Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Dimensions of educational stratification : non-standard employment, workplace task discretion, and educational… Pullman, Ashley Michelle 2017

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

Item Metadata

Download

Media
24-ubc_2017_november_pullman_ashley.pdf [ 3.41MB ]
Metadata
JSON: 24-1.0355840.json
JSON-LD: 24-1.0355840-ld.json
RDF/XML (Pretty): 24-1.0355840-rdf.xml
RDF/JSON: 24-1.0355840-rdf.json
Turtle: 24-1.0355840-turtle.txt
N-Triples: 24-1.0355840-rdf-ntriples.txt
Original Record: 24-1.0355840-source.json
Full Text
24-1.0355840-fulltext.txt
Citation
24-1.0355840.ris

Full Text

DIMENSIONS OF EDUCATIONAL STRATIFICATION: NON-STANDARD EMPLOYMENT, WORKPLACE TASK DISCRETION, AND EDUCATIONAL BELIEFS  by Ashley Michelle Pullman  B.A., Simon Fraser University, 2009 M.A., University of Victoria, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (EDUCATIONAL STUDIES)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   September 2017 © Ashley Michelle Pullman, 2017 ii  Abstract  My dissertation consists of three distinct yet interrelated studies. Its purpose is to extend research and theory on inequality by investigating three educational outcomes: non-standard employment, workplace task discretion, and intrinsic and extrinsic educational beliefs. As a body of work, my research generates insight into how the level and type of educational attainment affect divergent life course pathways. The first study examines gender inequality in early career part-time and temporary employment in Canada. Through two types of decomposition analyses, I research non-standard employment across four cohorts graduating between 1990 and 2010, studying the extent to which gender stratification within fields of study or systemic employment inequality contribute to dissimilar outcomes. I find that rates of non-standard employment vary substantially across disciplines. Furthermore, the over and under-representation of women in certain fields is a main factor explaining gender differences in temporary employment but cannot fully account for disparities in part-time employment. The second study researches the relationship between education and workplace task discretion in 30 countries. Through regression and decomposition analyses, I examine the direct association between education and task discretion and the extent to which skill and occupational sector function as mediators. I compare individual-agency and critical-institutional theoretical perspectives as explanations for direct and indirect associations. My findings mainly support critical-institutional accounts and yield evidence of a relative relationship between education and task discretion. That is, in contexts where task discretion is higher overall and more equal among occupations, education, skill, and occupational sector are less significant mechanisms of stratification.  The third study considers how intrinsic and extrinsic educational beliefs change over adulthood. My research is based on a longitudinal study that repeatedly surveyed the same graduating British Columbia high school cohort over 28 years. Through hierarchical growth modelling, I contrast demographic and experience-based explanations to consider the influence of social origin and individual education and employment participation over time. The findings suggest that both life course experiences and social position have an influence on initial educational beliefs in early adulthood and the rate of change over time. Additionally, educational beliefs are more variable in early adulthood and become more stable later in participants’ life courses.   iii  Lay Summary  My dissertation consists of three distinct yet interrelated studies of the relationships among education, employment, and social outcomes. The first study examines how rates of early career part-time and temporary employment differ among men and women graduating with bachelor’s degrees between 1990 and 2010 in Canada. The second study investigates how education credentials influence workplace task discretion in 30 countries through improving literacy skills and providing access to occupations where discretion is possible. The third study researches how gender, parental education, time spent in education, and months of employment affect change in educational beliefs over adulthood. Together my research generates insight into how the level and type of educational attainment results in unequal outcomes among individuals and across different contexts.  iv  Preface  The research presented in Chapter 3 is independent scholarship. I conducted all analysis at the British Columbia Inter-University Research Data Centre, which is part of the Canadian Research Data Centre Network (CRDCN). The Research Data Centres Program approved the project [certificate #15-SSH-BCI-4579]. The views expressed in this chapter do not necessarily represent those of the CRDCN or its partners. Chapter 4 was partially co-written and conceptualized with Janine Jongbloed (Université Bourgogne Franche-Comté). As the lead author, I conducted all statistical analysis and undertook approximately 75% of the writing and revisions. The research conducted in Chapter 4 is based on public use data of the International Assessment of Adult Competencies (PIAAC) generated by OECD member countries. The opinions and arguments expressed in this chapter do not necessarily reflect the official views of the OECD member countries. Chapter 5 employs data collected by Lesley Andres (University of British Columbia) as part of the Paths on Life’s Way project. Co-authorship of Chapter 5 is in recognition of Dr. Andres as principal investigator of the Paths project and her 28 years of longitudinal research and scholarship. Together, we conceptualized the direction of Chapter 5 and I assumed the role of lead author by conducting all analysis and writing. v  Table of Contents  Abstract .......................................................................................................................................... ii Lay Summary ............................................................................................................................... iii Preface ........................................................................................................................................... iv Table of Contents ...........................................................................................................................v List of Tables .............................................................................................................................. viii List of Figures ............................................................................................................................... ix List of Abbreviations .....................................................................................................................x Acknowledgements ...................................................................................................................... xi Dedication ................................................................................................................................... xiii Chapter 1: Introduction ................................................................................................................1 1.1 Purpose of Dissertation ................................................................................................... 1 1.2 Theorizing and Researching Inequality in Educational Outcomes ................................. 3 1.3 Outline of Dissertation .................................................................................................... 7 Chapter 2: Inequality of Educational Opportunities and Outcomes in Canada ...................10 2.1 Introduction ................................................................................................................... 10 2.2 Systems of Canadian Education From an International Comparative Perspective ....... 11 2.3 Educational Attainment and Intergenerational Mobility .............................................. 15 2.4 Educational Distinctions and Horizontal Inequality ..................................................... 20 2.5 Inequality in Educational Outcomes ............................................................................. 25 2.6 Conclusion .................................................................................................................... 28 Chapter 3: Gendered Pathways From School to Work: The Association Between Field of Study and Non-Standard Employment Outcomes....................................................................31 3.1 Introduction ................................................................................................................... 31 3.2 Literature Review.......................................................................................................... 35 3.2.1 Non-standard employment and inequality ................................................................ 35 3.2.2 Gender inequality in non-standard employment ....................................................... 37 3.2.3 Gender inequality in non-standard employment among recent graduates ................ 39 3.3 Research Rationale, Questions, and Methodological Approach ................................... 41 3.3.1 Data ........................................................................................................................... 43 vi  3.3.2 Measures ................................................................................................................... 43 3.3.3 Analysis..................................................................................................................... 46 3.4 Findings......................................................................................................................... 48 3.4.1 Descriptive findings .................................................................................................. 48 3.4.2 The mediation effect of field of study....................................................................... 53 3.4.3 Assessing the contributing factors to non-standard employment ............................. 57 3.5 Discussion ..................................................................................................................... 60 3.6 Conclusion .................................................................................................................... 65 Chapter 4: The Relationship Between Education and Workplace Task Discretion: An International Comparative Perspective .....................................................................................67 4.1 Introduction ................................................................................................................... 67 4.2 Literature Review.......................................................................................................... 69 4.3 Research Overview and Design .................................................................................... 73 4.3.1 Data ........................................................................................................................... 75 4.3.2 Methodological approach.......................................................................................... 77 4.3.3 Dependent variable ................................................................................................... 78 4.3.4 Independent variables ............................................................................................... 79 4.4 Results ........................................................................................................................... 81 4.4.1 Assessing the direct relationship between education and task discretion ................. 81 4.4.2 Assessing the indirect relationship between education and task discretion .............. 90 4.5 Discussion ..................................................................................................................... 97 4.6 Conclusion .................................................................................................................... 99 Chapter 5: A State of Mind or a State of Experience? Intrinsic and Extrinsic Educational Beliefs Over the Life Course .....................................................................................................102 5.1 Introduction ................................................................................................................. 102 5.2 Literature Review........................................................................................................ 104 5.2.1 Defining intrinsic and extrinsic beliefs ................................................................... 104 5.2.2 The cultivation and maintenance of educational beliefs ......................................... 106 5.3 Research Design and Overview .................................................................................. 109 5.3.1 Research questions .................................................................................................. 109 5.3.2 Data ......................................................................................................................... 111 5.3.3 Dependent and independent measures .................................................................... 112 vii  5.3.4 Analysis................................................................................................................... 113 5.4 Findings....................................................................................................................... 115 5.4.1 Descriptive findings ................................................................................................ 115 5.4.2 Unconditional means models .................................................................................. 117 5.4.3 Unconditional growth models ................................................................................. 117 5.4.4 Conditional growth models ..................................................................................... 120 5.5 Discussion ................................................................................................................... 126 5.6 Conclusion .................................................................................................................. 131 Chapter 6: Conclusion ...............................................................................................................133 6.1 Contributions and Key Findings ................................................................................. 133 6.2 Data and Analytical Limitations ................................................................................. 136 6.3 Theoretical Consequences and Possibilities ............................................................... 138 6.4 Future Research .......................................................................................................... 140 6.5 Conclusion .................................................................................................................. 142 References ...................................................................................................................................144 Appendices ..................................................................................................................................175 Appendix A : Chapter 3 Variable Overview........................................................................... 175 Appendix B : Chapter 4 OLS Regression Results .................................................................. 178 Appendix C : Chapter 4 Robustness Testing .......................................................................... 179 Appendix D : Chapter 4 KHB Results .................................................................................... 180 Appendix E : Chapter 4 Estimation of Country-Level Effects ............................................... 181 Appendix F : Chapter 5 Model Equations and Parameters ..................................................... 182 Appendix G : Chapter 5 Conditional Linear and Curvilinear Growth Models ....................... 183  viii  List of Tables  Table 3.1. Profile of gender segregation, 1990/1995 and 2005/2010 cohorts .............................. 49 Table 3.2. Estimation of the relationship between gender and temporary employment as mediated by field of study ............................................................................................................................ 54 Table 3.3. Estimation of the relationship between gender and part-time employment as mediated by field of study ............................................................................................................................ 55 Table 3.4. Contribution of each field of study to the overall mediating effect between gender and non-standard employment ............................................................................................................. 57 Table 3.5. Shapley decomposition of the D-Index: Temporary employment............................... 59 Table 3.6. Shapley decomposition of the D-Index: Part-time employment ................................. 60 Table 4.1. Pooled linear regression estimation of self-reported level of task discretion .............. 82 Table 4.2. Pooled estimation of the indirect relationship between education and task discretion 91 Table 5.1. Means and standard deviations of dependent and independent variables ................. 116 Table 5.2. Variance components and inter-class correlations..................................................... 117 Table 5.3. Results of unconditional growth models ................................................................... 118 Table 5.4. Extrinsic beliefs: Results of conditional growth models ........................................... 121 Table 5.5. Intrinsic beliefs: Results of conditional growth models ............................................ 123 Table 6.1. The relationship between task discretion and education ........................................... 178 Table 6.2. The effect of education as mediated by occupational sector and literacy assessment score ............................................................................................................................................ 180 Table 6.3. Pooled estimation of country average and occupational range in task discretion ..... 181 Table 6.4. Results of linear and curvilinear growth models measuring extrinsic beliefs ........... 183 Table 6.5. Results of linear and curvilinear growth models measuring intrinsic beliefs ............ 183  ix  List of Figures  Figure 2.1. Public expenditure on education as a percentage of GDP (2012-2013) ..................... 14 Figure 2.2. Intergenerational mobility (2012-2015) ..................................................................... 18 Figure 2.3. Gender segregation by postsecondary field of study (2014-2013) ............................. 23 Figure 2.4. Women’s earnings ratio relative to men, 2012-2014 ................................................. 27 Figure 3.1. Rates of temporary and part-time employment by field of study, 1990/1995 & 2005/1010 cohorts ......................................................................................................................... 50 Figure 3.2. Predicted percentage of male and female graduates employed in temporary positions by percentage of women within a field of study, 1990-2010 cohorts ........................................... 52 Figure 3.3. Predicted percentage of male and female graduates employed in part-time positions by percentage of women within a field of study, 1990-2010 cohorts ........................................... 53 Figure 4.1. Schema illustrating the potential relationships among education and task discretion 74 Figure 4.2. Individual country OLS regression results examining the direct relationship between task discretion and education ........................................................................................................ 84 Figure 4.3. Bivariate relationship between average self-reported task discretion and range among occupational sectors ...................................................................................................................... 87 Figure 4.4. Conditional effects of credential level by country ..................................................... 89 Figure 4.5. The mediation effect of occupational sector and literacy by education level ............ 93 Figure 4.6. The indirect effect of occupation by the country-specific distribution of task discretion ....................................................................................................................................... 95 Figure 4.7. The indirect effect of literacy by the country-specific distribution of task discretion 96 Figure 5.1. Predicted probability of agreement with each belief statement generated from unconditional growth models ...................................................................................................... 119 Figure 5.2. Gender and parental education difference in predicted probabilities of agreement with extrinsic belief statements ........................................................................................................... 122 Figure 5.3. Predicted probabilities of agreement with belief statements by level of postsecondary participation ................................................................................................................................ 126 Figure 6.1. The mediation effect of numeracy and technology skills by education level .......... 179  x  List of Abbreviations  CAD – Canadian dollar CIP – Classification of Instructional Programs CMEC – Council of Ministers of Education, Canada CRDCN – Canadian Research Data Centre Network GDP – Gross domestic product ICC – Inter-class correlation ILO – International Labour Organization  ISCED – International Standard Classification of Education  ISCO – International Standard Classification of Occupations  KHB – Karlson-Holm-Breen method NGS – National Graduate Survey OECD – The Organisation for Economic Co-operation and Development OLS – Ordinary least squares  PIAAC – Programme for the International Assessment of Adult Competencies PISA – Programme for International Student Assessment PUF – Public use file SE – Standard error STEM – Science, technology, engineering, and mathematics  UNESCO – The United Nations Educational, Scientific and Cultural Organization xi  Acknowledgements  First, and most importantly, it is necessary to thank the thousands of anonymous respondents in Canada and around the world who willingly take part in governmental and social science surveys. These total strangers are the foundation of my research.  During my studies, I worked with an outstanding supervisor and committee who continually propelled me to be a stronger researcher through their thoughtful criticisms and exemplar scholarship. My supervisor Lesley Andres was my biggest advocate. She contributed to my development in many ways, providing countless forms of learning, research, and scholarship support. Alongside excellent feedback, Sylvia Fuller modeled an orientation towards quality research and scholarship. Alison Taylor was especially helpful in the later stages of my Ph.D. and Hongxia Shan aided in the first half. Staff members in the Department of Educational Studies offered what was often weekly aid throughout the last four years, notably Shermila Salgadoe, Sandra Abah, Erin Hagen, Erika Hughes, and Alexandra Wozny. Credit is due to the exceptional teachers I had at the University of British Columbia, especially Jason Ellis, Elizabeth Hirsh, Anita Hubley, Yan Liu, André Elias Mazawi, Amy Metcalfe, Hans Pechar, Taylor Webb, Amery Wu, and Bruno Zumbo in the Departments of Measurement, Evaluation, and Research Methodology, Sociology, and Educational Studies. Bruno Zumbo was a particularly prolific teacher and generated initial and lasting encouragement, insight, and inspiration for many aspects of my research. I am indebted to the research analysts at the University of British Columbia’s Inter-University Research Data Centre, including Lee Grenon, Wendy Kei, and Cheryl Fu. Wendy Kei was especially supportive in navigating confidentiality policies and procedures, ensuring access to the software and packages necessary for my research, and carefully and expeditiously releasing my research results. A portion of my research would not have been possible without the statistical training offered by the German Social Science Infrastructure Services (GESIS) and Leibniz-Institute for the Social Sciences in 2015 and 2017. Anja Perry, Jan Paul Heisig, Ralph Carstens, and Tim Daniel gave especially helpful instruction on complex survey design and the possibilities and constraints of working with plausible values and replicate weights. xii  Family and friends have been both colleagues and saviors of my doctoral studies. My friendship with Janine Jongbloed has “flourished” over the last few years as we collaborated and checked (and rechecked) each other’s work, code, and analysis from other sides of the earth. Kari Grain was a primary support over the last four years, especially in reminding me that well-being and good research go hand-in-hand. Stephanie Glick provided excellent feedback on two chapters of this dissertation and demonstrated the fine balance of taking oneself seriously and not-too-seriously at the same time. Sneha Shankar’s perseverance and passion for the art and science of modelling, measurement, and research has been especially galvanizing in the later stages of doctoral research. Not least, Chris Nichols has scrupulously edited my work (although I take ownership of all errors and the use of the word “utilize”), endlessly discussed every idea it contains, continually been my biggest champion and support, while passionately cultivating a meaningful life with me. Lastly, the funding I have received has granted me one of the greatest necessities doctoral students require: time. Thank you to the University of British Columbia’s Four Year Doctoral Fellowship for funding the first year of my Ph.D.; the Donald and Ellen Poulter and Dean of Education Scholarships for financing a powerhouse computer and the software necessary for my research; and the Social Sciences and Humanities Research Council for granting me the Vanier Canada Graduate Scholarship, which supported the last three years of my doctorate, and awarding me a postdoctoral fellowship to continue my scholarship—work that I hope continues to make societal contributions to understandings of education and inequality. xiii  Dedication       For Jayne Pullman, who taught me that all things worth doing are taken one step at a time.   1  Chapter 1: Introduction  1.1 Purpose of Dissertation My dissertation is about educational inequality, viewed through the lens of measurable social outcomes. Its purpose is to extend stratification research and theory with respect to three educational outcomes: non-standard employment, workplace task discretion, and intrinsic and extrinsic educational beliefs. Traditionally, the study of educational inequality has focused primarily upon disproportionate access to specific types and levels of schooling. Such studies continue to be necessary given the persistence of social, economic, academic, and psychosocial barriers. Extending from research on access, however, I examine how divergence in the level and type of educational attainment results in an array of diverse and unequal life course outcomes. In the chapters that follow, I provide distinct studies that examine three topics: how field of study impacts the likelihood of temporary and part-time employment two years after graduation among men and women; how education level enables access to workplace task discretion, primarily through occupational sorting; and how demographic factors and postsecondary and labour market participation influence educational beliefs over a 28-year period. With rising levels of education in Canada and many other countries worldwide, it is necessary to shift attention towards credential attainment and social and economic outcomes. Just as education levels vary, so do the outcomes they produce. Many theoretical perspectives frame the alleviation of outcome-based inequalities as a matter of equalizing individual access to 2  education—such as Rawl’s (1971) notion of distributive justice or Sen’s (1992) capability approach. Reformers advocate for equalizing access to education as a way to influence outcomes (Terzi, 2014). Often implicit within an equal opportunity principle is that any form of disparity in outcomes is based on personal ability, responsibility, or preference (Arneson, 1989; Crompton, 2008; Howe, 2015). Such perspectives risk overlooking how other circumstances and social forces contribute to unequal outcomes. Although equalizing educational access is integral to the promotion of social mobility for marginalized groups, “there is no guarantee that an individual’s educational attainment will translate unproblematically into achieved social status” (Thompson & Simmons, 2013, p. 745). Further, a consideration of educational outcomes must be multi-faceted, as there is not just one social outcome that fully captures the intricacies of social inequality. This is what my dissertation aims to do. A manuscript-based dissertation consists of distinct, free standing papers that are of journal article length. Though there are overarching conceptual similarities, each research chapter discusses contemporary theory and contextual factors related to its distinct research area. As a groundwork for inquiry, I explore a broader conceptual framework of educational inequality in Chapters 1 and 2. I discuss the concept of educational inequality and how it is dependent on both the absolute and relative levels of attainment in society at large (Hirsch, 1977/2012). I frame educational inequality as a condition and process of social stratification functioning within and beyond schooling. Furthermore, I present demographic, social, and academic factors as influences on access, attainment, and outcomes. In Chapter 2, I offer additional insight into how demographic and contextual factors contribute to distinct forms of stratification—namely, by gender, class, and social context. In summary, the aim of Chapters 1 and 2 is to advance a 3  foundational understanding of theory and research on educational inequality, a lineage from which my scholarship emerges and on which it builds. 1.2 Theorizing and Researching Inequality in Educational Outcomes Historically, North American education systems have shifted from differentiated to progressive ideals that aim to provide “a common educational experience” through public primary (e.g., elementary) and secondary schooling (Coleman, 1968, p. 11). Differentiated education refers to the understanding that certain groups require specific forms of education based on perceived notions of difference, namely race, gender, ability, and class. Contrastingly, progressive policy aims to promote equal educational opportunities for historically under-represented groups (Baker & Velez, 1996; Finnie, 2008; Wang, 2013). Implicitly promoting egalitarian perspectives based on principles of meritocracy, progressive educational ideals often attempt to “level the playing field” rather than secure equitable outcomes (Roemer, 1998, p. 1).1 Nevertheless, discourses aligning with differentiated educational principles still circulate. Attempts to promote educational equality may be constructed as impeding the pursuit of excellence, a perspective that frames the unequal distribution of educational resources as a necessary incentive for competition and innovation.  The equalization of educational opportunities is rarely a silver bullet policy solution to address inequality, given the persistence of stratification within and beyond education systems.                                                  1 Equity and equality are similar but not synonyms terms. I use the term “equality” to refer to group-based similarities that are often quantitative in nature (e.g., total level of educational attainment). Contrastingly, “equity” surrounds more qualitative components of individual and group justice and fairness (e.g., assessing the type and level of education necessary for well-being). 4  Research and theory on the conditions of social stratification examine the ways in which strata, classes, or other social hierarchies differentiate among individuals and groups. For instance, differentiation within education systems leads to dispersion and variation in credentials (Allmendinger, 1989; Arum, Gamoran, & Shavit, 2007). More than simply the construction of social groupings, stratification is a social process “in which members of a population become stratified” (Kerckhoff, 2001, p. 3). Intersecting demographic attributes (e.g., gender or age), social markers (e.g., prestige or reputation), and economic resources (e.g., property or income) influence a range of social and economic outcomes. Among other things, opportunities specific to certain occupations, groups, and locations unequally shape the value of educational attainment. A focus on inequality of educational outcomes from a social stratification perspective encompasses a consideration of not only the consequences of unequal access and educational experiences but also conditions and processes that lead to dissimilar life chances (Lefranc, Pistolesi, & Trannoy, 2008).  Social stratification research relies upon comparative and process-based frameworks (Lambert, Connelly, Gayle, & Blackburn, 2016). Comparative stratification research examines the preservation and alleviation of inequality across social groups, societies, and time periods. Notably, educational research from a stratification perspective examines the equalization of attainment and new and persistent forms of educational inequality. The aim is to examine both social change and how “inequalities persist and endure” (Bottero, 2005, p. 3). Rather than only give a descriptive overview of maintenance or change in level or types of stratification, researchers examine what produces and sustains inequality. In this sense, stratification research seeks to develop theoretical and empirical insight into the various mechanisms that lead to the 5  creation and continuance of inequality. Of note, educational ideals and policy mandates are influential in reducing inequality—yet, their efficacy is often truncated by complex factors that contribute to ongoing social reproduction and the maintenance of group-based distinctions. Bourdieu and Passeron (1970/1990, 1964/1979) offer invaluable insight into conditions and processes of educational stratification. Through a theoretical construction of processes of legitimization, they consider how “pedagogical action” based upon “the cultural arbitrary of the dominant or of the dominated classes” (1970/1990, p. 5) leads to specific forms of social inequality. Bourdieu and Passeron (1964/1979) argue that processes of legitimization take place within everyday interactions, as privilege tied to social background results in habits, manners, skills, and attitudes that promote educational success. Possession of distinct and connected forms of capital are deemed to be “required to seize the ‘potential opportunities’ theoretically available to all” (Bourdieu, 1980/1990, p. 64).2 A cultural and social reproduction perspective argues that promoting equal access to education to mitigate inequality ignores that “abilities measured by scholastic criteria stem not so much from natural ‘gifts’ [...] but from the greater or lesser affinity between class cultural habits and the demands of the educational system or the criteria which define success within it” (Bourdieu & Passeron, 1964/1979, p. 22). Thus, Bourdieu and Passeron’s framework cast stratification as integral to the function of education itself.                                                  2 The authors detail four forms of capital generated through differing but connected forms of accumulated labour: 1) economic capital, which can be directly converted into money and resources; 2) cultural capital, which comprises formal and informal institutionalized, embodied, and objective qualifications; 3) social capital, which includes social obligations, connections, and networks that allow access to more collective forms of capital on the basis of social connections; and 4) symbolic capital, which comes to represent or depict the other three forms of capital to be legitimized (Bourdieu, 1998). 6  The walls of a school do not enclose educational inequality. Boudon (1974) offers important insight into how educational inequality interacts with employment and social demands, theoretically constructing a circular relationship between education and the labour market. Educational inequality is not only based on individual agency or structural constraints, but also on the aggregation of action that leads to unintended consequences (i.e., perverse effects) (Boudon, 1974). A primary concern for Boudon is how increasing credential levels do not necessarily change the relational level of inequality among people within a society. That is, there is a weak relationship between level of educational attainment (i.e., inequality of educational opportunity) and social status and opportunity (i.e., inequality of social opportunity). Boudon makes an important contribution to stratification scholarship by demonstrating that increasing attainment has a “perverse” effect on the relationship between education and the social outcomes it yields. Boudon argues that rising education at the societal level changes the relative relationship between attainment and opportunities. That is, educational outcomes are not constant over time.  Scholarship by Boudon, Bourdieu, and Passeron heavily influences scholarship on social reproduction (e.g., Demerath, 2009; DiMaggio, 1982; Khan, 2011; Lareau, 1987, 2002) and social mobility (e.g., Breen & Jonsson, 2005; Brown, 2013; Sturgis & Buscha, 2015; Thompson & Simmons, 2013). Additional research considers how mobility and reproduction impact inequality in various educational outcomes. Scholarship on inequality of educational outcomes examines a range of topics, such as hiring practices (Rivera, 2011, 2012), school-to-work transitions (Bynner & Parsons, 2002), and job satisfaction (Andres & Grayson, 2003). As discussed further in Chapter 2, stratification research also generates insight into how social 7  position intersects with educational and social opportunities (Gerber & Cheung, 2008; Kao & Thompson, 2003). Of crucial importance for my dissertation, scholarship on social reproduction and mobility provides a substantive basis for contemporary theorization of educational stratification as a context-specific and dynamic entity that exists in and beyond schools.  1.3 Outline of Dissertation As a manuscript-based dissertation, each of the three primary research chapters contends with a distinct type of educational outcome. To contextualize these three separate studies, I first give an overview of educational inequality in Canada through the lens of international comparison. In Chapter 2, I describe the Canadian system of education and explore how social position influences inequality in educational attainment and social opportunities. I demonstrate how differentiation in level of educational attainment and forms of credentials is necessary to consider when researching outcomes. Furthermore, I compare the Canadian system of education, levels of attainment, and social outcomes internationally. Indeed, what is often normalized in one context differs considerably in others. In Chapter 3, I research the relationship among field of study, gender, and part-time and temporary employment two years after postsecondary graduation in Canada. Using cross-sectional data, I compare non-standard employment outcomes for cohorts graduating with bachelor’s degrees between 1990 and 2010. Through two types of decomposition analyses, I examine the relationship between field of study and the likelihood of early career non-standard employment among male and female graduates. Furthermore, I consider the extent to which field of study contributes to gender inequality in non-standard employment compared to other known 8  factors, such as industry of employment and parenthood. I assess if gender differences in rates of non-standard employment are due to different characteristics and experiences of men and women or a more widespread form of social inequality that disproportionally affects all women regardless of what they study. Moving from employment type to employment experiences, Janine Jongbloed and I study the relationship between education level and workplace task discretion across 30 countries in Chapter 4. We use regression and decomposition analyses to explore if individual-agency or critical-institutional theoretical perspectives explain why education credentials lead to different self-reported levels of workplace task discretion. Contrasting the two frameworks, we research how skill assessment scores and occupational sector mediate the relationship between education and discretion across different contexts. Cross-nationally, the overall country level and the distribution of discretion among occupational sectors impact the extent to which education influences access to workplace task discretion. Whereas Chapter 3 investigates stratification within a single level of education by field of study, Chapter 4 examines educational inequality horizontally by level of credential.  In Chapter 5, Lesley Andres and I examine longitudinally how demographic factors and life course experience influence the educational beliefs of individuals over a 28-year period. We examine through growth modelling how gender, parental background, employment, and education relate to intrinsic and extrinsic educational beliefs. Complementing my previous chapters, we research the relationship among social location, education, and employment outcomes and individual orientations towards education and how it changes over time. Whereas stratification research and theory largely focus upon employment and socioeconomic outcomes, 9  Chapter 5 presents insight into a more subjective component of educational inequality. That is, it explores how self-enabling and self-restricting belief systems are part of the process of educational stratification. In Chapter 6, I engage with important critiques of educational stratification and outcome-based research. My conclusion explores important concerns surrounding how outcome-based education research risks “trickle-up” arguments that over-determine the ability for education to change forms of social inequality. There is also risk that educational outcome research will promote certain types or forms of education over others, with the assumption that they yield better social or economic life course pathways. Thus, alongside discussing my main findings and areas for future research in the conclusion of my dissertation, I examine possible critiques, aiming to work within a confluence of perspectives and establish a justification for why educational outcome research is necessary but must take into account its possibilities and limitations.  