UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Open data portals in northern New England states Paige, Bonnie E. 2017

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

Item Metadata


24-ubc_2017_november_paige_bonnie.pdf [ 520.94kB ]
JSON: 24-1.0355232.json
JSON-LD: 24-1.0355232-ld.json
RDF/XML (Pretty): 24-1.0355232-rdf.xml
RDF/JSON: 24-1.0355232-rdf.json
Turtle: 24-1.0355232-turtle.txt
N-Triples: 24-1.0355232-rdf-ntriples.txt
Original Record: 24-1.0355232-source.json
Full Text

Full Text

Open Data Portals In Northern New EnglandStatesbyBonnie E. Paigea thesis submitted in partial fulfillmentof the requirements for the degree ofMaster of Library and Information Studiesinthe faculty of graduate and postdoctoralstudies(Library and Information Studies)The University of British Columbia(Vancouver)August 2017c© Bonnie E. Paige, 2017AbstractAs the United States transitions from the Obama administration’s engage-ment with open government data to the Trump administration’s more closedinformation strategies, the future support for federal open government datais uncertain. An alternative target for open data initiatives is state-levelopen government data portals. This study provides preliminary informa-tion on state level open data, illustrating challenges faced by small, ruralstates in supporting an open data portal. The research investigates the cur-rent condition of state open data portals: whether their current form andthe laws supporting them are sufficient to support their intended use. Thisstudy also explores whether the effects of the national political climate canbe seen on state portals. This research uses a case study approach, focusingon the northern New England states: Maine, New Hampshire, and Vermont.The case studies use four main methods of investigation: content analysis todetermine the goals of the portal, consideration of the policies and contextinfluencing the portal based on the Open Data Policy Framework, invento-rying of the data based on the Open Data Barometer, and a review of savedcopies of the portals using the Internet Archive.Based on these methods, we found that these portals fall short of support-ing their stated goals. Problems with ambiguous licensing, unclear informa-tion organization, unclear project ownership, lack of support for data users,and minimal advertisement of the portal’s existence may have contributed tolow citizen engagement with the portals. Portal data is vulnerable as none ofthe states currently have laws that ensure data will be open and proactivelyiiprovided, although Vermont is considering such legislation. National politicsmay have an influence on state open data, as Maine’s portal ceased updatestwo days before the federal election.There is potential for those in the field of library and information sci-ence to contribute to state level portals through the provision of support forthe knowledge organization and information literacy aspects of the portalthat are currently lacking. This study also suggests that evaluative toolsmore specifically attuned to the state open data context would considerablystrengthen the analysis of future research.iiiLay SummaryThis research focuses on government data made available in open formats onstate websites in Maine, New Hampshire and Vermont. It details the dataavailable and the state laws related to open data, and considers whetherthe portals in their current form are able to support the goals they werecreated to address. It also looks at the influence of national politics on theseportals. The study provides preliminary information on state level open data,illustrating the challenges faced by small, rural states in supporting an opendata portal.ivPrefaceThis thesis is the original, unpublished, independent work of the author,Bonnie Paige.vTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . x1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . 32 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . 52.1 Adoption by Governments: Tensions . . . . . . . . . . . . . . 52.2 Adoption by Citizens: Barriers and Opportunities . . . . . . . 82.3 Portal Assessment Models and Frameworks . . . . . . . . . . . 92.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102.5 Expected Outcomes . . . . . . . . . . . . . . . . . . . . . . . . 113 Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123.1 Case Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 123.2 Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13vi3.3 Key Documents . . . . . . . . . . . . . . . . . . . . . . . . . . 133.4 Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . 153.5 Data Collection and Analysis . . . . . . . . . . . . . . . . . . 153.6 Challenges and Limitations . . . . . . . . . . . . . . . . . . . . 174 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204.1 Maine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204.1.1 Open Data Policy Framework . . . . . . . . . . . . . . 214.1.2 Open Data Barometer . . . . . . . . . . . . . . . . . . 234.1.3 History . . . . . . . . . . . . . . . . . . . . . . . . . . . 254.1.4 Maine Summary . . . . . . . . . . . . . . . . . . . . . 264.2 New Hampshire . . . . . . . . . . . . . . . . . . . . . . . . . . 264.2.1 Open Data Policy Framework . . . . . . . . . . . . . . 274.2.2 Open Data Barometer . . . . . . . . . . . . . . . . . . 284.2.3 History . . . . . . . . . . . . . . . . . . . . . . . . . . . 304.2.4 New Hampshire Summary . . . . . . . . . . . . . . . . 314.3 Vermont . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314.3.1 Open Data Policy Framework . . . . . . . . . . . . . . 324.3.2 Open Data Barometer . . . . . . . . . . . . . . . . . . 334.3.3 History . . . . . . . . . . . . . . . . . . . . . . . . . . . 354.3.4 Vermont Summary . . . . . . . . . . . . . . . . . . . . 354.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355 Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375.1 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . 375.1.1 Portal Design and Architecture . . . . . . . . . . . . . 385.1.2 Portals in Context . . . . . . . . . . . . . . . . . . . . 405.1.3 Data Types . . . . . . . . . . . . . . . . . . . . . . . . 435.2 Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465.3 Laws and Longevity . . . . . . . . . . . . . . . . . . . . . . . . 485.4 History and Change . . . . . . . . . . . . . . . . . . . . . . . . 50vii5.5 Frameworks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546.1 The State of State Open Data . . . . . . . . . . . . . . . . . . 546.2 Contribution of Research, Strengths and Weaknesses . . . . . 576.3 Application of Findings . . . . . . . . . . . . . . . . . . . . . . 576.4 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . 58References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Appendix A Dictionary for Open Data Motive Coding . . . . 65viiiList of TablesTable 3.1 Open Data Barometer . . . . . . . . . . . . . . . . . . . . . 16Table 3.2 Open Data Policy Framework . . . . . . . . . . . . . . . . . 18Table 4.1 Examples of datasets within the categories on data.maine.gov . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21Table 4.2 Dataset frequency for Open Data Barometer categories indata.maine.gov . . . . . . . . . . . . . . . . . . . . . . . . 24Table 4.3 Dataset frequency for Open Data Barometer categories inNew Hampshire portal . . . . . . . . . . . . . . . . . . . . . 29Table 4.4 Dataset frequency for Open Data Barometer categories indata.vermont.gov . . . . . . . . . . . . . . . . . . . . . . 34ixAcknowledgmentsThank you to Luanne Freund for your expert guidance, encouragement, andthoughtful revisions.Thank you to Lisa Nathan for providing the voice of skepticism about opendata through your comments.Thank you to Rick Kopak for helping me realize a thesis was a possible option,and to the SLAIS office staff for fielding my many emails during that process.Thank you to my parents, Lee and John Paige, for nearly three decades ofunwavering support and love.Thank you to Raj Saha, for all of the logistical, financial, and emotionalsupport that made this thesis possible.And greatest thanks go to Neil Tenzing Sage, for sitting in the SLAIS loungewhile I attended my open government data class before you could even openyour eyes.xChapter 1IntroductionThe global proliferation of information technology has led many governmentsto look into ways of incorporating technology in their governance. As anoutgrowth of the shift to e-government, many governments have chosen toadopt some form of open government, which is defined as “the opening up ofgovernment processes, proceedings, documents and data for public scrutinyand involvement” (OECD, 2017). While the term “open government” has ex-isted since the push for Freedom of Information legislation in the 1950s in theUnited States, it has taken on new importance as an endeavor for governmenttransparency through digital means. The adoption of digital technologies bygovernments increases opportunities for innovation and efficiency in govern-ment (Clarke & Francoli, 2014). Enthusiasm for the idea of open governmenthas spread throughout the international community, with 75 member nationsmaking concrete commitments towards open government through member-ship in the Open Government Partnership as of 2017 (Action plan, 2017).Open government can take shape through a number of initiatives, includ-ing releasing meeting minutes, providing e-services, and creating platformsfor deliberation (De Blasio & Selva, 2016). This thesis will focus on opengovernment as achieved through proactive release of open government data.Open government data is data collected by government organizations inthe process of their work that is made publicly available in a way that1facilitates reuse. These datasets can include information on social statis-tics, transport, meteorology, law, geography, patents and businesses (Ubaldi,2013). Open data can be used to improve public services, increase oversightof government finances, and inform citizens about the standards expected ofgovernment agencies (G. 8 Open Data Charter and Technical Annex - Gov.UK., n.d.). Open government data advocates primarily argue for its socialand economic benefits, rather than making the case for openness as a right(Ubaldi, 2013). Access to this data is usually provided through a governmentopen data portal, which allows stakeholders to access the data most relevantto their needs without restricting them to certain views of the data (Alonsoet al., 2009).In the United States, open government data was a major initiative ofthe Obama administration. President Obama’s first executive action was amemorandum on “Transparency and Open Government” which outlined hisvision for a government built on “transparency, public participation, and col-laboration” and tasked staffers with developing an action plan to implementthese goals (Obama, 2009). This was furthered by an executive order re-quiring US government data to be open by default (Obama, 2013). In theearly 2010s, open government data began to be adopted across many levelsof government, from federal agencies to municipalities. These initiatives weresupported by administrations across party lines, suggesting that open datacould transform into a shared American value.Trends in the federal government after the 2016 election, however, indicatea shift away from open data and transparency. Obama’s White House por-tal, now archived at open.obamawhitehouse.archives.gov, provided open datacollected at the White House, such as datasets on staff salaries and budget, aswell as showcasing President Obama’s commitment to open government byproviding information on United States open government initiatives and poli-cies. After the transition to the Trump administration, the open.whitehouse.gov url now redirects to whitehouse.gov/briefing-room/disclosures,2which includes a page for PDF versions of salary data and ethics waivers,as well as a contact form to request financial disclosures. Data on climatechange was removed from the Environmental Protection Agency website toalign the site with the administration’s priorities and has yet to be replaced(Mooney & Perin, 2017). The Trump administration has given little indi-cation as to what plans, if any, they may have to incorporate open data intheir governance. The current Open Government Partnership (OGP) plancommits the United States to actions through 2017, but OGP officials stillhave not heard whether the project will continue to be supported by theTrump administration (Kroll & Choma, 2017).Given the uncertainty of open government at the federal level in the U.S.,the importance of open data made available by lower levels of governmenthas increased. The G8 Open Data Charter suggests approaching open gov-ernment data in the widest sense possible, including opening local data bydefault (G. 8 Open Data Charter and Technical Annex - Gov. UK., n.d.).Open data at lower levels of government may also allow citizens to be thedrivers of the project, rather than be merely the recipients of the top-downdata release at the national level (Kassen, 2013). While there has been someresearch into open government data portals for large cities like New York andSan Francisco, there has been little investigation of state-level portals. Par-ticularly for smaller and more rural states that are not anchored by a largemetropolitan area, we argue that information made available by the statewould be the best way to facilitate the goals of transparency and innovationthrough open government.1.1 Research QuestionsThis study will examine three state open data portals to investigate how thepush towards open government data initiated by the Obama administrationand international organizations such as the Open Government Partnershiphas been interpreted ăand implemented at the state level. Research questions3include: Are state-level portals successful in supporting their stated goals?How are they used? Do policies currently enacted at the state level encouragefurther development of the portals, or is access to currently published datavulnerable? Have there been any changes in state-level portals reflectingthe change in federal administration priorities? Case studies will detail thecurrent level of adoption by state governments and determine the degree towhich they are following guidelines for implementation as outlined in theOpen Data Barometer (Open Data Barometer Global Report 4th ed , 2017).Products or publications created by citizens using the open data will alsobe included. By looking at the current citizen use of the portals, we canconsider whether the current status of the portals is sufficient to encouragereuse. Additionally, policy documents will be discussed. We will investigatewhether the current state policy framework would ensure continued accessto open data in the long term.4Chapter 2Literature ReviewWe will focus on literature discussing the adoption of open government databy governments and by citizens, as well as methods for the assessment ofportals and data policies.2.1 Adoption by Governments: TensionsClaims have been made that governments benefit from open data, as it al-lows government processes to function more efficiently and transparently,increasing legitimacy in the eyes of citizens (Ubaldi, 2013). In addition tosocial benefits like transparency, proponents argue that open data can pro-vide economic benefits