially supported me and this research: The Canadian Institutes of Health Research and the Western Regional Training Centre.  Last but not least, I would like to thank my family for their support. To my parents who believed I could do this and edited this entire document. To Vera Hansen Lander who showed me that there?s more to life than being a student.     1  1 Chapter 1: Introduction, literature review, and methodology 1.2 Introduction Research collaborations are growing in importance for both individuals and policy makers (Hoekman, Frenken and Tijseen, 2010; He, Geng and Campbell-Hunt, 2009; Wagner and Leydesdorff, 2005; Narin et al, 1991). Collaborations create networks of individuals that are connected together and transfer knowledge and norms (Ahuja 2000; Hansen, 1999; Uzzi, 1997; Goes and Park, 1997; Burt, 1992). Networks, in turn, are embedded within larger systems comprised of organisations and institutions. These systems create a landscape that influences network structure and, over time, network structure changes these systems, creating a feedback loop between the networks and the systems within which they operate (Owen-Smith and Powell, 2008). While policy makers often view networks of research collaborations as desirable, the academic literature is often unclear about the role of various factors in shaping these networks and related systems. I explored three gaps in the academic literature through my dissertation. First, unlike many previous analyses of research and development (R&D) systems that focused disproportionately on the private sector, I investigated the roles that different sectors play within these systems. Second, the academic literature offers no consistent method for drawing analytic boundaries around the system under study. I tested the relative importance of different boundaries. Third, most previous systems studies focused on organisation level interactions. I investigated how interactions between system structure, organisations, and individuals shape collaboration decisions. Within economics, analyses of the role of networks in R&D processes are disproportionately focused on the private sector and the firm as innovator (Edquist 2005; Nelson and Rosenberg 1993; Lundvall 1992). Within these models, universities are often portrayed as important initial producers of research that is then transferred to firms for development (Etzkowitz and Leydesdorff 2000; Etzkowitz et al 2000; Etzkowitz and Leydesdorff 1999). Non-commercial sectors, such as governments, hospitals, and non-governmental organisations (NGOs) support R&D activities of firms and universities but are not assumed to be active in R&D processes themselves. Because of the focus    2  on for-profit innovation, the implicit goal of these models is product development that is expected to increase economic prosperity. Studies of biomedical innovation that focus on the firm or interactions between the firm and universities in the development of new medical products such as pharmaceuticals and medical devices are examples of this perspective (Windrum and Garcia-Goni 2008; Hopkins 2006; Djellal and Gallouj 2005).  This firm-centric view fails to recognize the significant roles of the non-commercial sectors in R&D activities. This is particularly true for biomedicine where academic hospitals play an important role in R&D (Gelijns et al 2001). Linkages between hospitals and universities appear to be particularly crucial in non-commercial R&D, especially in developing novel practices and techniques (Lander and Atkinson-Grosjean 2011). Models which focus predominately on the roles of firms and universities may fail to properly include all sectors active within these R&D systems. Understanding the relative importance of different sectors is important because the dominant sector influences which institutions dominate the system (Owen-Smith and Powell, 2008; Scott, 2008). For example, for-profit and not-for-profit organisations have different motivations for collaborating in R&D (Gregersen, 1992). Owen-Smith and Powell (2004) found that systems anchored by not-for-profits were more open than those dominated by for-profit organisations. Knowing what the relative importance of the different sectors in a system under study thus increases understanding of the motivations and norms within the system. In Chapters 3-5 of my dissertation, I provide insight into both for-profit and not-for-profit collaboration motives by expanding network and systems analysis to include market and non-market organisations and related institutions. Drawing boundaries around R&D systems for analytic purposes is often challenging. Boundaries are generally based on knowledge, technology, geography, or some combination of these attributes (Ramlogan et al, 2007; Carlsson, 2006; Malerba, 2005; Wolfe and Gertler, 1998; Cooke, 1997). There is no agreement as to which boundaries better suit what study and why. Scholars sometimes discuss research collaboration as transcending geographic boundaries, motivated instead by a particular problem (Ramlogan et al, 2007; Carlsson, 2006; Malerba, 2005). In other cases,    3  collaboration is depicted as geographically constrained, allowing for dense, embedded relations important for fine-grained information exchange (Reagans and McEvly, 2003; Ahuja, 2000; Burt, 2000; Uzzi, 1997). Often presented as a dichotomy, these motivations to collaborate can instead be conceptualized as different dimensions of proximity between individuals (Boshcma, 2005; Ponds et al, 2007; Torre and Rallet, 2005). This multidimensional concept of proximity goes beyond mere geographic closeness to include individuals who are institutionally proximate by sharing organisational structures, regulations, or cultures. The importance of these different dimensions of proximity will be empirically tested in Chapter 2 to better understand which system boundaries best fit what study. Another identified gap that my dissertation addressed is the connection between individual perspectives and overall system structure. Many systems studies focus on organisational level interactions, ignoring the role of individuals in collaboration decisions (Powell and Colyvas, 2008; Scott, 2008; Edquist, 2005). I believe that a shortcoming of previous systems studies in a paucity of studies integrating individuals? perspectives and system structure together. This is because institutions at all levels can affect the ability of individuals to collaborate across organisations. Organisational and suborganisational institutions can be particularly important in their ability to affect how individuals act and make decisions (March and Simon, 1958). To address this gap, my dissertation drew on individual-level perspectives of system structure in Chapters 4 and 5. Bibliometric systems analyses have similarly taken organisations, rather than individuals, as the unit of analysis (see for example Hicks and Katz, 1996; Hoekman et al, 2010; Ponds et al, 2007; Sandstrom et al, 2000; Thijs and Glanzel, 2010; Wagner and Leydesdorff, 2005). Bibliometric studies based on the organisation level obscure collaborations within the same organisations and were unable to properly weight interorganisational studies by co-author. To address this gap, I conducted individual-level bibliometric analysis in Chapter 2. To address these gaps, this dissertation combined an innovation systems (IS) framework with economic geography, neo-institutional theory, Bourdieu?s concept of fields, and a social network perspective. The IS framework, organisational fields within    4  neo-institutional theory and Bourdieu?s fields all provide a systems level approach to networks and collaboration which I combined. These frameworks include organisations, their interactions and the institutions within which the organisations are embedded (Scott, 2008; Malerba, 2005; Bourdieu, 1985).  I further drew from neo-institutional literature to define institutions and apply Scott?s (2008) three pillars of institutions to my own work. From economic geography I added the concept of proximity to explore different boundaries around R&D systems.  I drew from social network analytic tools to identify actors and their relative importance within the systems. Finally I incorporated the social networks concepts of homophily, social capital, and boundary spanning.  I applied these analytic tools through mixed methods to a study of individuals involved in infection and immunity (I2) R&D within Metro Vancouver, Canada (Vancouver). In doing so, this dissertation analyses how people collaborate and how collaborations are influenced by institutions. In Section 1.2 of this introductory chapter I reviewed related literature. Section 1.3 outlined the dissertation?s research questions. Section 1.4 introduced the case study used in the dissertation. Section 1.5 provided an overview of methodology. Finally, Section 1.6 outlined the structure of the rest of the dissertation. 1.3 Related theoretical concepts 1.3.1 Systems level approaches Many different systems level approaches exist that investigate networks of individuals and organisations that are embedded within institutions. I briefly outline four such approaches below: IS, organisational fields, Bourdieu?s social field, and actor network theory.   1.3.1.1 Innovation systems The IS framework was first developed in the mid-1980s as both an academic and policy tool (Sharif, 2006). The IS insights utilized in this dissertation are that innovation relies on knowledge creation and translation between individuals that are often affiliated with different organisations and embedded in and affected by institutions. Most IS analyses include three main components: actors, their interactions, and institutions    5  related to these actors and their interactions (Edquist, 2005; Malerba, 2005). Innovation involves the co-evolution and interaction of the different components within the system (Malerba, 2005; Cooke, 1998). Varying from study to study, actors are often defined on the organisational level and include firms, universities, research labs, financial organisations, user groups, and entrepreneurs. IS analyses do not just identify key organisations and individuals but also explore the ways that actors interact to generate new innovations, are connected together, and transfer knowledge (Malerba, 2005; Lundvall, 1992). Implicit in the IS approach is the importance given to the transfer of new knowledge, especially R&D knowledge, between organisations that often have different objectives. Institutions create the environment within which actors function. The concept of institutions will be discussed in more detail in Section 1.2.4 but generally within the IS literature, institutions can take the form of written laws or informal cultural norms, be enforced through public government agencies, private professional organisations, or peers (Edquist and Johnson, 1997).  1.3.1.2 Organisational fields Neo-institutional theory broadly focuses on the impact of institutions on organisations within an organisational field. The concept of an organisational field was first proposed by DiMaggio and Powell (1983, p.148) to denote: ?key suppliers, resource and product consumers, regulatory agencies, and others that produce similar services or products.? For example, Owen-Smith and Powell (2004) outline the organisational field of Boston biotechnology to include biotechnology firms, venture capital companies, government agencies and public research organisations and explore how these organisations interact.  An organisational field includes a diversity of organisations that interact in a tight network within a particular arena or domain and connectedness between organisations within a field is a key concept (Wooten and Hoffman, 2008; DiMaggio and Powell, 1983). Institutions guide behaviour within organisational fields; these institutions are shaped by the organisations within the field in an iterative process (Wooten and Hoffman, 2008). The relative importance of different institutions, organisations and sectors within an organisational field affect the field?s overall tone and    6  governance system (Emirbayer and Johnson, 2008; Scott, 2008; Owen-Smith and Powell, 2004).  1.3.1.3 Bourdieu?s social field Bourdieu?s social field is a social topology. Each agent or group of agents is defined by their relative positions within that space. Different forms of capital1 shape the overall field structure and access to capital defines each agent?s position and its interactions with other agents within the social field (Bourdieu, 1985). The field acts as a coherent and specific playground with its own constantly evolving rules of the game. An ongoing power struggle exists between agents for domination over the social field. The core of these rules is coordinated by a field?s habitus which is the culmination of the past experiences of agents embedded in the field?s ?cultural unconscious? (Emirbayer and Johnson, 2008; Martin, 2003). A social field?s structure has been internalised by agents and is therefore largely taken for granted. Fuzziness in how different agents view the field?s structure enables agents to change their position within the field. Agents have varying abilities to rewrite field rules, based on their position within the social field?s structure and their access to different forms of capital.  1.3.1.4 Actor network theory Other systems perspectives for analysing research and development exist, most notably actor network theory (ANT). ANT involves an account of how knowledges, technologies and actors are produced through hetereogeneous actor-networks made of both human and non-human entities. No presupposition exists of what sorts of entities, structures, and forces should be considered part of a network; and no coherent framework or theories can be applied to the study of all actor-networks (Page, 2010; Law, 1992). Instead, each network is constructed based on four ?momements? of translation during which various human and non-human actors cohere together into an actor-network around a specific problem (Law, 1992; Callon, 1986). Thus the entities, structures and forces that are considered part of an actor-network depend entirely on                                             1 The concept of capital is further discussed in Section 1.2.5 below    7  the translation case under study. A key facet of ANT is that both human and non-human actors are afforded equal importance in creating an actor-network (Law, 1992). 1.3.1.5 Systems compared These three systems levels approaches are similar in many ways. One similarity is that all have a common focus on interactions between agents embedded within a system. While I was unable to find literature explicitly connecting the IS and organisational field concepts together, Emirbayer and Johnson (2008) see organisational fields as similar to industrial sectors, which are roughly congruent to the IS framework. Organisational fields have also been explicitly connected to Bourdieu?s social field (Emirbayer and Johnson, 2008; Wooten and Hoffman, 2008). While DiMaggio and Powell (1983) did not cite Bourdieu in their pivotal 1983 article, an earlier version of the paper (1982) lists Bourdieu as an inspiration for their organisational field concept. Key differences between the IS and organisational field models focus on how and why organisations within a system change. Based on evolutionary economics, organisations within the IS model respond to market pressures by innovating. This allows organisations to differentiate themselves from their competitors, creating a temporary monopoly of profits (McKelvey, 1997; Nelson and Winter, 1977). The IS focus on the market as the motivator of change may help to explain why the IS literature has traditionally been firm-centric. While multiple actors are seen as playing a role, implicit in most of these analyses is that the firm plays the central role as innovator (Edquist, 2005). Other sectors supply firms with components needed for innovation such as capital, research, knowledge and human capacity (Nelson, 2006; Nelson and Rosenberg, 1993). Sectors and specific organisations can also create a demand for innovation, feeding their preferences, insights and needs to the firm (Lundvall, 1992; Gregersen, 1992). This firm-centric focus means that the non-profit motivations of actors participating within IS analyses remains poorly explained; leaving explanatory gaps within this framework.  Neo-institutional theory has traditionally focused on how organisations, motivated by legitimacy, become increasingly similar over time through a process called    8  isomorphism (DiMaggio and Powell, 1983). Two types of isomorphism exist: competitive and institutional. Competitive isomorphism draws from an evolutionary approach (Hannan and Freeman, 1977) and, like IS, focuses on the market as the motivation for change. Institutional isomorphism has been the traditional focus of neo-institutional theory (Scott, 2008). It explores how non-market motives such as legitimacy bring about system-level change (Deephouse and Suchman, 2008). Because of its focus on the influence of legitimacy on organisational change, neo-institutional theory became a powerful tool for understanding the non-economic motivations of organisations. Non-profit organisations were perceived as particularly prone to institutional isomorphism and many of the initial analyses focused on non-profit organisations including the civil service (Tolbert and Zucker, 1983), hospitals (Fennel and Alexander, 1987), and school districts (Strang, 1987). Several factors differentiate Bourdieu?s social field from the organisational field including the concept of capital, institutions, the role of habitus in shaping the field, the role of power in the field, and the unit employed in analysis (Emirbayer and Johnson, 2008). First, capital, as a concept, is used in organisational theory but typically signifies economic capital and does not include the multiple forms of capital discussed by Bourdieu. For Bourdieu, a key component of capital is its relational nature; capital only gains value through the interactions between agents focused on gaining access to different forms of capital. This relational nature of capital has not been explicitly incorporated into the organisational field literature. Bourdieu?s social field does not explicitly include institutions. In their place, a field?s habitus shapes relations and structural power differentials.  Neo-institutional theory does not include Bourdieu?s concept of a field?s habitus, which is the culmination of the past experiences of individuals embedded in the field?s ?cultural unconscious? (Emirbayer and Johnson, 2008; Martin, 2003). Within neo-institutional theory the concept of a governance system for each organisational field plays a similar role to the field?s habitus. The role of power in shaping interactions within a field has not traditionally been a focus in organisational fields. More recently, the focus on isomorphism as a key mechanism within organisational fields has been questioned as academics explore the different ways that organisations respond to institutions, processes that can lead to increased    9  heterogeneity as opposed to homogenisation (Boxenbaun and Jonsson, 2008). As analyses increasingly focus on how organisations respond to institutional pressures, the organisational field became a source of power struggles, moving closer to Bourdieu?s concept of a field (Wooten and Hoffman, 2008). A final difference between the organisational field and Bourdieu?s social field is their unit of analysis. Organisational fields have traditionally focused on organisational level interactions (Powell and Colyvas, 2008; Scott, 2008). Conversely Bourdieu?s (1985, p.724) focus on ?agents and groups of agents? enables individual-level analysis and facilitates a more explicit integration of individual?s perspectives of field structure.  ANT views social structure as constantly changing because of power struggles; Law (1992) notes that this view is similar to Bourdieu?s concept of habitus. In many other ways, the focus of ANT diverges from the other system level approaches discussed here. ANT involves in-depth case studies of how networks cohere to explore a specific problem, called moments of translation (Murdoch, 1997). Thus moments of translation?for example the development of an anthrax vaccine (Latour, 1983) or a conservation strategy for St. Brieuc Bay scallops (Callon, 1986)?are actively chosen as the basis of analysis. The disadvantage of this approach is that, because ANT begins with a moment of translation and uses that moment to explore a network, ANT cannot be used for exploring the spectrum of translation moments within a network. Conversely, organisational fields, innovation systems and Bourdieu?s social field can begin by mapping a network and then exploring the diverse types of problematizations, and moments of translation, within this predefined network. In my own work here I combined insights from IS, organisational fields and Bourdieu?s social field, incorporating these with the concept of capital discussed in more detail in Section 1.2.5 below. Using this approach I began with a network and then explored different reasons why actors did or did not cohere within the network. I was thus able to better determine the spectrum of translation moments occurring in Vancouver and explore the extent that Vancouver?s I2 network had the connections commonly expected to exist within a biomedical organisational field.            10  1.3.2 System boundaries Defining system boundaries is challenging. Both the organisational field literature and Bourdieu?s social field are unclear on how boundaries of a field should be drawn. Bourdieu argues that field boundaries can only be investigated empirically and extend only so far as a field?s power relations (Emirbayer and Johnson, 2008). Early analyses of organisational fields focused on interorganisational, geographically constrained fields while later analyses focused on organisations, from geographically diverse regions, that played a similar role in society (Wooten and Hoffman, 2008).  Relatively clear, albeit multiple and sometimes conflicting, proposed boundaries exist within the IS literature (McKelvey, 1991). The three dominate IS boundaries?national innovation systems (NIS), regional innovation systems (RIS), and technological innovation systems (TIS)?each use different geo/political or techno/epistemic criterion to draw boundaries. NIS was the first IS framework to develop (Sharif, 2006; Freeman, 1988; Lundvall, 1988). It defines its boundaries nationally and is based on the premise that factors such as public policy, institutions, and R&D are often national and highly relevant for explaining the differences in national economic competitiveness (Mowery and Sampat, 2005; OECD, 1999; OECD, 1997; Nelson, 1993; Lundvall, 1992; McKelvey, 1991).  Defining its boundaries regionally, RIS arose to explain observed IS within a geographic cluster or relatively autonomous non-national region. The RIS framework builds off of the NIS framework (Braczyk, Cooke and Heidenreich, 1998; Cooke, 1998; Cooke, Uranga, and Etxebarria, 1997). It applies the model regionally based on the observation that collaboration often appears to be geographically constrained, allowing for dense, embedded relations important for fine grained information exchange and tacit knowledge sharing (Reagans and McEvly, 2003; Ahuja, 2000; Burt, 2000, Uzzi, 1997). While RIS studies are generally sub-national, such as Canadian provinces (Latouche, 1998), regions can transcend national boundaries as illustrated in the Great Lakes automotive cluster in North America (Wolfe and Gertler, 1998) or certain culturally autonomous regions such as the Basque region, which spans Spain and France (Cooke, Uranga and Etxebarria, 1997).     11  TIS boundaries are defined by epistemological and technology domains, based on the recognition that each industrial sector has their own techno-epistemic regime with different R&D intensities, industrial strategies, market structures, relevant institutions, and collaborators (Carlsson, 2006; Malerba, 2004; Breschi and Malerba, 1997).  Within each techno-epistemic regime specific ways of learning, knowledge bases, and technologies exist (Breschi and Malerba, 1997). Scientific collaboration often transcends geographic boundaries, focusing instead on a particular problem or technology (Ramlogan et al, 2007; Carlsson, 2006; Malerba, 2005). By focusing on a technology, TIS introduces knowledge and artifacts as additional analytic components within the IS framework. Proponents argue that, through globalisation, technological regimes are becoming increasingly important while geography?s importance is decreasing (Carlsson, 2006). Innovation analyses often combine regional and technological boundaries in order to include local institutional structures in the analysis. Examples of this include analyses of the software engineering cluster within Silicone Valley (Saxenian,1996) or the biotechnology cluster in Boston (Owen-Smith and Powell, 2004). 1.3.3 Proximity While the IS framework presents clear guides for drawing systems boundaries, it is often unclear which boundary to choose (Sharif, 2006; Edquist, 2005). More knowledge intensive sectors, such as biotechnology and finance, are often seen as clustering more while more traditional, less knowledge intensive sectors cluster less (Asheim and Gertler, 2005; McKelvey et al, 2004; Breschi and Malerba, 1997). The idea of proximity can be used to help understand the relation between different boundaries employed in IS analyses. Technological and geographic boundaries explore different motivations of people to collaborate and can be conceptualised as different dimensions of proximity between individuals. In essence, individuals will collaborate if they are proximate along a dimension.  Beginning in the 1990s, the French School of Proximity Dynamics have argued that the concept of proximity covers a number of dimensions (Boshma and Frenken, 2010; Frenken et al, 2009; Ponds et al, 2007; Boschma, 2005; Torre and Rallet, 2005).    12  Boschma (2005) introduced five different forms of proximity: cognitive, organisational, social, institutional, and geographical.  1. Cognitive proximity explores the extent that two agents share the same knowledge base (Frenken et al, 2009). It addresses the tacit, localised and cumulative nature of most knowledge. Knowledge creation involves bringing new knowledge together but, to be effective, agents need to have the absorptive capacity to receive this new knowledge. To effectively absorb new knowledge, an agent needs priori related knowledge (Lane and Lubatkin, 1998; Szulanski, 1996; Cohen and Levinthal,1990). This related knowledge puts them in cognitive proximity with the agent that they are receiving knowledge from. 2. Organisational proximity can be defined as the degree that two agents are affiliated with the same organisation (Frenken et al, 2009). Organisations will bring together people who might not otherwise associate (Feld, 1981). Individuals within the same organisation often share routines, norms, or knowledge (Frenken et al, 2009; Boshma, 2005; Torre and Rallet, 2005).   3. Social proximity draws on the concept of embeddedness in social network theory (Uzzi, 1997; Granovetter, 1985). In embedded relations, ties between agents often exist along multiple dimensions and include social and economic relations. Because embedded ties include multiple dimensions, trust in the tie increases and embedded ties are more likely to persist (Dahlander and McFarland, 2013).  4. Boschma (2005) called his fourth dimension of proximity institutional and defined it as including formal laws as well as cultural norms and values. To both differentiate this dimension from a broader conception of institutions (see my discussion below) and concretely operationalize this dimension, I have relabelled institutional proximity as sectoral and national proximity. Sectoral proximity can be defined as the extent to which agents operate under similar organisation level incentive structures (Frenken et al, 2009). Agents affiliated with the same nation will share formal laws, cultural norms and values putting them in national proximity to others affiliated with the same country. 5. Geographic proximity builds off of the concept of geographic concentration and here represents perceived geographic distance between two individuals.    13  Boschma?s (2005) five different dimensions of proximity in many ways overlap. All five can be categorised into two broader dimensions?geographic and institutional?in which cognitive, organisational, social, sectoral, and national proximity are all subsumed within a broader institutional category. This broader definition of institutions is in keeping with the description of institutions in Section 1.2.4 below. Boschma (2005) recognised this overlap and argued that what differs between his institutional dimensions are their level: social and cognitive proximity are shared at the individual level, organisational proximity at the organisation level, institutional proximity?here sectoral and national proximity?at the macro level. Thus I explore cognitive, organisational, social, sectoral and national proximity as dimensions of a broader institutional proximity category. Individuals can be proximate along multiple?and sometimes overlapping?institutional dimensions; for example if they belong to the same sector, nation, or profession. Individuals can be proximate to each other by working in the same sector, organisation or country (institutional) or within the same city (geographic).  Torre and Rallet (2005) argue that individuals are more likely to collaborate if they are in either geographical or institutional proximity. Boschma (2005) expands on this by arguing that geographical proximity can be complementary in building and strengthening the different dimensions of institutional proximity and the reverse: that institutional proximity can compensate for a lack of geographical proximity. As shown in Figure 1, geographic and institutional proximity can be theoretically viewed within an analytic grid that predicts the relative likelihood that two individuals will collaborate: individuals that are institutionally and geographically proximate have a high likelihood of collaborating. Individuals with close geographic proximity that are not institutionally proximate are less likely to collaborate as are individuals that are institutionally proximate but distant in geographic proximity; individuals that are neither institutionally nor geographically proximate will be the least likely to collaborate.    14  Figure 1: Likelihood of collaboration based on geographic and institutional proximity   The different IS boundaries map onto Figure 1. Both national and regional IS analyse different degrees of high geographic proximity irrespective of institutional proximity and are shown in the left column in Figure 1. TIS analyses institutional proximity irrespective of geographic proximity and are shown in the top row of Figure 1. Analysis of local industrial clusters, a combination of RIS and TIS would be focused on the top left cell of Figure 1. Testing the likelihoods of collaboration based on different dimensions of proximity may enable a better determination of which IS boundary best captures observed collaboration patterns. For example if, as predicted in Figure 1, close geographic and institutional proximity act as the best predictors for collaboration, then local industrial clusters may be the most applicable system boundary. If institutional proximity, irrespective of geographic proximity, acts as the best predictor for collaboration, then a TIS boundary may be the most appropriate. Conversely, if high geographic proximity irrespective of institutional proximity strongly predicts collaboration, then a RIS or NIS boundary could be chosen. 1.3.4 Institutions  Institutions feature dominantly in neo-institutional, IS, and proximity literature. Neither the IS nor the proximity literature explains their conception of institutions in very concrete terms. Neo-institutional theory, conversely, is focused on institutions and can be used to better understand the concept. Meyer and Rowan (1977) were the first to    15  discuss institutions within neo-institutional literature. They describe institutionalized rules as a type of myth. These myths are rationalized and impersonal prescriptions that describe behavior in a rule-like way while being so entrenched in society that they become taken-for-granted as legitimate. Institutions are thought of as implicit ?rules of the game? that prescribe behavior in a taken for granted and legitimate way. Institutions can both support and inhibit action and are produced and reproduced through specific activities and social interactions. Within a specific situation, institutions inhibit action by setting bounds on rationality and restricting perceived opportunities and alternatives, thereby increasing the probability of certain behaviour (Deephouse and Suchman, 2008; Greenwood et al, 2008; Scott, 2008; Wooten and Hoffman, 2008; Barley and Tolbert, 1997; Jepperson, 1991; DiMaggio and Powell, 1983; Giddens, 1979; Meyer and Rowan, 1977).  Scott (2008) draws on political science, sociology and economic approaches to the study of institutions and proposes that three pillars provide institutions with legitimacy: regulative, normative, and cultural-cognitive. Scott argues that moving from the regulative to the cultural-cognitive pillar means moving from conscious to unconscious and from legally enforced to more implicit taken-for-granted institutions. Table 1 outlines Scott?s three pillars. Individual institutions can draw from one or more of these pillars; drawing from more than one pillar increases an institution?s legitimacy. Conversely, different pillars can support conflicting institutions, leading to divergent conceptions of what is legitimate behavior. Often decoupling can occur within an organisation whereby certain components, i.e. departments or individuals, become more influenced by some institutions while other components will become more influenced by others. In resolving the tension between conflicting institutions, decoupling can cause divergent behaviour within a single organisation (Meyer and Rowan, 1977).    16  Table 1: Three pillars of institutions   Regulative Normative Cultural-Cognitive Basis of order Established rules Social obligations and expectations Constitutive schema Indicators Rules, laws, governance systems Values, goals, roles, norms Beliefs, identities, 'scripts' of action Adapted from Scott (2008) p.51 and p.79  Scott?s regulatory pillar is focused on explicit regulatory processes based on established rules. An authority exists to assess conformity to these institutions and implement sanctions if institutions are not followed. Because this pillar is explicit and based on authority, analyses often focus attention to the role of the state as a rule maker and enforcer (Equist and Johnson, 1997; Hodgson, 1994).  Scott?s normative pillar is based on the values, defined as conceptions of what is preferred, and norms, defined as ideas of how things should be done, within a particular social group. Normative systems define goals and objectives as well as the appropriate ways of achieving them, and, in doing so, can give rise to specific social roles. Conformity to normative institutions is largely seen as the result of self-evaluation; individuals feel good when they comply with normative institutions and bad when they do not. Scott?s cultural-cognitive pillar is based on how culture is cognitively internalized into shared concepts of social reality and meaning. Focus here is given to meanings attached to symbols such as words, signs, and gestures; and how these meanings can become perceived as objective and external to a particular social group, creating a unique epistemology. Scott believes that Meyer and Rowan?s (1977) institutionalized rules and much of neo-institutional theory fall within this category. Institutions supported by the cultural cognitive pillar are ultimately taken-for-granted as ?the way we do things.? 1.3.5 Capital A shortcoming of many systems theories is that organisations are often seen as the actors in analysis (Powell and Colyvas, 2008; Scott, 2008; Edquist, 2005). Less    17  work has gone into understanding the microfoundations of these systems analyses (Powell and DiMaggio, 1991). Organisations are instead treated as unitary ?black boxes? with limited analysis of the role of individuals within them. However, it is individuals who are influenced by institutions and their agency that affects ?organisational? behaviour (Berends et al., 2003). It is ultimately the individual who creates connections and translates knowledge between organisations. Institutions are often viewed as enacted through the interactions between individuals because the institutions that an individual has internalised only become apparent through their actions (Scott, 2008; Barley and Tolbert, 1997; Bourdieu, 1985; Giddens, 1984). Collaboration between individuals is a prime exemplar for understanding how institutions affect action, particularly if the collaboration involves individuals affiliated with different organisations or sectors that are influenced by divergent institutions. Because institutions enable and constrain individual?s action, they enable and constrain collaboration between individuals.  Both Giddens (1984) and Bourdieu (1986) argue that collaborations involve accessing the capital of another individual within existing institutional structures.2 Giddens (1984) argues that in addition to institutions, social structures such as organisational fields are made up of both human and nonhuman resources that can be used to enhance and maintain power. Giddens does little to explain what human and nonhuman resources are (Sewell, 1992). Bourdieu more fully develops a theory of resources through his concept of capital.  Bourdieu argues that capital underlies social structures in society and can present itself in multiple forms including economic, human, cultural, and social capital.  Economic capital represents money or financial resources. Human capital includes the stock of knowledge and competencies possessed by a person. Cultural capital can be embodied as symbolic capital of an individual, objectified into cultural objects, or institutionalised such as through an educational qualification. Social capital is made up of social obligations created and enforced group membership. Capital is important insofar as it is given value within social relations (Emirbayer and Johnson, 2008; Bourdieu, 1986).                                             2 This in some ways echoes the importance accorded to non-human actants in ANT. However, in ANT, human and non-human actors can play equally important roles in a network. Instead here capital is perceived as motivating individuals to collaborate within fields.    18  In scientific research, various types of capital shape interactions within the structure of the scientific field (Bourdieu, 1991). Combining questionnaires and interviews focused on exploring the motivations behind scientific collaborations, Melin (2000) found that scientists often collaborated to gain access to methods, equipment, or special competences. These can be viewed as falling within what Hackett (2005) calls a unique ?ensemble of research technologies? comprised of materials, techniques, instruments, ideas, and theories that define scientific research. Collaborations are often motivated by the desire to gain access to all or part of this ensemble of research technologies (Lander, 2011; Shrum et al 2007; Katz and Martin, 1997). Knowledge and techniques can be conceived as part of human capital for they represent two different ways of understanding within scientific research (Lander, 2011; Fujimura, 1996; Barley and Bechky, 1994; Etzkowitz, 1992). Knowledge as a resource focuses on possessing information related to understanding nature through theory development (Shrum et al., 2007; Barley and Bechky, 1994; Price, 1984). Techniques focus on understanding through hands-on experience based on the technical aspects of a project (Hackett 2005; Barley and Bechky 1994; Price, 1984).   In their study of 53 scientific collaborations in physical sciences including high energy physics, geophysics, space science, oceanography, materials research, and medical physics, Shrum et al. (2007) found that the most common reason for collaboration was to access physical resources, a motive echoed by Katz and Martin (1997). Physical resources can be seen as a facet of economic capital and can include specialised and often capital intensive equipment as well as access to data and different types of samples. Falling within Bourdieu?s concept of cultural capital, collaboration to increase popularity, visibility and recognition had also been identified as a motivating factor behind scientific collaboration (Dahlander and McFarland, 2013; Katz and Martin, 1997).  1.3.6 Social network analysis and theory The structure of a system of individuals or organisations can be perceived as a social network. Several academics propose that social network analysis is a useful methodological approach to mapping relations within organisational fields, Bourdieu?s    19  field, and IS (Emirbayer and Johnson, 2008; Owen-Smith and Powell, 2008; OECD, 1999). Much of the research that draws from a social network perspective explores how the structure of social networks influences the way that knowledge and norms are translated (see for example Owen-Smith and Powell, 2004; Ahuja, 2000; Hansen, 1999; Uzzi, 1997).  Social network analysis provides methodological tools for analysing interactions within a system. Interactions and structure, as opposed to individuals and their attributes, are the main units in social network analysis. Social network analysis uses a variety of methods and models to explore social network data. Several key terms are used within social network analysis. The social entities under study within a system, such as individuals, organisations, or countries, are called actors. The links between them are called relational ties. The system that these relational ties form is called the social network. A number of methods are commonly used within social network analysis. Specific methods describe network structure graphically through sociograms; visually through matrices; and numerically through measures such as centrality, density, betweeness and degree. These numeric measures can often be applied to the whole network or its actors. These descriptive tools help to create a map of a network under study that includes specific actors and their relative importance. Network models can also be used to test theories related to network structure and relations (Knoke and Yang, 2008; Scott, 2000; Wasserman and Faust, 1994). In my own dissertation I draw on several social network analytic tools to describe the I2 organisational field in Chapters 2, 3 and 4. Chapter 3 further explores collaboration within the I2 organisational field using social network analytic techniques. As discussed in more detail in Section 6.5, other social network metrics?such as betweenness and centrality?were not extensively used largely because of how my bibliometric data were sampled.   Social network methods and theory are closely tied and several social network theories are relevant to this study. Proximity is conceptually similar to homophily. Homophily is used in social network theory to explore preferential attachment between individuals who share similar traits. Like proximity, homophily can exist along multiple dimensions including race, gender, age, geographic propinquity, organisational foci,    20  values, or roles (Powell et al, 2005; McPherson et al, 2001). In their review of different dimensions of homophily, McPherson et al (2001) argue that organisational foci, isomorphic occupational roles, and geographic propinquity are dominant sources of homophily. Organisational foci are conceptually similar to organisational proximity. In organisational foci, individuals affiliated with the same organisation are expected to have more ties. This is because the focused activity that occurs through shared organisational affiliation puts people into contact, fostering the formation of personal relationships within organisations (Feld, 1981). Other forms of institutional proximity, such as shared routines, norms, or knowledge; are similar to isomorphic occupational roles, another dimension of homophily. Individuals in isomorphic occupational roles have equivalent job descriptions and the same role relations (McPherson et al, 2001). Burt (1987) argued that people in isomorphic occupational roles strongly influence each other while Lazaga and Van Duijin (1997) found that an individual?s structural role influences tie creation. Geographic propinquity, another dimension of homophily, is roughly equivalent to geographic proximity and closely related to small world theory, another social network theory. In small world theory individuals within networks cluster geographically with a relatively small number of remote links (Powell et al, 2005). In geographic propinquity, individuals that are geographically close are observed as having more ties (McPherson et al, 2001). Although geography is often viewed as decreasingly important with the advent of new technologies, Wellman (1996) found that ties between individuals are still often initially made through face-to-face encounters and even high technology communication shows geographic patterning. Because of the close conceptual similarities between homophily and proximity, I explore both under the rubric of proximity in Chapter 2.  Social network analyses have also explored the relational nature of capital, particularly focusing on developing elaborations of Bourdieu?s social capital that focus on network structure (Emirbayer and Johnson, 2008). Burt (2000) reviewed the connections between social network analysis and social capital and outlined two major network structural network models for social capital: closure and brokerage.    21  Coleman (1990, 1988) argued that a dense network where all actors are connected, something Burt (2000) called network closure, is a source of social capital. For example, in Figure 2 Network A is less dense than Network B because not all actors are connected in Network A; this means that Network B has more network closure and social capital.  Coleman believed that network closure increased social capital for two reasons. First, because information quality deteriorates as it moves through intermediaries, information quality is greater in dense networks. Second, in networks with dense ties, the repercussions of inappropriate behaviour involve a greater number of actors. This decreases inappropriate behaviour, increases trust between actors, and, through increased trust, the social capital of the whole network.  Figure 2: Closure and structural holes  Brokerage draws on Burt?s (1992) previous work where he argued that individuals who bridge network structural holes have the largest amount of social capital. Structural holes involve nonredundant ties between two networks. In Figure 2, a structural hole exists between Network A and B because they are not connected. Between Networks C and D, TG bridges this structural hole, providing nonredundant ties between the two networks. By bringing the two networks together, TG supplies each network with access to the other, and benefits from this bridging role. Drawing from Simmel (1922), Burt called TG tertius gaudens, ?the third who benefits.? Individuals in TG?s structural role have the largest amount of social capital within a network.     22  Conversely Lin (1999) argues that social capital can either be explored based on the network location of an actor?the approach taken by Burt (1992) and Coleman (1990)?or as resources. He believes that it is better to take a resource based approach to social capital because an actor?s network location does not guarantee that they will have access to the resources they need. Lin and Erickson (2008) also take a resource based approach so social capital, defining it as resources embedded in social relations and social networks.  This resource based conception of social capital resembles Bourdieu?s (1985, 1986) broad definition of capital. This moves beyond analysis of how the structure of an actor?s network ties can embody capital towards an exploration of the role of different types of capital within a field or network.  In the process Lin and Erickson (2008) and Lin (1999) begin to operationalize Bourdieu?s broader theory of capital while discounting the structural forms of social capital explore by Burt (1992) and Coleman (1990). In Chapter 5 of my dissertation, I explore how resources affect social relations, building off of the work of Boudrieu, Lin and Erickson. In Chapter 2 I draw from Burt?s (19992) structural view of social capital as I analyse whether a specific type of bridging?affiliation with multiple organisations?confers any sort of collaboration advantage to an actor.     1.4 Research questions Drawing from the literatures outlined above, my dissertation research focused on the following set of research questions applied to Vancouver?s global I2 R&D organisational field: 1. How do different types of proximity affect collaboration? 2. What reasons do people give for collaborating? 3. How do institutions such as policies, norms, and organisational culture affect collaboration? 1.5 Study focus In exploring these research questions, my dissertation analysed I2 research, development and collaboration in Vancouver. I took a systems approach that combined insights from IS, organisational fields, and Bourdieu?s social field. I brought efficiency    23  and legitimacy explanations for systematic change together and incorporated concepts of capital and individual agency. I called the system under study an organisational field. In instances where I focused solely on system structure, I called the system a network.  I2 is broadly the study of the body?s defence against infection (Murphy, Travers, and Walport 2008). As a techno-epistemic area, I2 incorporates research, clinical care, and industry, making it a good choice for mapping the role of different organisational types within an organisational field. Research in this area includes the study between pathogens and their hosts, innate and adaptive immune responses, auto-immune disorders, allergies, inflammation, and host resistance or susceptibility. Key commercial applications within the area of infection and immunity are vaccination, the development of antibacterial and antriretroviral compounds, and diagnostics. Clinical care centres on preventing, identifying and treating infection and/or immune responses. Governments have traditionally taken an active role in infection and immunity trying to control infectious disease through outbreak monitoring, improvements in sanitation, and vaccination programs.  I used IS to select an initial boundary for my study; by choosing I2, I selected TIS as an initial boundary and explored the relative importance of other types of institutional proximity, as well as geographic proximity, within this boundary. I selected I2 as the techno-epistemic boundary for my dissertation based on the belief that, as is argued in the TIS literature, specific trends and patterns exist within different technological regimes that are glossed over in national and regional innovation systems analyses. What was and was not included within the I2 organisational field under study was largely determined through the creation of an infection and immunity bibliometric database, a procedure described in greater detail in Appendix A. Within this field, the database was used to identify Vancouver-based organisations, their global collaborators, and Vancouver-based researchers. By taking this approach I avoided applying my own preconceptions of which actors would be actively involved in this organisational field, instead identifying organisations and individuals based on what emerged from the published interests of actors within my database.      24  Bounding my case techno-epistemically is, to a certain extent, investigating individuals who share certain cultural-cognitive beliefs related to this techno-epistemology. These institutions may lead to different actions, structures and norms, subtleties that would be obscured in a study that focused on all science or all biomedicine. The collaborations, experiences, and activities of Vancouver-based researchers form a second geographic boundary to the case study.  The importance of geography as a boundary in selecting collaborators is further explored in my dissertation. Focusing on Vancouver is in part practically motivated; in order to study the effect of regulative institutions within an organisational field, I needed a geographically bounded sample where institutions were shared among participants. Using geography as a boundary in my study, a boundary whose relevance is tested by exploring collaborations of these individuals both within and outside of Vancouver, I was able to critically assess insights in social network, neo-institutional, and IS theory that geography does matter in collaboration (see for example Powell et al, 2005; Bathelt et al, 2004; DiMaggio and Powell, 1983). By bounding my study both techno-epistemically and geographically, I have focused on the local and global collaborations of what could be considered to be a regional organisational field (Scott, 2008). This creates a relatively small community of organisations and individuals within Vancouver, enabling me to study key players and the connections between them, in more depth. Within Vancouver?s I2 organistional field, I focused on a subset of actors: individuals, and their affiliated organisations, conducting R&D. Other organisations and individuals, such as government policy makers, financial organisations, and markets, were considered only insofar as they created institutions that were identified by individuals as affecting individuals? ability to conduct R&D and collaborate.  1.5.1 Vancouver as a case study Many past similar systems analyses focus on cities and regions that are world-renowned in the field studied. While most cities and regions are not global leaders, science policy is often aimed at growing local R&D capacity. Analyses of IS in regions that are not global leaders help theory and policy expand to include these networks (Gilding, 2008; Rees, 2005; McKelvey et al., 2003). Vancouver is not a global leader but    25  shows promise in I2. Vancouver has historically relied on natural resource exploitation but more recently is striving to become a leader in the knowledge-based economy (Rees, 2005). Vancouver has a relatively strong presence of university R&D, clinical expertise, and biotechnology. Vancouver ranked third in Canada?s public R&D expenditures by city in 2006 (Holbrook and Clayman, 2006). Vancouver has large and relatively specialised clinical care and training, housing British Columbia?s medical fellowship program, BC Cancer Agency, and a majority of British Columbia?s hospital specialist programs. While there is no strong pharmaceutical sector, the city has a relatively well-established biotechnology sector, ranking 7th in North America by some measures (Salazar et al, 2008). The greater Vancouver area has extensive I2 research capacity that is acknowledged internationally and hosts world famous infectious disease researchers including two winners of the world?s premier award in antimicrobial research and two Howard Hughes International Scholars (BC Ministry of Health, 2003).  In addition, Vancouver is an ideal city to explore the different effects of geographic and institutional proximity. Vancouver suffers from what might be called a national ?tyranny of distance? (Gilding, 2008). Geographically close to American cities within ?Cascadia,? such as Seattle and Portland, Vancouver is geographically separated from other major Canadian cities by the Rocky Mountains and sheer distance. Calgary, the closet large Canadian city, is over a 12-hour drive away along treacherous winter roads. Perhaps because of this geographic distance, Rees (2005) found that within Vancouver?s biotechnology sector almost half (43%) of all collaboration occurred with the USA. Only 23% occurred within Greater Vancouver and 9% with the rest of Canada. In part to compensate for Canada?s geographical vastness, strong institutional structures have been put in place to encourage national R&D collaborations through programs such as its Networks of Centres of Excellence (Atkinson-Grosjean, 2006). Because geographic and national institutional proximity are not completely correlated for Vancouver, Vancouver provides a strong case study for understanding the different influences that different dimensions of proximity play.     26  1.5.2 Organisations and institutions While organisations are often called institutions in general parlance, in this dissertation institutions are conceptually different from organisations. While institutions can be seen as the implicit and explicit rules that influence action, organisations are a type of social collective. Pfeffer (1997) argues that organisations differ from other types of social collectives in several ways. First, an organisation is formed around a collective goal; the most basic of these goals is the organisation?s own survival. Second, an organisation can be distinguished from other types of social collectives by the nature of its boundaries. Organisations have clear boundaries. Inclusion within these boundaries and the degree to which these boundaries are permeable are both largely controlled by the organisation. This differs from other social collectives. Families, for example, are a social collective but not an organisation because a family does not have complete control over its membership. An individual can be born into a family and their very birth (generally) gives them family membership. Finally, Pfeffer (1997) argues that organisations are generally recognized by a government entity, which confers legitimacy to the organisation. This distinguishes organisations from other, less formal, social collectives that often are not granted the same status by the state.     