10  Chapter 2: Inequality of Educational Opportunities and Outcomes in Canada  2.1 Introduction  Access to education is an integral component to socioeconomic development, the promotion of equality, and social cohesion in contemporary society. Increasing the availability of learning opportunities is a principle policy directive used to support a range of social outcomes. For example, the Tuition Access Bursary program in New Brunswick offers free postsecondary tuition for families with an annual income under $60,000 CAD (Cromwell, 2016). Similarly, Nova Scotia residents on social assistance are eligible to receive up to one year of postsecondary tuition and living assistance (Laroche, 2016). Despite their merits, such programs are limited to targeting inequality of access rather than reckoning with educational outcomes, relying upon a unidirectional understanding of this process. In this chapter, I discuss multiple dimensions of educational inequality and emphasize the connections and differences between opportunities and outcomes, showing how it is necessary to consider new and persistent forms of inequality that arise in highly educated societies where inequality of opportunity and inequality of outcome exist side-by-side. The purpose of Chapter 2 is two-fold. First, I seek to contextualize the empirical research conducted in later chapters by providing an internationally comparative overview of the Canadian education system. Canada is a highly educated country with strong public investment in education, especially at the postsecondary level. Nonetheless, social mobility remains restricted to certain sectors of education and demographic groups. Both policy and research 11  rarely consider that improvements in overall levels of attainment are not necessarily synonymous with the promotion of equitable outcomes. Thus, the second purpose of Chapter 2 is to examine how outcomes vary among demographic groups. I focus predominantly upon how attainment and outcomes differ by social class and gender, two common group distinctions within educational scholarship.3  2.2 Systems of Canadian Education From an International Comparative Perspective  The governance of Canadian education is under exclusive provincial and territorial jurisdiction according to Section 93 of the Constitution Act.4 With no federal education department, 13 unique provincial and territorial systems of primary, secondary, and postsecondary education exist in Canada (Cameron, 1992; McEwen, 1995). Despite being typically characterized as decentralized, Canada often has common educational objectives promoted across the country, “fashioning a de facto pan-Canadian education policy framework” (Wallner, 2012, p. 851). In part, the confluence of provincial and territorial goals is due to the interprovincial Council of Ministers of Education, Canada (CMEC) (Wallner, 2014). Established in 1967, CMEC aims to generate and represent common provincial and territorial interests at a national level. Such goals include the need to fulfill international treaty obligations, facilitate                                                  3 There are other important social and demographic factors I do not assess in detail, particularly disability, sexual orientation, age, immigration status, and ethnicity. A considerable body of research examines educational access and outcomes among disabled peoples (Arim, 2015), by sexual orientation (Taylor et al., 2011; Waite & Denier, 2015), age of entry and attainment (Chesters & Watson, 2014), immigrations status (Adamuti-Trache & Sweet, 2014; Hou & Bonikowska, 2016), and Aboriginal ancestry (Arriagada, 2016; Feir, 2016a, 2016b).  4 The Canadian federal government does have some influence on provincial and territorial education sectors—for example, through funding research and development within postsecondary education (Shanahan & Jones, 2007) and the direct allocation of funding for Aboriginal education (Drummond & Rosenbluth, 2013; Usher, 2009). 12  interprovincial recognition of qualifications, oversee student assessment, and develop and report on Canada-wide educational indicators (CMEC, n.d.). A decentralized education system with a non-federal council to facilitate common objectives is especially unique when compared internationally. Authority over education also exists at the sub-state level in the United States. However, there is no pan-state governing council equivalent to CMEC (Wallner, 2012) and a national education department established after the 1965 Elementary and Secondary Education Act has increasing amounts of national influence (Vinovskis, 2015). Despite the unifying role of CMEC, notable differences among provincial and territorial education systems exist. The province of Quebec is the most distinct in terms of level of funding and system differences. All other provinces and territories offer primary and secondary education from grades one to 12. However, the Quebec system is based on 11 years of primary and secondary school, followed by two to three years of pre-university or vocational training within the Collège d'Enseignement Général et Professionnel (OECD, 2010). Several aspects of early-childhood education differ provincially and territorially, including the availability of full-day kindergarten and preschool and the allocation of childcare tax credits and government funding (Canadian Public Health Association, 2016; McCuaig & Akbari, 2014). The Quebec system of early-childhood education is again distinctive as the provincial government allocates more far more generous childcare and kindergarten funding compared to other jurisdictions (White & Prentice, 2016). Postsecondary education is also geographically distinct in Canada. Quebec, alongside Newfoundland and Labrador, also has comparably low postsecondary tuition fees—approximately half the national average (Statistics Canada, 2016, Table 1). Level of articulation between college and university sectors is also a notable regional system difference. Notably, 13  British Columbia and Alberta promote student transfer between colleges and universities, whereas Ontario upholds greater distinction between institutional types (Andres, 2001, 2015; Piche & Jones, 2016).  Both the structure and funding of Canadian education are part of a larger rationale of social policy that constitutes the welfare state. Education researchers use Esping-Andersen’s (1994, 1990) welfare-state typology to consider how international differences in levels of commodification and social expenditure impact inequality in attainment and outcomes (Hega & Hokenmaier, 2002; Pechar & Andres, 2011; Peter, Edgerton, & Roberts, 2010; Willemse & De Beer, 2012). Importantly, social spending on education is part of active labour market policy that seeks to improve skill and employability (Busemeyer, 2015). Examining levels of educational expenditure offers insight into two aspects of welfare state policy: the overall level of expenditure as an indicator of the priority of education in comparison to other types of social spending; and the intra-expenditure allocation of funding as a gauge of governmental prerogatives (West & Nikolai, 2013). Figure 2.1 compares government expenditure on education as a percentage of gross domestic product (GDP). It illustrates that Canadian primary and secondary expenditure is approximately average when compared to other Organisation for Economic Co-operation and Development (OECD) member countries. However, Canada has above average funding at the postsecondary level. Compared to Germany and Switzerland, Canada spends over two times more, as a percentage of GDP, within university and college sectors. As argued by Pechar and Andres (2011), high levels of state support for postsecondary education are generally characteristic of liberal-market regimes that promote equitable access through mass education. 14  The distinct Canadian system of skill formation promotes university and college level credentials, forms of postsecondary education that generally predominate over vocational education (Busemeyer, 2015). Discussed next, high levels of postsecondary attainment in Canada simultaneously generates social mobility while leaving certain forms of inequality intact.   Note: Expenditure includes funding for all public and private educational institutions. Data source: OECD (2017a).  Figure 2.1. Public expenditure on education as a percentage of GDP (2012-2013) 15  2.3 Educational Attainment and Intergenerational Mobility  At a national level, Canada has one of the highest rates of educational attainment in the world. By 2015, 90% of all Canadian residents between the ages of 25 and 64 had graduated from high school and 67% held a postsecondary credential—either a university degree (30%), college certificate or diploma (26%), or apprenticeship/vocational credential (11%) (Statistics Canada, 2016, Table A.1.1).5 A distinct feature of attainment in Canada is the high percentage of individuals with college certificates and diplomas (i.e., short cycle tertiary). The extensive public and private college system includes general and technical programs, which are more accessible through lower admission standards, fewer prerequisites, shorter training duration, and less costly tuition (Adamuti-Trache & Sweet, 2008; Cowin, 2017).6 Attainment rates at the college certificate and diploma level are much lower in other countries. The overall OECD average for attainment at the short cycle diploma/credential level is only 8%, over three times smaller than Canada (Statistics Canada, 2016, Table A.1.1). In the United States—a country often considered to be comparable to Canada in overall levels of education—only 11% of the population hold a college diploma or certificate as their highest credential, while 34% have degrees at the bachelor’s level or above (Statistics Canada, 2016, Table A.1.1).                                                  5 For consistency, I use the credential level framework set by the International Standard Classification of Education (ISCED). The ISCED classification system attempts to harmonize education levels cross-nationally but does not pertain to all forms of education, such as non-formal education (UNESCO, 2011).  6 There is not always a clear distinction between university and college sectors in Canada. Of note, some provinces have awarded degree granting rights to colleges and promote transferability of course credits between institutions (Andres, 2001; Gallagher & Dennison, 1995; Jones, 2009).  16  In Canada, educational attainment varies among social groups in distinct ways, particularly by generation, geography, gender, and class. Differences among age groups exist, due to both increasing levels of attainment and trends in the types of credentials earned. For instance, individuals aged 55 to 64 have similar levels of short-cycle college-level education compared to 25 to 34-year-olds in Canada; however, the older cohort is less likely to hold a bachelor’s degree (15% versus 25%) (OECD, 2016a, Table A1.2). Geographical inequality is also a marked feature of Canadian education. Approximately half of all Nunavut residents—approximately two-thirds of whom are Inuit—do not have a high school diploma compared to only 7% in British Columbia (Statistics Canada, 2016, Table A.1.1). Postsecondary attainment also varies regionally and provincially. In 2015, 33% of Ontario residents versus 18% in Newfoundland and Labrador held degrees at the bachelor’s level or above (Statistics Canada, 2016, Table A.1.3). A partial explanation for geographical differences is a persistent rural-urban divide. Rural residents are less likely to complete high school and attend postsecondary institutions, with student, family, and community characteristics partially explaining the gap (Andres & Looker, 2001; Looker, 2009; Newbold & Brown, 2015). As in most OECD member countries, women have higher levels of educational attainment in Canada (Andres, 2015; OECD, 2016a). Women are more likely to graduate from high school. There is a notable gender gap in the Northwest Territories, where 75% of women compared to 58% of men held high school diplomas by 2015 (Statistics Canada, 2016, Table A.2.1). Even in Quebec, the province with the highest high school graduation rate, a 7% gender gap exists (99% versus 92%). A larger proportion of women across Canada have also attained university degrees at the bachelor’s degree level or above (32% versus 27%) (Statistics Canada, 17  2016, Table A.1.2).7 A gender gap in attainment also exists among First Nations, Inuit, and Métis people. In 2011, 12% of Aboriginal women and 7% of Aboriginal men held a bachelor’s degree or higher (Arriagada, 2016, Table 10).8 Several factors explain the gender gaps in attainment: in high school girls self-report higher educational aspirations (Shapka, Domene, & Keating, 2012), spend more time on homework (Frenette & Zeman, 2007), take university preparatory classes at slightly higher rates (Krahn & Taylor, 2007), and enrol in postsecondary education sooner after graduation (Hango, 2011).  Parental education remains highly influential on levels of attainment in Canada. Canadian intergenerational mobility—that is, when a child holds a credential higher than his or her parent(s)—is large relative to other OECD member countries. Figure 2.2 depicts the percentage of individuals with a postsecondary certificate, diploma, or degree by the education level of their parents. Among parents without a high school diploma, 36% of their children hold postsecondary credentials in Canada. Higher levels of mobility in Canada correspond with prior research suggesting that greater attainment decreases intergenerational inequality in education (Beller & Hout, 2006; Breen, 2010). The relationship between education and social mobility also has inter-group ramifications. For example, higher attainment among women generates gender differences                                                  7 Importantly, the over-representation of women differs by level and credential type. As of 2014, women represented 60% of graduates at the bachelor’s level, 56% at the master’s level, and 45% of graduates at the doctoral level (OECD, 2016a, p. 71). Women are more likely to have college credentials (29% versus 22%) but less likely to have apprenticeship and short-cycle non-tertiary training (7% versus 15%) (Statistics Canada, 2016, Table A.1.2). 8 Comparably, in 2011, 28% of non-Aboriginal women and 25% of non-Aboriginal men held a bachelor’s degree or higher (Arriagada, 2016, Table 10).  18  in intergenerational mobility.9 Additionally, as parental education increases, so does the corresponding education of their children. The equivalent rate among individuals with parents who have attained postsecondary credentials is 72% in Canada, an above average level of intergenerational perpetuation.  Note: Percentages restricted to individuals aged 25 to 64. Data source: OECD (2017b). Figure 2.2. Intergenerational mobility (2012-2015)                                                  9 In 2012, intergenerational mobility among women aged 25 to 44 was 59% compared to 45% for men (Statistics Canada, 2016, chart E.1.3). Notably, gender differences vary considerably by province, ranging from a 29% to 25% gap in Alberta, Prince Edward Island, Quebec, and Saskatchewan, to 1% in Manitoba.  19  Even with high levels of mobility, intergenerational perpetuation—that is, when a child attains the same education level as his or her parent(s)—is also a prominent feature of educational attainment in Canada. In 2012, 73% of individuals aged 25 to 44 had the same postsecondary level as their parents (Statistics Canada, 2016, Table E.1.4). To reconcile how high levels of mobility and perpetuation exist side-by-side, it is necessary to consider the varying forms of class inequality in access to postsecondary education. Unequal attendance within university and college sectors provides insight into how different postsecondary streams perpetuate intergenerational and class inequality. First-generation students are more likely to attain college credentials at the certificate and diploma level (Krahn, 2017). Although rates of university degree attainment have grown over time in Canada, divides by parental education remain. In 1986, 12% of first-generation individuals aged 25 to 39 held a university degree (Turcotte, 2011, Table 1). Comparably, the rate for individuals whose parents held at least one university degree was 45%. By 2009, intergenerational perpetuation was still maintained: a 23% degree-attainment rate among the first-generation group compared to 56% of individuals with at least one university-educated parent (Turcotte, 2011, Table 1). A large body of Canadian research has considered the resource and contextual factors that contribute to intergenerational perpetuation. First, financial resources allow high-income parents to give increased opportunities that influence learning and later attainment. Children from high-income families are more likely to participate in formal and informal learning during the summer months, resulting in literacy skill improvement (Davies & Aurini, 2013). Class background influences the availability and type of extracurricular activities children (Ashbourne & Andres, 2015) and postsecondary students pursue (Lehmann, 2012a). Second, class 20  background shapes academic achievement and access. Advantaged high school students are more likely to take university preparatory courses (Krahn & Taylor, 2007), enrol in postsecondary schooling directly after high school (Hango, 2011), and expand their postsecondary choices by attending non-regional institutions (Frenette, 2004). Children of highly educated parents are more likely to attend higher ranked universities (Davies, Maldonado, & Zarifa, 2014), choose fields of study that have more advantageous career outcomes (Zarifa, 2012), and graduate with lower average student loan amounts (Kapsalis, 2006). It is apparent from research on the mechanisms of intergenerational perpetuation that, although Canada has high overall levels of attainment and intergenerational mobility, educational inequality persists. 2.4 Educational Distinctions and Horizontal Inequality Differences among institutions, credential types, and disciplines generate socioeconomic distinctions that perpetuate inequality. Furthermore, such distinctions are often undetectable within statistics on education level alone. Inequality among people with the same education level (i.e., horizontal stratification) is not only a characteristic of postsecondary sectors in Canada. Most primary and secondary students study a common comprehensive curriculum in public schools. In 2014, only 4% of students in Canada graduated from a vocational high school program compared to an OECD average of 49% (OECD, 2016a, p. 50).10 Nonetheless, more subtle forms of academic streaming exist at primary and secondary levels, such as private-public school distinctions. Only a small percentage of primary and secondary students attend private                                                  10 The OECD defines upper-secondary vocational programmes as ISCED level 3 with a vocational distinction. Importantly, level 3 does not capture students who participate in specialized high school training programs and classes but graduate without a vocational designation (Molgat, Deschenaux, & LeBlanc, 2011).  21  schools in Canada, yet their greater socioeconomic status translates into higher overall achievement in course marks and credential levels (Baker, 2014; Frenette & Chan, 2015). Alongside a public-private divide, the location of a school and neighbourhood characteristics are also influential. As an example, in 2006, Vancouver neighbourhood characteristics influenced kindergarten children’s school readiness and early development, even when accounting for family background factors (Oliver, Dunn, Kohen, & Hertzman, 2007). One of the most important neighbourhood differences was median income. Canadian institutional prestige at the postsecondary level is less distinctive when compared with the United States, a country characterised by high levels of elitism and selectivity among Ivy League schools (Davies & Hammack, 2005). Even so, there are important institutional distinctions in Canada. Income and expenditure disparity increased among Canadian universities between 1971 and 2006, although to a smaller extent than the United States (Davies & Zarifa, 2012). National rankings, such as the ones produced by Maclean’s magazine, generate distinctions within the postsecondary sector—especially for small and mid-sized institutions (Drewes & Michael, 2006). National and global rankings construct certain universities as offering superior and competitive education (Baker, 2014). Nevertheless, horizontal stratification is more complex than referring to a gradient of institutional prestige. Students do not necessarily aim to attend the highest ranked institution for which they are eligible, as socio-demographic factors influence postsecondary destinations. For example, international students quintupled their attendance in private colleges and training institutes between 2004 and 2014 (Illuminate Consulting Group, 2016). First Nations, Inuit, and Métis groups are more likely to study in institutions offering certificates, diplomas, or trades qualifications (Assembly of First Nations, 22  n.d.; Statistics Canada, 2015, Chart 13). Contemporary and historical attendance patterns construct institutional differences in relation to demographic characteristics of who does or ought to attend a particular type of school.  A prominent type of horizontal inequality surrounds gender differences in skill assessment. The most recent 2015 Programme for International Student Assessment (PISA) results indicated that no gender gap in high school science assessment scores exists, other than in Newfoundland and Labrador and Manitoba, where girls outperform boys (O’Grady, Deussing, Scerbina, Fung, & Muhe, 2016). Contrastingly, girls had higher literacy scores in all provinces, while boys had higher mathematic scores in half of all provinces (other than Prince Edward Island, Nova Scotia, New Brunswick, Manitoba, and Saskatchewan, where there is no significant gender difference) (O’Grady et al., 2016). Among Canadian adults, gender differences in literacy assessment scores disappear—although a small gender difference in numeracy skill remains across all age groups (Statistics Canada, 2013). Less studied are gender differences in technology-based skills. What research does exist suggests girls are more likely to self-report lower ability and interest in computers and technological domains (Hargittai & Shafer, 2006; Luu & Freeman, 2011). Not withstanding these perceptions, there is no overall gender difference in assessment scores for problem solving in technology-rich environments for adults in Canada (Statistics Canada, 2013).  Perceived and actual gender differences in skill and competency measures relate to segregation within postsecondary fields of study (Correll, 2001). As demonstrated in Figure 2.3, gender segregation among academic disciplines is present in most countries. Fewer women are in 23  engineering, manufacturing, science, math, and computer science disciplines. Contrastingly, more women study in health, welfare, humanities, and arts-based fields of study.   Data source: OECD (2017c).  Figure 2.3. Gender segregation by postsecondary field of study (2014-2013) It is important to emphasize that not all countries follow the same gender segregation trends. In Canada, rates of enrolment in science fields are similar among men and women. However, men outnumbered women 4.3 to 1 in engineering and manufacturing fields and women outnumbered men 5.4 to 1 in health and welfare fields (OECD, 2016a, p. 65). Estevez-Abe (2006) argues that 24  coordinated market economies, such as Germany, have greater levels of gender segregation due to the skill specificity of their vocational sectors. Yet in liberal-market economies like Canada, segregation also endures due to gender norms and the perceived portability and atrophy rate of certain skills and disciplines (Estevez-Abe, 2012). There is also evidence of segregation within sub-specializations and among other socio-demographic groups. Coarse discipline groupings, like the fields of study Figure 2.3 shows, risk overlooking gender segregation within the categories themselves. By way of illustration, while the number of men and women studying business and management has become more equal over time in Canada (Zarifa, 2012), human resource students are more likely to be women and entrepreneurship students are more likely to be men (Hunt & Song, 2013). Field of study segregation also impacts other groups. For instance, international students in the Canadian postsecondary sector are more likely to study in science, engineering, and business (McMullen & Elias, 2011; OECD, 2016a). The language profile of Grade 12 students is also a predictor of university preparatory course completion and type. In 2009, East and South Asian, European, French, Punjabi, Korean, and Vietnamese language speaking students were more likely to take math, Chinese and Korean language speakers were more likely to take physical sciences, and Chinese, Korean, Persian, Punjabi, and Tagalog language speakers were more likely to take life sciences in British Columbia high schools (Adamuti-Trache & Sweet, 2014). Finally, there is evidence of intergenerational perpetuation among postsecondary fields of study. For example, students from advantaged class backgrounds were more likely to graduate from professional degree programs—such as law, medicine, and dentistry—between the mid 1990s and early 2000s (Frenette, 2005). 25  2.5 Inequality in Educational Outcomes Research on various employment, social, and well-being outcomes by education level presents continual support for greater levels of attainment. On average, individuals with postsecondary credentials have better health outcomes (Raphael, 2009), higher proficiency scores in literacy, numeracy, and technology use (Statistics Canada, 2013), and lower unemployment rates (OECD, 2016a). University degree and college certificate holders experience fewer permanent and temporary layoffs compared to high school graduates (Frenette, 2015) and work more hours and weeks over time (Boudarbat, Lemieux, & Riddell, 2010). Between 1991 and 2001, male bachelor’s degree holders earned 1.75 times more than high school graduates, while their female counterparts earned 1.85 times more (Frenette, 2015, Table 1).11 As I discuss next, broad-based statistics downplay the extent to which demographic and social factors both within and beyond education influence outcomes. Inequality in attainment and differentiation does not stay contained within systems of education—rather, it propagates multiple forms of social inequality. Class stratification in education influences later earnings, employment, and life-course patterns. Given that individuals whose parents earned professional or university-level degrees have a higher likelihood of also completing the same level of education or above, this group can also expect to earn more and have higher rates of employment. Class and parental background effects are also influential on employment expectations (Lehmann, 2009), intergenerational                                                  11 College certificate holders also receive an earning premium. However, their average earnings are closer to high-school graduates than degree holders (Frenette, 2015; Li, 2006). 26  transmission of employers (Corak & Piraino, 2011), and the rate of student-loan default (Wright, Walters, & Zarifa, 2013). In other cases, parental education has little to no direct effect on social and economic outcomes, as the influence of parental education becomes indistinguishable from educational achievement (Boudarbat & Chernoff, 2010; Torche, 2011). That is, while the effect of parental background may influence an individual’s likelihood of gaining high-level credentials, it does not always result in distinguishable employment outcomes at the same education level (Hout, 1988).  Employment rates and earnings differ by gender in Canada. In 2015, male high school graduates between the ages of 25 and 64 had a 13% higher employment rate than female high school graduates (Statistics Canada, 2016, Table A.3.1). The gender employment gap converged at higher credential levels, shrinking to 8% at the bachelor and 5% at the master’s degree levels (Statistics Canada, 2016, Table A.3.1). Higher levels of education increase women’s labour market attachment, generate access to better employment opportunities, and reduce the overall gender gap in wages (Chaykowski & Powell, 1999). As shown in Figure 2.4, women in Canada without a high school diploma earn, on average, 61% of what men without a high school diploma earn. At higher educational levels, the gender wage gap diminishes. Yet even at the highest level, the gap is larger than approximately half of the OECD member countries shown. Additionally, the assumption that higher credentials reduce gender inequality in earnings is clearly not universally valid—lower education levels reduce the gender pay gap in many countries. There are several explanations for why higher levels of education do not necessarily reduce gender wage inequality, including field of study stratification (Bobbitt-Zeher, 2007; Leuze & Strauß, 2014; Ochsenfeld, 2014) and labour market variation (England, Gornick & Shafer, 2012). 27   Men=100%; Statistics calculated from average annual full-time full-year earnings among individuals aged 25-64. Data source: OECD (2016, p. 126). Figure 2.4. Women’s earnings ratio relative to men, 2012-2014  28  Research often constructs education as a powerful mechanism in reducing inequality in outcomes, from closing the gender earning gap to diminishing class advantage. However, the question of whether or not education reduces or generates inequality is dependent upon the type of comparisons made. For example, educational wage premiums comparing the earnings of high school and university graduates grew from 45% to 53% for women and 32% to 40% for men between 1980 and 2006 (Boudarbat et al., 2010). Women still earn less overall but have more to gain from university credentials given their larger intra-gender earning divide. Understandings of inequality are also based on where in the distribution a comparison is made. For instance, between 1991 and 2010, there was a $286,000 gender gap in cumulative earnings among bachelor’s degree holders (Frenette, 2015). Yet earners in the highest percentile drive discrepancy in wages. For example, while median earnings only differed by $17,000 among men and women between 1991 and 2010, the gender gap was $1,877,000 for the top 5% of earners (Frenette, 2015, Tables 7 & 8). Finally, prior research signals that gender inequality has subjective elements surrounding expectations and desires. Despite the increasing occupational aspirations of women between the 1970s and 1990s (Andres, Anisef, Krahn, Looker, & Thiessen, 1999), career expectations in terms of salary and promotion are still lower among female Canadian postsecondary students, especially in male-dominated fields of study (Schweitzer, Ng, Lyons, & Kuron, 2011).  2.6 Conclusion Social stratification research provides an important account of how educational institutions and systems disproportionately serve certain individuals. Some historically under-29  represented groups, such as women, have increased their overall education level. Furthermore, a greater number of first-generation students are gaining postsecondary credentials. Achievement trends interconnect with Canadian social policy, embedded in a liberal market context that promotes high levels of educational attainment. For many groups, it is no longer adequate to construct broad claims of marginalization when education is increasingly accessible. Yet, as I have demonstrated, accessibility does not mean that inequality has simply dissolved. Forms of under-representation and group differences remain, especially when examining more detailed levels or types of credentials. Additionally, we cannot take for granted that elements within the Canadian context are universal or that more education leads to less inequality, especially with respect to outcomes. Thus, social stratification research must be conscious of change in educational inequality and the nuanced, context-specific forms it may take.  Each of the next three research chapters deepens existing analysis on educational attainment and outcomes. In Chapter 3, I examine the relationship among gender, field of study, and non-standard employment. In Chapter 4, I consider how credential level influences task discretion in the workplace. In Chapter 5, I research the relationship between time spent in postsecondary schooling and educational beliefs over a 28-year period. My three separate studies show how level and type of educational attainment influences an individual’s employment experience and even worldview. Context and social position shape the relationship between education and a given outcome. In the following chapters, I study educational inequality as a dynamic entity that changes over time, context, and social background—predominantly considering gender and social reproduction. 30  In summary, it is well documented how educational attainment rates have transformed and differ among countries and groups. Less considered is how outcomes vary. As I demonstrate above, Canadian researchers largely study how income and employment rates differ by education level—important job-related factors that influence well-being in a range of ways. Yet, as I detail in the next three chapters, it is necessary to broaden the types of educational outcomes researched to understand the multifaceted realities of social inequality more fully. My purview of study is especially necessary given the high levels of educational attainment in Canada. Governmental and institutional educational policy has predominantly focused on promoting access to higher credential levels for under-represented groups (Kirby, 2007). Indeed, encouraging educational attainment continues to be an important way to promote equality. However, policy makers must also consider unequal outcomes to avoid casting education as the primary mechanism to reduce social inequality. It is necessary to support collective bargaining, redistributive tax policies, and forms of social protection for all individuals, regardless of their education level (Fortin, Green, Lemieux, Milligan, & Riddell, 2012). As my dissertation shows, more education cannot solve unequal outcomes that mirror deeply ingrained social inequalities.  31  Chapter 3: Gendered Pathways From School to Work: The Association Between Field of Study and Non-Standard Employment Outcomes  3.1 Introduction Women are over-represented in non-standard work—a form of employment that has grown since the 1990s (OECD, 2015a). Among all OECD countries, women are less likely to hold employment in full-time continuous positions compared to men, with slightly higher rates of temporary employment and much higher rates of part-time employment (OECD, 2016b, 2015a). Explanations for the gender gap in non-standard employment include disproportionate care and household responsibilities (Rose, Hewitt, & Baxter, 2013), norms and preferences (Booth & van Ours, 2013), occupational segregation (Leschke, 2015), and explicit and implicit discrimination within employment regulations (Fudge & Vosko, 2001) and by employers (Eagly & Steffen, 1986; Pedulla, 2016). An umbrella rationale capturing many explanations for the disproportionate share of women in non-standard employment surrounds gender differences in life course pathways (Oesterle, Hawkins, Hill, & Bailey, 2010). Under this framework, life decisions and occurrences are understood to set a course for later opportunities and barriers, which may include a greater incidence of non-standard employment.  Education has an integral influence on later life course trajectories and influences future employment opportunities. Yet educational pathways are often gendered, with women 32  surpassing men in level of attainment in most Western countries but remaining differentiated by fields of study (DiPrete & Buchmann, 2013; England & Li, 2006; OECD, 2016a). It is well established that education level and fields of study influence later employment and wages (Boudarbat & Connolly, 2013; Finnie & Frenette, 2003), with gender segregation among academic disciplines contributing to lower wages for women (Bobbitt-Zeher, 2007; Brown & Corcoran, 1997; Kim, Tamborini, & Sakamoto, 2015; Shauman, 2016). Despite greater levels of educational attainment corresponding to lower rates of non-standard employment (OECD, 2016b, 2015a), variation among fields of study remains understudied. To address this gap, I examine the presence of gender inequality in early career non-standard employment in greater depth, considering the extent to which academic disciplines contribute to different rates of non-standard employment for men and women two years after postsecondary graduation. I examine how the composition of male and female graduates differs, such as the field in which they study or the industry of their employment. Non-compositional elements are also necessary to consider—such as the impact of parenthood—as effects distinct from education-work trajectories may also propagate gender disparity in rates of non-standard employment.  Postsecondary fields of study have an influence on later occupational and employment outcomes, functioning as a gatekeeper to certain employment opportunities and industry sectors (Prix, 2009; Vuorinen-Lampila, 2016) with higher or lower rates of non-standard employment. Research shows that men are more likely to graduate from fields of study that bolster employment outcomes, particularly science, technology, engineering, and mathematics (STEM) (Han, Tumin, & Qian, 2016; Hu & Vargas, 2015, Kim et al., 2015; 33  Melguizo & Wolniak, 2012; Thomas & Zhang, 2005; Xu, 2013). Explanations for gender segregation within academic disciplines vary but often surround mechanisms of self-selection (Estevez-Abe, 2012; Polachek, 1981) or enduring stereotypes (Zafar, 2013). Casting field of study as a linchpin, researchers understand gendered pathways through education as mediating the relationship between gender and work, which may include a different likelihood of non-standard employment. Greater gender equality in the characteristics of graduating cohorts—namely within academic disciplines—would result in equalizing rates of non-standard employment. Alternative perspectives would argue that higher rates of non-standard employment among women is not simply a “product” of field of study and other graduate characteristics but rather is due to the gendered organization of work and social life. There is risk of overemphasizing the importance of field of study given existing gender inequality in paid and unpaid work. Academic disciplines may only partially explain the likelihood of non-standard employment given gendered work norms (Booth & van Ours, 2013) and historical and contemporary employment discrimination (Fudge & Vosko, 2001). The relationship between non-standard employment and field of study is not simply based on post-graduation outcomes specific to certain disciplines and industries; rather, it is due to expectations within and outside paid employment that generate differences among men and women (de Ruyter & Warnecke, 2008; Purcell, 2000; Smyth, 2005). In this sense, a direct relationship between gender and non-standard employment exists, one that functions regardless of field of study. Rather than refuting that non-standard employment differs by academic discipline, this 34  perspective emphasizes that compositional differences among men and women cannot fully explain persistent gender discrimination in work. In assessing the two explanatory frameworks, I examine the extent to which gender inequality in non-standard employment is 1) due to the differing characteristics of male and female graduates, a main component being field of study, or 2) is based on a systemic disparity that disproportionally affects women regardless of academic discipline. In other words, I query to what extent non-standard employment outcomes are due to the structural conditions of certain feminized fields of study or general gendered bias across all disciplines. An important way to assess both perspectives is change over time. Variation in the characteristics of female and male graduates may impact the likelihood of non-standard employment. Contrastingly, no change suggests gender inequality in rates of non-standard employment evades compositional variation in the characteristics of graduates. To assess change over time, I compare the early-career outcomes of individuals graduating from Canadian universities in 1990/1995 and 2005/2010. I separately examine part-time and temporary employment to study how the relationship among gender, field of study, and non-standard employment may differ by employment type. I research the extent to which field of study, industry, and other possible contributing factors explain gender inequality in non-standard employment using descriptive, binomial logistic regression, and non-linear decomposition analyses. Given the volume of research on early career earnings disparities, my inquiry offers insight into other important employment characteristics that result in divergent outcomes among graduates. 35  3.2 Literature Review 3.2.1 Non-standard employment and inequality The International Labour Organization (ILO) defines non-standard employment as “work that falls outside the scope of a standard employment relationship” such as “(1) temporary employment; (2) temporary agency work and other contractual arrangements involving multiple parties; (3) ambiguous employment relationships; and (4) part-time employment” (ILO, 2015, pp. 1-2). By definition, non-standard work differs from a “standard” employment relationship, which historically involved predictable working hours, normative career progression, and continuous and open-ended employment (Horemans, 2016). Non-standard employees may be privy to benefits and employment protection depending on their contracts; despite this, employment regulation often constructs a normative model of standard employment that increases insecurity for non-standard workers (Vosko, 2006). Discussed next, exclusion from a standard employment relationship often lowers job quality and propagates multiple forms of economic and social inequality. A central focus for researchers within the area of non-standard employment is job quality—defined as a “set of work and employment related factors that have a positive and direct effect on the worker’s well-being” (Boccuzzo & Gianecchini, 2015, p. 3). Of concern is how individuals and groups engaged in non-standard employment are at risk for lower quality work solely based on the form of their employment (Vosko, 2010). Individuals employed in temporary and part-time positions are at greater risk for variable and unsociable work hours (Glorieux, Mestdag, Minnen, & Vandeweyer, 2009; Heisz & LaRochelle-Côté, 2006; Kalleberg, 2000), earn less and receive fewer fringe and employment-provided 36  benefits (Barrett & Doiron, 2001; Kalleberg, Reskin, & Hudson, 2000; Marshall, 2003; Vosko, 2003, 2006), and are more likely to live in poverty (Horemans & Marx, 2013). Nevertheless, temporary and part-time positions are not intrinsically of lower quality. Non-standard employment also encompasses highly paid contract work (Doogan, 2009) and positions within professions and industries that allow for more flexible employment through the professional and class advantage of their workers (Gerstel & Clawson, 2014). Non-standard employment is a contributing factor to overall social and economic inequality in Canada. Non-standard workers have lower earnings compared to standard employees across all OECD countries and Canada has one of the highest earning gaps. Comparing 2012 median annual earnings, standard workers earned approximately $43,000 CAD, while full-time temporary workers earned $16,500 CAD and part-time temporary or permanent workers earned $11,000 CAD (OECD, 2015a, pp. 152-153).12 Median hourly wage differences between standard and non-standard workers was also one of the highest among OECD countries. For every dollar an individual employed in a standard position made in 2012, a part-time or temporary worker earned between $.62 to $.58 cents depending on his or her type of non-standard employment (OECD, 2015a, pp. 152-153). During this same year, two-thirds of non-standard employees were clustered within the lowest earnings decile in Canada, one of the highest levels of income polarization among OECD countries (OECD,                                                  12 Average earnings reflect all paid workers aged 15 to 64 but excluded employers, self-employed individuals, student workers, and apprentices.  