to governments by encouraging innovation, as well asoperational benefits by optimizing government processes (Janssen, Charal-abidis, & Zuiderwijk, 2012). The motivations of governments opening theirdata, however, are often more complex than simply viewing open data as apublic good. The use of open data can also be a strategic move to appearopen while pursuing a neoliberal agenda of privatization (Bates, 2014).Yang, Lo, and Shiang (2015) identified the legal and policy framework asthe most important factor inhibiting open data. Difficulty navigating var-ious privacy regulations and ensuring appropriate licensing discourage gov-ernments from opening their data. They also found that useful data in non-5machine readable formats, such as articles or pictures, pose a major problemfor data publication. While the information may be valuable, it is difficult toinclude these sources within an open data information environment withoutsignificant effort to prepare them into a machine-readable form. Pressurefrom authority, organizational norms of information sharing and concerns ofliability for data misuse also work to discourage the opening of data (Yang& Wu, 2016). In local initiatives, it may be more difficult to identify datathat is non-identifying and has copyright that allows for release (Conradie &Choenni, 2014). It may be easier for local projects to begin with data thatcould be publicly observed, such as the location of public art projects, towork through the basic technical aspects of the project before dealing withthe legal issues of releasing more sensitive data.A study by Yang and Wu (2016) shows that the organizational capabilitycoming from existing staff gives governments confidence to proactively opentheir data. Particularly, Lee (2013) found that stronger working relationshipsbetween IT and program staff improved the ability to collaboratively solveissues, leading to better e-government effectiveness. The proactive openingneeds to be facilitated by adequate resources and support in order to beeffective (Yang & Wu, 2016). Perceived usefulness motivates governmentsto open their data, but governments often underestimate data’s usefulnessby failing to consider uses such as inter-agency data sharing (Yang & Wu,2016). Other governments undertaking open data initiatives can provide apositive influence, as well as inspiration as to what datasets could be shared(Yang et al., 2015).When a government has decided it has the capacity to begin an open dataproject, the project is shaped the context and attitude in which it is beingcreated. De Blasio and Selva (2016) comparison of European open data policydocuments found that the two main models of open data approach the issuefrom either an innovation- or economic-based framework. The innovation-based policies focus on transparency and public-private collaboration, while6the economic models anchor open data policies within previously existingeconomic policies. De Blasio and Selva (2016) suggest that more democraticpolicy approaches, like collaborative projects involving other stakeholdersin government processes, are beginning to be implemented in some localgovernments. This may lead to a new participatory trend in open datapolicy.A case study comparison by Dawes, Vidiasova, and Parkhimovich (2016)explores open data as an ecosystem where the government interacts withdata producers, innovators and users. In addition to providing high-qualitydata, governments can provide consultations to steer stakeholders towardsreal community needs that could be addressed with data. They also foundthat taking advantage of a broad network of existing working relationshipswithin governments and agencies was a more effective foundation for opengovernment data that relying on select activists to push the effort forward.Smaller governments may encounter additional difficulties due to a lack ofresources and few locals with data expertise. One potential strategy is tocreate regional cooperatives to share costs and expertise with neighboringgovernments (Lassinantti, Bergvall-Kåreborn, & Ståhlbröst, 2014). Thesecooperatives currently exist primarily between municipalities, counties andorganizations, such as the Western Pennsylvania Regional Data Center 1, butthe same model could be applied to state-level portals.After the initial difficulties in getting buy-in are overcome and the philos-ophy is cemented, the project can proceed to actually publishing the govern-ment data. According to Janssen and Zuiderwijk (2014), it is more effectiveto incorporate multiple different business models into an open data project,rather than simply putting raw data into portal. Applications, aggregators,and service platforms can allow the data to be more accessible to users with-out high levels of comfort with technology. These services can bring govern-ment information into users everyday life, such as a trash collection alert or1www.wprdc.org7a restaurant health code violation notification system, or help users monitorgovernment, as with a legislation tracker. Different business models can alsoencourage different types of participation and dialogue. Although it may ini-tially be simpler to release information such as financial data that is alreadyin tabular form, it also provides additional value to citizens to publish opendata such as meeting transcripts and extract, analyze and visualize the data(Steinbauer, Hiesmair, & Anderst-Kotsis, 2016). This allows citizens to seetrends government deliberation without fully reading each text, which willencourage them to access the primary source data rather than relying solelyon infomediaries.2.2 Adoption by Citizens: Barriers and OpportunitiesOnce a government has embraced open data, it is often difficult to engagepotential users. Zuiderwijk, Janssen, and Dwivedi (2015) carried out a studyof open data users based on the Unified Theory of Acceptance and Use ofTechnology (UTAUT). The team found that a belief that open data wouldhelp the user succeed in their job and data use amongst respected peers andcolleagues are the most significant predictors of the intent to use open data.Perceived difficulty of using open data and a lack of work that requires the useof open data lessened the intent to use open data. Based on these findings, anopen government data portal could increase its use by clearly communicatingand advertising the potential value of its datasets, creating a user community,keeping portal navigation clear and consistent, and encouraging schools andbusinesses to use open data as part of their regular workflow.Engagement with the public can also be improved by opening up datasetsthat are most interesting to the target groups of citizens. According to Both(2012), potential users are most interested in data about city planning, ad-ministration and the environment. While this might vary by community,identifying these high interest datasets is an important aspect of attractingusers to the project. Since many users may not know enough about avail-8able data to make informed requests, civic organizations that work with thepublic, particularly with underserved populations, could play a role advo-cating for data they identify as useful to their constituents (Ubaldi, 2013).Balancing the needs of citizens with the goals of the government agenciesopening their data may require a participatory design or community infor-matics approach to creating the portal, to ensure that all stakeholders have areal chance to influence the design process (Maruyama, Douglas, & Robert-son, 2013). Although these approaches require a large amount of time fromboth portal creators and participants, the investment may be worthwhile ifit significantly improves the portal’s efficacy and likelihood of being usefulto the community.Data literacy can be a major hurdle for would-be users of an open dataportal. One potential way to overcome this is by having an “infomediary”who can process some of the open data into forms that are more understand-able to the average citizen and also provide some support for users learningto work with data (Chan, Johnson, & Shookner, 2016).2.3 Portal Assessment Models and FrameworksFor effective adoption, open data portals need to cater to the multiple formsof democratic activity and different user groups that the portal can support,and help uncover the information need behind the question asked (Ruijer,Grimmelikhuijsen, & Meijer, 2017). A model for understanding the contextof these intersecting roles is the Democratic Activity Model of Open DataUse, developed by Ruijer et al. (2017). In this model, the open data por-tal serves as a tool that connects the separate activity streams of citizensand governments for different types of democratic actions. As described inthe model, monitorial democracy largely relies on the proactive release ofinformation by the government, and continued oversight of this informationby citizens, often facilitated by the media. Deliberative democracy requiresdialog between citizens and governments about policy issues in which citizen9opinion has some possibility of contributing to substantive change. Participa-tory democracy requires the active contribution of citizens in bettering theircommunities in collaboration with the government. These different demo-cratic processes require a context-sensitive design for an open data portal.The evaluation of open data adoption is frequently done using bench-marks. Susha, Zuiderwijk, Janssen, and Gronlund (2015) compared the opendata benchmarks proposed by several organizations, and found that bench-marks designed to measure open data in one way, such as implementation,are not necessarily adequate to measure other aspects such as impact. There-fore, a deep understanding of what aspects of an open data project shouldbe measured is key to adequately selecting benchmarks. To measure the im-plementation of open data, Susha et al. (2015) suggest using the Open DataBarometer developed by the World Wide Web Foundation. For a researcheror portal creator, understanding the types of democratic activity the portalintends to support and the measure by which success is defined is critical forthe evaluation of a portal.2.4 SummaryGovernments may adopt open data portals to appear transparent, to encour-age innovation or to optimize processes. Concerns about licensing, formatand legality, along with a lack of in-house expertise needed for the project,often discourage governments from opening their data. Governments alsooften underestimate the utility of open data unless they are given examplesof what it can achieve. Those governments that do open their data usuallytake either an economic- or innovation-based approach, but a participatoryapproach is increasing in popularity. This approach is best achieved by pro-viding support to existing networks rather than relying on activists to driveprogress. Providing applications and service platforms in addition to theraw data enables more citizens to participate, and extracting data from textsources such as meeting minutes adds additional value to the portal. Regional10cooperatives can help governments share costs and talent.Citizens are more likely to use open government data if they believe itwill help them with their work, find it easy to use, and know respectedpeers are using it. Identifying high value datasets to release, either throughconsultation with citizens and organizations representing them or through aparticipatory design process, will increase citizen use of the portal. Providingsome support for users as they learn to work with data and creating moreuser-friendly products such as visualizations from the data can help morepeople get involved.An effective portal will be designed to support monitoring of, dialog with,and participation in the government, despite the differing data needs of theseactivities. The method of evaluating an open data portal needs to be ap-propriate to both the processes it is meant to support and the stage of opendata the researcher wishes to measure.2.5 Expected OutcomesBased on the existing research literature to date, we expect that the currentcontent and organization adopted by the state governments will vary widely,but that all states could benefit from more clearly articulated goals andpolicies. Furthermore, we expect that reuse of data is low and limited to asmall group of users. We expect that the policies governing state open dataare minimal, and that further legislation would be necessary to protect thefuture of state open data. We also expect many of the internal policies onopen data publication to not be available online, despite the importance ofthese documents for transparency (G. 8 Open Data Charter and TechnicalAnnex - Gov. UK., n.d.).11Chapter 3ResearchTo consider issues in state open data portals, we have conducted case studiesof three sample states- Maine, New Hampshire, and Vermont.3.1 Case SelectionThe lack of a clear partisan stance makes these New England states interest-ing cases in the context of the change in federal administration and its shiftwith regard to open data policy. The states are fairly centrist politically.All three states currently have Republican governors, despite having votedfor the most recent Democratic presidential candidate. Maine and Vermontare also the only states represented by Independent senators. The historyand present condition of these states are similar, making it easier to drawcomparisons between them. These states are among the least populous andmost rural in the US (Annual estimates of the resident population: April 1,2010 to July 1, 2016 , n.d.). Therefore, they lack the access to human capitalthat a major city like nearby Boston attracts to an open data project. Thecases were also chosen, in part, for their geographic proximity to the author.Naturally, the very qualities that singled these states out as “neutral” alsoset them apart from many states that have urban cores and different politi-cal makeups. It is also worth noticing that these states are considerably less12diverse than the United States as a whole, with a white population of over90% vs. 77% in the US generally (Annual estimates of the resident popu-lation: April 1, 2010 to July 1, 2016 , n.d.). With these cautions in mind,the selected cases can be illustrative of some of the issues in creating andsustaining a state open data portal.3.2 DesignThis study follows a multiple-case study design, as described in (Yin, 1994).In these cases, the unit of study is a state portal. A case study approach waschosen to allow for a deeper consideration of how policy and politics affectopen data portals than would have been possible in, for example, a surveyof all state portals. The ability to combine more specific data from contentanalysis of the portals and policies with qualitative observations about theserelationships is the case study’s strength. The study was exploratory innature.The case studies focused on portals in the Northern New England statesof Maine, New Hampshire and Vermont. We detailed the types and formsof data available in each portal and any current use by citizens, as well asdiscussed the policies governing their release. Information was all gatheredthrough reviewing documents either published or available through the Free-dom of Access/Information process. We predicted that these cases wouldfollow the replication logic of a literal