Organisations can have their own institutions that are shared by individuals affiliated with the organisation and influence these individuals? action (Berends et al., 2003; Barley and Tolbert, 1997; Williamson, 1994). Other institutions are specific to an organisational department, a sector, profession, social group, or society (Scott, 2008).  For example, specific hospitals, universities, and firms are considered to be organisations, not institutions, in this dissertation. However, there may be institutions that are shared by hospital, university, or firm sectors, with each sector including many organisations. Organisational institutions are important when individuals affiliated with different organisations collaborate. Like all institutions, organisational institutions affect how individuals make decisions, act, and limit their rationality (March and Simon, 1958). Based on organisation specific institutions, individuals belonging to different organisations may experience different concepts of what constitutes rational decision-   27  making and appropriate action. This divergence may influence the ability of individuals to collaborate across organisations. 1.6 Overview of design and methodology My research approach draws from case study mixed methods design. Case studies are a common approach for innovation systems analyses, which take the ?system? as the case study under analysis (Edquist, 2005). Creswell (2007) defines a case study approach as a methodology ?in which the investigator explores a bounded system (a case) or multiple bounded systems (cases) over time, through detailed, in-depth data collection involving multiple sources of information (e.g. observations, interviews, audiovisual material, and documents and reports) and reports a case description and case-based themes? (p.73). Sampling within case studies involves both choosing the case study(s) to be analysed and sampling evidence within the case(s). Stake (2005) argues that there are three general types of case studies: intrinsic case studies that are done out of interest for the particular case (evaluation case studies would fit in here), instrumental case studies that are done to provide larger insights or generalizations, and multiple case studies where more than one case is chosen and then compared. My dissertation takes an instrumental approach; in studying collaboration in Vancouver?s I2 network, I gained greater empirical insight into how research and development collaborations occur in regions that are not global leaders. All case studies involve a clearly bounded, complex, and holistic system that is the case under study (Stake, 2005; Creswell, 2007; Yin, 2003). Clearly bounding and justifying the boundaries of a case is a key challenge in case study research (Yin, 2003). As discussed in Section 1.4, my case study is bounded both geographically and techno-epistemically to include I2 R&D by Vancouver-based researchers.  Data used in case study research is highly varied and changes depending on the specific case (Creswell, 2007). Yin (2003) argues that data in case studies are predominately drawn from six sources: documents, archival records, interviews, direct observation, participant observation, and physical artefacts. These multiple data sources enable triangulation to be an essential component of case study research (Stake, 2005). Because case study research draws from so many data sources, many    28  case studies draw from mixed methods. Yin (2003) argues that mixed methods can effectively be used in case studies to collect a richer and stronger array of evidence. I used mixed methods in my dissertation. Mixed qualitative and quantitative methods can provide stronger inferences as one method offsets disadvantages of another and provides more diversity during analysis (Teddlie and Tashakkori, 2003). Because this study analyses a system of individuals working in different organisations, direct and participant observation was not a major data source. Instead, data for this research mainly drew from co-published articles, interviews, and documents. Greene, Carcelli, and Graham (1989) identify five purposes for mixed methods research: triangulation, complementary, development, initiation, and expansion. Triangulation is also an identified strength of case study research (Creswell, 2007; Yin, 2003). This case study uses a mixed methods approach for development, triangulation, and complementary purposes.  In developmental mixed methods, one method helps develop the second. As a developmental approach, I began with my quantitative investigation, using it as a basis for my qualitative work. Both analyses came together during the writing stage. I began by quantitatively mapping Vancouver?s I2 network drawing from a social network perspective and more traditional statistical methods. I used paper co-authorships as a proxy for collaboration based on the belief that ?texts are?crystallized social relations? (Campbell and Gregor, 2002, p.79). Through this approach, I created a relatively impartial map of the individuals and organisations, and their interactions, involved in Vancouver?s I2 network using UCINET 6.32 (Borgatti et al 2002) and Gephi 0.8.1 (Bastian et al, 2009). I explored geographic and institutional proximity through an individual-level analysis of I2 co-authorship rates based on a quasi-Poisson random effects regression (Hilbe, 2007). I further analysed sectoral co-authorship patternsthrough an organisation-level analysis of whether organisations affiliated with different sectors co-author more or less than expected based on a relational contingency table and ANOVA network analysis (Hanneman, 2005; Borgatti et al, 2002). These results are described in Chapters 2, 3, and 4.    29  My qualitative study built off the map of Vancouver?s I2 network created in my quantitative analyses. I used this map as a sampling framework to identify participants to interview, employing a stratified purposeful sample (Creswell, 2007) that preserved the same proportions of organisational affiliations observed in my quantitative data. This sampling technique was later supplemented with snowball sampling (Creswell, 2007). Methodologists do not generally recommend an ideal number of interviews for a case study (Crewell, 2007; Stake, 2005; Yin, 2003). I used sufficiency and saturation criterion to decide whether I had interviewed enough people (Seidman, 1998). Employing sufficiency criteria, I interviewed participants and sites until I believed that the experiences of the participants would resonate with others who were part of the I2 network but not sampled to be interviewed. Based on the saturation criteria, I continued to interview until I began to hear the same information repeated by participants. Yin (2003) recommends that interview questions in case study research are draw from related theoretical themes. My interview questions aimed to develop themes related to innovation systems and neo-institutional theory by exploring participants? work and roles; their reasons for collaborating, particularly between sectors; and the impact of institutions on collaboration. Drawing from institutional ethnography, I paid particular attention to encouraging participants to describe in detail ?the work? that is part of these collaborations in order to understand related institutions (Campbell and Gregor, 2002). Interviews were supplemented by analysis of related documents. Text is often seen as a fundamental representation of institutions because it is generalizable and translocally produced (Luken and Vaugh, 2006; Campbell and Gregor, 2002). The majority of these documents were organisational strategic plans, reports, and websites. These were used to help understand and describe the organisations and institutions that formed the organisational field under study and to create organisational profiles. These results are described in Chapters 4 and 5.  Complementary methods are when both methods are used to measure overlapping but different facets of the same phenomenon. My mixed methods approach is complementary because both qualitative and quantitative methods measure overlapping and different aspects of the same phenomena. Through quantitative means I mapped the global collaboration network of Vancouver?s I2 R&D collaborations and    30  analysed statistical trends within the data related to geographic and institutional proximity. Qualitatively, I looked at a smaller number of Vancouver I2 researchers in more detail, trying to understand the statistical trends seen within the larger system through an in-depth study of individuals? work and perceptions on collaboration patterns. Chapter 4 combines my quantitative and qualitative approaches through a mixed methods investigation of Vancouver?s I2 organisational field. My dissertation also employed mixed methods to assess the same phenomena in different ways, counteracting the biases of each method. By mapping collaborations both qualitatively and quantitatively, I tested how well bibliometic sociograms represent an individual?s collaboration patterns and organisational affiliations. Triangulation also occurred as I compared how institutions were discussed in different interviews and within institutional documents. My qualitative and quantitative results are compared in Section 6.3 of Chapter 6. Yin (2003) argues that all stages in case study research should be explicitly connected to theory. I drew on the organisational field framework to define my case and sample documents and individuals to interview. As per this framework, data generation particularly focused on learning about organisations, their interactions and how these interactions are affected by institutional structures. Case studies need to provide ?thick descriptions? of the case under study (Creswell, 2007). Part of this ?thick description? is putting the case within its relevant context (Yin, 2003). In structuring my dissertation, I included all aspects that Creswell (2007) argues makes for a good case study including a clear identification and description of the case, an illustration that the case is used to understand a research issue, and themes and assertions made from the case. No formal process exists for analysing case studies, although a final report generally includes the case description, context, themes, and lessons learned (Creswell, 2007). Combining these components with my theoretical framework formed the basis of my analytic strategy. Stake (2005) suggests that case study analysis should be interactive and iterative, beginning simultaneously with the data generation process. Following Creswell (2007), I began designing my final report around a description of my case and themes.     31  This dissertation brings together both the quantitative and qualitative components of my study into an analysis that moves from the global systems level, which maps and analyses I2 collaboration trends, to an analysis of a subset of the individuals who are part of that system, which explores collaboration in more detail and describes their perceptions on how sectoral and organisational institutions and capital influenced their collaboration decisions. My quantitative analysis was used to map Vancouver?s global I2 R&D collaborations, predominately focusing on my first research question: How do different types of proximity affect collaboration? My qualitative analysis studied Vancouver?s local I2 collaborations, primarily focusing on a subset of my research questions one to three: How does institutional proximity affect collaboration? How do people collaborate? What reasons do people give for collaborating? More details of the methods used in this dissertation can be found within individual results chapters (Chapters 2-5) and in Appendix A. 1.6.1 Reflexivity A researcher?s reflection on their own data-making role is called reflexivity and reflexivity plays an important role within qualitative research (Richards, 2005). For me, an important component of reflexivity in this research was acknowledging that power dynamics between the interviewer and participant can play a role within the interview setting (Karnieli-Miller et al, 2009). The majority of my interviews were with individuals in research, service, and research/service roles. Individuals in these roles often supervise university students, and likely view their relationship to PhD students to be that of teacher to student. Other participants included chief scientific officers and CEOs of biotechnology companies or relatively high-ranking administrators within the public service. A small number of participants included laboratory technicians and fellow students. In essence, the majority of my interviews involved interviewing elites where I was in a position of relatively little power. To a certain extent, the power asymmetry worked to my advantage. Some participants stated that they agreed to the interview to ?help? a PhD student in their research. These individuals may have felt in the interview that they were ?teaching? me, potentially going into more detail in their answers. Others likely did not reply to my request for an interview because of my relatively low rank. To    32  try and counteract this asymmetric dynamic, I dressed professionally for all interviews and tried to appear self-confident. However, I recognize that participants? attitudes towards me would have been different if I had been an established, middle aged, male researcher. My study assumes that collaboration between individuals and across sectors is important for knowledge translation and innovation. My questions similarly focus on the barriers and incentives, as well as the role that collaboration plays in individual?s day-to-day work. This impacts my final narrative which investigates how collaboration occurs without explicitly exploring whether collaboration, in itself, is a good thing. 1.7 Structure of the dissertation I explore Vancouver?s I2 organisational field in the chapters that follow. Chapters 2 and 3 focused on the structure of collaborations and strongly relied on quantitative analysis of Vancouver?s I2 network. Chapter 2 quantitatively analysed the role of institutional and geographic proximity in Vancouver?s global I2 R&D network, Chapter 3 explores sectoral proximity as a type of institutional proximity in Vancouver?s global I2 R&D network ignoring the effect of geographic proximity. Chapter 4 combined qualitative and quantitative methods and incorporated institutions and organisations into the analysis shifting focus to Vancouver?s I2 organisational field. Chapter 4 mapped the major organisations and regulative institutions involved in Vancouver?s local I2 R&D networks. In Chapter 5, drawing on qualitative methods, I discussed individual?s perspectives on how institutions and capital influence their collaboration decisions. Chapter 6 concludes. In drafting these chapters, both Chapters 2 and 3 were written in manuscript style; Chapter 3 has been published (Lander, 2013) and Chapter 2 is currently under review. These chapters thus read more like self-contained investigations. Chapters 4 and 5 were conversely drafted together as a case study. These chapters are thus more closely integrated together and will subsequently be edited and submitted for publication as a single manuscript.      33  2 Chapter 2: Proximity at a distance in I2 research 2.1 Introduction The importance of research collaborations appears to be growing for both scientists and policy makers (Hoekman et al, 2010; He et al., 2009; Wagner and Leydesdorff, 2005; Narin et al, 1991). While collaboration is generally viewed as desirable, the literature is unclear about with whom scientists should and do collaborate. Scientific collaboration is sometimes shown as transcending geographic boundaries, motivated by a particular problem or sector (Ramlogan et al, 2007; Carlsson, 2006; Malerba, 2005). In other cases, collaboration is depicted as geographically constrained, allowing for dense, embedded relations important for fine-grained information exchange (Reagans and McEvly, 2003; Ahuja, 2000; Burt, 2000; Uzzi, 1997). Often presented as a dichotomy, these motivations to collaborate can instead be conceptualized as different dimensions of proximity between individuals (Frenken, 2009; Ponds et al, 2007; Boshcma, 2005; Torre and Rallet, 2005). The focus of this chapter is this multidimensional concept of proximity that goes beyond mere geographic closeness to include individuals who are institutionally proximate by sharing organisational structures, regulations, norms, or cultures. I explored multiple dimensions of proximity through a case study of the world-wide collaborations formed by members of Vancouver, Canada?s I2 research network. I add to previous analyses of proximity by investigating the impact of proximity on collaboration usingindividuals as opposed to organisations as the unit of analysis  and by exploring multiple dimensions of institutional proximity and testing the relation between geographic and institutional proximity. I traced Vancouver?s I2 research collaborations through co-authorships sourced from the ISI-Web of Science, a recognised approach to mapping such networks (see for example Sun and Negishi 2010; Morlacchi et al. 2005; Meyer 2002; Murray 2002; Newman 2001; Sandstrom et al 2000). Co-authorship and citations of journal publications are often seen to represent the transfer of knowledge. Murray (2002), for example, emphasizes co-authorship as an informal means of knowledge translation.  In my study of proximity, I address a methodological shortcoming of previous studies which took organisations, rather than individuals, as the unit of analysis (see for    34  example Hoekman et al, 2010; Thijs and Glanzel, 2010; Ponds et al, 2007; Wagner and Leydesdorff, 2005; Sandstrom et al, 2000; Hicks and Katz, 1996). Until recently the ISI Web of Science did not provide individual-organisation affiliations and analyses of Web of Science data were unable to properly weight interorganisational collaboration by author. This obscured collaborations within the same organisation, a troubling oversight since collaborations within organisations require a high degree of geographic and institutional proximity (He et al, 2009). Previous analyses were also unable to capture to what extent individuals who had affiliations with more than one organisation facilitate collaborations between organisations, information that both strengthens and complicates analysis. For example, five scientists affiliated with the University of British Columbia might write an article with one scientist affiliated with both the University of British Columbia and Boston General Hospital. Previous analyses based on the organisational level counted this as one collaboration between the University of British Columbia and Boston General Hospital. When this collaboration is analysed at the individual level it can instead be viewed as entirely local; all authors are affiliated the University of British Columbia in Vancouver. Conversely, it could be analysed as a bridging collaboration. The scientist affiliated with the University of British Columbia and Boston General Hospital is acting as a bridge between the hospital and university sectors. Individuals in such bridging roles are often believed to hold a privileged place in networks (Burt, 1992). By conducting my analysis on the individual level I was able to investigate to what extent individuals in these bridging roles are able to facilitate interorganisational collaboration.   This study represents the first attempt to my knowledge to conduct a large-scale co-authorship network analysis based on individuals as opposed to organisations. By analysing collaborations between individuals based on organisational affiliations, I was able to add a methodological robustness that has previously been lacking to an analysis of the different dimensions of proximity. This adds accuracy and nuance to comparisons of institutional and geographic proximity and enables an exploration of the role that multiple affiliations play within these large networks.        35  This study?s focus on Vancouver, Canada allows, at least in part, for the untangling of geographic and national institutional proximities. Vancouver suffers from what Gilding (2008) has called a ?tyranny of distance.? Geographically close to American cities such as Seattle and Portland, Vancouver is geographically separated from other major Canadian cities by the Rocky Mountains and the sheer scale Canada?s landmass. Calgary, the closest Canadian city of comparable size, is 675 km away with Toronto, Ottawa, and Montreal an average of 3,500 km distant.  Perhaps because of its geographic separation from the rest of Canada, Rees (2005) found that for firms within Vancouver?s biotechnology sector almost half (41%) of all collaboration occurred with the USA. Only 23% occurred locally (within Greater Vancouver) and 9% with the rest of Canada. The final 27% occurred outside of North America. In part to compensate for the country?s scale, strong institutional structures have been put in place to encourage national research and development collaborations through programs such as the Networks of Centres of Excellence (Atkinson-Grosjean, 2006). Because geographic and national institutional proximity are not completely correlated for Vancouver, the city offers a strong case study for understanding the different influences that these two dimensions of proximity play. In addition, past analyses of R&D networks tend to focus on cities and regions that are world-renowned in a particular field (see for example Casper, 2007; Owen-Smith and Powell, 2004; Keeble, 1999). While these analyses are useful, ?global leaders? may collaborate differently, both internally and externally, than other networks (Gilding, 2008; McKelvey et al., 2003; Rees, 2005). A case study of Vancouver?s I2 network may thus advance understanding of how individuals in networks that are not global leaders interact, both locally and internationally. Using the case of Vancouver, this chapter investigates the impact of multiple dimensions of proximity. In addition to its theoretical contributions, this study has important implications for science policy. First, this study tests whether policy initiatives within Canada aimed at encouraging national collaboration have been successful by exploring the extent to which Vancouver authors collaborate nationally. In addition, this chapter offers insight to policy makers interested in promoting clusters in high technology areas such as biotechnology. In general, collaboration across sectors is seen as a prerequisite for knowledge translation between research and development. By exploring individuals?    36  collaborations across sectors and the degree to which intersectoral collaborations are localised within Vancouver, I tested whether an I2 research cluster exists locally. In Section 2.2, I review the literature and suggest a number of hypotheses. Section 2.3 describes the data sources and coding procedures in detail while Section 2.4 focuses on Vancouver?s I2 global collaborations and provides results of the effect of geographic and institutional variables on individual-level co-authorship rates. Section 2.5 summarizes these results and discusses their implications for my hypotheses. Section 2.6 concludes the chapter. 2.2 Proximity Collaborations enable knowledge creation, resource sharing, and act as information conduits for existing information (Bathelt et al, 2004; Ahuja, 2000; Hansen, 1999; Burt, 1992; Uzzi, 1997). Networks enable collaborations between a set of individuals or organisations. Much of the past network research has focused on how different types of connections, such as strong or weak ties, affect innovation and knowledge translation. Weak ties, which are distant and infrequent relationships, often provide non-redundant bridges between two or more networks and are seen as important conduits of novel information (Hansen, 1999; Burt, 1992; Granovetter, 1973). Conversely, strong ties, which are close relations based on multiple facets, are viewed as important for fine grained information transfer and increased sharing of tacit knowledge (Reagans and McEvily, 2003; Hansen, 1999; Uzzi, 1997). Less work has been done on understanding how geography affects information transfer within networks (Owen-Smith and Powell, 2004). Some academics argue that knowledge transfer related to a particular problem or sector transcends a geographic area (Malerba, 2005). For the case of biomedicine, Ramlogan et al (2007) argue that biomedical innovation occurs around a specific problem and is not constrained within geographic, technical, or institutional boundaries. The non-localized nature of knowledge translation may become especially important in an era of increasing globalization (Carlsson, 2006).     37  Other academics argue that geographic concentration is important for knowledge exchange, particularly in knowledge-intensive sectors such as biotechnology and finance (Asheim and Gertler, 2005; Owen-Smith and Powell, 2004; Cooke et al, 1997). Katz (1994) found that the likelihood of co-authorship decreases exponentially as the distance separating pairs of authors within nations increases. This is similar to the concept of ?small world? where networks cluster geographically with a relatively small number of remote links (Powell et al, 2005). Local networks enable the dense, embedded relations that are important for knowledge exchange; in particular, they allow for more fine grained information transfer and create a local ?buzz? based on more informal knowledge sharing (Bathelt et al, 2004; Owen-Smith and Powell, 2004; Reagans and McEvily, 2003; Ahuja, 2000; Burt, 2000; Uzzi, 1997). Face-to-face contacts are seen as particularly important where knowledge is tacit rather than codified (Fritsch and Kauffeld-Monz, 2010; Polanyi, 1967). The concept of geographic concentration has gained particular ground, building on the understanding that organisations from different sectors that are focused on a specific field of knowledge tend to ?cluster? in specific geographic regions. For the field of biotechnology, researchers have explored clusters in Australia (Gilding, 2008), Boston (Owen-Smith and Powell, 2004), Sweden (McKelvey et al, 2003), and Vancouver (Wixted and Holbrook, 2011; Salazar et al, 2008; Rees, 2005), among others.  More recent work has argued that a network can derive benefits from both geographic concentration and the bridging of dispersed geographic areas by having a dense network that is connected to individuals outside this dense network (Reagans and McEvily, 2003; Burt, 2000). Thus, both global and local networks can be important for knowledge translation but in different ways (Torre and Rallet, 2005; Bathelt et al, 2004; Owen-Smith and Powell, 2004).  Beginning in the 1990s, the French School of Proximity Dynamics has argued that the concept of proximity covers a number of dimensions (Boshma and Frenken, 2010; Frenken et al, 2009; Ponds et al, 2007; Boschma, 2005; Torre and Rallet, 2005). Boschma (2005) introduced five different forms of proximity: cognitive, organisational, social, institutional, and geographical.     38  1. Cognitive proximity explores the extent that two agents share the same knowledge base (Frenken et al, 2009). It addresses the tacit, localised, and cumulative nature of most knowledge. Knowledge creation involves bringing new knowledge together but, to be effective, agents need to have the absorptive capacity to receive this new knowledge. To effectively absorb new knowledge, an agent needs prior related knowledge (Lane and Lubatkin, 1998; Szulanski, 1996; Cohen and Levinthal,1990). This related knowledge puts them in cognitive proximity with the agent that they are receiving knowledge from. 2. Organisational proximity can be defined as the degree that two agents are affiliated with the same organisation (Frenken et al, 2009). Organisations will bring together people who might not otherwise associate (Feld, 1981). Individuals within the same organisation often share routines, norms, or knowledge (Frenken et al, 2009; Boshma, 2005; Torre and Rallet, 2005).   3. Social proximity draws on the concept of embeddedness in social network theory (Uzzi, 1997; Granovetter, 1985). In embedded relations, ties between agents often exist along multiple dimensions and include social and economic relations. Because embedded ties include multiple dimensions, trust in the tie increases and embedded ties are more likely to persist (Dahlander and McFarland, 2013).  4. Boschma (2005) called his fourth dimension of proximity institutional and defined it as including formal laws as well as cultural norms and values. To both differentiate this dimension from a broader conception of institutions (see my discussion below) and concretely operationalize this dimension, I have relabelled institutional proximity as sectoral and national proximity. Sectoral proximity can be defined as the extent to which agents operate under similar organisation level incentive structures (Frenken et al, 2009). Agents affiliated with the same nation have national proximity through shared formal laws, cultural norms, and values. 5. Geographic proximity builds off of the concept of geographic concentration and here represents perceived geographic distance between two individuals.  Boschma?s (2005) five different dimensions of proximity in many ways overlap. For example, organisational proximity contains a spatial component; agents working within the same organisation are geographically close together. In addition, all five can    39  be categorised into two broader dimensions?geographic and institutional?in which cognitive, organisational, social, sectoral, and national proximity are all subsumed within a broader institutional category. This definition of institutions is in keeping with Scott (2008, p.48) who defines institutions as ?regulative, normative and cultural-cognitive elements that, together with associated activities and resources, provide stability and meaning to social life.? Institutions are legitimated prescriptions and scripts that describe behaviour while being so entrenched in society that they become taken for granted. Institutions form a basis of what is considered legitimate and adhering to institutions increases legitimacy.  Organisations can have their own institutions that are shared by individuals affiliated with the organisation. Other institutions can be specific to an organisational department, a sector, profession, social group, or society (Scott, 2008; Berends et al, 2003; Barley and Tolbert, 1997).  Boschma (2005) recognises this overlap and argued that what differs between his institutional dimensions are their level: social proximity is shared at the individual level, organisational proximity at the organisation level, institutional proximity?here sectoral and national proximity?at the macro level. Thus I explore cognitive, organisational, social, sectoral and national proximity as dimensions of a broader institutional proximity category. Individuals can be proximate along multiple, and sometimes overlapping, institutional dimensions.  Torre and Rallet (2005) argue that individuals are more likely to collaborate if they are in either geographical or institutional proximity. Boschma (2005) expands on this by arguing that geographical proximity can compensate for a lack of institutional proximity and the reverse: that institutional proximity can compensate for a lack of geographical proximity. As shown in Figure 1, geographic and institutional proximity can be theoretically viewed within an analytic grid that predicts the relative likelihood that two individuals will collaborate. Individuals who are institutionally and geographically proximate have a high likelihood of collaborating; individuals with close geographic proximity who are not institutionally proximate are less likely to collaborate as are individuals who are institutionally proximate but distant in geographic proximity; Individuals who are neither institutionally nor geographically proximate will be the least likely to collaborate. Ponds et al (2007) tested these concepts for the case of Dutch research collaborations in four broad disciplinary areas: agriculture and food chemistry;    40  analysis, measurement and control technology; biotechnology; and information technology. Because scientific and industrial research involve different institutional norms, the authors argued that geographical proximity can compensate for institutional differences in intersectoral collaborations, thereby facilitating field based local clusters.  This chapter builds on the findings of Ponds et al (2007) by means of a case study of Vancouver, Canada?s I2 worldwide research collaborations. By choosing a specific scientific field, cognitive proximity is controlled for and is assumed to be small (Ponds et al, 2007). Geographic proximity is explored in this chapter by geographic region and by distance measured as binary radiuses centered on Vancouver. Institutional proximity is explored through shared organisational, sectoral and national affiliations between individuals. Individuals from the same organisation or city are considered to be both institutionally and geographically proximate. The analytic grid outlined in Figure 1 is articulated through four hypotheses: first, that geographically proximate individuals will show greater proclivity to collaborate; second, that institutionally proximate individuals will show a greater proclivity to collaborate; third, that individuals who are both institutionally and geographically proximate will show the greatest proclivity to collaborate; and, finally, that geographical proximity will be able to compensate for lack of institutional proximity, and the reverse: that institutional proximity will be able to compensate for a lack of geographical proximity.    2.3 Data and methods 2.3.1 Developing an I2 database Appendix A describes in detail the process used to create an I2 database used in this dissertation. This chapter focuses on collaborations between individuals during 2008-2011 inclusively, when it is possible to connect authors directly to organisations within ISI Web of Science bibliometic records.  Overall, the I2 filter developed for this analysis led to the extraction of 762 papers. The same author or organisation was often written multiple ways within the database. Authors? names were cleaned through truncation to last name and first initial. Organisational names were manually cleaned. The definition of organisation in this    41  analysis was fairly broad; for example, hospital and university centres and departments were subsumed within the larger hospital and university. Organisations physically located in different cities and/or countries that were part of the same organisation, for example, local offices of large multi-national companies, were treated as separate organisations.  Organisations were also coded by sector: hospital, university, firm, government, and non-governmental organisation (NGO). To increase reliability, the primary researcher and a second individual both coded organisations into sectors based on guidelines developed by the primary researcher. Sectoral affiliation was first determined by the organisation?s name; if affiliation remained unclear, organisations were researched online. In the guidelines, academic teaching hospitals were coded as hospitals. Large, multi-hospital administration systems were coded as government or firm depending on their ownership.  This was only really an issue in the USA. There, New Haven Healthcare System and Sunnybrook Health were coded as ?firm? while Veteran Affairs administrative units were coded as ?government.? Public research institutes with highly limited teaching capacity were coded as government. Organisations with non-profit status were coded as NGOs. Consortia of organisations that were non-profit in nature were coded as NGOs, an example is the Victorian Partnership for Advanced Computing, Australia. After both individuals had coded a subset of all organisations, sectoral coding was compared and showed almost perfect intercoder reliability with no significant difference in the coding (?=.809, p=.000).3 All organisations where coding differed were discussed by both coders and a common sector was agreed upon. The primary researcher subsequently coded additional organisations independently. The organisations? addresses were used to map an entity to its respective city and country and to classify organisations into local (Metro Vancouver), national (rest of Canada), USA, the rest of North America, Europe, Asia, Africa, Oceania, and South America categories.                                              3 See McGinn et al (2004) for guidelines on how to calculate and interpret the ? statistic    42  A subset of this larger database was then extracted for this analysis. Based on discussions with a local I2 expert, three types of papers-articles, letters, and reviews-were chosen to represent R&D collaborations (186 papers discarded). Additionally, within the years of analysis, authors were not always properly mapped to organisations. Papers were thus omitted where listed authors had no organisational affiliation or not all listed organisations were affiliated with an author (111 papers discarded). Finally, papers with 60 or more authors were excluded as outliers. For these ?big science? products it was assumed that meaningful collaboration could not have occurred between authors (2 papers discarded).  2.3.2  Variables and analytic techniques Tie creation is the dependent variable used in this analysis. It represents collaboration between two authors and is measured as a co-authorship between two authors on one paper. For example, a paper listing five co-authors would lead to 10 co-authorships listed in the database, one between each of the co-authors listed in the paper. Each author was affiliated with one or more organisations and was given a set of attributes related to their organisational affiliations: the organisation?s name and its city, country, and sector. I used UCINET 6.32 (Borgatti et al 2002), Gephi 0.8.1 (Bastian et al, 2009), and R (R Core Team, 2012) to analyse these data. Some procedures used in analysis, such as block modeling and density calculations, required a one-to-one relationship between authors and their affiliations on each of their publications.  These analyses were based on initial two-mode dataset that connected authors to the papers they co-authored. This matrix was converted into a one mode matrix through cross product analysis.4 The resulting one mode matrix connects authors to authors through co-authorship. Block model and density calculations were based on city affiliations of each author. Authors with multiple affiliations complicated these calculations. While most of the authors with multiple affiliations were affiliated with organisations in the same city, almost 8% (176) listed affiliations in two or more cities. I created a proxy ID that uniquely combined author and city Ids (AutCitID) to simulate a                                             4 See Borgatti and Everett (1997) and Wasserman and Faust (1994) for further discussions of two-mode matrices and their conversion to one-mode matrices    43  one-to-one relationship between authors and cities per paper. This proxy ID was used in block modeling and density calculations. A quasi-Poisson random effects regression is used to test the significance of hypotheses one through four. Table 2 presents variables used in this analysis. Vancouver authors and their potential co-author pairs are the units of observation. Each Vancouver author is compared to all authors within the database, excluding Vancouver authors with themselves. This violates the independence assumption of regression. The dependent variable in this analysis is the count of the number of co-authorships by these Vancouver author/potential co-author pairs. Most author pairs have no co-authorships. Poisson models are commonly used when modeling count data with a large number of zeros (Long, 1997). Newman (2003) has shown that co-authorships follow a Poisson distribution. In a Poisson model, variance is assumed to be identical to the mean and dispersion is fixed at 1. This can incorrectly estimate the actual variance in the data rendering model-based tests inaccurate. In a quasi-Poisson model the dispersion parameter is not assumed to be 1 but is instead estimated from the data. This leads to the same co-efficient estimates as the Poisson model but inference is adjusted for the data?s dispersion (Zeileis et al, 2008). Both fixed and random effects models can be used to control for interdependence of cases in regressions.  Hilbe (2011) argues that fixed effects are most appropriate when less than 20 groupings are used while random effects estimators are more efficient for larger groupings for the data.  I chose random effects because of the large number of groupings in my data.    44  Table 2: Variables for quasi-Poisson regression analysis Variable Description Number of Collaborations Dependent variable. Count of the number of co-authorships between the Vancouver author and potential co-author pair. OrgSame Reference group. The Vancouver author and potential co-author are affiliated with all of the same organisations in Vancouver. Geographic proximity VanDIFOrg A dummy variable indicating whether the Vancouver author and potential co-author pair are affiliated with completely different organisations in Vancouver (1='yes', 0='no'). Also illustrates institutional proximity. SH Drive A dummy variable indicating whether the potential co-author's geographically closet affiliation is outside Vancouver but within 250 km  (1='yes', 0='no') SH Fly A dummy variable indicating whether the potential co-author's geographically closet affiliation is greater than 251 and less than 700 km away from Vancouver (1='yes', 0='no') LH Fly A dummy variable indicating whether the potential co-author's geographically closet affiliation is greater than 701 and less than 4,600 km away from Vancouver (1='yes', 0='no') Global A dummy variable indicating whether the potential co-author's geographically closet affiliation is further than 4,599 km from Vancouver (1='yes', 0='no') Institutional proximity SectorDIF A dummy variable indicating whether the Vancouver author and potential co-author pair are affiliated with completely different sectors (1='yes', 0='no') SectorBR A dummy variable indicating whether the Vancouver author and potential co-author pair are affiliated with both the same and different sectors. In this case one or more of the pair has multiple affiliations and is bridging similar and different sectoral affiliations (1='yes', 0='no') CountryDIF A dummy variable indicating whether the potential co-author has any Canadian affiliations (1='no', 0='yes')  Through this regression I test if co-authorships are random or show preferences based on different proximity variables. The reference group used in this regression are when the Vancouver author and potential co-author are affiliated with all of the same organisations in Vancouver. Variables measure the different dimensions of institutional    45  proximity. I controlled for cognitive proximity by choosing I2 as a scientific field. I did not test for social proximity. To measure organisational proximity I took an approach similar to He (2008) and created a dummy variable indicating whether the Vancouver author and potential co-author pair are affiliated with completely different organisations in Vancouver. Previous authors have tested for sectoral proximity as a series of organisation-level dummy variables (Ponds, 2009; Ponds et al, 2007; Frenken et al, 2005). Based at the organisational-level, these variables could not differentiate between author pairs affiliated with completely different sectors and author pairs where one author?with multiple affiliations?bridges two sectors. Returning to the example above, if five scientists affiliated with the University of British Columbia write an article with one scientist affiliated with the University of British Columbia and with Boston General Hospital, this sixth author acts as a bridge between university and hospital sectors. To test if intersectoral collaboration is more likely to occur when an individual acts in such a bridging role, I created two sector variables. The first sector variable?sector different?acts as a dummy variable indicating whether a Vancouver author and their potential author pair are affiliated with entirely different sectors.  The second sector variable?sector bridge?acts as a dummy variable indicating whether a Vancouver author and their potential author pair are affiliated with some of the same sectors as well as some different sectors. In this case, one or both of the authors in the potential pair act as a bridge between a shared and a different sector. The sixth author described above acts in this bridging role. Based on Burt?s (1992) work, author pairs with this sector bridge attribute can be viewed as bridging network structural holes and should have a large amount of social capital.  I created an additional institutional proximity variable?country affiliation?that indicates whether the potential co-author is affiliated with Canada. Geographic proximity has been measured a variety of ways including travel time (Ponds et al, 2007), kilometric distance (Hoekman et al, 2010), or as a binary variable (Hoekman et al, 2010; He, 2008; Ponds, 2009). Several studies have used a gravity model for distance (Hoekman et al, 2010; Ponds et al, 2007). The gravity model is best described through the following formula (Ponds et al, 2007):    46  C = 	       (1) Here Cij?the gravity?represents the intensity of collaboration between authors divided by the distance between them. Authors? masses?Ai and Bj?symbolise each author?s total publications. Authors with greater mass have a stronger gravity. I ran regressions based on this gravity variable and found that it did not have good explanatory value. The gravity co-efficient was slightly greater than 1 which indicated that for every additional 1,000 km from Vancouver that a potential co-author was located they published 0.6% more with Vancouver authors. This small and counter intuitive finding hinted that that a non-linear relation may exist between distance and collaboration. I also included authors? masses as a regression variable. By weighting regressions by authors? masses I hoped to adjust for authors with large numbers of publications, providing a means to compensate for the Vancouver-based author publication bias within my sampling frame. I found that regressions with and without this weight did not materially differ with the largest difference in the sector bridging coefficient which increased by 3% when the weigh was added. On average, coefficients from regressions with and without this weight differed by approximately 1%. Torre and Rallet (2005) argue that geographic proximity should be binary, rather than a continuous distance variable, based on perceived distance between individuals. Distance is either ?close? or ?far.? England and India are both ?far? from Vancouver. To an individual deciding whether to travel to England or India, the difference in ?farness? between these two locations in kilometric distance is relatively unimportant. Seattle, two and a half hours away by car, is ?close.? Perception of nearness and farness is based on travel time using available transportation; a distance that is ?far? to drive may be ?close? to fly. Different measures of ?farness? thus depend on the type of transportation used.  Based on this concept of perceived distance, Lublinski (2003) define geographically proximate collaborators as within a radius of a two-hour drive in Germany. Ejermo and Karlsson (2006) differentiate between two types of ?farness,? driving and flying, in their analysis of geographic proximity in Sweden.  I took this more nuanced approach to distance and created series of binary rings centered at Vancouver that measure the distance of the affiliations of each potential co-   47  author from Vancouver. North Americans have their own perceptions of ?farness? that differ from Europeans. Because North America is so geographically vast, North American researchers may view cities to be ?close? while travelling a distance that European researchers would view as ?far.? To test whether North American researchers view distance differently than Europeans, I divided geographic distance into five radial distance categories: within Vancouver, short haul driving, short haul flying, long haul flying, and global. The institutional variable, Vancouver different organisation, described above, acted as the first ring where potential author pairs were not affiliated with any of the same organisations as the Vancouver author but also worked in Vancouver. Short haul driving is the farthest distance comfortably driven round trip in a day. Seattle, approximately 250 km south of Vancouver, is the outer distance. Short haul flying is the farthest distance comfortably flown round trip in a day. Calgary, approximately 700 km away and a two and a half hour flight from Vancouver, is at the outer limit. Long haul flying differentiates between travel times within and outside of North America based on the conjecture that North Americans perceive of travel within North America as long, but not necessarily ?far? in the way that they view travel outside of North America. Miami, 4600 km away, is at the outer limit of long-haul flying. All other distances are global.  Multiple affiliations affected the construction of all dependent variables as they are based on the organisational affiliations of authors. For dummy variables city not organisation and country, authors with multiple affiliations can be in more than one variable category potentially leading to an increased likelihood of the potential co-author sharing an organisational or Canadian affiliation. Distance between authors is affected if a potential co-author is affiliated with organisations in multiple cities; almost 8% of authors fall into this category. For these authors, I took the minimum distance of a potential co-author?s affiliations from Vancouver as the distance. Torre (2008) argues that temporary co-presence in the form of short or medium term visits are often enough to enable the information exchange necessary for local collaborations. Authors affiliated with organisations located in different cities likely spend part of their time physically located in each affiliated city, enabling ?local? collaborations to form.     48  2.4 Results 2.4.1 The geographic and institutional structure of collaboration During the period studied (2008-2011 inclusive), Vancouver?s I2 R&D network published 463 papers involving 2323 co-authors representing 630 organisations, located in 305 cities, in 49 countries, on 6 continents. Each paper listed a mean of 7.73 authors, 4.18 organisations, 3.20 cities, 1.89 countries, and 1.38 continents. Almost 25% of authors (535) were affiliated with more than one organisation. As shown in Table 3, the rate of multiple affiliations varied by region and was the highest in Africa (35%), followed by Canada (33%). The lowest observed rates were in the rest of North America category, which is made up of Mexico and the Caribbean (0%), followed by the USA (15%). In Vancouver, multiple affiliations mainly involved the public sector; university affiliations (227 authors) were the largest, while hospital (112 authors) and government (145 authors) affiliations had roughly the same numbers. Only 9 multiple affiliations in Vancouver involved firms. Table 3: Authors with multiple affiliations Vancouver 32.80% Rest Cda. 32.95% USA 14.63% Europe 20.86% Asia 24.11% Africa 34.62% Oceania 26.67% S. America 27.27% Rest N. Am. 0.00%  To explore relations between cities, Figures 3 and 4 map all cities involved in Vancouver?s I2 R&D network between 2008 and 2011. I created these sociograms by collapsing the one-mode author matrix into a city co-authorship matrix through a block modeling procedure based on the AutCitID. Table 3 outlines the number of authors affiliated with each region, their papers, and the density of co-authorships of authors from each region with Vancouver affiliated authors. The number of authors? affiliations in Table 4 adds up to more than the total number of authors because of multiple affiliations.    49  Co-authorship density is based on the AutCitID proxy and represents the proportion of ties between Vancouver authors and the authors of a given region that are actually present over the total number of possible author ties; a density of 1 means that all authors in Vancouver co-authored at least one paper with all authors of that region (Hannenman and Riddle, 2011). Density rates overall were low, which is in keeping with previous studies that found that scientific co-authorship networks are sparse (Newman, 2001).  Figure 3: Co-authorships outside of North America       50  Figure 4: Co-authorships in Canada and USA   Together Figures 3 and 4 and Table 3 show that authors in Vancouver co-author the most with others in the same city, then with others in Canada, followed by the USA and Europe. In fact, after Vancouver, 4 of the top 5 city affiliations of co-authors (Toronto, Montreal, Ottawa, Winnipeg, and London, England) are in Canada. Density was the highest in Vancouver, followed by co-authorships with authors in the rest of Canada. Of the top ten US collaborating cities by co-author, four are on the west coast (Seattle, La Jolla, San Francisco, and Los Angeles) while six are located in the eastern states of the USA (Boston, Bethesda, Philadelphia, New York, Pittsburgh, and Baltimore); the middle and southern states are less well represented. In North America overall co-authorships appear concentrated in the midwest and the north-east. Top cities in Europe include London, Oxford, Amsterdam, Leicester, and Geneva while, within Asia, the largest number of co-authorships is with Iran followed by China and Japan. Collaboration within Asia overall was quite low, only 18 papers included co-authors from Iran.     51  Table 4: Affiliations by region Region # Author # Paper Density Vancouver 811 463 0.0113 Rest Cda 528 174 0.0039 USA 458 134 0.0031 Europe 417 107 0.0030 Asia 112 37 0.0022 Africa 52 15 0.0044 Oceania 45 14 0.0023 S. America 11 5 0.0017 Rest N. Amer 8 2 0.0023  Some 75% of African co-authors are located in Uganda and Kenya and collaborations were highly focused on HIV/AIDS research, with 14 of 15 publications on this topic. Interestingly, density between Vancouver and Africa was the second highest in the dataset. Thus these collaborations are geographically focused with a small number of co-authors but include a relatively large proportion of Vancouver authors.   Table 5 outlines how Vancouver?s regional co-authors differ by sector. Here again co-authors may be counted more than once if they have multiple affiliations in multiple regions and sectors. Overall, university co-authorships dominated for all regions with governments the second most important sector for the rest of Canada, Asia, and Africa. Hospitals ranked third in importance in co-authorship and number of organisations except in Europe, South America, and Oceania where they ranked second.    