37  2015a, p. 167). Canada also has one of the highest household poverty rates among non-standard workers—35% compared with the OECD average of 22% (OECD, 2015a, p. 178).  3.2.2 Gender inequality in non-standard employment Rates of part-time and temporary employment vary by gender, age, and period. In Canada, between 1995 and 2007, 11% of new job growth was full-time temporary employment (OECD, 2015a, p. 146). During this same period, permanent and temporary part-time work grew by 4%. Women continued to be over-represented in part-time employment; however, their total share among all part-time workers decreased slightly from 69.3% to 67.5% between 1991 and 2009 (Ferrao, 2011, Table 7). By 2015, among individuals aged 25 to 55, 18.9% of all employed women and 5.5% of all employed men worked in part-time positions (Moyser, 2017, Table 3). Still, the share of involuntary part-time workers (as a percentage of all part-time employees) grew from 25.1% to 34.6% between 1990 and 2016 (OECD, 2017d, n.p.). In 2015, women were more likely to cite voluntary reasons for part-time employment, such as caring for children (25% versus 3%) (Moyser, 2017, Table 3). This same year, women aged 25 to 54 had slightly higher rates of temporary employment (10.4% versus 9.4%) (Moyser, 2017, Table 6), a relatively stable trend since the mid-1990s (OECD, 2017e, n.p.). Women earn less, on average, in temporary positions (Fuller & Vosko, 2008). Rates of temporary employment are also much larger for young adults aged 15 to 24. For example, in 1997, 24.5% of women and 25.6% of men aged 15 to 24 were employed in temporary positions (OECD, 2017e, n.p.). By 2016, the rate of temporary employment had grown to 31.9% of employed young women and 29.5% of employed young men. 38  The over-representation of women in non-standard positions connects to historical and contemporary employment regulation, gendered employment expectations, and occupational and industry stratification. The establishment of standard employment regulations in the post WWII period was based on the normative conception of a male breadwinner/female caregiver economic unit (Vosko, 2010). The historical lineage of standard employment policy continues to propagate the association of non-standard work with women (Cranford, Vosko, & Zukewich, 2006). Societal conceptions of the necessity of flexible and diminished hours for women to combine domestic and work responsibilities also support policy assumptions (Beaujot, 2002; Crompton & Harris, 1998). Markedly, the presence of a child in a household increase rates of non-standard employment for women (Vosko & Clark, 2009). Occupation and industry segregation also contribute to gender disparity in rates of non-standard employment (Fernandez-Mateo & King, 2011). For example, women are under-represented in blue-collar manufacturing and manual occupations that are traditionally more likely to offer full-time employment (Fortin & Huberman, 2002). Women are also over-represented in sectors with high rates of non-standard employment—notably the public sector (Fuller, 2005). Prior research on the feminization of professional spheres also signals that the influx of women into previously male-dominated professions does not necessarily eradicate gender inequality. Research on professions has assessed how employment trends within previously male-dominated sectors change when an increasing percentage of women enter (Bolton & Muzio, 2008). Two trends are important to highlight. First, gender segregation may be maintained within sub-specializations (Adams, 2010). Second, gender inequality may remain 39  with respect to working hours and temporary employment. For example, rates of temporary employment are higher overall for women in academic professions in the United Kingdom but lower in senior temporary positions (Bryson, 2004). This body of research highlights that although women may be making inroads into previously male-dominated areas of employment, gender inequality in non-standard employment may remain. 3.2.3 Gender inequality in non-standard employment among recent graduates Individuals leaving school have a higher risk of non-standard employment (OECD, 2015b). People in the period initially following graduation are more susceptible to non-standard employment as they undergo entry or re-entry into the labour market (de Lange, Gesthuizen, & Wolbers, 2014) and compete with older workers who tend to have more work experience (de Vries & Wolbers, 2005). Career progression within specific industries may also be influential on the risk of non-standard employment (Ferguson & Wang, 2014). For example, recent graduates seeking to gain experience within non-profit and creative sectors may expect early-career non-standard work (Charlesworth, 2010; Hennekam & Bennett, 2017; Salamon, 2015). Non-standard employment may offer industry-specific work experience, skill acquisition, or access to social networks. Importantly, certain fields of study connect to industries characterized by high (or low) rates of non-standard employment when graduates undergo school-to-work transitions. A possible contributing factor to gender differences in non-standard employment is horizontal segregation among postsecondary fields of study. Between 1979 and 2004, the proportion of women graduating from education, fine arts, and humanities disciplines 40  increased in Canada, while STEM fields remained largely male-dominated (Andres & Adamuti-Trache, 2007). Nonetheless, the increasing number of women in health and some science disciplines has contributed to desegregation in STEM fields. For instance, in 2011, 39% of STEM degree holders between the ages of 25 and 34 were women, compared with 23% of individuals aged 55 to 64 (Ferguson, 2016, p. 20). Even so, the use of coarse discipline categories masks the extent of change within STEM sub-fields: a 1% cohort difference in mathematics and computer science (from 29% to 30%), a 15% increase in engineering fields (from 8% to 23%), and a 24% increase in science and technology fields (from 35% to 59%) (Ferguson, 2016, chart 6). Notably, over the 1980s and 1990s, female enrolment in biology quadrupled (Andres & Adamuti-Trache, 2007). The proportion of women with medical degrees also increased: 62% of medical degree holders aged 25 to 34 are women, compared with 28% of 55 to 64-year-olds (Statistics Canada, 2013, Figure 1). Gender segregation by field of study contributes to several forms of later labour market inequality, including the gender wage gap (Bobbitt-Zeher, 2007; Brown & Cororan, 1997; Leuze & Strauß, 2014; Ochsenfeld, 2014), differences in occupational status (Triventi, 2013), and rates of unemployment (Reimer, Noelke, & Kucel, 2008; Reimer & Steinmetz, 2009). Graduates from male-dominated fields of study often have more advantageous labour market outcomes (Boudarbat & Connolly, 2013; Shauman, 2016). Yet, curiously, the relationship between field of study and non-standard employment has received almost no attention. What studies do exist illustrate that rates of non-standard employment vary by academic disciplines, with occupational and industry pathways offering a partial explanation (Giesecke & Schindler, 2008) along with the skill specificity of certain disciplines 41  (Lombardo, De Luca, & Passarelli, 2012). Such studies, however, do not examine gender differences in non-standard employment in detail and limit their research to a small number of disciplinary categories. 3.3 Research Rationale, Questions, and Methodological Approach As discussed in the introduction, compositional differences among men and women provide one explanation for gender disparity in rates of non-standard employment. Notably, industry and occupational segregation contributes to unequal rates of non-standard employment (Gerstel & Clawson, 2014; Leschke, 2015; Pedulla, 2016). Postsecondary fields of study are often a “pipeline” leading to employment segregation (Triventi, 2013). Thus, the over-representation of women in academic fields characterized by higher rates of non-standard employment would contribute to gender inequality. If gender segregation among disciplines propagates an unequal probability of non-standard employment, then it also offers a means of reduction. That is, equalization of the different characteristics of men and women over time would reduce the gender gap in rates of non-standard employment. Conversely, prior research also suggests women are at greater risk of non-standard employment due to the gendered organization of paid and unpaid work (Cranford et al., 2006; Fudge & Vosko, 2001). Such a perspective suggests that a relationship between gender and non-standard employment exists regardless of a graduate’s field of study. Gender segregation among academic disciplines may be a contributing factor, but it cannot fully explain systemic and often enduring bias. For example, male and female graduates from the same academic field may have a different likelihood of non-standard employment. Thus, 42  persistent forms of inequality—especially systemic discrimination and unequal caretaking responsibilities—will continue to contribute to lasting differences in the rates of non-standard employment among men and women even when accounting for field of study differences.  Contending with two possible explanations, my research seeks to answer the follow questions: 1) to what extent does field of study mediate the relationship between gender and non-standard employment? and 2) how variable is the mediating effect of field of study between 1990/1995 and 2005/2010 graduating cohorts? To answer both questions, I first examine the gender composition of 28 separate fields of study at the bachelor’s degree level among 1990/1995 and 2005/2010 graduating cohorts. Next, I assess how rates of non-standard employment two years after graduation differ by discipline and part-time and temporary work. I generate the predicted probability of temporary or part-time employment through logistic regression to study how rates differ by the percentage of female graduates within an academic discipline. In part two, I move to directly consider my research questions by assessing the mediating effect of field of study. I examine the extent to which each of the 28 disciplines mitigates or propagates overall gender inequality in rates of non-standard employment and contrast the two periods. Finally, in part three, I compare field of study to other possible characteristic and non-characteristic effects that may influence the likelihood of non-standard employment, such as industry and province of employment, and martial status and the presence of a child in a household. In this section, I estimate the relative importance field of study has in determining the likelihood of non-standard employment compared to other factors. 43  3.3.1 Data I use the Canadian National Graduate Survey (NGS), a cross-sectional study that provides insight into a respondent’s educational background and subsequent employment outcomes approximately two years after postsecondary graduation. The 1990 graduating cohort survey first used indicators measuring non-standard employment. They were repeated in later cohorts, surveyed approximately every five years. Respondents include graduates from recognized Canadian public postsecondary institutions. The NGS excludes graduates from private institutions (a small portion of the Canadian postsecondary system), apprenticeship programs, and terminal continuing education programs. A stratified simple random sample design produced the sample selection of graduates, with geographical location, level of educational attainment, and field of study used to generate sample strata. Trained interviewers collected data through telephone interviewing, and the overall response rate ranged from 79% for the 1990 cohort to 49% for the 2009/2010 cohort. Given that non-response was non-random, sampling weights correct for non-response, subsampling, and post-stratification error.  3.3.2 Measures I examine two types of non-standard employment as dependent variables: temporary and part-time work. Part-time employment is defined as working 29 hours or less a week, treated as “regular wage employment in which the hours of work are less than ‘normal’” (Kalleberg, 2000, p. 343). The NGS survey measures temporary employment through a question asking respondents if their primary job (the week prior to the survey) is permanent 44  or not. Seasonal, casual, short term, or temporary defines non-permanent work. Across all cohorts, survey questions capturing average hours of employment were consistent. Measures for temporary employment changed slightly, with some surveys distinguishing among the type of temporary employment (i.e., casual or seasonal), but did not affect the formation of a measure capturing a broad category of temporary employment. Given that I seek to contrast rates of standard and non-standard employment, my analysis excludes unemployed, unpaid, and self-employed respondents and graduates who were attending a postsecondary program or living abroad. Due to both space constraints and field of study differences connected to credential level, I focus my study on outcomes among bachelor’s degree graduates.   One of the main independent variables of interest is field of study. I assess 28 different academic fields at the bachelor’s degree level. The categorization of field of study is unified across cohorts and is based on a similar system used by Andres and Adamuti-Trache (2007) (See Appendix A for details on categorization).13 The NGS records field of study using the Classification of Instructional Programs (CIP) taxonomy. A major expansion of the CIP classification system took place in 2000, reflecting the widening range of academic fields. To unify across survey periods, I collapsed new fields of study into previously used broader categories. In most disciplines, there was only minimal change. However, in others—most markedly within business—sub-specializations unrecorded in 1990 and 1995 are unexamined. Between 1990 and 2010, the number of students graduating from                                                  13 Some small differences exist due to a low number of respondents graduating from certain fields (e.g., zoology) 45  interdisciplinary programs also grew. However, due to the small number of interdisciplinary graduates in 1990 and 1995, I have excluded this discipline from analysis. I also include other education, employment, and demographic independent variables that are likely to influence rates of temporary and part-time employment. A dummy variable measures gender, while another captures respondents who took additional education programs longer than three months in the period between initial graduation and the survey period (e.g., captures the effect of additional education and possible later labour market entry). I also include a continuous measure of the amount of government student loan debt as an indicator of potential financial vulnerability.14 Demographic controls include two dummy variables capturing household composition: if a respondent self-reported being married or in a common-law relationship, and if he or she had a dependent child at the time of being interviewed. All models restrict analysis to respondents aged 18 to 64 and I model age as a continuous variable. I also include a series of binary indicators measuring industry classification, based on the North American Industry Classification System. Finally, given regional variation in employment outcomes, I also control for the province or territory of primary residence at the time of being surveyed through a series of binary indicators.                                                   14 My measure of student loan debt is adjusted for inflation among the survey waves to 2010 dollars.  46  3.3.3 Analysis  In part one, I present a descriptive overview of the percentage and proportion of men and women graduating from each field of study among 1990/1995 and 2005/2010 cohorts.15 I then assess how rates of non-standard employment differ by discipline. Due to both the level of segregation and the range in rates of non-standard employment, I am unable to release descriptive microdata by gender or separately by temporary and part-time employment. For example, despite surveying hundreds of nursing and engineering graduates, the small percentage of either men or women in these fields leads to residual cell counts below the threshold allowed for data release. Given this limitation, I use binomial logistic regression to generate the predicted probability of temporary and part-time employment by the percentage of female graduates within each of the 28 disciplines. Binomial logistic regression models the probability of a binary dependent variable, whereas a logit-function captures the probability of non-standard employment. The marginal effects generate a predicted statistic of how the gender composition of field of study—plotted to show a range from 10% to 90% female graduates—influences the likelihood of temporary or part-time employment separately for men and women. In part two, I assess the extent to which field of study mediates the relationship between gender and non-standard employment through Karlson-Holm-Breen (KHB) (Kohler, Karlson, & Holm, 2011) non-linear decomposition analysis. Based on binomial logistic                                                  15 The percentage of women graduating from each discipline provides an interpretable measure of gender segregation widely used within prior research (e.g., England & Li, 2006). However, it is affected by variability in the overall and subject-specific proportion of graduates. 47  regression, a KHB approach treats field of study as a confounder variable mediating the relationship between gender and non-standard employment. It decomposes the relationship between gender and employment into total, direct, and indirect effects. The direct effect measures the relationship between gender and non-standard employment net of discipline, while the indirect effect captures if field of study functions as a mediator. I also assess the contribution of individual fields to the overall indirect effect. Analysis is conducted separately for part-time and temporary employment and across 1990/1995 and 2005/2010 cohorts. I also run two separate KHB models. The first controls only for cohort, while the second adds demographic, education, and employment indicators. The second model provides insight into how other covariates impact the extent to which field of study functions as a mediator between gender and non-standard employment.  In part three, I present non-linear Shapley decomposition analysis separately by gender and time period (Azevedo, Franco, Rubiano, & Hoyos, 2011; Shorrocks, 2013). This method decomposes the marginal contribution of gender, field of study, and other demographic, education, and employment indictors on the likelihood of temporary or part-time employment two years after graduation. It identifies what part of observed differences in rates of non-standard employment are due to the distribution of circumstances (i.e., composition effect) and the extent to which circumstances influence a given outcome (i.e., coefficient effect). The Shapley approach decomposes the dissimilarities index—commonly termed inequality index D. The D-index ranges from zero to one and stands for the percentage of circumstances (i.e., field of study or industry of employment) that need to be “reallocated” between respondents employed and not employment in non-standard positions 48  to achieve equal risk of this type of employment. Used within inequality of outcome research (Marrero & Rodríguez, 2013), Shapley decomposition provides insight into the extent to which each circumstance contributes to inequality of a given outcome and, in turn, how sensitive an outcome is to these circumstances. Finally, I conducted all analysis in Stata version 14 (StataCorp, 2015). 3.4 Findings 3.4.1 Descriptive findings Table 3.1 presents an overview of the percentage and proportion of men and women graduating with bachelor’s degrees by field of study among the 1990/1995 and 2005/2010 cohorts. The percentage of women ranged considerably across disciplines, highest in nursing and lowest in computer science and engineering. Female graduates were over-represented in many social science disciplines, while economics, geography, and political science/criminology were slightly male-dominated across both periods. Some fields had an increasing percentage of female graduates among 2005 and 2010 cohorts, particularly law, agriculture, and architecture. Medical disciplines, such as pharmaceutical sciences and public health, were also increasingly female-dominated. Contrastingly, the percentage of women in engineering and computer science remained low across both periods. The proportion and percentages indicate that computer science had more male graduates among 2005 and 2010 cohorts, while engineering had slightly more female graduates in the later period. Across both periods, approximately one-third of all men and women graduated from business and education. A smaller proportion of 49  women graduated with degrees in education, psychology, and sociology/anthropology over time, while a larger proportion graduated from nursing and business. Finally, fewer men graduated with education degrees in 2005 and 2010, while the proportion of men in business and engineering disciplines increased. Table 3.1. Profile of gender segregation, 1990/1995 and 2005/2010 cohorts  Percent female grad Proportion female grad Proportion male grad Field of study 90/95 cohorts 05/10 cohorts 90/95 cohorts 05/10 cohorts 90/95 cohorts 05/10 cohorts Nursing 96.4 92.4 4.7 9.4 0.3 1.2 Laboratory & rehab medicine 87.5 78.9 2.0 1.7 0.4 0.7 Social work & public policy 84.7 87.8 2.7 3.0 0.7 0.6 Psychology 81.5 81.8 8.4 5.8 2.8 2.0 Literature 78.6 76.9 6.4 4.7 2.6 2.2 Sociology & anthropology 76.4 75.5 7.3 3.7 3.4 1.9 Public health 72.9 82.3 2.3 2.1 1.3 0.7 Languages 72.1 72.6 3.7 4.9 2.1 2.9 Fine arts 71.9 66.9 2.6 4.0 1.5 3.1 Education 71.6 72.7 24.3 18.7 14.3 10.9 Pharmaceutical science 70.8 77.4 1.2 1.2 0.8 0.5 Media 60.3 65.5 1.8 3.2 1.7 2.6 Biology 57.7 65.9 3.0 4.2 3.2 3.4 History 56.3 49.9 2.7 2.3 3.1 3.5 Geography 53.9 42.8 2.3 0.9 2.9 1.8 Medical sciences 52.8 71.4 1.4 1.4 1.9 0.9 Political science & criminology 49.2 51.3 1.6 2.9 2.4 4.3 Law 48.9 60.5 2.3 2.3 3.6 2.3 Business 48.8 51.6 10.7 15.7 16.7 22.9 Agriculture 48.0 59.7 0.3 0.4 0.5 0.5 Philosophy & religion 47.2 37.8 1.2 0.6 2.0 1.5 Chemistry 45.8 52.4 1.0 0.4 1.7 0.6 Environment studies 43.0 49.3 0.8 0.6 1.6 1.0 Math 36.3 36.9 1.5 0.9 4.0 2.3 Economics 27.9 41.6 1.3 1.4 4.8 3.2 Architecture 27.8 61.6 0.2 0.5 0.7 0.5 Computer science 22.8 15.2 0.9 0.6 4.5 5.4 Engineering 14.6 20.3 1.7 2.8 14.6 16.9 Total 59.8 60.8 100 100 100 100 Among 1990 and 1995 graduates, 28% of women and 24% of men were employed in temporary positions two years after graduation. The corresponding figures fell for 2005 and 2010 cohorts, where 22% of women and 21% of men were employed in temporary positions. 50  There was a smaller percentage of graduates employed in part-time positions across both time periods: 15% of women and 11% of men from the 1990 and 1995 cohorts, and 11% of women and 8% of men from 2005 and 2010 cohorts. Despite the small overall gender gap, rates of non-standard employment varied by discipline. Figure 3.1 gives an overview of the percentage of both men and women employed in a temporary and/or part-time position two years after graduation by field of study. Given that the total percentage of non-standard employment was higher for the 1990/1995 cohorts (e.g., 33% versus 27%), many fields of study have a corresponding decrease over time.   Note: Weighted percentages generated from the total pool of employed graduates to compare the rate of standard and non-standard employment. Figure 3.1. Rates of temporary and part-time employment by field of study, 1990/1995 & 2005/1010 cohorts Nonetheless, the percentage employed in non-standard positions grew in a few fields (i.e., environmental studies and laboratory and rehabilitation medicine). In contrast, rates of non-standard employment markedly decreased in medicine and law. The most important finding, 51  however, is that rates of non-standard employment substantially differed by field of study: approximately 10% or less in pharmaceutical sciences and engineering and 40% or more in education, literature, biology, history, and fine arts across both periods. Figure 3.2 presents the predicted probability of temporary employment two years after graduating with a bachelor’s degree by the percentage of female graduates within a discipline. The intent is to illustrate the relationship between the gender composition of a field of study and the separate likelihood of temporary employment for men and women. As the percentage of female graduates increases within a discipline, all four cohorts demonstrate a clear upward trend in the predicted probability of temporary employment for both men and women. The steepness of the curve varies by cohort. Nevertheless, when the percentage of female graduates is 10%, the probability of temporary employment is lowest (between 9% and 19% depending on the cohort). In contrast, when the composition of a field is 90% women, the likelihood of temporary employment is highest (between 26% and 39%). Unadjusted for other contributing education, demographic, and industry factors, findings show that both men and women graduating from female-dominated fields of study face increased risk of temporary employment.  52   Note: Graphical representation of logistic regression results with error bars at the 95% confidence level.  Figure 3.2. Predicted percentage of male and female graduates employed in temporary positions by percentage of women within a field of study, 1990-2010 cohorts Figure 3.3 presents the corresponding predicted probability of part-time employment by the percentage of female graduates within a field of study. It depicts a similar trend to temporary employment: rates of part-time employment increase as the percentage of female graduates within a discipline grows. Men graduating from female-dominated disciplines are more likely to be employed in part-time positions than those graduating from male-dominated disciplines. Nonetheless, as the percentage of female graduates increases within a field of study, a small gender divide in rates of part-time employment emerges. Thus, like 53  temporary employment, female-dominated fields of study have higher rates of part-time employment. Yet, unlike temporary employment, the increased probability of part-time employment among female-dominated fields of study disproportionally impacts women.   Note: Graphical representation of logistic regression results with error bars at the 95% confidence level.  Figure 3.3. Predicted percentage of male and female graduates employed in part-time positions by percentage of women within a field of study, 1990-2010 cohorts 3.4.2 The mediation effect of field of study  The descriptive findings demonstrate that rates of non-standard employment vary by field of study and the percentage of women within a discipline. To further assess the 54  interconnection between gender and employment type, I decompose the relationship into direct effects (e.g., net of field of study) and indirect effects (e.g., the mediating effect of field of study). The significant total effect in Table 3.2 denotes that women have slightly higher odds of temporary employment across both time periods. The diminished odds ratios in Model 2 signal that additional covariates partially explain women’s higher risk of temporary employment. Controlling for field of study reduces the direct effect of gender. The indirect effect—capturing the extent to which field of study mediates the relationship between gender and temporary employment—is statistically significant for Models 1 and 2 and across both time periods. However, the corresponding odds ratios for the indirect effect indicate that the additional covariates help explain the gender gap. That is, only a small mediating effect of field of study on the odds of temporary employment remains in Model 2. Table 3.2. Estimation of the relationship between gender and temporary employment as mediated by field of study  1990/1995 cohort 2005/2010 cohort  Model 1 Model 2 Model 1 Model 2  Log  odds Odds  ratio Log  odds Odds  ratio Log  odds Odds  ratio Log  odds Odds  ratio Total effect .394***(.070) 1.48 .275***(.072) 1.32 .412***(.085) 1.51 .229**(.088) 1.26 Direct effect .153*    (.077) 1.17 .152    (.079) 1.16 .124    (.094) 1.13 .063  (.096) 1.06 Indirect effect .240***(.030) 1.27 .123*    (.049) 1.13 .288***(.050) 1.33 .166**(.063) 1.18 N 10,721 10,522 13,290 12,921 R2 .05 .10 .07 .11 Standard errors in parentheses; * p < .05, ** p < .01, *** p < .001 Model 1 controls: cohort Model 2 control: cohort, age, presence of child in household, married or in common-law relationship, additional education since graduation, student loan amount, province of employment, and industry of employment.  As illustrated in Table 3.3, the corresponding results for part-time employment signify that women have much higher odds of part-time employment, a relationship that functions both directly and indirectly through field of study. As with temporary employment, the results across both time periods are similar. Unlike temporary employment, the direct relationship remains statistically significant in both Models 1 and 2. The corresponding odds 55  ratios also suggest that the direct effect of gender is similar to the mediating effect of field of study. The odds ratios indicate that the subjects women graduate from partially explain their higher rates of part-time employment. However, their propensity for part-time employment also functions regardless of what disciplines they graduate from. It is also notable that the inclusion of covariates in Model 2 reduces the odds of the mediating effect of field of study rather than the direct effect of gender. Together the results establish that women have slightly higher odds of temporary employment, comparably higher odds of part-time employment, and that field of study mediates both relationships.  Table 3.3. Estimation of the relationship between gender and part-time employment as mediated by field of study  1990/1995 cohort 2005/2010 cohort  Model 1 Model 2 Model 1 Model 2  Log  odds Odds  ratio Log  odds Odds  ratio Log  odds Odds  ratio Log  odds Odds  ratio Total effect 1.192***(.115) 3.30 1.028***(.115) 2.80 1.158***(.147) 3.18 1.008***(.144) 2.74 Direct effect .607***(.116) 1.84 .592***(.119) 1.81 .509***(.156) 1.66 .480** (.155) 1.62 Indirect effect .585***(.070) 1.80 .436***(.082) 1.55 .649***(.090) 1.91 .529***(.106) 1.70 N 10,702 10,508 13,209 12,846 R2 .09 .10 .11 .13 Standard errors in parentheses; * p < .05, ** p < .01, *** p < .001 Model 1 controls: cohort Model 2 control: cohort, age, presence of child in household, married or in common-law relationship, additional education since graduation, student loan amount, province of employment, and industry of employment. I next consider how each field of study contributes to the overall mediation effect between gender and non-standard employment. Given the variability in rates of non-standard employment among disciplines, each field may either propagate or attenuate gender differences. Demonstrated in Table 3.4, nursing has a negative contribution on the overall mediating effect of field of study between gender and temporary employment. That is, across both time periods, it reduces the overall likelihood of temporary employment for female graduates. Contrastingly, the under-representation of women in engineering, business, and computer science has a large positive effect on the overall mediating effect of field of study. 56  Women’s over-representation in some fields of study, such as psychology, sociology/anthropology, and education, reduces the overall likelihood of temporary employment for women graduating in 1990 and 1995—yet the same disciplines increase the likelihood of temporary employment for the 2005 and 2010 cohorts. Given that the percentage of women in these academic fields has not dramatically altered, the source of change is presumably the employment outcomes they engender. Finally, assessing the contribution of each field of study to the total indirect effect highlights that most fields have a negligible impact. Rather, the under-representation of women in business, engineering, and computer sciences, disciplines that tend to lead to permanent jobs, and the over-representation of women in education (for the 2005 and 2010 cohorts) are most substantial. The corresponding results for part-time employment reveal that engineering, business, and computer science have large positive contributions to the overall mediation effect of field of study for the 1990 and 1995 cohorts. Contrastingly, business and computer science cease to be highly influential mediating fields for 2005 and 2010 graduates. Rather, nursing, psychology, and education—three fields where women are over-represented and where part-time work is common—have a large contribution to the gender gap in part-time employment for the 2005 and 2010 cohorts. Remarkably, nursing decreases the likelihood of temporary employment but increases the likelihood of part-time employment. Thus, my findings suggest that field of study contributes to gender difference in rates of non-standard employment through both the under-representation of women in certain fields where standard employment relationships are common, and their over-representation in disciplines where non-standard arrangements are more prevalent. 57  Table 3.4. Contribution of each field of study to the overall mediating effect between gender and non-standard employment  Temporary employment Part-time employment  90/95 cohort 05/10 cohort 90/95 cohort 05/10 cohort  Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Education -7.41 -17.31 27.57 33.44 -0.36 -1.12 17.65 16.75 Fine arts -1.36 -2.97 2.48 3.17 0.68 0.34 2.76 3.79 Media -0.08 -0.80 0.01 -0.54 0.00 0.05 1.63 1.67 Languages -0.98 -0.81 0.22 -3.44 -0.47 -0.42 4.55 4.72 History -0.15 -0.12 -4.91 -4.44 -0.23 -0.30 -3.92 -4.27 Literature -1.05 -8.34 7.35 6.70 1.04 0.27 7.64 6.97 Philosophy & Religion 2.51 7.15 -3.01 -3.28 0.27 0.60 -2.27 -2.59 Economics 8.67 16.44 -1.70 1.65 4.65 4.80 2.03 3.65 Geography 1.83 2.52 -1.81 -1.62 1.00 0.92 -0.81 -0.22 Law 2.34 8.25 0.04 0.91 2.90 5.05 0.09 0.53 Environmental studies 2.94 5.46 -1.11 -1.57 5.48 6.29 0.00 0.07 Political Sci. & Crim. 1.07 3.83 -2.28 -0.87 1.70 2.51 -1.48 -1.74 Psychology -14.97 -35.16 8.27 2.96 -1.43 -2.53 13.47 13.59 Sociology & Anthro. -9.25 -20.69 1.81 -1.55 -0.62 -0.91 5.64 5.93 Social work & Policy -2.73 -6.54 0.36 -1.77 -1.68 -2.06 2.25 1.94 Business 34.66 55.29 13.69 21.04 15.32 15.56 -0.06 1.46 Chemistry 0.17 -0.01 -0.41 -0.25 1.01 0.82 -0.06 0.02 Biology -0.18 -0.02 3.10 3.59 0.23 0.13 2.03 2.04 Architecture 1.54 2.69 0.00 0.08 1.95 2.28 -0.01 0.05 Engineering 94.55 118.87 48.85 62.65 58.16 55.49 28.22 26.74 Computer science 24.60 48.17 8.70 16.01 12.30 16.11 2.51 3.89 Medical sciences -1.26 -2.32 1.26 0.81 1.37 2.56 0.00 -0.04 Nursing -24.72 -45.40 -6.23 -22.96 -2.31 -3.79 14.34 12.98 Pharmaceutical science -5.26 -15.32 -3.84 -9.66 -1.03 -2.44 -0.51 -0.94 Public health -2.94 -6.14 2.52 1.29 -0.77 -1.03 4.30 3.41 Lab. & rehab. medicine -6.68 -15.68 1.93 0.78 -2.85 -3.43 2.39 1.58 Math 4.16 8.94 -2.87 -3.14 3.68 4.27 -2.38 -1.96 Total contribution 100 100 100 100 100 100 100 100 Reference group: Agriculture Note: Assessing the contribution of each field of study is an extension of KHB decomposition generated through the disentangle function.  3.4.3 Assessing the contributing factors to non-standard employment  Model 2 indicates that other demographic, educational, and employment characteristics attenuate the mediating effect of field of study. I examine confounding effects in greater detail by considering the extent to which a graduate’s academic discipline affects the likelihood of part-time and temporary employment compared to other covariates. Decomposing the D-index assesses the dissimilarity in observed characteristics among respondents employed in standard and non-standard employment. The D-index represents the 58  percentage of compositional change necessary to achieve equal risk of temporary or part-time employment among graduates.  Table 3.5 illustrates that 27.5% of the circumstances modeled would need to shift to equalize the likelihood of temporary employment for the entire 1990 and 1995 cohorts. The percentage grows to 31.5% for the 2005 and 2010 cohorts. Across both periods, field of study and industry of employment are the main differentiating effects between temporary and permanent workers. Among the 1990 and 1995 cohorts, province of employment is also a notable factor influencing the likelihood of temporary employment. There is a reduced effect of geography and age among the 2005 and 2010 cohorts, although additional education since graduation becomes more influential for the later cohorts. The presence of a dependent child in a household, marital status, cohort, and student loans account for only a small percentage of the observed characteristics leading to temporary employment. Table 3.5 also presents results separated by gender. Here I examine the extent to which the characteristics of men or women employed in temporary positions differ from respondents of the same gender employed in permanent positions. Overall the findings are similar to the pooled analysis: field of study and industry of employment are most substantial in affecting the likelihood of temporary employment. Across both time periods, the contribution of field of study is slightly larger for men, while industry of employment is slightly larger for women. Additional education since graduation is somewhat more influential for men graduating in 1990 or 1995, although I find the reverse for the later cohorts. Province of employment has a slightly larger determining effect on temporary 59  employment for women graduating in 1990 and 1995, but becomes equal to men in 2005 and 2010.  Table 3.5. Shapley decomposition of the D-Index: Temporary employment  1990/1995 cohorts 2005/2010 cohorts  All Women Men All Women Men D-index     0.275 0.263 0.291 0.315 0.292 0.338         Percentage of D-index explained by each variable Gender 4.085   3.410   Age 5.264 6.376 4.932 0.527 0.803 1.284 Dependent child 0.499 0.767 0.716 1.083 1.073 1.121 Marital status 4.194 4.350 5.369 6.263 5.568 6.436 Additional education 6.129 5.806 7.330 12.566 14.585 12.264 Student loan  4.999 4.190 5.804 2.057 2.146 1.431 Cohort 1.809 4.376 1.344 3.429 2.309 5.359 Field of study 26.157 24.240 29.218 30.479 29.015 31.879 Province 9.741 12.515 8.638 3.525 4.793 4.569 Industry 37.122 37.380 36.648 36.661 39.708 35.656 Table 3.6 provides the corresponding results for part-time employment. Like temporary employment, the D-index increases slightly between the two periods. The larger D-index indicates that a somewhat higher percentage of the observed characteristics modeled would need to shift to equalize the likelihood of part-time employment. Rates of part-time employment are lower than temporary employment, yet the D-index is comparably larger, as the observed characteristics of graduates are more likely to determine the likelihood of part-time employment. Across both periods, field of study and industry have the greatest influence on the likelihood of part-time employment for both men and women. The results for age, the presence of a dependent child in a household, marital status, and student loans generally account for only a small percentage of the observed factors leading to a greater likelihood of part-time employment. Province of residence, gender, cohort, and additional education since graduation have a larger effect across both time periods.  60  Table 3.6. Shapley decomposition of the D-Index: Part-time employment  1990/1995 cohorts 2005/2010 cohorts  All Women Men All Women Men D-index     0.374 0.352 0.384 0.410 0.385 0.429          Percentage of D-index explained by each variable Gender 12.791   9.844   Age 1.071 1.297 1.518 0.167 0.454 0.527 Child 3.394 5.404 2.682 3.117 4.457 4.782 Marital status 1.225 2.297 0.292 2.262 1.416 1.994 Additional education 5.024 4.680 7.534 11.249 16.401 9.223 Student loan  0.599 0.870 0.781 0.207 0.322 0.868 Cohort 4.047 10.389 2.444 1.937 6.531 3.592 Field of study 27.047 28.195 30.417 34.481 31.295 43.422 Province 6.191 7.890 5.238 5.729 6.326 4.423 Industry 38.611 38.977 49.095 31.007 32.798 31.168 Examining the results separately by gender again shows a similar trend to the pooled analysis. Across both time periods, the total D-index is larger for men. The contribution of field of study and industry is also greater for men graduating in 1990 and 1995, while industry of employment is more substantial among women graduating in 2005 and 2010. The presence of a dependent child in a household makes a greater contribution to the D-index for women graduating in 1990 and 1995, but is similar among men and women graduating in 2005 and 2010. The impact of cohort and province is also larger for women across both time periods, suggesting that period and geographical differences in rates of part-time employment have a weaker impact on male employment outcomes. Finally, as with temporary employment, additional education since graduation increases the likelihood of part-time employment for men graduating in 1990 and 1995, while additional education is more influential for women graduating in 2005 and 2010. 3.5 Discussion  Non-standard employment among recent postsecondary graduates continually receives media attention, with headlines proclaiming that “job churn” is a “new reality” 61  (Younglai, 2016) as graduates “juggle jobs of varying certainty” (Dougherty & Jones, 2016). Unfortunately, media rarely discusses the extent of variation by field of study and gender. Across the time periods I study, women are slightly more likely to hold temporary contracts and much more likely to work in part-time positions two years after graduating with a bachelor’s degree. Yet the overall rate of non-standard employment masks discipline-specific variation. For example, approximately 50% of all employed history graduates held temporary and/or part-time positions two years after graduation across all cohorts. Contrastingly, 10% or less of pharmaceutical science or engineering graduates held non-standard positions. Not all female-dominated disciplines are at risk for non-standard employment and not all male-dominated disciplines protect against non-standard employment. Nevertheless, my research suggests that as the discipline-specific percentage of women increases, rates of temporary and part-time employment also grow. In examining my overarching findings in greater detail, I discuss next what fields of study are most influential, possible sources of change over time, other influences on the likelihood for non-standard employment, and policy implications. One purpose of my research was to uncover the extent to which field of study mediates the relationship between gender and non-standard employment. I show that enrolment patterns within specific academic disciplines largely explain gender difference in rates of temporary employment and partially explain the higher chance of part-time employment for female graduates. The under-representation of women in engineering, business (among 1990/1995 graduates), and computer science propagates gender inequality in non-standard employment. Such fields of study generally lead to male-dominated professions that tend to offer well-paid standard positions (Gielen & Schils, 2015; Prescott & 62  Bogg, 2013). Female-dominated disciplines play a more complicated mediating role. Psychology, sociology/anthropology, and education reduced the overall gender gap in non-standard employment for the 1990 and 1995 cohorts, although the same disciplines increased the overall rate of non-standard employment for women graduating in 2005 and 2010. Contrastingly, nursing decreased the overall likelihood of temporary employment among female graduates across both periods but increased part-time employment for 2005 and 2010 graduates. Connecting to this finding, prior research suggests nurses are able to obtain greater workplace flexibility due to their class standing and professional advantage (Gerstel & Clawson, 2014) and professional self-regulation (Adams, 2010).  