Ambiguities of Open Government
For a concept that seems quite straightforward, the meaning of “open government” remains highly ambiguous and contested (Aitken, Reference Aitken2017; Clarke and Francoli, Reference Clarke and Francoli2014). To begin, open government is commonly associated with enhanced accountability. For example, when introducing Canada's Access to Information Act, then Minister of Communications Francis Fox argued that the legislation would make government more accountable to Parliament, the press and the electorate. In 2006, the Conservatives echoed the theme that open government means enhanced accountability, campaigning on increasing accountability in government by advocating for streamlined procedures under the Access to Information Act and extending its coverage to a greater number of public bodies.
Another conception of open government equates it with open data: promoting easier access to government data and statistics. In this regard, Ontario Premier Kathleen Wynne referred to her government's 2013 Open Government Initiative as revolving around the publication of government information and statistics: “We need to make information easier to find, understand and use, so that we can design services that deliver better results to the people of Ontario. We must also unlock public data so that you can help us solve problems and find new ways of doing things” (Benzie, Reference Benzie2013). Similarly, the federal Conservative government, which is generally understood to be “control minded” (Roy, Reference Roy2017: 439), launched a similar initiative in 2012. The Open Government portal on the Government of Canada's website identifies ‘open data’ as one of three core goals along with ‘open information’ and ‘open dialogue.’ The website is intended to facilitate the release of government data and statistics to users (Canada, 2017).
Often, however, open government is also linked to promoting participation in government decision making. For example, speaking about the Ontario's Open Government Initiative, former cabinet minister John Milloy argued this: “At its core, it's about involving people more in decision-making and the governing of our society” (Belgrave, Reference Belgrave2013). In her assessment of the Conservative government's progress on implementing the Open Government Partnership (OGP), Francoli (Reference Francoli2016) recommended that the federal government should move its implementation of the OGP increasingly toward citizen collaboration and empowerment. In the 2015 Liberal campaign platform, the party branded its promise to end the single-member plurality electoral system as bringing about “open and fair elections,” and this promise was included in the platform section on “fair and open government” (Liberal Party of Canada, 2015). Similarly, in 2014 Liberal leader Justin Trudeau was quoted as saying: “The kind of politics we need in this country is politics that is actually open, and has room for divergent opinions in trying to move forward” (Brieva and Martin, Reference Brieva and Martin2014). Moreover, the emergence of new media technology has broadened and amplified interest in finding ways to both promote and normalize increased civic engagement in politics and governance (Lee and Kwak, Reference Lee and Kwak2012).
Finally, another dimension of open government relates to the access that governments grant to journalists. For example, after awarding the Harper government an award for secrecy, the Canadian Association of Journalists wrote: “His [Harper] government departments now deal with media almost exclusively by email, often returning with unhelpful or useless responses containing little to none of the information actually requested” (Canadian Association of Journalists, 2012). Here, secretive government—the nominal opposite of open government—is equated with a failure to properly respond to journalists’ requests for information.
While these examples are Canadian, open government has an international character. One of President Obama's first initiatives was to issue a Memorandum on Transparency and Open Government (Obama, Reference Obama2009). To date, 75 countries have joined the Open Government Partnership (OGP). Reflecting the many dimensions identified above, the OGP describes four core commitments for members: first, increasing information; second, supporting civic participation; third, implementing high ethical standards (that is, combatting corruption); and lastly, making new technologies available to support openness and accountability (Open Government Partnership, 2011).
The scholarly literature on open government has noted these ambiguities. Mulgan argues that open government is often first equated with transparency but also implies increased participation. “‘Open government,’ in turn will be understood broadly to include both transparency and accountability … to work effectively, open government requires both transparency and accountability in all its stages” (Reference Mulgan2014: 6). Others argue that this can lead to disappointment and recriminations. For example, Yu and Robinson (Reference Yu and Robinson2012) note that governments can provide information in user-friendly formats such as machine-readable bus schedules (“open data”) but that this in no way guarantees accountability by decision makers to voters.
When policymakers and the public use the same term for both of these important benefits, governments may be able to take credit for increased public accountability simply by delivering open data technology … we acknowledge that this ambiguity may sometimes be beneficial, but ultimately argue that the term ‘open government’ has become too vague to be a useful label in most policy conversations (Yu and Robinson, Reference Yu and Robinson2012).