52  Table 5: Affiliations of Vancouver's co-authors by region and sector   Uni Govt Hosp Firm NGO Total Vancouver 551 272 191 24 0 1038 Rest Cda 350 158 117 12 18 655 USA 288 47 64 62 32 493 Europe 275 56 85 25 32 473 Asia 78 20 7 7 6 118 Africa 18 17 6 0 13 54 Oceania 32 4 12 0 6 54 S. America 8 0 4 0 1 13 Rest N.Amer 8 0 0 0 0 8 Total 1608 574 486 130 108 2906  Almost half of all firm co-authors and organisations (32 out of 73) were American, showing the importance of the American private sector in this network. Vancouver and European firms tied for second place in both co-authors and organisations (13 each), followed by firms in the rest of Canada. With the exception of Africa and Oceania, NGOs did not play an important role. Of interest is that co-authors in the rest of Canada come from relatively few organisations. Within Canada, government makes up the largest number of organisations (52), with both universities and hospitals measuring in the mid 30s.  2.4.2 The role of proximity in collaboration I further explored the effect of proximity on proclivity to collaborate through a regression analysis. Table 6 presents the results from the regression as incidence rate ratios. Vancouver author/potential co-author pairs where both authors are affiliated with all of the same organisations in Vancouver are the reference group. In Model 1, the quasi-Poisson random effects regression includes all geographic proximity variables. It shows that all geographic proximity variables decreased the number of co-authorships between the pairs compared to the reference group (p<0.001). Authors from Vancouver but who share none of the same organisational affiliations publish one tenth the amount that authors from Vancouver who share an organisational affiliation. In fact authors from Vancouver who share no organisational affiliation publish less together than Vancouver    53  authors do with authors outside of Vancouver. Outside of Vancouver, the greater the distance of the potential co-author, the lower the number of co-authorships. Of the geographic proximity variables, the number of co-authorships was the highest when potential authors were a short haul drive from Vancouver. These co-author pairs published almost 80% less than the reference group. Differences in incidence rate ratios between short haul flight, long haul flight and global travel were less pronounced; co-author pairs published between 86% and 89% less than the reference group. I ran Model 1 again with short haul flight as the reference group to see if significant differences existed between the short haul flight, long haul flight, and global travel variables. Differences between short and long haul flight were not significant while differences between short haul flights and global flights were (p<0.05). Long haul flight potential co-authors published 6% less than short haul flight potential co-authors while global potential co-authors published 20% less.     54  Table 6: Quasi-Poisson random effects incidence rate ratios   M1   M2   M3   Intercept 0.018 *** 0.022 *** 0.024 *** Geographic VanOrgSame REF  REF  REF  VanDIFOrg 0.092 *** 0.136 *** 0.106 *** SH Drive 0.202 *** 0.295 *** 0.311 *** SH Fly 0.143 *** 0.177 *** 0.252 *** LH Fly 0.135 *** 0.191 *** 0.126 *** Global 0.114 *** 0.185 *** 0.123 *** Institutional  AllSectorSame REF REF REF SectorDIF 0.380 *** 0.507 *** SectorBR 0.754 *** 0.656 *** Canada REF REF REF CountryDIF 0.800 *** 0.849 *** Interaction SctD/SHD 0.413 *** SctD/SHF 0.135 *** SctD/LHF 0.846 SctD/Glob 1.079 SctB/SHD 1.086 SctB/SHF 0.769 SctB/LHF 2.011 *** SctB/Global 1.530 *** Significance: p<0.001 '***', p<0.01 '**', p<0.05 '*' In Model 2, institutional proximity variables related to organisation, sector, and country are added to the quasi-Poisson random effects regression. The geographic coefficients remained roughly equivalent between Models 1 and 2, differing by less than 10%. This relative stability after I introduced the country variable illustrates that any potential co-linearity between the geographic and country variables does not change the geographic coefficients in any considerable way. All institutional variables significantly (p<0.001) decrease the expected number of co-authored papers with Vancouver authors. Author pairs where at least one author acts as a bridge between sectors are    55  more likely to publish together than authors from entirely different sectors; authors from entirely different sectors publish 62% less than the reference group compared to only 24% less when at least one author acted as a bridge. Authors affiliated with organisations outside of Canada publish over 20% less with Vancouver affiliated authors.  Model 3 adds two-way interaction terms between the sectoral affiliation variables and four geographic proximity variables outside of Vancouver. The Vancouver different organisation variable measures both geographic and institutional proximity and is co-linear with the sector variables, thus interaction between these variables was inappropriate. Here we see that both sector difference and bridging exacerbates the consequences of being as a distance. Among Vancouver authors, those from completely different sectors are 50% less likely to publish than those who share all sectoral affiliations. By multiplying the geographic coefficients and sector different/geographic interaction terms together I find that for distances outside of Vancouver, authors from completely different sectors are between 87% and 97% less likely to publish than those who share all sectoral affiliations to publish. This difference peaks for potential co-authors a short haul flight away. When at least one author acted as a bridge, Vancouver authors were 34% less likely to publish than authors who share all sectoral affiliations. By multiplying the geographic coefficients and sector bridge/geographic interaction terms together I find that for distances outside of Vancouver this rate decreases to between 66% and 81% in a non-linear fashion with the difference peaking for potential co-authors a short haul flight away. 2.5 Discussion  My regression analyses for Vancouver?s I2 research network support the hypothesis that collaborations are more likely to occur among geographically close individuals. Vancouver authors are most likely to co-author with others from Vancouver. Vancouver authors are more likely to co-author with individuals a short haul drive away than with authors further afield. There is less difference between the propensities to co-author papers with authors a short haul flight, long haul flight or further distance from Vancouver; hinting that geographic proximity matters the most for potential authors    56  within driving distance. Once a flight is involved, Vancouver authors may not differentiate as much between ?close? and ?far? geographic proximity. From Figures 3 and 4, it is apparent that a large proportion of Vancouver?s co-authors are located within North America, although this may be based on institutional as opposed to geographic proximity. Figure 4 shows that the strongest collaborations within North America are to the Eastern side of the continent, and not along the geographically closer west coast. This may be due, in part, to population density on the North American continent and established research traditions. The trend toward Eastern collaborations appears more prevalent in the USA, hinting that Canadian institutional structures and national policies may encourage collaborations with all regions of Canada. Outside of Canada, many of the top cities that are home to Vancouver?s co-authors may be explained by looking at global centers of academic excellence, while others are top centres for biotechnology (see Ernst and Young 2012 and Times Higher Education 2012 for rankings). Bethesda may be an exception; home to the US National Institutes of Health, it has a large research capacity.  Figure 3 shows a higher rate of co-authorships between Vancouver and Europe than with Asia, even though much of Asia (along the Pacific Rim) is geographically closer than Europe.  Other factors, such as institutional proximity through a shared history and culture, seem to play a more important role here.  My second hypothesis, that institutionally proximate individuals show a greater proclivity to collaborate, was well supported in the regression analysis. Authors from completely different sectors publish almost 62% less than authors completely in the same sector. Authors in bridging roles publish more together than authors in entirely different sectors. This hints that multiple affiliations may confer proximity to a potential co-author that, while not as strong as if the potential co-author shared all sectoral affiliations, may help to encourage collaboration across sectors. Multiple affiliations may help individuals bridge ?structural holes? (Burt, 1992). As shown in Table 2, multiple affiliation rates vary by regions hinting that the importance of these bridging authors also likely varies by region. Authors within Vancouver have the third highest rate of multiple affiliations and the importance of bridging is likely more pronounced in this sample than    57  it would be in a sample of authors from a region, such as the USA, where there is a lower multiple affiliation rate. Authors affiliated with organisations outside of Canada publish 15% less with Vancouver affiliated authors. Both Figure 4 and results in Table 3 show that Vancouver authors will co-author papers with other Canadians, even if the Canadian city is geographically distant.  The third hypothesis, that individuals who are both institutionally and geographically proximate show the greatest proclivity to collaborate, was somewhat supported in the analysis. Within Vancouver, co-authors share institutional and geographic proximity. Table 4 shows that density rates were the highest within Vancouver. In Table 6, potential co-authors affiliated with all the same Vancouver organisation acted as a reference group, the numbers of co-authorships for all other groups are lower. However, Vancouver based authors affiliated with different organisations are less likely to co-author papers together than Vancouver authors are to publish with authors outside of Vancouver. This result does not support the concept of clusters; when Vancouver authors look for collaborators outside of their own organisation they often look beyond Vancouver.  This finding may be exacerbated by the way data was sampled. While all Vancouver-based I2 authors are part of this sample irrespective of whether they co-author papers together, authors outside of Vancouver were extracted from realised instances of co-authorship and not from the universe of all infection and immunity potential co-authors. This sampling bias may have resulted in an under-estimation of the effect of geographic proximity and a relatively lower Vancouver different organisation co-efficient when compared to the other geographic proximity variables. The relatively low level of interorganisational co-authorships within Vancouver may also be an indicator that Vancouver?s I2 cluster is dominated by the public sector and that few authors are collaborating with firms and NGOs locally. Most Vancouver authors with multiple affiliations are affiliated with public sector organisations and their collaborations with other authors affiliated with one of the same organisations would not be recorded as interorganisational. Few Vancouver authors with multiple affiliations are affiliated with    58  firms and none are affiliated with NGOs. Collaborations recorded as interorganisational are more likely to be with these sectors. Low interorganisational collaboration rates within Vancouver may illustrate that Vancouver authors are looking beyond Vancouver to find firm and NGO based collaborators. Vancouver may not have a critical mass within these sectors to support high levels of collaboration. I weighted regressions by authors? masses as a means to indirectly account for the oversampling of Vancouver authors. On average, coefficients from regressions with and without this weight differed by only 1% hinting that my sampling framework did not materially change my results. The results in Table 6 support the fourth hypothesis that geographical proximity can compensate for a lack of institutional proximity and vice versa. For authors from entirely different sectors, Vancouver-based author pairs publish more together than author pairs where one author is outside of Vancouver. The same finding holds for author pairs where at least one author acted as a bridge: Vancouver based pairs published more than pairs where one author is outside of Vancouver. Thus geographic proximity appears to help to compensate for institutional distance.  While the concepts of geographic and institutional proximity have explanatory value in this analysis, other factors beyond the hypotheses tested affect the likelihood of co-authorships. Individual collaborations often involve bringing different resources to bear on a collaborative effort (Atkinson et al, 1998; Katz and Martin, 1997; Melin, 2000). The collaboration patterns seen in this network may result not only from proximity but also from other types of preferential attachment (Powell et al, 2005; Wagner and Leydesdorff, 2005). I submit that it is important to differentiate between resources available at multiple locations and relatively scarce resources. When resources are available in more than one location researchers may pick the location that is the most proximate to them for collaboration. In other instances, a particular researcher may collaborate to gain access to a relatively rare resource, irrespective of proximity. Co-authorships with American firms and Africans can be viewed in this light.    In Table 5, the high number of authors affiliated with American firms when compared to the number of firm affiliations overall is striking. This result supports Rees (2005) who found that Vancouver?s biotechnology firms predominantly collaborated with    59  American firms. Rees (2005, p.304) argued that Vancouver firms? collaborations aided in decreasing the cost and time while accessing experience required to clear regulatory hurdles in the market. I concur that this factor may be promoting Vancouver authors? collaborations with American firms. The USA is the largest market for pharmaceutical drugs and has a relatively complex regulatory environment that must be navigated if drugs are to reach the market. Collaboration with US firms creates institutional proximity by proxy with the USA regulatory environment thereby aiding in the smooth navigation within its labyrinthine structures.    Collaborations with Africa provide access to another relatively rare resource: AIDS patients. Over 90% of the articles co-authored with African affiliated researchers were related to AIDS research; 75% of co-authors were located within Uganda and Kenya. Sub-Saharan Africa is home to an estimated 70% of HIV-infected people worldwide even through the region has only 10% of the world?s population (Cohen, 2000). Western researchers have been increasingly collaborating with African researchers on projects to slow HIV?s spread in the region and change the course of disease for those infected (Cohen, 2000). However, research capacity within Africa is low overall (Nchinda, 2002). Thus, researchers in Vancouver interested in HIV collaborations with African researchers would necessarily focus on a relatively small number of individuals, thus explaining the high density of African co-authors seen in Table 4.     2.6 Conclusion This chapter sought to evaluate the effect of geographic and institutional proximity on proclivity to collaborate through a bibliometric study of Vancouver?s global I2 co-authorships. It found support for both concepts as well as for the dual concepts that geographical proximity can compensate for a lack of institutional proximity and the reverse.  The concept of proximity has explanatory value in this analysis. I argue that the strength of both may break down when dealing with a relatively scarce resource. Here researchers will preferentially collaborate with the few individuals that can provide access to the resource irrespective of their proximity.          60  This study has implications for the science policy community. First it illustrates that policy initiatives within Canada aimed at encouraging national collaboration appear to have been successful. Vancouver researchers show a preference for collaborating with individuals affiliated with Canadian organisations. However, my results offer a more sobering conclusion for policy makers interested in promoting clusters in high technology areas such as biotechnology. Overall, individuals within the I2 network were reluctant to collaborate across organisations; such ?partnerships? are often viewed as a prerequisite for knowledge translation between research and development. Within Vancouver itself, researchers appeared to show an even greater disinclination to collaborate across organisations when compared to collaborations with international co-authors. This result hints that Vancouver may lack sufficient capacity for a thriving biotechnology cluster and supports Rees? (2005) findings that Vancouver biotechnology firms looked elsewhere for collaborative partners. An unanticipated result of this study was the high level of multiple affiliations observed within this dataset and the large variance in multiple affiliation rates across regions. The high number of Canadians with multiple affiliations supports the previous finding of Katz and Martin (1997) that multiple affiliations in Canada are higher than for the UK or Australia. As cautioned by Katz and Martin (1997), however, it is unclear how multiple affiliations should be treated. Rather than representing collaborations between authors in different organisations, they are more likely to represent an agreement between organisations with which an individual is affiliated. For some, multiple affiliations may involve splitting work across organisations. For others, one or more affiliations may be little more than symbolic; nevertheless, the symbolism allows increased access for groups sharing institutions with that affiliation. For the remainder, one or more affiliation may have little or no impact on their propensity to collaborate.  From the data used in this chapter, it is unclear how substantive multiple affiliations are for a given author and how the nature of multiple affiliations varies by country or region. This has important implications when reviewing previous organisational based studies that were unable to capture the prevalence of multiple affiliations. Multiple affiliations also provide an important area for future research,    61  making up a large and growing trend among the scientific research population. To date it is unclear to what extent these individuals may act as bridges between the different organisations, sectors, and countries with which they are affiliated and how this may vary by region.        62  3 Chapter 3: Sectoral collaboration in biomedical research and development5 3.1 Introduction Networks play an important role in research and development (R&D) processes by transferring knowledge and norms between organisations (Ahuja 2000; Hansen 1999; Uzzi 1997; Goes and Park 1997; Burt 1992). Analyses of innovation, and the role of networks in innovation processes, are disproportionately focused on the private sector and the firm as innovator (Edquist 2005; Nelson and Rosenberg 1993; Lundvall 1992). Within these innovation models, universities are often portrayed as important initial producers of research that is then transferred to firms for development (Etzkowitz and Leydesdorff 2000; Etzkowitz et al 2000; Etzkowitz and Leydesdorff 1999).  Non-commercial sectors, such as governments, hospitals, and non-governmental organisations (NGOs), are portrayed as supporting R&D activities of firms and universities but are not generally assumed to be active in R&D processes themselves. Because of the focus on for-profit innovation, the implicit goal of these models is product development, leading to increased economic prosperity. Studies of biomedical innovation that focus on the firm or interactions between the firm and universities in the development of new medical products such as pharmaceuticals and medical devices are examples of this perspective (Windrum and Garcia-Goni 2008; Hopkins 2006; Djellal and Gallouj 2005).  However, the firm-centric view fails to recognize the significant roles of the non-commercial sectors in R&D activities, roles that need to be incorporated into innovation frameworks. This is particularly true for biomedical innovation where academic hospitals play an important role in R&D (Gelijns et al 2001). Linkages between hospitals and universities appear to be particularly crucial in non-commercial R&D, especially in developing novel practices and techniques (Lander and Atkinson-Grosjean 2011). These developments are difficult to measure, which may, in part, explain why most innovation frameworks focus on firm-based product innovation (Hopkins 2006).                                             5 A version of Chapter 3 has been published. Lander, B. (2013). Sectoral collaboration in biomedical research and development. Scientometrics, 94, 343?357.     63  I argue that innovation models which focus predominately on roles of firms and universities create an unrealistic representation of innovation processes by failing to include all sectors active within these R&D networks. Understanding the relative importance of different sectors is key to creating a realistic model of innovation processes because R&D networks dominated by different sectors function differently. For example, for-profit and not-for-profit organisations have different motivations for becoming involved in R&D networks (Gregersen 1992). Thus, models focused predominately on the profit motive inaccurately portray the motivations of non-commercial participants within these R&D networks. In addition, networks dominated by different sectors appear to be driven by different underlying norms of collaboration. Owen-Smith and Powell (2004) found that networks anchored by not-for-profit public research organisations were less proprietary than those dominated by for-profit organisations. This claim reinforces Dosi?s (1982) finding that not-for-profit based technological paradigms are relatively open. Because the relative importance of different sectors impacts the motivations and methods of collaboration, it is important to understand the role of different sectors in an R&D network if an accurate model is to be created.  This chapter compares these general theoretical arguments to Vancouver, Canada?s I2 R&D network. I chose Vancouver?s I2 R&D network as a case study to illustrate how a network made up of organisations which are not a global R&D leader, behaves. Vancouver has a fairly strong presence of university R&D, clinical expertise, and biotechnology but it is not world-renowned in biomedical innovation. The subfield of I2 has a basic research tradition as well as clinical and commercial applications. This chapter strives to create an accurate and systematic analysis of the role that different sectors play within the network by describing the sectors involved, their relative importance, and the extent of their collaboration. In doing so, this chapter not only outlines the characteristics of this specific R&D network but also discusses how insights from this case study can help develop a more robust model of different sector?s roles in R&D networks.      64  I use archival, cross-sectional bibliometric data to map the network through journal co-authorships. Mapping networks through co-authorship of journal publications has become a recognised approach (see for example Sun and Negishi 2010; Morlacchi et al. 2005; Meyer 2002; Murray 2002; Newman 2001; Sandstrom et al 2000). Co-authorship and citations of journal publications are often seen to represent the transfer of knowledge. Murray (2002) for example, emphasizes co-authorship as an informal means of knowledge translation.  By exploring how different sectors may influence the innovation pathways within a R&D network, this study has important implications for science and innovation policies. Such policies are often designed to encourage firm innovation, linkages between industry and universities, international competitiveness, and economic development (Thune 2007; Sharif 2006; Malerba 2004; Etzkowitz et al 2000; OECD 1999; Cooke 1998; OECD 1997). If non-commercial pathways exist within the innovation process, other motivations and objectives need to be recognised and supported within innovation policy frameworks. By mapping the roles of different sectors in R&D networks, this study represents a first step towards creating a more accurate model of innovation which can be of use to both academics and policy makers.  3.2 Innovation theory, organisations, and networks A variety of innovation frameworks underscore the importance that networks and linkages play in research and innovation. These frameworks often disproportionately focus on the firm as innovator. Below I briefly outline two such models: the innovation systems framework and the triple helix model. The IS framework originated in Europe in the 1980s and has been adopted by both academics and policy makers (Sharif 2006). IS generally includes two main components: institutions and actors. Innovation within the system involves the co-evolution and interaction of these components (Malerba 2005). Institutions are the common habits, routines, rules, or laws that influence relations between actors (Edquist and Johnson 1997). Actors can include organisations such as firms, universities, research labs, banks, user groups, and individual entrepreneurs. In reality most IS studies are firm centric: ?there seems to be general agreement that the main    65  components in SIs [IS] are organisations-among which firms are often considered to be the most important ones-and institutions.? (Edquist, 2005, p.189) Other actors are perceived as playing a role within this system by feeding into either the demand or supply side of the firm. On the supply side, non-firm actors supply essential components that firms need to innovate such as capital, research, knowledge, and human capacity (Nelson and Rosenberg 1993). On the demand side, users feed their preferences, insights and needs back into the innovation process (Lundvall 1992).  The triple helix model, another framework used to describe the role of linkages in research and innovation, focuses on complex, iterative, interactions between three helixes of innovation processes: universities, firms, and governments (Etzkowitz et al 2000). The triple helix model privileges the university, rather than the firm, and focuses on what is seen as an increasing blurring of boundaries between universities, firms, and government (Etzkowitz and Leydesdorff 1999). This model proposes that in addition to teaching and research tasks, universities increasingly have a ?third? mission of direct contributions to industry and economic development (Etzkowitz and Leydesdorff 2000). This expands a more traditional view of the university?s third mission of community engagement and service to include commercialisation (Atkinson-Grosjean and Douglas 2010). Thus, while the focus of the triple helix model is universities rather than firms, the innovation ?path? analysed within the triple helix model remains firm-based commercialisation of innovations leading to increased economic development.      But research can follow pathways other than firm based commercialisation on its journey to application. This is particularly true for biomedical R&D where research can be translated to application through direct university-hospital linkages that predominately exclude commercialisation activities (Lander and Atkinson-Grosjean 2011; Hopkins 2006).  This ?clinical pathway? exists in parallel to the commercial pathway and is identified by movement of problems and solutions back and forth between universities and hospitals (Lander and Atkinson-Grosjean 2011; Atkinson-Grosjean and Douglas 2010). Motivations to conduct this type of translation focus on health rather than economic benefits. The idea of a clinical pathway builds on the argument forwarded by Hicks and Katz (1996, p.304) that hospitals act as important    66  sites for the application of university research within what they call the ?biomedical innovation system? in contrast to the more commonly recognised firm-based innovation system. The biomedical innovation system, the clinical pathway of innovation, is dominated by interactions between two non-commercial organisational forms: hospitals and universities.6 A network not dominated by the need for profit may have different norms and standards than those described by the innovation systems and triple helix frameworks.  Based on this insight, the rest of this chapter explores the roles of different sectors and organisational types within one R&D network.  Vancouver?s I2 R&D network provides an illustrative example of the multiple innovation pathways that can occur within biomedicine. Looking at all biomedical R&D collaborations within Vancouver would likely be too large and heterogeneous, potentially glossing over differences between R&D disciplines. Thus, it is important to focus on a specific subfield within biomedicine. Broadly defined, I2 is the study of the body?s defence against infection (Murphy, Travers, and Walport 2008). Because of its strong basic research tradition as well as its clinical and commercial applications, I2 offered a strong case study of the different potential pathways of innovation. Many past analyses of R&D networks focus on cities and regions that are world-renowned in the field studied. While most cities and regions are not global leaders, science policy is often aimed at growing local R&D capacity. Analyses of R&D networks that are not global leaders helps theory and policy expand to include these networks (Gilding, 2008; Rees, 2005; McKelvey et al., 2003). Vancouver has historically relied on natural resource exploitation but more recently is striving to become a leader in the knowledge-based economy (Rees, 2005). Vancouver has a relatively strong presence of university R&D, clinical expertise, and biotechnology. Vancouver ranked third in Canada?s public R&D expenditures by city in 2006 (Holbrook and Clayman 2006). Vancouver has large and relatively specialised clinical care and training, housing British Columbia?s medical fellowship program, BC Cancer Agency, and a majority of British Columbia?s hospital specialist programs. While there is no strong pharmaceutical sector, the city has a relatively well-established biotechnology sector, ranking 7th in North America by some                                             6 Canada?s health and education systems are almost exclusively publicly funded.    67  measures (Salazar et al 2008). Thus, studying Vancouver?s I2 R&D network can help us understand sectoral collaboration trends in biomedical R&D networks. 3.3 Methods My analysis is based on an I2 database filtered and extracted from the ISI Web of Knowledge. Data were cleaned and coded to facilitate analysis. Appendix A outlines the process used in creating the I2 database in more detail. The filter created for data extraction was used to find articles on I2 R&D where at least one author was from Metro Vancouver. This filter consisted of four components: the years used for extraction, a definition of Metro Vancouver, a classification of papers that represent R&D collaborations, and an identification of I2 papers. For the first component of the filter, extraction was based on the years 2004-2009. A six year window was chosen to provide a relatively stable picture of R&D collaborations; six years is long enough to smooth out yearly variation in publication rates by specific authors and organisations while too short to illustrate the evolution of a researcher?s collaborations over time. For the second component of the filter, the Statistics Canada 2006 Census definition of Vancouver?s Census Metropolitan Area (CMA) 7 was used to define Metro Vancouver. For the filter?s third component, R&D papers were classified according to paper types specified by the ISI Web of Knowledge database. Based on discussions with a local I2 expert, three papers types?articles, letters and reviews?were chosen to represent R&D collaborations. To define I2 papers, the fourth component of the filter followed the biomedical subfield procedure recommended by Lewison (1999).  Lewison?s (1999) procedure enables the creation of a biomedical subfield database through a multi-stage process from two interrelated sources: specialist journals and generalist journals based on a keyword list. Per Lewison (1999), I developed keywords based on commonly occurring words in the titles of papers published in I2 specialist journals. Their precision and accuracy (sometimes called specificity and sensitivity) was assessed sequentially by two I2 experts. Precision here represents the proportion of papers retrieved by the                                             7 See http://www.statcan.gc.ca/pub/92f0138m/2003002/4225101-eng.pdf (Accessed September 28, 2010).    68  keyword list that are I2 related over total papers retrieved. Recall represents the proportion of I2 papers retrieved from generalist journals by the keyword list over the total I2 papers in the generalist journals (Montori et al 2005; Lewison 1999). I2 experts assessed the final keyword list as having a 78% precision and a 96% recall rate. Overall, the I2 filter developed for this analysis led to the extraction of 738 papers. The records in the database were subsequently cleaned and coded to facilitate analysis. The definition of organisation in this analysis was fairly broad; for example, hospital and university centres and departments were subsumed within the larger hospital and university. Organisations physically located in different cities and/or countries that were part of the same organisation, for example local offices of large multi-national companies, were treated as separate organisations.  Organisations were also coded by sector: hospital, university, firm, government, and NGO. To increase reliability, coding was conducted independently by the primary researcher and a second individual based on guidelines developed by the primary researcher. Sectoral affiliation was first determined by the organisation?s name and, if affiliation was unclear from the name, organisations were researched online. As part of these guidelines academic teaching hospitals were coded as hospitals. Large, multi-hospital administration systems were coded as government or firm.  This was only really an issue in the USA. There, New Haven Healthcare System and Sunnybrook Health were coded as firms while Veteran Affairs administrative units were coded as government. Public research institutes with highly limited teaching capacity were coded as government. Organisations with non-profit status were coded as NGOs. Consortiums of organisations, generally non-profit in nature, were coded as NGOs. The Victorian Partnership for Advanced Computing, Australia, serves as an example of this type of consortium. The results of the two independent coding processes showed almost perfect intercoder reliability with no significant difference in the coding (?=.809, p=.000)8. In total both coders agreed on 983 of the 1161 organisations coded. For the 178                                             8 See McGinn et al (2004) for guidelines on how to interpret the ? statistic    69  organisations where sectoral coding differed, the two coders reached a coding agreement after discussion.  Research collaboration between organisations is the key variable used in this paper. Because of the way ISI Web of Knowledge structures its records, this is measured by the organisations given as addresses by a paper?s authors.  Two organisations are described as co-authors if they are both listed in the address field of a paper. Individuals listing more than one organisational affiliation on a paper thus represent collaboration between these organisations. These individuals are seen here as ?boundary spanners? who facilitate the translation of knowledge between their organisational affiliations (Lander and Atkinson-Grosjean, 2011; Swan et al., 2007). Listing more than one organisational affiliation is likely fairly prevalent in these data; Katz and Martin (1997) found that in Canada 10-14% of papers listed more organisations than authors. Sector is a second variable employed in the analysis. The initial dataset used is a two-mode matrix that connected organisations to the papers they co-authored. This matrix was converted into a one mode matrix through cross product analysis.9 The resulting one mode matrix connects organisations to organisations through co-authorship. UCINET 6.32 (Borgatti, Everett, and Freeman, 2002) and Netdraw (Borgatti 2002) were used to analyse these data. 3.4 Sectors? involvement in Vancouver?s I2 network Between 2004 and 2009 Vancouver?s I2 R&D network published 738 papers involving co-authors representing 925 organisations. These organisations were located in 413 cities, 56 countries, and 6 continents. The average number of papers connected to each organisation through co-authorship was 3.11 while the average number of organisations listed for the co-authors on each paper was 3.9.  Figure 5 presents all organisations that were involved in Vancouver?s I2 R&D network between 2004 and 2009. The shape of the organisation?s nodes represents their sector: hospitals= circles, universities= squares, government=boxes, firms=up                                             9 See Borgatti and Everett (1997) and Wasserman and Faust (1994) for further discussions of two-mode matrices and their conversion to one-mode matrices    70  triangles, and NGOs= diamonds. The size of each node represents the number of times researchers from one organisation co-authored an article with researchers from another organisation within the dataset. There are several interesting features of Figure 5. The ties that exist form a fairly well connected co-authorship network. There are relatively few isolated organisations, only one isolated dyad and one isolated triad. Instead, most organisations appear to be relatively well connected to other organisations within the network, with a core group facilitating much of this connection. This core group of organisations is further explored in Figure 6, which shows only the organisations within the top quartile; organisations with 23 or more co-authorships with another organisation. Table 7 further shows the relative importance of different sectors in this network by outlining the number of organisations, number of co-authors, and normalised two-mode degree calculations for each sector. Normalised two-mode degree calculations aid in understanding the relative importance of each sector. Degree calculations represent the number of connections each node has; in the two-mode case this represents the number of connections of each node of one type (here sector) to the nodes of another type (here papers). Normalisation divides the number of observed connections by the total number of possible connections. 10 In Table 7, the normalised two-mode degrees represent the number of papers each sector was involved in as a proportion of the total papers in the network.                                                10 See Borgatti and Everett (1997) p.254 for a further explanation of the calculation and interpretation of two-mode degree statistics.    71  Figure 5: Vancouver's global I2 by organisation  Universities visually dominate this network and many play a key role in connecting different organisations together. This dominance of universities, shown visually in Figures 5-6, is further reinforced in Table 7, which demonstrates that 344 universities are part of the network, making up 37.2% of the total organisations involved. Based on two-mode degree calculations, the university sector was involved in 87.9% of total publications within this dataset. The relative importance of universities in the network is further underscored by the proportion of total universities that are part of the network?s core; Figure 6 shows that 32% of the universities in this network are located at its core. Key universities in this network include the University of British Columbia, the University of Toronto, the University of Calgary, and the University of California in San Francisco.  Perhaps more interesting is the relatively large presence of both hospitals and governments in this network, in contrast to the small presence of firms. 210 hospitals were involved. The hospital sector was involved in 47% of publications. A large proportion of these hospitals (30.5%) belong to the core shown in Figure 6. Key hospitals include St. Paul?s Hospital (Vancouver); Women?s and Children?s Hospital    72  (Vancouver); Vancouver General Hospital, and Sunnybrook Health Sciences Centre (Toronto).  Figure 6: The core of Vancouver's global I2 network  (top quartile)  Hospital representation is followed by that of government organisations. Some 166 government organisations were involved, with at least one government organisation co-authoring 38.8% of publications. Government organisations make up a surprisingly small proportion of organisations that are part of the core shown in Figure 6, only 12.7% (21) of total government organisations are part of this figure. Government organisations that are most active researchers in the network are the BC Centre for Disease Control, BC Cancer Agency, USA National Institutes of Health and USA Centre for Disease Control.  While firms are the traditional focus of the biomedical innovation literature, they have a relatively weak presence here. Only 125 firms are part of this network, with 22 firms at the core group of organisations shown in Figure 6. At least 1 firm was involved in 15.4% of all publications. Only one of the most active firms in the network, Viridae    73  Clinical Science Inc., was located in Vancouver. The others, 3M Centre, Dynavax, and the New Haven Healthcare System, were located within the USA.  NGOs make up the smallest group within this network with a total representation of 80. Only 14 of these organisations are part of the core group of organisations. At least one NGO was involved in 12.5% of all publications within this network. None of the most active NGOs were from Vancouver. Key NGOs include the Fred Hutchinson Cancer Research Center (Seattle), the Netherlands HIV Foundation, and the International AIDS Society (Geneva). Table 7: Sectoral activity Sector #Orgs Normalised 2M Degree University 344 0.879 Hospital 210 0.470 Government 166 0.388 Firm 125 0.154 NGO 80 0.125  3.5 Co-authorship within and between sectors Observed rates of co-authorships within and between sectors are shown in Table 8. All numbers in Table 8 are expressed as column percentages. This allows the reader to interpret co-authorship rates based on each sector?s relative size. Thus, each row in the first column represents the co-authorships between hospitals and the sector listed in that row divided by total hospital co-authorships and taken as a percentage; the second column is co-authorships between universities and all other sectors divided by total university co-authorships taken as a percentage and so on. This leads to an asymmetric table that underscores the relative importance of each sectoral co-authorship relationship when compared to the sector listed in the column as opposed to rates in absolute numbers.    74  Table 8: Observed rates of co-authorship within and between sectors weighted by sector Hospital University Firm Government NGO Hospital 25.13% 28.51% 21.49% 17.76% 20.63% University 50.17% 34.00% 52.64% 52.07% 51.35% Firm 7.25% 10.09% 9.25% 4.91% 8.13% Government 10.81% 18.00% 8.86% 16.88% 15.83% NGO 6.64% 9.40% 7.76% 8.38% 4.06% Observed frequencies are expressed as column percentages.     The ?2 statistic is often used to find whether observed proportions differ from expected proportions. Two sectors with large co-authorship numbers are expected to co-author a large number of papers together while a sector with large co-authorship numbers and a sector with small co-authorship numbers are expected to co-author a smaller number of papers together. If the observed and expected rates of co-authorship differ significantly, organisations are believed to show a preference in which sectors they co-author with. Network data requires special statistical considerations because cases are not independent. This violates an assumption that underlies the ?2 statistic and increases the probability of a type-1 error, meaning that there is an increased likelihood of rejecting the null hypothesis even when it is true. Relational contingency table analysis has been developed as a network alternative to ?2. Relational contingency table analysis similarly creates a table that gives a count of the number of ties within and between groups and then compares that to a set of expected values. Unlike a traditional ?2 approach, relational contingency table analysis uses permutation trials to calculate expected values, assuming that the ties are independent and randomly distributed throughout the groups, and the significance levels of expected compared with observed values. This enables the calculation of robust statistics for interdependent case network data (Hanneman, 2005; Borgatti, Everett, and Freeman, 2002; Cliff and Ord, 1973). Based on relational contingency table analysis, Table 9 shows the odds ratio of observed over expected rates of sectoral co-authorships.  A related network approach to relational contingency table analysis is an ANOVA structural blockmodel density test (Borgatti, Everett, and Freeman, 2002). This model    75  tests whether the average density of ties in different matrix blocks are significantly different from each other.11 This ANOVA test also uses permutation methods to calculate significance (Snijders and Borgatti, 1999). Results from this ANOVA test are shown in Table 10. To aid in comparison, Tables 8 to 10 have been shaded in a similar gradient: larger numbers are darker. Table 9: Observed over expected rates of co-authorships within and between sectors Hospital University Firm Government NGO Hospital 1.95 1.18 0.47 0.53 0.67 University 1.18 1.73 0.70 0.94 1.02 Firm 0.47 0.70 0.69 0.24 0.45 Government 0.53 0.94 0.24 1.28 0.65 NGO 0.67 1.02 0.45 0.65 0.70 ?2(4)=1624.114, p<.001, number of permutations: 1000  Table 10: Mean tie densities of co-authorships within and between sectors Hospital University Firm Government NGO Hospital 0.0398 0.0286 0.0092 0.0108 0.0128 University 0.0286 0.0454 0.0136 0.0215 0.0218 Firm 0.0092 0.0136 0.0125 0.0045 0.0084 Government 0.0108 0.0215 0.0045 0.0259 0.0127 NGO 0.0128 0.0218 0.0084 0.0127 0.0133 Structural blockmodel: R2=.002, p<.001, number of permutations:5000  Based on Table 8, universities are the most important co-author for all sectors.  Hospitals act as the second most important co-author for all sectors. Firms are relatively less important, acting as fourth, or even fifth, most important co-author for all sectors except firms. Tables 9 and 10 explore whether sectors show a sectoral preference in co-authorship and test if this preference is significant. Co-authorship trends in both Tables 9 and 10 are broadly similar. From Table 9 we see that preferential co-authorship appears to exist (?2 (4)=1624.114, p<.001) and from Table 10 we see that                                             11 For examples of other analyses using this approach see Conti and Dorein, 2010 and Cross, 2001    76  while this ANOVA model explains only 2% of variance of mean tie density, it is significant. Intra-sectoral co-authorship is higher than expected for hospitals, universities and government. This phenomenon is most pronounced in the hospital sector where hospitals are almost twice as likely to co-author papers with other hospitals than expected. Preferential intra-sectoral co-authorship is second highest among universities where universities are almost one and three-quarters as likely to co-author papers with other universities than expected. For firms and NGOs, preferences for intra-sectoral co-authorship are second to their inter-sectoral co-authorship preferences with universities.  Overall, sectors appear to show a relatively high preference for collaborating with universities. This trend is most pronounced in hospital-university co-authorships; hospitals and universities were almost twenty percent more likely to co-author papers together than expected. This was followed by NGOs who were two percent more likely to co-author papers with universities than expected. 3.6 Discussion Perhaps the most interesting finding of this study is that universities and hospitals co-author much more than expected statistically. Not only do universities and hospitals collaborate more than expected, but university-hospital co-authorships make up the second largest number of sectoral co-authorships in absolute numbers after university-university co-authorships. Thus, university-hospital collaborations appear to be a key component within this biomedical R&D network. The importance of university-hospital collaborations shown here echoes a similar bibliometric study performed by Hicks and Katz (1996) who found that hospitals and universities co-authored more than expected across the UK science base. The results of these two studies cannot be directly compared, however, as the Hicks and Katz (1996) study is based on a different analytic strategy.  Hicks and Katz (1996) used regression across 9 sectors to calculate differences between observed and expected rates of sectoral collaboration in all scientific fields in the UK. In addition to the five sectors used here, Hicks and Katz (1996) included Research Council, Special Health Authority and British Postgraduate Medical Research Groups, Polytechnics and Other as additional sectors. While their coding schema differed from the one used here, the Hicks and Katz (1996) definitions of    77  hospitals and universities appear broadly similar. This finding also seems to reinforce qualitative studies that have underscored the importance of university-hospital linkages in biomedical innovation (Lander and Atkinson-Grosjean 2011; Hopkins 2006; Gelijns et al 2001).  The high presence of government organisations in the network was initially surprising but makes sense given the nature of the field. Governments are key agencies in infection monitoring and control through organisations such as national centres for disease control. They play active roles in funding biomedical research, and fund medical care in much of the Western world. Also, different sectors may use publications to facilitate different objectives. For example, government organisations may focus more on publishing best practice protocols, hospital on identifying new standards of care, industry on clinical trial results, and universities on novel research. The relatively low levels of intra-sectoral co-authorship observed among NGOs are unsurprising given that many associations and consortiums are developed precisely to facilitate collaboration. Others likely lack to resources to conduct extensive research independently. Because of the emphasis on the firm and university-industry collaborations in the innovation literature (Thune 2007; Edquist 2005; Owen-Smith and Powell 2004; Etzkowitz et al 2000), it is interesting to note that co-authorships between firms and universities do appear to be relatively important, particularly for firms. However, firms played a relatively weak role in this network; only NGOs had a weaker presence than firms. Beyond absolute numbers, firms had a relatively weak presence in the network?s core and in overall co-authorship rates.  In addition, firms located outside of Vancouver were more active in the sector than local firms. Only one of the top four firms in the network was located in Vancouver. These findings may, in part, be biased through the use of research papers as a data source; firms are generally more strongly represented in patent data (Meyer 2002). Analyses based on patent data would conversely be firm biased. While this may be the general case, in fields such as biomedicine, firms publish more in academic journals than firms in other fields. Godin (1996), for example, found that firms involved in drug development follow this pattern. This trend towards industry publication in biomedicine    78  has likely increased as firms face increasing pressure to disseminate their clinical trial results. Bourgeois et al (2010) found that 67% of industry-funded drug trials were published in academic journals. In addition to firms, hospitals, government organisations, and NGOs may display a similar bias against publication. Thus, while these results may under-report firm activity relative to universities, they may also under-report hospital, government, and NGO activity relative to universities.  One interpretation of the relatively low presence of firms within this network is that Vancouver?s I2 R&D network is a deficient innovation system because firms are relatively weak in this network. The relatively small presence of the private sector within Canadian innovation systems has been noted by others (McFetridge 1993). Following from this interpretation, policymakers in Canada will need to encourage more university-industry linkages in an effort to correct this weakness. I argue for a different interpretation of this result. I suggest these results indicate that alternative innovation pathways, dominated by the public rather than private sectors, exist and flourish. In Canada, as in much of the Western world, the majority of medical treatment and care is provided by the public sector. It is often this sector, its doctors, nurses, hospitals, and patients, that not only produces biomedical innovations but is also the final user of these innovations (Windrum and Garcia-Goni 2008; Gelijns et al 2001). A successful IS, and related public policy, would focus on connecting R&D producers, innovators, and users, wherever they are found, whether in organisations located within the private or the public sector.       Using the ISI Web of Knowledge as a data source has admitted shortcomings. Not all co-authored articles from Vancouver will be found because the ISI Web of Knowledge does not include all journals. The ISI Web of Knowledge is a comprehensive and quality database. Composed of nearly 12,000 journals, the ISI Web of Knowledge includes journals based on publishing standards and citation data. In addition, co-authorships may not accurately represent collaborations; individuals can be designated as co-authors who did not significantly contribute to a document, for example as a form of patronage. This has become less of a problem recently as journals clarify co-authorship rules. Other researchers may informally collaborate but not be named as co-   79  author an article. Thus results of this mapping cannot be considered a complete map of Vancouver?s I2 R&D network but rather a rough approximation that can be used to indicate the direction for future explorations. It is important to note the relatively low densities calculated in Table 10 and the poor predictive capability of this ANOVA model. Low densities here simply reflect the large size of this network and the way data were sampled. Because this analysis is not based on a whole network but rather a Vancouver centric network, it is impossible for organisations outside of Vancouver to achieve densities approaching one. This likely lowered the predictive ability of this ANOVA model. Here, the ANOVA model is used for triangulation purposes, showing similar density trends to the relational contingency table odds ratios. The ANOVA model is not explicitly used for predictive purposes, thus its low predictive ability is not of great issue in this particular analysis. 