On one hand, certain male-dominated disciplines were persistent in mediating the relationship between gender and non-standard employment between 1990 and 2010. Women continued to be under-represented in these disciplines and thus were not privy to the standard employment outcomes they provide. On the other hand, the percentage of women in academic fields that increased their likelihood of non-standard employment among 2005 and 2010 graduates did not drastically alter. My findings suggest that change in discipline-specific employment outcomes is more likely to impact the probability of non-standard employment than compositional change among graduating cohorts. Fluctuation in overall rates of non-standard employment is a notable trend in Canada, especially due to industry and economic downturn (Altonji, Kahn & Speer, 2016; Kahn, 2010). Of note, the cross-sectional design of the data utilized in this chapter does not directly capture the effect of the 2008 economic crisis, as 2005 graduates were interviewed in 2007 and 2010 graduates were interviewed in 2012. Although overall employment levels returned to pre-downturn levels by 63  2010 (Statistics Canada, 2012), employment rates among young adults fell between 2008 and 2014 (Bernard, 2015), while rates of part-time (Caranci & Marple, 2017) and temporary (DePratto & Bartlett, 2015) employment fluctuated. Graduates from certain fields of study may be at greater risk for early career non-standard employment; however, this risk cannot be constructed as simply characteristic of a discipline but rather is part of a school-to-work pathway impacted by industry and economic forces. Nevertheless, disciplines of study and types of skills are defined, valued, and rewarded in different ways (Armstrong, 2013; England, 1992; Steinberg, 1990) and may be more susceptible to labour market fluctuations (Caranci & Marple, 2017). Field of study is not the only influence on the likelihood of non-standard work. Part three of my findings show that industry of employment is also a notable characteristic effect dovetailing with previous research studying the relationship between field of study and industry (Morgan, 2008; Prix, 2009; Roksa & Levey, 2010). As discussed above, the likelihood of non-standard employment relates to industry and occupation-specific expectations and workplace practices. Giesecke and Schindler (2008) argue that structural relationships differentiating among certain fields of study generate “positional differences” in temporary employment “at the level of industry sectors” (p. 285). Nevertheless, I find that field of study still explains a sizable portion of the D-index even when controlling for industry of employment. That is, field of study influences the likelihood of non-standard employment even when accounting for a respondent’s industry of employment. Finally, other characteristic and composition factors had less of a definitive influence on the likelihood of non-standard employment. The nature of the data and the specific life course period it 64  captures may be a contributing factor to the comparatively small impact of family formation. Importantly, early career influences on non-standard employment will differ from those at other critical junctures in the life course. Inequality in non-standard employment outcomes by gender and field of study has implications for university, industry, and governmental policy. University policy often aims to equalize the ratio of men and women within academic disciplines. For example, the University of British Columbia set an explicit target to balance the percentage of men and women in engineering by 2020 (Qualizza, 2015). Additionally, certain industries promote gender equality in hiring and workplace practices. For example, the Ontario College of Teachers and the Ontario Teachers’ Federation (2004) formed a strategic plan with specific government recommendations to bolster the number of men entering teaching professions. Such policy initiatives remain important, as gender segregation within certain academic disciplines partially propagates different rates of non-standard employment among men and women. However, policy seeking to alter the “supply” of graduates is unable to address asymmetrical non-standard employment outcomes between fields. To alter unequal outcomes among disciplines, it is necessary to equalize “positional differences” in school-to-work pathways through policy that promotes standard employment relations between workers and employers. Compared to other OECD countries, Canada has weak employment regulation regarding non-standard employment, such as the type of work or eligibility for fringe benefits (OECD, 2013b). Implementing policy that aims to equalize employment practices, regardless of contract type, will promote high-quality employment among all workers (Jackson, 2006).  65  3.6 Conclusion  Researching employment outcomes two years after graduation does not fully capture labour market integration and longitudinal change. Importantly, I am unable to conclude if the gender gap in non-standard employment would change after the two-year survey period. More recent changes to the NGS limits examining change in gender inequality over time; however, prior research suggests gender inequality increases when comparing outcomes two and five years after graduation for the 1982, 1986, and 1990 NGS cohorts (Finnie, 2000). Without further longitudinal data, I am also unable to assess the consequences of early career non-standard employment. Nevertheless, research from other contexts highlights divergent outcomes originating from early non-standard career experiences (Allen & Van der Velden, 2007; Gebel, 2010; Scherer, 2004). In some cases, early career non-standard employment has no adverse effect on later employment outcomes (McGinnity, Mertens, & Gundert, 2005). In other cases, early career non-standard employment results in lower wages over time (Elia, 2010). Further, research suggests that the consequences of early career non-standard employment disproportionally impact women (de Vries & Wolbers, 2005). It is also necessary to highlight that no single employment measure can fully assess the level of job quality for recent graduates. Job quality within temporary and part-time positions ranges, with some positions offering high pay and fringe benefits. Still, there is evidence that experiences within non-standard employment differ for men and women. Men employed in non-standard positions—especially as contract workers—earn more, have greater wage returns over time, and receive more promotions (Drolet, 2002; Fernandez‐Mateo, 2009; Zeytinoglu & Cooke, 2008). Given this, it will be necessary for future research 66  to consider how various aspects of job quality differ among fields of study. Not all university graduates have advantageous employment outcomes given the high level of employment and wage variation by field of study (Foley & Green, 2015; Kim et al., 2015). Future research must continue to assess how social and economic outcomes vary among graduates at the same education level.  In Canada, postsecondary education increases the likelihood of full-time employment (Frenette & Morissette, 2014). Despite this trend, my research finds that a notable proportion of recent graduates work in non-standard positions. High rates of non-standard employment run contrary to the perception that advanced credentials facilitate access to standard employment contracts (de Vries & Wolbers, 2005). I demonstrate that a degree does not result in access to permanent and full-time early-career employment for all graduates. Notably, gender and field of study are key components in the distribution of non-standard employment among recent graduates. Both the over and under-representation of women in certain key fields partially explains inequality in non-standard employment. Although there are a range of other explained and unexplained factors, field of study propagates divergent pathways that lead to inequality in likelihood of non-standard employment.   67  Chapter 4: The Relationship Between Education and Workplace Task Discretion: An International Comparative Perspective   4.1 Introduction Over 40 years ago, Bowles and Gintis (1976/2011) argued that research in education must consider both labour power and “the demands of working people—for literacy, for the possibility of greater occupational mobility, for financial security, for personal growth, for social respect” (p. 240). Within the study of education and work, various perspectives take up Bowles and Gintis’s appeal by examining how education and skills not only generate productive and political forms of labour (Collins, 1979) but also have the power to transform work (Baker, 2009). Of central importance for Bowles and Gintis (1976/2011) is “the degree to which workers have control over planning, decision making, and execution of production and tasks, as well as sufficient autonomy to express their creative needs and capacities” (pp. 68-69). They argue that credentials generate social distance among workers by unequally distributing workplace discretion and other aspects of employment. Education yields the skills necessary for task discretion while simultaneously generating access to employment where it is available. The relationship between education and workplace task discretion differs by country. Country differences in the level of task discretion within the same types of jobs connect to the “different logics” of national employment systems “institutionalized in management, 68  training, bargaining” (Dobbin & Boychuk, 1999, p. 258). Research on education systems links higher levels of workplace task discretion to countries that emphasize industry-specific skills and on-the-job training (Edlund & Grönlund, 2010; Esser & Olsen, 2012). Contrastingly, other studies offer opposing evidence. For example, Germany has comparably low levels of workplace discretion despite promoting industry-specific skills through vocational education (Gallie, 2007). Beyond the type of training promoted, country differences connect to the historical and contemporary promotion of Taylorism, Fordism, and other managerial practices and technological change that seek to routinize work (Boxall & Winterton, 2015). Country-specific occupational structures influence both the overall level of workplace discretion and the extent to which it ranges across different occupations. The context-specific availability—both overall and among occupations—may affect the efficacy of education in determining access to task discretion. Through analysis utilizing the OECD Programme for the International Assessment of Adult Competencies (PIAAC), we consider the relationship between education and workplace task discretion in an international comparative context. We show that task discretion is an aspect of job quality unevenly distributed by country, occupational sector, literacy level, and education credentials. We first examine the direct association between education and task discretion in relation to the distribution of workplace discretion within a country. In this sense, we frame the relationship between task discretion and education as “interactive and territorially embedded” (James, Guile, & Unwin, 2013, p. 244). A comparative framework allows for an exploration into how the overall level of task discretion and its distribution among occupations within a country influences the relationship between 69  education and task discretion. Second, we compare individual-agency and critical-institutional theoretical perspectives by considering the extent to which skill and occupational sector mediate the relationship between education and task discretion. We also consider how processes of mediation are based on the country-specific distribution of workplace task discretion. We present our research in four parts: First, we explore the interplay among education, task discretion, occupation, and skill in our literature review. Next, we detail the research purpose, data, and modelling approach. Our last two sections provide the research findings and reengage the original research questions in our discussion and conclusion. 4.2 Literature Review Broadly, task discretion is the ability for workers “to participate in making decisions about their jobs and working conditions” (Kalleberg, Nesheim, & Olsen, 2009, p. 99) and, more specifically, “the extent to which employees are able to exercise independent initiative and judgment over their job tasks” (Zhou, 2014, p. 6587). Given that individuals with greater discretion report higher workplace satisfaction (Gallie, 2013; Green, 2004, 2008), management and organizational theory frame task discretion as an essential feature of a “high performance” workplace. It increases employee satisfaction, investment, and productivity (Appelbaum, Bailey, Berg, & Kalleberg, 2000; Becker & Huselid, 1998). Thus, task discretion is “a nonmonetary labour market reward” (Petrie & Roman, 2004, pp. 590-591). Not everyone has equal access to workplace task discretion. Where you live matters, with Nordic countries reporting the highest levels among highly industrialized nations (Dobbin & Boychuk, 1999; Edlund & Grönlund, 2010; Lorenz & Lundvall, 2011). Socio-demographic 70  aspects are also important, with women (Adler, 1993; Halldén, Gallie, & Zhou, 2012) and certain ethnic and racial groups (Petrie & Roman, 2004; McCrate, 2007) reporting lower levels. Two broad theoretical standpoints consider the relationship between education and task discretion: individual-agency perspectives and critical-institutional approaches. Such perspectives characterize task discretion as either a skill or a characteristic of employment. An individual-agency perspective explicitly or implicitly relies on a human capital framework. It casts education as an investment that produces skills benefiting an individual and the society in which he or she resides (Schultz, 1960). Education enables the ability to preform work tasks that require discretion and increases the “effective agency” of individuals (Mirowsky & Ross, 1998, p. 415). For example, Spenner (1990) argues that “autonomy-control” is a workplace skill, as it requires learned attributes such as “independence, leadership, and problem solving” (p. 404). An individual-agency perspective thus casts autonomy-control skills as individual attributes learned through training and education (Lorenz & Lundvall, 2011). Other researchers working from an individual-agency perspective highlight task discretion as an important component of workplace well-being necessary to promote widely (Mustosmäki, Anttila, Oinas, & Nätti, 2011). As Green (2006) writes, “an individual whose job involves choosing a set of tasks t for a wider set T can be taken to have a higher quality of work life than one whose job precisely prescribes that tasks t will be performed” (p. 13). However, within job quality literature there is ambiguity concerning the relationship among education, skill, and task discretion. First, skill level—which researchers often measure 71  through using education credentials as a proxy—is predicted to increase the discretion afforded to workers, as assumption not always supported empirically (Green, 2006). Second, directionality between skill and discretion is often unclear. That is, task discretion also promotes skill use, especially literacy, communication, numeracy, and problem-solving, and is not just a result of previously learned skills and competencies within education (Green, 2012). Literacy is especially influential on workplace participation as it enables “higher forms of communication that are entailed in facilitating learning” (2012, p. 39). Following an examination of individual-agency perspectives, two lines of inquiry arise: first, the separation between skills and the organization of work itself (Guile, 2002), and second, the extent to which education and/or skills facilitate access to task discretion. In response, critical-institutional perspectives situate task discretion as a feature of workers’ struggle for power within the labour process. Beginning with Braverman (1974/1998), scholars cast both the rise of monopoly capitalism and changing use of technology in the workplace as mechanisms contributing to processes of deskilling and reduced worker discretion (Lewis, 2007). The inability for individuals to utilize their knowledge and skills is one aspect of “subjective underemployment” (Livingstone, 2004, p. 85). Nevertheless, both Burawoy (1979) and Boltanski and Chiapello (2005) assert that post-Fordist contexts increasingly offer discretion in the workplace to prevent dissent, raise productivity, and control the labour power of workers. However, the implementation of workplace practices promoting discretion vary among occupation and countries. Several factors explain within and between country differences, such as the size, power, and history of trade unions, 72  employment policies, technology use, and type of industry sectors (Holman & Rafferty, 2017). Critical scholarship emphasizes both the structural and indirect aspects of the education-discretion association. Of concern is how individuals face varying types of workplace practices based upon the opportunities their education level affords. Systems of education and employment allocate status by propagating hierarchical divisions of labour (Arrow, 1973; D. K. Brown, 2001; Collins, 1979; Willis, 1977). Rather than understanding education as a means of endowing individuals with abilities—as individual-agency perspectives argue—critical-institutional approaches tend to follow a “screening hypothesis,” which argues that schools sort pupils in accordance with “characteristics employers are more likely to accept as evidence of greater productivity” (Sobel, 1982, p. 261). That is, credentials themselves, rather than the skills learned through education, bestow access to jobs with high task discretion. Nonetheless, processes of occupational sorting are context dependent given that the allotment of jobs to specific credentials varies from country-to-country (Henseke & Green, 2017). In summary, two broad perspectives frame the relationship between task discretion and education: one casts task discretion as an outcome of human capital or skill acquisition and focuses on freedom of choice at the individual level by, while the other provides a more structuralist account and focuses on how education credentials provide opportunities to higher quality employment where discretion is more readily available. 73  4.3 Research Overview and Design Our study explores the relationship between education and workplace task discretion from a comparative perspective. The literature review delineated two divergent theories surrounding the role of education:  a) An individual-agency framework, which assumes the connection between education and task discretion is based on education and skill acquisition, and  b) A critical-institutional approach, which alternatively argues that the relationship between education and task discretion is based on occupational sorting.  As arrow B1 in Figure 4.1 represents, an individual-agency approach predicts that skill will mediate the relationship between education and task discretion. Given the availability of indicators, we assess the effect of education on task discretion as mediated by literacy assessment scores. Contrastingly, critical-institutional perspectives suggest education functions as a “sieve” (Stevens, Armstrong, & Arum, 2008, p. 129), sorting individuals into various occupations that then create varied opportunities for task discretion. Arrow B2 in Figure 4.1 depicts the relationship between education and task discretion as mediated by occupational sector.  A direct relationship between education and task discretion would also partially support a critical-institutional perspective, which theorizes that credentials have a signaling effect beyond only occupational sorting. Nevertheless, we cannot distinguish arrow A from an individual-agency framework that would argue a direct relationship suggests other ability domains. Thus, any direct effect that remains after accounting for both skill and occupational 74  sector may be due to either unexplained ability or modes of educational sorting functioning beyond occupational sector. In another sense, the remaining direct relationship provides indication of how large the mediating effects of skill and occupational sector are in comparison to other ways education may influence task discretion.  Figure 4.1. Schema illustrating the potential relationships among education and task discretion We research the interplay between education and task discretion across national contexts, assuming that our schema is not universally constant, especially as skill and educational outcomes vary among countries (Busemeyer, 2015; Di Stasio, Bol, & Van de Werfhorst, 2016). Within each country of study, we examine the average level and the occupational range in reported level of task discretion (i.e., the gap between high and low task discretion occupations). Country-level indicators offer insight into whether education has an “absolute” or “relative” association with task discretion. In other words, if indeed education has an essential relationship with task discretion then there will be no interconnections among the direct and mediated pathways and the country-specific level and 75  distribution. Thus, from our framework of inquiry, three research questions arise: 1) does education have a direct influence on task discretion? 2) does skill level, as measured by a standardized literacy score, and/or occupational sector mediate the relationship between education credentials and task discretion? and 3) how do the three relationships differ across contexts?  4.3.1 Data PIAAC offers a wealth of survey and proficiency data to examine the relationship among education, skill, and task discretion across contexts. As of early 2017, approximately 216,250 non-institutionalized adults between the age of 16 and 65 have been surveyed in regional and minority languages around the world (OECD, 2013a, p. 2016). PIAAC allows for cross-country comparison of demographics, education, work, self-reported skill use, and standardized assessments in literacy, numeracy, and problem solving in technologically-rich environments. Round one of data collection took place between 2011 and 2012 and surveyed approximately 5,000 individuals in 24 countries and regions.16 Round two was conducted between 2014 and 2015, bringing the total number of countries and regions surveyed to 33. Round three, not yet released at the time of publication, will generate data for six additional countries.                                                  16 The one exception is Canada, where, due to the aim of capturing regional and population diversity, approximately 27,000 individuals were surveyed. 76  Each country was mandated to cover at least 95% of the entire population in their sampling frame, with a target response rate of 70% and a minimum response rate of 50%. Sampling differed in each country and was based on household or registry strategies. The survey is comprised of three main sections. Part one is a 45-minute background questionnaire examining demographic characteristics, education, employment, skill use outside of employment and, in some cases, country-specific questions. Subsequently, a five-minute module sorts participants into paper- or computer-based competency assessment modules. Within each assessment area, each respondent is allotted only a part of each test, with his or her final score imputed through 10 plausible values. Although inconsistent at the individual level, the use of plausible values minimizes measurement error at the population level (OECD, 2013a). Additional complexity when using PIAAC data surrounds handling the 81 replicate weights correcting for both country-specific sampling strategy and non-response.17 Given the complex survey design, statistical programs aid in the modelling process. Consistent with the recommendations of OECD analysts, analysis within our chapter uses both the “Repest” (Avvisati & Keslair, 2014) and “Piaactool” (Pokropek & Jukubowski, 2013) commands in Stata version 14 (StataCorp, 2015) to obtain correct estimates and robust standard errors. Our research uses the PIAAC public use file (PUF), a subset of the full national master databases. Analysis excludes Australia and Jakarta (Indonesia), countries that did not                                                  17 For most countries nonresponse is minimal. However, it is higher in some regions, specifically 17.7% in Cyprus, 5.2% in Flanders (Belgium), 4.5% in Lithuania, and 4.2% in the United States (OECD, 2016c). 77  release PUF. We also did not include Russia, due to the lack of availability of key indicators. Concerns of data quality for Russia have also led others to remove it from comparative studies (Heisig & Solga, 2015). Given our focus on task discretion in employment, we exclude unemployed individuals from all analyses. PIAAC uses the International Labour Organization’s definition of employment, defining individuals who worked at least one hour in the previous week as employed. We remove both respondents who worked less than one hour in paid employment and who self-described as unemployed from our research. 4.3.2 Methodological approach We initially study the direct relationship between education and task discretion through three approaches. First, we use pooled fixed effects and individual country ordinary least squares linear regression models (Models 1, 2, and 3) to investigate how select covariates impact the relationship between task discretion and education. Second, we descriptively compare countries by examining both the overall average and the occupational range in task discretion at the country level. Third, we explore how the results of Model 2—which includes all controls other than occupational sector—relate to the country-level average and occupational range in task discretion. We use a fixed effects regression approach in our pooled analysis to account for unobserved country-level heterogeneity, a common approach among PIAAC-based research (e.g., Hanushek, Schwerdt, Wiederhold, & Woessmann, 2015). We choose fixed effects rather than multilevel modelling due to constraints surrounding the sampling weights. PIAAC sampling weights are proportional to each country but scaled to its population size to facilitate estimation of the county-specific 78  population total. Thus, individual responses from large countries (e.g., Canada) would have weights disproportional to cases from small countries within multi-level modelling. The second part of our findings section examines the indirect relationship between education and task discretion. First, we conduct pooled fixed effects and individual country mediation analyses (Models 5 and 6) using the Karlson-Holm-Breen (KHB) method (Kohler, Karlson, & Holm, 2011). The KHB method describes the degree to which control variables—that is, occupational status (Z, see Figure 4.1) and literacy score (W)—mediate or explain the relationship between educational credentials (X) and task discretion (Y). That is, we decompose the effect of education on task discretion into direct, indirect, and full effects. Finally, we replicate part one by examining how the results of our KHB analysis relate to the country-level average and occupational range in task discretion. 4.3.3 Dependent variable A section of the PIAAC questionnaire measures perceptions of workplace skills and behaviours. We used three background questionnaire questions to form a derived indicator of task discretion. The source questions asked respondents “to what extent can you choose or change 1) the sequence of your tasks? 2) how you do your work? and 3) the speed or rate at which you work?” Respondents answered using a 5-point scale: “1) not at all, 2) very little, 3) to some extent, 4) to a high extent, and 5) to a very high extent.” We transformed responses to the three questions into a standardized summative measure with 125 unique responses ranging from -5.71 to 3.93. We assessed the reliability of the derived scale using Cronbach’s alpha and Pearson’s correlation coefficients. Item-test correlations were similar 79  across all indicators, ranging from .88 to .84, while item-rest correlations ranged from .73 to .65. The scale reliability coefficient alpha was .83 across all countries. At the individual country level, alpha was highest in South Korea (α=.92) and lowest in Finland (α=.71). 4.3.4 Independent variables The main independent variables of interest are formal educational credentials, literacy assessment score, and occupational sector. The PUF captures educational level through the International Standard Classification of Education (ISCED), a classification system that allows for comparison across different systems of education. Due to coarsened data and cross-national differences, we compare four levels of education: 1) lower secondary or less (ISCED levels 1, 2, and 3c short) as the reference group; 2) upper-secondary (ISCED levels 3a, 3b, and 3c long); 3) non-tertiary and professional diplomas (ISCED levels 4a, 4b, 4c, and 5b); and 4) tertiary bachelor’s and research degrees (ISCED levels 5a and 6). We examined several different methods of categorizing educational credentials but discrepancy among countries was a limiting factor. Namely, the PUF does not capture respondents with credentials at ISCED level 4 in Cyprus, France, South Korea, and the Netherlands.   One of the strengths of PIAAC is the ability to include standardized test results that measure skill level. Given the strong correlations among the three test areas, we only model literacy assessment score. As conceptualized by the OECD (2013a), literacy skill is “the ability to understand, evaluate, use and engage with written texts to participate in society, achieve one’s goals, and develop one’s knowledge and potential” (p. 61). Importantly, we did not arbitrarily choose literacy assessment scores. Literacy is one of the fastest growing 80  workplace skills and is foundational to the other types of skills (Green, 2012). Finally, literacy score serves as a proxy for individual skill and by no means fully captures the nuanced nature of individual ability. We weigh the drawbacks of this standardized indicator with the benefit of cross-country and group comparisons. We operationalise occupational sector through the International Standard Classification of Occupations (ISCO). The ISCO classification system enables an examination of task discretion for 1) managers, 2) professionals, 3) technicians and associate professionals, 4) clerical and support workers, 5) service and sales employees, 6) skilled agriculture, forestry, and fishery workers, 7) craft and tradespeople, 8) assemblers, plant, and machine operators, and 9) elementary occupations (such as cleaners and laborers). Due to the very small number of PIAAC respondents employed in the armed forces (n=627, .48% of the overall sample) we re-categorized these personnel to be part of a tenth category capturing missing occupational data. To examine task discretion across groups, we model occupational sector as a series of dummy variables. Plant and machine operators are the reference group, as they comprise the employment sector with the lowest level of task discretion across all countries. Finally, to take into consideration demographic and employment differences, we model several control variables: 1) gender, 2) age (in 10-year increments), 3) native or non-native speaker status, 4) part-time employment (29 hours or less a week), 5) public and non-profit sector employment, and 6) self-employment.  81  4.4 Results 4.4.1 Assessing the direct relationship between education and task discretion To examine the direct relationship between education and self-reported task discretion, we first consider the overarching association across all countries. Next, we consider how the direct relationship differs among individual countries and how the country-specific distribution of task discretion correlates to these findings. Model 1, illustrated in Table 4.1, examines the bivariate relationship between education and self-reported workplace task discretion across all countries. Without the inclusion of additional controls, education correlates with increased task discretion in the workplace. Compared to the reference group, individuals with education credentials above lower secondary self-report greater levels of workplace task discretion. The size of each education coefficient grows at higher credential levels, with individuals holding a bachelor’s degree or above self-reporting the highest level of workplace task discretion. Model 2 controls for literacy assessment score, gender, non-native speaker status, age, part-time employment, public sector/non-profit employment, and self employment.18 In comparing Models 1 and 2, all educational levels continue to exert a significant influence on task discretion; however, the coefficients diminish by 23% to 25%. The coefficient measuring literacy assessment score is significant but small, a finding that should be viewed with caution given the interrelation between education and literacy score. Notably, the                                                  18 Given that literacy assessment score has only a small effect when already accounting for education level, we do not include a separate model assessing this relationship (i.e., separating it from other control variables). Rather, our KHB analysis examines the relationship between education and skill more thoroughly.  82  inclusion of education controls in a model reduces the magnitude of proficiency coefficients (Holzer & Lerman, 2015). Table 4.1. Pooled linear regression estimation of self-reported level of task discretion  Model 1 Model 2 Model 3 Upper-secondary1 .467***(.043) .351***(.042) .166***(.043) Diploma1 .874***(.067) .673***(.067) .308***(.065) Degree1 1.320***(.049) 1.010***(.059) .423***(.050) Literacy assessment score  .006***(.001) .004***(.001) Women  -.040    (.035) -.148***(.035) Non-native speaker  -.456***(.080) -.432***(.080) Age 25-342  .258**(.064) .247***(.060) Age 35-442  .447***(.065) .394***(.065) Age 45-542  .465***(.068) .394***(.065) Age 55 and above2  .571***(.077) .465***(.072) Part-time employment  -.219***(.047) -.152*   (.047) Public sector employment  -.220***(.040) -.312***(.039) Self employment  1.622***(.050) 1.512***(.049) Managers3   2.361***(.086) Professionals3   1.689***(.091) Technicians and associate professionals3   1.696***(.082) Clerical and support3   1.463***(.091) Service and sales3   1.212***(.087) Skilled agriculture, forestry, and fishery3   1.242***(.102) Craft and trades3    .980***(.084) Elementary occupations3   .630***(.139) Occupational sector missing3   1.219***(.131) Intercept -1.174 (3.920) -2.441 (4.150) -2.775 (4.728) R2 .08 .16 .19 N=125,123; * p < .05, ** p < .01, *** p < .001; Robust standard errors in parentheses; Dummy country indicators included in all models. 1Reference group: Lower secondary or less.  2 Reference group: Age 24 and under. 3Reference group: Assemblers and plant and machine operators. In Model 3, occupational sector further weakens the influence of education on self-reported task discretion. Individually held credentials still have a significant positive relationship with level of task discretion, but the coefficient size diminishes by 53% to 58%. Bachelor’s degree holders self-report the highest average level of workplace task discretion, but occupational sector attenuates their educational advantage. Occupational sector of employment has a clear influence on level of task discretion, with managers self-reporting the highest average level when compared with assemblers and plant and machine operators. 83  Finally, the explanatory power of all three models, as portrayed by the R2, grows from .08 in Model 1 to .19 in Model 3. We next assess the direct relationship between education and task discretion within individual countries. Model 1, shown in Figure 4.2, separately examines the bivariate relationship between education and self-reported workplace task discretion for each country (see Appendix B for full models). Across most countries, education level has a significant positive relationship with self-reported task discretion. Nevertheless, there is no significant relationship at the upper-secondary level in eight countries (Lithuania, Poland, Ireland, Japan, Greece, Turkey, Finland, and Sweden), nor at the diploma level in four countries (Lithuania, Japan, Greece, and Turkey). Comparing across countries, findings show that the strength of the relationship between task discretion and education varies considerably. The coefficient representing credentials at the bachelor’s degree level or above is largest in the Slovak Republic (β=2.87, SE=.25, p<.001) and smallest in Sweden (β=.36, SE=.12, p<.01). Generally, the coefficients representing the relationship between education at the diploma level and task discretion are smaller. Still, the difference between the effects of a degree versus a diploma is narrow or non-existent in several countries (Germany, Italy, Finland, Norway, and Sweden) with overlap in the confidence intervals in many countries. Finally, the explanatory power of Model 1, as illustrated by the R2 reported in Appendix B, ranges from .00 in Sweden to .13 in Singapore and England/North Ireland.   84   Graphical representation of OLS results (see Appendix B for full models). Only coefficients significant at the 95% confidence level with corresponding confidence intervals shown. Figure 4.2. Individual country OLS regression results examining the direct relationship between task discretion and education 85  As shown in Figure 4.2, education continues to have a significant but reduced influence on self-reported task discretion in most countries with the addition of demographic and employment controls in Model 2. A few exceptions exist: the coefficient measuring diploma is now significant in Greece (β=.62, SE=.27, p<.05) and the coefficients measuring education at the bachelor’s degree level or above are larger in seven countries (Cyprus, South Korea, Finland, Turkey, Sweden, Ireland, and Poland). In Model 2, there is no significant relationship between task discretion and upper-secondary education in 11 countries, nor at the diploma level in three countries (Lithuania, Turkey, and Japan). As in Model 1, education at the bachelor’s degree level results in greater levels of self-reported task discretion in most countries, largest in the Slovak Republic (β=2.52, SE=.27, p<.001) and smallest in Norway (β=.35, SE=.12, p<.05). However, in an increasing number of countries, respondents with a diploma report similar or slightly higher average levels of task discretion compared with degree holders, with increased overlap among the confidence intervals. As the R2 indicates, the explanatory power of Model 2 ranges from .05 in Finland to .25 in South Korea. In most countries, occupational sector reduces the magnitude of the coefficients measuring education. As Model 3 in Figure 4.2 depicts, the relationship between task discretion and all education levels becomes or remains non-significant in nine countries (Chile, the United States, Japan, New Zealand, Greece, Turkey, Finland, Norway, and Sweden). Regardless of the effect of occupational sector, a direct relation between task discretion and education exists at the upper-secondary and diploma level in 13 countries and the degree level in 18 countries. Nonetheless, compared with Model 2, all credential coefficients have diminished in size with increased overlap in the confidence intervals. There 86  may be a direct relationship between education and self-reported levels of task discretion in some countries, yet comparing the results of Models 2 and 3 signals that occupational sector of employment is an influential factor across all countries. Finally, the explanatory power of Model 3 ranges from .09 in Finland to .28 in South Korea and the Slovak Republic. To consider what accounts for country differences in the relationship between education and task discretion, we examine next how the distribution of task discretion varies among countries. As illustrated by the vertical axis in Figure 4.3, the weighted average level of task discretion among PIAAC respondents is highest in Finland, Japan, and Austria and lowest in Italy, Lithuania, and Greece. Along with the average level of task discretion within a country, the horizontal axis in Figure 4.3 represents the distribution of task discretion across occupational sectors within each country. As an important marker of overall inequality, the range is largest in the Slovak Republic and smallest in Finland. On one hand, inequality in task discretion is distinct within each country, both in terms of the average level reported and the range between occupations. Yet, on the other hand, the bivariate relationship in Figure 4.3 suggests that higher levels of average task discretion across all occupational sectors narrows the gap among occupations within a country. Indeed, there is a moderate negative correlation between the two measures (r=-.40, R2=.16).19 The ability of education to generate access to employment with high levels of task discretion may also be dependent upon both the overall level and range between occupations within a country.                                                  19 Greece has a powerful outlier effect on the bivariate relationship between the average occupational task discretion and the range. When removed, the correlation increases to -.52 with an R2 of .27.  87   Figure 4.3. Bivariate relationship between average self-reported task discretion and range among occupational sectors  88  Next, we examine the extent to which the country-level average and occupational range influence the relationship between education and task discretion. Figure 4.4 illustrates the conditional effect of credential level by plotting the educational coefficients from Model 2 (on the y-axis) by both the average level of task discretion and the overall occupational range (on the x-axis). Net of demographic and employment controls, the negative correlation between education and the average level of task discretion in a country grows at higher credential levels, smallest at the upper-secondary level (R2=.02) and largest at the bachelor’s degree level and above (R2=.29). An even stronger relationship exists between education and the range in task discretion among occupations, a positive moderate-to-high correlation at the upper-secondary (R2=.41), diploma (R2=.44), and degree (R2=.67) levels. Our findings suggest that education has a smaller direct effect on an individuals’ self-reported level of task discretion in countries with higher average levels of task discretion; that is, the educational advantage is truncated when task discretion is ubiquitous. Conversely, a large range in task discretion among occupations heightens the direct advantage of higher credentials.20                                                    20 Pooled statistical models that include an interaction term between each education level and the country-level average or range allow us to test descriptive trends parametrically. Appendix E shows a significant negative relationship between task discretion and the degree-average interaction terms exists (both with and without occupational controls). Corresponding with Figure 4.4, we find a significant positive relationship between task discretion at the degree and diploma credential-range interaction terms (both with and without occupational controls). 89   Educational coefficients from Model 2 on the y-axis, country level average and range on the x-axis.  Coefficients control for gender, age, non-native speaker status, literacy, self-employment, public sector employment, and part-time employment.  