replication, which means that manyof the same issues are present across the case studies, suggesting underlyingsimilarities (Yin, 1994).3.3 Key DocumentsFor each research question restated below, we will rely on several key docu-ments.131. How has the push towards open government data initiated by the Obamaadministration and international organizations such as the Open Gov-ernment Partnership been interpreted and implemented at the statelevel?This question is addressed through introductory materials on the opendata portal websites. This background information provides evidenceof what open government data means to the creators of the portals.State level policies also inform this line of questioning.2. Are state-level portals successful in supporting their stated goals? Howare they used?This information is available primarily through looking at the portalsthemselves and considering what they publish, with what frequency,and to what degree it facilitates reuse. Projects using the data, or alack thereof, are also included in answering this question.3. Do policies currently enacted at the state level encourage further devel-opment of the portals, or is access to currently published data vulnera-ble?This question uses information from policy documents, and also in-cludes federal policy documents and newspaper articles to put statepolicies in context.4. Have there been any changes in state-level portals reflecting the changein federal administration priorities?Looking at previous iterations of the portals through the Internet Archivedocuments any shifts in the portals since the change in federal admin-istration.143.4 DocumentationGeneral note collection was done through a word processor document file,in which all relevant observations while looking through materials are noted.Two reading lists were kept in a citation manager, one for background lit-erature and one for items actively involved in the research such as policydocuments and portal screenshots. A spreadsheet was used for content anal-ysis to track information gathered from the portals and the policies.3.5 Data Collection and AnalysisData collection was done through a mixed-methods approach, combiningdocument analysis and content analysis (Krippendorf, 1980). The coding forthe content analysis, included as Appendix 1, assumes knowledge of socialscience research and U.S. politics.We performed several content analyses as a part of this research. First,themes in the introductory information on state open data portals were cate-gorized to look for patterns in the portal creators’ stated goals. Based on theliterature, expected major themes were economic accountability and citizeninnovation. Units in this section were syntactical, focusing on words thatoccur within the dictionary of terms for coding.Next, the content of the portals themselves was analyzed for informationcontent, format, publication date, and licensing information. We lookedthrough the units of the datasets for patterns in content type and publicationschedule and compared these portals to benchmarks available from the OpenData Barometer (Open Data Barometer Global Report 4th ed , 2017). Thesebenchmarks focus on inventorying whether different types of data are open,reusable, and updated regularly, as detailed in Figure 1. We also noted ifthe data is available from the government but not included in the portal.Coding for all other questions followed a simple strategy of 1-present, 0-absent for unambiguous reporting (Krippendorf, 1980). We used the Open15Table 3.1: Open Data BarometerDoes the data exist?Is it available online from government in any form?Is the dataset provided in machine-readable and reusable formats?Is the machine-readable and reusable data available as a whole?Is the dataset available free of charge?Is the data openly licensed?Is the dataset up to date?Is the dataset being kept regularly updated?Was it easy to find information about this dataset?Are data identifiers provided for key elements in the dataset?Data Barometer Research Handbook to ensure we evaluated the resourcesin a way that is consistent with other researchers to increase the ease ofcomparison between portals. We also, however, adjusted the questions fromthe Open Data Barometer to better suit the examination of state portals bychanging the initial question “Does the data exist?” to “Is the data availableon the portal?”. While this does decrease the ability to directly compare withother research, it is necessary to keep the scope of the thesis focused on theportals themselves rather than all state government information. To addressany misrepresentation coming from this focus on portals, the question “Is thedata available online from the government in any form?” was answered witha link to the appropriate data, whether on or off the portal. This strategyalso has the benefit of identifying data for possible future inclusion. The useof the Open Data Barometer also helped with closure, as it was easy to findpatterns in what types of information are absent from the portals. If anyprojects made use of open data from the portals, we checked for patterns inthe content of datasets that were reused.The final content analysis for this project was of the policies governingthese portals. To guide this investigation, we used the Open Data Policy16Framework developed by Zuiderwijk and Janssen (2014). The frameworkallows for comparison of the context and content of different policies, as wellas their performance indicators. The elements of the framework are listed inTable 3.2.Document analysis, or close reading of text with a goal of interpretingmeaning, was used throughout to support the content analysis, as well asbeing the main approach for the archived versions of the portal sites (Olsen,2012). Any differences between previous versions of the portals and theircurrent state were noted and compared.3.6 Challenges and LimitationsThe main challenge we experienced in the process of this research was dif-ficulty communicating with portal staff. We were never able to establishcontact with staffers knowledgable about any of the portals, which signif-icantly limited the scope of information we had hoped to access. Specificpolicies relating to issues like how data was selected for inclusion and howfrequently data was updated were not available online, and without thesecontacts in government we had to extrapolate these policies based on whatwas evident in the portal. It may be that these policy documents do notexist, but the impossibility of confirming this was a frustrating result of thelack of communication.The other major obstacle this project encountered is the changing land-scape of open data in the United States. Federal open data is in the processof changing to fit the current administration’s priorities. This may meanthat some of the context provided in this thesis becomes obsolete beforepublication.The completed research still has limitations. It is difficult to generalize theresults of case studies, as they focus on the case’s unique aspects. Althoughthe cases may provide a snapshot of certain portals and their policies, noconclusion can be drawn about state level open data in general from this17Table 3.2: Open Data Policy FrameworkCountryLevel of Government/ Mission TypeKey motivations, policy objectivesOpen data platform launchResource allocation and economic contextLegislationSocial and political context, culture in which the opening of data is institutionalizedPolicy strategy and principles for opening dataPolicy measure and instrumentsProcessing of data before publicationAmounts of open dataTypes of open dataWay of presenting dataFee charged for data accessTarget group(s) for the open dataTechnical standards and formats for open dataProvision of metadataTypes of data not publicizedTechnical support for the use of publicized dataActive encouragement of data reuse and promotion of open dataData qualityData licenseAvailability of data without application or registration and without requiring user detailStructure of relationship between information suppliers and usersUsages of publicized dataRisks of publicizing data (possible negative impacts)Benefits of publicizing data (possible positive impacts)18work. The states selected also create limitations, as the information gatheredhere may not be relevant to open data portals in more diverse and denselypopulated states.19Chapter 4ResultsIn this chapter, the results of the three cases, Maine, New Hampshire, andVermont, are presented separately. Results are compared and synthesized inthe following Discussion chapter.4.1 MaineThe Maine open data portal, available at data.maine.gov, is a site poweredby the commercial subscription-based platform Socrata, which provides apre-made layout for open data catalogs as well as tools to create charts andmaps without additional software.The site includes 38 datasets and visualizations. The site’s landing pagehighlights the categories of government, finances, recreation, and statistics,implying that these are the priority datasets of the site creators. Examples ofdatasets illustrative of the contents of these categories are included in Table4.1.The site also has a prominent link to the financial information hosted atopencheckbook.maine.gov. Although Open Checkbook is hosted separately,the importance it is given on data.maine.gov’s landing page suggests thatthe two sites are intended to work in tandem. On data.maine.gov, the onlyintroductory text is a nine word tagline within the welcome banner- “Access20Table 4.1: Examples of datasets within the categories on data.maine.govData Category Examples of datasets with categoryGovernment - Veterans’ services offices- Median household income dataFinances - State expenses 2011- State revenue 2004 2009Recreation - Maine lighthouses- Wildlife management districtsStatistics - Maine population by counties- Crime by countyto public information to facilitate transparency and knowledge”. Contentanalysis of this phrase shows the intended uses of this portal as stated are fortransparency and education. For Maine Open Checkbook, we determinedthat the first two paragraphs of the “About” page are introductory in na-ture. These paragraphs give equal attention to issues of transparency andfiscal responsibility, most simply expressed in the portal’s own explanationof the idea behind its creation, “Every Maine citizen has a right to knowhow their hard earned tax dollars are spent”. Given that this portal housesopen financial data, it is unsurprising that the introductory content focuseson transparency in government spending. This page’s introductory contentalso includes a minor theme of participation, as it “encourage(s) Mainers toshare their proposals to save money in state government”.4.1.1 Open Data Policy FrameworkIn applying the criteria outlined in Zuiderwijk and Janssen (2014) Open DataPolicy Framework, it becomes clear that while the portal is relatively userfriendly, it lacks the user support, legal framework, and depth of data to be21fully utilized. The Maine portal was created in 2014. The legal context inwhich the portal exists consists of national laws combined with the state’sFreedom of Access laws. It is clear on the Open Checkbook page that thestate’s Republican governor at the time of creation (and also at the timeof this study), Paul LePage, was in support of open financial data as apart of transparency initiatives. While the Open Checkbook is funded andrun by the Office of the State Controller, the data.maine.gov portal seemsto be generally under the auspices of the state government office InforME(Information Resources Maine), which builds state websites and applications,but whether they provide the funding and oversight for the portal or merelythe technical know-how is not stated. There is no information providedabout the strategy used in opening data, or how data is processed beforepublication.The design of the portal does seem to take user needs into account to somedegree in its presentation. The Socrata platform provides tools for informa-tion visualization within the portal, making it easier for users to understandthe data provided with minimal background or additional software. Thereare tools for filtering, which for example can display only the deer killedafter 2011 within the larger hunting dataset provided on the portal. Thereare also tools for creating calendars, maps, graphs and charts. These toolsare quite user-friendly, as options are presented as check boxes, and datathat does not make sense on a given axis is greyed out from selection. Thedata formats vary, including CSV, RSS, KML, Shapefile, and many otheravailable formats. Basic metadata such as creator, last update and source isusually included with the datasets. Based on the information about sourcesprovided, the risks of publishing data on the portal should be minimal, asthe data on the portal is generally gathered from previous publications bygovernment entities.While no information about technical support is included on the portal,there is a general InforMe help desk that might be able to assist with the por-22tal (InforMe 2014) Attempts to reach the help desk through email, however,were not successful. There is no record of any promotion of reuse of dataon the Maine portal online. Usage statistics from the portal indicate thatthree community users have used the data to create filtered views, maps, orcharts with Socrata. No applications or outside uses of the data have beenmentioned on the portal. Because it is possible for portal creators to see howdata is being used within the portal, it creates a limited feedback loop inwhich creators could provide more instances of popular datasets.4.1.2 Open Data BarometerApplying the Open Data Barometer’s evaluation criteria, which focuses onthe types of data provided as well as their licensing and currency, to thedataset records on the data.maine.gov catalog, reveals that four types ofdata are available within the portal. Map data, census data, governmentspending, and crime statistics are all represented within the portal. Themost frequent dataset type, as shown in Table 4.2, is government spendingdata.All of these datasets are provided in machine-readable, reusable, down-loadable data formats at no charge. All of the datasets are licensed as beingin the public domain, but this information was somewhat difficult to uncover.No details are given as to whether this licensing allows for modification ofthe data or requires attribution. The datasets do all include basic metadataand some include a link to the original data source. Some useful metadata,like tags and categories, are provided for some datasets but not consistently.Datasets do not seem to be up to date, besides map data which is unlikelyto need major changes with time. Census information came from the mostrecent full census, 2010, but does not include any population statistical in-formation gathered since then, such as the American Community Survey(American Community Survey , n.d.). All other datasets used data collectedno later than 2013. The datasets frequently show the same created and last23Table 4.2: Dataset frequency for Open Data Barometer categories indata.maine.govData Category # of Datasets in PortalMap 1Census 2Government spending 3Crime Statistics 1updated dates, meaning that they have not been updated since being addedto the portal. Additionally, many of