3.7 Conclusion This chapter analysed worldwide co-authorships between hospitals, universities, firms, governments, and NGOs for Vancouver?s I2 R&D network. It mapped these co-authorships and compared observed and expected co-authorship rates. Results show that universities are top players in this network, followed by hospitals, government organisations, firms, and then NGOs. Most sectors collaborated intra-sectorally more than expected, given their relative size, and less than expected with other sectors. Echoing the emphasis traditionally placed on university-firm collaboration in the innovation systems framework and the triple helix model, university-firm collaborations appeared to be important, particularly for firms. Overall, however, the private sector played a rather limited role within this network. Hospital-university co-authorship showed the highest rate of co-authorship across sectors, a level that was statistically above expectation. Networks are key vehicles for organisations to gain access to new information and co-authorship can be seen as a good proxy for these knowledge-sharing networks. This study begins to explore the role that different sectors play within one biomedical R&D network. The results of this study show that current models of innovation that privilege the firm cannot appropriately model Vancouver?s I2 R&D network. If results of    80  this case study hold for other R&D networks, this has important implications, not only for theoretical R&D models, but for science and innovation policy tools. This is because previous research has shown that dominant sectors influence the norms and knowledge sharing practices of the entire network (Owen-Smith and Powell 2004). Thus, policy makers need to be cognizant of the true character of the networks that their policy tools affect and ensure that tools are tailored towards the realities of these networks. The results of this case study need to be tested in other biomedical and non-biomedical R&D networks in order to move from Vancouver?s I2 R&D network to new theory and policy. This can be done by looking at other subfields and geographic areas. While general sectoral trends are shown through this analysis, moving from an analysis of the larger network trends to an exploration of specific collaborations could also improve further understanding of the motivations behind specific collaborations. It would also better show how sectoral norms and rules may affect collaborations and the network as a whole. This approach would help build insight into why certain sectors collaborate more or less than expected and could build on previous qualitative studies of the micro-dynamics of R&D relationship formation and development (Thune 2007). Such an analysis, conducted in tandem with the sort of bibliometric analysis described here, would also improve insights into whether co-authorship analyses are an accurate illustration of collaborations between non-university sectors.    81   4 Chapter 4: Collaboration and institutions in Vancouver?s local I2 organisational field 4.1 Introduction Within this chapter and the next I combine my quantitative and qualitative analyses to explore how institutions affect individuals? actions and collaborations within an organisational field. Previous chapters focused on the broad trends of Vancouver?s I2 global collaborations. Within this chapter, I move from this general description to a more in-depth and localised study, and, in doing so, from the macro to the meso to the micro levels of analysis. I take individuals embedded within Vancouver?s I2 organisational field as my unit of analysis and focus on how they perceive that their interactions are affected by institutions. This chapter sets the stage for an analysis that continues into Chapter 5. It does so by outlining key theoretical concepts and methods, as well as organisations, institutions, and structures in Vancouver?s I2 research organisational field. This chapter draws on institutional literature?particularly neo-institutional theory from organisational theory? in its study of collaborations. Institutional literature explores how individuals act and interact within an institutional environment that shapes everything from their goals and motivations to demands on their time to their ability to work with other individuals (Scott, 2008; Barley and Tolbert, 1997; Jepperson, 1991; Giddens, 1979). Neo-institutional literature often focuses on organisations? interactions within an organisational field. Organisational fields include different organisations, such as suppliers, producers, consumers, and regulatory agencies, that interact within a general epistemic and geographic area and which are affected by specific institutions and an overarching governance system (Emirbayer and Johnson, 2008; Owen-Smith and Powell, 2008; Scott, 2008; Wooten and Hoffman, 2008; Owen-Smith and Powell, 2004; DiMaggio and Powell, 1983).  I focused on the micro-foundations of organisational fields and institutions. This adds to the existing neo-institutional literature that has predominately used organisations, as opposed to individuals, as a unit of analysis (Powell and Colyvas,    82  2008; Scott, 2008). Individuals affiliated with different organisations make connections within an organisational field. By focusing on these individuals, I am better able to understand how they perceive that sectoral and organisational institutions influence their actions and interactions. Interactions between individuals affiliated with different organisations within an organisational field form an IS that creates new products and processes (Edquist, 2005; Malerba, 2005; Cooke et al., 1997; Lundvall, 1992). A strong innovation system leads to increased economic and social prosperity, and in a biomedical field such as I2, to improved human health. Most studies of biomedical innovations have focused on translations between universities and firms (Hopkins, 2006; Gelijns and Rosenberg, 1994; Blume, 1985). Here I expand analysis beyond this traditional firm-university focus by including interactions with government, hospitals and NGOs as well as universities and firms.  In this and the next chapter I sampled I2 as an organisational field to study and focused on a subset of this field, individuals involved in research and development. Other organisations and institutions are included insofar as they affect these individuals. Owen-Smith and Powell (2008) argue that organisational fields are nothing more than social networks that transfer information between organisations. Organisational fields are thus structured and integrated by social networks (DiMaggio and Powell, 1983). Building on this insight, this chapter will draw from a social network perspective to map Vancouver?s I2 organisational field, and, within this context, individual interactions within this field will explored. In answering my research questions I begin with an overview of the relevant theory related to institutions and organisational fields. I then describe the methods used for this study followed by an outline of the main organisations and institutions within Vancouver?s I2 organisational field. This is followed by a discussion of how individuals perceive that organisational and sectoral institutions and capital influence their actions. I argue that participants don?t believe that organisational affiliations affect collaboration decisions. However, participants collaborate to gain access to capital, capital that is often only available in certain organisations. While organisationally specific regulative institutions only appear to superficially influence collaboration decisions, normative and cultural-cognitive institutions influence participants? goals and thereby their reasons for collaborating. Less obvious and more    83  entrenched, normative and cultural-cognitive institutions appear to influence collaboration decisions more than regulative institutions. 4.2 Institutions, organisational fields, and collaboration 4.2.1 Institutions Drawing from the study of institutions within political science, sociology, and economics, Scott (2008, p.48) defines institutions as ?regulative, normative and cultural-cognitive elements that, together with associated activities and resources, provide stability and meaning to social life.? Institutions are legitimated prescriptions and scripts that describe behaviour while being so entrenched in society that they become taken for granted. Institutions form a basis of what is considered legitimate and adhering to institutions increases legitimacy. Legitimacy is perceived as particularly important for individuals and organisations that were not motivated by the market. Institutions form the bedrock of social life but change over time. They can both support and inhibit action and are produced and reproduced through specific activities and social interactions. Within a specific situation, institutions inhibit action by setting bounds on rationality and restricting perceived opportunities and alternatives, thereby increasing the probability of certain behaviour (Deephouse and Suchman, 2008; Greenwood et al, 2008; Scott, 2008; Wooten and Hoffman, 2008; Barley and Tolbert, 1997; Jepperson, 1991; DiMaggio and Powell, 1983; Giddens, 1979; Meyer and Rowan, 1977).  As outlined in Table 1, Scott (2008) proposes three pillars of institutions? regulative, normative, and cultural-cognitive?and argues that moving from the regulative to the cultural-cognitive pillars means moving from conscious to unconscious and from legally enforced to more implicit taken-for-granted institutions. Specific institutions draw from one or more pillar; drawing from additional pillars increases an institution?s legitimacy. Conversely, different pillars can support conflicting institutions, leading to divergent conceptions of what is legitimate behaviour. This allows individuals to selectively choose from institutions, enabling different decisions to be taken in reaction to the same situation (Scott, 2008; Haunschild and Chandler, 2008; Thornton and Ocasio, 2008; Friedland and Alford, 1991; Giddens, 1984). Often decoupling can    84  occur within an organisation whereby certain components, i.e. departments or individuals, become more influenced by some institutions while other components will become more influenced by others. Resolving a potential tension between conflicting institutions, decoupling can cause divergent behaviour within a single organisation (Meyer and Rowan, 1977). Scott?s regulatory pillar is focused on explicit regulatory processes based on established rules. An authority exists to assess conformity to these institutions and implement sanctions if institutions are not followed. Because this pillar is explicit and based on authority, analyses often focus attention to the role of the state as a rule maker and enforcer (Equist and Johnson, 1997; Hodgson, 1994). Scott?s normative pillar is based on the values, defined as conceptions of what is preferred, and norms, defined as ideas of how things should be done, within a particular social group. Normative systems define goals and objectives as well as the appropriate ways of achieving them, and, in doing so, can give rise to specific social roles. Conformity to normative institutions is largely seen as the result of self-evaluation; individuals feel good when they comply with normative institutions and bad when they do not. Scott?s cultural-cognitive pillar is based on how culture is cognitively internalized into shared concepts of social reality and meaning. Focus here is given to meanings attached to symbols such as words, signs and gestures; and how these meanings can become perceived as objective and external to a particular social group (Meyer and Rowan, 1977). Institutions supported by the cultural-cognitive pillar are deeply embedded and taken-for-granted.  While organisations and institutions are often used interchangeably in general parlance, for the purposes of this chapter the two are conceptually distinct. Organisations can have their own institutions that are shared by individuals affiliated with the organisation and influence these individuals? action (Berends et al., 2003; Barley and Tolbert, 1997; Williamson, 1994). Other institutions are specific to an organisational department, a sector, profession, social group, or society (Scott, 2008). For example, individuals in hospitals, universities, and firms may have institutions that are only shared within their department, others unique to their organisation as a whole,    85  some shared with other organisations within the same sector, and others shared with individuals within the same city or country irrespective of their own organisational affiliation. Organisational institutions are important when individuals affiliated with different organisations collaborate. Like all institutions, organisational institutions affect how individuals make decisions, act, and limit their rationality (March and Simon, 1958). Based on organisation specific institutions, individuals belonging to different organisations may experience different concepts of what constitutes rational decision-making and appropriate action. This divergence may influence the ability of individuals to collaborate across organisations.  4.2.2 Organisational fields Neo-institutional theory focuses on the impact of institutions on organisations within an organisational field. The concept of an organisational field was first proposed by DiMaggio and Powell (1983, p.148) to denote: ?key suppliers, resource and product consumers, regulatory agencies, and others that produce similar services or products.? An organisational field includes organisations with the same and different purposes, which interact related to a particular techo-epistemic context and are affected by an overarching governance system (Scott, 2008). The boundaries of organisational fields are often defined through affiliation, competition, and shared memberships (Owen-Smith and Powell, 2008). Organisational fields are conceptually similar to industrial sectors, which are defined as all organisations that play a role within a particular area of production (Emirbayer and Johnson, 2008), to Bourdieu?s concept of the field (Emirbayer and Johnson, 2008), and to innovation systems, which are defined as organisations and individuals interacting within an institutional environment (see for example Edquist, 2005; Malerba, 2005; Cooke et al., 1997; Lundvall, 1992).  Connectedness between organisations within a field is a key concept (DiMaggio and Powell, 1983). For example, Owen-Smith and Powell (2004) outline the organisational field of Boston biotechnology to include biotechnology firms, venture capital companies, government agencies, and public research organisations and explore how these organisations interact. Institutions guide behaviour within organisational fields, these institutions have, in turn, been shaped by the organisations    86  within the field in an iterative process (Wooten and Hoffman, 2008). The relative importance of different institutions, organisations and sectors within an organisational field affect the field?s overall tone and governance system (Emirbayer and Johnson, 2008; Scott, 2008; Owen-Smith and Powell, 2004). The concept of governance system is similar to Bourdieu?s concept of a field?s habitus, which is the culmination of the past experiences of individuals embedded in the field?s ?cultural unconscious? (Emirbayer and Johnson, 2008; Martin, 2003). Boundary spanning units can help to bring together different types of organisations within an organisational field (Colyvas and Powell, 2006) while boundary spanning individuals can often enable knowledge translation and collaboration between one epistemic culture and another (Swan et al, 2007; Ben-David, 1960).  A shortcoming of many institutional theories is that organisations are often seen as the ?actors? in analysis (Powell and Colyvas, 2008; Scott, 2008). Less work has gone into understanding the microfoundations of neo-institutionalism (Powell and DiMaggio, 1991). Organisations are treated as unitary ?black boxes? with limited analysis of the role of individuals within them. However, it is individuals who are influenced by institutions and their agency that affects ?organisational? behaviour (Berends et al., 2003). It is ultimately the individual who creates connections and translates knowledge between organisations. Institutions are often viewed as enacted through the interactions between individuals because the institutions that an individual has internalised only become apparent through their actions (Scott, 2008; Barley and Tolbert, 1997; Bourdieu, 1985; Giddens, 1984). Collaboration between individuals is a prime exemplar for understanding how institutions affect action, particularly if the collaboration involves individuals affiliated with different organisations or sectors that are influenced by differing institutions. Because institutions constrain individual?s action, they enable and constrain collaboration between individuals. This focus here is on the individual and how institutions from the sub organisational to organisational field level affect individual action.     87  4.3 Methods I used Vancouver?s I2 R&D organisational field as a case study in the following analysis. Because much of culture and law can be considered institutions, Jepperson (1991) argues that what is identified as an institution in a study is often defined by the study?s focus. My research focused on I2 R&D collaboration and thus concentrated on institutions across all three pillars that affected I2 R&D. Particular attention was given to how individuals perceived that organisational and sectoral institutions influenced their actions. As described in more detail below, interview participants often identified these institutions. I used mixed methods for both developmental and complementary purposes (Greene et al., 1989). In developmental mixed methods one method is used to help the development of the second method. In complementary methods both methods are used to measure overlapping but different facets of the same phenomenon. As described in greater detail in Section 4.3.2.1 below, my approach was developmental because I primarily sampled participants to interview from my quantitative database. I also began most interviews by showing participants sociograms based on my quantitative database. My approach was complementary because my quantitative analysis focused on mapping the individuals, organisations, and sectors involved in Vancouver?s I2 network; assessing their relative importance; and exploring their interactions based on I2 co-authorships. This created a map of the organisational field under study. Within this map, my qualitative analysis focused on describing organisations and individuals involved in this case and key relevant regulatory institutions and then explored my second and third research questions:   2 What reasons do people give for collaborating? 3 How do institutions such as policies, norms, and organisational culture affect collaboration? Combined, my quantitative and qualitative analyses thus provided a richer description and understanding of the case under study.     88  4.3.1 Quantitative approach I drew from a social network perspective to create an outline of Vancouver?s local I2 organisational field structure. Owen-Smith and Powell (2008) argue that social networks create ties that transfer information between organisations. Much of the research in social network theory explores how the structure of social networks influences the way that knowledge and norms are translated. Social network analysis thus provides methodological tools for analysing individuals?, organisations?, and sectors? interactions within a larger system. Details of the construction, coding, and cleaning of the primary database used in this chapter can be found in Appendix A.   I used three subsets of this primary database in analysis. With the first subset I created a sampling framework to select participants to interview. This subset included all authors that could be connected to a Vancouver organisation between 2007 and 2009. This sampling dataset was made up of 514 authors with many authors having more than one organisational affiliation. Because interview sampling occurred beginning in February 2011, years 2010 and 2011 had not yet been added to the primary I2 database and could not be part of the sampling framework. I describe interview sampling in more detail in Section 4.3.2.1 below. I employed a second subset of my I2 database to create a sociogram of the co-authors of each of participants interviewed. As described in more detail in Section 4.3.2.2 below, these sociograms were discussed during the majority of my interviews. This subset was the same subset as used in Chapter 3 and focused on publications between 2004 and 2009.  I created a third subset of my I2 database to draft the tables and sociogram found in Section 4.4 below. This was based on the subset that I used in Chapter 2 and included 2008-2011 where authors can be connected to organisations. I used this subset because, as I argued in Chapter 2, authors could not be connected to organisations reliably in Web of Science data before 2008 and author based co-authorship analysis is more accurate than organisation based co-authorship analysis. Within this subset there were 811 Vancouver-based authors affiliated with 34 distinct Vancouver-based organisations. As outlined in Chapter 2, a large number of these    89  authors (32.8%) had more than one affiliation. Out of the participants that I interviewed, described in Section 4.3.2 below, 52.6%(20) had multiple affiliations. All participants with multiple affiliations were affiliated with a university (UBC or SFU), all but one were not physically located at either university. They were located instead at either a government or hospital organisation that they were affiliated with. To more accurately reflect the physical location of co-authors in my analysis, I created a new adjusted dataset that took out the university affiliations of individuals who listed multiple affiliations on a single publication. Controlling for multiple affiliations decreased the number of authors affiliated with universities by 21% from 551 to 435. This adjusted dataset is used in Section 4.4. In Section 4.4 I used block modelling techniques, described in more detail in Chapter 2 and 3 above, to collapse authors into their respective organisations. 4.3.2 Qualitative approach 4.3.2.1 Interview sampling  Methodologists do not generally propose an ideal number of interviews for a case study (Crewell, 2007; Stake, 2005; Yin, 2003). I used sufficiency and saturation criterion to decide whether I had interviewed enough people (Seidman, 1998). Employing sufficiency criteria, I interviewed participants and sites until I believed that the experiences of the participants would resonate with others who were part of the I2 network but not sampled to be interviewed. Based on the saturation criteria, I continued to interview until I began to hear the same information repeated by participants. As shown in Figure 7, three rounds of interview recruitment were conducted. In Rounds 1 and 2, physical letters were sent to all individuals introducing them to my study, describing my interest in their participation, and outlining their potential role. I then followed up with individuals either via email or phone to ascertain their interest. These individuals were generally quite busy and I contacted them up to three times trying to attain a response. Individuals that stated that they were not interested were not contacted again. Sampling and interviews for Rounds 1 and 2 occurred between February and May 2011. One Round 3 interview occurred in May 2011 with the other five occurring between June and October 2012.    90   Figure 7: Interview sampling  During Round 1 of recruitment, I employed a stratified purposeful sampling technique (Creswell, 2007; Stake, 2005) whereby I tried to roughly reproduce the sectoral distribution in the sampling dataset by randomly sampling approximately twice the number of ideal interviews from each sector based on a random number generator. Table 11 shows the sectoral distribution of the sampling dataset, my initial ideal distribution of participants by sector, and the final distribution of participants by home sector. In my ideal distribution of participants, the private sector has more than its proportional share of participants based on the belief that five was likely the minimum to satisfy the sufficiency and saturation sampling criterion. Hospitals also received a larger proportion of ideal participants for theoretical reasons: I believed that the role of universities in biomedical innovation is already well understood and one of my key interests in this study was further understanding the role that hospitals play in biomedical innovation. For each individual sampled, I researched contact information through publications and organisational websites. Individuals no longer located in Vancouver or where I was unable to find sufficient contact information were excluded. In addition, firms posed a sampling challenge in part because the majority (8 out of 15) of    91  the individuals within the sampling dataset belonged to one firm. Based on my sufficiency sampling criteria, I decided to elicit perspectives from a variety of firms and discounted all but one individual from the main firm, expanding my firm sample to include all individuals affiliated with other firms.  Table 11: Sectoral affiliations of sampling dataset, ideal participants, and participants  Sampling dataset   Sector # Authors % Ideal participants Participants University 321 50% 10 6 Hospital 145 22% 10 10 Government 158 24% 5 14 Private Sector 23 4% 5 6 NGO 0 0% 0 2 Total 647 100% 30 38 Note: In sampling dataset, authors with multiple organisational affiliations are counted once for each affiliation.   I contacted 35 individuals during Round 1 of recruitment, leading to 18 interviews. As interviews began, I realised that only 6 of the 18 individuals I was interviewing had their physical ?home? at the organisation from which I had sampled them. Many of the differences were due to multiple affiliations while others were due to individuals changing jobs since publishing the articles in my dataset. For example, two individuals affiliated with the private sector in my sampling dataset now worked at NGOs. To try and remedy this, during Round 2 of recruitment I researched the ?home? organisation of individuals within the dataset and discounted individuals whose ?home? organisations had been disproportionately sampled during the first round of recruitment. During Round 2, 22 individuals were contacted leading to 14 interviews. Drawing from grounded theory (Charmaz, 2005), during Round 3 I purposefully recruited 6 individuals to further explore themes developed during an initial analysis and to expand my understanding of Vancouver?s I2 organisational field. Five of these individuals were found through snowball sampling techniques while the final participant was found through opportunistic sampling (Creswell, 2007). Three of these were administrators whom I interviewed to    92  learn more about regulative institutions related to organisations within the field. The three other participants belonged to firms. Recruitment of individuals affiliated with firms from my sampling dataset had yielded limited results and I felt it necessary to expand my sampling in the private sector to meet sufficiency and saturation sampling criterion. Overall the success rate for my recruitment was slightly over 60%. The far right column in Table 11 shows a final breakdown of individuals interviewed by their home sector. Home organisational affiliations are not given to try and respect participants? anonymity. Brief descriptions of participants interviewed can be found in Appendix B. All participant names used in this dissertation are pseudonyms. 4.3.2.2 Interviews It is important to acknowledge that power dynamics between the interviewer and participant can play a role within the interview setting (Karnieli-Miller et al, 2009). The majority of my interviews were with individuals in research, service, and research/service roles. Individuals in these roles often supervise university students, and likely view their relationship to PhD students to be that of teacher to student. Other participants included chief scientific officers and CEOs of biotechnology companies or relatively high-ranking administrators within the public service. A small number of participants included laboratory technicians and fellow students. In essence, the majority of my interviews involved interviewing elites where I was in a position of relatively little power. To a certain extent, the power asymmetry worked to my advantage. Some participants stated that they agreed to the interview to ?help? a PhD student in their research. These individuals may have felt in the interview that they were ?teaching? me, potentially going into more detail in their answers. Others likely did not reply to my request for an interview because of my relatively low rank. To try to counteract this asymmetric dynamic, I dressed professionally for all interviews and acted in a self-confident manner.   Participants, for the most part, were quite busy. I employed several strategies to mitigate the time constraints of participants. First, I followed up numerous times with each individual contacted. Second, I conducted all interviews at a time and location chosen by the participant. Third, I underscored that the time commitment for each    93  interview should only be an hour and respected any and all time constraints of the participants. Fourth, I employed a semi-structured interview strategy, an interview style that is often the best choice when interviewing busy individuals, enabling all interview themes to be covered in a relatively efficient manner (Bernard, 2006). Thirty-six of the interviews were in-person at a time and location chosen by the participant. Two additional interviews were conducted by phone at the request of the participant. The majority of in-person interviews were conducted in the office of the participant. Other interviews were conducted in coffee shops, conference rooms, lunchrooms, and empty offices. Following appropriate informed consent procedures, consent forms were sent to each participant at least 24 hours before the interview. These were discussed and signed at the beginning of each interview. I offered each participant a copy of the consent form for his or her records. For Rounds 1 and 2 of interviews I created two sociograms for each participant. The first was an egonet of each participant and their co-authors both within and outside of Vancouver. To create this sociogram, I found the paper ID for all papers that listed the participant as an author within my dataset and extracted a table that included the names of all authors for each paper ID, paper IDs, paper titles, and publication years. I manually cleaned author names to first initial and last name and used this table to create a two-mode sociogram in NetDraw that connected the interview participant to their co-authors through paper IDs. The second sociogram was based on a table that connected paper IDs for all papers that listed the participant as an author to the organisations connected to each paper ID. This was used to create a two-mode sociogram in NetDraw that connected organisations through paper IDs. Samples of both sociograms can be found in Appendix C. Yin (2003) recommends that interview questions in case study research are draw from related theoretical themes. My interview questions aimed to develop themes related to innovation systems and neo-institutional theory by exploring participants? work and roles; explanations of why they would collaborate, particularly between sectors; and the impact of institutions on collaboration. Drawing from institutional ethnography, I paid particular attention to encouraging participants to describe in detail ?the work? that is part    94  of these collaborations in order to understand related institutions (Campbell and Gregor, 2002). My initial interview guide was piloted with an individual involved in I2 R&D and interview questions were modified based on feedback from that interview. Interview questions evolved slightly over time. Interviews lasted from half an hour to over an hour.  Appendix C outlines a sample interview guide. For interviews where I was able to create a two mode co-author/paper egonet, each interview began by my asking the participant whether they believed this sociogram accurately reflected their collaborations. This served both as a triangulation method for my bibliometric analysis and to encourage the participant to begin thinking about their collaborations. Co-authors represented in the sociogram, and collaborations associated with specific publications in the sociogram, were often discussed in the interview. This sociogram frequently acted as a visual prop during discussions with participants who pointed to specific co-authors and papers to illustrate their point.  All interviews included basic demographic survey type questions exploring an individual?s educational and work background, organisational affiliations, and which organisation the participant considered ?home.? While all interviews focused on collaboration, interviews closer to the beginning of the process focused more on descriptions of a specific collaboration and challenges and incentives of this collaboration. Interviews occurring later focused more on descriptions of reasons for collaborating with different sectors and the nature of the participant?s research and work. Probing questions were asked as needed and interview questions? order changed to fit the flow of the interview. Round 3 interviews with participants affiliated with firms followed a similar format, although sociograms were neither created nor discussed during the interview. Round 3 interviews with administrators focused on specific and detailed questions related to the organisations that each administrator was affiliated with. 4.3.2.3 Organisational and institutional documents Interviews were supplemented by analysis of related documents. Text is often seen as a fundamental representation of institutions because it is generalizable and translocally produced (Luken and Vaugh, 2006; Campbell and Gregor, 2002). The majority of these documents were organisational strategic plans, reports and websites.    95  These were used to help understand and describe the organisations and institutions that formed the organisational field under study and to create organisational profiles. I identified the majority of these documents based on institutions and policies discussed during interviews. Participants gave me many of these documents. I searched for other texts to create more fulsome organisational descriptions. Together these institutional and organisatonal documents, combined with interview transcripts, were used to create an overview of Vancouver?s I2 organisational field, explore the motivations of individuals within the field to collaborate, and understand how these collaborations were influenced by organisations.  4.3.2.4 Qualitative analysis No formal process exists for analysing case studies, although a final report generally includes the case description, context, themes, and lessons learned (Creswell, 2007). Combining these components with my theoretical framework formed the basis of my analytic strategy. Stake (2005) suggests that case study analysis should be interactive and iterative, beginning simultaneously with the data generation process.  The first stage in my analytic process occurred within hours of each interview when I wrote field notes for each interview. Written as a flow of consciousness, these notes ranged in length from 2-6 single-spaced typed sheets. The goal of these notes was to write as much about the interview as possible. A key component of these field notes was to include an observation component for each interview, something that would be lost in the audio recordings and subsequent transcriptions. Observations often included dress of participants, particular speech or behaviour, the setting and relevant physical objects, as well as personal feelings (Bernard, 2006). In addition to these observations, my field notes summarised the interview and outlined initial themes and points of interest stemming from the interview. I reviewed these field notes and cut and pasted their text into a common template?creating interview descriptions?as a second stage in analysis. This template included participant?s pseudonym, date, and location of the interview; physical and emotional observations from the interview; interview summary; themes identified from interview; thoughts related to the interview; and notes for future interviews.    96  The third stage of my analysis involved transcription. Thirty-six of the interviews were recorded and transcribed. I transcribed one of the interviews owing to poor sound quality the day after the interview occurred and had all others professionally transcribed. One study participant declined to have their interview recorded and recording equipment malfunctioned in another interview. In these two cases, detailed field notes are used as the record of the interview. To increase the reliability (Kvale, 1996) of all transcription, I personally checked all transcribed transcripts against the audio recording.  The fourth stage in my analysis drew from a condensation (Kvale, 1996) or vignette (Seidman, 1998) approach. In these approaches, analysis focuses on condensing interview text into essential meaning statements and combining these meaning statements together into a narrative that tells a story gleaned from the interview. As I checked interview transcripts against audio recordings, I expanded interview summaries in my interview descriptions into narratives. My field notes were relatively good at identifying the essential meanings from interviews and putting these into a rough narrative. To these in italics I added specific quotes, names, numbers, and points missed during my initial summary. During this review I also added information to the themes and thoughts sections of each description, creating expanded interview descriptions. As I reviewed transcripts I also created standardised profiles of participants, which included their pseudonym, sex, home organisation, affiliations, titles, education, as well as descriptions of their research focus, models for research, workload balance, organisational collaborations, and additional notes. This was used as the basis for Appendix B.  The fifth stage of my analysis drew from a data categorisation approach (Seidman, 1998; Kvale, 1996). In data categorisation key themes, often called codes, are identified from interview transcripts. These are then used to tag specific text in interview transcripts, often with the aid of specialised software. I began developing a code list after I had checked the transcription of the first 18 interview transcripts and developed interview descriptions for each interview. I consolidated all of the themes listed in the themes section of these 18 interview descriptions into a single document. I grouped these themes into meta themes through an iterative process and wrote    97  descriptions of each meta theme and listed the relevant interviews. This resulted in 14 meta themes. I considered these meta themes during verification of later interviews and added new themes as necessary. After reviewing all interviews, I expanded this coding list to include 17 meta themes and 41 sub themes. This final coding list is given in Appendix D. Codes were both descriptive and thematic (Richards, 2009).  My descriptive and thematic analyses drew both from the institution and organisational fields literature as well as letting key themes emerge from the data. Drawing from the organisational field literature, descriptive codes focused on the key organisations that were part of Vancouver?s I2 organisational field. From the interviews descriptive codes were added related to the roles of participants and the types of research they conducted. These emerged as important to participants? identity and their decisions of with whom, and towards what ends, they collaborated. Analytic codes drew from institutional literature to explore the institutions identified by participants as influencing collaboration. Additional analytic codes that emerged from interviews investigated the reasons participants gave for collaborating and the different pathways identified by participants between research and application. As analytic themes further emerged from the data, particularly those related to the reasons participants gave for collaborating, I returned to the existing literature to expand my theoretical framework to include these emerging themes. My analysis thus continued in an iterative fashion, using theory to help guide my findings and my findings to help expand my theoretical framework.    Following Creswell (2007), I began designing my final report around a description of my case and themes. In addition to participant profiles, I began creating organisational profiles based on organisational documents. I loaded all interviews, interview descriptions, organisational profiles and my coding list into Atlas.ti as my sixth stage of analysis. I continued my categorisation analysis as I coded all interview descriptions based on my coding list. Writing memos is a key stage in qualitative analysis; these memos are separate documents that contain observations about ideas, methods, and documents that emerge from your analysis (Richards, 2005). I began writing descriptive and theme based memos within Atlas.ti as the initial draft of the    98  results outlined below.  Through memos I expanded on draft organisational profiles based on interview summaries and interview transcripts. I also began writing thematic memos combining and expanding on meta code descriptions by reviewing coded interview summaries.  Analysis continued through an iterative distillation and expansion process. As I expanded on descriptions, thoughts, and themes identified from interview summaries in specific memos, I often reread and coded the full interview transcripts. To increase the validity (Kvale, 1996) of this analysis, throughout this process, the original audio interview recordings were listened to again and key thoughts written in a physical notebook; these thoughts were later incorporated into memos.  4.4 Vancouver?s I2 organisational field Drawing on my I2 database, I map Vancouver?s I2 organisational field here. Table 12 summarises the number of organisations and authors within Vancouver?s I2 network by sector. Even after multiple affiliations with universities are controlled for, universities still have the largest number of authors within the network. This is followed by government organisations. Firms have the largest number of organisations within the network but the smallest number of authors. This hints at the fact that firms within the network are quite small.  Table 12: Vancouver authors and organisations by sector Sector # Org # Auth Uni 6 435 Govt 10 272 Hosp 5 191 Firm 13 24  Table 13 lists all Vancouver organisations within the network, their sectoral affiliation, the number of co-authorships (also known as degree) researchers affiliated with each organisation have globally, the percent of these co-authorships that are in Vancouver, and the number of Vancouver organisations each organisation is connected to through the co-authorships of its affiliated researchers. Acronyms of organisational    99  names can be found in Appendix E. UBC dominates the network. In addition to having the largest number of co-authorships, through its affiliates, it is connected to almost 55% of other organisations within the network. St. Paul?s comes in a far second with 63% of the number of co-authorships and less than 56% of the organisational connections of UBC. BCCDC is in third place in terms of total co-authorships while CFRI is in third place for number of Vancouver organisations it is connected to. Overall, firms are the least connected. Only one firm, Tekmira, is in the top 10 organisations ranked by degree.  Firms showed high variation of local compared to non-local co-authorships. Some published entirely with locally affiliated individuals. Others published entirely non-locally. Overall, firms co-authored the least with individuals affiliated with other Vancouver organisations. Slightly over half (56%) of their co-authorships were local. Government organisations co-authored the most locally (73%).     100  Table 13: Co-authorships for researchers affiliated with Vancouver organisations Organisation Sector Total % Van # Van Orgs UBC Uni 5379 65% 18 St Paul's Hosp 3405 63% 10 BCCDC Govt 1506 66% 8 CFRI Govt 1455 76% 9 VCH Govt 827 81% 8 BCCA Govt 783 68% 8 CWHC Hosp 647 54% 8 SFU Uni 305 52% 4 VGH Hosp 152 60% 10 Tekmira Firm 151 77% 1 BC Biomed Labs Ltd Firm 38 89% 2 Glover Medical Clinic Firm 30 0% 0 UBC Hospital Hosp 30 20% 1 AlCana Technol Inc. Firm 29 52% 1 BC Institute of Women & Children?s Health Govt 27 81% 1 Fraser Health Govt 25 56% 3 Liver and Intestinal Research Centre Firm 18 0% 0 Infectious Control Network Govt 16 100% 1 PHSA Govt 16 88% 3 Gain Medical Centre Firm 15 7% 1 CMMT Uni 12 17% 0 Occupational Health and Safety Govt 12 100% 1 BC Ministry of Health Govt 11 91% 3 Chemokine Therapeutic Firm 11 0% 0 Richmond General Hospital Hosp 11 100% 2 iCo Therapeutics Firm 8 63% 2 Trinity Western University Uni 8 38% 1 West Coast Clinical Research Corporation Firm 8 13% 1 BCIT Uni 7 57% 2 Douglas College Uni 6 100% 0 Sirius Genomics Inc. Firm 6 100% 2 Bernadette Stringer Consulting Firm 4 0% 0 Protiva Biotherapeutics Firm 2 100% 0 Viroforce System Firm 1 100% 1     101  In Figure 8 I visually show the importance of different organisations within the network using Gephi. To create this sociogram, individuals were grouped into their respective organisation through a block modelling procedure. Individuals with more than one affiliation in the adjusted dataset are counted once for each affiliation. Circles represent organisations. Circle size is based on the degree, here representing local co-authorships, of each organisation and the top ten organisations by degree are labeled. The colour of different circles represents the sectoral affiliation of each organisation: universities are purple, hospitals are pink, government organisations are green and firms are blue. Lines between organisations represent the number of co-authorships between organisations based on affiliations of researchers, thicker lines represent a greater number of co-authorships. As can be seen from Figure 8, UBC and CFRI have the strongest interorganisational tie, followed by ties between UBC and BCCA and then by ties between CFRI and CWHC. In Figure 8, most of the strong ties within the network involve UBC. Ties between most other organisations within the network are relatively weaker. Five firms within the network are isolates, meaning that they are not connected to any other Vancouver-based organisation. The two university based isolates are Douglas College, a local community college, and the CMMT, which is a collaboration between UBC and CWHC. Some of the firms within the network have relatively strong ties to organisations other than UBC. The tie seen between a firm and government organisation on the bottom left is between West Coast Clinical Research and Fraser Health, one of the province?s health authorities, described in more detail below. iCo Therapeutics, bottom middle, is connected to both UBC and BCIT. The Gain Medical Centre, upper right, is connected to VGH and Sirius Genomics Inc., top middle, is connected to St. Paul?s and BCCA.      102  Figure 8: Organisations in Vancouver's local I2 network   4.4.1 Public presence within the field Within Canada, hospitals and universities are both publically financed. Thus hospitals, universities and government organisations can all be viewed as public organisations, albeit with different degrees of autonomy from direct government control. As seen in Figure 8, public sector organisations dominate Vancouver?s I2 network. All public organisations operate within specific government Ministries and mandates that help to create institutional norms within the organisation. Occupation Health and Safety is part of WorkSafeBC, an independent agency created as part of the BC Workers Compensation Act with a board appointed by the Ministry of Labour.     103  UBC and SFU are both part of the province?s postsecondary education and fall under the mandate of the Ministry of Advanced Education as well as Labour Market Development (Brimacombe, 2010). UBC is a public university with two main campuses: one located in Vancouver and another in Kelowna, BC. Its Vancouver campus is home to more than 37,000 undergraduates and 10,000 graduate students.12 The primary campus of SFU is in Burnaby (a suburb of Vancouver) with other campuses located in Surrey (another Vancouver suburb) and in downtown Vancouver. SFU teaches over 30,000 undergraduates and 5,000 graduate students (IRP, 2013). UBC and SFU administration is mandated by the BC Government University Act and is composed of a convocation, which confers degrees and diplomas; a board, responsible for property management and revenue; a senate, responsible for academic operations; and Faculties (University Act, 1996). Both universities are considered corporations, although the BC Business Corporations Act is only applied to them selectively (University Act, 1996). The main function and duty of both universities is instruction of knowledge and conduct original research (University Act, 1996, 47.2). Related to I2, SFU does not have a Faculty of Medicine, and, as a result, does not have a medical school. The UBC Faculty of Medicine includes 633 full time faculty, 4,860 clinical faculty, and 1,287 graduate students.13  All other public organisations within this network fall within the BC Ministry of Health. The mandate of the BC Ministry of Health is to administer access to medically necessary hospital and physician services as entreated by the Canada Health Act. The BC Ministry of Health has devolved decision-making authority for much of the province?s health care management and delivery to six non-profit health authorities. The health authorities are responsible for local health services such as public health, mental health, home care, residential care and hospital care. Five of these are regional, with two, Vancouver Coastal Health (VCH) and Fraser Health, covering the greater Vancouver Area that is the focus of this study. The Provincial Health Services Authority (PHSA) is the sixth health authority. It is responsible for providing province-wide specialized                                             12 See http://www.ubc.ca/about/ Accessed January 24, 2013 13 See http://med.ubc.ca/about/facts-figures, Accessed January 24, 2013    104  services. While health authorities share a common mandate related to health care management and delivery, some have also incorporated basic and clinical research, education and training into their own mandate (see for example Powell and Cranston, 2010). Each health authority is organized differently. With regionalisation of health care in BC, much administration has moved from individual organisations to the regional health authorities, with each health authority structuring administration differently. With the exception of St Paul?s Hospital, all of the traditional teaching hospitals now have no separate or distinct governance structure within the health authorities (Bressler and Campbell, 2010) although, as entreated in the BC Hospital Act, each hospital must have its own superintendent (Hospital Act, 1996). Within VCH, Providence Health Care, a Catholic health care provider that operates more than five sites in the greater Vancouver area, acts as a separate organisation with its own board in accordance with a dominational agreement between the province and its faith-based providers (Bressler and Campbell, 2010). Providence Health Care, in many ways, acts as an additional health authority within Vancouver. Overall Providence Healthcare has approximately 22,319 inpatient admissions and 94,802 emergency room visits as well as specialty procedures such as 13 heart transplants and 100 kidney transplants a year (Providence Health Care, 2012b). Providence Health Care administers St Paul?s Hospital. Located in downtown Vancouver, St. Paul?s Hospital is an acute care teaching and research hospital with 520 acute care beds. Every day, up to 300 patients visit St. Paul?s emergency room (Baron, 2012) and, in 2011, more than 1,660 babies were born at St. Paul?s (O?Connor, 2012). St. Paul?s is home to the Heart and Lung Institute, the Heart Centre, the Diabetes Teaching and Treatment Centre, the Centre for Health and Evaluation and Outcome Sciences, a Palliative Care Unit and the James Hogg Research Centre.  Related specifically to I2, St Paul?s HIV/AIDs and Addiction Program is the largest comprehensive program of its kind in Canada, providing care to more than 65% of the seropositive people in British Columbia.14 The program is the headquarters of the national and provincial offices of the Canadian HIV Trials Network. There is an                                             14 See http://www.providencehealthcare.org/health-services/hivaids accessed January 22, 2013    105  infectious disease clinic, called ?10C? on the 5th floor of St. Paul?s to treat HIV-positive individuals. St. Paul's also has a 25-bed inpatient unit to assess, treat, and manage patients who are acutely medically ill and HIV positive (Providence Health Care, 2012a). St. Paul?s also houses a Centre for Excellence in HIV/AIDs. Established in 1992, the Centre is a provincially mandated organisation that integrates HIV/AIDs treatment, research, and education with approximately 100 staff and reports directly to the CEO of St. Paul?s (Providence Health Care, 2009). As part of its treatment mandate, the Centre manages the procurement and distribution of antiretroviral drugs for BC; monitors clinical, laboratory and epidemiological impacts of highly active antiretroviral therapy (HAART) across the country (except for Quebec); and generates BC?s HIV/AIDs Therapeutics Guidelines (Provincial Health Care, 2009). Research at the Centre includes population and epidemiology work, laboratory research, international research, and clinical trial activities (Providence Health Care, 2009). Education includes a biannual update related to HIV/AIDs research and treatments, workshops and newsletters (Providence Health Care, 2009). VCH administers Vancouver General Hospital (VGH), Richmond General and UBC Hospital. VCH has one CEO. Three chief operating officers are responsible for different geographic areas. VGH and UBC hospital are in one geographic area. Research across organisations is coordinated by the Vancouver Coastal Health Research Institute, which has formed centres based on different research themes. These centres act as administrative entities that bring people together around a particular topic, allocate resources within each centre, act as a training forum for students, and enable researchers to apply for funding together. One of these centres specifically focuses on I2.  PHSA administers the British Columbia Centre for Disease Control (BCCDC), British Columbia Cancer Agency (BCCA), Women?s Health Research Institute, Child and Family Research Institute (CFRI), Provincial Infection Control Network, and Children and Women?s Health Centre (CWHC). Organisations within PHSA each have their own head, are organized differently, and integrate research into their organisations    106  in different ways. A chief administrative officer, research, helps to coordinate research across PHSA organisations.  BCCDC was established in 1997 and later incorporated as a branch agency of PHSA in 2002 (Bibby, 2009). It ?is responsible for preventing and controlling communicable disease and promoting environmental health for British Columbia.? (PHSA, 2007, p.1). It achieves this by supporting the regional health authorities, the BC Ministry of Health, and acting as the scientific support arm of the Provincial Health Officer (Bibby, 2009). In 2010/11, BCCDC had 32 researchers and was awarded $3,412,744 in research grants, $2,413,453 in operating grants and was involved in training 63 students (Chesney, 2011). Most of the work in BCCDC does not involve directly seeing patients. Instead, responsibilities at BCCDC include: public health information system management, program consultation, disease surveillance, epidemiological analysis, policy analysis, best practice guidelines, outbreak investigation, education, disease control and prevention planning and services, research and evaluation. The exception is the work of Clinical Prevention Services, called 'Service' by people within BCCDC, which provides patient care related to sexually transmitted infection, HIV prevention, Tuberculosis, as well as non-patient care related to Hepatitis. Satellite clinics affiliated with BCCDC also provide Tuberculosis and Hepatitis care across the province. Concurrent to the establishment of BCCDC the UBC CDC was established. Today the two organisations are essentially one and the same; however, the separate title serves to underscore the teaching and research mandate that is part of BCCDC. As Jason argues: ?the affiliation, it was very meaningful, because it really accented that BCCDC had roles, not only in the delivery of public health services, but also in education, training and in research.?  CWHC includes BC Children?s Hospital and BC Women?s Hospital, which are located on one campus in midtown Vancouver. BC Children?s Hospital provides specialized pediatric care for British Columbia, as well as pediatric services, such as an emergency room, for Vancouver-based children. BC Women?s Hospital provides primary and specialized care for women, newborns, and their families. BC Women?s    107  Hospital delivers approximately 7,000 babies a year, nearly 20% of BC?s babies.15 In addition to maternity care, BC Women?s Hospital offers services ranging from breast screenings to HIV care to osteoporosis programs.       