Clarke and Francoli (Reference Clarke and Francoli2014) offer perhaps the best distillation of the concept of open government into four dimensions: access to information, accountability, public participation and improving public services.
Finally, the addition of meanings to the term open government has, as yet, gone unconsidered in public debate and in the scholarly literature, but we consider it to be a potentially productive pursuit. Another meaning of “open” is to have an honest and frank conversation about the advantages and disadvantages of a course of action. In this regard, open government might be a process that emphasizes the costs and benefits or trade-offs associated with any given policy.
The concept of open government is multidimensional and is linked to many other valued concepts such as accountability, transparency, better governance and participation. It is also a contested concept that points to different outcomes ranging from the highly optimistic view that open government is an important mechanism to transform government to the more cynical view of “open washing,” in which open government is a merely a branding exercise with no substance (Aitken, Reference Aitken2017). Several questions have arisen, however. First, does one definition of open government predominate in Canadian politics? Second, do some groups privilege some definitions at the expense of others? Third, and lastly, what do the answers to the first two questions suggest for the prospects for improving public policy in Canada?
To begin to answer these questions, we conducted an exploratory survey of Canadian political journalists, elected federal and provincial parliamentarians and political bloggers. The findings indicate that these key political actors have decidedly different things in mind when prompted to think about the meaning of open government. While something approaching a superficial consensus equates open government with increased access to information, important competing conceptions underlie this approach. Competing definitions of open government reflect important political divisions in Canadian politics that are heavily influenced by actors’ self-interests. The implications of these findings, while exploratory, are significant. We suggest that the self-interested nature of the competing conceptions of open government act as an important barrier and previously unknown source of resistance to more open governance in Canada. Exploring such sources of resistance to open government has been identified as “a critical contour for research” in the fields of public administration and open government (Roy, Reference Roy2017: 441).
This article advances through three stages. First, we consider some theoretical reasons that political actors might define open government differently. Here, we focus on arguments from institutional incentives, generational changes, ideological commitments and differences between French and English Canadian journalists. Second, the modelling strategy and results are discussed. Third, we conclude with a discussion about some preliminary implications of these findings with some avenues for future research.
The Institutional Imperative
The hypothesis that we are most keen to test is that there are meaningful distinctions in how people define open government depending on the role they play in Canadian politics. This hypothesis is derived from institutional theories of human behaviour that explain preferences and behaviours with reference to the institutional contexts in which political actors move (Hall and Taylor, Reference Hall and Taylor1996). While there are distinctions within institutionalist theories about the extent to which institutions shape behaviour by structuring the incentives (rational choice institutionalism) or by inculcating and emphasizing certain norms of behaviour (sociological theories), our results are not intended to settle the debate between these two explanations. There are prima facie reasons to think that both might play a role in shaping how journalists and politicians define open government. A great deal of work on the sociology of journalism has identified how both interests and incentives play important roles in shaping news coverage (Fengler and Ruß-Mohl, Reference Fengler and Stephan2008; Gandy, Reference Gandy1982; Gans, Reference Gans1979; Quinn, Reference Quinn2012; Reese, Reference Reese2001). On the other hand, considerable evidence documents the importance of institutional norms to news production (Bennett, Reference Bennett1990; Galtung and Ruge, Reference Galtung and Ruge1965; Tuchman, Reference Tuchman1972). Similarly, it is not difficult to imagine why politicians might resist definitions of open government that put constraints on their behaviour by creating a standard whereby they are expected to be available for scrutiny and accountability on a regular basis (Epps, Reference Epps2008). For example, politicians may have to rely on unsavoury horse trading to reach desirable policy goals. Under such conditions, extreme openness may make their job of legislating more difficult. Moreover, in theory, secrecy inside parliamentary caucuses allows backbench members to hold its parliamentary leadership to account. Openness in the sense of complete transparency may reduce the backbench capacity of parliamentarians to constrain their caucus leadership. Therefore, we hypothesize that respondents will prefer definitions of open government that are congenial to their interests and, specifically, that journalists will privilege definitions that emphasize information and access more than politicians will.