these datasets share a created or lastupdated date around April 16, 2015, which shows that there was an initialpush to populate the site around this time. The most recent additions of newdatasets was in February 2016, over a year ago at the time of research. Noupdates were made from November 2016 to the date of this research (July2017).The portal also includes a variety of datasets that are not represented bythe Open Data Barometer’s metrics. Information about hunting and light-houses falls outside of the government focus of the Open Data Barometer,but is included in the portal as a part of the recreation focus. The locationsof Veteran’s Service offices add human services information to the portal,while datasets about railroads and natural gas pipelines add infrastructuredata. It is also important to note what information inventoried by the OpenData Barometer was not present in the portal. Some of these data types,like education performance data, are available in open data formats on theirdepartment’s website1. Others, like budget data, are only available throughnon-open formats, like PDFs or a website search engine. Still others, likehealth performance data, do not have any data easily available online forpublic review. Even among those types of data present in the portal, therewas often a more complete data repository elsewhere online. For example,1http://www.maine.gov/doe/data/index.html24while the portal had some basic map data, the Office of GIS website hasmuch more detailed datasets available for download2.4.1.3 HistoryData from the Internet Archive provides an overview of the development ofthe portal over time. Examining the 19 saved copies of data.maine.gov inthe Internet Archive’s Wayback Machine3, it seems that the Maine portal hasnot undergone major changes since its initial creation. Between the initialupload of 20 datasets in 2015 and the next archived version of the portalin 2016, 20 datasets were added. Some datasets, such as a heat map ofmoose-involved crashes, were taken down. Some minor changes in designalso appear to have been made by Socrata for the data catalog presentation.Besides these adjustments to the datasets and layout, the page has remainedunchanged, including the items highlighted on the front page and majordesign elements such as the header photo.In addition to looking at the historical version of data.maine.gov, wealso found reference to a previous open data portal, Maine DataShare4.This portal has not been updated since 2010, and most links redirect todata.maine.gov, suggesting that data.maine.gov was intended as replace-ment for DataShare. However, the site is still in existence, and its contentsinclude a variety of datasets that are not replicated on data.maine.gov,such as the locations of boat launches and data on families unable to affordhousing. Whether this data was intentionally left off of the new site or justnever was migrated is unclear. Given the broken image links and seven yearswithout updating, it seems likely that this site has been abandoned, but noinformation is included encouraging visitors to check data.maine.gov or ex-plaining that it is no longer being maintained.2http://www.maine.gov/megis/catalog/3web.archive.org4http://www.maine.gov/cgi-bin/data/index.pl254.1.4 Maine SummaryWhile Maine’s portal shows some intention of user-friendly design, the por-tal is missing some elements necessary to support its goals of transparencyand education. While some data of special interest to Maine residents andvisitors is included, a great deal of important datasets are not. Technicalsupport and education is lacking, as are clear details of licensing and legalframework encouraging open data. Reuse of portal data has been limited.Maine previously had a different open data portal, which was abandonedwithout fully migrating the data. The current portal was not updated forover 8 months.4.2 New HampshireNew Hampshire’s open data has not been worked into a formal portal likeMaine and Vermont’s. Instead, a directory provides links to 117 open datasetsmade available by 17 different departments5.The introductory information available on this site is in a PDF on themain open source page: “DoIT Open Data Standards Policy”. The secondsection of the policy, titled “Open Standards/Open Data Formats”, is themost applicable to open data. The content analysis of this section showsa focus on fiscal responsibility with an additional theme of transparency.According to this document, the driving motivation for New Hampshire’sadoption of open data is to use a “cost-effective, build once, use many times’approach” to data, as well as to “facilitate the interoperability of State sys-tems” and “support transparency in government” (Open source software anddata formats policy , 2012).5https://www.nh.gov/doit/open-source/data-sets.htm264.2.1 Open Data Policy FrameworkAnalyzing the New Hampshire portal through the Open Data Policy Frame-work (Zuiderwijk & Janssen, 2014) shows that although New Hampshire hasa strong policy commitment to open data formats, this does not necessarilytranslate into a strong open data portal because of the choice not to employuser-centered design. No date is available for the platform launch, as NewHampshire’s data is not on a purpose-built portal, but the website copyrightdate is 2015. The state legal context for the open data is robust, as NewHampshire has a law, RSA 21-R:10-13, which requires that all New Hamp-shire data comply to open standards “unless specific project requirements, orexcessive cost, preclude use of an open data format” (RSA 21-R:10-13 , 2012).This law also has a requirement that all existing data not available in openformat be reviewed every four years to assess the feasibility of converting itto an open format. The bill was sponsored by former State RepresentativeSeth Cohn, a member of the libertarian Free State movement (Cohn, 2012).This law adds an additional layer to national laws and state Right-to-Knowlaws.Since the data is hosted on the sites of individual agencies, and oversightof the data collection and publication comes from these agencies, they es-sentially provide the funding for the project as well, with the Departmentof Information Technology providing the minimal funding to host the singlepage listing the data. Because state data is required to be in open formats, itis already in an appropriate format to be linked to on an open data directorywithout additional preparation. Acceptable types of open data, such as CSV,SHP, and XML, are specifically defined by the IT department (Open sourceresources and open data formats , 2015).The method of presenting the data does not take a user-centered ap-proach, but rather reflects a more traditional government-centric approach.Data is presented by the government department that hosts it, rather thanby topic or any other category that is more user needs-based. The list format27aids users in identifying and accessing available datasets, but does not addvalue for users with respect to using the data, as a platform with capabilitiesfor visualization would. The directory itself also does not provide any meta-data aside from the title. The metadata standards adhered to by the con-tributing departments vary, so no expectation of metadata can be assumedin this directory. This same problem occurs with the data license: the lackof any overarching licensing information on the directory makes it unclearwhat is available for reuse and what might have more restrictive licensingconditions, such as prohibiting commercial reuse. Even when a user followsthe link to a department’s landing page of datasets available for download, li-censing information is not clearly stated and the information provided rarelygoes beyond the title of the dataset. The model for the directory is decen-tralized, with individual departments deciding what information to provideand how. For users, this is a low-engagement, informational model withoutthe opportunity to give feedback, share, or be aware of previous uses of thedata. There is no known support for working with the data, and the onlycontact information provided on the directory is the email of the site’s web-master. Individual department sites also show no sign of providing supportfor data usage or contact information for staff who can address data-relatedquestions. There is no apparent promotion of the open data on this site orits active reuse. There is no evidence that this data has been used, nor arethere any applications or products built with it that can be discovered.4.2.2 Open Data BarometerSince no data is actually hosted by New Hampshire’s open data directory,for the purposes of this thesis any dataset the page links to counts as “inthe portal”. Inventorying the datasets using the criteria of 15 dataset typesfrom the Open Data Barometer shows that the datasets cover five differenttopics. Links are provided for datasets on census data, budget, governmentspending, education performance and environmental statistics. As shown in28Table 4.3: Dataset frequency for Open Data Barometer categories inNew Hampshire portalData Category # of Datasets in PortalCensus 25Budget 1Government spending 3Education Performance 9Environmental Statistics 14Table 4.3, census data is the most frequently occurring type of dataset, witha significant number of environmental statistics and education performancedatasets as well.All of these datasets are provided in machine-readable, reusable, down-loadable data formats at no charge. No clear licensing information is in-cluded, neither on the portal itself nor on the pages hosting each specificdataset. All datasets appear to be up to date and regularly updated. Sincethe datasets are maintained by the individual departments, this updatingschedule would depend on the information practices of the departmentsrather than on the creators of the open data site. The directory does notinclude explanatory information about the datasets.Additional datasets that do not fall under the Open Data Barometer cat-egories highlighted on New Hampshire’s open data page could support bothgovernmental and personal business. Bank charters, employment data andprojections, and inspections data could all be useful to the business commu-nity. Polling places and traffic information are useful to citizens. Interlibraryloan statistics and plant species round out the data provided. Data notlinked to on the page but easily available from the government includes GISdata, election results, and government contracts. Additional datasets areavailable through a search feature, such as environmental data. Some data,while nominally available to access, is extremely difficult to get, such as deeds29information that requires detailed troubleshooting tips and additional soft-ware downloads to be accessed. Other datasets, like the company register,require a log-in for access. No New Hampshire data sources could be foundfor international trade statistics, public transit, or crime.4.2.3 HistoryThe Internet Archive has limited information on the history of the NewHampshire open data directory. A check of the single saved copy of thesite on the Internet Archive shows that the New Hampshire site has largelyremained the same over time. The one archived version of the site is fromSeptember 2015. The only difference between this saved page and the currentversion, as of July 2017, is that one department and dataset-the InsuranceDepartment and New Hampshire Population Under 65 By Zip Code- hasbeen removed from the current page. Aside from this, the page remainsunchanged in content or layout, although Internet Archive saved copies ofthe data the directory links to on the agency pages suggest that the dataitself has sometimes been updated in that time.While the open data directory on the Department of Information Tech-nology website is the most current and complete source for New Hampshireopen data, it worth noting that the list of state open data portals com-piled by the federal government at www.data.gov/open-gov does not linkto this site. Instead, it links to nhopengovt.org, a now-defunct open dataportal for government financial information run by a think tank. The lastarchived version of the page that was actually related to open data was inApril 2016. The site, run by the Josiah Bartlett Center for Public Policy,housed a database of 4.5 million state financial transactions (NHOpenGovtlets you Google your government , n.d.). Although the database is no longeravailable, the datasets seemed to be grouped into state spending, payroll,and pension data. There is no indication of whether this data was moved toa new location online or has been lost. The nhopengovt.org web address,30from October 2016 to present, now sells motor fleet insurance.4.2.4 New Hampshire SummaryNew Hampshire has chosen to host a directory of open data on state agencywebsites, rather than a full open data portal, in support of its goals of fiscalresponsibility and transparency. Due to a law that requires all governmentdata to be made available in open formats, there is a great deal of open dataavailable from New Hampshire. The decentralized model means that thingslike metadata and licensing are inconsistent. The directory format is notuser-friendly, as it does not categorize data in a way the facilitates discoveryor offer any support for working with the data.4.3 VermontThe Vermont open data portal, available at data.vermont.gov, includes 160datasets and visualizations hosted by Socrata.The site’s landing page highlights select datasets, as well as providing alist of 12 categories of data. The portal itself does not include any explana-tory text, but a page on the Department of Information and Innovation sitedoes provide this background information (Open Data Portal , 2017). Thecontent analysis of the “What is this Service?” and “What is Included?” sec-tions required contingency analysis, as many words included in the codingdictionary were included in negative phrases (eg. no cost). With the contin-gency analysis controlling for such statements, the portal appears to focuson participation and education. The initial description of the purpose is to“publish and store publicly accessible data,” which implies providing accessas an end rather than a means. However, further discussion promotes us-ing visualizations to “share the data in a meaningful way and in context,”encouraging users to actively create new information products and use thetools to communicate information.314.3.1 Open Data Policy FrameworkUsing the Zuiderwijk and Janssen (2014) Open Data Policy Framework, ananalysis of Vermont’s portal shows that although the portal does utilise user-friendly design, the policy surrounding open data is still developing and thesupport system for users is unclear. Vermont’s portal was created in 2014.Although there was an attempt to introduce legislation to make Vermont an“open data democracy state”, this legislation has not been adopted (OpenGovernment Data Legislation, 2017). This means that currently, the portalis supported just by national laws and the state Public Records law. Inaddition to this legal context, there was a public interest around open data,with a summit that occurred in 2013 (Vermont Open Data Summit , 2017).Vermont’s major city, Burlington, also has its own open data portal6 whichfurther demonstrates interest in open data within the state. Vermont’s