BCCA provides a province-wide, population-based cancer control program for the residents of British Columbia and the Yukon through five regional treatment centres across British Columbia. I2 is not the explicit focus of BCCA, although I2 relates to much of its work. In 2011, BCCA treated over 18,000 new patients (BCCA, 2012) and employed over 275 researchers (Chesney, 2011). Services within BCCA include cancer screening, genetic counseling, diagnostic services, surgical oncology, radiation, chemotherapy, and supportive care (BCCA, 2012). Research at BCCA ranges from cancer control to molecular oncology and includes a tumour tissue repository. BCCA also houses the Genome Science Centre, which specializes in large scale ultrahigh throughput sequencing applications. In addition to focusing on cancer related research, the Genome Science Centre provides a technology platform for Canadian genomics initiatives.16 Other Ministries including Health Living and Sport (responsible for health promotion for the province) and Children and Family Development (responsible for promoting health and wellbeing in children and families) have mandates tangentially related to some public organisations within this network (Brimacombe et al, 2010). 4.4.2 Private presence within the field The private sector dominates this field in absolute number of organisations but not in number of co-authors. The private sector is supported and constrained by government organisations and policies. BC?s Innovation Council and the Ministry of Small Business and Economic Development are the two organisations that are responsible for the research and innovation mandates for the province related to commercialisation. No major pharmaceutical company has an office in Vancouver. Before the Great Recession of 2008, Vancouver biotechnology had been hailed as a                                             15 See http://www.bcwomens.ca/AboutUs/BCWomens/FactsAboutUs.htm, access January 28, 2013 16 See http://www.bccrc.ca/dept/cmsgsc, accessed January 24, 2013    108  budding cluster (Salazar et al., 2008). Participants interviewed who have been part of Vancouver?s biotechnology sector agreed that the sector had been hit hard during the 2008 financial crisis. It was challenging to find companies working within I2. As Adele, someone who has worked in biotechnology in Vancouver, said; ?BC academically has a lot of strengths in infectious disease. As so that?s why it?s unfortunate that a bunch of company names are not rattling off my tongue.?  During the Great Recession, capital dried up globally for biotechnology (Ernst and Young, 2012). Simultaneously, problems in the pharmaceutical sector made it increasingly challenging for biotechnology companies to get contracts from larger pharmaceutical companies, another potential source of cash. These factors made it hard for Vancouver biotechnology companies to remain afloat and for new companies to be founded. The two largest biomedical companies in Vancouver, QLT and Angiotech, both took hard hits over the past several years. Angiotech is a medical device company. It filed for bankruptcy protection in January 2011. While it has emerged from bankruptcy protection, the process left it greatly weakened. QLT began 2012 with 214 employees and ended the year with 38 after a series of layoffs. During the year, it also sold its flagship product, Visudyne, which fights age related blindness (G&M, 2012). Participants thought that the collapse of these and other companies had caused first a glut and then an exodus of skilled biotechnology labour: A lot of folks lost their jobs.  There was a bit of a glut in the employment market in biotech and then not enough employment opportunities, so people leave.  There?s not enough venture financing to kind of revitalize the market.  So kind of a constellation of factors, I think, conspiring against us. (Bill)  Participants that were affiliated with biotechnology companies viewed their companies as actors on a global stage more than they viewed them as part of a local network, ?we think of ourselves and we operate globally? (Bill). One of the affiliated companies followed a centralised organisational structure. Spun out of UBC, this company provides a platform technology that can be applied to a variety of disease areas and also develops therapeutics using its platform. This company conducts much of its research and production internally with approximately 35 researchers working    109  onsite at the time of my interview in spring 2011. The company may keep much of its work internal in order to maintain full ownership of their platform. Organisations often work with this company in order to develop or test something on their platform.  The other four affiliated companies followed what participants called a virtual model. The physical shell of these companies was small with between two and twenty permanent staff. Almost all of their work was outsourced through contracts, generally to other companies as opposed to public organisations. Peggy described this as ?really a hub and spokes model that we have whereby our hub is these five people, but we have about three organisations at a time working for us on a contract research basis.? Only one of these companies maintained a small research and development lab for critical path items. This model was viewed as being that of the ?modern company? which, instead of vertical integration, involved specialising in one part of a larger value chain:  I don't think the modern company is vertically integrated. So the modern company doesn't dig a resource out of the ground, mill it up, make steel out of it, build the buildings with it, and sell the buildings on the other end. You know, they don't do all those things. That's sort of an old method. I think the modern company does one of those pieces very wide. So it will dig not just iron ore, it also digs copper and gold, and so it's very wide, but it doesn't do any of the other bits and pieces (Lane).  For biotechnology companies, participants viewed their part of the value chain as advancing early stage development of a product. As Peggy stated ?we stick to our knitting, which is the development and clinical advancement of the product.?  This virtual model was seen as advantageous in order to lower infrastructure costs, access leading specialised expertise, control cash burn, and outsource risk. As a result, most of the employees in each company took on a project management role, supervising other organisations that were conducting the work. Three of these companies were focused on therapeutic development while the two others focused on diagnostics. The primary market of focus for all companies was the USA. The exit strategy of the four virtual companies appeared to be sale to a larger company; ?so this company was basically built to be sold? (Pete).     Beyond biotechnology companies, other private organisations are part of the Vancouver I2 organisational field. While doctors? visits are paid almost exclusively by the    110  government within Canada, many doctors who work outside of hospitals run their own private practice, such as the Glover and Gain Medical Clinics seen in Table 13, and could be considered part of the private sector. One individual interviewed worked in such a private practice. In addition, companies specialising in contract work from lab analysis, such as BC Biomedical Labs Ltd., to manufacturing to data analysis are contracted by biotechnology firms and can be located within Vancouver.  No contract work companies were part of my sampling dataset. When deciding on what types of companies to sample during my third sampling round, I decided to expand my sampling of biotechnology companies to reach saturation in this area. 4.4.3 Boundary spanning within the field Within Vancouver, organisations and formal institutions have developed to encourage collaboration across sectors with the majority of these boundary spanning efforts focused on facilitating collaboration between the universities and hospitals or between the public and private sectors. It appears that no formal organisations or institutions developed to encourage research across the health authorities; ?I would say that the collaboration with the other health authorities really primarily works researcher to researcher.? (Meghan)  Formalised collaboration agreements, called affiliation agreements, facilitate collaboration between UBC and the health authorities. Affiliation agreements predominately focus on hospitals agreeing to accept UBC medical students for training and giving these students patient access (PHSA, 2002). Affiliation agreements implement stipulations, outlined in the BC Hospital Act, that hospitals providing primarily acute care must provide facilities for giving clinical instruction to UBC medical students and members of the teaching staff (Hospital Act, 1996, 45). Affiliation agreements also mandate that hospitals provide reasonable academic space, including research labs and offices, for UBC staff and students. In return, UBC is responsible for the curriculum and discipline of its students and provides university appointments to Health Authority staff involved in UBC teaching programs (PHSA, 2002). Healthcare staff who are involved in clinical instruction through these affiliation agreements generally are given an affiliation with UBC in addition to their hospital affiliation. As discussed in more detail    111  below, the meaning of a university affiliation?and its associated institutions?varied greatly between interview participants that were physically located in healthcare organisations, based in part on each participants day-to-day work and perceived role. With the regionalisation of health care in BC, affiliation agreements now exist between UBC and five of the seven regional health authorities: VCH, PHSA, Fraser Health Region, Vancouver Island Health Authority and Providence Health Care (Brimacombe, 2010). With the exception of Providence Healthcare, UBC is not formally represented on health authority boards (Bressler, 2010).  Research affiliation agreements for intellectual property (IP) have developed separately from the main UBC/health authority affiliation agreements (see for example UBC, 2003). Generally these IP affiliation agreements place the responsibility on UBC to administer and enforce its Patent Policy in order to protect the IP rights of the health authority. The health authority reserves the right to take over IP administration in the future (UBC, 2003). Because healthcare researchers generally have UBC affiliations through affiliation agreements, they are obliged to report IP to UBC under its Patent Policy. Under IP affiliation agreements, health authorities and UBC share the direct costs of IP development. Any resulting revenue is first used to pay back direct costs and then split 50% to researchers, and 25% each to UBC and the health authority (UBC, 2003). BCCA is the exception. An UBC/BCCA affiliation agreement, predating the UBC/PHSA affiliation agreement, specifies that individuals working at BCCA have the right to develop their own IP commercially through the BCCA Technology Development Office (TDO). TDO facilitates relationships between BCCA and the private sector. Other researchers working within the health authorities with a UBC affiliation report to the UBC University Industry Liaison Office (UILO). The UBC UILO thus administers IP for UBC faculty and affiliates and negotiates contracts, material transfers, and licensing agreements between UBC faculty and organisations in the private, not-for-profit, and government sectors. Both technology transfer offices can be seen as acting as organisational boundary spanners between the public and private organisations within this organisational field (Colyvas and Powell, 2006). The UBC UILO also spans UBC and health authority work.    112  Within Canada, all research involving human subjects must be approved by a research ethics board (REB) at organisations that receive money from one of Canada?s three major public funding agencies (referred to as the ?Tri-Council?): Canadian Institutions of Health Research, Natural Sciences and Engineering Research Council of Canada, and the Social Sciences and Humanities Research Council of Canada. This policy applies even if the research is not funded by the Tri-Council (CIHR et al, 2010).  To fulfill these and other requirements pertaining to human participant protection, UBC has developed behavioural and clinical REBs, which have the authority to review research involving human subjects that occurs under the auspices of UBC (UBC, 2011). UBC affiliated REBs have been established at BCCA, CWHC, and Providence which function independently of, but in coordination with, the UBC REBs. These boards all use a common application form, and, through agreement, only one board normally supervises each application and subsequent project (UBC 2011; Bellward et al, 2007).  The UBC clinical REB reviews research that involves surgery, clinical interventions, exercise programs, and/or the analysis of clinical data at UBC?s Vancouver campus, Vancouver Coastal Health Authority sites, and clinical research that does not fall under the purview of site-specific REBs at BCCA, CWHC, and Providence.17  Research approval often occurs concurrently within the health authorities and hospitals related to resource allocation within the hospitals. UBC affiliated REBs function under the auspices of the UBC Office of Research Services; comply with UBC REB policies; report to UBC, Vice-President Research; and have been authorised by the UBC Vice-President Research to function as REBs.18 In addition to reporting to UBC, these boards report to their affiliated organisations and have their own terms of reference (BCCA, 2009). Thus, affiliated REBs act as boundary spanning organisations, complying to both UBC and organisation specific policies and ?bringing? UBC policies into other organisations.  The Centre for Drug Research and Development (CDRD), an established non-profit, acts as a newer boundary spanning organisation between academia and industry                                             17 See http://research.ubc.ca/ethics/clinical-research-ethics-board, accessed March 19, 2013  18 See http://www.bccancer.bc.ca/RES/REB/Intro.htm, accessed March 19 2013    113  within Vancouver. CDRD works in the pre-clinical space, trying to take ideas through a proof of concept and functioning in the area that has sometimes been called the ?valley of death? in drug development, past the traditional academic phase in concept development but before venture capital and pharmaceutical companies begin to take interest in a new prospect. Through a group called the search and evaluation team, CDRD finds academics with promising ideas to develop. After an idea is identified, a contract outlining general expectations and timelines related to the initial development of the idea is written between the academic principal investigator and CDRD. Some of these projects are very small, focused on testing concept viability. Other ?incubation? projects last between three and six months and focus on a key question. Large projects last between six and nine months and further develop ideas. Intellectual property stays with the university or investigator depending on university intellectual property policies. CDRD offers their time and expertise to help in development, working with the primary investigator to bring the idea forward. This can either be done in house by a group of permanent researchers or by identifying other academics, generally in Canada, that can bring their own related expertise to further develop the project through a collaboration that is often relatively academic in nature. The primary investigator generally devotes related time and consumables, such as assays, models, and mediums, from their lab.   Incorporated in 1995, CFRI is located on the CWHC campus.19 CFRI is dedicated to children?s and women?s health concerns and is the largest research institute of its kind in Western Canada. Registered as a non-profit society, CFRI was co-developed by UBC and PHSA and is considered a partnership between UBC, PHSA, and CWHC by UBC (Farrar, 2011). PHSA considers CFRI to be the research arm of Children?s Hospital and its financials are rolled into Children?s Hospital and then up to PHSA. The CFRI board of directors reports to both UBC and PHSA (Farrar, 2011). In 2010/11, CFRI had over 200 affiliates who spent at least part of their week conducting research (Chesney, 2011). In that year, approximately 53% of CFRI?s annual revenues came from government, 22% from donations and foundations, 13% from UBC and 12% from industry and government grants and contracts (Chesney, 2011). While some                                             19 See http://www.cfri.ca/aboutus/default.asp, accessed January 24, 2013    114  affiliates work full time on research, others work as clinician-scientists who are funded to do both research and clinical care, and others still are funded to conduct clinical care. Child immunity research is one of CFRI?s areas of focus. As part of this focus, the Vaccine Evaluation Centre exists within CFRI. The Vaccine Evaluation Centre provides evaluations to help health authorities select the best vaccines and immunization programs. A group of investigators affiliated with the Vaccine Evaluation Centre work together, strategizing what sorts of grants and industry contracts to apply for. These can be administered through the Vaccine Evaluation Centre which then provides support services such as administration, nurse coordination, data management, and statistical analysis based on a portion of the grant money affiliated investigators bring in. Individuals working within CFRI view their ?home? organisations differently depending on their role with some viewing their home as CFRI and others as UBC. Culturally, CFRI appears to have strong ties to UBC to the extent that some researchers working within CWHC feel a divide between themselves and CFRI: And although, on paper, it?s clear how we?re supposed to be interacting with CFRI, in practice I don?t really understand it? I think it?s a conflict between PHSA, which is delivering health services, and the university, which has an academic mission. (Helo) CMMT is physically located in CFRI. Like CFRI, CMMT is considered part of Children?s Hospital and acts as a collaborative effort between UBC and Children?s Hospital.  4.5 Discussion and conclusion Chapter 4 introduced an analysis that continues into Chapter 5. Here I defined the focus of these two chapters, outlined key theoretical concepts of institutions and organisational fields, described the methods used in analysis, and combined my quantitative analyses to outline Vancouver?s I2 organisational field. In doing so, I have provided a background for the ?case? under study by describing dominant sectors, organisations, and formal institutions. This chapter begins to explore how formal institutions affect the governance of this organisational field and the interactions between individuals affiliated with different organisations. Scott (2008) argues that each organisational field has its own governance structure based on the public and private organisations within it. This ?cultural    115  unconscious? (Emirbayer and Johnson, 2008; Martin, 2003) helps to dictate how interactions occur within the field. Owen-Smith and Powell (2004) contend that fields dominated by public organisations will interact differently from fields dominated by private organisations as the institutions of the ruling sector prevail.  Vancouver?s I2 organisational field is dominated by the public sectors, and within the public sector, by academic organisations and related institutions. While the private sector experienced a great shock with the Great Recession and appears to be in almost constant flux, the public sector appears to be relatively stable. UBC is the central organisation within this field. Both UBC and SFU work under an official mandate focused on the instruction of knowledge and conducting original research. Almost all of the remaining public sector organisations within this field work under the BC Ministry of Health and are involved in healthcare. The mandate of the BC Ministry of Health is to administer access to medically necessary hospital and physician services. Contrary to a more traditional concept of one academic hospital affiliated with a university, multiple public organisations involved in I2 research fall within the BC Ministry of Health. These organisations are also actively involved in the province?s health care system. Separate health authorities administer local health services relatively autonomously from the Ministry of Health. Each health authority has a similar overarching mandate, but has also created individual organisational mandates and is administered differently. Formal organisational structures within these authorities often blur.  Although universities and the Ministry of Health have different mandates, lines between the two often seem to blur, through organisations such as CFRI, and formal institutions such as agreements related to affiliations, IP and ethics. These relationships are complex, evolving, and exist at multiple organisational levels. However, much of the formal institutional blurring between universities and hospitals appear focused on incorporating clinical training into the hospital setting and university based research structures?such as technology transfer and ethics boards?into service based research. These ?spanning? structures can thus be perceived as reinforcing the organisational mandates of universities as research focused and the Ministry of Health as focused on clinical care. These spanning structures appear to have limited focus on    116  knowledge translation between academic research and problems in healthcare. Regulative academic institutions spilled from academia into healthcare organisations through affiliation agreements and the use of UILO offices and REB offices run by UBC. Universities, to a certain extent, take ownership of research, whether it occurs in healthcare organisations or in academia; healthcare organisations take ownership of clinical training.  Firms within the network are large in number of organisations but small in size. The private sector appeared particularly fragile in the wake of the recent financial crisis. Relatively nimble when compared to public sector organisations, the majority of organisations that are part of this field have adopted a virtual model that relies on collaboration as the modus operandi. Worldwide, the biotechnology sector experienced a shock after the financial crisis as capital has dried up. The virtual model adopted by Vancouver-based firms appears to be illustrative of a ?new normal? within the biotechnology sector worldwide as firms restructured to address capital shortages (Ernst and Young, 2012). The reaction of firms within Vancouver to this crisis thus acts as an important illustration of how less developed biotechnology clusters responded to the financial crises and their ability to weather the financial storm. Boundary spanning organisations facilitate collaboration between the public and private sector through organisations such as CDRD and the UBC UILO. Much of this work appears focused on translating ideas from academia to the private sector. This bridging role of UILOs between universities and the private has become institutionalised (Colyvas and Powell, 2006). More recently, UBC?s UILO has begun expanding its bridging role beyond the private sector to include what it dubs ?knowledge mobilisation? to translate ideas from academia to both to industry and non-industry partners (UBC, 2012) Chapter 5 builds off of the theories and descriptions developed within this chapter by exploring key themes related to how institutions structure collaborations within this organisational field.      117  5 Chapter 5: The microfoundations of collaborations within Vancouver?s I2 field  5.1 Introduction In Chapter 4, I outlined the theoretical underpinnings of neo-institutional theory by describing institutions and organisational fields. Drawing on both qualitative and quantitative data, I then outlined key organisations, institutions, and structures of Vancouver?s I2 organisational field. In this chapter, I build off the theoretical and descriptive work of Chapter 4 to further explore the microfoundations of Vancouver?s I2 organisational field. I do this by drawing predominantly from interview data to investigate the perspectives of participants regarding their work, their reasons for collaborating, and how organisational and sectoral institutions affect their actions. This chapter explores my second and third research questions:  2 What reasons did people give for collaborating? 3 How do institutions such as policies, norms, and organisational culture affect collaboration? Figure 9 shows the participants whom I interviewed for this study and their co-authors in I2. This figure includes all of the papers where an individual that I interviewed acted as a co-author and is based on the subset of my database, described in Chapters 2 and 4, where authors could be connected to organisations between 2008 and 2011. This subset was used because of its increased accuracy. However, it is different from my interview sampling subset and three interview participants are excluded because the authors and addresses on their papers did not properly match. The six participants sampled in Round 3 though snowball sampling and opportunistic sampling techniques are not part of my I2 database and are also excluded. To create Figure 9, I searched for all of the articles published by the participants I had interviewed within my subset and exported this list of papers, authors, and their organisational affiliations. I coded authors based on whether they had been interviewed, were affiliated with a Vancouver-based organisation, had any Canadian-based organisational affiliations or met none of the criteria. Two mode author/paper matrices were transformed into one mode author/author matrices in UCINET. This one mode matrix was then exported to Gephi where the sociogram was drawn.    118  In Figure 9 blue circles represent interview participants, green circles represent co-authors affiliated with Vancouver-based organisations, purple circles represent co-authors affiliated with a Canadian organisation, and red circles represent co-authors affiliated with an organisation from somewhere else in the world. As can be seen in Figure 9, ten (34%) of the participants interviewed had directly co-authored a paper with another participant. Twenty-two (75%) of the participants were connected to other participants through Vancouver affiliated co-authors. Canadian affiliated co-authors did not connect additional participants to the group and adding all co-authors only connected one additional participant. It?s important to realise that while all participants shown in Figure 9 published in the area of I2, the primary focus of each participant?s work was not necessarily I2. For example, the primary focus of participants included Alzheimer?s Disease, Huntington?s Disease, and ovarian cancer. Thus, some participants may be better connected to other networks, or even to the participants shown here, but through research focused on other subjects.     119  Figure 9: Participants interviewed and their co-authors   Appendix C provides an overview of some of the characteristics of the participants interviewed for this study; characteristics further summarised in Table 14. Backgrounds included PhDs (16), MDs (9), MD/PhDs (5), Bachelors (3), Masters (4), and Nursing (1). Official roles of participants included service/researcher (10), researcher (10), administrator (9), research support (4), service (4), and student (1). Out of the 38 participants interviewed, 14 were women. Administrative, research and research support roles were relatively evenly divided between the sexes. The majority (12 out of 14) of individuals interviewed in the service and service/researcher roles were men. All but one of these positions was physically located within healthcare organisations. Here service is seen as someone who provides service roles through a healthcare organisation to the general population. While the majority of this is clinical    120  care, participants were also involved with other service based activities such as diagnostic testing, epidemiological surveillance, and autopsies.   Table 14: Summary characteristics of participants Female Male Hospital 5 13 Administration 1 1 Researcher 2 3 Research Support 1 1 Service 3 Service/Researcher 5 Student 1 University 3 3 Researcher 2 2 Research Support 1 Service/Researcher 1 Firm 2 4 Administration 1 3 Researcher 1 Service 1 Government 2 4 Administration 1 Research Support 1 Service/Researcher 1 3 NGO 2   Administration 2 Total 14 24  As can be seen in Table 14, the greatest variety of participants? roles was within the hospitals. Of particular interest is the relation between individuals in researcher, researcher/service and service roles. Participants in all three groups were physically located in hospitals and had both hospital and universities titles. While these participants often looked similar on paper, their perceived roles and relations to both their affiliated university and hospital were often highly varied.  For example, participants in researcher roles generally viewed their primary affiliation to their hospital, participants in service roles as to their hospital, and participants in research/service roles as somewhere in between, depending on the participant. Participants who were    121  predominately involved in service work discussed how their university affiliation was necessary for them to practice in their hospital and brought with it a requirement to be involved in medical residency training but carried little other meaning to them:  If you want a university appointment, you have to do something for them. And in many medical centres not having an appointment is not an alternative. Your ability to work in a hospital is tied to having an appointment. So essentially you?re held hostage of the university to do some teaching for them. (Saul) Terry, a service/researcher working in government, similarly views his UBC affiliation as having little meaning, flippantly stating that UBC gives ?me an email address, that?s it. Sometimes they pay for parking, sometimes they don?t.? The line between roles often appeared to be fuzzy. This is particularly true for the division between service/researcher and service. All participants classified in service were involved in research activities. However, the official role of participants classified in service was their service work; research activities were supposed to be done ?off the side of their desk and [they] not even are compensated for it? (Meghan). All individuals classified as service/researcher were ?expected? to conduct research as part of their role. Because individuals in service/researcher roles inhabit both the service and research realms they are often perceived as playing an important role in bridging research and clinical practice (see for example Lander et al, 2010). While the participants that I interviewed in service/research roles did inhabit both realms, the specific roles of each participant had generally developed in organic and ad hoc ways, defying broader classification. The degree to which a research expectation was formally written into participants? job description and supported either by protected research time or dedicated resources varied between individuals as did their backgrounds, service work and research interests. For example, while three of the service/researchers interviewed were involved in what could be called ?translational? research?for example performing diagnostic services and conducting research related to improving these services?three others focused on more ?basic? research?for example studying the underlying biological mechanisms of a disease in cell and animal models.  There were other roles that were necessary for research to be carried out.  This included research support roles such as lab managers, technicians, and data stewards    122  as well as organisational administration roles. I counted participants who worked in companies that followed the virtual and semi-virtual model as working in administration because most of their work involved overseeing contracts with other organisations. To a certain extent, individuals who worked as a principal investigator (PI) running their own wet research lab acted as administrators as well because most of their worked involved administration related to the activities of their lab. This appears to be part of a larger trend where PIs are predominately involved in what has been called ?articulation work,? which includes grant writing, drafting publications, and presenting findings at conferences (Fujimura, 1996).  However, PIs remain closely connected to the day-to-day work within their labs as well as the lab?s direction and identity (Cetina, 1999). Because of this, I chose to classify PIs either as researchers or, for those also involved in service, as service/researchers. Research conducted by participants ran the gamut from basic research on cell lines within wet labs to basic research on patient samples within wet labs to epidemiological and public health research focused studies using patient data in dry labs to clinical trial involvement to systematic reviews of other studies. While some participants focused on one research model or method, others were involved in a variety of different research models and projects. For example, one participant involved in the study of a particular disease is involved in both basic research based on mouse models and clinical trials. Other participants study processes in animal models and then further explore findings in patient samples.   The length of collaborations varied from short-term to decades-long collaborations. Regarding the nature of his ties, Eric argues: ?They?re organic in the sense that they tend to persist if they?re useful and die if they?re not. And some of the more durable ones are decades old.? Tommy co-directs a decade long research program with the professor in the office next to his. Sam moved to Vancouver with a longstanding collaborator who is now based at UBC. Saul and a collaborator at BCCA have been involved in running a tissue bank for over ten years. Ben describes an ongoing collaboration with a UBC professor that initially began in 1984.     123  Participants also discussed how they believed that particular research areas have a collaborative ethos that facilitates their own collaborations. PIs working within research labs often didn't see collaborations within the labs as 'real' collaborations, a result keeping with previous studies (Lander, 2011). ?Real? collaborations only occurred with researchers from another research lab. For lab technicians and students working within scientific research labs, the majority of their collaborations appear to be within their lab with their PI acting as both a bridge and gatekeeper to external collaborations. Laura notes that her PI sets up external collaborations and provides introductions and she then contacts the other lab based on her PI?s guidance. A fine line also appears to exist between what is perceived as a collaboration and what is perceived as a relationship established to gain access to a sought after resource. Many academics only perceived of relationships as collaborations if the people they worked with provided intellectual insight in addition to resources. This is different from collaborations involving firms where firms see any and all types of connections as a form of collaboration.   Participants explained that collaborations were often successful if you trusted that the person you were working with would create good and reliable data and that you each had defined roles on the project. Participants diverged in opinion on whether an underlying friendship and complimentary personalities created a 'gel' that lead to more successful collaborations. Ben argued that ?the difference is, working with different people, are defined by different people?s personalities, not by where they?re located.? Conversely Felix stated that "some of the best people have had terrible personalities, they don?t get along with anybody.  But that doesn?t mean they can?t do good experiments.  So you go by what they produce, not whether or not they?re pleasant colleagues."  5.2 Capital and collaborations: What I have is what you need Participants felt that it was essential for them to collaborate in their work: "I always collaborate with people...it?s impossible to work by yourself." (Terry). Most participants said that the organisational affiliation or geographic location of individuals didn't make a difference when choosing collaborators: ?it doesn?t make a whit of difference to me whether someone?s in UBC or St. Paul?s or Children?s or Women?s    124  Hospital or in industry. It?s simply a matter of, is there common ground scientifically, that likely will benefit from an interaction.? (Ben)  Instead, a theme that emerged from the interviews was that participants look for collaborators who share a common interest but bring different and complementary strengths to the collaboration: "I think that the aim of any collaboration is to use resources that you don?t have that your collaborators do" (Gauis). Participants believe that each potential collaborator had their own specialty and strength. In collaborating you accessed this strength rather than having to develop the strength yourself. Lee argues that "we pretty much look for the best guy in the world at whatever he does and go to him."  This concept of complementary strengths echoes the belief of both Giddens (1984) and Bourdieu (1986) that collaborations involve accessing the capital of another individual within existing institutional structures. Giddens (1984) argues that in addition to institutions, social structures such as organisational fields are made up of both human and nonhuman resources that can be used to enhance and maintain power.  From my own interviews I found that participants collaborated to access capital that fell into several broad categories: knowledge, techniques, physical resources, and status. Bourdieu argues that capital can present itself in multiple forms including human, economic and cultural. Human capital includes the stock of knowledge and competencies possessed by a person. Economic capital represents money or financial resources. Cultural capital can be embodied as symbolic capital of an individual, objectified into cultural objects, or institutionalised such as through an educational qualification. These forms of capital echoed the types of resources identified during interviews.  Participants discussed how they often collaborate with others who have related but unique knowledge. Lee, for example, collaborates with European researchers who are expert in a different bacteria strain from the one used in his research lab. Terry, who studies disease in humans, often finds collaborators who study the same disease in animals. Crystal, an epidemiologist, collaborates with clinicians to understand the clinical presentation of disease and more basic scientists to understand its underlying mechanisms.      125  A large amount of the work that occurs in labs involves discovering or perfecting experimental techniques (Price 1984). Techniques can unify researchers working on different projects that draw from different knowledge bases (Lander, 2011). In this study Sharon explains "most of the research[ers], they use different ideas, but the techniques are the same." Technicians often possess much of the technical skill within a lab and thus their collaborations often revolve around sharing techniques (Barley and Bechky, 1994). In my interviews, collaborations based on sharing techniques was identified as particularly important to individuals involved in research support roles, many of whom worked as technicians: "So my job is not [to] give them ideas, my job is just [to] teach them, [to] show them the techniques." (Sharon)  Knowledge and techniques can be conceived as examples of human capital and comprise two different ways of understanding within scientific research (Lander, 2011; Fujimura, 1996; Barley and Bechky, 1994; Etzkowitz, 1992). Knowledge focuses on understanding nature through theory development; techniques on understanding through hands-on experience based on the technical aspects of a project (Shrum et al., 2007; Hackett, 2005; Barley and Bechky, 1994; Price, 1984).    Interview participants often identify access to physical resources as motivating collaborations. Physical resources could include capital-intensive pieces of equipment and infrastructure such as use of an electron microscope or mass spectrometer. Laura believes that access to the mass spectrometer that her lab owns and she runs motivates many individuals to collaborate with her lab. Access to this capital resource was particularly important when the lab was established because mass spectrometers were relatively rare. Because many collaborations center on this resource, most of Laura?s collaborations involve her running tests for others on the lab?s mass spectrometer. Another common physical resource that motivated collaborations appeared to be access to patients, either directly for clinical trials, to their samples for more basic research, or to their data for epidemiological work. In their study of 53 scientific collaborations in the physical sciences including high energy physics, geophysics, space science, oceanography, materials research, and medical physics, Shrum et al. (2007) found that the most common reason for collaboration was to access    126  physical resources, which can be seen as a facet of economic capital. This motive was echoed by Katz and Martin (1997). Physical resources can include specialised and often capital intensive equipment as well as access to data and different types of samples. Other participants discussed the importance of working with collaborators in order to increase the recognition or credibility of their work. Virgil argues that high end journals are increasingly demanding articles that take scientific research through a whole spectrum of activities from basic research to toxicity tests to clinical trials. Because no one researcher has expertise along this full spectrum, publications in these higher end journals often necessitate collaboration.  Crystal cites the advantages of working with one collaborator as "their expertise and their credibility." Access to prestige seemed to be particularly important for participants working in firms. Firms would try to publish articles with established academics in order to increase their firm's profile:  When you're a start-up and nobody really knows who you are or what the protein is that you're working on, for us it's a level of endorsement, right? People know who they are. They don't know who we are and so it increases viewership or interest. (Peggy).  Falling within Bourdieu?s concept of cultural capital, collaboration to increase popularity, visibility, and recognition had also been identified as a motivating factor behind collaboration (Dahlander and McFarland, 2013; Katz and Martin, 1997).  Collaboration is often based on access to multiple types of capital. For example, Virgil collaborates with a chemist who creates chemical components that Virgil's lab then tests. Through this collaboration Virgil gains access to an ensemble of research technologies including knowledge, techniques and physical resources.  Participants described how it is easier or harder to substitute one potential collaborator for another depending on what you are collaborating to gain access to. Collaborations centred on gaining proprietary knowledge were the hardest to substitute one potential collaborator for another while collaborators based on gaining access to a relatively widely available resource were generally more substitutable: "frequently companies have something proprietary that it's hard to get somewhere else...to go get patients for a study, I could probably have the top 300 [hospitals], and if number 5 is    127  being difficult, I'll just go to number 6." (Lane). Physical resources were often substitutable as more than one potential collaborator has access to the same physical resource. Other times, a physical resource was specific, being either proprietary or relatively rare. Because Bill's company had developed a proprietary platform technology, individuals often worked specifically with his company to gain access to their platform. The Genome Science Centre at BCCA provides high throughput sequencing not available in many locations, making its physical resource high in demand and often hard to access.  5.3 Sectors and capital Individuals working within a specific organisation or sector often had access to capital that other organisations and sectors did not. This specialisation often motivated collaborations between individuals working in different sectors "because they [different sectors] do different things." (Lane). 5.3.1 Healthcare Participants affiliated with hospitals had access to patients. Patients were used in clinician trials and patient samples were analysed in more basic research. Saul believes that because of improvements in instrumentation and diagnostics tools, research based on human samples had been increasing over the last 10 years both in the amount of research done and its relative importance. He thinks that basic research based on human samples is becoming popular enough to replace research based on animal models in some research fields: "the ability to do amazing things with human samples, technically, has opened up all sorts of avenues of research that simply weren?t there ten years ago." As shown in Table 14, participants involved in research who were physically located in hospitals ran the gamut from researcher to service/researcher to service to research support to student. Regardless of their role, participants' physical location in the hospital gave them access to patients in ways that other participants did not. This held even for participants in roles such as researcher or student that were not directly connected to patients through their work.     128  Participants in researcher roles who worked in hospitals describe how they benefited from their physical location by being able to gain access to patient samples for their research: "we have a tremendous opportunity here to have access to patient samples." (Sookie) For participants working in service/researcher roles and service roles their work in the clinic connected them with patients who could act as a resource for their research. Eric, for example, acted as a bridge between basic researchers who had tumour samples and linked, anonymised, clinical data that he had access to through his job. This link enabled researchers to connect specific characteristics of the tumours to specific types of cancer and/or results. Collaborations between firms and individuals affiliated with hospitals similarly appeared focused on patient access; "so they have access to patients I think is the primary benefit of working with them" (Lane).  Individuals who work in service and research/service roles and who also interact directly with patients were often able to bring intellectual insights to projects with regard to how disease presented in patients. Terry, who runs a diagnostic lab, and Crystal, an epidemiologist, involve physicians in their research when they need to know details of a specific clinical case such as how an infection physically manifested itself in a patient.  While most individuals in service/research roles affiliated with the BCCDC do not see patients, they are often able to access patient data. Because of its role in monitoring disease outbreaks for the province, BCCDC 'holds' key provincial communicable disease data that physicians are, based on the BC Public Health Act, required to report. Individuals working at BCCDC appeared able to trade their access to patient communicable disease data in much the same way as participants affiliated with hospitals traded their patient access.  5.3.2 Universities Participants in healthcare and industry discussed the specific capital that motivated collaborations with researchers physically located in universities. These include access to equipment, grant money, intellectual property, and individuals with the time and money to conduct early discovery work. As discussed above, academics often have access to particular capital intensive equipment that others did not. This appears to be particularly important for individuals working in healthcare organisations in service    129  or service/research roles with little or no research resources. These individuals can trade their access to patients and patient data with more basic researchers who have access to research equipment and expertise these participants, in a more service oriented role, lack: "I?m out interacting with the public and children, and drawing blood, and sticking vaccines into their arms, and recording all sorts of side effect data from that...A lot of the blood we draw has to go to a basic scientist to be analyzed" (Helo). William, who works in a lab run by a clinician-scientist, explained how his lab works with a nearby hospital to gain patient samples; "So he [the doctor who runs a clinic] is the gatekeeper, so you go through him to see any patient with [a specific disease] in BC." In collaborations with basic researchers William?s lab, in turn, becomes the gatekeeper for patient samples, providing basic researchers access to patient samples: "now the tables are turned because these are people [basic researchers] who don't have any access to patient samples." (William). Participants in service and service/research roles explained that the majority of their collaborations with more basic researchers occurred locally:  One question to ask me is whether I have collaborations with basic scientists. And the answer is, in my particular research field, not very many....whereas [in a centre that I'm affiliated with], the guy called [name], does a lot of basic immunology research that?s now tied into [my research area]...and so I?m able to collaborate with him onsite. Whereas, if his expertise was somewhere else in the country, that would be a bit more difficult. (Helo)   For others in government and hospitals, collaborations with academics gave them access to research grants from public funding organisations. Sam is unable to apply for research grants independently because of his official work title. Crystal and Terry felt that they are unable to apply for these research grants because only a portion of their work involved research: "with my kind of work, it?s very difficult to go after the big grants, because you have to commit so much time? it?s very, very tough, very, very tough to find that kind of time. You need [to be] full-time to do those things." (Terry). Collaborations in the network that focused on a local Cryptococcal disease outbreak serve to illustrate how individuals affiliated with government and universities can collaborate together. Crypotococcal disease is due to infection with spores of Cryptococcus gatti, a fungus that was previously associated with tropical and    130  subtropical areas until it emerged on Vancouver Island in 1999. Disease due to this fungus is relatively rare, but incubation can last several months resulting in pneumonia or meningitis.20 This collaboration began in 2001-2002 during a local outbreak.  Each collaborator brought their own strengths to this collaboration, strengths that were in part gained from their organisational affiliations. The role in this collaboration of individuals working in the BCCDC was to track cases of Cryptococcal disease within BC and get clinical isolates of the pathogen. This involved epidemiological work, mapping of outbreak locations and diagnosing clinical samples within a lab. An environmental hygienist at UBC collected and processed environmental isolates related to Cryptococcus gatti from the soil and air, complementing the clinical samples collected by BCCDC. Another researcher at UBC was able to conduct a molecular characterisation of both isolates within his research lab, expanding the understanding to the mechanisms underlying the infection. Both academic and clinical artifacts resulting from this work: several papers were published in scientific journals and BCCDC created guidance forms clinicians within BC could use in diagnosing and treating Cryptococcus gatti infection. All firms interviewed based their business at least in part on intellectual property initially developed in universities or hospitals which had then been acquired or licensed by the firm. These firms' very existence depended on their relationships with academics. For all of these companies, their initial development of this intellectual property had occurred as collaboration with the research lab that had developed the intellectual property, a collaboration that ended as development moved further towards application:  For the first year and a half, we actually still used [name] lab and [name] lab to do all of the internal research, because we were just an office space so we had no lab facility. And then at the time the company was about four persons, and then we grew to six, and at that point in time we decided okay, we want to take it out of the university hands. (Pete)                                              20 For more information see: http://www.bccdc.ca/dis-cond/a-z/_c/CryptococcalDisease/Cryptococcus+gattii.htm. Access February 15, 2013    131  For individuals in healthcare and industry, the focus of academics on understanding how things worked and tracing processes to their underlying mechanisms was an expertise that acted as a key resource in collaborations:  The MDs are really more "I'm treating the patient" and "I'm collecting the blood" and getting the patient's consent and things like that, but the PhDs are really the ones who are figuring it out...Even in the applied part, you don't know all the answers, and so the PhDs are working on 'What are the answers?' (Lane).   5.3.3 Firms Participants affiliated with healthcare or university organisations and who collaborated with industry described doing so to gain access to capital such as money or proprietary knowledge and physical resources. For example, Eric discussed how commercial clinical trials help to subsidise his other clinical trial work. Eric explained that he is involved in what he sees as three different kinds of clinical trials: small locally initiated trials, trials run by a national non-profit group, and commercial clinical trials. Small locally initiated trials are generally based on the treatment results of a specific patient and are used to see if these results are applicable to a larger subgroup. Often, there is no real money for these types of trials but organisational approval is needed to use the drugs for an unapproved treatment. Trials run by non-profit groups are generally multi-centre and involve per-case funding that normally is not enough to cover the full costs of the trial but enough to defray them. Commercial trials are for a specific drug company or contract research organisation and involve testing a specific drug. All clinical trials need a certain amount of infrastructure in place including clinical trial nurses, radiologists, pathologists, and so on. In general, the commercial clinical trials pay better than the other two types of trials and a certain number of commercial clinical are necessary to keep the clinical trial infrastructure in place: "so they [the two types of clinical trials] have to be kind of subsidized indirectly by the pharma trials" (Eric).  Helo took this idea of income generation even further, arguing that income from clinical trials "represents millions of dollars that could be brought into the BC economy...employing people to do the research" (Helo). Felix and Ben discussed collaborating with individuals in industry to gain access to specific and often proprietary    132  assays or other physical resources. Cedric discussed working with a pharmaceutical company on a project interpreting a specific genome that they were studying. More of a scientific collaboration, this relationship didn't involve any money or confidentiality agreements but rather access to data that had been collected by the company. Sookie, Virgil and Jason all discussed wanting to work with industry on projects because they believed that these collaborations provided the only means to move their ideas towards application:  I?m quite hopeful actually that we will have something that we can pass off to big pharma or maybe a biotechnology company, to ultimately take to the next stage.  