To rule out other explanations for variations in definitions, we tested this hypothesis against alternative hypotheses derived from three other important contemporary theories of Canadian politics. First, as the concept of open government touches on fundamental questions of the relationship between citizens and the state, it would not be surprising to see variation based on left–right ideology. Although most members of the mass public lack cohesive ideological commitments, we used this survey to investigate elite attitudes in contexts in which ideology has been shown to serve as a stronger organizing force (Converse, Reference Converse2006; Jennings, Reference Jennings1992). However, it is not entirely clear what that specific relationship might look like. Do those on the left prefer definitions of open government that emphasize access and information more than those on the right? Both left and right ideologies tend to be suspicious of the possibilities of democratic processes being co-opted by special interests. Both tend to be suspicious of the possibility of special interests corrupting the policy process, the left of corporations, while the right wing is often concerned with the excessive influence of unions, minorities and public-sector workers. Additionally, both left- and right-wing forces can, at times, make commitments to greater public participation. For example, the right-wing Reform Party in Canada made decision making by referendum and other forms of direct democracy key elements of its platform. On the other hand, the New Democratic Party (NDP) has also emphasized themes of “open government.” The former NDP Member of Parliament Pat Martin introduced the Open Government Act as a private member's bill in 2008 and 2011. Consequently, there are no clear expectations about what the relationship between left–right ideology and preferences for definitions of open government might be. However, given the fact that our primary hypothesis is that position influences definitions of government and that the left–right distinction is foundational in elite politics, it should be explored, if only as a control.
A second rival theory is that of generational change. Generational change remains one of the most profound sources of political socialization (Inglehart and Abramson, Reference Inglehart and Abramson1995). In Canada, Nevitte (Reference Nevitte1996) has argued that the shift to a wealthier, more educated advanced industrial economy has brought about a “decline of deference” in Canadian society. Reflecting this, young Canadians display a clear preference to engage in politics via interest groups rather than in the formal party system; they view this approach as a more direct route to participation and influence (O'Neill, Reference O'Neill2007). While trends toward greater citizen participation may be evident among young Canadian politicians, existing research suggests that the opposite may manifest among Canadian journalists. Pritchard et al. (Reference Pritchard, Brewer and Sauvageau2005) found that journalists under 30 years of age were significantly less likely to value investigating government and public institutions to discuss policy while it is still being discussed and to be sceptical of the actions of public officials. This finding suggests that deference to authority may not decline uniformly with age. Thus, it is not entirely clear how generational differences might influence preferences for definitions of open government. A “decline of deference” is as much about scrutiny and skepticism of authority as it is about participation. Because of this unclear relationship and the fact that our primary interest is the hypothesis of institutional imperatives, we formed no hypotheses at the outset about what patterns might exist between age and preferences for definitions.
Lastly, given Canada's bilingual character, it would not be surprising to find that journalists, parliamentarians, and bloggers might define open government differently depending on their linguistic group. Important differences in public opinion between French and English Canadians have been well documented (Lambert and Curtis, Reference Lambert and Curtis1993; Lipset, Reference Lipset1986, 1990; Taylor, Reference Taylor1993; Zheng and Baer, Reference Wu and Baer1996). More specifically, in their surveys of Canadian journalists, Pritchard et al. (Reference Pritchard, Brewer and Sauvageau2005) found evidence of distinct cultures of journalism in French and English Canada. Specifically, they found that, by 2003, English-Canadian journalists had reported that accurately reporting the views of public officials, providing analyses of complex problems and giving members of the public a chance to express their views were less important than they were in 1996. We interpreted this to mean that English-Canadian journalists had increasingly subscribed to a watchdog and confrontational vision of journalism and more so than French-Canadian journalists. More specifically, we hypothesized that English-Canadian journalists would rank the definitions information and access higher than would French-Canadian journalists.
Although we suspected at least a surface level of agreement about open government being related to the provision of information, we hypothesized that underneath this surface level of consensus, important differences in how journalists, politicians and bloggers conceptualize open government would emerge. If so, then open government may, in fact, operate more like a buzzword than an operating principle in Canadian politics, allowing different actors to invest it with meanings that serve their own, rather than a general, public interest (Kaufer and Carley, Reference Kaufer and Carley1993). Moreover, truly ‘open government’ may never be achieved because relevant political actors disagree about what the actual end goal is. On the other hand, such ambiguity also indicates the existence of an issue that could routinely and repeatedly appear on the public agenda because of the potential for perpetual disagreement over its meaning.