portalis run by the state’s Department of Information and Innovation, which statesthat the portal has no ongoing maintenance costs (Open Data Portal , 2017).The portal follows an opt-in strategy for government departments, which canchoose to include any non-private information in the portal. Those uploadingthe data are responsible for ensuring sensitive data is not shared. Since thedata is opt-in from government agencies, there should be very little risk topublishing data on the portal.The portal is somewhat user needs driven, with Socrata’s built-in visu-alization tools to encourage reuse and frequent use of tagging and catego-rization to aid discovery. The portal also makes full use of its landing page,highlighting several datasets that might be of interest to users. All data pro-vided within the portal is in open formats, but the portal also links to datain closed formats such as PDFs. Basic metadata is included with all datasets,although not all include links to the original data source. Training and sup-port are offered by the Department of Information and Innovation, but it is6data.burlingtonvt.gov32unclear if this is available for all users or only government employees wishingto add datasets to the portal. The portal forms a limited feedback loop, asportal creators can see what datasets are popular and reused, and commu-nity users can contribute through creating visualizations. The licensing forthe datasets varies, with most sets using either the Open Database Licenseor Public Domain, but many datasets provide no licensing information at all.There are no clear instances of the portal being promoted or the reuse of theopen data being encouraged. Usage statistics show that the portal to datehas 14 community users who have contributed 18 filtered views, maps andcharts. No applications or other projects created using this data have beenrecorded.4.3.2 Open Data BarometerThe application of Open Data Barometer criteria to the datasets includedin the data.vermont.gov catalog shows the portal includes seven types ofdata. These datasets include map data, census data, budget, health sectorperformance, crime, environmental statistics, and public contracts. As shownin Table 4.4, census data is the most common dataset type on the Vermontportal, followed by crime statistics.All of these datasets were provided in machine-readable, reusable, down-loadable data formats at no charge. Data identifiers are present for alldatasets. Licensing information varied by dataset. Some, such as healthperformance data, had an unspecified license. Those datasets that did havea license listed used Open Data Commons Open Database License. Thislicense allows the user to share, create with, and adapt the data, as long asthe data is attributed, also provided under the same license, and providedin a open format (ODC Open Database License (ODbL) summary., n.d.).Most datasets appear to be up to date, but some datasets do not show signsof being regularly updated. Census data includes up to 2014, and has notbeen updated since its original addition to the portal in 2015. Health data33Table 4.4: Dataset frequency for Open Data Barometer categories indata.vermont.govData Category # of Datasets in PortalMap 1Census 9Budget 1Health Sector Perf. 1Crime Statistics 6Environmental Statistics 3Public Contracts 3also has not been updated since being added to the portal. Although en-vironmental data has been updated after its addition to the portal, it hasnot been updated for 10 months. Information about the datasets is usuallyeasy to find, with links provided, but for some data, like census and budget,the information is lacking. New datasets have been added and updates havebeen made during the time of writing, suggesting that the portal is still inactive use and development.Additional datasets made available by the Vermont portal include fuelprices, library locations, trade licenses, traffic and accident reports, pub-lic records requests, infrastructure, childcare providers, baby names, SNAPretailers, energy use, hospital spending, public works requests, house priceindex, police incidents, and hazardous waste sites. Some of these data topicsare recent additions to the portal, leading to the possibility that additionaldata topics could be expected in the future. The portal did not include sev-eral types of data that were available from the government but not in opendata form, such as election data, which is available through a search functionon a government website, and transit data, which is available in a PDF. Tradedata did not seem to be available from the Vermont government. Notablyabsent from both the portal and the Internet at large is data on government34spending.4.3.3 HistoryExamining the 13 iterations of data.vermont.gov on the Internet Archive,there have been some changes in this portal. The portal is still being up-dated with new datasets. The total number of datasets and visualizationsincreased from 116 in August 2016 to 143 in November 2016 to 160 in July2017. Frequently changing data such as traffic stops are kept updated to themost recent year. One major subtraction from archived versions of the site isthe removal of a link to the Governor’s Dashboard, a page providing progressreports on various issues from the office of the governor. The link was re-moved from data.vermont.gov some time between August and November2016. That site itself was taken down some time between December 2016and April 2017 after being unavailable in August 2016 due to a redesign.4.3.4 Vermont SummaryVermont’s portal is intended to support participation and education. Theportal includes tools to support visualizations, and has fairly complete tag-ging and categorization of datasets. Vermont introduced legislation in 2015for proactive release of open data, but it has not been adopted as of yet. Newand updated datasets continue to be added in the portal.4.4 SummaryThis chapter presented the results of three case studies of the provisionof state-level open data. These case studies included three analytical ap-proaches: considering the policy and design aspects of the portals using theZuiderwijk and Janssen (2014) Open Data Policy Framework, categorizingthe data presented following the criteria of the Open Data Barometer, and ex-35ploring the history of changes made to the portal using the Internet Archive.These cases demonstrate a range of responses to the trend of open data andopen government in the United States. Maine’s case shows initial investmentin open data, followed by stalled growth and engagement, possibly resultingfrom a lack of clear oversight of the portal. New Hampshire’s case shows a dif-ferent approach, in which providing open format data is a legally-mandatedpriority, but moving beyond supplying data into encouraging citizen engage-ment with the data does not seem to be a goal of the directory. Analysis ofthe Vermont case shows the most consistent interest in engaging citizens inopen data, with updates to the portal continuing throughout the period ofstudy, but the portal still has issues related to oversight and consistency.36Chapter 5DiscussionsThis chapter compares the results of the three cases and relates the insightsgained to the initial research questions and existing body of open data lit-erature. The initial section will discuss the current implementation of theportals and whether this supports the portal’s stated goals as determined bythe content analysis. Next will be a discussion of how the portals are used bycitizens. Following this, we will consider the laws governing the open dataportals and whether these are sufficient to guarantee the longevity of theportal. We will then discuss the changes to the portal throughout time andwhether these show the influence of political changes on the federal level.Finally, we will evaluate the frameworks and criteria used in this researchand consider their applicability to the state open data portal context.5.1 ImplementationHow has the push towards open government data initiated by the Obamaadministration and international organizations such as the Open GovernmentPartnership been interpreted and implemented at the state level? Are state-level portals successful in supporting their stated goals?All of the northern New England open data portals were created withinroughly the same time frame, 2014 or 2015. This timing is likely linked to37Obama’s 2013 order making federal data open by default (Obama, 2013).These portals were created at a time of momentum for the open data move-ment. Although the states were developing these portals within the samecontext, their implementation of open data ideas varies.5.1.1 Portal Design and ArchitectureThe portals represent three different models for presenting open data in-formation. Vermont’s portal is most in line with the strategy taken at thefederal level, with all data presented through a single portal with a consis-tent user interface. Maine’s portal is similar, but the prominence given tothe separately-hosted Open Checkbook site leads to what is essentially a dualportal. The financial data on Open Checkbook is an important supplementto data.maine.gov, but it is a separate site with a completely different userinterface. The reasons that Maine chose to follow this dual structure areunclear, but given that, according to the Internet Archive, Open Checkbookwas created a year before data.maine.gov, we would speculate that the de-velopment process was going on concurrently in two separate governmentoffices, and that merging the products was deemed too challenging. NewHampshire’s list of datasets is the simplest way to present open data infor-mation. All of the data is centrally accessible, and the portal creators do notneed to perform ongoing work to maintain the datasets. This model, how-ever, does not provide the same consistency of formatting and navigation ashaving all data uploaded in the same portal. It also lacks the additional fea-tures that a platform provides. New Hampshire’s minimal approach signalsthat this project is largely to fulfil the letter of the law, and the directoryshows the least evidence of interest in centralization or collaboration acrossdepartments out of the projects studied. Vermont’s seeming enthusiasm foropen data translates well into a successful portal. Maine is between the twoextremes, with their choice to use Socrata signaling their interest in user par-ticipation in working with open data but their organizational choices making38it difficult for users to succeed.Of the portal design and architecture choices, Vermont’s seems the mosteffective at supporting user needs, with a design that supports monitorialdemocracy through frequently updated datasets and a minor degree of par-ticipatory democracy by enabling the reuse of data (Ruijer et al., 2017). TheSocrata platform used by the Maine and Vermont open data portals doesprovide opportunities for users to filter data and create visualizations withinthe portal, but does not provide additional ways of using data for those lesscomfortable with technology (Janssen & Zuiderwijk, 2014). The method oforganizing information Maine employs does not cater to users without a pre-existing knowledge of the extent of the data. The application of categoriesand tags is inconsistent, which, combined with the choice to have some in-formation on a separate portal, decreases the likelihood that a user will finddatasets of interest. New Hampshire’s page does not have the added valueof visualizations, and its design does not seem to reflect any user orienta-tion. Information is presented by department, rather than by a method oforganization that has more meaning to the user. There are no keywords,subject categories, or search bars included in this portal, which essentiallymeans that in order to find a dataset, a user would need to already knowwhat department would have collected it or read through the entire page.Vermont includes tags and categories for nearly every dataset, making itsimple for users to find data through several different pathways. The infor-mation architecture within these portals can make information easier to find,as in Vermont’s case, or add additional confusion. Providing services basedon categories useful to the user, such as life events, improves the effective-ness of an e-government service, but requires a reframing of government asone collaborative entity rather than siloed departments (Vintar, Kunstelj, &Leben, 2002). The services examined in this study do not seem to have em-braced this change of vision. This is clearest in the case of New Hampshire’sdepartment-centered directory but even in the case of Vermont, categoriza-39tion provides some pathway for users but does not make explicitly clear howthe data could be applicable to users.The front pages of the portals also show varying strategies towards opendata. Although New Hampshire does not have such a landing page beforeits data catalog, Vermont and Maine both have datasets highlighted visuallyat their main URL before presenting the full data catalog. Maine focuses oncontent areas (Government, Finance, Recreation and Statistics) and contenttypes (Datasets, Charts, Maps and Open Checkbook). Vermont instead fea-tures individual datasets, with a menu of categories to the right. One majoradvantage of Vermont’s approach is that as additional categories of data areadded to the portal, they can be added to the category menu without requir-ing a redesign of the homepage. The static nature of the selected datasetson the homepage, however, is a missed opportunity, as only those datasetshave ever been able to benefit from the serendipitous discovery that resultsfrom being featured on a landing page. While the content areas of Maine’slanding page are useful, the content types do not seem natural categories fora user and raise questions as to why the portal creators would want to high-light the types of information products rather than the information. Thismight be due to a disconnect between the agencies producing the data andthe employees carrying out the project.5.1.2 Portals in ContextThe government entities responsible for the portal are often not clearly iden-tified. As might be expected, the state IT offices play an important rolein the portals. The New Hampshire Department of Information Technologyhosts the open datasets page, although since they do not host or overseeany of the individual datasets, this was primarily a one-time investment ofeffort to compile the links. The Vermont Department of Information and In-novation oversees data.vermont.gov, with other governments departmentsbeing able to request accounts to add datasets. While the accompanying40documentation for the portal stresses that there are no ongoing costs associ-ated with the portal, we believe they mean to an agency choosing to includetheir data, and that the Department of Information and Innovation is pay-ing for the platform and any administrative staff time. Maine’s portals haveboth the clearest and the most obscure ownership. Open Checkbook statesclearly that it is run and funded by the Office of the State Controller’s trans-parency program and provides contact information for the portal staff. Incontrast, data.maine.gov provides no information about who is responsiblefor the portal. The contact email is for the creative team at InforME, theoffice in charge of Maine’s websites and