Something that I feel is quite likely to ultimately succeed. So I think I would feel more satisfied if I was able to pass off something? pass on to something that ultimately was translated into a product that improved human health. (Jason)  Thus collaboration with firms was viewed as the pathway to translate academic research into a product. Four of the companies affiliated with participants were organised as virtual models where the majority of company work was contracted out to other organisations. This work was often contracted to other firms because "they have expertise that we don't have or they have infrastructure that we don't have." (Lane) Companies affiliated with participants worked with other firms for a variety of purposes ranging from data analysis to organising clinical trials to producing test products to certification to performing related tests. Peggy described working with organisations where personal connections previously existed. Lane contracted anywhere in North America or beyond based on price and quality. Intellectual property had similarly been outsourced. Companies had first been established and then looked for viable IP, related to a predetermined strategic direction of the company and specific criteria, to license. While some of these IP searches had been global in nature, all five companies licensed IP within Vancouver either from UBC or the hospitals. Lane?s company? X?serves an example of how firms collaborate with healthcare organisations and other firms. X is developing a genetic diagnostic that will help to predict the relative effectiveness of a specific drug when used in treatment. The company plans to bundle their diagnostic with this drug, offer the diagnostic for free, and take a share of the total sales of the drug based on the belief that the combination of the    133  diagnostic with the drug will produce more revenue than either alone.  X has set up an agreement with the company that owns this drug but the two companies are still separate. The diagnostic that they are developing is based on intellectual property developed in one of Vancouver's academic hospitals?Hospital Y?and first licensed, then acquired, through the UBC UILO office. X had also bought a patient database that they used in initial hypothesis testing from Hospital Y before beginning clinical trials. X?s diagnostic will be used to treat very sick patients, patients that are generally only found in academic hospitals. X has been working directly with academic hospitals throughout North America and Europe to conduct clinical trials. X has just completed a Phase 4 clinical trial that analysed the effectiveness of their diagnostic based on blood samples from 3,000 patients collected from 15 academic hospitals in the UK, France, Spain, US, Canada, Germany, and Switzerland. All samples were sent to a private lab in North Carolina to be genotyped. Kits used in these trials were made through contracts with other firms. If the Phase 4 trial results are favourable, they will be submitted to the FDA for approval so that X?s diagnostic can be listed on the label of the bundled drug. 5.4 Collaboration and the regulative institutional pillar As outlined in Chapter 4, institutions can both support and constrain action. This section explores the different ways that institutions that draw from the regulative pillar were perceived to constrain the collaborations of participants across and within sectors, as well as locally, nationally and internationally. Participants recognised that institutions could act as barriers to collaboration. Many institutions explicitly identified during interviews drew from the regulative institutional pillar. These institutions ran the gamut from pay structures to patient confidentiality and research ethics policies to the contract and licensing processes to organisational intellectual property rights policies to overhead rates to regulations related to cross-border collaborations. Helo and Donald identified the fee for service model used to remunerate many clinicians in BC as a barrier to working with clinicians. Under this model, clinicians are paid based on each clinical service rendered and research work is not directly compensated. Participants believed that this makes it difficult to motivate clinicians paid under a fee for service model to perform work that they would not necessarily be paid to    134  do on research projects: "anything that takes away from their patient flow in an office is money out of their pocket." (Helo)  Barriers related to conducting patient based research with hospitals and universities included policies focused on confidentiality, privacy, and research ethics. Confidentiality rules developed by specific organisations or REBs sometimes made it difficult to share patient data during collaboration (Crystal). Dealing with REBs, necessary for patient based research involving hospitals or universities, often involved long and complex administrative processes. In collaborations involving multiple organisations, multiple research ethics boards were often required to approve a single research proposal, each asking for different information in a slightly different form. Helo explains:  So you?ll have a multi-centre [clinician trial] project that has to go to each and every IRB [Research Ethics Board]. So there?s huge in-built delays. And then if one IRB says, ?Well, we want this to be changed.? that means that all the other IRBs have to consider the change. So there?s no coordination. Several barriers were identified in collaborations between industry and academia. Pete discusses how academic organisations were not designed to process the reams of paperwork necessary to submit a candidate drug or diagnostic to the FDA for approval, one of the reasons why collaborations between his company and academics focused only the initial discovery stages development:  All clinical work, you're governed by regulatory groups once you start getting close to man, even in your tox [toxicology tests]. And it's done by the FDA or the EMA, and these sites have to have good laboratory practice or good clinical practice. Universities have none of that, so that leaves the academic world behind.  The complex and formal contracts and licensing agreements that have developed that dictate interactions between academics and industry were identified as often constraining the relationships that they were initially designed to encourage. Universities were often perceived as driving hard bargains in negotiations with firms: ?many of them [universities] are much, much larger than your average biotech company and can be as aggressive or more aggressive than your average biotech company?they [universities] certainly behave as corporations and they?re not giving anything away to anybody.? (Bill) Bill, Samuel, and Pete believed that organisational IP    135  policies could act as a potential barrier during collaborations with biotechnology companies which are generally careful to secure their IP rights and may hesitate to work with organisations if they think that their IP ownership will not be protected: "some research organisations do things for fee-for-service. In Canada, that's a little bit harder at times. UBC always wants to have some portion of the pie." (Pete). The 25% overhead rate charged by UBC for university contracts was also perceived as a barrier making Vancouver less competitive compared to locations such as Central and Eastern Europe where overhead rates are generally believed to be lower. One participant identified restricted access to physical areas within a building as discouraging informal interactions between individuals, thereby decreasing the likelihood of collaborating. Collaborations with individuals affiliated with government organisations may require security clearance. Bill's company collaborates with a US government biosafety level 4 research lab. Biosafety level 4 research labs are used for work with dangerous and exotic agents that pose a high risk of individual infections and gaining access to this lab, as a Canadian company, was initially challenging.  Others collaborating with Americans had to deal with regulations related to sending samples across the border.  While participants discussed how institutions that drew from the regulative pillar could act as barriers to collaborations, most did not believe that these institutions stopped collaborations from happening between individuals: "the only reason I wouldn?t collaborate with them is if I didn?t like them or there was? you know, they did something to slight us. I would collaborate with everybody in the world if I could." (Gauis). Instead, institutions drawing from the regulative pillar sometimes acted as 'hurdles' or 'details' that needed to be worked out in order for a collaboration to be successful. In addition, institutions drawing from the regulative pillar could be strategically be used by individuals who didn?t want to collaborate as an obvious excuse for why the collaboration would not work. For example, Tommy talked about trying to gain access to data controlled by another group where that group felt that the collaboration would not help them achieve their own goals: ?where collaboration gets difficult, is when people I think feel that their interests are threatened or they might get criticized and they?re not so collaborative.? Instead of willingly collaborating, Tommy described how these potential collaborators used institutions that drew from the regulatory pillar such as    136  bureaucracy to create barriers against the potential collaboration, purposefully making things needlessly complex. Conversely, 'failed' or 'bad' collaborations weren't described as situations where institutions drawing from the regulative pillar and bureaucracy had caused the collaboration to fail. Instead, they involved instances of collaborators violating expected and standard scientific norms, institutions that would more clearly drew from the normative pillar. For example both Saul and Felix described failed collaborations as involving collaborators stealing scientific data and publishing it or doctoring results;  You send back the results and sorry there?s nothing there and they went ahead a published it anyways. Without the negative data...Not that they promised to do anything, but that?s just really bad science. There are lots of other people who you work with who would never think of doing that. So you just don?t work with people like that. (Saul)  For Lee and Laura, a failed collaboration involved a case not where the collaboration failed but where it failed to achieve the desired results: "if you?re going to actually be collaborating, usually the collaboration works, but the project won?t work." (Lee)  5.5 Sectors and the normative and cultural-cognitive institutional pillars  While often not explicitly identified as a barrier to collaboration, through participants? discussions it emerged that the true barrier to a successful collaboration was finding someone who not only had complementary capital but who believed that, through collaboration, they would be able to further their own goals using methods and means they deemed to be appropriate: "the hurdle is getting somebody who you know has got a resource that?s invaluable to change their mindset so it just doesn?t get wasted" (Gauis). An individual?s mindset relates to what they perceive as their goals and objectives. In order for a collaboration to be successful, in addition to providing access to needed capital, potential collaborators must view the collaboration as fulfilling their own goals and objectives; "there are challenges in making collaborations work, because they?re human endeavours where people have egos and different goals and other problems" (Virgil). As discussed in Chapter 4, institutions drawing from the normative pillar define an individual?s roles, related goals and objectives, and what they view as the appropriate ways of achieving these goals and objectives. Successful collaborations    137  involve a balancing act. On the one hand, two potential collaborators need to have access to contrasting and mutually desired types of capital; on the other hand, potential collaborators need to share institutions that draw from the normative and cultural-cognitive pillars.  From interviews, it appeared that participants working in different organisations and sectors often had divergent goals, hinting that sectors draw from different institutions. This led to a potential tension in collaboration decisions. Individuals affiliated with another sector had access complementary types of capital; however, these individuals were often influenced by different institutions drawing from the normative?and to a lesser extent the cultural-cognitive?pillars.   Some of the institutions drawing from the normative pillar manifested themselves in explicit goals written in organisational mandates. Others were more implicit. In general terms, participants affiliated with academic organisations discussed their primary goals as being to obtain publications and grants, participants affiliated with biotechnology companies discussed their goals as moving towards making an profit, , and participants affiliated with healthcare organisations discussed their goals as being providing service.  For participants strongly influenced by academic institutions their primary motivation was publishing papers and getting grants. The concept of publications as the ?currency? in the scientific community has been noted by others (see for example Haeussler and Sauermann, 2013). This was particularly true for those that ran research labs. Participants perceived grants to be awarded based on publication history and published papers thus acted as the ?currency? leading to ?profit? in academia. Academics with more publications and grants were perceived as more profitable. Academics running research labs could also go ?bankrupt,? forcing them to lay off their staff and shut down their research lab if they were unable to gain enough research money.  Loftier, more long-term goals often identified as academic institutions focused on furthering ?the science.? ?Science? here was equated with ?asking questions of nature and seeking answers? (Felix) and advancing knowledge in a specific field in a linear way. These longer-term goals appeared more strongly aligned with the cultural-cognitive    138  pillar, reinforcing the belief that scientists establish truths through the scientific method with the ultimate goal of advancing knowledge. In working towards this goal, many participants appeared to draw from the cultural-cognitive pillar in developing relatively informal relationships based on the premise of open knowledge sharing. These institutions, and related norms, goals, and beliefs about the nature of knowledge, appeared to dominate participants who viewed primary affiliation was with a university, irrespective of whether they worked in universities or hospitals.  For participants affiliated with biotechnology companies, their goals and motivations focused on making a profit by developing a product or service. A unique feature of these biotechnology companies was that they have not yet made a profit: ?so a company like ours, we?re a biotechnology company and biotechnology is an interesting business model in that we are funded by investors who are really investing in the promise of the technology.? (Bill)  Instead, the shorter-term goal of these participants focused on developing their product or service to a stage either where it could make a profit or where another company became convinced that their organisation could eventually make a profit and buy the company. To achieve this goal, it was important for companies to develop products relatively quickly and cheaply while maintaining quality. Based on a ?promise? of future profit, affiliated biotechnology companies needed to remain legitimate.  Drawing from the cultural-cognitive pillar, development work focused on understanding how things worked through the scientific method and applying this understanding to creating a working product or service. In working towards these goals, participants appeared to build from the cultural-cognitive pillar in developing formal relationships based on well-defined IP ownership policies. Participants across sectors perceived that the primary goals of clinicians were to treat patients. Clinicians were not perceived as particularly interested in discovering why something worked but only that it would work effectively in treating their patients. From discussions with participants, it appeared that very few ?pure? clinicians collaborated with participants on research related work. Participants involved in service/research and service work collaborated with clinicians in their service work but generally had limited    139  interactions with ?pure? clinicians in their research. While some of the firms interviewed used patients recruited from ?pure? clinicians in their clinical trials, these interactions were mediated by contract research organisations. When asked why they didn?t work with ?pure? clinicians, many participants explained that clinicians were too busy; ?their [clinician?s] priority is to treat people and to, you know, get them out the door, and work within that system.? (William) It appeared as though most clinicians did not view research collaborations as helping them achieve their goals. There were some exceptions. Often in these cases clinicians would work with researchers providing samples and clinical insight for research projects but would not be actively part of the project. These relationships appeared to work best when participants worked with clinicians who treated patients with fatal conditions, such as cystic fibrosis or HIV/AIDs, where no cure currently exists. Participants believed clinicians were interested in collaborating on such projects because they hoped the project would help improve the clinician?s ability to treat their patients: ?it's not a great disease, it's a terrible disease, and they want to do as much as they can to help these patients who have this disease.? (William). Because most participants were unable to find ?pure? clinicians interested in collaboration they worked instead with individuals physically located in healthcare organisations in order to access patient based resources. As discussed above, participants in researcher roles that were physically located in healthcare organisations viewed their primary affiliation as with a university. These participants drew from academic institutions. Participants who worked within research/service roles and who ran research labs?such as Virgil, Ben and Hoyt? also appeared to predominately adopt academic institutions and related norms and goals in their research work. This may be because publishing papers and getting grants was necessary to keep their lab running: [My goals are] getting grants so we can keep the lab open, writing papers so we can be successful in getting our grants... if you asked me the simplest equivalence of my goal for the next six months, then it?s to secure a CIHR grant and to review another grant. That?s the goal. It?s not to make sick kids healthy. (Virgil)     140  Even while academic institutions dominated their research goals, these participants? service roles still appeared to shape how they viewed research problems and prioritised questions, helping them to bridge the service and research institutions that influenced their different roles: ?it [my clinical work] affects how I think about my research and the ideas that I have and what I think is interesting and important.? (Ben). Others in researcher/service and service roles?such as Crystal, Lafayette, Sam, and Terry?appeared to balance academic and service institutions more, perhaps because they did not rely as heavily on grants to conduct their research. Participants located within healthcare organisations drew on academic institutions to a greater or lesser extent. Participants physically located within hospitals were able to provide university-based academic collaborators with an otherwise unavailable and desired resource: patient samples. Because they drew from academic institutions, these participants also shared institutions with their university-based collaborators. The ability of participants in healthcare organisations to draw on academic institutions may have enabled them to collaborate more easily with individuals located in universities, helping to resolve the tension between accessing contrasting capital and sharing institutions that draw from the normative and cultural-cognitive pillars through collaboration.  All participants working within healthcare organisations had a connection to both the healthcare organisation and to a university. The relative importance of these two connections varied from participant to participant. As discussed in Chapter 4, organisations within the Ministry of Health work under a mandate to administer access to medically necessary hospital and physician services. Other mandates, existing at the departmental, organisational, or health authority level often included research. Within the health authorities, tensions between the mandates of research and service appeared to cause decoupling (Orton and Weick, 1990; Thompson, 1967) within the system related to individual action, roles, and funding for individuals in service and service/research roles. For example, institutions drawing from the cultural-cognitive and normative pillars often appeared to encourage participants to conduct research while participants? salaries, drawing from the regulative pillar, often paid the same participants only to see patients. Historically, a certain ?wiggle room? within the system enabled individuals to navigate these institutional tensions and conduct research ?off the side of    141  their desk? in addition to clinical care. As healthcare budgets tighten, causing a constraint on resources, and as centralisation within the health authorities continues, participants discussed how this ?wiggle room? appears to be decreasing, leading to a potential increase in tensions between academic and service institutions within healthcare organisations. French and Miller (2012) similarly found tensions between innovation and care mandates Toronto, Canada based hospitals. The BCCDC acted as an example of how service and research institutions can be bridged. Research has been explicitly embedded into BCCDC?s service role. Organisationally, BCCDC has expanded its mandate beyond service to include research and teaching (Strategic Plan, 2007). Research done by individuals at BCCDC runs the gamut from epidemiological to bioinformatics to environmental to translational based on patient samples to 'wet' labs based on cell lines and animal models. While research is part of the BCCDC mandate, professionals are not necessarily explicitly given time that is set aside for them to conduct research. Instead, they are offered a certain degree of flexibility combined with an implicit expectation that research will be part of their work:  It?s kind of hard to explain?but our peers do research and publish. Our leader has done a lot of research, has published a lot, and there?s pressure from all of that?indirect from peers but direct from our leadership?to publish. (Crystal)  BCCDC appears to have been successful in combining research and service in part by encouraging their staff to build their research off of their service work:  All of our professionals have large service roles, but the critical thing is to use the problems of your service delivery, and create those problems into a researchable question and then articulate that as a research proposal, raise the funds, acquire the collaborations, generate the finding and then solve the problem. (Jason) As such, much of the research done at BCCDC appears, to a certain extent, to be convenience based, taking ideas that were collected for service purposes a little bit further into the research realm. Tensions can exist in the expectations, and support for, research and service goals at BCCDC. While there is flexibility in work schedules in order to do some research, there is limited monetary support.     142  5.6 Discussion Through my interviews, I learned that while participants do not perceive that organisation specific institutions affect collaborations, descriptions of actual collaboration practices imply that they do. Supporting the contentions of Bourdieu (1986) and Giddens (1984), participants appeared to collaborate to gain access to complementary capital in order to further their own goals within an organisational field structured by institutions. Participants identified capital that formed the basis of collaborations as broadly falling into Bourdieu?s (1986) multiple forms of capital including economic, human, cultural, and social. Capital that formed the basis of collaboration often included capital within a unique ?ensemble of research technologies? (Hackett, 2005) comprised of materials, techniques, instruments, and ideas that define scientific research.  Table 15 outlines some of the institutions identified through participants and the institution?s dominant pillar. Institutions strongly associated with the regulative pillar such as IP law, academic tenure policies, and confidentiality agreements formed obvious incentives and barriers for collaboration. These institutions were also purposefully drawn on to create barriers against collaborations that an individual did not want to be part of. However, institutions drawing from the regulative pillar were relatively easily surmounted if two individuals wanted to collaborate. Institutions strongly drawing from the normative and cultural-cognitive pillars influenced participants? roles and goals as well as accepted ways of achieving these goals. It was these institutions that more strongly influenced collaborative behaviour than institutions drawing on the regulative pillar. Collaborations that were described as ?failed? were collaborations that violated institutions drawing from the normative and cultural-cognitive pillars, involving individuals that violated scientific norms related to the scientific method and plagiarism or that produced ?bad? scientific results. Successful collaboration thus involved a balancing act. Potential collaborators needed to have access to desired and different capital. Potential collaborators also needed to share institutions that drew from the normative and cultural-cognitive pillars or develop collaborations that fulfilled both collaborators? goals. These institutions often varied by sector.     143  Table 15: The three pillars in I2 organisational field Regulative Normative Cultural-Cognitive IP law Organisational mandates Organisational ethos Academic tenure policy Strategic plans Scientific method   REB policies Perceived job expectations Medicine as a craft FDA approval process Career goals White lab coat Security clearance rules Definition of roles Knowledge as shared Confidentiality rules  Expected proper conduct Knowledge as proprietary Fee for service Scientific norms Understanding why things work Contracts  Clinical norms Using things that work Material transfer agreements Making things work  The profit motive has been identified as spurring collaborations within both the innovation systems and neo-institutional literatures (DiMaggio and Powell, 1983; Nelson and Winter, 1977). Within my own analysis, participants affiliated with biotechnology companies collaborated based upon a profit goal. Participants strongly influenced by academic institutions, to a certain extent, were also focused on generating profit within publication and grant-based currencies. Haussler and Sauermann (2013) similarly argued that publications act as an academic currency. DiMaggio and Powell (1983) identified a second non-profit motivation based on legitimacy, a motivation that they argue takes on greater importance with non-profit players. This motivation was also identified in my study. Legitimacy also motivated collaboration by participants affiliated with biotechnology companies. Here, collaboration with academic researchers focused on publication was perceived as increasing firm legitimacy, an important trait for companies that are often not yet profitable but instead rely on creating the ?promise? of one day generating a profit.  Owen-Smith and Powell (2004) argue that public sector organisations generally work under a more open model of knowledge sharing while the private sector works under a more closed knowledge sharing model. These conjectures were echoed by participants who generally collaborated with other academics through relatively informal relationships with open knowledge sharing and with the private sector through formal contracts with explicit intellectual property clauses. This hints at divergent dominant institutions in the public and private sector drawing from the cultural-cognitive pillar.     144  Dominant sectors of an organisational field create a dominant governance systems and institutions within organisational fields (Scott, 2008; Owen-Smith and Powell, 2004). Within Vancouver?s I2 organisational field, academic institutions appeared to dominate. Some institutions related to service work were incorporated into this organisational field but their influence appeared to be relatively muted. Firms, and their related institutions, similarly played a relatively weak role. For this organisational field to effectively address the needs and goals of all sectors, academic institutions need to become less dominant. This would not just increase the importance of the private sector within the field but also the importance of healthcare organisations. While participants affiliated with healthcare organisations play a strong role within this organisational field, service based needs and goals do not appear to be strongly incorporated into the current system. Beyond patient resources, individuals affiliated with healthcare organisations need to incorporate service-based norms, values, and goals more strongly into their research work.  The importance that patient samples played as a desired form of capital connected to the healthcare sector echoes the findings of French and Miller (2012) who argued that, within Toronto, Canada, patient populations were seen as distinctive hospital based resources and that hospitals acted as ?obligatory passage points? in biomedical innovation, brokering access to patient populations and care infrastructure. In connecting researchers with patient resources, individuals affiliated with healthcare organisations were able to act as what Burt (1992) calls ?tertius gaudens,? individuals who gain power by bringing groups of people together who would otherwise not interact. Geographic proximity appeared to be important for this ?tertius gaudens? role. The emphasis in interviews of both physical presence of researchers in hospitals to gain access to patient samples and proximity between service and service/researchers with researchers for increased success in collaborations ran counter to the belief of participants that geography played no role in collaborations. It is unclear why connections based on patient samples between participants located in healthcare and university organisations appeared so localised, although several factors likely contribute to this trend. Geographic proximity may provide a necessary connection that enables individuals to transcend potential institutional differences. Individuals working in more    145  service based roles may be unable to attend conferences and meetings, which are a traditional way for academics to meet, in the same way that academics can. Local connections may also be necessary to facilitate the use to patient samples, which might be more difficult to access and transport long distances. The same reliance on geographic proximity to establish service/academic connections does not seem to exist for connections between industry and service. Focused predominately on clinical trials, these connections often require patient numbers that are impossible to get in only one location and instead span Europe and North America, often facilitated through contract research organisations (CRO): "so we'll pick one CRO and they'll be in charge in an umbrella fashion to run the study, to monitor it, and to make sure the quality's there." (Pete). Distinct institutions influenced individuals affiliated with the private sector and academia and these distinctions delineated clear and separate research approaches within the two spheres. Institutions within healthcare did not appear to similarly differentiate between academic and healthcare approaches to research. Instead, academic institutions dominated the research practices within the organisational field?s public organisations. Regulative academic institutions?such as the UBC UILO and REBs? were applied to research conducted within healthcare organisations; participants affiliated with healthcare organisations discussed often adhering to academic institutions when conducting their research. For individuals affiliated with healthcare organisations, health specific institutions governed their service roles and academic institutions often dominated their research. Perhaps because a well-defined healthcare-based approach to research did not exist, ?pure? clinicians were not perceived as actively involved in research. The influence of academic institutions on research based in healthcare organisations may be an instance of boundary work and demarcation (Gieryn, 1983) as UBC works to ?claim? research based in healthcare organisations as its own. Adhering to academic institutions while conducting research may also be an effective way for individuals physically located within healthcare organisations to decouple their research and service work. Because research is not part of the healthcare mandate, individuals conducting research within healthcare organisations can resolve potential institutional tensions by choosing to only follow    146  academic institutions when conducting research. Multiple affiliations played a complex role as a boundary-spanning institutional structure between healthcare organisations and universities. These multiple affiliations meant different things to different people. The purpose of creating multiple affiliations as a boundary spanning mechanism appeared strongly focused on clinical education rather than on translating healthcare needs to academic researchers and academic research findings into the healthcare setting.  5.7 Conclusion Figure 9 shows that despite the global reach of this organisational field, participants appeared relatively well connected locally; 75% of the participants were either directly connected to other participants or were connected together through Vancouver affiliated co-authors. I found that many participants believed that collaboration is essential for their work. While participants do not perceive that organisation specific institutions affect collaborations, these institutions often do. Participants disagreed on how important personality was in shaping collaboration decisions. Many of the collaborations described by participants involved complex decades-long relationships. Participants collaborated to gain access to capital such as knowledge, techniques, physical resources, and status in order to further their own goals. Some of this capital, such as proprietary knowledge or prestige, were more specific vis a vis who was chosen as a collaborator; other types of capital, such as money or infrastructure, were less specific. This finding supports the conjecture of Giddens (1984) and Bourdieu (1986) that collaboration involves accessing the capital of another individual within existing structures. Capital identified in interviews fell into Bourdieu?s (1986) concept of multiple forms of capital including economic, human, cultural, and social capital. Collaborations appeared to be successful if they were able to bring complementary capital and achieve the goals of all parties. The capital available to an individual, as well as their perceived roles and goals, appeared to be influenced by organisational structures, sectoral affiliations, and related institutions.     147  Individuals affiliated with universities were perceived as having access to capital such as expensive research equipment, funding for conducting research, intellectual property, and expertise in specific experimental techniques and in the scientific method more generally. Normative institutions shaped the goals of university-affiliated researchers towards publishing articles in peer reviewed journals often with the objective of gaining more research grant money.  Cultural-cognitive institutions encouraged university-affiliated researchers to achieve these goals under a relatively open knowledge system focused on understanding how things work through the scientific method.      Healthcare organisations were perceived as having access to capital related to patients and their data as well as expertise in how patients present a disease and react to treatments. Normative institutions shaped the goals of individuals affiliated with healthcare organisations towards fulfilling their service role, which was often focused on treating patients. Cultural-cognitive institutions encouraged these individuals to achieve these goals under a relatively open knowledge system focused on using things that work. Individuals affiliated with biotechnology and pharmaceutical companies were perceived as having access to capital such as money, proprietary knowledge and artifacts, and expertise in moving ideas to commercial application. Normative institutions shaped the goals of these individuals towards moving a product towards development quickly while keeping costs low and quality high. Cultural-cognitive institutions encouraged individuals affiliated with biotechnology and pharmaceutical companies to achieve these goals under a closed knowledge system, applying enough understanding to problems in order to make them work. Scott (2008) argues that tensions can exist between institutions. While, both firms and universities draw from a cultural-cognitive institution based on the scientific paradigm, tensions exist between the sectors related to a culture of open versus closed knowledge. Within healthcare, a cultural-cognitive tension exists between viewing healthcare as a craft or as a science. Normative institutional tensions exist for many participants affiliated with healthcare organisations between their service role and    148  research expectations these have led to role ambiguity and organisational decoupling for some participants working in service and service/research roles. Regulative institutions create obvious incentives and barriers to collaborations but appeared relatively easily surmounted or ignored by participants. Normative and cultural-cognitive institutions influenced participants? goals and their conceptions of the proper ways of achieving these goals, thereby fundamentally influencing their motivations to collaborate. Less obvious and more entrenched, normative and cultural-cognitive institutions appeared to influence collaborations more than regulative institutions.  The importance of long-term relationships as forming the basis of many of the collaborations described by participants appeared as a theme in this research.  This supports the findings of other researchers that successful relationships are often built on multiple facets including trust, fine grained information transfer and joint problem solving that create an investment in the relationship, making its dissolution more challenging (Uzzi, 1997). Others (Dahlander and McFarland, 2013) have argued that the processes of tie formation and tie persistence may differ. My analysis appeared to support the contention that tie formation and tie persistence processes differ, but this was not a topic extensively explored in my analysis. Further work needs to be done to understand how these processes differ, and how institutions affect tie formation and tie persistence.   By focusing on interactions within Vancouver?s I2 organisational field, this chapter explores the strengths and weaknesses, the incentives and the barriers, of interactions within this organisational field. This has implications not only for this particular case study, but for other biomedical organisational fields. My work offers a sobering conclusion for policy makers interested in promoting clusters in high technology areas such as biotechnology. Despite their own perceptions, participants appeared less likely to collaborate across sectors. These intersectoral partnerships are often viewed as a prerequisite for a successful innovation system. The institutions that most strongly affected collaboration decisions were normative and cultural-cognitive, institutions that are challenging to influence directly through policy.      149  6 Chapter 6: Discussion and conclusion 6.1 Introduction Research collaborations are increasingly important for scientists and science and technology policy. Collaborations between and within sectors facilitate R&D by transferring knowledge between individuals and their affiliated organisations. This transfer of knowledge, in turn, drives innovation. My dissertation described and analysed collaborations between and within the sectors of Vancouver?s I2 organisational field. I applied theoretical perspectives from the IS framework, economic geography, neo-institutional theory, and social networks to my study Vancouver?s I2 organisational field. I focused on a subset of actors within the IS framework: researchers and their affiliated organisations. Other organisations and individuals, such as government policy makers, financial bodies, and market forces, were considered only insofar as they create institutions that affect an individuals? ability to conduct collaborative R&D.  I explored the three research questions: 1. How do different types of proximity affect collaboration? 2. What reasons do people give for collaborating? 3. How do institutions such as policies, norms, and organisational culture affect collaboration? These questions were addressed through a mixed methods study of Vancouver?s I2 organisational field. Building from the insight that systems of organisations are social networks that transfer information, I drew from a social network perspective to map co-authorships of Vancouver?s I2 organisational field. I interviewed a subset of these researchers to increase my understanding of their work, research, reasons for collaborating and the effect of institutions on these factors.  In using a mixed methods approach, I brought quantitative and qualitative components of my study together in a multi-level analysis that mapped local, national, and international collaborators in Vancouver?s I2 organisational field and analysed individual?s perspectives on how work and collaborations are influenced by institutions.    150  6.2 Summary of main findings I used quantitative bibliometric analysis in Chapters 2 and 3. In Chapter 2, I investigated my first research question: How do different types of proximity affect collaboration? I explored geographic and institutional proximity through an individual-level analysis of I2 co-authorship rates based on a quasi-Poisson random effects regression. Geographic proximity was operationalized through regional coding and binary distance radiuses centred on Vancouver. Co-authors with shared organisational, sectoral and national affiliations were coded as institutionally proximate. I found that collaborations were more likely to occur among geographically close individuals. Relative distance mattered most between driving and flying radiuses, while differentiation between short and long haul flights made little difference. Institutional proximity increased proclivity to co-author. Individuals who were both institutionally and geographically proximate showed the greatest proclivity to collaborate. Geographic proximity compensated for institutional proximity and vice versa. Unexpectedly, a large proportion (almost 25%) of authors within this network was affiliated with more than one organisation. I observed regional variation in multiple affiliation rates. In Chapter 3 I further analysed sectoral co-authorship patterns. I investigated sectors? relative importance and tested if organisations affiliated with different sectors co-authored more or less than expected. I did this through an organisation-level analysis of I2 co-authorship rates based on a relational contingency table and ANOVA network analyses. I found a marked preference for organisations to collaborate within, as opposed to across, sectors. This trend was most pronounced in hospitals. All sectors show a relatively high preference for collaborating with universities, a trend that was most marked in hospital-university co-authorships.  Combining results from Chapter 2 and 3, I found that universities dominated the network both in number of affiliated co-authors and organisations. The private sector and NGOs played a relatively weak role in the network. The results of Chapter 2 and 3 diverged slightly. In Chapter 2, based on both the number of individual-level and organisation-level sectoral affiliations, the government sector ranked second after the university sector. In Chapter 3, based on the number of organisation-level sectoral    151  affiliations, the hospital sector ranked second after the university sector. The datasets used in the two chapters differed. The Chapter 3 dataset included more years and fewer exclusion criteria and was approximately 65% larger than the Chapter 2 dataset. This variance in dataset size likely accounts for differences in the sectors? rank order and number of organisation-level sectoral affiliations between the two datasets. However, the proportion of organisations affiliated with each sector over total number of organisations varied by less than 3.5% between the two datasets.  In Chapter 4, I used mixed quantitative bibliometric and qualitative interview and document analysis to outline the main organisations and dominant regulative institutions within Vancouver?s I2 organisational field. Non-commercial organisations dominate this organisational field. The majority of government organisations fall under the BC Ministry of Health and its six health authorities. Universities had the largest number of affiliated authors followed by government organisations. Firms had the largest number of organisations but the smallest number of authors. Since 2008 private sector presence has fallen and biotechnology companies have adopted a virtual or semi-virtual organisational model. I observed that organisations and formal institutions have developed to encourage collaborations across sectors and organisations. These include affiliation agreements between UBC and the health authorities, local UILOs, CDRD, and CFRI.  I used qualitative interview and document analysis in Chapter 5 to address my second and third research questions: What reasons do people give for collaborating? How do institutions such as policies, norms, and organisational culture affect collaboration? I drew on Scott?s (2008) three pillars of institutions?regulative, normative, and cultural-cognitive?and?based on a theme that emerged from my analysis?combined Scott?s concept of institutions with Giddens? (1984) and Bourdieu?s (1986) concepts of capital. I found that researchers collaborated to gain capital necessary to achieve goals according to the methods and means they deemed appropriate. Although many researchers explicitly stated that organisational affiliation played no role in collaboration decisions, I found that organisational affiliations implicitly shaped collaboration choices. Organisations and sectors provided individuals with access to    152  distinct capital and organisation-specific and sector-specific institutions shaped individuals? goals as well as the their accepted methods and means for achieving these goals. These institutions drew from the normative and cultural-cognitive pillars.  Participants perceived that individuals with university affiliations often had access to capital in the form of expensive research equipment, intellectual property and expertise; they pursued goals related to publishing and securing grants; and they worked in a system that encouraged relatively open knowledge based on the scientific method. Participants perceived that individuals affiliated with healthcare organisations had access to capital such as patients, patient data, and clinical expertise; they pursued goals related to their service roles and patient treatment; and they worked in a relatively open knowledge system that focused on using things that work. Participants perceived that individuals with firm affiliations had access to capital such as money, proprietary knowledge, and expertise in moving ideas to application; they pursued goals that emphasized speedy product development while keeping costs low and quality high; and they worked in an environment that encouraged a closed knowledge system that applied enough understanding to problems to make them work. Because participants physically located in healthcare organisations also had university affiliations, both healthcare and academic institutions influenced their actions; the relative importance of healthcare and academic institutions varied by individual. While regulative institutions could act as barriers to collaborations, these barriers were perceived to be easily surmounted. Rather, it was institutions drawing from the normative and cultural-cognitive pillars that influenced goals, motivations, and approaches, and acted as stronger barriers to collaboration.    6.3 Cross cutting themes 6.3.1 Proximity Combined, my results chapters paint a complex, nuanced, and sometimes contradictory picture of Vancouver?s I2 organisational field. The French School of Proximity Dynamics argues that multiple dimensions of proximity increase the likelihood of collaboration, an assertion conceptually similar to homophily, a term used in social    153  network analysis (Ponds et al., 2007; Powell et al, 2005; Boshcma, 2005; Torre and Rallet, 2005; McPherson et al, 2001). My own analysis demonstrated qualified support for this multi-dimensional understanding of proximity. Quantitatively, institutional proximity and geographic proximity increased co-authorship. The importance of geographic proximity supports previous regional innovation, cluster, propinquity, and institutional analyses (see for example Ponds et al, 2007; McPherson et al, 2001; Braczyk et al, 1998; Porter, 1998; Cooke et al, 1997). Qualitative analysis painted a more nuanced picture. While participants explicitly discounted institutional and geographic proximity as factors in collaboration decisions, both variables appeared to implicitly shape participants? choices. Participants believed that collaboration   provided access to complementary capital as a means of achieving goals in appropriate ways. The idea of collaboration as a means to gain access to complementary capital supports previous analyses of individuals? motivations to collaborate (see for example Dahlander and McFarland, 2013; Shrum et al, 2007; Katz and Martin, 1997; Bourdieu, 1986; Giddens, 1984).  Scott?s (2008) framework of regulative, normative, and cultural-cognitive institutional pillars, combined with the concept of capital, fit well with my qualitative findings and help to elucidate motives behind my quantitative finding that individuals appear more inclined to collaborate intrasectorally. Sectors and organisations often subscribed to different institutions. Institutions drawing from the normative pillar resulted in divergent perceived roles and goals between sectors. Institutions drawing from the cultural-cognitive pillar lead to varying perspectives of the openness of knowledge, expertise, and acceptable methodologies.  While participants affiliated with different sectors may have access to diverse and complementary capital, their sectoral affiliations often influenced their goals and accepted ways of doing things, leading to divergence and influencing collaboration decisions. Individuals affiliated with the same sector often shared goals and epistemological standards making collaboration easier.    Boschma (2005) argues that geographic proximity can compensate for a lack of institutional proximity and vice versa. My quantitative analysis in Chapter 2 showed support for this hypothesis; similarly, in Chapter 5, the individuals interviewed    154  underscored the importance of geographic proximity in facilitating specific types of collaboration. These included collaborations by academic researchers focused on gaining access to patient samples and data; collaborations where individuals in service roles in healthcare worked with individuals in research roles in universities; and collaborations between firms and academics focused on licensing or acquiring intellectual property that formed the basis of a product?s development.  6.3.2 Scarcity and prestige  Factors such as scarcity and prestige appeared to strongly influence collaboration decisions. In Chapter 2 I conjectured that it is important to differentiate resources available at multiple locations from relatively scarce resources. When resources are available in more than one place, researchers may pick the location that is proximate to them to collaborate. However, when a resource is scarce, a researcher may collaborate to gain access to a resource, irrespective of proximity. These insights relating proximity to resource access move us towards a more coherent understanding of the effect of both proximity and resource access on collaboration decisions. In Chapter 2, I proposed that co-authorships between Vancouver-based researchers with, first, American firms and, second, African collaborators could be viewed in this light. Collaborations between Vancouver-based researchers and American firms may help the former gain American market approval by creating institutional proximity by proxy with the USA regulatory environment within which these American firms are already embedded. This echoes the findings of Rees (2005) that Vancouver-based biotechnology firms predominantly collaborated with American firms and these collaborations aided in decreasing cost and time of R&D by allowing the firms to access the experience required to clear regulatory hurdles. Co-authorships within Africa were predominately located in Uganda and Kenya and focused on HIV/AIDs. As argued in Chapter 2, these collaborations could be viewed as providing Vancouver-based researchers with access to another relatively rare resource: HIV/AIDs patients.21                                             21 Vancouver?s Downtown Eastside has high HIV/AIDs infection rates that are estimated to be roughly equivalent to infection rates in Botswana and to reach 30% among intravenous drug users (Providence Health Care, 2009). Through the work of Vancouver?s Centre for Excellence in HIV/AIDs, doctors, and    155  Interviews with participants reinforced the findings in Chapter 2 that scarcity plays in shaping collaboration decisions. In Chapter 6, I proposed that access to proprietary knowledge and capital assets, such as specialized laboratories and large-scale equipment, could be considered as access to relatively scarce resources. Because of the value placed on IP in the private sector, firms were generally involved in collaborations that expanded or utilized their IP base, sometimes licensing or acquiring proprietary knowledge from universities, hospitals, or other firms, or granting access to their own proprietary knowledge, which was often held in the form of assays, drugs, or specialised research equipment. In relation to capital assets, study participants indicated the importance of access to specialized laboratories and capital equipment, such as those at the Genome Sciences Centre at BCCA, assets which are not available in many locations. In Chapter 2, I also suggested that factors such as prestige might influence collaboration decisions. I hypothesized that the large number of co-authorships observed between Vancouver-based researchers and those in the northeast of North America could potentially be viewed in this light, as the northeast is recognised as a cluster of academic excellence. This hypothesis is supported in the interviews where participants discussed finding appropriate collaborators based on collaborators? prestige. This factor may be more important for researchers who are not yet established in their careers and feel the need to gain prestige through association. For example, one participant who discussed the necessity of working with ?the best? was a relatively junior academic researcher. The increasing expectations of higher end scientific journals that article submissions include a spectrum of activities from basic research to toxicity tests to clinical trials may also necessitate collaboration. Prestige also appears to be a key motivation for individuals working in the private sector who actively form connections with academics in order to add prestige to their project and borrow academic legitimacy. This falls within Bourdieu?s (1986) concept of cultural capital and supports the findings                                                                                                                                             community organisations, Downtown Eastside HIV/AIDs patients are involved in intervention and treatment programs. Some interview participants discussed perceived research fatigue among Downtown Eastside HIV/AIDs patients. Conversely, HIV/AIDs patients in Africa are often not part of active interventions and treatments and likely have not experienced research fatigue. This makes them appealing research subjects.    