Figure 4.4. Conditional effects of credential level by country  90  4.4.2 Assessing the indirect relationship between education and task discretion The results of Models 1, 2, and 3 demonstrate that the direct relationship between education and task discretion weakens with the inclusion of covariates, especially measures of occupational sector. What is less evident is the distinction between education and literacy proficiency scores. PIAAC researchers largely take two approaches to assessing the interconnection between skill and education: 1) As in Models 2 and 3 above, both assessment score and education are modeled as predictors and the effects of collinearity are accepted or ignored (Reder, 2015); or 2) path models are used to examine the mediating relationship between education and skill on a given outcome variable (Smith & Fernandez, 2015). We take the second approach, examining the extent to which occupational sector and literacy skill mediate the relationship between education and task discretion. Mimicking the prior section, we first present cross-country findings, followed by individual country models and an examination of how the country-specific distribution of task discretion correlates to the mediating power of occupational sector and literacy skills. Table 4.2 presents the results of two KHB models separately examining the mediating effect of occupational sector and literacy assessment score on the relationship between education and task discretion. The total effect is an additive portrayal of both the direct effect of education on task discretion, net of all controls, and the indirect effect functioning through occupational sector (Model 4) and literacy assessment score (Model 5). Of note, the direct effect gives the same coefficients as Model 3. The results show that, across all countries, both occupational sector and literacy assessment score mediate the relationship between education and task discretion. In Model 4, the indirect effect of occupational sector grows at each 91  additional level of education and is largest at the bachelor’s degree level and above. The mediating effect of literacy assessment score is smaller than occupational sector and also amplifies at higher credential levels. The results of Models 4 and 5 establish that alongside a direct relationship between education and task discretion—where individuals self-report higher average levels if holding advanced credentials regardless of their occupational sector and literacy score—education also indirectly influences task discretion through increasing literacy and providing access to occupational sectors where workplace discretion is more readily available. Table 4.2. Pooled estimation of the indirect relationship between education and task discretion    Model 4 Occupation  Model 5 Literacy Upper-secondary1 Total effect .351***(.042) .253***(.043) Direct effect .166***(.043) .166***(.043) Indirect effect .186***(.015) .087***(.011) Diploma1 Total effect .673***(.067) .435***(.062) Direct effect .308***(.065) .308***(.065) Indirect effect .365***(.023) .127***(.015) Degree1 Total effect 1.010***(.059) .616***(.055) Direct effect .423***(.060) .423***(.060) Indirect effect .588***(.033) .193***(.023) * p < .05, ** p < .01, *** p < .001; Robust standard errors in parentheses.  1Reference group: lower secondary. Country, demographic, and occupational controls included.  Figure 4.5 graphically presents the individual country results of KHB decomposition Models 4 and 5 (see Appendix D for full models). In all countries other than Spain, occupational sector mediates the relationship between education and self-reported workplace task discretion. Across most countries, the indirect effect of occupational sector is largest at the bachelor’s degree level or above, yet there is notable overlap between diploma and degree levels of education in several countries (Slovak Republic, Slovenia, Czech Republic, 92  Italy, Chile, Flanders/Belgium, Turkey, and Denmark). Occupational sector also mediates the relationship between credentials at the upper-secondary level and task discretion in most countries, although it is not significantly different from lower-secondary levels in four countries (Lithuania, Poland, Chile, and Denmark). In some countries, the coefficients measuring the direct and indirect effect through occupation are similar in size (i.e., Slovenia). Nonetheless, in most countries the indirect effect of occupational sector is larger. Together our findings suggest although education may directly influence level of task discretion, the more powerful indirect pathway is the way in which credentials provide access to occupational sectors where task discretion is more readily available. As illustrated in the bottom half of Figure 4.5, the mediating effect of literacy assessment score is smaller than occupational sector for most countries.21 Furthermore, there is notable overlap among credential levels and no significant indirect effect in 10 countries. In some contexts, such as Chile and Spain, the mediating effect of literacy skills is similar to or greater than occupational sector. Nonetheless, in most countries with a significant effect, literacy mediates the relationship between education and task discretion alongside other direct and indirect (via occupational sector) relationships between education and workplace task discretion. That is, literacy is not a full explanation for the relationship between education and task discretion.                                                   21 As a robustness check, we also assessed if the other PIAAC skill assessment domains exhibited similar mediating effect in Appendix C. Overall, the results were similar for both numeracy and problem solving in technology-rich environments.  93   Graphical representation of KHB results for the indirect effects of occupational sector and literacy score (see Appendix D for full models).  All demographic and occupational controls included, as well as occupational sector (Model 6) or literacy score (Model 7).  Only coefficients significant at the 95% confidence level with corresponding confidence interval shown (p < .05). Figure 4.5. The mediation effect of occupational sector and literacy by education level   94  Lastly, we examine correlations among the indirect effects of occupational sector and literacy assessment score and a country’s average level and occupational range in self-reported workplace task discretion. As demonstrated in Figure 4.6, the coefficients capturing the mediating effect of occupational sector (on the x-axis) have a moderate-to-strong positive correlation with the country-level occupational range in task discretion (R2=.28 to .47) and a much smaller negative correlation with the overall average (R2=.09 to .16). Displayed next in Figure 4.7, the coefficients measuring the mediating effect of literacy assessment score (on the x-axis) have no correlation with occupational range but exhibit a moderate negative correlation with the average level of task discretion within a country (R2=.33 to .39). Our correlative findings suggest that higher overall levels of task discretion within a country lessen the mediating effects of literacy, while a greater range in task discretion between occupations increases the mediating effect of occupational sector.   95   Figure 4.6. The indirect effect of occupation by the country-specific distribution of task discretion  96   Figure 4.7. The indirect effect of literacy by the country-specific distribution of task discretion  97  4.5 Discussion Our research refutes a purely individual-agency framework and presents support for critical-institutional perspectives for understanding the relationship between education and task discretion. Across most countries, the mediating effect of occupational sector is much larger than skill. Our findings indicate that educational credentials have the power to sort individuals into occupational sectors that are characterized by more or less workplace task discretion. That is, education increases opportunities through enhancing access to occupational sectors where task discretion is more readily available. In some countries, occupational sorting works in tandem with a smaller mediating effect of literacy and a remaining unexplained direct effect between education and task discretion. This evidence suggests an individual-agency framework functions in some countries, as skills gained through education also boost self-reported levels of workplace task discretion. Yet the mediating effect of literacy does not offer a complete explanation for the relationship between education and task discretion—rather, it functions alongside occupational sorting. Our second main research finding is that the relationship between education and task discretion is dependent on the country context. Both the overall level of self-reported task discretion and the range among occupations provide an explanation for country differences surrounding the direct and indirect relationships between education and task discretion. First, the direct influence of postsecondary education and the mediating effect of literacy is smaller in countries with higher overall levels of task discretion. Second, the greater the occupational range within a country, the larger the direct and occupationally-mediated effect of education 98  on task discretion. The correlational relationships suggest that the relationship between education and task discretion is relative, based on both the distribution of task discretion within a country and the overall average. Equitable access to workplace task discretion functions like a rising tide that lifts all boats; that is, the power of education diminishes and become less of a stratifying force when task discretion is higher overall and less unequal among occupations.  There are several important institutional factors argued to influence skill and educational outcomes. Hanushek et al. (2015) find that returns on skill are lower in countries with large public sectors, greater union density, and strong employment protection. Heisig and Solga (2015) demonstrate that characteristics of a country’s education system—such as the extent of primary or secondary-school tracking or size of vocational sector—influence the level of skill inequality. Such aspects are characteristic of certain welfare and production regimes, country-specific factors that support and promote specific forms of skill formation and trajectories (Busemeyer, 2015; Estevez-Abe, Iversen, Soskice, 2001). Gallie’s (2007) research casts doubt on the strength of a welfare production regime perspective in explaining cross-country differences in task discretion. Nevertheless, employment equality clearly does matter. Our research highlights that the relationship among education, skill, and task discretion is weaker in countries where there is greater equality among workers. Furthermore, our findings have implications for theories of educational sorting (Stevens et al., 2008) and screening (Sobel, 1982). Given country-specific differences, researchers cannot assume education has an “absolute” association with task discretion or, potentially, other employment outcomes. 99  We offer further insight into the connection among education, skill, and country context by drawing attention to the importance of considering the availability of employment and social outcomes. The direct and indirect relationships among skill, education, and occupation are relative to levels of inequality among individuals within a country. When occupations with high levels of task discretion are scarce, education acquisition matters more. As also suggested by other researchers (Fuller & Unwin, 2006), our research counters frameworks that propose individuals must possess certain qualities to be qualified to take on workplace tasks where discretion is necessary. Specifically, we suggest that the level and distribution of task discretion are key contextual variables framing the relationship between education and task discretion. Notably, as Pena (2015) argues, skill level itself does not determine unequal standing in society. Policy focusing solely on increasing skill and education levels risks overlooking the necessity of addressing inequality in workplace practices. Social policy must aim to counter educational stratification by promoting aspects of high quality employment, such as workplace task discretion, for all workers, regardless of their education levels. 4.6 Conclusion In Chapter 4, we illustrate the importance of education, alongside and interacting with individual skill and occupational sorting, for workplace task discretion. Employing a composite measure of task discretion, we demonstrate that in many international contexts education does not simply “enable” individuals to exercise workplace discretion. Rather, it functions indirectly through occupational sorting, and to a lesser extent through literacy enhancement. In particular, occupational sector has a large mediating effect on the 100  relationship between education and task discretion in the majority of countries, providing strong support for the critical-institutional and education signalling explanations. Trends in the distribution of task discretion by educational level, and differences in the predictors of task discretion more generally, show that societal arrangements structure individuals’ possibilities for workplace discretion in lasting ways.  Several factors limit our analysis, including the cross-sectional nature of the data, the lack of macro-level control variables for relevant country characteristics, and the inability to account for cultural differences. We recognized the first limitation, weighed against the richness of the PIAAC dataset, from the outset. Nonetheless, it leaves open the possibility of reverse causation, in particular regarding literacy scores and task discretion. As mentioned above, it is possible that task discretion enables informal learning that encourage greater literacy (Garrick, 1998). Second, as highlighted in our discussion section, additional country-level effects may shed light on further contextual factors that impact the relationship between education and task discretion. Lastly, our chapter does not account for the ways in which cultural differences among countries may influence the availability of and expectations for task discretion in the workplace. Indeed, Inglehart and Welzel (2005) argue that societal shifts towards post-industrialization place greater emphasis on self-expression and agency. Rather than directly measuring an absolute level of task discretion, country-specific cultural effects, workplace practices, and survey response styles may influence our self-reported measure. From a comparative perspective, we show that task discretion is unequally dispersed both within and among countries. We do not attempt to capture a complete account of the 101  asymmetrical international distribution of task discretion, especially as other important factors play significant roles, such as workplace organization and the strength of organized labour. Rather, our purpose has been to highlight the multiple and complex ways education provides access to workplace task discretion. It is necessary to further explore the interactions between education and the labour market to understand the mechanisms underlying the allocation of task discretion across contexts. Our research makes an important contribution to this aim. Education and skill do not offer a complete account of why some workers are afforded more or less task discretion. Rather, the relationship between education and task discretion is dependent upon the ways in which education sorts individuals in the labour market and overall access to and inequality between occupational sectors.    102  Chapter 5: A State of Mind or a State of Experience? Intrinsic and Extrinsic Educational Beliefs Over the Life Course  5.1 Introduction Educational beliefs are central to understanding people’s motivation and likelihood of educational participation. Opportunity structures—what, if any, educational choices are available to an individual—shape pathways through schooling, but are not the only influential factor. The degree to which opportunities align with individual dispositions, educational aspirations, and expectations matters as well (Andres et al., 1999). Scholars have long looked to educational beliefs as important in the creation and maintenance of inequality, with beliefs forming a type of “practical sense” at the individual level that guides action and choice (Bourdieu, 1980/1990, p. 67). Beliefs also function as “structuring” mechanisms, producing systems of dispositions that “generate and organize practices and representations” (p. 53) concerning the purpose, motivation, and rationale of education. Dispositions towards education thus provide a framework for understanding the persistence of educational inequality as it relates to the reproduction of consciousness (Bowles & Gintis, 1976/2011), what choices are “rational” (Breen & Goldthorpe, 1997), and practices of self-cultivation (Demerath, 2009). Research and scholarship on educational beliefs has focused primarily on school-aged children (e.g., Hegna, 2014) or postsecondary attendees (e.g., Mullen, 2010). There remains 103  no insight into how educational beliefs change over adulthood, especially in response to life course experiences and social position. This is increasingly important to understand, as individuals engage and reengage with education at multiple points in their lives. Further, educational beliefs can be extrinsically or intrinsically motivated. Whereas, extrinsic beliefs capture educational outcomes, intrinsic beliefs surround alignment towards skills and learning itself (Tomlinson, 2013). Extrinsic or intrinsic educational beliefs may be influenced in different ways by various employment, education, and demographic factors. In this sense, people’s orientation is part of their “continually constructed” educational disposition (Bourdieu, 1980/1990, p. 60). Given that past research has not considered change or consistence in educational beliefs beyond early adulthood, we study how social background factors and life course experience influence variation in intrinsic and extrinsic educational beliefs from age 25 to 46. To present insight into how extrinsic and intrinsic educational beliefs vary among individuals and change over adulthood, we pursue two primary explanations: “demographic” theories, which emphasize life course circumstances connected to social origin, and “experience” perspectives, which consider how life course activity promotes certain worldviews. We pose two research questions: 1) how do parental education and gender influence the level and variability over time of extrinsic and intrinsic educational beliefs? And 2) how do postsecondary and labour market participation influence the level and variability over time of extrinsic and intrinsic educational beliefs? We present our examination in five parts. First, drawing on research from the sociology of education and work, we define key terms and explore research on the interplay between education and 104  beliefs. Parts two and three outline our research purpose, questions, and design. Our primary mode of analysis is hierarchical growth modelling, a unique and insightful approach for longitudinal and repeated measured data. After detailing the research findings, the discussion and conclusion re-engage the original research questions and discuss model assumptions and limitations.  5.2 Literature Review 5.2.1 Defining intrinsic and extrinsic beliefs Beliefs form taken-for-granted systems “about how one ought not to behave, or about some end-value of existence worth or not worth attaining” (Rokeach, 1968, p. 124). A belief is defined as “a structure of socialized feeling, contingently allied to discursive practices” (Cromby, 2012, p. 945). Under this definition, it is important to frame the notion of values and beliefs as intimately intertwined: values are expressions of personhood that form desirable actions or choices guiding behaviour (Hitlin & Piliavin, 2004). Beliefs, as precursors to values, are socially constructed through one’s experience and social position. Nevertheless, beliefs are not simply individually-held entities—they are part of cultural schemas imbedded in specific contexts. A range of factors influence stability or changeability in beliefs at the societal level, such as media (Happer & Philo, 2013) and social inequality (Newman, Johnston, & Lown, 2015). Finally, a belief may have an intrinsic or extrinsic orientation, insofar as it represents an inherent good or instrumental motivator, respectively. Intrinsic and extrinsic beliefs are not necessarily dichotomous or opposing (Reiss, 2012). Nevertheless, examining intrinsic and extrinsic beliefs separately is a way to understand their distinct influences.  105  Within the sociology of work, gender and parental education are two influential “demographic” indicators influencing beliefs. Women typically self-report a higher intrinsic and lower extrinsic work orientation when compared to men (Hjort, 2015; Krahn & Galambos, 2014; Lindsay & Knox 1984; Marini, Fan, Finley, & Beutel, 1996), with gender role and socialization factors accounting for the trend (Kalleberg & Marsden, 2013). The impact of parental education is less conclusive. Depending on the study, high level parental education has a strong (Halaby, 2003; Johnson & Mortimer, 2011) or weak (Hjort, 2015) positive relationship with both intrinsic and extrinsic work beliefs. Additionally, both gender and parental education are influential demographic factors on educational participation (David, Ball, Davies, & Reay, 2003; Dickson, Gregg, & Robinson, 2016). Rather than suggesting gender or parental education has a biological or genetic influence, such studies establish that social position lead to specific forms of socialization, experiences, and opportunities. Life course experiences may affect the influence of demographic factors on intrinsic and extrinsic beliefs. “Experience” indicators are trajectories of life course roles and activities that vary by age and cohort (Elder, Johnson, Crosnoe, 2003). Individual experiences range over childhood and adulthood; for example, educational participation tends to be highest earlier in the life course, while employment increases after leaving school. Educational and employment experiences also vary by cohort, as historical change influences the availability of opportunities. Longitudinal studies reduce such confounding effects while simultaneously enabling different experiences among individuals to be assessed. For instance, comparative cohort studies reveal that although full-time employment generally 106  increases extrinsic work beliefs, generational differences are influential on the relationship (Krahn & Galambos, 2014). Life course experiences, such as employment and educational participation, may also be influential on initial educational beliefs and how they change over time. 5.2.2 The cultivation and maintenance of educational beliefs A range of historical and contemporary educational scholarship has considered how educational beliefs are a component of opportunity structures (Bowles & Gintis, 1976/2011; Holm Jæger, Karlson, & Reimer, 2013), rational action (Breen & Goldthorpe, 1997; Elster, 1983), and the formation and maintenance of habitus (Ball, Davies, David, & Reay, 2002; Bourdieu & Passeron, 1970/1990). Research on opportunity structures suggests that possibilities and barriers shape educational beliefs. The perception of barriers diminishes educational aspirations (Shapka et al., 2012) and shapes “beliefs about how schooling will pay off” (Downey, 2008, p. 109). A rational action perspective argues that two mechanisms influence educational beliefs: primary effects, such as direct social barriers to schooling, and secondary effects, such as individual decision-making processes that arise from such barriers (Breen & Goldthorpe, 1997; Goldthorpe, 1996). As Sullivan (2006) argues, “we cannot generate any hypotheses about action simply on the basis that it is rational without making any claims about the beliefs and desires held by the actor” (p. 272). Action is “rational” if it is the best way of fulfilling one’s desire, yet such desires must be in line with one’s beliefs. Rather than assuming a “rational” cultivation of beliefs, other educational researchers consider the influence of dispositions and habitus in the formation and maintenance of belief 107  systems that guide educational choices (Hatcher, 1998). As part of an individual’s habitus, beliefs provide an explanation for the formation and execution of educational choices, framed as “a choice of lifestyle and a matter of ‘taste’” (Ball et al., 2002, p. 53). Beliefs emerge as “deeply normalised grammars of aspiration” (p. 69) and can be connected to a range of educational experiences and action, such as the performance of identity (Archer, Hollingworth, & Halsall, 2007) and the academic subjects students study (Mullen, 2010). Beliefs link individuals and groups “to the world through the medium of symbolic meaning” (Borhek & Curtis, 1975, p. 5), while “objective structures are incorporated into the body and the ‘choices’ constituting a certain relation to the world are internalized” (Bourdieu, 1977, p. 662). Prior research has illustrated the intricate process of belief formation, maintenance, and change, from Macleod’s (1987) research on how peer group belief systems may function to devalue conventional educational success, to Demerath’s (2009) fieldwork that examines how middle-class belief systems that idealize self-cultivation unequally position individuals. Prior research continually highlights that educational choices are not only based on utility maximization but also emerge from a framework of conscious and unconscious beliefs. An emerging body of research studies how extrinsic and intrinsic educational beliefs differ by gender and social class. Sullivan (2006) finds no significant parental class difference in educational beliefs among Grade 11 students, but her work shows that girls had slightly more positive intrinsic beliefs towards education. Nevertheless, parental education and class is influential in other studies. Mullen’s (2010) research, juxtaposing how students at elite and state universities understand their educational choices, finds that a “distance from necessity” (p. 201) shapes modes of educational distinction. Elite students cultivated their 108  “intellectual and personal qualities for the intrinsic satisfaction of doing so” (p. 209) rather than “the extrinsic goal of getting a good job or preparing for a specific career” (p. 73). Woodfield (2012) reports similar results when studying mature female students who are much more likely to promote a “love of learning” (p. 101) rather than employability as their reason for returning to school. Working-class identity promotes extrinsic beliefs surrounding employability (Lehmann, 2012b) and engaging in “forms of learning that have more immediate economic relevance while discarding those that are likely to have minimal bearing on their anticipated working lives” (Tomlinson, 2013, p. 98). Given the small body of educational research that directly considers intrinsic and extrinsic beliefs, it is worthwhile to examine additional scholarship from the sociology of work that considers this subject.  Within the sociology of work, educational choices are the “first step in realising one’s preferences for work” (Dæhlen, 2005, p. 398). A proliferation of scholarship examines intrinsic and extrinsic employment beliefs in ways that have deep implications for education. Research focuses on how orientations towards employment are shaped, change over time, and react to experiences and level of labour market participation (Jin & Rounds, 2012; Johnson, Sage, & Mortimer, 2012; Krahn & Galambos, 2014; Twenge, Campbell, Hoffman, & Lance 2010). Researchers consider a multidimensional, two-way relationship between work and education: on one hand, intrinsic and extrinsic orientations influence educational outcomes and trajectories (Johnson & Elder, 2002), while, on the other hand, education level and aspirations influence the construction of beliefs (Kalleberg & Marsden, 2013; Krahn & Galambos, 2014). As Hjort (2015) writes, “while doing well in school can strengthen students’ intrinsic work values, a commitment to learning, creativity, and the development of 109  skills may also be conducive to academic success to begin with” (p. 307). Importantly, extrinsic and intrinsic education and work beliefs are co-constituting.  Both postsecondary and employment experiences influence extrinsic and intrinsic work orientations. Education leads to opportunities for higher quality employment with intrinsic benefits (Johnson et al., 2012; Lindsay & Knox, 1984) and extrinsic rewards (Johnson & Elder, 2002). Kalleberg and Marsden (2013) argue that individuals with postsecondary education “are more apt to regard obtaining a well-paying and secure job as non-problematic, and so are freer to prioritize ‘higher order needs’ such as accomplishment” (p. 266). Employment experiences also influence change in beliefs over time. For example, unemployment and underemployment weakened extrinsic values in a longitudinal study of young adults conducted between 1991 and 2009 (Johnson et al., 2012). Even though our chapter does not study work beliefs, prior research from the sociology of work is influential for two important reasons: first, it offers insight into how intrinsic and extrinsic beliefs are operationalized; and second, it is pertinent when we reverse the role of education to consider how labour market participation influences educational beliefs over time. 5.3 Research Design and Overview 5.3.1 Research questions  Prior explorations within the sociology of education and work present initial indication of how life course experience and social position may shape intrinsic and extrinsic educational beliefs. As discussed above, past research primarily considers educational beliefs in childhood and early adulthood and offers little insight into how beliefs change in response 110  to later education and labour market participation. Gender and parental education are social background effects capable of transmitting and reproducing educational beliefs through early socialization. Nevertheless, how demographic factors influence change in educational beliefs over time remains unexplored. Thus, our first research question asks the following: how do parental education and gender influence the level and variability over time of extrinsic and intrinsic educational beliefs? As discussed in our literature review, prior research suggests that women and individuals with university-educated parents hold higher intrinsic beliefs, while men and individuals whose parents have not attended university hold higher extrinsic beliefs. Yet the extent to which the effects of gender and parental education continue to influence beliefs over adulthood is unclear. Notably, Lucas (2001) argues parental status has less of an effect over the life course as it becomes mediated by other outcomes. A primary explanation for the diminished effect of demographic factors over time is the influence of life course outcomes and experiences. This leads to our second research question: how do postsecondary and labour market participation influence the level and variability over time of extrinsic and intrinsic educational beliefs? There is little insight into how education and work increase or diminish educational beliefs for adults. Research on youth shows that perceived barriers to higher education diminish educational aspirations (Shapka et al., 2012). Thus, we can predict that higher levels of postsecondary participation promote more positive beliefs towards education. Nonetheless, it remains unknown whether education has a similar influence on extrinsic and intrinsic educational beliefs. Above, scholarship within the sociology of work finds that experiences of unemployment or low levels of labour market participation weaken extrinsic work beliefs and high levels of 111  participation strengthen beliefs. Our research can assess if work and schooling influence educational beliefs in the same way. 5.3.2 Data The Canadian Paths on Life’s Way longitudinal cohort survey is a rich database for considering educational beliefs. The Paths study follows the multiple life course activities of participants who graduated from high school across the province of British Columbia in 1988 (Andres, 2002, 2013). The provincial government designed the study to capture the transition between high school and postsecondary schooling and employment. As time went by, the mandate broadened as participants entered adulthood and started careers and families. Over 28 years (1988-2016), mailout surveys focused on education, employment, and family patterns. The surveys capture month-by-month activity data, employment histories and experiences, family formation and change, and inquires into the justification and importance of such experiences.  Alongside demographic and background information, our research employs two key forms of data generated through the survey. Starting in 1993, five years after high school graduation, participants reported levels of agreement or disagreement with a series of belief statements surrounding education. The study also gathered detailed information on higher education and labour market participation through month-by-month activity data indicating employment (full time, part-time, or unemployed) and/or school attendance (full time or part-time). By 2016, a total of 510 respondents answered all survey waves, a sufficient sample size for the analysis below (Maas & Hox, 2005). Over time, attrition produced sample bias 112  toward women, respondents from more educated backgrounds, and individuals with better high school performance.22  5.3.3 Dependent and independent measures  Participants first responded to a set of Likert-type educational belief statements five years after high school graduation. Respondents indicated the extent to which they agreed with each statement using a five-point balanced scale: 1) strongly disagree, 2) disagree, 3) no opinion, 4) agree, and 5) strongly agree. We chose six belief statements to separately examine as dependent measures among the 15 statements consistent across survey waves. Our rationale for selecting statements was theory driven and, following Sullivan (2006), we choose three items focusing on skill and inclination towards education as statements orientated towards intrinsic beliefs: 1) “postsecondary education is not for me”; 2) “my education has improved my communication skills”; and 3) “my education has improved my reasoning skills.” We selected three items on income, lifestyle, and work outcomes as statements representing extrinsic beliefs: 1) “I can’t get ahead these days without postsecondary education”; 2) “to attain the lifestyle I want, I must have a university degree”; 3) “I need a university degree to earn a decent income.”23 Identical belief statements and                                                  22 Attrition was higher early in the study. The response rate for the original stratified random sample of high school graduates (N=10,000) was 53% (N=5345) in 1989. The response rate was 42% in 1993 (N=2220), 48% in 1998 (N=1055), 69% in 2003 (N=733), 78% in 2010 (N=574), and 89% in 2016 (N=510).  23 We excluded statements that did not focus on personal orientation towards education (“Canada’s future economic competitiveness is dependent upon a highly skilled work force,” “these days, people require higher levels of education than they did in the past,” and “given the way things are, it will be much harder for people in my generation to live as comfortably as previous generations”), items that only loosely associated to education (“to stay gainfully employed in the future, I must be highly skilled in a given field/specialty,” and “it is important that my job be related to my field of study or specialization”), and articles that did not align fully with  113  response formats were repeatedly surveyed in 1993, 1998, 2003, 2010, and 2016, capturing change in educational beliefs over five separate time points.  Independent variables include both demographic and experience indicators. Demographic variables were operationalized through two dummy variables: gender (women=1) and parental education (one or more parent completed a bachelor’s degree or higher=1). To examine how employment and educational experience relate to beliefs, summative measures capture months of work and participation in formal education. We count each month a participant self-reported full-time employment and/or enrollment in a postsecondary institute and measure part-time employment or postsecondary enrolment as half a month (i.e., .5). Examining the number of months enrolled in postsecondary education captures level of participation rather than highest credential earned. Summative measures do not fully capture a “credential effect” (Bills, 2016, p. 65) but offer a “tangible representation of investment in education” (Hauser & Featherman, 1976, p. 100). Additionally, measuring both employment and education through months of participation yields comparable measures between employment and education over adulthood. 5.3.4 Analysis Our study uses ordinal growth curve analysis, a form of multilevel/mixed modelling that examines multiple responses given over time to the same ordinal outcome variable.                                                  either intrinsic or extrinsic educational beliefs (“I expect to re-enter the post-secondary system more than once over my life time,” “my education has been useful in helping me find a job,” “my education has improved my career prospects,” and “I believe I am better off with a post-secondary education”).  114  Ordinal growth modelling is based on two levels of analysis that generate proportional odds using a cumulative logit link function for responses (see Appendix F for model equations).24 We conducted all analysis in Stata version 14 using robust standard errors adjusted for individual-level clustering (StataCorp, 2015). In growth analysis, level one captures how an outcome variable constituting each belief statement increases, decreases, or remains constant over time.25 How the predictor “time” is specified impacts if trajectories of change are assumed to be linear (Singer & Willett, 2003). We model a linear measure of time as a random effect within an unstructured covariance matrix. We also include a fixed quadratic term for time to capture curvilinear change in beliefs.26 To interpret initial 1993 responses more easily, measures of time begin at zero and are proportionally coded to mirror elapsed time between surveys (i.e., 1998=1.2, 2003=2.6, 2008=3.6, 2016=4.6).  Given that education and labour force participation vary over the life course—with individuals often studying in early adulthood and increasing their frequency of employment upon graduation—level one includes fixed time-varying measures of months of education and employment prior to each survey period. We centre both indicators around the group-mean (i.e., averaged at the individual level) to capture change over an individual’s life course                                                  24 Unlike logistic regression, ordinal approaches provides cumulative probabilities (i.e., P(1)=responses to 1, P(2)=responses to 1 and 2, P(3)=responses to 1, 2 and 3, P(4)=responses to 1, 2, 3, and 4, P(5)=1). The logit function provides thresholds/cutpoints that are interpreted as the difference necessary to adjust from each probability function to the next. Importantly, parameterization in Stata does not include a constant term, as it is incorporated into the cutpoint. 25 Level one also accounts for differences between the true and observed trajectories through random error terms that rely on the two-level structure of the model (for a deeper discussion on the composite residual term see Raudenbush & Bryk, 2002). 26 See Appendix G for a comparison of linear and curvilinear growth models.  115  and deviation from their overall average at each survey period. Centring also controls for variation in education and employment participation from age 25 to 48. For instance, a respondent may self-report higher educational beliefs while going to school. The inclusion of time-variant measures is necessary for two reasons: our research seeks to examine change over a longer period, and education and employment participation varied among individuals.  Level two examines how change varies among individuals and includes covariates that explain the individual-level variance in the level one model. We model dummy variables for gender and parental education and continuous measures capturing the average months of employment and education (i.e., total months divided by five survey periods). Interaction terms between level-two dummy variables and time/time2 capture the extent to which demographic indicators influence the direction and rate of change in beliefs. Measures of average employment and education capture a stable underlying individual orientation to education and work. Hence, time-varying measures of education and employment participation at level one accounts for change over a participant’s life course and time-invariant measures at level two enable comparison among individuals. 5.4 Findings 5.4.1 Descriptive findings Table 5.1 provides the means and standard deviations of all dependent and independent variables. Average responses decreased slightly over subsequent survey waves (i.e., less agreement) among all extrinsic belief statements. Level of agreement was highest for the statement “I can’t get ahead these days without postsecondary education” and lowest for “I 116  need a university degree to earn a decent income.” Intrinsic belief statements were more constant over time. Participants were most likely to agree with the statement “My education has improved my communication skills” at all data collection points. On average, participants disagreed that “postsecondary education is not for me.” Nevertheless, respondents were slightly more likely to agree with the statement over time.  Table 5.1. Means and standard deviations of dependent and independent variables  Years after high school graduation Extrinsic belief statements 5 10 15 22 28 ‘I can’t get ahead these days without postsecondary education’  4.12 (1.11) 3.95 (1.07) 3.93 (1.05) 3.95 (1.07) 3.88 (1.07) ‘to attain the lifestyle I want, I must have a university degree’ 3.55 (1.40) 3.40 (1.37) 3.31 (1.27) 3.42 (1.23) 3.40 (1.24) ‘I need a university degree to earn a decent income’ 3.33 (1.34) 3.13 (1.32) 3.08 (1.19) 3.18 (1.15) 3.12 (1.15) Intrinsic belief statements      ‘Postsecondary education is not for me’ 1.49 (.83) 1.55 (.80) 1.62 (.84) 1.59 (.77) 1.66 (.83) ‘My education has improved my communication skills’ 4.23 (.89) 4.19 (.82) 4.09 (.84) 4.15 (.82) 4.18 (.78) ‘My education has improved my reasoning skills’ 4.09 (.85) 4.02 (.82) 3.94 (.85) 3.97 (.76) 4.05 (.83) Level 1 indicators      Months of postsecondary prior to survey 27.96 (14.80) 9.60 (11.63) 5.19 (11.40) 2.75  (6.89) 1.38  (4.95) Months of employment prior to survey 28.41 (14.67) 36.20 (15.27) 51.20 (18.24) 62.76 (18.59) 59.52 (19.63) Level 2 indicators      Total years of postsecondary education (months/12) 3.91 (2.47) Total years of employment (months/12) 19.84 (4.67) Gender (women=1)  .61 (.48) Parental education (PSE degree or higher=1) .34 (.47) Standard deviation in parentheses; N=510; response range for belief statements=1 (strongly disagree) to 5 (strongly agree). Labour market and postsecondary participation was more changeable. Over the 28-year period, months of employment increased and months of postsecondary education decreased. All participants worked at some point during the total 28 years. Respondents also had high levels of postsecondary participation. One year after high school graduation 19% had not yet entered a postsecondary institution. However, over the 28-year period studied, this proportion reduced to 4%. The low proportion of non-attendance captures highly 117  variable levels of participation, from individuals who only took short-term courses to those who received advanced degrees. By 2016, 59% of the sample held baccalaureate degrees or higher.  