e-government services, but emailsto this account were never returned. While it makes sense that the portalservices were all created by IT teams, keeping the portals as an IT projectmay lessen the portal’s exposure for both citizens and government workers,as IT is not the most visible component of government. A partnership withthe IT department and other government programs may be needed, as Lee(2013) found that e-government services are most successful when these in-terdepartmental working relationships are strong. A lack of clear ownershipand departmental focus contributes to many of the issues on the portal suchas unclear procedures.The processes the departments use for selecting and preparing the datafor inclusion are not clearly stated anywhere. The closest thing to such astatement is Vermont’s reiteration that any data added by an outside de-partment needs to be checked thoroughly for any identifying information.New Hampshire is not required to do a great deal of preparation at the stageof inclusion, as all of their state data is in open formats by default. Data onall of the portals seems to largely have been uploaded based on conveniencefor the portal creator, but more thought may have gone into the selectionthan is initially apparent. None of the portals seem to meet Kassen (2013)goal of having the release of data driven by user input rather than a top-down method. Making the criteria and process for adding open data on41these portals public would improve transparency and the effectiveness of theportals.None of the portals publish any sort of updating schedule, so the updatingpractices must be gathered from what has occurred on the portals to date.New Hampshire is the most hands-off with its update strategy, with all of theresponsibility for keeping data up to date borne by the outside departments.Maine does not seem to have a plan for frequent updating. Many of thedatasets have never been updated since they were added to the portal, andeight months have passed since the last update to any dataset. Given thelength of time since the last update and the minimal development since theportal’s initial creation period, it seems likely that the project is no longerbeing pursued. Vermont has the most consistent updating of the portals, butit is difficult to determine whether this follows a set schedule, or is up to thediscretion of the dataset owners or portal managers. Making the updatingschedule clearer would help users ensure they are using the most up to datedata.Based on the content analysis, transparency seems to be the most com-mon stated aim of these portals, followed by education, fiscal responsibility,and participation. The Maine portal’s aim to support “transparency andknowledge” is not clearly evident in the portal, as there is little support toaid a novice user in building knowledge and much of the information thatwould be necessary to transparency is missing. New Hampshire also fallsshort on its goal of transparency, although it is more successful in fulfillingits aims of being cost-effective and interoperable. While Vermont does makesome visible attempt to “share the data in a meaningful way,” this could befurther developed by providing more background information, as well as bet-ter tools to support understanding. Those portals that include education asa part of their mission do have more opportunities to do visualizations withintheir portals, perhaps supporting these educational goals. Participation inthe portal seems to be supported only in a narrow interpretation of creating42these visualizations, rather than a more robust form in which dialog wouldbe encouraged within the portal. As the portals seem to be struggling to up-hold their basic transparency goals, a collaborative approach to the portal,as described in De Blasio and Selva (2016), seems like an unlikely possibilityfor these portals.5.1.3 Data TypesThe types of data represented within the portal do show some pattern. Allof the portals provide access to census data as their first- or second-mostfrequent category of dataset. This data was probably added to facilitate thecomparison of other datasets to population statistics. Maine and Vermontboth have some map data represented, although both states have dedicatedGIS portals as well. This information is also likely to have been uploadedprimarily in support of other datasets, enabling users to overlay data ontostate maps. Maine and Vermont both also provide access to crime data. Thiscould partially be because these states have the two lowest crime rates in thenation, leading this to be a dataset they would want to promote (Violentcrime rate by state, 2017). Maine and New Hampshire both publish dataon government spending on their portals, not including all of the additionalfinancial information provided on Open Checkbook. New Hampshire’s intro-ductory information, as well as the information on Maine’s Open Checkbook,argues for open data on economic grounds, which may explain the emphasison these datasets. New Hampshire and Vermont both provide a detailed bud-get and environmental statistics. Additionally, there were some categories ofdata that only one of the selected portals published: New Hampshire’s ed-ucation performance data and Vermont’s public contracts and health sectorperformance.Several categories of data did were not provided by any portal. Some ofthese types of data, like international trade data and public transit tables,aren’t particularly relevant to small, rural states. Other types of data, such43as legislation, might not have been included due to the technical challengesthat would occur when presenting all-text data in these portals. Althoughthe confines of the portal could support a dataset with the numbers and titlesof bills and links to the full text, the text analysis described by Steinbaueret al. (2016) would likely be beyond the scope of possibility for these por-tals. Land ownership data and company registers may have been kept offthe portal because of privacy concerns, but as the information is availablethrough government-provided searches elsewhere, it may just have not beenviewed as a priority for upload. Election results, which are available for allstates, and already available in open formats for download for New Hamp-shire and Maine, seem to have also just not made it onto the portal at thispoint despite the ease of adding these datasets.All of the portals also contained datasets that the Open Data Barometerhad not included in its evaluative criteria. Maine included datasets abouthunting and wildlife management that fit its reputation as a hub for out-door sporting, as well as information about veteran’s services to cater tothe high percentage of Maine residents who are veterans (Chokshi, 2014).New Hampshire’s additional datasets focus more on its economic interests,presenting bank charters, employment projections, and inspections, but alsoinclude other datasets that could prove useful to citizens such as traffic dataand the location of polling places, plus information for cultural enrichment,such as interlibrary loan statistics and data on native plants. As Vermont’sportal has the most datasets of the states studied, it is unsurprising that italso has the most varied additional datasets, of which we will only highlighta few. A major focus of this portal that could be replicated on other portalsis licensing information for tradespersons and child care providers. Thesedatasets are frequently viewed and frequently updated, showing that theyare useful to the community. Other popular datasets show the locations ofcommunity resources such as libraries and retailers who accept SupplementalNutrition Assistance Program (SNAP) payments. A dataset detailing pub-44lic records requests could be useful for transparency, but does not seem tobeing updated with regularity. Some datasets seem as if they were added toenable the portal creators to troubleshoot early on, but were never deleted,such as the Social Security name data. None of the datasets provided by thestates that were not included in the Open Data Barometer were repeatedacross portals, which implies that these datasets are reflective of the states’unique characters. Overall, the data included seemed to follow the patternidentified by Yang and Wu (2016) in which datasets are added based ontheir perceived usefulness, but portal owners fail to recognize a great deal ofpotentially useful data.The datasets varied with regard to providing metadata and appropriatelylinking to source materials. On the Maine and Vermont portals, the Socratainterface automatically generates some metadata, such as the date the portalwas last updated, but metadata such as tags and keywords are often not pro-vided by the dataset uploader. The dataset owners on these portals also oftenneglect to link to the original source of the data, making it difficult to ver-ify information. New Hampshire does not provide any metadata within theportal beyond title. The individual datasets, however, sometimes includeuseful metadata depending on the department. The lack of metadata formany datasets may be because the information is still hosted by the agencyproviding that information- they may see less need to formally cite their ownagency as the source. In all of the portals, inconsistent metadata can beconfusing to users and make it unnecessarily difficult to find required infor-mation. The lack of consistent metadata is unsurprising given the investmentof time and money that would be required for high quality documentationand indexing. Without this investment, however, it is unlikely that the portalwill be useable for the average citizen. There is potential for state libraries tocontribute their expertise to open data projects by taking over the indexingand metadata provision for the datasets.The current datasets in their existing form are inadequate on all the por-45tals to support the innovation and optimized government processes that arepromoted by open data advocates as reason for governments to create a por-tal (Janssen et al., 2012). While the portals may express intentions to createtransparency, in practice they do not provide the necessary support to makethis a reality.5.2 UseHow are they used?It is difficult to tell to what degree the data from the portals is used. TheNew Hampshire portal does not provide any usage information, so there is noway to determine how frequently it is accessed. For the Maine and Vermontportals, dataset views and downloads can give some window into how muchthe information is being accessed. Vermont’s most viewed item, a filteredview of fuel prices, was accessed 10, 918 times and downloaded 82 times.Maine’s most viewed item, a map of Maine lighthouses, was accessed 7,555times and downloaded 59 times. This means that, at least at the passive levelof looking at data, Vermont’s portal is getting more use. Given the topics ofthe most commonly accessed datasets, this may be due to Vermont providinginformation that is more useful in daily life. Vermont’s portal also has morecommunity-created products in the portal. Vermont has 14 contributingusers who created 18 filtered views, charts, or maps. Maine only has threecommunity users who have created six information products. There is noevidence of an interactive user community on either portal, as the platformdoes not have features that support interaction like message boards, or evenan ability to comment on datasets. Adding in some collaborative featuresmight increase engagement.Additional factors inhibit the portal’s potential for reuse. It is difficultfor users to determine what kind of technical support is provided for theportals. It is possible that state information technology help desks, like46InforME and the Department of Information and Innovation, would providesome technical support to users, but it is unclear whether they are onlyavailable to help government officials add datasets or if their services extendto public use. Even if public support is a part of their mandate, in practicethey are unresponsive to inquiries. The Socrata platform does provide sometutorials to support users in the basics of working with data, but the link tothese tutorials is not featured prominently on either the Maine or Vermontportals, and is instead hidden in the filters tab of the dataset viewer. NewHampshire does not seem to provide any support for their data, and theonly contact is the webmaster of the site. Without a clear pathway for aportal user seeking support, only the most technologically confident are likelyto be able to contribute to the portal. A potential solution to this issue ispartnering with state libraries to provide this support and training, as well ascreate information products understandable to those with lower data literacy(Chan et al., 2016). Librarians with a background in data stewardship andinformation literacy instruction could assist users of the portal without theneed for extensive retraining of existing portal technical staff.A lack of consistent licensing information is also problematic for datareuse on all portals. According to Ubaldi (2013), licenses allowing all reuse,including commercial reuse, are essential to realizing the benefits of open gov-ernment. Ambiguous licensing will lessen participation in the portals. NewHampshire’s portal does not provide licensing information on the page, andthe licensing information is inconsistently included with the datasets them-selves. The Socrata portal has a space for licensing information, but bothMaine and Vermont leave the license unspecified for many of the datasets.For those datasets that do specify license, many in both the Maine and Ver-mont portals are licensed Public Domain, with some additional datasets forVermont using the Open Database license. Without this information clearlystated, users do not know the terms to which they need to adhere whenreusing data.47A common problem to all three portals is a lack of advertising and com-munication about the portals. It is difficult to serendipitously discover theportal through looking at other government websites, so most traffic wouldhave to come through users specifically searching out open data. Commonstrategies for promoting open data, such as hackathons or app creation con-tests, do not seem to have been tried in relation to these portals. It alsodoesn’t seem like there has been any push to draw attention to the portalsthrough social media. As a result, the user interaction with the portals hasbeen minimal, with no recorded instances of the data being reused to makeapplications or projects. If the portal creators undertook efforts to promotethe portal, and then publicised the resulting applications or projects thatused the data, it might bring more users into the portal and create a morevibrant culture of use. Since perceived usefulness for work is the main moti-vator for open data users, portal owners should focus on making the portal’sdata a part of users daily workflow, starting by finding ways to use it forinternal operations in government (Zuiderwijk et al., 2015).5.3 Laws and LongevityDo policies currently enacted at the state level encourage further developmentof the portals, or is access to currently published data vulnerable?Given Yang et al. (2015) finding that the largest factor inhibiting opendata release is difficulty navigating laws, it is unsurprising that the