Methods
To test these hypotheses, we developed a questionnaire that included a series of potential definitions of open government, derived from our survey of the contemporary debate. In Table 1, the first column contains the precise definition we asked respondents to rank; the second column contains the variable name used in the text; and the third column contains the broader concept or dimension of open government we discerned from our literature review. This survey was distributed to a sampling frame of Canadian parliamentarians, journalists and bloggers (detailed below).
Table 1. Definitions of Open Government
Participants were asked to rank the definitions from their most preferred definition to their least preferred definition. Respondents were not required to rank all definitions and ties were not allowed. We relied on rankings of the different dimensions rather than ratings because we wanted to minimize the probability of satisficing and to effectively force respondents to make a choice (Krosnick, Reference Krosnick1991; Krosnick et al., Reference Krosnick1996).
We established our sampling frame from two sources. For elected legislators, we downloaded names and e-mail addresses from every elected provincial legislative assembly and the House of Commons in 2014. To contact journalists and online bloggers, we used the CISION media database, a proprietary database that is commonly used by media and public relations professionals to contact journalists and online commentators. To include journalists and bloggers interested in public affairs and politics, we used a search string that ensured that journalists with interest in provincial, federal, local government, politics or public affairs were included. Ultimately, our sampling frame included 2,369 journalists, bloggers and politicians.
We devised a web questionnaire using the LimeSurvey platform, and we conducted a pilot study with university faculty and some journalists who were excluded from the sampling frame. Participants in the sampling frame were sent three separate invitations to participate. The survey was entirely anonymous, but to encourage participation, respondents were invited to record their e-mail addresses to have their names entered in a drawing to choose a charity to receive a $100 donation. Because of the opt-in nature of the sampling process, the data quality must be considered. One measure of the quality of a sample is the response rate. Invitations to participate were distributed six times between the summer of 2014 and February 2015. Overall, 352 respondents returned fully completed survey, for a response rate of 12.9%. The low rate was primarily attributable to politicians, which is perhaps understandable (Table 2). While the overall response rate was low, journalists and bloggers participated at rates comparable to those in similar studies (Pritchard et al., Reference Pritchard, Brewer and Sauvageau2005). While the response rate for politicians was very low, the response rate is an indicator of a survey's ability to test substantive hypotheses (Holbrook et al., Reference Holbrook and Lepkowski2007).
Table 2. Distribution of Respondents and Raw Response Rates by Position
A second way to assess data quality is to compare the demographic composition of the sample with the demographic composition of the populations of interest. Table 3 shows the distribution of the three positions (journalists, parliamentarians and bloggers) by gender and age in the sample compared to two different estimates of the same groups in the population. There were some noticeable differences in the gender composition of the sample sub-groups. There were more male journalists in the sample than in the population but there were more female bloggers and parliamentarians than in the population. Because journalists were the largest subgroup of the sample, the overall sample skewed somewhat male. Overall, the sample is moderately less male than female.Footnote 2 In addition, Table 4 shows a linguistic distribution of the sample that reflects the Canadian population as a whole.Footnote 3
Table 3. Sample demographics compared to population estimates from the 2011 Census (age, gender), National Household Survey (Occupation) and Giasson, Jansen and Koop, Reference Giasson, Jansen and Koop2014 (all information for bloggers)
Table 4. Distribution of respondents by language group compared to the 2011 Census
Lastly, we considered the possibility that the decision to participate in the survey might be linked to the key variable of interest, namely the preferences for definitions of open government. However, we were not overly concerned with this issue. First, evidence has been accumulating that opt-in panels estimate population parameters as well as samples constructed randomly (Breton et al., Reference Breton, Cutler, Lachance and Mierke-Zatwarnicki2017). Second, we were not primarily interested in estimating the distribution of preferences for different definitions in the population. Instead, we looked for differences in how groups ranked definitions. As such, for our sample estimates to be biased in ways that would invalidate our conclusion, it would have to be the case that, for example, journalists who do participate in the survey have different rankings than politicians who do not. Instead, we found it more likely that preferences for certain definitions were distributed in similar ways among nonrespondents and respondents across the three populations of interest.
Findings
Table 5 shows the mean rankings and standard deviations for each item. Information was the most commonly accepted definition, followed by accessible and officials, and these findings were not surprising.