5.4.2 Unconditional means models  We first fit unconditional means models with no predictors at level one or two to assess the extent of within- and between-person variation. Model 1 determines if and to what extent there is variation that covariates included in later modelling can explain. Shown in Table 5.2, unconditional means models generated inter-class correlations (ICC) between .38 and .62. Hence, depending on the belief statement, from 38% to 62% of the total score variation is attributable to differences among respondents, while the remainder is attributable to differences in intra-individual change over time. Given that all ICCs are not close to either 0 or 1, Model 1 indicates that there is significant inter and intra-individual variation among educational beliefs that may be explained by both time and covariates. Table 5.2. Variance components and inter-class correlations Extrinsic belief statements Variance τ00 ICC ‘I can’t get ahead these days without postsecondary education’  2.054 .38 ‘To attain the lifestyle I want, I must have a university degree’ 4.843 .60 ‘I need a university degree to earn a decent income’ 3.132 .49 Intrinsic belief statements   ‘Postsecondary education is not for me’ 5.363 .62 ‘My education has improved my communication skills’ 3.033 .48 ‘My education has improved my reasoning skills’ 3.461 .51 With an ordinal outcome variable, the residual variance is assumed to have a standard logistic distribution (mean of 0 and a variance of 2 / 3=3.29) (O’Connell, 2010). Thus, the calculations are based on ICC=τ00/ τ00+3.29; N=510 5.4.3 Unconditional growth models  Model 2 examines variation in growth trajectory parameters through the addition of linear and quadratic slopes of time. Shown in Table 5.3, the cumulative logits for measures of 118  linear time are statistically significant and negative for all three extrinsic belief statements. Additionally, the corresponding quadratic terms are statistically significant and positive. These measures of time indicate that not only were participants, on average, less likely to believe that education was necessary for achieving career and lifestyle goals over time, but the rate of change decelerated. That is, change in extrinsic educational beliefs was most notable in early in adulthood.  Table 5.3. Results of unconditional growth models  Get ahead  Lifestyle Income  Comm. Reasoning Not for me Direction of change (time)  -.386*** (.107) -.441*** (.112) -.398*** (.109) -.402*** (.110) -.487*** (.113) .385** (.128) Rate of change (time2)  .053* (.027) .073*** (.021) .063** (.020) .072*** (.021) .091*** (.023) -.036 (.024) Thresholds       δ1 -5.408*** (.228) -5.302*** (.264) -4.654*** (.234) -7.110*** (.372) -7.180*** (.299) 1.304*** (.202) δ2 -3.104*** (.167) -1.859*** (.195) -1.256***  (.173) -4.683*** (.233) -4.658*** (.218) 4.331*** (.262)  δ3 -2.329*** (.152) -.763** (.192) -.188 (.171) -3.221*** (.193) -2.953*** (.181) 6.478*** (.321) δ4 .411** (.139) 1.472*** (.189) 2.363*** (.182) .192*** (.151) 1.330*** (.161) 8.495*** (.492) Variance components       Initial status 3.747  (.597) 9.492 (1.080) 7.396 (.878) 4.875 (.711) 5.107 (.748) 8.527 (1.346) Time .161  (.043) .231 (.045) .267 (.049) .146 (.039) .155 (.048) .164 (.049) * p < .05; ** p < .01; ***p < .001; Robust standard errors in parentheses; N=510. The findings for intrinsic educational beliefs are mixed. Participants were more likely to believe that “postsecondary education is not for me” as time passed, with no difference in the rate of acceleration or deceleration. Change over time for statements corresponding to communication and reasoning skills was similar to extrinsic beliefs. That is, participants were slightly less likely to agree that education improved their communication and reasoning skills over time, with decelerating rate of change. As illustrated in Figure 5.1, trajectories of change were similar but the overall level of agreement differed. We generate predicted probabilities 119  from the adjusted thresholds to explore the likelihood of agreeing with each belief statement over time. Confirming our descriptive findings, participants held strong positive intrinsic values, high level of agreement that postsecondary education is necessary to get ahead, and less agreement that it was necessary for their income and lifestyle over time.   Note: The negatively worded statement “postsecondary education is not for me” captures disagree/highly disagree responses.  Figure 5.1. Predicted probability of agreement with each belief statement generated from unconditional growth models "Get ahead""Lifestyle""Income"40%50%60%70%80%90%25 30 35 42 48% agree or strongly agreeAgeExtrinsic educational beliefs"Communication skills""Reasoning Skills""Not for me"40%50%60%70%80%90%25 30 35 42 48% agree or strongly agreeAgeIntrinsic educational beliefs120  5.4.4 Conditional growth models  The main model of interest, Model 3, adds predictors at both levels one and two to assess the effect of gender, parental education, and education and employment participation. For all models, the linear time becomes non-significant. The shift in significance signals that our independent variables explain the negative change in educational beliefs. As Table 5.4 and Table 5.5 depict, while controlling for all other covariates, gender has a statistically significant effect on initial responses to four belief statements. On average, women self-report higher levels of agreement with all three extrinsic belief statements at age 25. Women are also more likely to disagree that postsecondary is not for them. The odds ratios specify the magnitude of the gender effect. Women are more than 2.69 to 4.06 times more likely to initially agree with extrinsic belief statements and approximately half as likely to agree that education is not for them compared to men. Contrastingly, there is no gender difference in initial level of agreement to the intrinsic belief statements capturing the necessity of education for communication and reasoning skills.  Gender does not influence trajectories of change over time among intrinsic educational beliefs. However, there is significant gender difference in change over time for two extrinsic belief statements capturing the necessity of education to get ahead and attain a lifestyle desired. The negative coefficients for the interaction with linear time and the positive coefficients for the interaction with quadratic time indicate that on average, women’s beliefs decreased more over time, while also decelerating at a more rapid rate. Figure 5.2 demonstrates that the gender difference in responses to both belief statements was greatest from age 25 to 35, with a trend of convergence over time. 121  Table 5.4. Extrinsic beliefs: Results of conditional growth models  Get ahead Lifestyle  Income  coef or coef or coef or Rate of change (time)  -.400 (.207)  .299 (.231)  -.081 (.225)  Quadratic rate of change (time2) -.084* (.038)  -.047 (.041)  .014 (.041)  Effects of covariates on intercept       Gender1 1.402*** (.281) 4.06 1.233*** (.348) 3.43 1.00** (.328) 2.69 Parental education2 .622* (.276) 1.86 1.212*** (.361) 3.36 .858** (.333) 2.36 Mean months of education .110*** (.015) 1.12 .208*** (.019) 1.23 .116*** (.017) 1.12 Mean months of employment .015* (.007) 1.02 .020* (.010) 1.02 .012 (.009)  Interaction with time       Gender1 -.781*** (.219) .46 -.512* (.233) .60 -.216 (.300)  Parental education2 -.626** (.222) .53 -.678** (.233) .51 -.295 (.223)  Interaction with time2        Gender1 .134** (.043) 1.14 .090* (.044) 1.09 .035 (.043)  Parental education2 .112** (.049) 1.12 .104* (.044) 1.11 .037 (.043)  Time varying measures       Deviation from mean months of education3 .012* (.006)  .019** (.006)  .010 (.006)  Deviation from mean months of employment3 .005 (.003)  .007 (.004)  .005 (.003)  Thresholds       δ1 -2.536*** (.537)  -1.014 (.659)  -2.034*** (.601)  δ2 -.210 (.511)  2.442*** (.650)  1.368* (.596)  δ3 .578 (.509)  3.553*** (.665)  2.441*** (.601)  δ4 3.350*** (.511)  5.801*** (.879)  4.989*** (.616)   Variance components       Initial status 3.178 (.528) 6.461 (.877) 6.090 (.771) Rate of change .160 (.043) .207 (.043) .255 (.048) * p < .05; ** p < .01; ***p < .001; Robust standard errors in parentheses; N=510. 1 Women=1; 2 PSE degree or higher=1; 3education and employment group mean centred. 122    Figure 5.2. Gender and parental education difference in predicted probabilities of agreement with extrinsic belief statements The effect of parental education on extrinsic and intrinsic educational beliefs follows a similar trend to gender. Shown in Table 5.4, participants who have at least one parent with a postsecondary degree or higher are more likely to initially self-report higher agreement with all extrinsic belief statements. The corresponding odds ratios highlight that respondents with highly educated parents were 1.86 to 3.36 times more likely to initially agree with the statements. Contrastingly, parental education has no statistically significant influence on responses to intrinsic belief statements concerning skills in Table 5.5. Nonetheless, parental 45%55%65%75%85%95%25 30 35 42 48% agree or strongly agreeAge‘I can’t get ahead…’WomenMenParents w/ BAParents no BA45%55%65%75%85%95%25 30 35 42 48% agree or strongly agreeAge‘to attain the lifestyle…’WomenMenParents w/ BAParents no BA123  education was influential on responses to the statement “postsecondary education is not for me”—on average individuals from highly educated households were more likely to initially disagree with this statement at age 25.  Table 5.5. Intrinsic beliefs: Results of conditional growth models  Communication  Reasoning  Not for me   coef or coef or coef or Rate of change (time)  -.375 (.203)  -.409 (.219)  -.070 (.242)  Quadratic rate of change (time2) .092* (.038)  .066 (.041)  .036 (.044)  Effects of covariates on intercept       Gender1 .281 (.295)  -.448  (.304)  -.775* (.367) .46 Parental education2 .242 (.309)  .156  (.308)  -.775*  (.392) .46 Mean months of education .148*** (.020) 1.16 .174*** (.019) 1.19 -.266*** (.024) .77 Mean months of employment -.006 (.009)  .005 (.009)  -.009 (.011)  Interaction with time       Gender1 .247 (.219)  .188  (.233)  .033  (.257)  Parental education2 .051 (.232)  -.128  (.248)  .497 (.288)  Interaction with time2        Gender1 -.063 (.043)  -.021 (.046)  -.006 (.049)  Parental education2 -.026 (.045)  .046 (.050)  -.091 (.055)  Time varying measures       Deviation from mean months of education3 .014* (.006)  .015*  (.006)  -.026*** (.008)  Deviation from mean months of employment3 .000 (.003)  .005  (.004)  -.006  (.004)  Thresholds       δ1 -5.506*** (.659)  -5.373*** (.676)  -2.677*** (.725)  δ2 -3.075*** (.599)  -2.864** (.649)  .358 (.713)  δ3 -1.608** (.593)  -1.152 (.654)  2.495* (.730)  δ4 2.111*** (.595)  3.159*** (.661)  4.488*** (.842)   Variance components       Initial status 3.877 (.602) 3.672 (.634) 5.250 (.929) Rate of change .136 (.038) .151 (.047) .150 (.048) * p < .05; ** p < .01; ***p < .001; Robust standard errors in parentheses; N=510. 1 Women=1; 2 PSE degree or higher=1; 3education and employment group mean centred. 124  Change over time by parental education also follows a similar trajectory to gender. As illustrated in Figure 5.2, respondents who have a parent with a degree or higher have a 22% higher predicted probability of agreeing with the statement “to attain the lifestyle I want, I must have a university degree” five years after high school graduation. Notably, the divide closes to 12% higher predicted probability of agreement 28 years later. Respondents from highly educated backgrounds are also more likely to agree that postsecondary education is necessary to get ahead, although the difference again narrows over time. Like gender, change in extrinsic beliefs by parental education takes place in early adulthood, with deceleration over time. Contrastingly, differences by parental education in response to the statements “postsecondary education is not for me” and “I need a university degree to earn a decent income” remain constant over time. Interpreting the influence of employment and postsecondary participation on educational beliefs differs from demographic indicators given that experience can be understood as both time varying and as an overall orientation. The time-varying measures in Table 5.4 and Table 5.5 indicate the effect of individual deviation from their average education or employment participation; that is, how change in participation over time influenced individual responses. For all beliefs (other than “I need a university degree to earn a decent income”), each standard deviation increase in months of postsecondary participation above a participant’s overall average level generated more favorable educational beliefs. Likewise, each standard deviation decrease in months below average generated less favorable educational beliefs. The education coefficients at level one suggest the timing of postsecondary participation is influential on responses. Indeed, educational beliefs decreased 125  in early adulthood at an accelerated rate in the unconditional growth models, a trend that follows the common pattern of postsecondary participation. Contrastingly, change in employment participation did not have any effect on reported educational beliefs.  The coefficients for average months of education and employment estimates the change in cumulative logits, while holding all other variables constant, for initial agreement with each belief statement. Measures of average participation also offer insight into total level of participation over the 28-year period. To ease interpretation, we generated the predicted probabilities for low (bottom third) and high (top third) months of postsecondary participation for meaningful group comparisons. Figure 5.3 shows that respondents with the greatest level of postsecondary participation have the highest extrinsic and intrinsic educational beliefs, with the same decelerating negative change for extrinsic beliefs. Nevertheless, intrinsic educational beliefs remain more constant over time for both groups. For individuals with the lowest one-third of participation, agreement that postsecondary education is necessary for a desired income or a lifestyle is particularly low. Finally, average months of employment has a small effect on two extrinsic belief statements. Even so, comparing the coefficients between education and employment suggests the effect of labour market participation is less influential. 126    Note: The negatively worded statement “postsecondary education is not for me” captures disagree/highly disagree responses.   Figure 5.3. Predicted probabilities of agreement with belief statements by level of postsecondary participation 5.5 Discussion Our research assesses how demographic factors and life course experiences influence educational beliefs. We ask: 1) how do parental education and gender influence the level and variability over time of extrinsic and intrinsic educational beliefs? And 2) how do 20%30%40%50%60%70%80%90%100%25 30 35 42 48% agree or strongly agreeAgeRespondents with high postsecondary participationGet ahead Life style IncomeCommunication Reasoning Not for me20%30%40%50%60%70%80%90%100%25 30 35 42 48% agree or strongly agreeAgeRespondents with low postsecondary participationGet ahead Life style IncomeCommunication Reasoning Not for me127  postsecondary and labour market participation influence the level and variability over time of extrinsic and intrinsic educational beliefs? In response to question one, two important research findings warrant further discussion. First, unlike in research on work beliefs, we demonstrate that women had higher levels of extrinsic educational beliefs. Second, the influence of parental education on beliefs is dependent upon the type of statement queried. Regarding our second question, two research findings are necessary to discuss in greater depth. First, participation in education serves to self-reinforce beliefs. Given prior research, the importance of education in shaping beliefs is not surprising. Nonetheless, it has important implications for understanding the maintenance of educational inequality over the life course. Second, average labour market participation has a small statistically significant effect on two extrinsic educational beliefs, yet there are important cofounding effects to consider.  As discussed in the literature review, women self-report higher intrinsic and lower extrinsic beliefs with respect to orientations towards work (Hjort, 2015; Kalleberg & Marsden, 2013; Krahn & Galambo, 2014; Lindsay & Knox, 1984; Marini et al., 1996). Our research does not support the same gendered trend in regard to educational beliefs. Rather, women self-report higher agreement with all three extrinsic belief statements and are more likely to disagree that postsecondary is not for them. A potential explanation for our findings surrounds the relationship between pro-school attitudes and gender. Prior research demonstrates women generally have more positive attitudes towards education compared to men (Jones & Myhill, 2004; Legewie & DiPrete, 2012). Our research reveals that, among adults, the most notable gender difference surrounds extrinsic belief statements. To contextualize our finding, it is necessary to consider gender differences in educational 128  outcomes. As in other contexts, women in Canada with only a high school diploma have median cumulative earnings that are approximately half the size of their male counterparts over a 20-year period (Ostrovsky & Frenette, 2014). Although a gender earning gap exists at all levels of education, it begins to narrow with higher credentials. Thus, one explanation is that gender differences in educational beliefs reflect social inequality in educational outcomes. Individuals support education to a different extent based on its perceived necessity and women are required to participate in postsecondary education to secure advantageous outcomes. The influence of parental education differs by the belief statement asked. Respondents with one or more university-educated parent are more likely to self-report higher initial extrinsic educational beliefs in early adulthood, with no difference for intrinsic beliefs. For certain statements, the effect of parental education is quite large, up to a 21% difference in the likelihood of agreeing with the statement “to attain the lifestyle I want, I must have a university degree.” However, the gap in extrinsic belief responses among respondents with and without highly educated parents narrows over time. This change confirms prior findings that suggest the effect of parental education diminishes over the life course as later confounding experiences and events become more influential (Black, Devereux, & Salvanes, 2003; Lucus, 2001). Given that parental education is influential on the likelihood of attaining postsecondary education (Andres, Adamuti-Trache, Soon, Pidgeon, & Thomsen, 2007), social background factors are difficult to disentangle from the later experiences and opportunities they afford. In early adulthood, the perception of education is, on average, more positive for all respondents and even more positive for women 129  and respondents with one or more highly educated parents. Nevertheless, our findings suggest that demographic indicators have a greater influence in early adulthood and life course experiences influence later change in educational beliefs as demographic differences narrow over time. Our research confirms that postsecondary participation functions as a self-reinforcing mechanism shaping both intrinsic and extrinsic beliefs. In contextualizing our findings, it is important to highlight that Canada has high rates of postsecondary participation (OECD, 2016a). Individuals in Canada with more education are also the most likely to engage in education and training in adulthood. Our research exposes a potential source of influence supporting the ratcheting effect between education and beliefs. With greater belief in the intrinsic and extrinsic benefits of education, education is more readily seen as a means to improve skills, raise income, and get ahead. Nonetheless, it is important to remain open to two explanations. First, as discussed above in relation to gender differences, some occupational groups may not require high levels of postsecondary participation to gain necessary skills, earn an acceptable income, and attain a lifestyle deemed adequate. For example, postsecondary education may not be necessary in Canadian provinces with large energy and resource industry sectors, where wage growth for individuals—mainly men—without university-level education is possible (Fortin & Lemieux, 2015). When compared with prior research, our research also suggests a second explanation. Given that people who have less postsecondary participation are more likely to self-report lower intrinsic and extrinsic educational beliefs, one’s orientation towards education may produce divergent levels of participation that propagate inequality (Bourdieu & Passeron, 1970/1990; Mehan, 130  1992). Much as the literature review highlighted prior research examining the importance of individual alignment towards education, our research contributes evidence that the formation and maintenance of belief systems continue through adulthood. The final notable finding surrounds how labour market participation throughout the life course influences intrinsic and extrinsic educational beliefs. Time varying increases or decreases in employment have no effect on educational beliefs. Nevertheless, overall employment has a small statistically significant effect on two extrinsic belief statements but no influence on intrinsic beliefs. Our study supports prior qualitative research that finds orientations towards education are formed and maintained regardless of labour market experience and expectations (Read & Oselin, 2008). Nonetheless, there are two avenues for further research that surround the limitations of how we operationalized labour market participation. First, our analysis does not capture how short periods of non-voluntary unemployment, economic downtown, or income instability may have short-term influences on educational beliefs. Our research takes a comprehensive view over adulthood and thus is not sensitive to short-term change. Second, lower levels of labour market participation may be due to a greater number of months in education and gendered labour market experiences. Indeed, in our research, 85% of respondents in the bottom one-third of labour market participation are women. Thus, more research disentangling the effect of education, gender, and labour market participation on educational beliefs will be necessary before concluding that there is no significant effect.  131  5.6 Conclusion There are three main theoretical and empirical limitations necessary to highlight in our study. First, due to data restrictions, we do not consider the influence of cultural background and ethnicity on educational beliefs. Prior research suggests that educational beliefs differ across minority and cultural groups (Francis & Archer, 2005; Mirza, 2009; Reay, 1998). Future research should consider how other demographic aspects influence extrinsic and intrinsic beliefs towards education. The second limitation surrounds the implicit suggestion that higher levels of agreement with extrinsic and intrinsic beliefs is preferred, a bias that “can deny the importance of freedom and choices, including the freedom to make what may be considered sub-optimal choices” (Watts, 2013, p. 504). Importantly, we do not assess the degree to which lower extrinsic and intrinsic beliefs result in adverse life course outcomes. Finally, our findings are not causal. We cannot establish educational participation or demographic characteristics as the “source” of change in extrinsic and intrinsic beliefs. In summary, it is necessary to highlight that postsecondary education is largely a private investment made by individuals and families, though it is also promoted by governments to improve skills, employability, and well-being. Measures to increase accessibility are widespread in Canada and abroad, including low tuition institutions and programs, low barrier admission policies, and government provided student loans. Yet, as our chapter explored, such policy may not be able to counter belief systems established early in adulthood and maintained over time. We find that greater postsecondary participation strengthens beliefs in the necessity of education for both intrinsic and extrinsic outcomes. Group differences in level of agreement forms early in adulthood. Women and individuals 132  with highly educated parents have higher intrinsic and extrinsic beliefs over adulthood—a trend dependent on the type of belief statement surveyed. The aim of our chapter has been to bridge scholarship between the sociology of education and work on extrinsic and intrinsic beliefs, extend prior research to consider belief change over adulthood, and offer avenues for further research that considers how educational beliefs are both reflective and generative of experiences in the courses of peoples’ lives. 133  Chapter 6: Conclusion  6.1 Contributions and Key Findings My dissertation has demonstrated that postsecondary participation has the power to reflect or even propagate social division and inequality. My purpose was not to dispel or contradict the notion that education is a common good with widespread benefits for individuals and social progress. Rather, I have illustrated that educational credentials are relative to the individuals who hold them and the contexts where they are earned. This is why it is integral for governmental and institutional policy to consider why the same or similar education leads to disparate life chances. I have shown that credentials are differentially rewarded in particular contexts and circumstances. Belief in an absolute or intrinsic value of education may serve to inform ruminations on policy, yet these must also involve considerations of the unequal distribution of educational outcomes. My research has revealed that the worth of a credential is not fixed. Rather, value functions in response to the relationship between attainment and opportunities. It is necessary for practitioners, policy makers, and key figures shaping educational discourse to account for unequal outcomes that cannot be fully accounted for or mitigated by equalizing opportunities.  As the first two chapters of my dissertation discussed, most scholarship on educational outcomes focuses on occupational status and earnings—two central aspects of inequality. A primary contribution of my research has been to expand stratification research to study the relationship between education and less considered non-monetary outcomes. I 134  showed in Chapter 3 that rates of non-standard employment vary by field of study and among men and women. Chapter 3 also established that the mediating effect of academic disciplines and the extent of gender inequality are dependent upon the type of non-standard employment. Whereas field of study fully explained gender differences in rates of temporary employment, women were still more likely to hold part-time positions even after accounting for their academic disciplines and other covariates. Chapter 3 also revealed that field of study is one of the most important contributing effects to the likelihood of non-standard employment in the early careers of graduates in the 1990s and 2000s. Given that non-standard employment has implications for earnings and other employment characteristics, social policy that aims to promote work equity in Canada must address discrepancies in non-standard employment among industries, fields of study, and by gender. I demonstrate that recent graduates face different employment prospects due to both their demographic and educational characteristics. Thus, rather than promoting the inherent worth of a bachelor’s degree, social policy must account for the unequal ways academic disciplines are rewarded.  Scholarship on educational outcomes often implicitly suggests that the benefits of education are universal. In Chapter 4, Janine Jongbloed and I explored how the relationship among education, skill, occupation, and workplace task discretion was dependent upon varying contexts. We showed that across most of the 30 countries studied, education influenced the availability of task direction within employment through occupational sorting. That is, higher credentials provided access to occupations where task discretion was more available. The skills gained through education did have a small influence in some contexts. Nonetheless, there was little indication that workplace task discretion was an attribute of 135  employment only highly educated and skilled individuals were capable of performing. We further emphasized this point by demonstrating that both the overall level of task distraction within a country and how widely it ranges among occupations influenced the direct and indirect relationships between education and task discretion. Education was less influential in contexts where task discretion was more readily available among occupational sectors. Our findings have important policy, research, and theoretical implications. Often researchers and social policy focus primarily on how the distribution of education in a society leads to unequal outcomes (e.g., Fujihara & Ishida, 2016). We showed that future research and policy must also take into account the distribution of outcomes. The context-specific availability of social goods, high quality employment, and other necessities for well-being have the potential to generate greater or lesser educational inequality. Importantly, the stratifying tendency of education diminishes when well-being equalizes.  The relative value of education is not only based on external outcomes but also an individual’s position and worldview. In Chapter 5, Dr. Lesley Andres and I explored how intrinsic and extrinsic educational beliefs changed over adulthood in relation to social position and experience. Women and individuals from highly educated backgrounds were more likely to self-report agreement that postsecondary education was necessary to have a desired lifestyle and earn a decent income. Yet the same demographic markers had an insignificant effect on beliefs surrounding the agreement that education improves communication and reasoning skills, a commonly held principle among the group studied. The most important finding, however, was the self-referential nature of educational participation and beliefs. Among all intrinsic and extrinsic belief statements studied, 136  postsecondary participation increased the likelihood of more a positive view of education. Our findings complement prior research that suggests some adult learners are at risk of disengaging with formal learning due to their lack of subjective alignment with education (Crossan, Field, Gallacher, & Merrill, 2003). Policy aiming to increase postsecondary participation but neglecting how orientations to education differ will result in a “Matthew effect”—namely, the tendency for “advantages to accumulate through time” (Rigney, 2010, p. 4). Policy cannot treat positive intrinsic and extrinsic educational beliefs as universal. Rather, it must account for how educational beliefs are relative to an individual’s social position and life course experiences.  6.2 Data and Analytical Limitations The availability and accessibility of the data influenced all three research chapters. Chapter 3 was based on Statistics Canada microdata and all findings underwent vetting prior to release. Multivariate models posed little confidentiality risk. However, frequency data was subject to more strict vetting rules. In Chapter 3, some fields of study were highly male or female-dominated and limited my ability to release rates of part-time and temporary employment among men and women separately. Confidentiality of individuals who participate in surveys was paramount and necessarily limited my analysis. It was essential for me to balance the demands of confidentiality with capturing nuances in fields of study.  In Chapter 4, Janine Jongbloed and I used publicly released PIAAC data that maintained confidentiality through the type and formation of variables released. There were two major difficulties we faced when using PIAAC survey data. First, coarsened data 137  removed variance that may have been pertinent for our analysis (Gordon, 2015). For example, our education grouping, which clustered together university degrees at the bachelor’s, master’s, and doctoral level, may have missed important intra-group differences. Second, country-specific confidentiality standards limited the public use international data. As an illustration, some countries gave the specific age of a respondent while others released categorical age groupings. Moreover, no public use data were released for Australia, a country that participated in the first round of PIAAC. Not only was Australia excluded from our research, it is also missing from the majority of PIAAC research conducted outside the OECD. In Chapter 5, Dr. Lesley Andres and I used microdata available through the Paths on Life’s Way longitudinal study. I signed a confidentiality agreement and engaged in ongoing discussions with Dr. Andres on how best to use the Paths data and respect the confidentiality of participants. We faced two major considerations. First, it was essential to balance the richness captured by longitudinal data with the disparate nature of life course events and pathways (Andres, 2012). No life course trajectory was the same among participants: individuals entered and left schooling and the labour market at unique times, or married and had children years apart from one another. Chapter 5 aimed to balance variation and similarity among the cohort. Second, while NGS data only captured data on outcomes two years after graduation, the Paths on Life’s Way data collection began after high school graduation and thus did not capture educational beliefs prior to adulthood. All research chapters in my dissertation produced empirical evidence on educational stratification. My purpose aligned with quantitative approaches that aim to contribute 138  explanatory and descriptive insight into the patterns and structures of inequality (Hout & DiPrete, 2006; Scott & Siltanen, 2016). It is important to highlight that my analytical framework did not intended to produce universal “scientific” explanations. Quantitative research may be perceived as constructing “value-neutral” and “objective” (Harding, 1987, p. 182) knowledge through the interrelationship between measurement and conceptualization. Indeed, my research involved operationalizing concepts in ways that generated forms of representativeness (Chang & Cartwright, 2008). The principles of social division on which quantitative research explicitly and implicitly relies has been substantively critiqued. Such principles include 1) naturalisation, which casts identity markers as unchangeable; 2) collective attributions, which assumes homogeneity of groups; and 3) relationality, which frames social difference as oppositional (Anthias, 1998, 2001; Penn, 2016). Such critiques are necessary reminders of the limitations and partiality of my analytical and research framework. 6.3 Theoretical Consequences and Possibilities  My research invites consideration of the purpose of education, especially surrounding its social efficacy. Focus on educational outcomes within my own research and the research of others shifts attention towards the ways learning and experiences within a classroom shape later outcomes. Disparate outcomes are often met with critique and reform within educational institutions, such as attempts to formalize and improve accountability in schools (Willis & Kissane, 1997) and generate more “effective” curriculum (Biemans, Biemans, Nieuwenhuis, Poell, Mulder, & Wesselink, 2004). Such efforts have the power to reshape learning by increasing emphasis on “outcome” (Andrish, 2002), “competence” (Biemans et al., 2004), or 139  “capability” (Wheelahan & Moodie, 2011) based education. Outcome-based education involves clarifying learning expectations and the formation of performance based curriculum (Morcke, Dornan, & Eika, 2013). Research that focuses on employment and other outcomes has the power to influence curriculum objectives, even when sources of inequality or difference are influenced by social factors that cannot be simply addressed through the “right” educational reform. Brown (2001) argues that increased emphasis on outcome-based learning connects to the influence of neoliberal welfare reform on education. Brown frames individual-state relations as increasingly based upon a skills nexus. That is, the state offers access to training while an individual takes up the responsibility of forming and marketing his or her skills. As Fisher and Rubenson (2014) write, education is projected “at times as a mechanism for social inclusion and equality, at others as an instrument for labour force development” (p. 335). Future research must consider that directing attention towards individual skill and education risks circumventing labour market and social regulation that would promote socioeconomic protection for all individuals, regardless of level or type of education. My dissertation has showed that credential levels, fields of study, and time spent in schooling generates difference and even inequality. Such inequality, however, is connected to contextual and demographic attributes often separate from education. There is risk in promulgating an implicit suggestion that more of the right “type” of education would alleviate unequal outcomes.  Future research, policy, and practice must consider the intricate connection between education and outcomes, a relationship influenced by a range of other factors. Education is 140  often constructed as a means to provide skills, meet employer needs, and promote better socioeconomic outcomes for individuals and groups. Nonetheless, it is inadequate to frame education as the only means through which to address social inequality. Constructing education as the main mechanism to counter social ills risks overlooking the power of other social provisions, forms of social policy, and redistributive strategies. Further, overemphasizing its effects can set educational reform on a path to failure, as it cannot ultimately overcome other political and economic realities. My dissertation illustrated in multiple ways that education does not universally satisfy social, employer, and individual needs. Nevertheless, education has the power to engender new and exacerbate old forms of inequality, from generating new distinctions via vertical and horizontal segregation, to propagating gender and class distinctions. There is clearly much at stake in crafting educational policy in conversation with an understanding of other contemporary social forces.  6.4 Future Research My dissertation begins to reckon with the relationship between education and skill both in Canada and internationally. As Chapter 4 explored, a human capital framework implicitly merges the distinction between education and skill, as schooling represents an investment or asset with incremental returns. The Mincer (1974) estimation influences this assumption, assuming years of education signal a gradient form of human capital. In this sense, investment in education is a proxy to estimate individual skill. The Mincer framework, however, is susceptible to ability bias. For example, individual-level unobserved ability casts years of schooling as an endogenous measure (Flores-Lagunes & Light, 2010; Harmon, 141  Oosterbeek, & Walker, 2003). As noted in Chapter 4, a common solution to address this concern is the inclusion of skill, ability, or competency measures to “partition” the effect of skill from education. Skill and schooling measures are often highly correlated, leading to collinearity that risks biased coefficient estimates, lowers statistical power, inflates standard error estimates, and increases the chance of type II error (Cawley, Heckman, & Vytlacil, 1998). In consideration of such methodological and theoretical concerns, Janine Jongbloed and I assessed the relationship between education and skill through KHB decomposition analysis in Chapter 4. Theoretically, a KHB approach shows the extent to which education credentials themselves or the literacy skills gained through education influence self-reported workplace task discretion. Chapter 4 offered insight into the relationship between skill and education and also raised important areas for further study. It is necessary to assess the relationship between competency assessment scores and more detailed education measures that were not possible to capture in Chapter 4 (e.g., other credential levels and fields of study). Chapter 3 demonstrated that more detailed field of study measures allow for an examination of how the composition of graduates varies among credentials. Average skill level rises with higher credentials, but it also varies substantially within each education level (Statistics Canada, 2013). Rather than simply casting education as the production of a specific form of skill, a more detailed examination of credential types and their corresponding composition of graduates would allow for insight into how skill levels vary and are rewarded in diverse ways. For example, skill may be differentially compensated according to demographic, education, geographical, employment, and other life course 142  attributes and experiences. The relationship between education, skill, and an outcome is not necessarily unidirectional. Skill interlinks with workplace training and other experience and daily activities that differ among social groups (Rose, 2004). It is necessary for future qualitative and quantitative research to consider how other life events impact skill-based practices, access to workplace training, and postsecondary participation.  International comparative research is also an avenue for future research. As Chapters 2 and 4 explored, educational policy varies internationally and influences the availability of opportunities, the extent to which inequality is mitigated or propagated, and the types of skill promoted. Given that universal accounts of educational outcomes can rarely be established, it is necessary to consider how and why certain contexts enable greater equality. It is important to assess to what degree contextual differences are due to specific features of an education system or other, more widespread institutional differences, such as welfare state provisions, the availability and type of employment, and even social attitudes and beliefs. Finally, it will be necessary through qualitative and quantitative research to consider how outcomes vary in terms of cultural and social orientation. Both item response and phenomenological research will be necessary to further understand group and context-based differences in orientations towards education and the outcomes expected.       6.5 Conclusion  I have made, in this dissertation, a strong commitment to understanding social divisions and stratification as variable and relative processes: Chapter 3 examined non-standard employment outcomes across cohorts and fields of study; Chapter 4 considered 143  workplace task discretion outcomes across place and credential level; and Chapter 5 studied how educational belief outcomes vary longitudinally and by length of postsecondary participation. My approach to understanding educational inequality provides insight into change over time, contexts, and adulthood. Nevertheless, my research is not completely immune from critiques of social stratification theory—in part, due to the methodological approach of quantitative stratification research. I try, where possible, to contend with the assumptions that undergird the perspectives on which I rely.  