156  of other researchers that visibility and recognition can motivate collaborations (Dahlander and McFarland, 2013; Katz and Martin, 1997).  6.3.3 Vancouver?s I2 organisational field The relative status of different organisations and sectors within an organisational field can affect the organisational field?s overall tone (Emirbayer and Johnson, 2008; Scott, 2008; Owen-Smith and Powell, 2004). Vancouver?s I2 organisational field is dominated by the public sector. Throughout the preceding chapters I have drawn attention to a key finding of this study, which is the relatively weak presence of both the private and NGO sectors within Vancouver?s I2 organisational field. Others have similarly pointed to a weak role played by the private sector in Canada?s R&D (McFetridge, 1993). A key finding in my research was that hospitals and government organisations, and their connection to universities, dominate Vancouver?s I2 organisational field. This finding reinforces the concept of the hidden research system (Hicks and Katz, 1996) discussed in Chapter 3. In the IS literature, commercial organisations are the innovators and commercial goals motivate all action including collaborations: innovation is profit-driven and takes place in a market-based competitive environment (McKelvey, 1997; Lundvall, 1988; Nelson and Winter, 1977). These commercial institutions appeared to similarly motivate the private sector but not the public sector within Vancouver?s I2 organisational field.  The public sector?s domination within this organisational field could be based in part on the location chosen for this case study. In Chapters 2 and 3 I found that firms located outside Vancouver were more active in the network than local firms. This finding reinforces comments by interview participants that they perceive Vancouver?s biotechnology sector to be relatively weak, particularly since the Great Recession.  However, the study provides useful insights into how an organisational field dominated by non-commercial institutions operates, particularly in biomedicine and health. These findings are important because in Canada, as in much of the Western world, the majority of medical treatment and care is provided by the public sector, which not only produces biomedical innovations but which also acts as their final user (Windrum and Garcia-Goni 2008; Gelijns et al 2001).      157  Within the network, those working in a healthcare organisation were directly connected to a university through affiliation agreements. Based on these agreements, a variety of institutions drawing from the regulatory pillar focused on research-related administrative procedures?including ethics reviews and intellectual property rights?were administered through UBC.  The meaning of their university affiliation varied among participants along a spectrum. Some viewed themselves as university employees; their physical location in a healthcare organisation had little meaning. Others saw themselves as hospital or government employees, for whom a university affiliation was essentially meaningless. These differences in perception appeared, to a large extent, to revolve around day-to-day work patterns and formal and informal roles. Thus, for some of the participants, it would be valid for analytic purposes to incorporate academic hospitals into universities, a practice sometimes used in innovation analyses, whereas for other participants the distinction remained valid.   6.4  Importance of work My dissertation provided several novel empirical, methodological, and theoretical insights. First, I theoretically and empirically integrated individual?s perspectives and organisational field structure through a multi-level mixed methods approach. Second, I provided novel insight into how multiple often conflicting institutions within an organisational field interact and influence individuals. Third, I conducted a large-scale  bibliometric analysis using individuals as the unit of analysis that addressed a methodological shortcoming of many similar analyses that used organisations as the unit of analysis.  An identified shortcoming of many previous neo-institutional studies was that they took organisations, as opposed to individuals, as a unit of analysis (Powell and Colyvas, 2008; Scott, 2008; Powell and DiMaggio, 1991). Scholars argued that incorporating Gidden?s (1984) theory of structuration or Bourdieu?s field theory (1986, 1985) into neo-institutional theory provided a theoretical bridge that connected individual agency, organisational field structure, and institutions (see for example Scott, 2008; Emirbayer and Johnson, 2008; Berends et al, 2003; Barley and Tolbert, 1997). Incorporating Giddens (1984) and Bourdieu (1986) into neo-institutional theory expands    158  the theory into multiple levels that connect individuals to institutions and organisational field structures. Through mixed methods, my dissertation successfully spanned these multiple levels of analysis and connected individual?s perspectives to organisational field structure and institutions. I know of no other empirical neo-institutional study that employed this multi-level approach. This innovative approach helped validate and expand components of neo-institutional theory and contributed novel empirical insights.  I empirically showed how collaborations are based on interplays between institutions and available capital. This theoretically connected Scott?s (2008) three institutional pillars to Bourdieu (1986) and Giddens (1984) concept of capital and validated the explicit incorporation of Bourdieu and Giddens into neo-institutional theory (Scott, 2008; Emirbayer and Johnson, 2008; Barley and Tolbert, 1997). I expanded existing theory by connecting resource scarcity to proximity through the conjecture that as the scarcity of a resource increases the influence of institutional and geographic proximity on collaboration decisions will decrease. Empirically, my mixed methods approach exposed an existing tension between participants? perceptions of how institutions influence their collaboration decisions and reality. While participants often believed that institutional differences between sectors and organisations did not influence collaboration decisions, institutions, particularly those that drew from the normative and cultural-cognitive pillars, affected action. This supports the theoretical assumptions that institutions are entrenched and taken-for-granted and that the influence of institutions drawing from the normative and cultural-cognitive pillars becomes unconscious (Scott, 2008). Leicht and Fennel (2008) argue that an existing gap in neo-institutional analyses is how market and non-profit institutions intertwine within biomedical organisational fields. I addressed this identified gap by exploring the multiple and often conflicting institutions within Vancouver?s I2 organisational field and analysing their influence on individuals. I outlined the dominant types of capital and institutions associated with specific organisations and sectors and analysed their impact on the structure of Vancouver?s I2 organisational field. An unexpected result was that tensions exist not only between market and non-profit institutions but also between academic and clinical    159  institutions. This finding provided a novel empirical insight into the subtle ways that institutions integrate and separate individuals affiliated with healthcare organisations and universities.  Distinct institutions influenced individuals affiliated with the private sector and academia that delineated clear and separate research approaches within the two spheres. Institutions within healthcare did not appear to similarly differentiate between academic and healthcare approaches to research. Instead, academic institutions dominated the research practices of individuals both universities and healthcare organisations. Academic institutions drawing from the regulative pillar, such as those related to IP development and ethical conduct of research, were used in research conducted within healthcare organisations. Participants affiliated with healthcare organisations discussed adhering to academic institutions that drew from the normative and cultural-cognitive pillars when conducting their research. With some exceptions?such as the BCCDC?the main differences between how research was conducted in healthcare organisations and universities did not relate to divergent institutions but to their contrasting capital. Affiliations with healthcare organisations gave researchers access to patients and patient data. For individuals affiliated with healthcare organisations, health-specific institutions governed their service roles and academic institutions their research roles. Perhaps because there did not appear to be distinct institutions that outlined a healthcare-specific research approach, ?pure? clinicians were not perceived as actively involved in research, creating a division between clinicians and clinician-scientists.  In Chapter 2, I addressed several methodological shortcomings of previous bibliometric studies that used organisations as their unit of analysis (see for example Hicks and Katz, 1996; Hoekman et al, 2010; Ponds et al, 2007; Sandstrom et al, 2000; Thijs and Glanzel, 2010; Wagner and Leydesdorff, 2005). This organisation-level approach was problematic because it obscured collaborations within the same organisation, failed to weight interorganisational collaborations by number of authors affiliated with each organisation, and did not differentiate between interorganisational collaborations involving two or more individuals affiliated with different organisations and    160  interorganisational collaborations involving one individual affiliated with multiple organisations. I remedied these three shortcomings by using individuals, as opposed to organisations, as the basis of my bibliometric analysis, adding a methodological robustness to large-scale bibliometric analyses that had previously been lacking. This was the first attempt, to my knowledge, to conduct a large-scale bibliometric analysis using individuals as the unit of analysis. In using individuals as my unit of analysis, I also created the first comprehensive summary of regional variations in multiple affiliation patterns as well as the role of multiple affiliations in bridging collaboration between sectors. The relatively high levels of multiple affiliations observed in Chapter 2, combined with results from my qualitative interviews which found that the perceived importance of multiple affiliations varied by individual, hint that multiple affiliations play a key, complex, and overlooked role in bibliometric analyses. 6.5 Shortcomings in work  Several shortcomings exist in my dissertation. Questions remain about the generalisability of my case study to cities other than Vancouver and to techno-epistemic areas other than I2. Generalisability is a common challenge in case study research (Creswell, 2007; Ragin and Becker, 1992). This is particularly true because Vancouver has a relatively small biotechnology presence (Salazar et al, 2008). Case studies allow researchers to explore specific processes in-depth, and, in doing so, offer insights and understandings that are often unachievable through other methods (Flyvbjerg, 2006). As an in-depth descriptive approach, case studies teach in a way that many people learn, providing information that can be taken from the specific case and generalised into larger truths (Stake, 2005).  Many major Western cities have a university, hospital, and biomedical private sector that are all integrated, to a certain extent, into an organisational field. While certain aspects of this case are likely unique to Vancouver, other insights are reflective of wider trends or realities in biomedical organisational fields. For example, my dissertation showed how a mid-sized biotechnology cluster weathered the recent financial crisis. After the Great Recession, venture capital dried up globally. Like the Vancouver-based companies that were part of this study, biotechnology companies    161  around the world have tried a variety of measures, including switching to a virtual organisational model, to address funding shortages (Ernst and Young, 2012). Other writers have similarly noted the challenges that exist in developing commercial applications from academic ideas (Butler, 2008) and other countries, most notably the USA through its new American National Centre for Advancing Translational Sciences (Wadman, 2013), are creating centres similar to CDRD to help address these challenges. University industry liaison offices that encourage translation between universities and the private sector have become institutionalised throughout the Western world (Colyvas and Powell, 2006). The complex, multi-organisational, networked academic hospitals observed in Vancouver have been observed in all of Canada?s major cities (Brimacombe et al, 2010).  My use of the Web of Science as a data source to initially map co-authorships had several shortcomings. The Web of Science does not include all journals, thus not all co-authored articles were found through this method. Co-authorship may not accurately represent collaborations. Researchers may include authors on their publications as a form of patronage, not as a result of actual collaboration in the research that resulted in the specific paper. Other important collaborations may never lead to a co-published article. As was pointed out in interviews, key individuals involved in research processes, for example nurses involved in patient recruitment, may not be listed as co-authors of papers. However, Haeussler and Sauermann (2013) found that, in addition to substantive intellectual contributions, co-authorship also reflects technical contributions and the provision of data or materials. Properly controlling for multiple affiliations became an unanticipated challenge in my quantitative analysis. My analysis in Chapter 2 showed that a large proportion of researchers were affiliated with multiple organisations, a rate that varied by country and region.The way that I sampled publications for my quantitative database also created some challenges for analysis. While all of the publications of Vancouver affiliated authors between 2004 and 2011 within the area of infection and immunity were sampled, only the publications of authors not affiliated with Vancouver that included an author from Vancouver as a co-author were included. Because of this, the publications of authors not affiliated with Vancouver, and their connections to authors also not affiliated with Vancouver, were incomplete.    162  This made it challenging to incorporate some of the more traditional social network analysis metrics such as centrality, betweennes, and closeness into my analyses. Firms are generally more strongly represented in patent as opposed to publication data (Meyer 2002). Analyses based on patent data would conversely be firm biased. While this may be the general case, in fields such as biomedicine, firms publish more in academic journals than firms in other fields. Godin (1996), for example, found that firms involved in drug development follow this pattern. This trend towards industry publication in biomedicine has likely increased as firms face increasing pressure to disseminate their clinical trial results. Bourgeois et al (2010) found that 67% of industry-funded drug trials were published in academic journals. From my own interviews I learned that, while publications were viewed as an important marketing technique for participants affiliated with biotechnology companies, it was a technique only employed after successful results had been collected. While this is similar to the positive result bias noted in publications more generally (Olson et al, 2002), it may have greater implications for start-up biotechnology companies that may wait years before publishing initial results.  Some of the discrepancies between my qualitative and quantitative results are probably due to shortcomings in my quantitative dataset. Because my quantitative data were unable to discern the purpose of a particular collaboration, ?key? local intersectoral collaborations were not distinguishable from other collaborations. For example, all the firms affiliated with participants I interviewed collaborated with universities and hospitals throughout the world. However, at least in part, the intellectual property on which their business was based had been initially acquired and developed with a local university or hospital. Thus, key collaborations between the commercial and public sectors appeared to be highly localised.  My quantitative dataset?s inability to differentiate these nuances meant that the importance of co-location and clusters in encouraging intersectoral collaboration may have been underemphasized in Chapter 2, but emerged in the qualitative analysis in Chapter 6. By combining qualitative and quantitative results, I was able to explore results that neither method sufficiently captured on its own. Using these complementary and triangulation mixed methods approaches (Greene et al, 1989) my qualitative results were able to compensate for the shortcomings discussed in my quantitative dataset.     163  6.6 Areas for future research Several areas of future research present themselves based on my findings. Because my dissertation focused on the Vancouver I2 organisational field, one approach would be to expand analysis to include other cities and techno-epistemic areas. This expansion could be achieved by conducting similar case studies in other cities and comparing and contrasting results. This would help illuminate which of the themes described within Vancouver?s I2 organisational field hold across cases. Publication data could also be expanded to a global database. This would facilitate a more traditional social network analytic approach that could include individual or organisational-level attributes such as centrality, betweenness, and closeness and explore correlations between these attributes and co-authorship rates. Additional individual-level attributes, such as the sex of an individual based on their name, could also be incorporated into the database. Another approach would be to include other types of data such as patents and formal contracts. Interviews with participants affiliated with firms showed that, while firms do publish, publications are used differently than they are in the academic sphere. Rather than a ?currency,? firms become involved in publications as a marketing tool once they believe they have something to market. Patents mark an earlier stage in intellectual collaboration processes within the commercial sphere. Combining publications and patents would create a more balanced picture of the academic and commercial spheres of collaboration. Contracts represent an even better physical manifestation of firm collaboration patterns. While contract information is not publicly available and more challenging to obtain, adding contract data would help map collaborations that are not necessarily intellectual but may be focused on production or analysis of research results. If firms continue their virtual business model, these types of collaborations will become increasingly important. This expanded dataset could be used to analyse the structure of collaboration within the network, testing different models such as geographic, sectoral, and cultural/linguistic clustering; and core-periphery structures.  Another potential research area relates to the meaning and effect of multiple affiliations on collaboration. An unanticipated result of this study was the high level of    164  multiple affiliations observed both within Vancouver and in other regions. Within Vancouver, I attempted to connect individuals to their ?home? organisation by controlling for the university affiliation of individuals with multiple affiliations. While this crude measure likely led to greater accuracy, my qualitative research showed that the importance placed on different affiliations varied dramatically by individual. Individuals with the same formal affiliations often perceived that these affiliations had different meanings depending to a large extent on their day-to-day work and informal roles. For some, multiple affiliations involved splitting work across organisations, for others it brought work expectations, and for others still it carried little meaning. It is unclear how reflective this ?fuzziness? of the roles associated with multiple affiliations in Vancouver is of the roles associated with multiple affiliations within other cities and countries. Chapter 2 showed that multiple affiliations are pervasive and need to be better understood. Future research aimed at understanding the role that multiple affiliations play in different cities and countries is necessary to ensure more accurate bibliometric analysis. It is also necessary to better understand the extent that multiple affiliations facilitate bridging between different organisations, a role likely better addressed through qualitative methods.  Another approach would be to further explore individuals? motivations to collaborate by focusing on more detailed understanding of how individuals are embedded in institutional cultures and organisations. To be effective, this analysis would conduct more in-depth studies of one or two specific organisations and institutions rather than creating the broader overview presented here. Through this approach, analysis would likely move beyond an organisational theory systems-level perspective instead drawing from other more micro-level literatures such as organisational behaviour.    Within my dissertation, hospital-based research was shown to span a wide spectrum from PhD researchers that were not involved in any clinical practice to clinicians dabbling in research off the side of their desk. Pathways integrating research with clinical care appeared ad hoc. Further studies should be conducted aimed at understanding how research and clinical practice integrate within teaching hospitals and    165  the impact of organisational structures and policies on integration. This could be studied through in-depth embedded qualitative studies of specific hospitals or hospital departments.  6.7 Conclusion Through a mixed methods multi-level study of researchers within Vancouver?s I2 organisational field my dissertation explored three research questions: 1. How do different types of proximity affect collaboration? 2. What reasons do people give for collaborating? 3. How do institutions such as policies, norms, and organisational culture affect collaboration? I found that universities dominated the network of Vancouver?s I2 global co-authorships while the private sector and NGOs played a relatively weak role. I explored geographic and institutional proximity through an analysis of I2 co-authorship rates using individuals as the unit of analysis. I found that collaborations were more likely to occur among geographically and institutionally close individuals, with shared sectoral affiliation being the best predictor of co-authorship. Almost 25% of authors were affiliated with more than one organisation. I analysed sectoral co-authorship patterns through an organisational-level analysis. I found marked preference for organisations to collaborate within, as opposed to across, sectors. Sectors showed a relatively high preference for collaborating with universities, a trend that was most pronounced in hospital-university co-authorships.  In Vancouver?s local I2 organisational field non-commercial organisations dominate, with the majority of non-commercial organisations reporting to the BC Ministry of Health. Universities had the largest number of affiliated authors. Firms had the largest number of organisations but the smallest number of authors. Organisations and formal institutions, such as affiliation agreements, UILOs, CDRD and CFRI, had developed to encourage intersectoral collaborations. Combining Scott?s (2008) three pillars of institutions with the concept of capital (Bourdieu, 1986; Giddens, 1984) I found that researchers collaborated to gain capital necessary to achieve goals. Organisations    166  and sectors shaped collaboration patterns because they provided access to specific capital and possessed their own institutions. While regulative institutions could act as barriers to collaborations, these were easily surmounted. Normative and cultural-cognitive institutions acted as stronger barriers to collaboration. I was also able to compare my qualitative and quantitative results. Quantitative and qualitative results did not generally diverge and provided complementary insights; I was able to use results from one method to help interpret findings from the other. In some instances qualitative findings helped give quantitative findings further depth. Qualitative interview results hinted that the reasons behind local intersectoral collaborations may differ from global intersectoral collaboration motivations; my quantitative analysis was unable to differentiate between collaboration motives. Quantitative results similarly helped to interpret qualitative findings; comparing perceptions from individual interviews to aggregate patterns. In my qualitative results participants often explicitly stated that organisational affiliation did not affect collaboration decisions; however, participants? discussions of particular collaborations illustrated that organisational affiliations mattered. Quantitative results supported this second interpretation, showing that affiliation did influence collaboration trends. Combined qualitative and quantitative results underscored the influence that resource scarcity and prestige play on collaboration decisions as well as how the public sector, and, within the public sector, academic institutions, dominate Vancouver?s I2 organisational field. My dissertation provided several novel empirical, methodological, and theoretical insights. A key contribution was my dissertation?s integration of individual?s perspectives and organisational field structure through a multi-level mixed methods approach. I was able to show how collaborations are based on interplays between institutions and available capital, empirically validating previous theoretical calls to incorporate the concept of capital into neo-institutional theory (see for example Scott, 2008; Emirbayer and Johnson, 2008; Berends et al, 2003; Barley and Tolbert, 1997). Empirically, this approach exposed a tension between participants? perceptions of how institutions influence their collaboration decisions and reality. My theoretical and methodological approach can also illustrate to other researchers how multiple levels of analysis can be used to integrate field structure and individual?s perspectives into an institutional case    167  study. My dissertation serves as an example of how an IS framework can be combined with the economic geography concept of proximity, neo-institutional theory, and perspectives from social network analysis to create a more robust analytic framework. I addressed a gap in neo-institutional analyses by exploring how different institutions intertwine within biomedical organisational fields. I found that tensions between academic and clinical institutions exist and that academic institutions dominate healthcare-based research which has access to unique resources in the form of patients and patient data. I addressed several methodological shortcomings of previous bibliometric studies by taking individuals, as opposed to organisations, as my unit of analysis. This was the first attempt, to my knowledge, to conduct a large-scale individual level bibliometric analysis. In doing so, I was able to create the first comprehensive summary of regional variations in multiple affiliation patterns. The results of my dissertation provide several insights that are useful in both assessing the health of Vancouver?s biomedical organisational field and in drafting future regional and national science policies.  I showed that while co-publishing maps idea based academic collaboration quite well, other types of collaborations that are necessary for a healthy organisational field?such as collaborations with the private sector?are not represented very well by co-authorships.  I believe that the dominant role academic institutions play in this network potentially shows a weakness in its structure. The weak role of firms makes translation along a ?commercial pathway? (Lander and Atkinson-Grosjean, 2011) more challenging because there are few firms involved in the process. Participants accepted that firms are necessary to move drugs, devices, and some diagnostics from theoretical to practical use because they have the expertise and economic resources required. While firms located outside of Vancouver can take on this developmental role, those in my small sample of biotechnology companies appeared to license most of their intellectual property locally from universities and hospitals. Because there is perceived limited local capacity, potentially valuable discoveries then get trapped in a congested pipeline. CDRD was established in 2007 in part to compensate for this weakness by building local private sector capacity. It    168  is unclear, as yet, how successful it will be. At the time of writing, CDRD had launched only one start-up company and out-licensed three novel therapies since its inception.22   The domination of academic institutions in the public sector is also worrying because it may decrease the effectiveness of the ?clinical pathway,? along which new clinical practices and some diagnostics are often mobilized (Lander and Atkinson-Grosjean, 2011).  Practitioners from healthcare organisations have the ability to inform their research with clinical insights, thereby strengthening the connection between academic research and clinical practice. This appears to be the approach taken by service/researchers at BCCDC. Other healthcare-based researchers and service/researchers may be discounting clinical institutions in favour of academic institutions in their research work. In this case, much of their bridging advantage is gone. Instead of combining research and service to make both, together, more than each can be alone, service/researchers addressing service and research demands separately may have trouble keeping up with colleagues who have only a service or a research demand on their time. The current system appears to obscure and downplay many of the strengths of doing research in health-based organisations which instead become obligatory passage points in accessing patients as resources. This result suggests that the BC Ministry of Health, and other Health Ministries worldwide, may want to reassess the role of research within healthcare settings by further integrating clinical institutions into research practices. Several approaches may help to further this integration. Drawing from the regulatory pillar, affiliation agreements between universities and healthcare organisations can more explicitly address research expectations of staff in addition to teaching, while providing nominal support, or increasing ?wiggle room? for research-based activities. Explicitly incorporating research and/or innovation into health care mandates would help support this shift. Perhaps, more importantly, normative and cultural institutions would need to evolve to more comprehensively integrate research into the clinical sphere. Within the BCCDC, using each employee?s service work as                                             22 See http://www.cdrd.ca/about-us/ (Access March 4, 2013)    169  ?experiments of nature? that act as springboards to research activities has been an effective approach that could be adopted elsewhere.   My work offers a sobering conclusion for policy makers interested in promoting clusters in high technology areas such as biotechnology. Overall, individuals within the I2 organisational field were reluctant to collaborate across sectors. Yet, such ?partnerships? are often viewed as a prerequisite for knowledge translation between research and development. A stronger private sector presence within this organisational field may be necessary for increased intersectoral collaborations. For funding agencies, such as the Networks of Centres of Excellence, that actively encourage intersectoral collaboration as a means of knowledge translation, my work has important implications for the development of funding policies. Policies adopted by these organisations to date have predominantly drawn from the regulative pillar, often adding criteria for grants without changing their underlying goals or allocation criteria, changes that would draw more strongly from the normative pillar. There has been some evidence that intersectoral collaborations based on meeting these grant criteria may be superficial (Atkinson-Grosjean, 2006). Indeed, my finding that most public research collaborations are relatively open, informal, and focused on ?why? questions appear to contradict newer grant criteria focused on utility and IP protection.  My dissertation shows how concepts of proximity, institutions and resources interact and influence collaboration decisions of researchers within and between sectors. This helped further the understanding of interactions between individuals influenced by different institutions. Those interested in exploring the interactions of these forces can find inspiration from my own work as they bring further insight to the study of collaborations and institutions.       170  7 References Ahuja, G. (2000). Collaboration networks, structural holes and innovation: A longitudinal study. Administrative science quarterly, 45(3), 425?455. Asheim, B. T., & Gertler, M. S. (2005). The geography of innovation: Regional innovation systems. In J. Fagerberg, D. C. Mowery, & R. R. Nelson (Eds.), The Oxford handbook of innovation (pp. 291?317). Oxford: Oxford University Press. Atkinson, P., Batchelor, C., & Parsons, E. (1998). Trajectories of collaboration and competition in a medical discovery. Science, technology, & human values, 23(3), 259?284. Atkinson-Grosjean, J. (2006). Public science, private interests: Culture and commerce in Canada?s Networks of Centres of Excellence. Toronto: University of Toronto Press. Atkinson-Grosjean, J., & Douglas, C. (2010). The third mission and the laboratory: How translational science engages and serves the community. In P. Inman & H. G. Schuetze (Eds.), The community engagement and service mission of universities (pp. 309?323). Leicester: Naice. Barley, S. R., & Bechky, B. A. (1994). In the backrooms of science: The work of technicians in science labs. Work and occupations, 21, 85?126. Barley, S. R., & Tolbert, P. (1997). Institutionalization and structuration: Studying the links between action and institution. Organisation studies, 18, 93?117. Bastian, M., Heymann, S., & Jacomy, M. (2009). Gephi: An open source software for exploring and manipulating networks. In International AAI conference on weblogs and social media. Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Progress in human geography, 28(1), 31?56. BC Cancer Agency. (2009). University of British Columbia-BC Cancer Agency Research Ethics Board. Vancouver. Retrieved from    171  http://www.bccancer.bc.ca/NR/rdonlyres/5F62664F-1273-4292-941D-F412B45062B9/32518/UBCBCCATermsofRefV420090216.pdf BC Cancer Agency. (2012). Fact sheet. Retrieved from http://www.bccancer.bc.ca/NR/rdonlyres/352EEEBD-6405-4CEF-824A-89282CB83B1A/59513/2012BCCAFactSheet1.pdf Bellward, Gail, Brownman, George, Suedfeld, Peter, & Fedoroff, Ingrid. (2007, March 27). One board of record letter. Ben-David, J. (1960). Roles and innovations in medicine. The American journal of sociology, 65(6), 557?568. Berends, H., Boersma, K., & Weggeman, M. (2003). The structuration of organisational learning. Human relations, 59(9), 1035?1056. Bernard, R. H. (2006). Research methods in anthropology (4th ed.). United States of America: Altamira Press. Bibby, M. (2009). Advocating strategies for government sponsored public health agencies: The BCCDC A Case Study (Master of Public Health). Simon Fraser University, Vancouver. Blume, S. (1985). The significance of technological change in medicine: An introduction. Research policy, 14, 173?177. Bordons, M., & Zulueta, M. (1997). Comparison of research team activity in two biomedical subfields. Scientometrics, 40(3), 423?436. Borgatti, S. P. (2002). NetDraw: Graph visualization software. Harvard: Analytic Technologies. Borgatti, S. P., & Everett, M. G. (1997). Network analysis of 2-mode data. Social networks, 19(3), 243?269. Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). UCINET 6 for Windows: Software for social network analysis. Harvard: Analytic Technologies.    172  Boschma, R. (2005). Proximity and innovation: A critical assessment. Regional studies, 39(1), 61?74. Boschma, Ron, & Frenken, Koen. (2010). The spatial evolution of innovation networks: A proximity perspective. In The handbook of evolutionary economic geography (pp. 120?136). UK: Edward Elgar Publishing Ltd. Bourdieu, P. (1985). The social space and the genesis of groups. Theory and society, 14(6), 723?744. Bourdieu, P. (1986). The forms of capital. In J. G. Richardson (Ed.), The Handbook of theory and research for the sociology of education (pp. 241?258). New York: Greenwood Press. Bourdieu, P. (1991). The peculiar history of scientific reason. Sociological forum, 6(1), 3?26. Bourgeosi, F. T., Murthy, S., & Mandl, K. (2010). Outcome reporting among drug trials registered in ClinicalTrials.gov. Annals of internal medicine, 153(3), 158?166. Boxenbaun, E., & Jonsson, S. (2008). Isomorphism, diffusion and decoupling. In R. Greenwood, C. Oliver, R. Suddaby, & K. Sahlin (Eds.), The SAGE handbook of organisational institutionalism (pp. 49?77). Los Angeles: SAGE. Braczyk, H.-J., Cooke, P., & Heidenreich, M. (1998). Regional innovation systems. Reading, UK: UCL Press. Breschi, S., & Malerba, F. (1997). Sectoral innovation systems: Technological regimes, Schumpeterian dynamics, and spatial boundaries. In C. Edquist (Ed.), Systems of innovation: Technologies, institutions and organisations (pp. 130?156). London: Pinter Publishers. Bressler, B., & Campbell, B. (2010). Securing the future of Canada?s academic health sciences centres. Academic Health Sciences Centres National Task Force. Brimacombe, G., & et al. (2010). Three missions one future: Optimizing the performance of Canada?s academic health sciences centres. Association of Canadian Academic Healthcare Organisations.    173  British Columbia Government. (1996), Hospital act., ? Chapter 200 British Columbia Government. (1996), University act., ? Chapter 468  British Columbia Ministry of Health. (2003). A national centre for disease control in British Columbia. BC Ministry of Health. Retrieved from http://www.health.gov.bc.ca/library/publications/year/misc/diseasecontrol.pdf Burt, R. (1987). Social contagion and innovation: Cohesion versus structural equivalence. American sociological review, 92, 1287?1335. Burt, R. (1992). The social structure of competition. In N. Nohria & R. G. Eccles (Eds.), Networks and organisation: Structure, form and action (pp. 57?91). Boston, MA: Harvard Business School Press. Burt, R. S. (2000). The network structure of social capital. Research in organisational behaviour, 22, 345?423. Butler, Declan. (2008). Crossing the valley of death. Nature, 453(12), 840?842. Callon, M. (1986). Some elements of a sociology of translation: domestication of the scallops and the fishermen of St. Brieuc Bay. In J. Law (Ed.), Power, action and belief: A new sociology of knowledge? (pp. 196?223). London: Routlage. Campbell, M., & Gregor, F. (2002). Mapping social relations: A primer in doing institutional ethnography. Aurora, Ontario: Garamond Press. Canadian Institutes of Health Research, Natural Sciences and Engineering Research Council of Canada, & Social Sciences and Humanities Research Council of Canada. (2010). Tri-council policy statement: Ethical conduct for research involving humans. Canada. Retrieved from http://www.pre.ethics.gc.ca/pdf/eng/tcps2/TCPS_2_FINAL_Web.pdf Carlsson, B. (2006). Internationalization of innovation systems: A survey of the literature. Research policy, 35, 56?67. Casper, S. (2007). How do technology clusters emerge and become sustainable? Social network formation and inter-firm mobility within the San Diego biotechnology cluster. Research policy, 36, 438?455.    174  Cetina, K. K. (1999). Epistemic cultures: How the sciences make knowledge. United States of America: Harvard University Press. Charmaz, K. (2005). Grounded theory in the 21st century: Applications for advancing social justice studies. In N. K. Denzin & Y. S. Lincoln (Eds.), The SAGE Handbook of qualitative research (3rd ed., pp. 537?545). Thousand Oaks, California: SAGE Publications Inc. Chesney, E. (2011). PHSA Research Metrics: 3rd Annual Report. Vancouver: PHSA. Cliff, A. D., & Ord, J. K. (1973). Spatial autocorrelation. London: Pion. Cohen, J. (2000). Ground zero: AIDS research in Africa. Science, 288(5474), 2150?2153. Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative science quarterly, 35(1, Special Issue: Technology, Organisations, and Innovation), 128?152. Coleman, J. S. (1988). Social capital in the creation of human capital. The American journal of sociology, 94(Supplement), S95?S120. Coleman, J. S. (1990). Commentary: Social institutions and social theory. American sociological review, 55(3), 333?339. Colyvas, J. A., & Powell, W. W. (2006). Roads to institutionalization: The remaking of boundaries between public and private science. Research in organisational behaviour, 27, 305?353. Conti, N., & Doreian, P. (2010). Social network engineering and race in a police academy: A longitudinal analysis. Social networks, 32, 30?43. Cooke, P. (1998). Introduction: Origins of the concept. In H.-J. Braczyk, P. Cooke, & M. Heidenreich (Eds.), Regional innovation systems (pp. 2?27). Reading, UK: UCL Press. Cooke, P., Uranga, M. G., & Etxebarria, G. (1997). Regional innovation systems: Institutional and organisational dimensions. Research policy, 26, 475?491.    175  Creswell, J. W. (2007). Qualitative inquiry and research design (2nd ed.). London: SAGE Publications Inc. Cross, R., Borgatti, S. P., & Parker, A. (2001). Beyond answers: Dimensions of the advice network. Social networks, 23, 215?235. Dahlander, L., & McFarland, D. A. (2013). Ties that last: Tie formation and persistence is research collaborations over time. Administrative science quarterly, 58(1), 69?110. Deephouse, D. L., & Suchman, M. (2008). Legitimacy in organisational institutionalism. In R. Greenwood, C. Oliver, R. Suddaby, & K. Sahlin (Eds.), The SAGE handbook of organisational institutionalism (pp. 49?77). Los Angeles: SAGE. DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organisational fields. American sociological review, 48(2), 147?160. Djellal, F., & Gallouj, F. (2005). Mapping innovation dynamics in hospitals. Research policy, 34(6), 817?835. Dosi, G. (1982). Technological paradigms and technological trajectories: A suggested interpretation of the determinants and directions of technical change. Research policy, 11(3), 147?162. Edquist, C. (2005). Systems of innovation: Perspectives and challenges. In J. Fagerberg, D. C. Mowery, & R. R. Nelson (Eds.), The Oxford handbook of innovation (pp. 181?208). Oxford: Oxford University Press. Edquist, C., & Johnson, B. (1997). Institutions and organisations in systems of innovation. In C. Edquist (Ed.), Systems of innovations: Technologies, institutions and organisations (pp. 41?63). London: Pinter Publishers. Ejermo, O., & Karlsson, C. (2006). Interregional inventor networks as studied by patent coinventorships. Research policy, 35, 412?430. Emirbayer, M., & Johnson, V. (2008). Bourdieu and organisational analysis. Theory and society, 37, 1?44.    176  Erickson, B. H. (1988). The relational basis of attitudes. In Social structures: A network approach (pp. 83?98). Cambridge: Cambridge University Press. Ernst & Young. (2012). Beyond borders: Global biotechnology report 2012 (No. FN0011). EYGM Ltd. Etzkowitz, H. (1992). Individual investigators and their research groups. Minerva, 30(1), 28?50. Etzkowitz, H., & Leydesdorff, L. (1999). The future location of research and technology transfer. Journal of technology transfer, 24, 111?123. Etzkowitz, H., & Leydesdorff, L. (2000). The dynamics of innovation: From national systems and ?mode 2? to a triple helix of university-industry-government relations. Research policy, 29, 109?123. Etzkowitz, H., Webster, A., Gebhardt, C., & Terra, B. R. C. (2000). The future of the university and the university of the future: Evolution of ivory tower to entrepreneurial paradigm. Research policy, 29, 313?330. Farrar, D. H. (2011). Memorandum: Senate approval of UBC research institutes (Vancouver UBC Senate document No. Item 11a) (pp. 1?31). Vancouver: University of British Columbia. Feld, S. (1981). The focused organisation of organisational ties. American journal of sociology, 86, 1015?35. Fennell, M. L., & Alexander, J. A. (1987). Organisational boundary spanning in institutionalized environments. Academy of management journal, 30(3), 456?476. Flyvbjerg, Bent. (2006). Five misunderstandings about case-study research. Qualitative inquiry, 12, 219?245. Freeman, C. (1988). Japan: A new national system of innovation? In G. Dosi, C. Freeman, R. R. Nelson, & L. Soete (Eds.), Technical change and economic theory (pp. 331?348). London: Pinter Publishers. French, M., & Miller, F. A. (2012). Leveraging the ?living laboratory?: On the emergence of the entrepreneurial hospital. Social science and medicine, 72, 717?724.    177  Frenken, Koen, Hardeman, Sjoerd, & Hoekman, Jarno. (2009). Social scientometrics: Towards a cumulative research program. Journal of informetrics, 3, 222?232. Frenken, Koen, Holzl, W, & De Vor, F. (2005). The citation impact of research collaborations: The case of European biotechnology and applied microbiology (1988-2002). Journal of engineering management and technology, 22(1-2), 9?30. Friedland, R., & Alford, R. R. (1991). Bringing society back in: Symbols, practices, and institutional contradictions. In W. W. Powell & P. J. DiMaggio (Eds.), The new institutionalism in organisational analysis (pp. 232?263). Chicago: University of Chicago Press. Fritsch, M., & Kauffeld-Monz, M. (2010). The impact of network structure on knowledge transfer: An application of social network analysis in the context of regional innovation networks. Annals of regional science, 44, 21?38. Fujimura, J. H. (1996). Crafting science: A sociohistory of the quest for the genetics of cancer. Cambridge, Massachusetts: Harvard University Press. Gelijns, A. C., & Rosenberg, N. (1994). The dynamics of technological change in medicine. Health Affairs, 13(3), 28?47. Gelijns, A. C., Zivin, J. G., & Nelson, R. R. (2001). Uncertainty and technological change in medicine. Journal of health politics, policy and law, 25(5), 913?924. Giddens, A. (1979). Central problems in social theory. Berkeley, CA: University of California Press. Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Great Britain: University of California Press. Gilding, M. (2008). ?The tyranny of distance?: Biotechnology networks and clusters in the antipodes. Research policy, 37, 1132?1144. Glass, G. V., & Hopkins, K. D. (1996). Statistical methods in education and psychology (3rd ed.). Needham Heights, USA: Allyn & Bacon.    178  Godin, B. (1996). Research and the practice of publication in industries. Research policy, 25(4), 587?606. Goes, J. B., & Park, S. H. (1997). Interorganisational links and innovation: The case of hospital services. The academy of management journal, 40(3), 673?696. Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. The American journal of sociology, 91(3), 481?510. Greene, J. C., Caracelli, V. J., & Graham, W. F. (1989). Toward a conceptual framework for mixed-method evaluation designs. Educational evaluation and policy analysis, 11(3), 255?274. Greenwood, R., Oliver, C., Sahlin, K., & Suddaby, R. (2008). Introduction. In R. Greenwood, C. Oliver, K. Sahlin, & R. Suddaby (Eds.), The SAGE handbook of organisational institutionalism (pp. 1?46). Los Angeles: SAGE. Gregersen, B. (1992). The public sector as a pacer in national systems of innovation. In B.A. Lundvall (Ed.), National systems of innovation: Towards a theory of innovation and interactive learning (pp. 129?145). London: Pinter Publishers. Grieryn, Thomas. (1983). Boundary work and the demarcation of science from non-science: Strains and interests in professional ideologies of scientists. American sociological review, 48(6), 781?795. Hackett, E. J. (2005). Essential tensions: Identity, control and risk in research. Social studies of science, 35(5), 787?826. Haeussler, Carolin, & Sauermann, Henry. (2013). Credit where credit is due? The impact of project contributions and social factors on authorship and inventorship. Research policy, 42, 688?703. Hannan, M. T., & Freeman, J. (1977). The population ecology of organisations. American journal of sociology, 82(5), 929. Hannenman, R. A., & Riddle, M. (2011). A brief introduction to analysizing social network data. In J. Scott & P. J. Carrington (Eds.), The SAGE handbook of social network analysis (pp. 331?340). Cornwall, Great Britain: SAGE Publications Ltd.    179  Hannenman, R.A., & Riddle, M. (2005). Introduction to social network analysis. Riverside, CA: University of California. Retrieved from published in digital form at http://faculty.ucr.edu/hanneman/ Hansen, M. (1999). The search-transfer problem: The role of weak ties in sharing knowledge across organisational subunits. Administrative science quarterly, 44(1), 82?111. Haunschild, P., & Chandler, D. (2008). Institutional-level learning: Learning as a source of institutional change. In R. Greenwood, C. Oliver, R. Suddaby, & K. Sahlin (Eds.), The SAGE handbook of organisational institutionalism (pp. 624?649). Los Angeles: SAGE. He, Z.L., Geng, X.S., & Campbell-Hunt, C. (2009). Research collaboration and research output: A longitudinal study of 65 biomedical scientists in a New Zealand university. Research policy, 38, 306?317. Hicks, D., & Katz, J. S. (1996). Hospitals: The hidden research system. Science and public policy, 23(5), 297?304. Hilbe, J. M. (2011). Negative binomial regression (2nd ed.). New York: Cambridge University Press. Hodgson, G. M. (1994). The return of institutional economics. In N. J. Smelser & R. Swedberg (Eds.), The handbook of economic sociology (pp. 58?76). Princeton, N.J.: Princeton Universtiy Press. Hoekman, J., Frenken, K., & Tijssen, R. J. W. (2010). Research collaboration at a distance: Changing spatial patterns of scientific collaboration within Europe. Research policy, 39, 662?673. Holbrook, J. A., & Clayman, B. P. (2006). Research funding by city: An indicator of regional technological competitiveness? Research evaluation, 15(3), 221?231. Hopkins, M. (2006). The hidden research system: The evolution of cytogenetic testing in the National Health Service. Science as culture, 15(3), 253?276.    180  IRP. (2013). Simon Fraser University: Fingertip statistics. Retrieved from www.sfu.ca/irp Jepperson, R. L. (1991). Institutions, institutional effects and institutionalism. In W. W. Powell & P. DiMaggio (Eds.), The new institutionalism in organisational analysis (pp. 143?163). Chicago: University of Chicago Press. Karnieli-Miller, O., Strier, R., & Pessach, L. (2009). Power relations in qualitative research. Qualitative health research, 19, 279?288. Katz, J. S. (1994). Geographical proximity and scientific collaboration. Scientometrics, 31(1), 31?43. Katz, J. Sylvan, & Martin, B. R. (1997). What is research collaboration? Research policy, 26, 1?18. Keeble, D. (1999). Collective learning processes, networking and institutional thickness in the Cambridge region. Regional studies, 33(319), 332. Knoke, D., & Yang, S. (2008). Social network analysis (2nd ed.). Thousand Oaks, California: SAGE Publications. Kvale, Steinar. (1996). InterViews: An introduction to qualitative research interviewing. London: SAGE Publications Inc. Lander, B. (2011). Between theory and craft: Exploring the role of co-operation within scientific research labs. Spontaneous generations: A journal for the history and philosophy of science, 5(1), 58?74. Lander, B. (2013). Sectoral collaboration in biomedical research and development. Scientometrics, 94, 343?357. Lander, B. & Atkinson-Grosjean, J. (2011). Translational science and the hidden research system in universities and academic hospitals: A case study. Social science and medicine, 72, 537?544. Lander, B., Hanley, G. E., & Atkinson-Grosjean, J. (2010). Clinician-scientists in Canada: barriers to career entry and progress. PLoS one, 5(10).    181   Lane, P. J., & Lubatkin, M. (1998). Relative absorptive capacity and interorganisational learning. Strategic management journal, 19(5), 461?477. Latouche, D. (1998). Do regions make a difference?: The case of science and technologies policies in Quebec. In H.J. Braczyk, P. Cooke, & M. Heidenreich (Eds.), Regional innovation systems (pp. 319?345). Reading, UK: UCL Press. Latour, B. (1983). Give me a laboratory and I will raise the world. In K. Knorr-Cetina & M. Mulkay (Eds.), Science observed: Perspectives on the social study of science (pp. 141?170). London: Sage. Law, J. (1992). Notes on the theory of actor-network: Ordering, strategy, and heterogeneity. Systems practice, 5(4), 379?393. Lazega, E., & Van Dujin, M. (1997). Position in formal structure, personal characteristics and choices of advisors in a law firm: A logistical regression model for dyadic network data. Social networks, 19, 375?97. Leicht, K.T., & Fennell, M.L. (2008). Institutionalism and the professions. In Greenwood, R., Oliver, C., Suddaby, R., & Sahlin, K. (Eds.), The SAGE handbook of organisational institutionalism (pp. 431?448). Los Angeles: SAGE Publications Inc. Lewison, G. (1999). The definition and calibration of biomedical subfields. Scientometrics, 46(3), 529?537. Lin, N. (1999). Building a network theory of social capital. Connections, 22(1), 28?51. Lin, N., & Erickson, B. H. (2008). Theory, measurement and the research entreprise on social capital. In N. Lin & B. H. Erickson (Eds.), Social capital: An international research program. Oxford University Press: Oxford. Long, J. S. (1997). Regression models for categorical and limited dependent variables. United States of America: SAGE Publications Inc. Luken, P. C., & Vughan, S. (2006). Standardizing childrearing through housing. Social problems, 53(3), 299?331.    182  Lundvall, B. A. (1988). Innovation as an interactive process: From user-producer interaction to the national system of innovation. In G. Dosi, C. Freeman, R. R. Nelson, G. Silverberg, & L. Soete (Eds.), Technical change and economic theory (pp. 349?369). London: Pinter Publishers. Lundvall, B. A. (1992). National systems of innovation: Towards a theory of innovation and interactive learning. London: Pinter. Malerba, F. (2004). Sectoral Systems of innovation: Basic concepts. In F. Malerba (Ed.), Sectoral systems of innovation: Concepts, issues and analyses of six major sectors in Europe (pp. 9?41). Cambridge, UK: Cambridge University Press. Malerba, F. (2005). Sectoral Systems: How and why innovation differs across sectors. In J. Fagerberg, D. C. Mowery, & R. R. Nelson (Eds.), The Oxford handbook of innovation (pp. 381?405). Oxford: Oxford University Press. March, J. G., & Simon, H. A. (1958). Organisations. USA: Wiley. Martin, J. L. (2003). What is field theory? The American journal of sociology, 109(1), 1?49. McFetridge, D. G. (1993). The Canadian system of industrial innovation. In R. R. Nelson (Ed.), National innovation systems: A comparative analysis (pp. 299?323). Oxford: Oxford University Press. McGinn, T., Wyer, P. C., Newman, T. B., Keitz, S., Leipzig, R., & Guyatt, G. (2004). Tips for learners of evidence-based medicine: Measures of observer variability (kappa statistic). CAMJ, 23(11), 1369?1373. McKelvey, M. (1991). How do national systems of innovation differ?: A critical analysis of Porter, Freeman, Lundvall and Nelson. In G. M. Hodgson & E. Screpanti (Eds.), Rethinking economics: Markets, technology and economic evaluation (pp. 117?137). Great Britain: Edward Elgar. McKelvey, M. (1997). Using evolutionary theory to define systems of innovation. In C. Edquist (Ed.), Systems of innovation: Technologies, institutions and organisations (pp. 200?222). London: Pinter Publishers.    