Table 5. Mean rank and standard deviations for each definition of open government. Respondents were asked to rank items from 1 (highest) to 7 (lowest).
However, important differences emerged when the rankings made by members of different groups were compared. Figure 1 shows the distribution of rankings by position, testing our key hypothesis. All groups ranked information highly. However, beneath this level of agreement, we identified important differences. For example, journalists’ median rank for information was 1, while government and opposition ranked it 3 and 2.7, respectively. For journalists, the next two most popular definitions were accessible and officials (median ranks 2 and 3). All of these items in some way reflect a conception of open government that fits with how journalists do their daily jobs; namely, they try to gather documentary evidence and interview human sources. By contrast, while journalists and opposition politicians assigned high rankings to accessible (median ranks 2 and 3), bloggers and government politicians did not (median ranks 4 and 4). Again, this ranking is consistent with what journalists and opposition politicians do in the day-to-day activities of Canadian politics, namely, scrutinizing the government in public. Journalists do so in press conferences, while opposition politicians do so in Question Periods and other legislative fora. Moreover, government and opposition together generally ranked responsive and debate about as highly as information, which again, reflects much of what politicians do, which is to try to mobilize public opinion to influence public policy by debate and persuasion. However, both assigned officials very low rankings and government politicians a particularly low ranking.
Figure 1. Distribution of Definition Rankings by Position
On the surface, we identified two clusters of definitions. Journalists preferred definitions of open government that privilege information and news sources (for example, information, accessible and officials), while opposition and government parliamentarians preferred definitions of open government that privileged information, public participation and debate (for example, information, responsive and debate). Bloggers tended to resemble opposition parliamentarians, more than any other group. Surprisingly, bloggers ranked social very low; instead, government politicians assigned social a higher ranking than did other groups.
Figure 1 provides strong if somewhat impressionistic evidence that institutional position does, in fact, influence how elites define open government. We also tested differences in a more formal way. One complicating factor was the ranked nature of the data. That is, if a respondent assigned an item a rank of 1, then the respondent could not assign the value of 1 to any other item. Had we asked respondents to rate their level of agreement with each of the definitions, a series of standard ordinary least squares regression models might have been more appropriate for our analysis.Footnote 4 The most well-developed modelling strategy to deal with ranked data is the exploded logit or rank-ordered logit model (Chapman and Staelin, Reference Chapman and Staelin1982; Hausman and Ruud, Reference Hausman and Ruud1987; Punj and Staelin, Reference Punj and Staelin1978). Allison and Christakis (Reference Allison and Christakis1994) noted that this model is identical to the Cox partial likelihood method of modelling survival times (Cox and Oakes, Reference Cox and Oakes1984). To make this possible, the data were required only to be reshaped such that there was one row for a respondent–rank pairing that included the rank the respondent assigned to the definition and all covariates of interest. Then, continuing with the terminology of the Cox proportional hazards model, the rank assigned to each definition was considered the time until the event occurred (that is, it was ranked). A higher rank (for example, 1) indicates a shorter unit of time; a lower rank (for example, 7) indicates a longer period. Because of this conceptualization, 2,758 cases were included in the data set, each row being a respondent–definition pair. For each case, the numeric value was the rank assigned by that respondent to the respective definition. Importantly, this is a semiparametric modelling strategy; it only produces information about how respondents prefer one definition to another, not how much they absolutely prefer a definition.
To begin, we specified a baseline model that included only six dichotomous variables, one for each possible definition except for debate. This model was set as a reference category and was set to 0 for all respondents. We chose debate to be the reference category because its average rank was in the middle of the overall range (Table 5). The results of this model are shown in the left-most column of Table 6. The coefficients here are the changes in the log odds that an item was ranked higher than debate. A positive sign means that an item was more frequently ranked higher than debate, and a negative sign suggests the opposite. Notably, the size of the coefficients in column 1 reflects the ordering of mean ranks shown in Table 5, validating this modelling strategy.