In summary, attempts to locate the “source” of educational inequality will always offer a partial account. As my dissertation has showed, several divergent accounts explain key aspects of educational inequality, including fields of study, social context, and even belief systems and worldviews. It will be necessary, as highlighted by other scholars (e.g., Edgerton & Roberts, 2014), to further engage with micro-macro tensions surrounding how structural forms of inequality emerge as everyday social realities. Education is many things in society. It is vehicle of socialization, a site of skills training, a precursor to employment, a stratifying force, and, conversely, a manner through which inequality can be ameliorated. It holds value as a social good and produces consequences unintended and potentially destructive. It is the task of researchers to acknowledge such tensions and contradictions when studying the multifarious aspects and outcomes of education. Such an aim will continue to direct my future pursuits.   144  References Adams, T. L. (2010). Gender and feminization in health care professions. Sociology Compass, 4(7), 454–465. Adamuti-Trache, M., & Sweet, R. (2008). Vocational training choices of women: Public and private colleges. Gender and Education, 20(2), 167–182.  Adamuti-Trache, M., & Sweet, R. (2014). Science, technology, engineering and math readiness: Ethno-linguistic and gender differences in high-school course selection patterns. International Journal of Science Education, 36(4), 610–634.  Adler, M. A. (1993). Gender differences in job autonomy: The consequences of occupational segregation and authority position. Sociological Quarterly, 34(3), 449–465. Allen, J., & Van der Velden, R. (2007). Transitions from higher education to work. In U. Teichler (Ed.), Careers of university graduates: Views and experiences in comparative perspectives (pp. 55–78). Dordrecht: Springer Netherlands. Allmendinger, J. (1989). Educational systems and labor market outcomes. European Sociological Review, 5(3), 231–250. Altonji, J. G., Kahn, L. B., & Speer, J. D. (2016). Cashier or consultant? Entry labor market conditions, field of study, and career success. Journal of Labor Economics, 34(S1), S361–S401. Andres, L. (2001). Transfer from community college to university: Perspectives and experiences of British Columbia students. The Canadian Journal of Higher Education, 31(1), 35–74. Andres, L. (2002). Paths on life’s way: Transitions of British Columbia young adults in a changing society—base line study (1988) and first follow-up (1989). Vancouver: Department of Educational Studies, University of British Columbia. Andres, L. (2012). Designing and doing survey research. London: Sage. Andres, L. (2013). Summary report: No credential? Post-secondary participation and life activity patterns of the Class Of ’88. Vancouver: BC Council on Admissions and Transfer. Andres, L. (2015). Taking stock of 50 years of participation in Canadian higher education. In M. Shah, A. K. Bennett, & E. Southgate (Eds.), Widening higher education participation: A global perspective (pp. 15–33). Oxford: Elsevier. 145  Andres, L., & Adamuti-Trache, M. (2007). You’ve come a long way, baby? Persistent gender inequality in university enrolment and completion in Canada, 1979-2004. Canadian Public Policy, 33(1), 93–116. Andres, L., & Grayson, J. P. (2003). Parents, educational attainment, jobs and satisfaction: What’s the connection? A 10-year portrait of Canadian young women and men. Journal of Youth Studies, 6(2), 181–202. Andres, L., & Looker, E. D. (2001). Rurality and capital: Educational expectations and attainment of rural, urban/rural and metropolitan youth. Canadian Journal of Higher Education, 31(2), 1–46. Andres, L., Adamuti-Trache, M., Soon, E. S., Pidgeon, M., & Thomsen, J. (2007). Educational expectations, parental social class, gender, and post-secondary attainment: A ten year perspective. Youth and Society, 39(2), 135–163. Andres, L., Anisef, P., Krahn, H., Looker, D., & Thiessen, V. (1999). The persistence of social structure: Cohort, class and gender effects on the occupational aspirations and expectations of Canadian youth. Journal of Youth Studies, 2(3), 261–282. Andrich, D. (2002). A framework relating outcomes based education and the taxonomy of educational objectives. Studies in Educational Evaluation, 28(1), 35–59. Anthias, F. (1998). Rethinking social divisions: Some notes towards a theoretical framework. The Sociological Review, 46(3), 505–535.  Anthias, F. (2001). The concept of social division and theorising social stratification: Looking at ethnicity and class. Sociology, 35(4), 835–854. Appelbaum, E., Bailey, T., Berg, P., & Kalleberg, A. L. (2000). Manufacturing advantage: Why high performance work systems pay off. Ithaca, NY: Cornell University Press. Archer, L., Hollingworth, S., & Halsall, A. (2007). ‘University’s not for me – I’m a Nike person’: Urban, working-class young people’s negotiations of ‘style’ identity and educational engagement. Sociology, 4(1), 219–237. Arim, R. (2015). A profile of persons with disabilities among Canadians aged 15 years or older, 2012 (Cat. No. 89-654-X). Ottawa, ON: Statistics Canada. Retrieved March 15, 2017, from www.statcan.gc.ca/pub/89-654-x/89-654-x2015001-eng.htm Armstrong, P. (2013). Puzzling skills: Feminist political economy approaches. Canadian Review of Sociology, 50(3), 256–283.  Arneson, R. J. (1989). Equality and equal opportunity for welfare. Philosophical Studies, 56(1), 77–93. 146  Arriagada, P. (2016). First Nations, Metis and Inuit Women (Cat. No. 89-503-X). Ottawa, ON: Statistics Canada. Retrieved November 17, 2016, from www.statcan.gc.ca/pub/89-503-x/2015001 /article/14313-eng.htm Arrow, K. J. (1973). Higher education as a filter. Journal of Public Economics, 2(3), 193–216. Arum, R., Gamoran, A., & Shavit, Y. (2007). More inclusion than diversion: Expansion, differentiation, and market structure in higher education. In Y. Shavit, R. Arum, & A. Gamoran (Eds.), Stratification in higher education: A comparative study (pp. 1–35). Stanford: Stanford University Press.  Ashbourne, D., & Andres, L. (2015). Athletics, music, languages, and leadership: How parents influence the extracurricular activities of their children. Canadian Journal of Education, 38(2), 1–35. Assembly of First Nations. (n.d.). Fact Sheet: First Nations post-secondary education. Ottawa, ON: Author. Retrieved January 8, 2017, from www.afn.ca/uploads/files/pse-fact-sheet.pdf  Avvisati, F., & Keslair, F. (2014). REPEST: Stata module to run estimations with weighted replicate samples and plausible values. Boston, MA: RePEc/Ideas Statistical Software Components. Retrieved January 7, 2016, from www.ideas.repec.org/c/boc/bocode/ s457918.html Azevedo, J. P., Franco, S., Rubiano, E., & Hoyos, A. (2011). HOI: Stata module to compute Human Opportunity Index. Boston, MA: RePEc/Ideas Statistical Software Components. Retrieved July 19, 2016, from www.ideas.repec.org/c/boc/bocode/s457191.html Baker, D. P. (2009). The educational transformation of work: Towards a new synthesis. Journal of Education and Work, 22(3), 163–191. Baker, J. (2014). No Ivies, Oxbridge, or grandes Écoles: Constructing distinctions in university choice. British Journal of Sociology of Education, 35(6), 914–932.  Baker, T. L., & Velez, W. (1996). Access to and opportunity in postsecondary education in the United States: A review. Sociology of Education, 69(extra issue), 82–101.  Ball, S. J., Davies, J., David, M., & Reay, D. (2002). ‘Classification’ and ‘judgement’: Social class and the ‘cognitive structures’ of choice of higher education. British Journal of Sociology of Education, 23(1), 52–72. Barrett, G. F., & Doiron, D. J. (2001). Working part time: By choice or by constraint. Canadian Journal of Economics, 34(4), 1042–1065. 147  Beaujot, R. (2002). Earning and caring: Demographic change and policy implications. Canadian Studies in Population, 29(2), 195–225. Becker, B. E., & Huselid, M. A. (1998). High performance work systems and firm performance: A synthesis of research and managerial implications. Research in Personnel and Human Resources Journal, 16, 53–101. Beller, E., & Hout, M. (2006). Welfare states and social mobility: How educational and social policy may affect cross-national differences in the association between occupational origins and destinations. Research in Social Stratification and Mobility, 24(4), 353–365. Bernard, A. (2015). Youth labour force participation: 2008 to 2014. (Cat. No. 11-626-X). Ottawa, ON: Statistics Canada. Retrieved September 18, 2017, from http://www.statcan.gc.ca/pub/11-626-x/11-626-x2015052-eng.pdf Biemans, H., Nieuwenhuis, L., Poell, R., Mulder, M., & Wesselink, R. (2004). Competence-based VET in the Netherlands: Background and pitfalls. Journal of Vocational Education and Training, 56(4), 523–538. Bills, D. B. (2016, March). Congested credentials: The material and positional economies of schooling. Research in Social Stratification and Mobility, 43, 65–70. Black, S. E., Devereux, P. J., & Salvanes, K. G. (2003). Why the apple doesn’t fall far: Understanding intergenerational transmission of human capital. American Economic Review, 95(1), 437–449.  Bobbitt-Zeher, S. (2007). The gender income gap and the role of education. Sociology of Education, 80(1), 1–22. Boccuzzo, G., & Gianecchini, M. (2015). Measuring young graduates’ job quality through a composite indicator. Social Indicators Research, 122(2), 453–478. Boltanski, L., & Chiapello, E. (2005). The new spirit of capitalism. New York, NY: Verso. Bolton, S., & Muzio, D. (2008). The paradoxical processes of feminization in the professions: The case of established, aspiring and semi-professions. Work, Employment and Society, 22(2), 281–299. Booth, A. L., & van Ours, J. C. (2013). Part-time jobs: What women want?. Journal of Population Economics, 26(1), 263–283. Borhek, J., & Curtis, R. (1975). A sociology of belief. New York, NY: John Wiley & Sons. 148  Bottero, W. (2005). Social stratification: Social division and inequality. New York, NY: Routledge. Boudarbat, B., & Chernoff, V. (2010). The determinants of education-job match among Canadian university graduates (IZA Discussion Papers, No. 4513). Bond, Germany: The Institute for the Study of Labor. Retrieved October 15, 2016, from www.nbn-resolving.de/urn:nbn:de:101:1-20091105842 Boudarbat, B., & Connolly, M. (2013). The gender wage gap among recent post-secondary graduates in Canada: A distributional approach. The Canadian Journal of Economics, 46(3), 1037–1065. Boudarbat, B., Lemieux, T., & Riddell, W. C. (2010). The evolution of the returns to human capital in Canada, 1980-2005. Canadian Public Policy, 36(1), 63–89. Boudon, R. (1974). Education, opportunity, and social inequality: Changing prospects in Western society. New York, NY: John Wiley & Sons.  Bourdieu, P. (1977). The economics of linguistic exchanges. Social Science Information, 1(6), 645–668. Bourdieu, P. (1990). The logic of practice. Sandford, CA: Sandford University Press. (Original work published 1980)  Bourdieu, P., & Passeron, J. C. (1979). The inheritors: French students and their relation to culture. Chicago: University of Chicago Press. (Original work published 1964) Bourdieu, P., & Passeron, J. C. (1990). Reproduction in education, society and culture. London: Sage. (Original work published 1970)  Bourdieu, P. (1998). The forms of capital. In A. H. Halsey, H. Lauder, P. Brown, & A. S. Wells (Eds.), Education: Culture, economy and society (pp. 46–58). Oxford: Oxford University Press. Bowles, S., & Gintis, H. (2011). Schooling in capitalist America: Educational reform and the contradictions of economic life. New York, NY: Heymarket. (Original work published 1976) Boxall, P., & Winterton, J. (2015). Which conditions foster high-involvement work processes? A synthesis of the literature and agenda for research. Economic and Industrial Democracy, early release. doi: 10.1177/0143831X15599584. Braverman, H. (1998). Labor and monopoly capital: The degradation of work in the twentieth century. New York, NY: New York University Press. (Original work published 1974) 149  Breen, R. (2010). Educational expansion and social mobility in the 20th century. Social Forces, 89(2), 365–388. Breen, R., & Goldthorpe, J. H. (1997). Explaining educational differentials: Towards a formal rational action theory. Rationality and Society, 9(3), 275–305. Breen, R., & Jonsson, J. O. (2005). Inequality of opportunity in comparative perspective: Recent research on educational attainment and social mobility. Annual Review of Sociology, 31, 223–243. Brown, C., & Corcoran, M. (1997). Sex-based differences in school content and the male-female wage gap. Journal of Labor Economics, 15(3), 431–465. Brown, D. K. (2001). The social sources of educational credentialism: Status cultures, labor markets, and organizations. Sociology of Education, 74(extra issue), 19–34. Brown, P. (2001). Skill formation in the twenty-first century. In P. Brown, A. Green, & H. Lauder (Eds.), High skills: Globalization, competitiveness, and skill formation (pp. 1–55). Oxford: Oxford University Press. Brown, P. (2013). Education, opportunity and the prospects for social mobility. British Journal of Sociology of Education, 34(5-6), 678–700. Bryson, C. (2004). The consequences for women in the academic profession of the widespread use of fixed term contracts. Gender, Work & Organization, 11(2), 187–206. Burawoy, M. (1979). Manufacturing consent: Changes in the labor process under monopoly capitalism. Chicago: University of Chicago Press.  Busemeyer, M. R. (2015). Skills and inequality: Partisan politics and the political economy of education reforms in Western welfare states. Cambridge: Cambridge University Press. Bynner, J., & Parsons, S. (2002). Social exclusion and the transition from school to work: The case of young people not in education, employment, or training (NEET). Journal of Vocational Behavior, 60(2), 289–309. Cameron, D. (1992). Higher education in federal systems: Canada. In D. Brown, P. Cazalis, & G. Jasmin (Eds.), Higher education in federal systems (pp. 45–68). Kingston: Institute of Intergovernmental Relations.  Canadian Public Health Association. (2016). Early childhood education and care. Ottawa, ON: Author. Retrieved March 28, 2017, from www.cpha.ca/uploads/policy/ecec_e.pdf 150  Caranci, B., & Marple, J. (2017). Canada’s part-time conundrum. Toronto: TD Economics Observation. Retrieved July 2, 2017, from www.td.com/document/PDF/ economics/ special/Part_Time_Conundrum.pdf Cawley, J., Heckman, J., & Vytlacil, E. (1998). Cognitive ability and the rising return to education (NBER Working Paper Series no.6388). Cambridge, MA: The National Bureau of Economic Research. Retrieved December 17, 2016, from www.nber.org/ papers/w6388.pdf Chang, H., & Cartwright, N. (2008). Measurement. In M. Curd & S. Psillos (Eds.), The Routledge companion to philosophy of science (pp. 367–375). London: Routledge. Charlesworth, S. (2010). The regulation of paid care workers’ wages and conditions in the non-profit sector: A Toronto case study. Industrial Relations, 65(3), 380–399. Chaykowski, R. P., & Powell, L. M. (1999). Women and the labour market: Recent trends and policy issues. Canadian Public Policy, 25(supplement), S1–S25. Chesters, J., & Watson, L. (2014). Returns to education for those returning to education: Evidence from Australia. Studies in Higher Education, 39(9), 1634–1648. Coleman, J. (1968). The concept of equality of educational opportunity. Harvard Educational Review, 38(1), 7–22. Collins, R. (1979). The credential society: An historical sociology of education and stratification. New York, NY: Academic Press.  Corak, M., & Piraino, P. (2011). The intergenerational transmission of employers. Journal of Labor Economics, 29(1), 37–68. Correll, S. J. (2001). Gender and the career choice process: The role of biased self-assessments. American Journal of Sociology, 106(6), 1691–1730. Council of Ministers of Education, Canada. (n.d.). About. Toronto: Author. Retrieved March 6, 2017, from www.cmec.ca/11/About/index.html Cowin, R. J. (2017). Public policy and the structural development of postsecondary education British Columbia, Canada, 2016 – 2015 (Unpublished doctoral dissertation). University of British Columbia, Vancouver, Canada.  Cranford, C. J., Vosko, L. F., & Zukewich, N. (2003). The gender of precarious employment in Canada. Industrial Relations, 58(3), 454–482. Cromby, J. (2012). Beyond belief. Journal of Health Psychology, 17(7), 943–957.  151  Crompton, R. (2008). Class and stratification. Cambridge: Polity Press.  Crompton, R., & Harris, F. (1998). Explaining women’s employment patterns: ‘Orientations to work’ revisited. British Journal of Sociology, 49(1), 118–136. Cromwell, A. (2016, Dec 13). New Brunswick government defends tuition access bursary. Global News. Accessed March 5, 2017, from www.globalnews.ca/news/3124958 /new-brunswick-government-defends-tuition-access-bursary/ Crossan, B., Field, J., Gallacher, J., & Merrill, B. (2003). Understanding participation in learning for non-traditional adult learners: Learning careers and the construction of learning identities. British Journal of Sociology of Education, 24(1), 55–67. Dæhlen, M. (2005). Change in job values during education. Journal of Education and Work, 18(4), 385–400. David, M. E., Ball, S. J., Davies, J., & Reay, D. (2003). Gender issues in parental involvement in student choices of higher education. Gender and Education, 15(1), 21–36. Davies, S., & Aurini, J. (2013). Summer learning inequality in Ontario. Canadian Public Policy, 39(2), 287–307.  Davies, S., & Hammack, F. M. (2005). The channeling of student competition in higher education: Comparing Canada and the US. The Journal of Higher Education, 76(1), 89–106. Davies, S., & Zarifa, D. (2012). The stratification of universities: Structural inequality in Canada and the United States. Research in Social Stratification and Mobility, 30(2), 143–158.  Davies, S., Maldonado, V., & Zarifa, D. (2014). Effectively maintaining inequality in Toronto: Predicting student destinations in Ontario universities. Canadian Review of Sociology, 51(1), 22–53. de Lange, M., Gesthuizen, M. J. W., & Wolbers, M. H. J. (2014). Consequences of flexible employment at labour market entry for early career development in the Netherlands. Economic and Industrial Democracy, 35(3), 413–434. de Ruyter, A., & Warnecke, T. (2008). Gender, non-standard work and development regimes: A comparison of the USA and Indonesia. Journal of Industrial Relations, 50(5), 718–735.  152  de Vries, R. M., & Wolbers, M. H. J. (2005). Non-standard employment relations and wages among school leavers in the Netherlands. Work, Employment & Society, 19(3), 503–525. Demerath, P. (2009). Producing success: The culture of personal advancement in an American high school. Chicago: University of Chicago Press. DePratto, B., & Bartlett, R. (2015). Precarious employment in Canada: Does the evidence square with the anecdotes? Toronto: TD Economics Observation. Retrieved September 19, 2017, from www.td.com/document/PDF/economics/special/PrecariousEmployment.pdf Di Stasio, V., Bol, T., & Van de Werfhorst, H. G. (2016, March). What makes education positional? Institutions, overeducation and the competition for jobs. Research in Social Stratification and Mobility, 43, 53–63. Dickson, M., Gregg, P., & Robinson, H. (2016, October). Early, late or never? When does parental education impact child outcomes? The Economic Journal, 126, F184–F281. DiMaggio, P. (1982). Cultural capital and school success: The impact of status culture participation on the grades of US high school students. American Sociological Review, 47(2), 189–201. DiPrete, T. A., & Buchmann, C. (2013). The rise of women: The growing gender gap in education and what it means for American schools. New York, NY: Russell Sage Foundation. Dobbin, F., & Boychuk, T. (1999). National employment systems and job autonomy: Why job autonomy is high in the Nordic countries and low in the United States, Canada, and Australia. Organization Studies, 20(2), 257–291.  Doogan, K. (2009). New capitalism? The transformation of work. Cambridge: Polity. Dougherty, I., & Jones, E. (2016, Oct 18). University graduates juggle jobs of varying certainty. The Globe and Mail. Retrieved January 17, 2017, from www.theglobeandmail.com/news/national/education/canadian-university-report/university-graduates-juggle-jobs-of-varying-certainty/article32408472/ Downey, D. B. (2008). Black/White differences in school performance: The oppositional culture explanation. Annual Review of Sociology, 34, 107–126. Drewes, T., & Michael, C. (2006). How do students choose a university?: An analysis of applications to universities in Ontario, Canada. Research in Higher Education, 47(7), 781–800. 153  Drolet, M. (2002). The “who, what, when and where” of gender pay differentials (The Evolving Workplace Series, No. 71-584-MPE, No. 4). Ottawa, ON: Statistics Canada and Human Resources Development Canada. Drummond, D., & Rosenbluth, E. K. (2013). The debate on First Nations education funding: mind the gap (Policy Studies Working Paper No. 49). Kingston, Ontario: Queen’s University. Retrieved March 9, 2017, from www.qspace.library.queensu.ca/bitstream/ handle/1974/14846/Drummond_et_al_2013_Debate_on_First_Nations.pdf;jsessionid=0D3B72680B058243798714218808AEA6?sequence=1 Eagly, A. H., & Steffen, V. J. (1986). Gender stereotypes, occupational roles, and beliefs about part‐time employees. Psychology of Women Quarterly, 10(3), 252–262. Edgerton, J. D., & Roberts, L. W. (2014). Cultural capital or habitus? Bourdieu and beyond in the explanation of enduring educational inequality. Theory and Research in Education, 12(2), 193–220.  Edlund, J., & Grönlund, A. (2010). Class and work autonomy in 21 countries: A question of production regime or power resources? Acta Sociologica, 53(3), 213–228. Elder, G. H., Johnson, M. K., & Crosnoe, R. (2003). The emergence and development of life course theory. In G. H. Elder, M. K. Johnson, & R. Crosnoe (Eds.), Handbook of the life course (pp. 3–19). New York, NY: Springer. Elia, L. (2010). Temporary/permanent workers’ wage gap: A brand-new form of wage inequality? Labour, 24(2), 178–200. Elster, J. (1983). Ulysses and the sirens: Studies in rationality and irrationality. Cambridge: Cambridge University Press. England, P. (1992). Comparable worth: Theories and evidence. New York, NY: Aldine. England, P., & Li, S. (2006). Desegregation stalled: The changing gender composition of college majors, 1971-2002. Gender & Society, 20(5), 657–677.  England, P., Gornick, J., & Shafer, E. F. (2012). Women’s employment, education, and the gender gap in 17 countries. Monthly Labour Review, 135(4), 3–12. Esping-Andersen, G. (1990). The three worlds of welfare capitalism. Oxford: Polity Press. Esping-Andersen, G. (1994). Welfare states and the economy. In N. J. Smelser & R. Swedberg (Eds.), The handbook of economic sociology (pp. 711–720). Princeton: Princeton University Press. 154  Esser, I., & Olsen, K. M. (2012). Perceived job quality: Autonomy and job security within a multi-level framework. European Sociological Review, 28(4), 443–454. Estevez-Abe, M. (2006). Gendering the varieties of capitalism. World Politics, 59(1), 142–175. Estevez-Abe, M. (2012). Gendered consequences of vocational training. In M. R. Busemeyer & C. Trampusch (Eds.), The political economy of collective skill formation (pp. 259–283). London: Oxford University Press.  Estevez-Abe, M., Iversen, T., & Soskice, D. (2001). Social protection and the formation of skills: A reinterpretation of the welfare state. In P. A. Hall & D. Soskice (Eds.), Varieties of capitalism: The institutional foundations of comparative advantage (pp. 145–183). Oxford: Oxford University Press. Feir, D. L. (2016a). The intergenerational effects of residential schools on children’s educational experiences in Ontario and Canada’s Western provinces. The International Indigenous Policy Journal, 7(3), 1–44. Feir, D. L. (2016b). The long‐term effects of forcible assimilation policy: The case of Indian boarding schools. Canadian Journal of Economics, 49(2), 433–480. Ferguson, S. J. (2016). Women in Canada: A gender-based statistical report – Women and education: qualifications, skills and technology (Cat. No. 89-503-X). Ottawa, ON: Statistics Canada. Retrieved January 11, 2017, from www5.statcan.gc.ca/olc-cel/olc.action?ObjId=89-503-X201500114640&ObjType=47&lang=en&limit=0 Ferguson, S. J., & Wang, S. (2014). Graduating in Canada: Profile, labour market outcomes and student debt of the class of 2009–2010 (Catalogue no. 81-595M – no. 069). Ottawa, ON: Statistics Canada.  Fernandez‐Mateo, I. (2009). Cumulative gender disadvantage in contract employment. American Journal of Sociology, 114(4), 871–923. Fernandez-Mateo, I., & King, Z. (2011). Anticipatory sorting and gender segregation in temporary employment. Management Science, 57(6), 989–1008. Ferrao, V. (2011). Paid work (Cat. No. 89-503-X). Ottawa, ON: Statistics Canada. Retrieved July 7, 2017, from www.statcan.gc.ca/pub/89-503-x/2010001/article/11387-eng.htm Finnie, R. (2000). From school to work: The evolution of early labour market outcomes of Canadian postsecondary graduates. Canadian Public Policy, 26(2), 197–224. Finnie, R. (2008). Who goes? Who stays? What matters?: Accessing and persisting in post-secondary education in Canada. Kingston, ON: Queen’s University Press. 155  Finnie, R., & Frenette, M. (2003). Earning differences by major field of study: Evidence from three cohorts of recent Canadian graduates. Economics of Education Review, 22(2), 179–192. Fisher, D., & Rubenson, K. (2014). Conclusion. In D. Fisher, K. Rubenson, T. Shanahan, C. Trottier (Eds.), The development of postsecondary education systems in Canada (pp. 334–352). Montreal & Kingston: McGill-Queen’s University Press.  Flores-Lagunes, A., & Light, A. (2010). Interpreting degree effects in the returns to education. Journal of Human Resources, 45(2), 439–467. Foley, K., & Green, D. A. (2015). Why more education will not solve rising inequality (and may make it worse). In D. A. Green, W. C. Riddell, & F. St-Hilaire (Eds.), Income inequality: The Canadian story (pp. 349–398). Ottawa, ON: The Institute for Research on Public Policy. Fortin, N. M., & Huberman, M. (2002). Occupational gender segregation and women’s wages in Canada: An historical perspective. Canadian Public Policy, 28(supplement), S11–S39. Fortin, N. M., & Lemieux, T. (2015). Changes in wage inequality in Canada: An interprovincial perspective. Canadian Journal of Economics, 48(2), 682–713. Fortin, N., Green, D., Lemieux, T., Milligan, K., & Riddell, C. (2012). Canadian inequality: Recent developments and policy options. Canadian Public Policy, 38(2), 121–145.  Francis, B., & Archer, L. (2005). British-Chinese pupils’ and parents’ constructions of the value of education. British Educational Research Journal, 31(1), 89–108. Frenette, M. (2004). Access to college and university: Does distance to school matter? Canadian Public Policy, 30(4), 427–443.  Frenette, M. (2005). The impact of tuition fees on university access: Evidence from a large-scale price deregulation in professional programs (Cat. No. 2005263e). Ottawa, ON: Statistics Canada, Analytical Studies Branch. Frenette, M. (2015). An investment of a lifetime? The long-term labour market premiums associated with a postsecondary education (Cat. No. 11F0019M – No. 359). Ottawa, ON: Statistics Canada, Analytical Studies Branch Research Paper Series. Frenette, M., & Chan, P. C. W. (2015). Academic outcomes of public and private high school students: What lies behind the differences? (Cat. No. 11F0019M – No. 367). Ottawa, ON: Statistics Canada, Analytical Studies Branch Research Paper Series. 156  Frenette, M., & Morissette, A. (2014). Wages and full-time employment rates of young high school graduates and bachelor’s degree holders, 1997 to 2012 (Cat. No. 11F0019M – No. 360). Ottawa, ON: Statistics Canada, Analytical Studies Branch Research Paper Series.  Frenette, M., & Zeman, K. (2007). Why are most university students women? Evidence based on academic performance, study habits and parental influences (Cat. No. 11F0019MIE – No. 303). Ottawa, ON: Statistics Canada, Analytical Studies Branch Research Paper Series.  Fudge, J., & Vosko, L. F. (2001). Gender, segmentation and the standard employment relationship in Canadian labour law, legislation and policy. Economic and Industrial Democracy, 22(2), 271–310. Fujihara, S., & Ishida, H. (2016, March). The absolute and relative values of education and the inequality of educational opportunity: Trends in access to education in postwar Japan. Research in Social Stratification and Mobility, 43, 25–37 Fuller, A., & Unwin, L. (2006). Expansive and restrictive learning environments. In K. Evans, P. Hodkinson, H. Rainbird, & L. Unwin (Eds.), Improving workplace learning (pp. 27–48). London: Routledge. Fuller, S. (2005). Public sector employment and gender wage inequalities in British Columbia: Assessing the effects of a shrinking public sector. Canadian Journal of Sociology, 30(4), 405–439. Fuller, S., & Vosko, L. F. (2008). Temporary employment and social inequality in Canada: Exploring the intersections of gender, race and immigration status. Social Indicators Research, 88(1), 31–50. Gallagher, P., & Dennison, J. D. (1995). Canada’s community college systems: A study of diversity. Community College Journal of Research and Practice, 19(5), 381–393. Gallie, D. (2007). Production regimes and the quality of employment in Europe. Annual Review of Sociology, 33, 85–104. Gallie, D. (2013). Direct participation and the quality of work. Human Relations, 66(4), 453–473. Garrick, J. (1998). Informal learning in the workplace: Unmasking human resource development. London: Routledge. Gebel, M. (2010). Early career consequences of temporary employment in Germany and the UK. Work, Employment & Society, 24(4), 641–660. 157  Gerber, T. P., & Cheung, S. Y. (2008). Horizontal stratification in postsecondary education: Forms, explanations, and implications. Annual Review of Sociology, 34, 299–318. Gerstel, N., & Clawson, D. (2014). Class advantage and the gender divide: Flexibility on the job and at home. American Journal of Sociology, 120(2), 395–431. Gielen, A. C., & Schils, T. (2015). Non-standard employment patterns across occupations in the Netherlands. In W. Eichhorst & P. Marx (Eds.), Non-standard employment in post-industrial labour markets: An occupational perspective (pp. 52–88). Northampton: Edward Elgar Publishing. Giesecke, J., & Schindler, S. (2008). Field of study and flexible work: A comparison between Germany and the UK. International Journal of Comparative Sociology, 49(4-5), 283–304. Glorieux, I., Mestdag, I., Minnen, J., & Vandeweyer, J. (2009). The myth of the 24-hour society: Nonstandard work hours in Belgium, 1966-1999. Social Indicators Research, 93(1), 177–183. Goldthorpe, J. H. (1996). Class analysis and the reorientation of class theory: The case of persisting differentials in educational attainment. British Journal of Sociology, 47(3), 481–505.  Gordon, R. A. (2015). Regression analysis for the social sciences. London: Routledge. Green, F. (2004). Work intensification, discretion, and the decline in well-being at work. Eastern Economic Journal, 30(4), 615–625. Green, F. (2006). Demanding work: The paradox of job quality in the affluent economy. Princeton: Princeton University Press. Green, F. (2008). Work effort and worker well-being in the age of affluence. In R. J. Burke & C. L. Cooper (Eds.), The long work hours culture: Causes, consequences and choices (pp. 115–136). Bingley, UK: Emerald Publishing. Green, F. (2012). Employee involvement, technology and evolution in job skills: A task-based analysis. Industrial and Labor Relations Review, 65(1), 36–67. Guile, D. (2002). Skill and work experience in the European knowledge economy. Journal of Education and Work, 15(3), 251–276.  Halaby, C. N. (2003). Where job values come from: Family and schooling background, cognitive ability, and gender. American Sociological Review, 68(2), 251–278. 158  Halldén, K., Gallie, D., & Zhou, Y. (2012). The skills and autonomy of female part-time work in Britain and Sweden. Research in Social Stratification and Mobility, 30(2), 187–201.  Han, S., Tumin, D., & Qian, Z. (2016, September). Gendered transitions to adulthood by college field of study in the United States. Demographic Research, 35, 929–960. Hango, D. (2011) Length of time between high school graduation and enrolling in postsecondary education: Who delays and who does not? (Cat. No. 81-004-X) Education Matters: Insights on Education, Learning and Training in Canada, 8(4), 1–14.  Hanushek, E. A., Schwerdt, G., Wiederhold, S., & Woessmann, L. (2015, January). Returns to skills around the world: Evidence from PIAAC. European Economic Review, 73, 103–130. Happer, C., & Philo, G. (2013). The role of the media in the construction of public belief and social change. Journal of Social and Political Psychology, 1(1), 321–336.  Harding, S. G. (1987). Feminism and methodology: Social science issues. Bloomington & Indianapolis: Indiana University Press. Hargittai, E., & Shafer, S. (2006). Differences in actual and perceived online skills: The role of gender. Social Science Quarterly, 87(2), 432–448. Harmon, C., Oosterbeek, H., & Walker, I. (2003). The returns to education: Microeconomics. Journal of Economic Surveys, 17(2), 115–156. Hatcher, R. (1998). Class differentiation in education: Rational choices? British Journal of Sociology of Education, 19(1), 5–24.  Hauser, R. M., & Featherman, D. L. (1976). Equality of schooling: Trends and prospects. Sociology of Education, 49(2), 99–120. Hega, G. M., & Hokenmaier, K. G. (2002). The welfare state and education: A comparison of social and educational policy in advanced industrial societies. German Policy Studies, 2(1), 143–173. Hegna, K. (2014). Changing educational aspirations in the choice of and transition to post-compulsory schooling–A three-wave longitudinal study of Oslo youth. Journal of Youth Studies, 17(5), 592–613. Heisig, J. P., & Solga, H. (2015). Secondary education systems and the general skills of less-and intermediate-educated adults: A comparison of 18 countries. Sociology of Education, 88(3), 202–205.  159  Heisz, A., & LaRochelle-Côté, S. (2006). Work hours instability. Perspectives on Labour and Income, 19(1), 1–18. Hennekam, S., & Bennett, D. (2017). Creative industries work across multiple contexts: Common themes and challenges. Personnel Review, 46(1), 68–85. doi:10.1108/PR-08-2015-0220 Henseke, G., & Green, F. (2017). Cross-national deployment of “graduate jobs”: Analysis using a new indicator based on high skills use. In S. W. Polachek, K. Pouliakas, G. Russo, K. Tatsiramos (Eds.), Skill mismatch in labor markets (pp. 41–79). Bingley, UK: Emerald Publishing. Hirsch, F. (2012). Social limits to growth. London: Taylor & Francis. (Originally work published 1977) Hitlin, S., & Piliavin, J. A. (2004). Values: Reviving a dormant concept. Annual Review of Sociology, 30, 359–393. Hjort, J. L. (2015). Intrinsic and extrinsic work motivation among US and Norwegian high school students. Young, 23(4), 293–312. Holm, A., M. Jæger, M., Karlson, K. B., & Reimer, D. (2013). Incomplete equalization: The effect of tracking in secondary education on educational inequality. Social Science Research, 42(6), 1431–442. Holman, D., & Rafferty, A. (2017). The convergence and divergence of job discretion between occupations and institutional regimes in Europe from 1995 to 2010. Journal of Management Studies, early release. doi: 10.1111/joms.12265 Holzer, H. J., & Lerman, R. I. (2015). Cognitive skills in the U.S. labor market: For whom do they matter? Washington, DC: AIR-PIAAC. Retrieved November 18, 2016, from www.piaacgateway.com/researchpapers/ Horemans, J. (2016). Polarisation of non-standard employment in Europe: Exploring a missing piece of the inequality puzzle. Social Indicators Research, 125(1), 171–189. doi:10.1007/s11205-014-0834-0  Horemans, J., & Marx, I. (2013). Should we care about part-time work from a poverty perspective? An analysis of the EU15 countries. In M. Koch & M. Fritz (Eds.), Non-standard employment in Europe: Paradigms, prevalence and policy responses (pp. 169–189). Houndmills, UK: Palgrave Macmillan. Hou, F., & Bonikowska, A. (2016). Educational and labour market outcomes of childhood immigrants by admission class (Cat. No. 11F0019M – No. 377). Ottawa, ON: 160  Statistics Canada. Retrieved March 21, 2017, from www.statcan.gc.ca/pub/11f0019m /11f0019m2016377-eng.pdf  Hout, M. (1988). More universalism, less structural mobility: The American occupational structure in the 1980s. American Journal of Sociology, 93(6), 1358–1400. Hout, M., & DiPrete, T. A. (2006). What we have learned: RC28’s contributions to knowledge about social stratification. Research in Social Stratification and Mobility, 24(1), 1–20. Howe, K. R. (2015). The meritocratic conception of educational equality: Ideal theory run amuck. Educational Theory, 65(2), 183–201. Hu, A., & Vargas, N. (2015). Economic consequences of horizontal stratification in postsecondary education: Evidence from urban China. Higher Education, 70(3), 337–358. Hunt, G., & Song, F. (2013). Gender and specialty in business management education. The Canadian Journal of Higher Education, 43(1), 129–145. Illuminate Consulting Group. (2016). 2016 British Columbia international education intelligence report. San Carlos, CA: Author. Retrieved March 18, 2017, from www.bccie.bc.ca/wp-content/uploads/2016/07/icg-bc-intelligence-report-2016.pdf Inglehart, R., & Welzel, C. (2005). Modernization, cultural change, and democracy: The human development sequence. Cambridge: Cambridge University Press. International Labour Organization. (2015). Report for discussion at the meeting of experts on non-standard forms of employment. Geneva: Author. Retrieved January 9, 2017, from www.ilo.org/wcmsp5/groups/public/@ed_protect/@protrav/@travail/ documents/meetingdocument/wcms_336934.pdf Jackson, A. (2006). Regulating precarious labour markets: What can we learn from new European models. In L. F. Vosko (Ed.), Precarious employment: Understanding labour market insecurity in Canada (pp. 276–298). Montreal: McGill-Queen’s University Press. James, L., Guile, D., & Unwin, L. (2013). Learning and innovation in the knowledge-based economy: Beyond clusters and qualifications. Journal of Education and Work, 26(3), 243–266. Jin, J., & Rounds, J. (2012). Stability and change in work values: A meta-analysis of longitudinal studies. Journal of Vocational Behavior, 80(2), 326–339. 161  Johnson, M. K., & Elder, G. H. (2002). Educational pathways and work value trajectories. Sociological Perspectives, 45(2), 113–138. Johnson, M. K., & Mortimer, J. T. (2011). Origins and outcomes of judgments about work. Social Forces, 89(4), 1239–1260. Johnson, M. K., Sage, R. A., & Mortimer, J. T. (2012). Work values, early career difficulties, and the US economic recession. Social Psychology Quarterly, 75(3), 242–267. Jones, G. (2009). Sectors, institutional types and the challenges of shifting categories: A Canadian commentary. Higher Education Quarterly, 63(4), 371–383. Jones, S., & Myhill, D. (2004). ‘Troublesome boys’ and ‘compliant girls’: Gender identity and perceptions of achievement and underachievement. British Journal of Sociology of Education, 25(5), 547–561. Kahn, L. B. (2010). The long-term labor market consequences of graduating from college in a bad economy. Labour Economics, 17(2), 303–316. Kalleberg, A. L. (2000). Nonstandard employment relations: Part-time, temporary and contract work. Annual Review of Sociology, 26, 341–365.  Kalleberg, A. L., & Marsden, P. V. (2013). Changing work values in the United States, 1973–2006. Social Science Research, 42(2), 255–270. Kalleberg, A. L., Nesheim, T., & Olsen, K. M. (2009). Is participation good or bad for workers? Effects of autonomy, consultation and teamwork on stress among workers in Norway. Acta Sociologica, 52(2), 99–116.  Kalleberg, A. L., Reskin, B. F., & Hudson, K. (2000). Bad jobs in America: Standard and nonstandard employment relations and job quality in the United States. American Sociological Review, 65(2), 256–278. Kao, G., & Thompson, J. S. (2003). Racial and ethnic stratification in educational achievement and attainment. Annual Review of Sociology, 29(1), 417–442.  Kapsalis, C. (2006). Who gets student loans? (Cat. No. 75-001-XIE). Perspectives on Labour and Income, 18(2), 12–18. Kerckhoff, A. C. (2001). Education and social stratification processes in comparative perspective. Sociology of Education, 74(extra issue), 3–18. Khan, S. R. (2011). Privilege: The making of an adolescent elite at St. Paul’s school. Princeton, New Jersey: Princeton University Press.  162  Kim, C., Tamborini, C. R., & Sakamoto, A. (2015). Field of study in college and lifetime earnings in the United States. Sociology of Education, 88(4), 320–339. Kirby, D. (2007). Reviewing Canadian post-secondary education: Post-secondary education policy in post-industrial Canada. Canadian Journal of Educational Administration and Policy, 65(1), 1–24.  Kohler, U., Karlson, K. B., & Holm, A. (2011). Comparing coefficients of nested nonlinear probability models. Stata Journal, 11(3), 420–438. Krahn, H. (2017). Choose your parents carefully: Social class, post-secondary education, and occupational outcomes. In E. Grabb, J. G. Reitz, & M. Hwamg (Eds.), Social inequality in Canada (pp. 90–103). Don Mills, Ontario: Oxford University Press.  Krahn, H., & Galambos, N. L. (2014). Work values and beliefs of ‘Generation X’ and ‘Generation Y’. Journal of Youth Studies, 17(1), 92–112. Krahn, H., & Taylor, A. (2007). “Streaming” in the 10th grade in four Canadian provinces in 2000. Education Matters: Insights on Education, Learning and Training in Canada, 4(2), 1–10. Lambert, P., Connelly, R., Gayle, V., & Blackburn, R. M. (2016). Introduction. In P. Lambert, R. Connelly, R. M. Blackburn, & V. Gayle (Eds.), Social stratification: Trends and processes (pp. 1–9). London: Routledge. Lareau, A. (1987). Social class differences in family-school relationships: The importance of cultural capital. Sociology of Education, 60(1), 73–85. Lareau, A. (2002). Invisible inequality: Social class and childrearing in black families and white families. American Sociological Review, 67(5), 747–776.  Laroche, J. (2016, Oct 28). Program offers 1 year of free university to Nova Scotians on social assistance. CBC. Retrieved January 12, 2017, from www.cbc.ca/news/canada /nova-scotia/welfare-assistance-university-community-college-books-tuition-aid-community-services-1.3825959?cmp=rss Lefranc, A., Pistolesi, N., & Trannoy, A. (2008). Inequality of opportunities vs. inequality of outcomes: Are Western societies all alike?. Review of Income and Wealth, 54(4), 513–546. Legewie, J., & DiPrete, T. A. (2012). School context and the gender gap in educational achievement. American Sociological Review, 77(3), 463–485. 163  Lehmann, W. (2009). University as vocational education: Working-class students’ expectations for university. British Journal of Sociology of Education, 30(2), 137–149.  Lehmann, W. (2012a). Extra-credential experiences and social closure: Working-class students at university. British Educational Research Journal, 38(2), 203–218. Lehmann, W. (2012b). Working-class students, habitus, and the development of student roles: A Canadian case study. British Journal of Sociology of Education, 33(4), 527–546. Leschke, J. (2015). Non-standard employment of women in service sector occupations: A comparison of European countries. In W. Eichhorst & P. Marx (Eds.), Non-standard employment in post-industrial labour markets: An occupational perspective (pp. 324–352). Northampton: Edward Elgar Publishing. Leuze, K., & Strauß, S. (2014). Female-typical subjects and their effect on wage inequalities among higher education graduates in Germany. European Societies, 16(2), 275–298.  Lewis, T. (2007). Braverman, Foucault and the labor process: Framing the current high‐skills debate. Journal of Education and Work, 20(5), 397–415. Li, C. (2006). Canada’s private colleges: The lesser known players in postsecondary education (Cat. No. 11-621-MIE – No. 036). Ottawa, ON: Statistics Canada, Income Statistics Division and Socio-Economic Analysis and Modeling Division. Lindsay, P., & Knox, W. E. (1984). Continuity and change in work values among young adults: A longitudinal study. American Journal of Sociology, 89(4), 918–931. Livingstone, D. W. (2004). The education-jobs gap: Underemployment or economic democracy. Toronto: University of Toronto Press. Lombardo, R., De Luca, G., & Passarelli, G. (2012). Field of study and transition into a stable job: The case of a university in southern Italy. International Journal of Economics and Finance, 4(2), 69–84. Looker, E. D. (2009).