legalfoundations of these less-developed portals do not provide clear legal guidancefor an open data project. Of the three states, New Hampshire is the onlystate with laws specifically relating to open data. The effectiveness of RSA21-R:10-13 2012 is evident since all datasets included on the portal are in openformats. Even beyond the portal, government provided data upholds the lawand provides an open format option, even if they also provide informationin a formatted document such as a PDF. New Hampshire, however, does48not have any laws that push for data to be published online in the firstplace. While New Hampshire’s Right to Know laws do guarantee accessto most governmental records and electronic communications, there is norequirement of proactive release (Right to Know Law , 2017). This meansthat while government information may be open, it also may not be availableunless specifically requested.Although Vermont does not currently have any legislation to support opendata or proactive data release beyond the state’s Public Records law, “AnAct relating to creating an Office of Public Policy, ” which requires data bemade available for analysis by the public by 2025, is still at committee (Anact relating to the creating an Office of Public Policy, H. 435 , 2015). Thisbill requires that the Office of Public policy it creates will “establish the Stateof Vermont as an open data democracy state whereby the Office shall redactany personal identifying information from data the State collects and shallorganize and share the State’s data with the public online in a manner thatenables the public to perform analyses with that data”. If this bill were tobe passed into law, Vermont would have an exceptionally strong legal basisfor an open data program.Maine does not seem to have initiated any dialog around open data lawsthus far. In fact, in 2013 three bills proposed in the state legislature soughtto exempt additional record types from basic Freedom of Access requests(Mistler, 2013). The minimal legal framework for open data may be a con-tributing factor to Maine’s portal including the fewest datasets out of thenorthern New England portals. Existing legislation in New Hampshire andproposed legislation in Vermont could serve as an example of how open datacan be promoted through law, if Maine were to ever decide this issue is astate priority.Based on the current laws, open data is vulnerable to a change in politi-cal opinion in all states studied. Although New Hampshire requires data inopen formats, it does not require that data be published or displayed in a49common directory. Maine and Vermont currently offer no reason to believethat open data portals will outlast their current champions. The portals of allthree states could easily disappear, as has happened with many predecessors.5.4 History and ChangeHave there been any changes in state-level portals reflecting the change infederal administration priorities?Each of the portals has followed a different trajectory of change throughtime. New Hampshire has left its portal virtually unchanged since its cre-ation, only removing one dataset. Maine had an initial flurry of activityin 2015, followed by a slow trickle of additions to the portal. Vermont hassteadily added new material to its portal, keeping the content of the portalin a consistent state of expansion. Some of these differences may be due tothe culture of support around the portals at the time of their creation. Ver-mont has the strongest open data culture, as the only state studied to host asummit on open data and create an open data portal on a municipal level inaddition to the state portal. Relying on a small group of activists rather thangetting full buy-in from the existing social networks of government agencies,however, is an unstable foundation for an open data project (Dawes et al.,2016). There is not enough sustained participation from enough people toindicate that this isn’t the case in Vermont. Maine’s governor supportedopen financial data as an extension of his fiscal conservatism, but there hasbeen little indication of interest in open data by government employees orcitizens otherwise. New Hampshire does not have a history of an open dataculture beyond its open data law. The law’s libertarian sponsor suggests thatperhaps this portal is an extension of New Hampshire’s libertarian politicalleanings, with the government providing the minimum support possible andexpecting market forces to do the rest. These cultural factors are importantin understanding the motivations behind the portals, given the criticisms50of Bates (2014) that suggest many governments participate in open dataprojects in order to pursue a neoliberal agenda of privatization and appeartransparent to gain political favor without commitment to full transparency.It seems likely that Maine and New Hampshire fall somewhere within thispattern, rather than pursuing an open data agenda for idealistic reasons.History also illuminates the fates of the portals’ failed open data prede-cessors. Maine’s DataShare seems to have been discontinued as it has beenleft without update for seven years and many links on the page redirect todata.maine.gov. However, not all of the data provided on this portal wascopied over to the new portal, and there is no redirect for the home page orinformation on DataShare that points users towards the updated informa-tion. The previous New Hampshire site, nhopengovt.org, is now an insurancesales site. The data that was once available through this portal does notseem to have been preserved anywhere online, so New Hampshire citizenshave lost easy access to the financial transparency data this portal had pro-vided. Other sites, most notably data.gov, still link to the nhopengovt.orgaddress, highlighting the problem of having an open data site run by an or-ganization outside the government that may allow the URL to be purchasedby other organizations for unrelated purposes. Both portals show the needfor a plan to retire an open data site, rather than allowing it to go fallow.We do not see clear evidence of the influence of the federal government’snew attitude towards open data affecting the state level portals. There are,however, two instances of change where the timing coincides with the changein administration. The removal of Vermont’s Governor’s dashboard site oc-curred in the same time period as the 2016 election. Rather than reflectingthe federal change, we think this removal is more likely to do with the pri-orities of Vermont’s new governor elected at that time. In Maine, the lastupdate of data.maine.gov was November 6, two days before the presidentialelection. While this is in no way a conclusive signal that the state is movingaway from open data because of the federal administration, the timing is co-51incidental enough to merit further consideration, especially since there wasno co-occurring change in state government.5.5 FrameworksWhile the Open Data Policy Framework and Open Data Barometer are themost appropriate tools we were able to find currently in the literature forexploring and comparing these portals, neither tool is ideally suited to theexamination of state-level open data. This is consistent with the cautionsabout applying evaluation tools outside of their context expressed by Sushaet al. (2015). While the Open Data Policy Framework created by Zuider-wijk and Janssen (2014) raises many important questions about the opendata portal, its criteria assume a more developed portal, and therefore manyof its questions are unanswerable for the portals studied. This tool alsoseems as if it would be more useful for a self-study or study in which theresearchers had the ability to personally interview portal staff, as many ques-tions require information that is unpublished, or information that requiresexhaustive searching, such as inventorying any usage of the data, where itis difficult to determine whether all relevant information has been found. Atool more specifically suited to the aims of this study would consider detailsof the published policies governing the portal, such as a specific checklistof policy elements and incorporate more factors from the political situationsurrounding the portal, such as political movements active in the state.The Open Data Barometer was too wide-reaching for this study, leadingus to modify it by narrowing the focus to only information provided on theportal. Even with this modification, this is an imperfect measure for stateopen data. Of the 15 categories of data the portal inventories, 6 categorieswere not present within any of the portals. While some of these categoriescould be useful additions to a state portal, like election results, several areinapplicable in this context, such as international trade data. This does notnecessarily mean, however, that these metrics would need to be edited out in52order to be useful for state-level research, as international trade data mightbe useful and available in a large state like California. This highlights a majordifficulty of research on a state level- what is appropriate for evaluating theportal of a large, resource rich state is unlikely to translate to small ruralstates such as the ones in this study.53Chapter 6Conclusions6.1 The State of State Open DataIn the northern New England states, the future of the state-level data portalsis uncertain. Two to three years after the creation of the portals, Vermontcontinues to update and improve its portal, Maine seems to have stoppedits updates, and New Hampshire has made almost no updates to its originaldirectory. Vermont’s model, a single portal with software that supports users’creation of visualizations, seems to be more effective for users than Maine’sdual portals or New Hampshire’s list site.The most common goal for the portals is transparency, but exactly whattransparency looks like is unclear. Is simply providing open data enough?Or would true transparency require advertising the available information andmaking sure users have the technical knowledge to make use of the data?Defining what transparency means in this context and what transparency issupposed to accomplish would help focus any future portal development onwhere it could have the largest impact. The same principle applies to theadditional portal goals of education, fiscal responsibility, and participation-without clearly defining what these goals will achieve, it is difficult to deter-mine what a successful portal will look like.Several common issues occur across portals that limit the ability of the54general public to use the data. One issue is the incomplete and inconsistentlicensing information. Without making the terms for reuse extremely clear,few will understand whether or not they have a legal right to create newproducts with the data. The ambiguity of portal ownership is also a con-cern. Although all of the portals have some relationship with their state’sinformation technology department, finding a contact person who is able toanswer questions about the portal is extremely challenging. It seems as if theportals are not considered a critical project in any of the states and thereforedon’t truly belong to any specialized working unit. Related to this difficultycontacting any portal leadership is a lack of technical support. Portals donot provide any assistance to users having difficulty navigating the portal.The training materials the Maine and Vermont portals provide are minimaland hidden, decreasing the likelihood that they will be used.Because of these issues, none of the portals show evidence of leading tonoteworthy reuse of the data. While Vermont’s portal leads in use of thedata, this is achieved with only a small number of visualizations and nolarger projects. None of the states have significantly invested in advertisingand promoting the portals. The datasets provided can also affect the level ofreuse. Most information on the portal seems to have been selected based onthe convenience of the portal managers, leading to some datasets with lessvalue being included while datasets that would be more technologically chal-lenging to include but have a large potential for reuse are excluded from theportal. Some datasets that are only on one portal, such as licenses for trades-people and childcares on the Vermont portal, show a high level of communityinterest based on number of views. This could be an area of expansion forother portals. Datasets that uniquely represent the state’s character, such asMaine’s lighthouse map, are also popular. The lack of citizen awareness ofthe portal and the creators’ prioritization of “easy” datasets combined withthe previously identified lack of clarity in and support for the portal lead tothe low levels of reuse.55While all states have some sort of Freedom of Information law to ensurethat citizens have access to government information, this does not guaranteeproactive moves such as uploading government data to a portal. New Hamp-shire is the only state that has specific open data legislation, which requiresall data produced by the government to exist in an open data format. Thishas the advantage of ensuring that data is ready for publication and reusewithout preparation by the data portal, but it does not mean that a givendataset will be available in general. Proposed legislation in Vermont wouldrequire the state government to remove identifying information from dataand then publish it in a format conducive to analysis by the public. A legalframework like this could help open data be adopted as a part of the normalwork flow of government, rather than being viewed as an extra that relies onthe interest of government workers to continue.The effects of the political changes at the federal level were not clearlyevident. The change on the portals that coincided most clearly with thechange in administration is in Maine. The last dataset added in Maine wasadded two days before the general election, and no updates or uploads haveoccurred since that point. Whether this is a direct result of the culture cre-ated by the federal government’s decreased support of open data, or merelya coincidence in timing is unclear, but it does signal that the initial pushfor open data in Maine has not led to a lasting inclusion of open data asa part of government operations. The influence of state politics on portals,however, was noticeable, whether this came in the form of Vermont’s removalof a transparency site with the change of governor or New Hampshire’s lib-ertarian approach to the portals. The influence of political factors on opendata portals suggests that in order to be viable long-term, portals would needto be woven into the state legal framework and become integrated with thedaily operations of government.566.2 Contribution of Research, Strengths and WeaknessesThe case studies provide a snapshot of some of the difficulties and successesthat smaller states experience in creating and maintaining an open dataportal. Since state portals have not been the object of as much researchattention as national or municipal portals, identifying and articulating someof the challenges they face is an important first step to research.One strength of this research is the similarity between states and its trans-lation into a similarity of issues in the portals. The lack of engagement withthe portals is unsurprising given the rural nature of the