183  McKelvey, M., Alm, H., & Riccaboni, M. (2003). Does co-location matter for formal knowledge collaboration in the Swedish biotechnology-pharmaceutical sector? Research policy, 32, 483?501. McKelvey, M., Orsenigo, L., & Pammolli, F. (2004). Pharmaceutical analysed through the lens of a sectoral innovation system. In F. Malerba (Ed.), Sectoral systems of innovation: Concepts, issues and analyses of six major sectors in Europe (pp. 73?120). Cambridge, UK: Cambridge University Press. McPherson, M., Smith-Lovin, L., & Cook, J.M. (2001). Birds of a feather: Homophily in social networks. Annual review of sociology, 27, 415?444. Melin, G. (2000). Pragmatism and self-organisation: Research collaboration on the individual level. Research policy, 29, 31?40. Meyer, J. W., & Rowan, B. (1977). Institutionalized organisations: Formal structure as myth and ceremony. American journal of sociology, 83(2), 340?363. Meyer, M. (2002). Tracing knowledge flows in innovation systems. Scientometrics, 54(2), 193?212. Montori, V. M., Wilcynski, N. L., Morgan, D., & Haynes, R. B. (2005). Optimal search strategies for retrieving systematic reviews from Medline: Analytical survey. British medical journal, 330(7482), 68?71. Morlacchi, P., Wilkinson, I. F., & Young, L. C. (2005). Social networks of researchers in B2B marketing: A case study of the IMP group 1984-1999. Journal of business-to-business marketing, 12(1), 3?34. Mowery, D. C., & Sampat, B. N. (2005). Universities in national innovation systems. In J. Fagerberg, D. C. Mowery, & R. R. Nelson (Eds.), The Oxford handbook of innovation (pp. 209?239). Oxford: Oxford University Press. Murdoch, J. (1997). Inhuman/nonhuman/human: Actor-network theory and the prospects for a nondualistic and symmetrical perspective on nature and society. Enviornment and planning D: Society and space, 15, 731?756.    184  Murphy, K., Travers, P., & Walport, M. (2008). Janeway?s Immunobiology (7th ed.). New York, NY: Garland Science, Taylor and Francis Group. Murray, F. (2002). Innovation as co-evolution of scientific and technological networks: Exploring tissue engineering. Research policy, 31, 1389?1403. Narin, F., Stevens, K., & Whitlow, E. S. (1991). Scientific cooperation in Europe and the citation of multinationally authored papers. Scientometrics, 21(3), 80?93. Nchinda, T. C. (2002). Research capacity strengthening in the South. Social science and medicine, 54, 1699?1711. Nelson, R. R. (2006). Reflections on ?the simple economics of basic scientific research?: Looking back and looking forward. Industrial and corporate change, 15(6), 903?917. Nelson, R. R., & Rosenberg, N. (1993). Technical innovation and national systems. In R. R. Nelson (Ed.), National innovation systems: A comparative analysis (pp. 3?22). Oxford: Oxford University Press. Nelson, R., & Winter, S. G. (1977). In search of a useful theory in innovation. Research policy, 6, 36?76. Newman, M. E. J. (2001). The structure of scientific collaboration networks. PNAS, 2, 404?409. Newman, M. E. J. (2003). The structure and function of complex networks. Society for industrial and applied mathematics, 45(2), 167?256. O?Connor, E. (2012, August 26). Special Deliveries: St. Paul?s Hospital a Centre for treating and delivering the highest-risk mothers in the province. The Province. Vancouver. Organisation for Economic Co-operation and Development. (1997). National innovation systems. Paris: OECD Organisation for Economic Co-operation and Development. (1999). Managing national innovation systems. Paris: OECD    185  Orton, J. D., & Weick, K. E. (1990). Loosely coupled systems: A reconceptualization. The academy of management review, 15(2), 203?223. Owen-Smith, J., & Powell, W. W. (2004). Knowledge networks as channels and conduits: The effects of spillovers in the Boston biotechnology community. Organisation science, 15(1), 5?21. Owen-Smith, J., & Powell, W. W. (2008). Networks and institutions. In R. Greenwood, C. Oliver, K. Sahlin, & R. Suddaby (Eds.), The SAGE handbook of organisational institutionalism (pp. 596?623). Los Angeles: SAGE. Page, J. (2010). Power, science and nature in the great bear rainforest: An actor-network analysis of integrated natural resource management project. University of British Columbia, Vancouver. Pestana, A. (1992). Spanish performance in life sciences: A comparative appraisal of the scientific production of Spain and five other European countries in 1989. Scientometrics, 24, 95?114. Pfeffer, J. (1997). New directions for organisational theory: Problems and prospects. New York, NY: Oxford University Press. PHSA. (2002). Master affiliation agreement between The University of British Columbia and Provincial Health Services Authority (Agreement). Vancouver. PHSA. (2007). BC Centre for Disease Control: Strategic plan. Polanyi, M. (1967). The tacit dimension. New York: Doubleday. Ponds, R., Oort, F. van, & Frenken, K. (2007). The geographical and institutional proximity of research collaboration. Regional science, 86(3), 423?443. Ponds, R. (2009). The limits to internationalisation of scientific research collaboration. Journal of technology transfer, 34, 76?94. Porter, M.E. (1998). Clusters and the new economics of competition. Boston, Massachusetts: Harvard Business School Press.    186  Powell, W. W., & Colyvas, J. A. (2008). Microfoundations of institutional theory. In R. Greenwood, C. Oliver, & K. Sahlin (Eds.), The SAGE handbook of organisational institutionalism (pp. 276?298). Los Angeles: SAGE. Powell, W. W., White, D. R., Koput, K. W., & Owen-Smith, J. (2005). Network dynamics and field evolution: The growth of interorganisational collaboration in the life sciences. American journal of sociology, 110(4), 1132?1205. Powell, W.W., & DiMaggio, P.J. (1991). Introduction. In The new institutionalism in organisational analysis. Chicago, Il: University of Chicago Press. Powell, W., & Cranston, L. (2010). PHSA strategic plan. Vancouver: PHSA. Price, D. D. (1984). The science/technology relationship, the craft of experimental science, and policy for the improvement of high technology innovation. Research policy, 13, 3?20. Providence Health Care. (2009). British Columbia Centre for Excellence in HIV/AIDs: Strategic plan, 2009-2014. Vancouver. Providence Health Care. (2012a). Fact sheet: HIV/AIDS program. Vancouver. Providence Health Care. (2012b). Providence Health Care annual community report 2011-2012. Vancouver. QLT cutting 42 per cent of work force, to take $2-million charge. (2012, December 6). The Globe and Mail. Vancouver. R Core Team. (2012). R: A language and environment for statistical computing. Vienna, Austria: R foundation for statistical computing. Retrieved from http://www.R-project.org Ragin, C. C., & Becker, H. S. (1992). What is a case? Exploring the foundations of social inquiry. New York: Cambridge University Press. Ramlogan, R., Mina, A., Tampublolon, G., & Metcalfe, J. S. (2007). Networks of knowledge: The distributed nature of medical innovation. Scientometrics, 70(2), 459?489.    187  Reagans, R., & McEvily, B. (2003). Network structure and knowledge transfer: The effects of cohesion and range. Administrative science quarterly, 48(2), 240?267. Rees, K. (2005). Interregional collaboration and innovation in Vancouver?s emerging high tech cluster. Tijdschrift voor economische en sociale geografie, 96(3), 298?312. Richards, Lyn. (2009). Handling qualitative data: A practical guide (2nd ed.). Cornwall, Great Britain: SAGE Publications Ltd. Salazar, M., Bliemel, M. J., & Holbrook, J. A. (2008). A comparison of R&D indicators for the Vancouver biotechnology cluster. Journal of commercial biotechnology, 14(3), 233?246. Sandstrom, A., Pettersson, I., & Nilsson, A. (2000). Knowledge production and knowledge flows in the Swedish biotechnology innovation system. Scientometrics, 48(2), 179?201. Saxenian, A. (1996). Regional advantage: Culture and competition in Silicon Valley and Route 128. United States of America: First Harvard University Press. Scott, J. (2000). Social network analysis (2nd ed.). Great Britain: Sage Publications. Scott, W. R. (2008). Institutions and organisations: Ideas and interests (3rd ed.). United States of America: SAGE Publications Inc. Seidman, Irving. (1998). Interviewing as qualitative research: A guide for researchers in education and the social sciences (2nd ed.). New York, NY: Teachers College Press. Sewell, William H., Jr. (1992). A theory of structure: Duality, agency and transformation. American journal of sociology, 98(1), 1?29. Sharif, N. (2006). Emergence and development of the national innovation systems concept. Research policy, 35, 745?766. Shrum, W., Genuth, J., & Chompalov, I. (2007). Structures of scientific collaboration. United States of America: MIT Press.    188  Simmul, G. (1922). Conflict and web of group affiliations. (Wolff, K.H & Bendix, R., Trans.). New York, NY: Free Press. Snijders, T. A., & Borgatti, S. (1999). Non-parametric standard errors and tests for network statistics. Connections, 22(2), 161?170. Stake, R. E. (2005). Qualitative case studies. In N. K. Denzin & Y. S. Lincoln (Eds.), The SAGE Handbook of qualitative research (3rd ed., pp. 443?467). London: SAGE Publications Inc. Strang, D. (1987). The administrative transformation of American education: School district consolidation, 1938-1980. Administrative science quarterly, 32(3), 352?366. Sun, Y., & Negishi, M. (2010). Measuring the relationships among university, industry and other sectors in Japan?s national innovation system: A comparison of new approaches with mutual information indicators. Scientometrics, 82(3), 677?685. Swan, J., Goussevskaia, A., Newell, S., Robertson, M., Bresnen, M., & Obembe, A. (2007). Modes of organizing biomedical innovation in the UK and US and the role of integrative and relational capabilities. Research policy, 36, 529?547. Szulanski, G. (1996). Exploring internal stickiness: Impediments to the transfer of best practice within the firm. Strategic management journal, 17, 27?43. Teddlie, C., & Tashakkor, A. (2003). Major issues and controversies in the use of mixed methods in the social and behavioral sciences. In Tashakkori, A. & Teddlie, C. (Eds.), Handbook of mixed methods in social and behavioral research (pp. 3?50). USA: SAGE Publications Inc. Thijs, B., & Glanzel, W. (2010). A structural analysis of collaboration between European research institutes. Research evaluation, 19(1), 55?65. Thompson, J. D. (1967). Organisations in action. New York: McGraw-Hill. Thornton, P. H., & Ocasio, W. (2008). Institutional logics. In R. Greenwood, C. Oliver, K. Sahlin, & R. Suddabry (Eds.), The SAGE handbook of organisational institutionalism (pp. 99?129). Los Angeles: SAGE.    189  Thune, T. (2007). University-industry collaboration: The network embeddedness approach. Science and public policy, 34(3), 158?168. Times Higher Education (2012). The world university rankings. Retrieved from http://www.timeshighereducation.co.uk/world-university-rankings/2012-13/world-ranking Tolbert, P., & Zucker, L. G. (1983). Institutional sources of change in the formal structure of organisations: The diffusion of civil service reform, 1880-1935. Administrative science quarterly, 28, 22?39. Torre, A., & Rallet, A. (2005). Proximity and localisation. Regional studies, 39(1), 47?59. UBC. (2003). Research affiliation agreement for intellectual property: University of British Columbia and Vancouver Coastal Health Authority. Vancouver. UBC UILO. (2012). The University of British Columbia-University Industry Liaison Office: Strategic Plan. Vancouver: UBC. UBC. (2011). Research ethics board standard operating procedures., ? 101 Authority and purpose. Uzzi, B. (1997). The social structure and competition in interfirm networks: The paradox of embeddedness. Administrative science quarterly, 42, 35?67. Wadman, M. (2013). Medicine man. Nature, 494, 24?26. Wagner, C. S., & Leydesdorff, L. (2005). Network structure, self-organisation, and the growth of international collaboration in science. Research policy, 34, 1608?1618. Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications. New York, NY: Cambridge University Press. Wellman, B. (1996). Are personal communities local? A Dumptarian reconsideration. Social networks, 18, 347?54.    190  Whittington, Kjersten Bunker, Owen-Smith, Jason, & Powell, Walter W. (2009). Networks, propinquity, and innovation in knowledge-intensive industries. Administrative science quarterly, 54, 90?122. Williamson, O. E. (1994). Transaction cost economics and organisational theory. In N. J. Smelser & R. Swed (Eds.), The handbook of economic sociology (pp. 77?107). Princeton, N.J.: Princeton University Press and Russell Sage Foundation. Windrum, P., & Garcia-Goni, M. (2008). A Neo-Schumpeterian model of health services innovation. Research policy, 37, 649?672. Wixted, B., & Holbrook, J. A. (2011). Innovation, cities and place: An empirical study of the knowledge system in Vancouver and its place on the Pacific Rim. In H. Melkas & V. Marmaakorpi (Eds.), Practice-based innovation: Insights, applications and policy implications (pp. 323?344). Berlin: Springer. Wolfe, D., & Gertler, M. (1998). The regional innovation system in Ontario. In H.J. Braczyk, P. Cooke, & M. Heidenreich (Eds.), Regional innovation systems (pp. 99?195). Reading, UK: UCL Press. Wooten, M., & Hoffman, A. J. (2008). Organisational fields: Past present and future. In Royston Greenwood, C. Oliver, R. Suddaby, & K. Sahlin (Eds.), The SAGE Handbook of organisational institutionalism (pp. 99?130). Los Angeles: SAGE. Yin, R. K. (2003). Case study research: Design and methods (3rd ed.). Thousand Oaks, California: SAGE Publications Inc. Zeileis, A., Kleiber, C., & Jackman, S. (2008). Regression models for count data in R. Journal of statistical software, 27(8), 1?25.       191  8 Appendix A: Creating an I2 database I began by developing a database of I2 publications from the ISI Web of Science where at least one author was from Metro Vancouver. Figure 10 outlines the process used in creating the database. I began by creating a filter to find and extract publications from the ISI Web of Science based on three components: range of years used for extraction, a definition of Metro Vancouver, and an identification of I2 papers. The years chosen to extract publications were between 2004 and 2011, inclusive. Initial extraction was based on the years 2004-2009. I chose a six year window to provide what I believed was a relatively comprehensive but static picture of research and development collaborations; five years is long enough to smooth out yearly variation in publication rates by specific authors and organisations while short enough to create a relatively static illustration of research and development collaborations. At the time of initial extraction (October 2010), I was unable to extract articles after 2009. Before 2008, it was not possible to connect authors directly to organisations within ISI Web of Science bilbiometric records. New protocols implemented in 2008 connected authors to organisations for the first time. To capitalize on this change, I later expanded my extraction years to include 2010 and 2011. As described in more detail in the separate chapters, Chapter 3 is based at the organisation-level and uses the initial extraction (2004-2009) while Chapter 2 is based at the individual-level and uses the later years of data (2008-2011).     192  Figure 10: Creation of an I2 database Years: 2004-2011 Vancouver: census metropolitan areaInfection and immunity: Specialist journals and keywords1450 papers1042 papersBook chapter: 2Proceedings paper: 3Correction: 2Editorial material: 62Meeting abstract: 293Proceedings paper: 46 Vancouver was defined according to the Statistics Canada 2006 Census definition of Vancouver?s Census Metropolitan Area (CMA)23 to include Bowen Island, West Vancouver, North Vancouver, Vancouver, Burnaby, Richmond, Delta, Surrey, Langley, Coquitlam, Port Coquitlam, Pitt Meadows, White Rock, New Westminster, Port Moody, Lions Bay, Maple Ridge, Semiahmoo, Tsawwassen, Musqueam, Burrard Inlet, Mission, Capilano, Barnston Island, Seymour Creek, Katzie, McMillan Island, Matsqui, and Whonnock. I2 papers were filtered following the biomedical subfield procedure recommended by Lewison (1999). Lewison?s (1999) procedure enables the creation of a biomedical subfield database through a multi-stage process from two interrelated sources: specialist journals and generalist journals based on a keyword list. While many similar analyses are based solely on specialist journal list data extractions (see for example Sandstrom et al, 2000; Bordons and Zulueta, 1997; Pestana, 1992), this approach of                                             23 See http://www.statcan.gc.ca/pub/92f0138m/2003002/4225101-eng.pdf (Accessed September 28, 2010).    193  combining specialist and generalist journals is much more comprehensive. Including papers from generalist journals is particularly important because an estimated two thirds of biomedical papers are in generalist as opposed to specialist journals and the most highly cited papers typically appear within generalist journals (Lewison, 1999).  First, a list of specialist I2 journals were compiled. This combined the Thomson Reuters Essential Science Indicators Immunology Journal List and the Ulrichs Web Global Serials Directory for Allergology and Immunology as well as Communicable Diseases. Combined, this created a specialist I2 journal list comprised of 137 journals which are listed in Table 15 at the end of Appendix 1. All papers from these journals were assumed to be I2 related. I extracted all papers (840) from these journals between 2004 and 2009 where at least one author was from Metro Vancouver.   To develop my initial I2 keyword list I took all the words used in all 840 titles, separated the words, and ranked them by the number of times they appeared in the 840 titles. Adjectives and words such as ?activate?, ?adherence? or ?adult? that were clearly relevant to more than just I2 papers were discounted from the keyword list. Similar words were grouped together and word stems were sought to create a more parsimonious keyword list. This resulted in a list of 49 potential keywords.  The precision and accuracy (sometimes called specificity and sensitivity) of these keywords were assessed sequentially by two I2 experts. Precision here represents the proportion of papers retrieved by the keyword list that are I2 related over total papers retrieved. Recall represents the proportion of I2 papers retrieved from generalist journals by the keyword list over the total I2 papers in the generalist journals (Montori et al 2005; Lewison 1999). To perform this assessment I first developed a list of generalist scientific and medical journals based on the Web of Science Journal Citation Reports. This was based on the top 100 generalist medical journals within the categories of medical laboratory technology; medicine, general and internal; and medicine, research and experimental combined with the top 30 journals within the multidisciplinary sciences category. The assessment of ?top? journals was based on impact factor. Eleven journals were subsequently taken off of this list as not related to biomedicine (i.e. the Proceedings of the Royal Society for Mathematical, Physical and Engineering Sciences).    194  See Table 16 at the end of Appendix 1 for a full list of generalist journals used in this extraction. Experts were given two lists of papers extracted from these generalist journals. Both lists included papers from 2009 where at least one author was from Metro Vancouver and included the first three authors for each paper, its title and journal of publication. The first list, used to assess the precision of the keywords, included all papers extracted from generalist journals using the keywords. The second list, used to assess the accuracy of keywords, extracted all papers published within New England Journal of Medicine, Lancet, Science, or Nature in 2009 where at least one author was from Vancouver. Experts were asked to review both lists and mark whether each paper was relevant to I2, non-relevant, or if the expert was unsure. Based on the assessment of the first expert, the keywords list was modified to 45 keywords shown in Table 17 at the end of Appendix 1. This was used to extract another list from generalist journals to assess keyword precision. This list, and the second list used to access the accuracy of keywords, was then given to a second expert who similarly marked the relevance of each paper. Assessment by the second expert led to a precision of 78% and recall of 96% for my keyword list. These keywords were then used to extract 237 papers from the generalist journal list between 2004 and 2009 where at least one author was from Metro Vancouver. These 203 papers were combined with the 840 papers to create my initial relational I2 database (1043 total) in MySql Workbench. A.1 Coding and cleaning the I2 database The records in this database were subsequently cleaned and coded to facilitate analysis. This included editing the addresses into common organisations and coding these organisations into different sectors and continents. Because the addresses appearing in the database were entered by the paper?s authors, the address of the same organisation would be written differently by different authors. This was manually edited. The level used to define an organisation in this analysis was fairly broad; for example hospital and university centres and departments were subsumed within the    195  larger hospital and university. Organisations with different physical locations, for example Gilead Sci Inc, Cambridge, England and Gilead Sci Inc, Foster City, USA, were treated as separate organisations.   Organisations were also coded by sectors: hospital, university, firm, government, and non-governmental organisation (NGO). To increase reliability, coding was conducted independently by me and a second individual based on guidelines that I developed.  Sectoral affiliation was first determined based on the name of the organisation, for example, organisations with hospital in their title were classified as hospital. When organisational names were unclear, organisations were researched online to determine their sector. Additional rules of thumb were applied. Academic teaching hospitals were classified as hospitals. Small private medical practices and clinics were classified as firms. Organisations, such as Genome Canada, that were non-profit but followed explicit government mandates and were funded almost exclusively through one branch of government, were classified as governmental.  Large, multi-hospital administration systems were coded as government or firm.  This was only really an issue in the USA. There, New Haven Healthcare System and Sunnybrook Health were coded as firms while Veteran Affairs administrative units were coded as government. Public research institutes with highly limited teaching capacity were coded as government. Organisations with non-profit status were coded as NGOs. Consortiums of organisations, generally non-profit in nature, were coded as NGOs. The Victorian Partnership for Advanced Computing, Australia, serves as an example of this type of consortium. The results of the two independent coding processes showed almost perfect intercoder reliability with no significant difference in our coding (?=.809, p=.000). We coded 1161 organisations (13 organisations were discounted from the Kappa calculations because its organisational coding was left blank by one coder). Of these organisations, both coders agreed on the coding of 983 organisations and disagreed on the coding for 178 organisations. For the organisations where sectoral coding did differ, we discussed each organisation together and came to an agreement regarding the proper sectoral coding of that organisation. The cities and countries used in the addresses of each paper were further used to map organisations to their respective city,    196  country, and continent based on the United Nations Geoscheme definitions of continents. This dataset is used in Chapter 3. A.2 Adding data to the database I later expanded analysis to include 2010 and 2011 in order to capitalize on changes to ISI Web of Science protocols and connect authors to organisations. Extraction from the ISI Web of Science used the same definition of Vancouver, generalist and specialist journals, and keywords outlined above. I manually cleaned address fields and coded 450 new organisations myself based on the same guidelines that I had previously developed. This created the final I2 database shown in Figure 10. Based on discussions with a local I2 expert, three types of papers?articles, letters and reviews?were chosen to represent R&D collaborations. As shown in Figure 10, this led to 408 papers being discarded.  A subset of this dataset, based on 2008-2011, was used in Chapter 2. Additional details of variables and analytic techniques used in quantitative analysis can be found in Chapters 2 and 3.        197  A.3 References Bordons, M., & Zulueta, M. (1997). Comparison of research team activity in two biomedical subfields. Scientometrics, 40(3), 423?436. Creswell, J. W. (2007). Qualitative inquiry and research design (2nd ed.). London: SAGE Publications Inc. Greene, J. C., Caracelli, V. J., & Graham, W. F. (1989). Toward a conceptual framework for mixed-method evaluation designs. Educational evaluation and policy analysis, 11(3), 255?274. Lewison, G. (1999). The definition and calibration of biomedical subfields. Scientometrics, 46(3), 529?537. Montori, V. M., Wilcynski, N. L., Morgan, D., & Haynes, R. B. (2005). Optimal search strategies for retrieving systematic reviews from Medline: Analytical survey. British medical journal, 330(7482), 68?71. Pestana, A. (1992). Spanish performance in life sciences: A comparative appraisal of the scientific production of Spain and five other European countries in 1989. Scientometrics, 24, 95?114. Sandstrom, A., Pettersson, I., & Nilsson, A. (2000). Knowledge production and knowledge flows in the Swedish biotechnology innovation system. Scientometrics, 48(2), 179?201. Yin, R. K. (2003). Case study research: Design and methods (3rd ed.). Thousand Oaks, California: SAGE Publications Inc.       198  Table 16: Specialist journals used in creating dataset Journal Title Source Acta Microbiologica Et Immunologica Hungarica Ulrich Advances In Immunology Essential AIDS Essential AIDS Patient Care And Stds Ulrich Aids Research And Human Retroviruses Essential Allergologia Et Immunopathologia Ulrich Allergologie Ulrich Allergy Essential Allergy And Asthma Proceedings Ulrich American Journal Of Infection Control Ulrich American Journal Of Reproductive Immunology Essential Annals Of Allergy, Asthma, & Immunology Ulrich Annual Review Of Immunology Essential Antiviral Research Ulrich Antiviral Therapy Ulrich Applied Immunohistochemistry & Molecular Morphology Essential Archivum Immunologiae Et Therapiae Experimentalis Essential Autoimmunity Essential Autoimmunity Reviews Essential BMC Infectious Diseases Ulrich BMC Immunology Essential Brain, Behavior, And Immunity Ulrich Canadian Journal Of Infectious Diseases & Medical Microbiology Ulrich Cellular & Molecular Immunology Essential Cellular Immunology Essential Central European Journal Of Immunology Essential Chemical Immunology Essential Clinical & Developmental Immunology Essential Clinical & Diagnostic Laboratory Immunology Essential Clinical & Experimental Allergy Essential Clinical & Experimental Immunology Essential Clinical & Vaccine Immunology Essential Clinical Immunology Ulrich Clinical Infectious Diseases Ulrich Clinical Reviews In Allergy & Immunology Ulrich Critical Reviews In Immunology Essential Current HIV Research Ulrich Current Opinion In Immunology Essential    199  Journal Title Source Current Opinion In Infectious Diseases Ulrich Current Topics In Microbiology And Immunology Essential Developmental And Comparative Immunology Essential Diagnostic Microbiology And Infectious Disease Ulrich Emerging Infectious Diseases (Online) Ulrich Epidemiology And Infection Ulrich European Journal of Immunogenetics Essential European Journal of Immunology Essential European Journal of Inflammation Essential Exercise Immunology Review Essential Expert Review of Anti-Infective Therapy Ulrich Expert Review of Clinical Immunology Essential Expert Review of Vaccines Essential Fems Immunology And Medical Microbiology Essential HIV Medicine Ulrich Human Immunology Essential Human Vaccines Essential Hybridoma Essential Hybridoma and Hybridomics Essential Immunity Essential Immuno-Analyse & Biologie Specialisee Essential Immunobiology Essential Immunogenetics Essential Immunologic Research Essential Immunological Investigations Essential Immunological Reviews Essential Immunologist Essential Immunology Essential Immunology & Allergy Clinics Of North America Essential Immunology & Cell Biology Essential Immunology Letters Essential Immunopharmacology Essential Immunopharmacology & Immunotoxicology Essential Immunotherapy Essential Indoor & Built Environment Essential Infection Essential Infection & Immunity Essential Infectious Disease Clinics of North America Ulrich Inflammation Essential Inflammation Research Essential    200  Journal Title Source Innate Immunity Essential International Archives Of Allergy & Immunology Essential International Immunology Essential International Immunopharmacology Ulrich International Journal of Immunogenetics Essential International Journal of Immunopathology And Pharmacology Ulrich International Journal of Immunopharmacology Essential International Journal of Immunotherapy Essential International Journal of Infectious Diseases Ulrich International Journal of STD & AID S Ulrich International Reviews of Immunology Essential Iranian Journal of Immunology Essential JAIDS-Journal of Acquired Immune Deficiency Syndromes Essential Journal of Allergy Ulrich Journal of Allergy & Clinical Immunology Ulrich Journal of Autoimmunity Essential Journal of Clinical Immunology Essential Journal of Clinical Virology Ulrich Journal of Endotoxin Research Essential Journal of Hospital Infection Ulrich Journal of Immunoassay Essential Journal of Immunoassay & Immunochemistry Essential Journal of Immunological Methods Essential Journal of Immunology Essential Journal of Immunotherapy Essential Journal of Immunotoxicology Ulrich Journal of Infection Essential Journal of Infectious Diseases Essential Journal of Inflammation-London Essential Journal of Innate Immunity Essential Journal of Interferon And Cytokine Research Essential Journal of Investigational Allergology And Clinical Immunology Ulrich Journal of Leukocyte Biology Essential Journal of Microbiology Immunology And Infection Essential Journal of Reproductive Immunology Essential Mabs Essential Medecine Et Maladies Infectieuses Ulrich Mediators of Inflammation Essential Medical Microbiology And Immunology Ulrich    201  Journal Title Source Microbes & Infection Essential Microbial Pathogenesis Essential Microbiology & Immunology Ulrich Molecular Immunology Essential Mucosal Immunology Essential Nature Immunology Essential Nature Reviews Immunology Essential Parasite Immunology Essential Scandinavian Journal of Immunology Essential Scandinavian Journal of Infectious Diseases Essential Seminars In Immunology Essential Seminars In Immunopathology Ulrich The Lancet Infectious Diseases Ulrich Transplant Immunology Ulrich Transplant Infectious Disease Essential Trends In Immunology Essential Vaccine Essential Vector-Borne & Zoonotic Diseases Essential Veterinary Immunology & Immunopathology Ulrich       202  Table 17: Generalist journals used in creating dataset Title Type Impact Factor (2010) Advances in Clinical Chemistry Medicine 3.406 Advances in Experimental Medicine And Biology Medicine 2.02 American Journal of Managed Care Medicine 2.737 American Journal of Medicine Medicine 4.466 American Journal of Preventive Medicine Medicine 4.235 Amyloid: The International Journal of Protein Folding Disorders Medicine 2.115 Anais Da Academia Brasileria De Ciencias Science 1.074 Annals of Clinical Biochemistry Medicine 1.917 Annals of Family Medicine Medicine 4.13 Annals of Internal Medicine Medicine 16.225 Annals of Medicine Medicine 4.246 Annals of the New York Academy of Sciences Science 2.67 Annual Review of Medicine Medicine 9.94 Archives of Internal Medicine Medicine 9.813 Archives of Medicial Research Medicine 1.884 Archives of Pathology Laboratory Medicine Medicine 2.558 Biomedicine Pharmacotherapy Medicine 2.238 BMC Medicine Medicine 3.985 British Journal of General Practice Medicine 2.442 British Medical Bulletin Medicine 2.9 British Medical Journal Medicine 13.66 Canadian Medical Association Journal Medicine 7.271 Cancer Gene Therapy Medicine 3.126 Cell Transplantation Medicine 5.126 Chinese Science Bulletin Science 0.917 Cleveland Clinic Journal of Medicine Medicine 2.149 Clinica Chimica Acta Medicine 2.535 Clinical Biochemistry Medicine 2.019 Clinical Chemistry Medicine 6.263 Clinical Chemistry and Laboratory Medicine Medicine 1.886 Clinical Science Medicine 3.982 Clinical Trials Medicine 1.917 Cochrane Database of Systematic Reviews Medicine 5.653 Complexity Science 0.948 Critical Reviews in Clinical Laboratory Sciences Medicine 4.48 Current Medical Research and Opinion Medicine 2.498 Current Molecular Medicine Medicine 5.096    203  Title Type Impact Factor (2010) Current Opinion in Molecular Therapeutics Medicine 3.452 Current Science Science 0.782 Cytotherapy Medicine 2.204 Discrete Dynamics in Nature and Society Science 1.577 Disease Markers Medicine 2.026 European Journal of Clinical Investigation Medicine 2.643 Experimental & Molecular Medicine Medicine 2.304 Experimental Biology & Medicine Medicine 2.635 Experimental Diabetes Research Medicine 2.574 Experimental Hematology Medicine 3.106 Expert Opinion on Biological Therapy Medicine 3.215 Gene Therapy Medicine 4.745 Herald of the Russian Academy of Sciences Science 0.552 HFSP Journal Science 1.6 Human Gene Therapy Medicine 4.202 Internal and Emergency Medicine Medicine 2.371 Internal Medicine Journal Medicine 1.786 International Journal of Bifurcation and Chaos Science 0.918 International Journal of Clinical Practice Medicine 2.245 International Journal of Molecular Medicine Medicine 1.98 JAMA Medicine 28.899 Journal of Biological Regulators & Homeostatic Agents Medicine 3.167 Journal of Biomedical Science Medicine 2.007 Journal of Bone & Mineral Metabolism Medicine 1.894 Journal of Cellular & Molecular Medicine Medicine 5.228 Journal of Clinical Investigation Medicine 15.387 Journal of Experimental Medicine Medicine 14.505 Journal of Gene Medicine Medicine 2.968 Journal of General Internal Medicine Medicine 2.654 Journal of Internal Medicine Medicine 5.942 Journal of Medicine Medicine 5.054 Journal of Molecular Medicine Medicine 5.004 Journal of Pain and Symptom Management Medicine 2.423 Journal of the American Board of Family Medicine Medicine 2.106 Journal of the Royal Society Interface Science 4.241 Journal of the Royal Society of New Zealand Science 0.846 Journal of Translational Medicine Medicine 3.407 Journal of Women?s Health Medicine 1.77 Laboratory Investigation Medicine 4.602    204  Title Type Impact Factor (2010) Lancet Medicine 30.758 Laryngoscope Medicine 2.018 Life Sciences Medicine 2.56 Mayo Clinic Proceedings Medicine 4.973 Medical Clinics of North America Medicine 2.182 Medical Journal of Australia Medicine 2.894 Melanoma Research Medicine 2.061 Molecular Aspects of Medicine Medicine 6.492 Molecular Genetics and Metabolism Medicine 2.897 Molecular Medicine Medicine 5.02 Molecular Therapy Medicine 6.239 Nanomedicine: Nantechnology, Biology and Medicine Medicine 5.44 Nature   Science 34.48 Nature Medicine Medicine 27.136 Naturwissenschaften Science 2.316 New England Journal of Medicine Medicine 47.05 Orphanet Journal of Rare Diseases Medicine 5.825 Pain Medicine Medicine 2.393 Palliative medicine Medicine 2.031 PLOS Medicine Medicine 13.05 Preventive medicine Medicine 3.172 Proceedings of the National Academy of Sciences Science 9.432 Progress in Natural Science Science 0.704 QJM: An International Journal of Medicine Medicine 2.627 Scandinavian Journal of Primary Health Care Medicine 2.205 Science Science 29.747 South African Journal of Science Science 0.506 Statistics in Medicine Medicine 1.99 Stem Cell Reviews Medicine 5.083 Stem Cells and Development Medicine 4.146 The Scientific World Journal Science 1.658 Therapeutic Drug Monitoring Medicine 2.429 Translational Research Medicine 2.062 Trends in Molecular Medicine Medicine 11.049 Trials Medicine 2.02 Vaccine Medicine 3.616 Wound Repair and Regeneration Medicine 2.781 Xenotransplantation Medicine 2.711    205       206  Table 18: Keywords used in creating the database antibio* antigen* autoimmu* coinfect* co-infect* FOXP3* HLA* IFN* immun* Not immunohistochemical* infect* inflammat* influenza interferon* interleukin* LL-37 monocyt* pathogen* thymic thymocyte* vaccin* viral virolog* viru* allergi* asthma* abacavir antiretroviral* cryptococc* enfuvirtide HAART efavirenz* HIV* nevirapine* Chlamydia coxsackievirus* hepatitis* leishmania* Mycobacterium pneumococcal pneumoni*    207  Salmonella* SARS staphylocc* streptococc* tuberculosis       208  9 Appendix B: Summary of participants interviewed (All names are pseudonyms) Name Home Sector Ed Role  Summary of work Adele NGO PhD Administration Meets with professors about potential projects, builds teams for research projects.  Andy Hosp PhD ServiceResearcher Involved in infectious disease diagnostic and research lab. Research based on lab results, patient samples, cell lines and virus cultures. Involved in clinical trials.  Arlene Uni PhD Researcher Public health researcher focused on how infectious disease patients use care based on surveys and focus groups. Also teaches.  Ben Hosp MD ServiceResearcher Clinician with dedicated research time. His research lab is based on cell lines and parasites isolated from patients.  Bill Firm PhD Researcher Manages research side of company whose work runs the gamut from cell lines to pre-clinical trials to clinical trials.  Cedric Uni PhD Researcher His research lab focuses on genome sequencing of pathogens.  Crystal Govt MD ServiceResearcher Service work with no formal dedicated research time but research encouraged.   Donald Govt PhD Administration Currently retired. Previously helped to administer a health authority.  Eric Hosp MD Service Clinician with no dedicated research time. Involved in clinical trials or more basic patient sample based research. Felix Hosp MD/PhD Researcher His research lab is focused on a specific disease based on patient samples.  Gauis Hosp PhD Researcher His research lab is based almost    209  Name Home Sector Ed Role  Summary of work entirely on patient samples.  Helo Hosp MD/PhD ServiceResearcher Clinician with unofficial dedicated research time. Research based on patient data, involved in clinical trials and epidemiological studies.  Hoyt Hosp MD ServiceResearcher Clinician with dedicated research time. His research lab focuses on a specific disease and ranges from mouse models to clinical trial involvement.  Jason Govt MD ServiceResearcher Service work mainly focused on administration. His research lab studies a specific disease mainly in animal models.  Jessica Hosp PhD Researcher Genetic epidemiologist focused on effect of genetics on disease based on patient data.  Kara NGO PhD Administration Manages research projects, builds research teams.  Kima Uni PhD Researcher Environmental hygienist. Keeps database of community based hazards.  Lafayette Govt MD ServiceResearcher Conducts literature reviews and contracts out research to find best evidence to inform policy decisions.  Lane Firm Bsc Administration Spends majority of time in meetings and nurturing relationships.  Laura Uni Msc ResearchSupport Within research lab, analyses cell lines using a mass spectrometer.  Lee Uni PhD Researcher His research lab studies a specific disease using bacteria models.  Meghan Govt MBA Administration Helps administers a health authority.  Pam Hosp Bsc Student PhD research analyses immune responses based on patient data.     210  Name Home Sector Ed Role  Summary of work Peggy Firm Bsc Administration Involved in everything in her company from IP negotiations to academic meeting posters to negotiating contracts.  Pete Firm Msc Administration NA.  Rene Govt MD ResearchSupport Manages a database of infectious diseases. Responsible for report generation used in service and research work.  Rhonda Hosp RN Administration Administers research centre.  Russell Uni MD/PhD ServiceResearcher Conducts systematic reviews of existing research.  Sam Hosp MD Service Clinical service with no official dedicated research time. Involved in clinical trials related to service and more basic research based on samples collected from service.  Samuel Firm PhD Administration Spends majority of time in meetings and nurturing relationships.  Saul Hosp MD Service Clinician with no dedicated research time. Helps to run tissue bank and works on research projects related to tissue bank samples.  Sharon Hosp MD/PhD ResearchSupport In research lab uses human descriptive statistics, patient samples and animal models to study immunity.  Sookie Hosp PhD Researcher Her research lab looks at cell lines, animal models and patient samples.  Tara Firm PhD ServiceResearcher Works as psychologist in private practice. Previously worked as hospital psychologist with dedicated research time interviewing patients regarding drug adherence.     211  Name Home Sector Ed Role  Summary of work Terry Govt PhD ServiceResearcher Runs a diagnostic lab. No formal dedicated research time but research encouraged and based on diagnostic lab and additional samples.   Tommy Hosp PhD Researcher Public health researcher studying how patients use health services using interviews and database analysis.  Virgil Hosp MD/PhD ServiceResearcher Clinician with dedicated research time. In his research lab studies cell lines, animal models and patient samples.  William Hosp Msc ResearchSupport In a research lab studies cell lines, animal models and patient samples.        212  10 Appendix C: Sample interview guide and related sociograms Introduction of myself including my name, program, and working name of my dissertation:  how individuals involved in I2 research collaborate in Vancouver.  Definition of collaboration used in study: Most of the questions that I?m going to ask you are about your collaborations. People often think of different things when they hear the word ?collaboration.? For this interview, I?m not just interested in the people who you have formal research collaborations with but also the people who you work with and interact with, perhaps less formally, as part of your research. Introduction of sociogram. Explanation that picture was created for this interview, won?t be shown to others or used in publications. To explain how the picture works: black dots=articles, red dots=co-authors, size depends on number of articles they co-authored with you. Time of diagram 2004-2009. Tell participant that I?d like to use picture as basis to ask you questions about your collaborations within Vancouver, OK? Questions 1. First off, do you think that this (the sociogram) is an accurate illustration of the people who you collaborate with?  2. Are there other people who you have important collaborations with that aren?t part of this diagram? Probes a. People who you work on grants with? b. People who you talk to on a regular basis about your work? Ask advice from? c. People who you work on projects with? 3. Can you give me an example of a collaboration you have that you think works well? Probes: a. How do you collaborate with ()? b. In what ways did you collaborate with ()? c. How did you decide to collaborate with ()? d. What do you think makes the collaboration work well?  i. Specific policies from your organisation? ii. Specific policies from funders?    213  iii. Other policies? iv. Cultural norms? i.e. words for things, standards of evidence, times of working, objectives for working together? v. Similar goals/interests e. Is there anything that you think makes the collaboration more challenging?  f. Specific policies from your organisation? g. Specific policies from funders? h. Other policies? i. Cultural norms? 4. Can you give me an example of collaboration that you were part of that didn?t work well? Probes a. How do you collaborate with ()? b. In what ways did you collaborate with ()? c. How did you decide to collaborate with ()? d. What do you think made the collaboration challenging?  i. Specific policies from your organisation? ii. Specific policies from funders? iii. Other policies? iv. Cultural norms? i.e. words for things, standards of evidence, times of working, objectives for working together? v. Similar goals/interests 5. Can you tell me about any collaborations that you have with people in hospitals? a. Common reason for collaborating? b. Are they generally clinicians/clinician-scientists/scientists? 6. Can you tell me about any collaborations that you have with people in firms? a. Common reason for collaborating? b. Are they generally clinicians/clinician-scientists/scientists? 7. Can you tell me about any collaborations that you have with people in government?  a.  Common reason for collaborating? b. Are they generally clinicians/clinician-scientists/scientists?    214  8. Can you describe a typical week day to me? Shifting gears bit to talk about your work a bit more generally 9. What is your top goal for the next 6 months? 10. What is your top goal for the next year?  I know we?re almost out of time, but I wanted to ask you a few basic questions before we finished:  11. So you have a (PhD, MD, Msc)? Any other degrees? Where did you get your degrees? 12. What organisation do you consider to be your ?home?? 13. What organisation do you list on your publications? 14. Are there other organisations that you are affiliated with? 15. What is your official title here? 16. How long have you been working here? 17. Who pays your salary? 18. Who pays for your research? 19. What was your previous job? 20. You yourself are a (PhD/MD/other) are there other (PhDs/MDs/whatever they aren?t) that you work with on a daily basis here?  Thank-you once again for taking the time to speak with me. Do you have any questions yourself or anything else that you?d like to say that you think would be important for my study?        215  Figure 11: Two-mode egonet of participants interviewed and co-authors connected by papers   Here red circle represent co-authors and black squares represent papers. The relative size of different co-authors are based on their two-mode degree. The participant interviewed appears as the large circle in the middle. Egonets shown during interviews would have included the full name of the participant and first initial, last name of all co-authors. These were excluded from this sample egonet to help protect anonymity.    216  Figure 12: Two-mode sociogram of organisations affiliated with participant and co-authors connected by papers    Here red circles represent the organisational affiliations of authors for each paper. Black squares represent papers. Size of both squares and circles are based on two-mode degree calculations. This sociogram was created in tandem with Figure 11 but was only shown to participants if they began discussing the organisational affiliations and collaborations associated with Figure 11.      217  11 Appendix D: Coding list used in analysis Meta Code Meta Code Details Code Roles: Description of individuals role within the system While it is difficult to generalise the roles that all individuals interviewed played within the system, there seemed to be a few common roles. There were 'basic' academic researchers who had always worked in hospitals. There were clinical-researchers with their foot in the clinic and another foot in basic research, and there were 'pure' clinicians who were generally not interested in being involved in research. There were individuals who controlled access to a specific source of patient samples who had longstanding research collaborations with more basic researchers. ? Administration ? Researcher ? Service/Researcher ? Service ? Research Support ? Student Network overview The network in question resides in Canada. A relatively small country, researchers working within a specific research area such as I2 feel like they know everyone else within their city working within this area and likely within their country. Within this network, the UBC hospital appears to play a relatively unimportant role. In addition there has been a weakening of Vancouver's biotechnology sector, perhaps since as early as 2000. Hospitals within this system appeared to play two roles, one was as a direct collaborator on the more basic research side and the other was as a venue for clinical trials, here accessed through clinical research organisations. As discussed in the access to patients theme, colocation of researchers in hospitals appears to play a key role in helping to gain access to patient data. Other academic researchers worked in hospitals but seemed to feel like they could just as easily work in a university setting.      218  Meta Code Meta Code Details Code Orgs: Key organisations within network   ? Private Sector ? UBC ? UBC Hospital ? SFU ? BCCDC ? St. Pauls ? CFRI ? BC Cancer ? CDRD ? PHSA ? BC Ministry of Health ? VGH  ? Women?s and Childrens  Research: Description of types of research conducted: Overview and one for each type of research Differentiation seems to exist in the eyes of the individuals interviewed between different 'types' of research that can be seen as running a spectrum from basic (based on cell lines and animal models), translational (essentially basic research conducted on patient samples/data) clinical (clinical trials) and public health. I have decided to differentiate between translational and commercialisation because the focus is on a commercialisable product. ? Basic ? PHEpi ? Translational ? Clinical Trials ? Reviews ? Commercialisation Facilitating Collaboration Meetings were a major vehicle that was listed in facilitating collaborations. Specific meetings within organisations were designed to bring together academics and clinicians around a certain topic. Informal disease specific research groups also aided in bringing together individuals in the province around a specific topic. NCEs can play an important role in this, bringing people together around a theme. Collaboration is seen as being 'trendy' for research funding bodies right now and may be added to an application to increase its chances of being funded. (See also shaping research agendas.) Publication/collaboration guidelines were sometimes seen as important for helping collaborations run smoothly. Jason talks about finding a collaborator by reviewing their grant for UBC.      219  Meta Code Meta Code Details Code Gatekeepers Different individuals acted as gatekeepers within this system. Clinician-scientists, nurses and clinical directors acted as gatekeepers to access more 'pure' clinicians and patient data. PIs also generally acted as gatekeepers for collaborations with any/all individuals within their labs.   Institutional structures Institutional structures appeared to act as a barrier in multiple ways. The affiliation and title that clinicians had with UBC greatly affected their ability to conduct research and apply for research grants. In addition, this study interviewed 3 individuals who were unable to work in their area where they were trained in China because of difficulty in certification in Canada, specifically related to MDs. These individuals were now all working in research related to their clinical training.  ? As barrier ? Between UBC and other public organisations ? Immigrants barred from previous jobs Motivations to collaborate: Collaboration a way to get a ?big? publication. Gaining access to patient samples is an important reason that individuals gave to collaborate. Within these collaborations, 'researchers' often did not play a role and the goal of the collaboration was not 'research.' Instead, nurses, research co-ordinators, and directors of medical centres often played important roles. Working within a hospital often gave researchers an 'edge' in this pursuit as hospitals acted as the data source for patient data. Specific people: clinician-scientists or directors of specific clinical centres, often acted as gatekeepers within this process (see gatekeeper). While other collaborations were often global, collaborations that involved access to patient samples were generally highly localised. Collaboration based on complementary expertise/gain access to a resource you don?t have access to yourself (knowledge vs. technique vs. resource) patients and potentially publications as a subtheme. ? Big publications ? Access to patients ? Complementary access    220  Meta Code Meta Code Details Code Norms and goals For firms, the bottom line appeared to be the main goal with product development mainly focused on the USA. For firms, publications appeared to play an important role as third party verification, useful in generating funding and other forms of buy-in. However, protecting intellectual property remained paramount. Publication could only occur after patents were secured and if intellectual property would not become violated. Collaborations with industry were based always on contracts, often with specific milestones. If a contract could not be agreed on, a collaboration would not occur. Increased privacy. Hospitals, universities, and BC Cancer were seen by some as all part of the academic community: collaborations happen informally between these groups. For individuals within this community, authorship order was often seen as quite important as it formed the basis of reputations. Bad collaborations were often linked to the violation of scientific norms, such as academic fraud. In addition, 'true' collaborations were seen as intellectually based, involving individuals from outside of one?s own lab. Authorship was important for reputations, getting tenure and also securing grant money, something that was essential in order to keep labs running. Getting grants was seen as an important goal for university academics. 'Official' goals of academia seem to be education, training and research, although in practice everyone talks about their research goals. ? Private sector ? Academic community ? Government service organisations    221  Meta Code Meta Code Details Code Tensions in goals Tensions between different sectoral goals could affect collaborations between hospitals, universities, government and firms. Often these tensions involved a divergence between corporate goals (bottom line) and academic goals (producing papers). This was especially acute when firms felt the need for secrecy in order to protect their intellectual property. Labs in universities and firms often different in their hierarchical structure, often being more 'flatly' structured in universities. Capital investments also differ in labs between these two organisational types: purchases in university based labs are often based on the PI's discretion, provided they can secure a capital grant. Within firms, capital purchases can often happen more quickly with less bureaucracy, but are more tightly tied to their bottom line. UILOs play an interesting role in the tension of goals between the public and private sectors because they are located in public organisations but are seen as embodying private goals. Some people interviewed found this role useful in helping public sector employees navigate unknown goals, others found that it made things worse, and that they shouldn't have a revenue generating objective. The public sector had taken on the wrong 'part' of private sector goals, the focus on revenue generation and IP protection as opposed to use of research. ? Public and private sector ? Between academic and application ? Within public sector Shaping research agendas Through funding, foundations and the government play an important role in shaping research agendas. Certain research themes cannot continue without funding. Current research funding bodies? preferences towards collaboration may thus increase collaboration at least at the symbolic level.   Universality of techniques Universality of techniques even when knowledge area differs.   Egonet verification Methodological comment on whether egonet accurately reflected collaborations   Pathway   ? Research to policy ? Research to practice ? Research to product   Experiments of nature        222    12 Appendix E: Acronyms Acronym Name BCCA BC Cancer Agency BCCDC British Columbia Centre for Disease Control CFRI Child and Family Research Institute CMMT Centre for Molecular Medicine and Therapeutics CWHC Child and Women's Health Centre I2 Infection and immunity IP Intellectual property IS Innovation systems NGO Non-governmental organisation PHSA Provincial Health Services Authority R&D Research and experimental development SFU Simon Fraser University TDO BC Cancer Agency Technology Development Office UBC University of British Columbia UILO UBC University-Industry Office VCH Vancouver Coastal Health VGH Vancouver General Hospital  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0074093/manifest

Comment

Related Items