Regional differences in Canadian rural‐urban participation rates in post‐secondary education (MESA Project Research Paper 2009‐3). Toronto, ON: Canadian Education Project. Lorenz, E., & Lundvall, B. Å. (2011). Accounting for creativity in the European Union: A multi-level analysis of individual competence, labour market structure, and systems of education and training. Cambridge Journal of Economics, 35(2), 269–294. 164  Lucas, S. R. (2001). Effectively maintained inequality: Education transitions, track mobility, and social background effects. American Journal of Sociology, 106(6), 1642–1690.  Luu, K., & Freeman, J. G. (2011). An analysis of the relationship between information and communication technology (ICT) and scientific literacy in Canada and Australia. Computers & Education, 56(4), 1072–1082. Maas, C. J., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling. Methodology, 1(3), 86–92. MacLeod, J. (1987). Ain’t no making’it: Aspirations and attainment in a low-income neighborhood. Boulder: Westview Press.  Marini, M. M., Fan, P., Finley, E., & Beutel, A. M. (1996). Gender and job values. Sociology of Education, 69(1), 49–65. Marrero, G. A., & Rodríguez, J. G. (2013, September). Inequality of opportunity and growth. Journal of Development Economics, 104, 107–122. Marshall, K. (2003). Benefits of the job. Perspectives on Labour and Income, 4(5). Retrieved July 19, 2016, from www.statcan.gc.ca/pub/75-001-x/00503/6515-eng.html McCrate, E. (2007). The racial gap in autonomy at work. In M. Kim (Ed.), Race and economic opportunity in the twenty-first century (pp. 101–125). London: Routledge.  McCuaig, K., & Akbari, E. (2014). The early childhood education report: Its time for preschool. Toronto, ON: Atkinson Centre for Society and Child Development, University of Toronto. Retrieved March 12, 2017, from www.oise.utoronto.ca/ atkinson/About_Us/Staff_Faculty/Kerry_McCuaig/index.html#sthash.ymqvC8Nj.dpuf McEwen, N. (1995). Accountability in education in Canada. Canadian Journal of Education, 20(1), 1–26. McGinnity, F., Mertens, A., & Gundert, S. (2005). A bad start? Fixed-term contracts and the transition from education to work in West Germany. European Sociological Review, 21(4), 359–374. McMullen, K., & Elias, A. (2011). A changing portrait of international student in Canadian universities (Cat. No. 81-004-X). Education Matters: Insights on Education, Learning and Training in Canada, 7(6), 1–14. Ottawa, ON: Statistics Canada. Retrieved March 22, 2017, from www.statcan.gc.ca/pub/81-004-x/2010006 /article/11405-eng.htm#f 165  McNeish, D., Stapleton, L. M., & Silverman, R. D. (2017). On the unnecessary ubiquity of hierarchical linear modeling. Psychological Methods, 22(1), 114–140. Mehan, H. (1992). Understanding inequality in schools: The contribution of interpretive studies. Sociology of Education, 65(1), 1–20. Melguizo, T., & Wolniak, G. C. (2012). The earnings benefits of majoring in STEM fields among high achieving minority students. Research in Higher Education, 53(4), 383–405. Mincer, J. (1974). Schooling, experience, and earnings. New York, NY: Columbia University Press.  Mirowsky, J., & Ross, C. E. (1998). Education, personal control, lifestyle and health: A human capital hypothesis. Research on Aging, 20(4), 415–449. Mirza, H. (2009). Race, gender, and educational desire: Why black women succeed and fail. New York, NY: Routledge. Molgat, M., Deschenaux, F., & LeBlanc, P. (2011). Vocational education in Canada: Do policy directions and youth trajectories always meet? Journal of Vocational Education & Training, 63(4), 505–524.  Morcke, A. M., Dornan, T., & Eika, B. (2013). Outcome (competency) based education: An exploration of its origins, theoretical basis, and empirical evidence. Advances in Health Sciences Education, 18(4), 851–863. Morgan, L. A. (2008). Major matters: A comparison of the within‐major gender pay gap across college majors for early‐career graduates. Industrial Relations: A Journal of Economy and Society, 47(4), 625–650. Moyser, M. (2017). Women and paid work (Cat. No. 89-503-X). Ottawa, ON: Statistics Canada. Retrieved July 7, 2017, from www.statcan.gc.ca/pub/89-503-x/2015001/article /14694-eng.htm Mullen, A. L. (2010.) Degrees of inequality: Culture, class, and gender in American higher education. Baltimore: John Hopkins University Press.  Mustosmäki, A., Anttila, T., Oinas, T., & Nätti, J. (2011). Job quality trends in Europe—Implications for work-life balance. In S. Drobnič & A. M. Guillén (Eds.), Work-life balance in Europe: The role of job quality (pp. 17–43). New York, NY: Palgrave Macmillan. 166  Newbold, K. B., & Brown, W. M. (2015). The urban-rural gap in university attendance: Determinants of university participation among Canadian youth. Journal of Regional Science, 55(4), 585–608. Newman, B. J., Johnston, C. D., & Lown, L. (2015). False consciousness or class awareness? Local income inequality, personal economic position, and belief in American meritocracy. American Journal of Political Science, 59(2), 326–340. O’Connell, A. A. (2010). An illustration of multilevel models for ordinal response data. Proceedings of the Eighth International Conference on Teaching Statistics. Retrieved April 4, 2016, from www.iaseweb.org/documents/papers/icots8/ ICOTS8_4C3_OCONNELL.pdf O’Grady, K., Deussing, M. A., Scerbina, T., Fung, K., & Muhe, N. (2016). Measuring up: Canadian Results of the OECD PISA Study. Ottawa, ON: Council of Ministers of Education, Canada.  Ochsenfeld, F. (2014). Why do women’s fields of study pay less? A test of devaluation, human capital, and gender role theory. European Sociological Review, 30(4), 536–548. Oesterle, S., Hawkins, J. D., Hill, K. G., & Bailey, J. A. (2010). Men’s and women’s pathways to adulthood and their adolescent precursors. Journal of Marriage and Family, 72(5), 1436–1453. Oliver, L. N., Dunn, J. R., Kohen, D. E., & Hertzman, C. (2007). Do neighbourhoods influence the readiness to learn of kindergarten children in Vancouver? A multilevel analysis of neighbourhood effects. Environment and Planning A, 39(4), 848–868. Ontario College of Teachers & Ontario Teachers’ Federation. (2004). Narrowing the gender gap: Attracting men to teaching. Toronto: Author. Retrieved June 15, 2017, from www.oct.ca/-/media/PDF/Attracting%20Men%20To%20Teaching/EN/ Men_In_Teaching_e.pdf Organisation for Economic Co-operation and Development. (2010). PISA 2000 and the Canadian context. Paris: Author. Organisation for Economic Co-operation and Development. (2013a). OECD skill outlook 2013: First results from the survey of adults. Paris: Author. Organisation for Economic Co-operation and Development. (2013b). OECD Indicators of employment protection. Paris: OECD Stat. Retrieved from December 5, 2016, from www.stats.oecd.org/# 167  Organisation for Economic Co-operation and Development. (2015a). In it together: Why less inequality benefits all. Paris: Author. Organisation for Economic Co-operation and Development. (2015b). Education policy outlook 2015: Making reforms happen. Paris: Author.  Organisation for Economic Co-operation and Development. (2016a). Education at a glance: 2015 indicators. Paris: Author.  Organisation for Economic Co-operation and Development. (2016b). LMF1.6: Gender differences in employment outcomes. Paris: Author. Retrieved January 7, 2017, from www.oecd.org/els/soc/LMF_1_6_ Gender_differences_in_employment_ outcomes.pdf Organisation for Economic Co-operation and Development. (2016c). The survey of adult skills: Reader’s companion (2nd Edition). Paris: Author. Organisation for Economic Co-operation and Development. (2017a). Education finance indicators: Public expenditure on education as a percentage of GDP. Paris: OECD Stat. Retrieved March 5, 2017, from www.stats.oecd.org/# Organisation for Economic Co-operation and Development. (2017b). Intergenerational mobility in education. Paris: OECD Stat. Retrieved March 11, 2017, from www.stats.oecd.org/# Organisation for Economic Co-operation and Development. (2017c). Share of women graduates by field. Paris: OECD Stat. Retrieved March 17, 2017, from www.stats.oecd.org/# Organisation for Economic Co-operation and Development. (2017d). Incidence of involuntary part time workers. Paris: OECD Stat. Retrieved July 2, 2017, from www.stats.oecd.org/# Organisation for Economic Co-operation and Development. (2017e). Incidence of permanent employment. Paris: OECD Stat. Retrieved from July 2, 2017, from www.stats.oecd.org/# Ostrovsky, Y., & Frenette, M. (2014). The cumulative earnings of postsecondary graduates over 20 years: Results by field of study (Catalogue no. 11-626-X — No. 040). Ottawa, ON: Statistics Canada. Retrieved July 7, 2016, from www.statcan.gc.ca/pub/11-626-x/11-626-x2014040-eng.htm Pechar, H., & Andres, L. (2011). Higher education policy and welfare regimes: International comparative perspectives. Higher Education Policy, 24(1), 25–52. 168  Pedulla, D. S. (2016). Penalized or protected? Gender and the consequences of nonstandard and mismatched employment histories. American Sociological Review, 81(2), 262–289. Pena, A. A. (2015). Revisiting the effects of skills on economic inequality: Within- and cross-country comparisons using PIAAC. Washington, DC: AIR-PIAAC. Retrieved October 15, 2016, from www.piaacgateway.com/researchpapers/ Penn, R. (2016). Rethinking class analysis: Some reflections on current issues and possible new forms of empirical research. Contemporary Social Science, 11(2-3), 113–124. Peter, T., Edgerton, J. D., & Roberts, L. W. (2010). Welfare regimes and educational inequality: A cross‐national exploration. International Studies in Sociology of Education, 20(3), 241–264. Petrie, M., & Roman, P. M. (2004). Race and gender differences in workplace autonomy: A research note. Sociological Inquiry, 74(4), 590–603.  Piche, P. G., & Jones, G. (2016). Institutional diversity in Ontario’s university sector: A policy debate analysis. Canadian Journal of Higher Education, 46(3), 1–17. Pokropek, A., & Jukubowski, M. (2013). PIAACTOOLS: Stata module to provide PIAAC tools Boston, MA: RePEc/Ideas Statistical Software Components. Retrieved December 2, 2015, from www.ideas.repec.org/c/boc/bocode/s457728.html Polachek, S. W. (1981). Occupational self-selection: A human capital approach to sex differences in occupational structure. The Review of Economics and Statistics, 63(1), 60–69. Prescott, J., & Bogg, J. (2013). Gendered occupational differences in science, engineering, and technology careers. Hershey, PA: Information Science Reference. Prix, I. (2009). Gender and recent graduates’ occupational stratification: The interactive role of the educational and employment sectors in four countries. Comparative Education, 45(4), 545–568. Purcell, K. (2000). Gendered employment insecurity? In E. Heery & J. Salmon (Eds.). The insecure workforce (pp. 112–139). London UK: Routledge. Qualizza, A. (2015, November 3). UBC Engineering sets goal of 50 per cent female enrollment by 2020. Vancouver: The UBYSSEY. Retrieved January 8, 2017, from www.ubyssey.ca/ news/engineering-sets-goal-fifty-percent-2020/ Raphael, D. (2009), Social determinants of health: Canadian perspectives. Toronto, ON: Canadian Scholars’ Press 169  Raudenbush S., & Bryk, A. (2002). Hierarchical linear models. Newbury Park, CA: Sage Rawls, J. (1971). A theory of justice. Cambridge, MA: Harvard University Read, J. N. G., & Oselin, S. (2008). Gender and the education-employment paradox in ethnic and religious contexts: The case of Arab Americans. American Sociological Review, 73(2), 296–313. Reay, D. (1998). Class work: Mothers’ involvement in children’s schooling. London, UK: Falmer. Reder, S. (2015). Digital inclusion and digital literacy in the United States: A portrait from PIAAC’s survey of adult skills. Washington, DC: AIR-PIAAC. Retrieved November 25, 2016, from www.piaacgateway.com/researchpapers/ Reimer, D., & Steinmetz, S. (2009). Highly educated but in the wrong field? Educational specialisation and labour market risks of men and women in Spain and Germany. European Societies, 11(5), 723–746. Reimer, D., Noelke, C., & Kucel, A. (2008). Labor market effects of field of study in comparative perspective an analysis of 22 European countries. International Journal of Comparative Sociology, 49(4-5), 233–256. Reiss, S. (2012). Intrinsic and extrinsic motivation. Teaching of Psychology, 39(2), 152–156. Rigney, D. (2010). The Matthew effect: How advantage begets further advantage. New York, NY: Columbia University Press. Rivera, L. A. (2011). Ivies, extracurriculars, and exclusion: Elite employers’ use of educational credentials. Research in Social Stratification and Mobility, 29(1), 71–90. Rivera, L. A. (2012). Hiring as cultural matching the case of elite professional service firms. American Sociological Review, 77(6), 999–1022. Roemer, J. E. (1998). Equality of opportunity. Cambridge, MA: Harvard University Press.  Rokeach, M. (1968). Beliefs, attitudes and values: A theory of organization and change. San Francisco, CA: Milton.  Roksa, J., & Levey, T. (2010). What can you do with that degree?: College major and occupational status of college graduates over time. Social Forces, 89(2), 389–415. Rose, J., Hewitt, B., & Baxter, J. (2013). Women and part-time employment: Easing or squeezing time pressure?. Journal of Sociology, 49(1), 41–59. 170  Rose, M. (2004). The mind at work: Valuing the intelligence of the American worker. New York, NY: Viking.  Salamon, E. (2015). (De)valuing intern labour: Journalism internship pay rates and collective representation in Canada. tripleC: Communication, Capitalism & Critique, 13(2), 438–458. Scherer, S. (2004). Stepping-stones or traps? The consequences of labour market entry positions on future careers in West Germany, Great Britain and Italy. Work, Employment & Society, 18(2), 369–394.  Schultz, T. W. (1960). Capital formation by education. The Journal of Political Economy, 68(6), 571–583. Schweitzer, L., Ng, E., Lyons, S., & Kuron, L. (2011). Exploring the career pipeline: Gender differences in pre-career expectations. Industrial Relations, 66(3), 422–444. Scott, N. A., & Siltanen, J. (2016). Intersectionality and quantitative methods: Assessing regression from a feminist perspective. International Journal of Social Research Methodology, early release. doi:10.1080/13645579.2016.1201328 Sen, A. (1992). Inequality reexamined. New York, NY: Oxford University Press. Shanahan, T., & Jones, G. A. (2007). Shifting roles and approaches: Government coordination of post‐secondary education in Canada, 1995-2006. Higher Education Research & Development, 26(1), 31–43. Shapka, J. D., Domene, J. F., & Keating, D. P. (2012). Trajectories of educational aspirations through high school and beyond: A gendered phenomenon. Canadian Journal of Education, 35(1), 239–258. Shauman, K. A. (2016). Gender differences in the early career outcomes of college graduates: The influence of sex-type of degree field across four cohorts. The Russell Sage Foundation Journal of the Social Sciences, 2(4), 152–193. Shorrocks, A. F. (2013). Decomposition procedures for distributional analysis: A unified framework based on the Shapley value. Journal of Economic Inequality, 11(1), 1–28.  Singer, J. D., & Willett. J. B. (2003). Applied longitudinal data analysis: Modelling change and event occurrence. New York, NY: Oxford University Press. Smith, W. C., & Fernandez, F. (2015). Education and wage gaps: A comparative study of immigrant and native employees in the United States and Canada. Washington, DC: AIR-PIAAC. Retrieved November 22, 2016, from www.piaacgateway.com/ researchpapers/ 171  Smyth, E. (2005). Gender differentiation and early labour market integration across Europe. European Societies, 7(3), 451–479. Sobel, I. (1982). Human capital and institutional theories of the labor market: Rivals or complements?. Journal of Economic Issues, 16(1), 255–272. Spenner, K. I. (1990). Skill, meanings, methods, and measures. Work and Occupations, 17(4), 399–421.  StataCorp. (2015). Stata statistical software: Release 14. College Station, TX: Author. Statistics Canada. (2012). Economic downturn and educational attainment (Cat. No. 81-599-X). Ottawa, ON: Author. Retrieved September 18, 2017, from http://www.statcan.gc.ca/pub/81-599-x/81-599-x2012009-eng.pdf Statistics Canada. (2013). Education in Canada: Attainment, field of study and location of study (Cat. No. 99-012-X2011001). Ottawa, ON: Author. Retrieved March 5, 2017, from www12.statcan.gc.ca/nhs-enm/2011/as-sa/99-012-x/99-012-x2011001-eng.pdf Statistics Canada. (2015). Aboriginal Statistics at a Glance, 2nd Edition (Cat. No. 89-645-x2015001). Ottawa, ON: Author. Retrieved March 19, 2017, from www.statcan. gc.ca/pub/89-645-x/89-645-x2015001-eng.pdf Statistics Canada. (2016). Education indicators in Canada: An international perspective, 2016. Ottawa, ON: Author. Retrieved March 3, 2017, from www.statcan.gc.ca/daily-quotidien/161215/dq161215b-eng.pdf  Steinberg, R. J. (1990). Social construction of skill: Gender, power, and comparable worth. Work and occupations, 17(4), 449–482. Stevens, M. L., Armstrong, E. A., & Arum, R. (2008). Sieve, incubator, temple, hub: Empirical and theoretical advances in the sociology of higher education. Annual Review of Sociology, 34, 127–151.  Sturgis, P., & Buscha, F. (2015). Increasing inter‐generational social mobility: Is educational expansion the answer?. The British Journal of Sociology, 66(3), 512–533. Sullivan, A. (2006). Students as rational decision-makers: The question of beliefs and attitudes. London Review of Education, 4(3), 271–290. Taylor, C., Peter, T., McMinn, T. L., Elliott, T., Beldom, S., Ferry, A., Gross, Z., Paquin, S., & Schachter, K. (2011). Every class in every school: The first national climate survey on homophobia, biphobia, and transphobia in Canadian schools. Final report. Toronto, ON: Egale Canada Human Rights Trust. Retrieved March 21, 2017, from www.egale.ca/wp-content/uploads/2011/05/EgaleFinalReport-web.pdf 172  Terzi, L. (2014). Reframing inclusive education: Educational equality as capability equality. Cambridge Journal of Education, 44(4), 479–493. Thomas, S. L., & Zhang, L. (2005). Post-baccalaureate wage growth within four years of graduation: The effects of college quality and college major. Research in Higher Education, 46(4), 437–459. Thompson, R., & Simmons, R. (2013). Social mobility and post-compulsory education: Revisiting Boudon’s model of social opportunity. British Journal of Sociology of Education, 34(5-6), 744–765. Tomlinson, M. (2013). Education, work and identity: Themes and perspectives. London, UK: Bloomsbury. Torche, F. (2011). Is a college degree still the great equalizer? Intergenerational mobility across levels of schooling in the United States. American Journal of Sociology, 117(3), 763–807. Triventi, M. (2013, June). The role of higher education stratification in the reproduction of social inequality in the labor market. Research in Social Stratification and Mobility, 32, 45–63.  Turcotte, M. (2011). Intergenerational education mobility: University completion in relation to parents’ education level. Canadian Social Trends, 92, 37–43. Twenge, J. M., Campbell, S. M., Hoffman, B. J., & Lance, C. E. (2010). Generational differences in work values: Leisure and extrinsic values increasing, social and intrinsic values decreasing. Journal of Management, 36(5), 1117–1142. United Nations Educational, Scientific and Cultural Organization. (2011). Revision of the international standard classification of education. Paris: Author. Retrieved October 21, 2016, from www.uis.unesco.org/Education/ Documents/UNES CO_GC_36C-19_ISCED_EN.pdf Usher, A. (2009). The post-secondary student support program: An examination of alternative delivery mechanisms. Toronto, ON: Educational Policy Institute. Retrieved March 1, 2017, from www.educationalpolicy.org/publications/ pubpdf/INAC.pdf  Vinovskis, M. (2015). From A Nation at Risk to No Child Left Behind: National education goals and the creation of federal education policy. New York, NY: Teachers College Press. Vosko, L. F. (2003). Precarious employment in Canada: Taking stock, taking action. Just Labour, 3(fall), 1–5. 173  Vosko, L. F. (2006). Precarious employment: Towards an improved understanding of labour market insecurity. In L. F. Vosko (Ed.), Precarious employment: Understanding labour market insecurity in Canada (pp. 3–39). Montreal, QC: McGill-Queen's University Press. Vosko, L. F. (2010). Managing the margins: Gender, citizenship, and the international regulation of precarious employment. New York, NY: Oxford University Press. Vosko, L. F., & Clark, L. F. (2009). Canada: Gendered precariousness and social reproduction. In L. F. Vosko, M. MacDonald, & I. Campbell (Eds.), Gender and the contours of precarious employment: Developing common understandings across space, scale, and social location (pp. 26–42). New York, NY: Routledge.  Vuorinen-Lampila, P. (2016). Gender segregation in the employment of higher education graduates. Journal of Education and Work, 29(3), 284–308. Waite, S., & Denier, N. (2015). Gay pay for straight work: Mechanisms generating disadvantage. Gender & Society, 29(4), 561–588. Wallner, J. (2012). Political structures, social diversity, and public policy comparing mandatory education in Canada and the United States. Comparative Political Studies, 45(7), 850–874. Wallner, J. (2014). Learning to school: Federalism and public schooling in Canada. Toronto, ON: University of Toronto Press Wang, F. (2013). Educational equity in the access to post-secondary education: A comparison of ethnic minorities in China with Aboriginals in Canada. Interchange, 44(1), 45–62.  Watts, M. (2013). The complexities of adaptive preferences in post-compulsory education: Insights from the fable of the fox and the grapes. Journal of Human Development and Capabilities, 14(4), 503–519. West, A., & Nikolai, R. (2013). Welfare regimes and education regimes: Equality of opportunity and expenditure in the EU (and US). Journal of Social Policy, 42(3), 469–493. Wheelahan, L., & Moodie, G. (2011). Rethinking skills in vocational education and training: From competencies to capabilities. Darlinghurst, NSW: NSW Board of Vocational Education and Training. Retrieved March 7, 2017, from www.skillsboard. nsw.gov.au/sites/default/files/rethinking_skills.pdf 174  White, L. A., & Prentice, S. (2016). Early childhood education and care reform in Canadian provinces: Understanding the role of experts and evidence in policy change. Canadian Public Administration, 59(1), 26–44. Willemse, N., & De Beer, P. (2012). Three worlds of educational welfare states? A comparative study of higher education systems across welfare states. Journal of European Social Policy, 22(2), 105–117. Willis, P. E. (1977). Learning to labor: How working class kids get working class jobs. New York, NY: Columbia University Press. Willis, S., & Kissane, B. (1997). Systemic approaches to articulating and monitoring student outcomes: Are they consistent with outcome-based education?. Studies in Educational Evaluation, 23(1), 5–30.  Woodfield, R. (2012). Mature women students, study motivation and employability. In T. Hinton-Smith (Ed.), Widening participation in higher education (pp. 91-107). London, UK: Palgrave Macmillan. Wright, L., Walters, D., & Zarifa, D. (2013). Government student loan default: Differences between graduates of the liberal arts and applied fields in Canadian colleges and universities. Canadian Review of Sociology, 50(1), 89–115. Xu, Y. J. (2013). Career outcomes of STEM and non-STEM college graduates: Persistence in majored-field and influential factors in career choices. Research in Higher Education, 54(3), 349–382. Younglai, R. (2016, Oct 28). For young Canadian, a new reality: Dealing with ‘job churn’. The Globe and Mail. Retrieved January 11, 2017, from www.theglobeandmail.com/ report-on-business/economy/jobs/canadian-youth-working-low-paid-temporary-jobs/article32579528/ Zafar, B. (2013). College major choice and the gender gap. Journal of Human Resources, 48(3), 545–595. Zarifa, D. (2012). Choosing fields in an expansionary era: Comparing two cohorts of baccalaureate degree-holders in the United States and Canada. Research in Social Stratification and Mobility, 30(3), 328–351. Zeytinoglu, I. U., & Cooke, G. B. (2008). Non-standard employment and promotions: A within genders analysis. Journal of Industrial Relations, 50(2), 319–337. Zhou, Y. (2014). Task discretion. In A. C. Michalos (Ed.), Encyclopedia of quality of life and well-being research (pp. 6587–6588). Dordrecht, Netherlands: Springer. 175  Appendices Appendix A  : Chapter 3 Variable Overview Dependent variables Name Variable type Description Code Temporary Dummy Temporary worker during survey week 1=yes Part-time Dummy Worked 29 hours or less during survey week 1=yes  Main independent variables  Name Variable type Description Code  Female Dummy Gender 1=female 1 Education Dummy Education, health/physical education 1=yes 2 Fine arts Dummy Fine arts, music, preforming arts, general arts 1=yes 3 Media & Communications Dummy Commercial arts, graphic arts, other applied arts, mass media studies, journalism 1=yes 4 Languages Dummy Languages (excluding French), classics, classical and dead languages, humanities, liberal arts 1=yes 5 History Dummy History 1=yes 6 Literature Dummy English languages and literature, French languages and literature, library and record science (small cell count) 1=yes 7 Philosophy & religion Dummy Religious studies and philosophy 1=yes 8 Economics  Dummy Economics 1=yes 9 Geography Dummy Geography, urban and area studies  1=yes 10 Law Dummy Law and jurisprudence 1=yes 11 Environmental studies Dummy Environment studies technologies, forestry, fisheries and wildlife management 1=yes 12 Political science & criminology Dummy Political science and criminology  1=yes 13 Psychology Dummy Psychology, counselling services  1=yes 14 Sociology & anthropology Dummy Sociology, anthropology  1=yes 15 Social work & public policy Dummy Social work, public policy, social services  16 Business Dummy Commerce, business, management, specialized administration studies 1=yes 17 Agriculture Dummy Agricultural science 1=yes 18 Chemistry Dummy Chemistry, biochemistry, chemical technology 1=yes 19 Biology Dummy Biology, botany, biophysics, zoology 1=yes 20 Architecture  Dummy Architecture and landscape architecture 1=yes 21 Engineering Dummy Electronic and electrical engineering, civil engineering, aeronautical and aerospace engineering, engineering science/other, 1=yes 176   Name Variable type Description Code design/systems engineering, industrial engineering, industrial engineering technologies, mining and metallurgical engineering, mechanical engineering, chemical engineering 22 Computer science Dummy Computer science and data processing  1=yes 23 Medical sciences Dummy Medicine (general and basic medical science), Paraclinical sciences, surgery and surgical specialities/residencies, optometry, dentistry, veterinary medicine/science 1=yes 24 Nursing Dummy Nursing, pre-nursing, nursing assistance 1=yes 25 Pharmaceutical sciences Dummy Pharmacy and pharmaceutical sciences 1=yes 26 Public health Dummy Public health, work and family studies, consumer studies, nutrition, adult/child development and aging, consumer studies 1=yes 27 Laboratory & rehabilitation medicine Dummy Medical equipment and prosthetics, medical laboratory and treatment technologies, rehabilitation medicine and therapy, other health professions 1=yes 28 Math Dummy Mathematics, actuarial science and applied, physics, geology, metallurgy and materials science, meteorology, astronomy and astrophysics 1=yes  Other independent variables:  Name Variable type Description Code Additional education Dummy Taken additional programs since graduation 1=yes Student loan Continuous  Student loan amount at graduation  Age Continuous Age at graduation range Child Dummy Respondent has dependent children 1=yes Couple Dummy Marital status: in common-law or married relationship 1=yes Province Newfoundland Dummy Primary residence Newfoundland and Labrador at time of survey 1=yes PEI Dummy Primary residence PEI at time of survey 1=yes Nova scotia Dummy Primary residence Nova Scotia at time of survey 1=yes New Brunswick Dummy Primary residence New Brunswick at time of survey 1=yes Quebec Dummy Primary residence Quebec at time of survey 1=yes Ontario Dummy Primary residence Ontario at time of survey 1=yes Manitoba Dummy Primary residence Manitoba at time of survey 1=yes Saskatchewan Dummy Primary residence Saskatchewan at time of survey 1=yes Alberta Dummy Primary residence Alberta at time of survey 1=yes 177  Name Variable type Description Code British Columbia Dummy Primary residence BC at time of survey 1=yes Territories Dummy Primary residence Nunavut, Yukon, or Northwest Territories at time of survey 1=yes Industry Public_ind Dummy Industry classification-Public administration  1=yes Other_ind Dummy Industry classification -Other services (except public administration)  1=yes Food_ind Dummy Industry classification-Accommodation and food services  1=yes Ed_ind Dummy Industry classification-Educational services  1=yes Health_ind Dummy Industry classification-Health care and social assistance  1=yes Info_ind Dummy Industry classification-Information, culture and recreation  1=yes Finance_ind Dummy Industry classification-Finance, insurance, real estate and leasing  1=yes Profess_ind Dummy Industry classification-Professional, scientific and technical services  1=yes Manag_ind Dummy Industry classification-Business, building and other support services  1=yes Agr_ind Dummy Industry classification-Agriculture  1=yes Resource_ind Dummy Industry classification-Forestry, fishing, hunting, mining, oil and gas  1=yes Utilities_ind Dummy Industry classification-Utilities  1=yes Construct_ind Dummy Industry classification-Construction  1=yes Manufact_ind Dummy Industry classification-Manufacturing – durables and non-durables 1=yes Wholesaletrade_ind Dummy Industry classification-Wholesale trade  1=yes Retailtrade_ind Dummy Industry classification-Retail trade  1=yes Transport_ind Dummy Industry classification-Transportation and warehousing  1=yes 178  Appendix B  : Chapter 4 OLS Regression Results Table 6.1. The relationship between task discretion and education  Robust standard errors in parentheses; * p < .05, ** p < .01, *** p < .001; Reference group: lower secondary Model 1: Bivariate relationship between education and task discretion. Model 2: Controls for gender, age, non-native speaker, literacy, self-employment, public sector employment, part-time employment and income Model 3: All controls plus occupational sector.   179  Appendix C  : Chapter 4 Robustness Testing  All demographic and occupational controls included.  Only coefficients significant at the 95% confidence level shown with corresponding confidence interval (p < .05). Figure 6.1. The mediation effect of numeracy and technology skills by education level  00.20.40.60.811.21.4βNumeracyDegree Diploma Upper Secondary0.000.200.400.600.801.001.201.40βProblem solving in technology-rich environmentsDegree Diploma Upper Secondary180  Appendix D  : Chapter 4 KHB Results Table 6.2. The effect of education as mediated by occupational sector and literacy assessment score  Robust standard errors in parentheses.  Reference group: lower secondary. All demographic and occupational controls included in models, as well as occupational sector (Model 6) or literacy score (Model 7). * p < .05, ** p < .01, *** p < .001  181  Appendix E  : Chapter 4 Estimation of Country-Level Effects Table 6.3. Pooled estimation of country average and occupational range in task discretion  Average  Occupational range  No occupational controls With occupational controls  No occupational  controls With occupational  controls Mean 1.01***(.06) 1.01***(.06) Range -.58***(.06) -.53***(.06) Up.sec.*Mean  .01    (.06) .05    (.06) Up.sec.*Range .07    (.08) .03    (.07) Diploma*Mean -.18*   (.08) -.14    (.08) Diploma*Range .47***(.09) .39***(.09) Degree*Mean -.22***(.06) -.19** (.06) Degree*Range .30***(.08) .23***(.08) Upper-sec.  .29***(.04) .12** (.04) Upper-sec.  .50***(.04) .32***(.04) Diploma  .63***(.07) .27***(.07) Diploma  .81***(.06) .46***(.06) Degree  .97***(.05) .39***(.07) Degree  1.02***(.05) .50***(.06) R2 .16 .19 R2 .12 .15 N=125,123; * p < .05, ** p < .01, *** p < .001; Cluster-robust standard errors in parentheses.27  Reference group: lower secondary. Occupational range and average task discretion centred.  All models control for gender, age, non-native speaker status, literacy, self-employment, public sector employment, and part-time employment.                                                     27 We use cluster-robust standard errors (also termed the sandwich estimator or empirical standard errors) to account for country clustering. Unlike the other pooled models included in our chapter, fixed effect modelling would introduce unacceptable levels of multicollinearity due to the perfect correlation between country-level average/range measures. The use of a single-level model with cluster-robust standard errors advantageous when a researcher seeks to examine variability across clusters and random effects are not of substantive research interest (McNeish, Stapleton, & Silverman, 2017).  182  Appendix F  : Chapter 5 Model Equations and Parameters  Level 1  Level 2  Un-conditional model Prob[Rti <= 1|πi] = ϕ*1ti = ϕ1ti Prob[Rti <= 2|πi] = ϕ*2ti = ϕ1ti + ϕ2ti Prob[Rti <= 3|πi] = ϕ*3ti = ϕ1ti + ϕ2ti + ϕ3ti Prob[Rti <= 4|πi] = ϕ*4ti = ϕ1ti + ϕ2ti + ϕ3ti + ϕ4ti Prob[Rti <= 5|πi] = 1.0 ϕ1ti = Prob[BELIEF(1) = 1|πi] ϕ2ti = Prob[BELIEF(2) = 1|πi] ϕ3ti = Prob[BELIEF(3) = 1|πi] ϕ4ti = Prob[BELIEF(4) = 1|πi] log[ϕ*1ti/(1 ϕ*1ti)] = π0i log[ϕ*2ti/(1 ϕ*2ti)] = π0i + δ2 log[ϕ*3ti/(1 ϕ*3ti)] = π0i + δ3 log[ϕ*4ti/(1 ϕ*4ti)] = π0i + δ4 π0i = β00 + r0i δ2 δ3 δ4 Conditional model Prob[Rti <= 1|πi] = ϕ*1ti = ϕ1ti Prob[Rti <= 2|πi] = ϕ*2ti = ϕ1ti + ϕ2ti Prob[Rti <= 3|πi] = ϕ*3ti = ϕ1ti + ϕ2ti + ϕ3ti Prob[Rti <= 4|πi] = ϕ*4ti = ϕ1ti + ϕ2ti + ϕ3ti + ϕ4ti Prob[Rti <= 5|πi] = 1.0 ϕ1ti = Prob[BELIEF(1) = 1|πi] ϕ2ti = Prob[BELIEF(2) = 1|πi] ϕ3ti = Prob[BELIEF(3) = 1|πi] ϕ4ti = Prob[BELIEF(4) = 1|πi] log[ϕ*1ti/(1 ϕ*1ti)] = π0i + π1i*(TIMEti) + π2i*(TIME2ti)  log[ϕ*2ti/(1 ϕ*2ti)] = π0i + π1i*(TIMEti) + π2i*(TIME2ti) + δ2 log[ϕ*3ti/(1 ϕ*3ti)] = π0i + π1i*(TIMEti) + π2i*(TIME2ti) + δ3 log[ϕ*4ti/(1 ϕ*4ti)] = π0i + π1i*(TIMEti) + π2i*(TIME2ti) + δ4 π0i = β00 + r0i π1i = β10 + r1i δ2 δ3 δ4 Final model  Prob[Rti <= 1|πi] = ϕ*1ti = ϕ1ti Prob[Rti <= 2|πi] = ϕ*2ti = ϕ1ti + ϕ2ti Prob[Rti <= 3|πi] = ϕ*3ti = ϕ1ti + ϕ2ti + ϕ3ti Prob[Rti <= 4|πi] = ϕ*4ti = ϕ1ti + ϕ2ti + ϕ3ti + ϕ4ti Prob[Rti <= 5|πi] = 1.0 ϕ1ti = Prob[BELIEF(1) = 1|πi] ϕ2ti = Prob[BELIEF(2) = 1|πi] ϕ3ti = Prob[BELIEF(3) = 1|πi] ϕ4ti = Prob[BELIEF(4) = 1|πi] log[ϕ*1ti/(1 ϕ*1ti)] = π0i + π1i*(TIMEti) + + π2i*(TIME2ti) + π3i*(EMPLOYMEti) + π4i*(EDUCATIOti)  log[ϕ*2ti/(1 ϕ*2ti)] = π0i + π1i*(TIMEti) + π2i*(TIME2ti) + π3i*(EMPLOYMEti) + π4i*(EDUCATIOti) + δ2 log[ϕ*3ti/(1 ϕ*3ti)] = π0i + π1i*(TIMEti) + π2i*(TIME2ti) + π3i*(EMPLOYMEti) + π4i*(EDUCATIOti) + δ3 log[ϕ*4ti/(1 ϕ*4ti)] = π0i + π1i*(TIMEti) + π2i*(TIME2ti) + π3i*(EMPLOYMEti) + π4i*(EDUCATIOti) + δ4 π0i = β00 + β01*(SEXi) + β02*(PARENT_EDi) + β03*(MYEARS_EDi) + β04*(MYEARS_EMPi) + r0i π1i = β10 + β11*(SEXi) + β12*(PARENT_EDi)  r1i π2i = β20 + β21*(SEXi) + β22*(PARENT_EDi) π3i = β30 π4i = β40 δ2 δ3 δ4 All level-1 continuous variables are centred around the group mean.  183  Appendix G  : Chapter 5 Conditional Linear and Curvilinear Growth Models Table 6.4. Results of linear and curvilinear growth models measuring extrinsic beliefs    Get ahead Lifestyle Income  Model  2a Model  2b Model  2a Model  2b Model  2a Model  2b Rate of change (time)  -.135*** (.031) -.386*** (.107) -.092** (.034) -.441*** (.112) -.098** (.034) -.398*** (.109) Quadratic rate of change (time2)   .053* (.027)  .073*** (.021)  .063** (.020) Thresholds       δ1 -5.225*** (.214) -5.408*** (.228) -5.041*** (.238) -5.302*** (.264) -4.442*** (.215) -4.654*** (.234) δ2 -2.925*** (.149) -3.104*** (.167) -1.613*** (.173) -1.859*** (.195) -1.052***  (.153) -1.256***  (.173) δ3 -2.153*** (.131) -2.329*** (.152) -.527** (.169) -.763** (.192) .010 (.153) -.188 (.171) δ4 .570*** (.120) .411** (.139) 1.687*** (.171) 1.472*** (.189) 2.544*** (.169) 2.363*** (.182) Variance components       Initial status 3.624 (.577) 3.747 (.597) 9.228 (1.047) 9.492 (1.080) 7.239 (.857) 7.396 (.878) Rate of change .149  (.042) .161  (.043) .220 (.044) .231 (.045) .260 (.048) .267 (.049) * p < .05; ** p < .01; ***p < .001; Robust standard errors in parentheses; N=510. Table 6.5. Results of linear and curvilinear growth models measuring intrinsic beliefs    Communication Reasoning Not for me  Model  2a Model  2b Model  2a Model  2b Model  2a Model  2b Rate of change (time)  -.063 (.033) -.402*** (.110) -.058 (.032) -.487*** (.113) .208*** (.044) .385** (.128) Quadratic rate of change (time2)   .072*** (.021)  .091*** (.023)  -.036 (.024) Thresholds       δ1 -6.857*** (.351) -7.110*** (.372) -6.858*** (.285) -7.180*** (.299) 1.176*** (.177) 1.304*** (.202) δ2 -4.434*** (.208) -4.683*** (.233) -4.341*** (.202) -4.658*** (.218) 4.191*** (.239)  4.331*** (.262)  δ3 -2.980*** (.171) -3.221*** (.193) -2.651*** (.159) -2.953*** (.181) 6.332*** (.299) 6.478*** (.321) δ4 .699*** (.133) .192*** (.151) 1.573*** (.147) 1.330*** (.161) 8.346*** (.475) 8.495*** (.492) Variance components       Initial status 4.716 (.684) 4.875 (.711) 4.898 (.714) 5.107 (.748) 8.329 (1.314) 8.527 (1.346) Rate of change .132 (.037) .146 (.039) .138 (.046) .155 (.048) .152 (.046) .164 (.049) * p < .05; ** p < .01; ***p < .001; Robust standard errors in parentheses; N=510.  

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.24.1-0355840/manifest

Comment

Related Items