states. The north-ern New England states would need to devote serious resources to trainingand support in order to get those citizens who may not rely on informationtechnology for their everyday work to be confident in their use of the portal.A weakness of this study was the lack of input from government officials.We had hoped to be able to access internal documents related to the portals,in order to get clearer ideas of how they approached updating, anonymizing,and other issues, but the only department supervising a portal we were ableto get in contact with, Maine’s, seemed to know almost nothing about theportal. Portal owners for New Hampshire and Vermont simply never replied,so whether they have these documents or not is unknown. In any case, thedifficulty of getting in contact with the portals for research questions high-lights their general lack of communication, which could prove problematicfor users.6.3 Application of FindingsThis research could be useful to those in government tasked with creating ordeveloping open data portals. Comparing the three portals highlights eachportal’s strengths and weaknesses, a practice that could be extended by aportal developer to other state portals to gain additional ideas for improve-ment. The primary recommendation for data portals coming out of this57research is to extend transparency to the portal’s own practices by makinginformation such as the criteria for inclusion, updating schedule, licensinginformation and contact information clearly available. The types of data in-cluded in the portals could also inspire portal developers to add new ideas.Open data practitioners should also note the importance of promotion, andinclude this in their plan from the outset. The findings of this research canalso provide background to those seeking to develop open data laws in statelegislature. Librarians can apply this research to a consideration of how theycould contribute their knowledge organization and information literacy in-struction skills to improve state open data.6.4 Future ResearchBecause the research presented is preliminary, many areas for future researchcan be explored. One major potential avenue is looking at the effect of na-tional political shifts throughout a larger sample of portals. Tracking changesin many state portals over time and correlating them with differing politicalclimates could provide interesting insights into the effect of national govern-ment policies on state level portals. If more portals show Maine’s pattern ofceasing updates in the time period of the 2016 election, this could provideevidence that there is a political factor in this change. Related to the con-sideration of the national political context, the effect of the politics withina state could be an interesting area of research. This study included opendata portals with conservative, liberal and libertarian influences, and explor-ing how these different guiding ideologies affect portals of a larger sample ofstates could clarify the relationship between politics and product.Another area worth future investigation is the correlation between suc-cessful municipal open data portals and successful state portals. In this study,Vermont had the most developed portal, and was also the only state to havea municipal-level open data site in any of its cities. If these conditions occurtogether frequently, it could also reveal what underlying factors may lead to58the success of both levels of portal, whether this is an active citizen interest,champions in government, or another driving force.Further research could also consider how the regional cooperatives suchas those municipal cooperatives studied by Lassinantti et al. (2014) could betranslated into state level portals. Although it might be difficult for differentstates with different legal frameworks for open data to share one portal,there could be areas in which they could collaborate. For example, theNew England portals all share a lack of support for data users. A possiblesolution could be staff tasked with assisting users for multiple portals, sothe cost would be shared. This cooperative model might also require morestandardization of design between portals. Research building on municipalportal research and examining the feasibility at the state level could improveservice delivery.Beyond this, more exploration of the general usefulness of state data por-tals is necessary. The portals investigated in this study were not heavily used,and did not lead to the development of applications or projects. Whetherthis is an artefact of these being smaller, less developed portals or whetherthis is common to many state portals is an important question for open datascholarship. If state data portals are generally ineffective, a new approachto sub-federal open data in areas without large municipalities would need tobe considered.To support the investigation of the usefulness of state data portals, it mayalso be necessary to develop evaluation tools specifically attuned to the statelevel. Some of the dataset types in the Open Data Barometer were clearlyinapplicable to the state context, suggesting that a method developed forevaluating state level data might provide more useful insights. If this methodof evaluation is going to investigate usefulness, it may need to move beyondtallying of datasets to include measures of use and engagement.59ReferencesAction plan. (2017). Retrieved April 27, 2017, from https://www.opengovpartnership.org/country/united-states/action-planAn act relating to the creating an office of public policy, H. 435. (2015). Retrievedfrom http://legislature.vermont.gov/bill/status/2016/H.435Alonso, J. M., Ambur, O., Amuito, M. A., Anzanon, O., Bennett, D., Flagg, R.,& Sheridan, J. (2009). Improving access to government through better use of theweb. Retrieved June 15, 2017, from https://www.w3.org/TR/egov-improving/American community survey. (n.d.). Retrieved July 24, 2017, from https://www.census.gov/programs-surveys/acs/data.htmlAnnual estimates of the resident population: April 1, 2010 to july 1, 2016.(n.d.). Retrieved June 5, 2017, from https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmkBates, J. (2014). The strategic importance of information policy for the con-temporary neoliberal state: The case of open government data in the unitedkingdom. Government Information Quarterly , 31 (3), 388-395. Retrieved fromhttps://doi.org/10.1016/j.giq.2014.02.009Both, W. (2012). Open data - what the citizens really want. The Jour-nal of Community Informatics, 8 (2). Retrieved July 24, 2017, from http://ci-journal.net/index.php/ciej/article/view/814Chan, M., Johnson, P., & Shookner, M. (2016). Assessing the use of governmentopen data and the role of data infomediaries: The case of nova scotia’s communitycounts program. JeDEM - eJournal of eDemocracy and Open Government , 8 (1),1-27.Chokshi, N. (2014). What each state’s veteran population looks like. Retrieved July6024, 2017, from https://www.washingtonpost.com/blogs/govbeat/wp/2014/11/11/what-each-states-veteran-population-looks-like-in-10-maps/?utm_term=.365a7ac15f00Clarke, A., & Francoli, M. (2014). What’s in a name? JeDEM - eJournal ofeDemocracy and Open Government , 6 (1), 248-266.Cohn, S. (2012, August). Who is a freestater? - guest article by N.H. Rep.Retrieved July 27, 2017, from http://freekeene.com/2012/08/06/who-is-a-freestater-guest-article-by-nh-rep-seth-cohn/Conradie, P., & Choenni, S. (2014). On the barriers for local government releasingopen data. Government Information Quarterly , 31 . Retrieved from https://doi.org/10.1016/j.giq.2014.01.003Dawes, S. S., Vidiasova, L., & Parkhimovich, O. (2016). Planning and designingopen government data programs: An ecosystem approach. Government Infor-mation Quarterly , 33 (1), 15-27. Retrieved from https://doi.org/10.1016/j.giq.2016.01.003De Blasio, E., & Selva, D. (2016). Why choose open government? motivations forthe adoption of open government policies in four european countries. Policy andInternet , 8 (3). Retrieved from https://doi.org/10.1002/poi3.118G. 8 Open Data Charter and Technical Annex - Gov. UK. (n.d.). Re-trieved June 4, 2017, from https://www.gov.uk/government/publications/open-data-charter/g8-open-data-charter-and-technical-annexJanssen, M., Charalabidis, Y., & Zuiderwijk, A. (2012). Benefits, adoptionbarriers and myths of open data and open government. Information Sys-tems Management , 29 (4), 258-268. Retrieved from https://doi.org/10.1080/10580530.2012.716740Janssen, M., & Zuiderwijk, A. (2014). Infomediary business models for connectingopen data providers and users. Social Science Computer Review , 32 (5), 694-711.Retrieved from https://doi.org/10.1177/0894439314525902Kassen, M. (2013). A promising phenomenon of open data: A case study of thechicago open data project. Government Information Quarterly , 30 (4), 508-513.Retrieved from https://doi.org/10.1016/j.giq.2013.05.012Krippendorf, K. (1980). Content analysis: An introduction to its methodology.61Beverly Hills: Sage Publications.Kroll, A., & Choma, R. . (2017, April). Donald Trump is mak-ing the United States an anti-corruption laughingstock. Retrieved June 2,2017, from http://www.motherjones.com/politics/2017/04/trump-ethics-anti-corruption-foreign-corrupt-practices-actLassinantti, J., Bergvall-Kåreborn, B., & Ståhlbröst, A. (2014). Shaping local opendata initiatives: Politics and implications. Journal of Theoretical and AppliedElectronic Commerce Research; Curicó, 9 (2), 17-33.Lee, J. (2013). Exploring the role of knowledge networks in perceived e-government:A comparative case study of two local governments in korea. American Reviewof Public Administration, 43 (1), 89-108. Retrieved from https://doi.org/10.1177/0275074011429716Maruyama, M., Douglas, S., & Robertson, S. (2013). Design teams as changeagents: Diplomatic design in the open data movement. In (p. 1860-1869).Mistler, S. (2013, March). In maine, ‘sunshine’ law’s fate looks dim. Portland PressHerald. Retrieved July 20, 2017, from http://www.pressherald.com/2013/03/02/sunshine-laws-fate-looks-dim-in-maine_2013-03-03/Mooney, C., & Perin, E. (2017, April). Epa website removes climate sci-ence site from public view after two decades. Retrieved June 2, 2017, fromhttps://www.washingtonpost.com/news/energy-environment/wp/2017/04/28/epa-website-removes-climate-science-site-from-public-view-after-two-decades/NHOpenGovt lets you Google your government. (n.d.). Retrieved July24, 2017, from http://www.jbartlett.org/nhopengov-lets-you-google-your-governmentObama, B. (2009, January). Executive order – transparency and open government.Retrieved May 20, 2017, from https://obamawhitehouse.archives.gov/the-press-office/transparency-and-open-governmentObama, B. (2013, May). Executive order – making open and machine read-able the new default for government information. Retrieved June 14, 2017,from https://obamawhitehouse.archives.gov/the-press-office/2013/05/09/executive-order-making-open-and-machine-readable-new-default62-government-ODC Open Database License (ODbL) summary. (n.d.). Retrieved July 24, 2017,from https://opendatacommons.org/licenses/odbl/summary/OECD. (2017). Open government. Retrieved June 2, 2017, from https://www.oecd.org/mena/governance/open-government.htmOlsen, W. (2012). Data collection: Key debates and methods in social research.London: SAGE Publications Ltd. doi: doi:10.4135/9781473914230.n14Open data barometer global report 4th ed. (2017). Retrieved July 27, 2017, fromhttp://www.opendatabarometer.orgOpen Data Portal. (2017). Retrieved July 24, 2017, from http://dii.vermont.gov/infrastructure/hosted/open-dataOpen Government Data Legislation. (2017). Retrieved July 16, 2017, fromhttp://www.ncsl.org/research/telecommunications-and-information-technology/open-data-legislation.aspxOpen source resources and open data formats. (2015). Retrieved July 24, 2017,from https://www.nh.gov/doit/open-source/data-software.htmOpen source software and data formats policy. (2012). Retrieved July 28,2017, from https://www.nh.gov/doit/open-source/documents/open-source-policy.pdfRight to know law. (2017). Retrieved July 20, 2017, from https://www.nhmunicipal.org/service/right-to-knowRSA 21-R:10-13. (2012). Retrieved July 24, 2017, from http://www.gencourt.state.nh.us/rsa/html/I/21-R/21-R-13.htmRuijer, E., Grimmelikhuijsen, S., & Meijer, A. (2017). Open data for democ-racy: Developing a theoretical framework for open data use. Government In-formation Quarterly , 34 (1), 45-52. Retrieved from https://doi.org/10.1016/j.giq.2017.01.001Steinbauer, M., Hiesmair, M., & Anderst-Kotsis, G. (2016). Making com-puters understand coalition and opposition in parliamentary democracy. InH. J. Scholl et al. (Eds.), Electronic government: 15th IFIP WG 8.5 Inter-national Conference, EGOV 2016, Guimarães, Portugal, September 5-8, 2016,Proceedings (pp. 265–276). Cham: Springer International Publishing. Retrieved63from https://doi.org/10.1007/978-3-319-44421-5_21 doi: doi:10.1007/978-3-319-44421-5_21Susha, I., Zuiderwijk, A., Janssen, M., & Gronlund, A. (2015). Benchmarks forevaluating the progress of open data adoption: Usage, limitations, and lessonslearned. Social Science Computer Review , 33 (5), 613-630. Retrieved fromhttps://doi.org/10.1177/0894439314560852Ubaldi, B. (2013). Open government data: Toward empirical analysis of opengovernment data initiatives. OECD Working Papers on Public Governance(22).doi: doi:10.1787/5k46bj4f03s7-enVermont Open Data Summit. (2017). Retrieved July 16, 2017, from http://vcgi.vermont.gov/events/open_data_2013Vintar, M., Kunstelj, M., & Leben, A. (2002). Delivering better quality publicservices through life-event portals. In Proceedings of the 10th nispacee annualconference: Delivering public services in cee countries: Trends and developments.Violent crime rate by state. (2017). Retrieved July 22, 2017, from http://www.worldatlas.com/articles/the-most-dangerous-states-in-the-u-s.htmlYang, T.-M., Lo, J., & Shiang, J. (2015). To open or not to open? determi-nants of open government data. Journal of Information Science, 41 (5), 596-612.Retrieved from https://doi.org/10.1177/0165551515586715Yang, T.-M., & Wu, Y.-J. (2016). Examining the socio-technical determinants in-fluencing government agencies’ open data publication: A study in taiwan. Gov-ernment Information Quarterly , 33 (3), 378-392.Yin, R. K. (1994). Case study research: Design and methods. Sage Publications.Zuiderwijk, A., & Janssen, M. (2014). Open data policies, their implementationand impact: A framework for comparison. Government Information Quarterly ,31 (1), 17–29. Retrieved from https://doi.org/10.1016/j.giq.2013.04.003Zuiderwijk, A., Janssen, M., & Dwivedi, Y. K. (2015). Acceptance and use pre-dictors of open data technologies: Drawing upon the unified theory of accep-tance and use of technology. Government Information Quarterly , 32 (4). doi:doi:https://doi.org/10.1016/j.giq.2015.09.00564Appendix ADictionary for Open DataMotive Cod-ingAll words should also be counted if occurring in a different form eg. trans-parency or transparent• Economic: Financial, savings, dollars, money, spending• Innovation: Innovate, rethink, streamline,• Transparency: Accountability, right to know, disclosure• Participation: Discussion, interactive, share, reuse, remix, create• Education: Knowledge, student, educator, class, training65


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items