Table 6: Modelling Ranked Definitions of Open Government
Note: *p < 0.05**p < 0.01***p < 0.001
Model 3 includes gender as a control because the sample contains more male journalists than in the Canadian population. None of the coefficients were significant and are not reported for readability
Because changes in log odds can be difficult to interpret, it is easier to understand the odds ratio, which can be obtained by exponentiating the coefficients. Information and accessible were significantly more likely to be ranked higher than debate. Information was 2.58 times more likely to be ranked higher than debate, while accessible was 1.66 times more likely to be ranked higher than debate. By contrast, the coefficients for officials and responsive were not statistically significant, which suggests that respondents ranked these items similarly to debate. Social and trade-offs, however, were much less preferred than debate. The odds ratio for social was 0.12, and for trade-offs it was 0.6. These rankings can be expressed in a percentage form by subtracting the odds ratio from 1 and multiplying by 100. As such, these data suggest that a respondent was 88% less likely to rank social higher than debate and 40% less likely to rank trade-off higher than debate.
However, given the theoretical and practical questions presented in previous sections and the apparent differences visible in Figure 1, we sought to determine whether the preferences that respondents exhibited differed by other variables of interest, particularly by the position that respondents occupied. Our results are shown in the second column in Table 6. In this model, we added a categorical variable for the position a respondent occupies, be it journalist, opposition parliamentarian, government parliamentarian or blogger. Opposition was set to be the reference category. This categorical variable was run through the model with each dichotomous variable representing the different definitions of open government. Thus, the coefficients in column 2 represent the change in log odds that a respondent in that position ranked that item higher than debate compared to opposition politicians. For example, the coefficient for information:journalist was 0.57. This means that journalists were more likely (1.78 times more likely) to rank information higher than debate than were opposition politicians. By contrast, government parliamentarians were significantly less likely to prefer information to debate. With an odds ratio of 0.267, government parliamentarians were 73% less likely than opposition parliamentarians to prefer information to debate. The same pattern was evident for accessible. Journalists were more likely than opposition politicians to prefer accessible to debate, and government parliamentarians were significantly less likely than opposition politicians to do so.
It is evident that journalists, opposition parliamentarians and government parliamentarians do prefer different definitions of open government, and it appears that there are significant differences between types of parliamentarians as well. However, we identified an important difference between politicians as a group and journalists. Journalists were significantly less likely than opposition politicians to prefer social to debate, whereas there were no differences between opposition and government politicians. Interestingly, bloggers showed no such preference for social as a definition of open government. The fact that politicians, rather than bloggers, had a greater tendency to rank social media higher than debate perhaps reflects the fact that politicians view online social media as an “easy” way to lend credence to claims that they are practicing open government without making any substantial reforms. In short, we found significant differences in how people in different positions in Canadian politics preferred to define “open government.” Moreover, strong evidence indicates that they do so in ways that are congruent with their institutional interest.
While the forgoing analysis provides strong evidence of the important role that institutional imperatives play in shaping preferences for definitions of open government, it did not rule out the role of confounding variables. For example, at the time our sample was constructed, the Conservative Party was in power in Ottawa. Although we also included provincial parliamentarians in our sample, the sheer size of the House of Commons compared to provincial legislatures may have skewed the sample toward ideologically conservative politicians who happened to also be in government.
We tested this hypothesis by including variables from three competing explanations adduced at the outset: ideology, generational norms and linguistic differences.Footnote 5 The results that emerged by including variables that measure these concepts are shown in the right-most column of Table 6. In this model, each variable was run with the dichotomous variables representing different definitions of open government. Several findings emerged. Most importantly, these alternative explanations did not eliminate the significant differences in the groups’ relative preferences for information, officials and social, which was evident in the second model. Even accounting for three competing explanations does not substantially change the central finding that preferences for definitions of open government are attributable to institutional position. Journalists were still more likely than opposition parliamentarians to rank information higher than debate, and government parliamentarians were less likely to do so. Government parliamentarians were still less likely to rank accessible and officials higher than debate. Moreover, journalists were less likely than opposition parliamentarians to rank social higher, while there were no differences between the two sets of politicians.
That said, important findings emerged once competing explanations were included. First, some evidence indicated that one's position on the left–right scale structured definitions of open government. The coefficient for trade:economic was positive and significant. For each single-unit shift to the right on the left–right scale, a respondent was 1.09 times more likely to rank trade higher than debate. By contrast, the coefficient for information: economic was negative (−0.07). For each single-unit shift to the right, a respondent was 8% less likely to rank information higher than debate. This finding is theoretically plausible. A core central tenet of the economic left is that social classes dominate politics. To remedy such class inequalities, a first step is to expose the influence of the dominant class on politics, which requires access to government information.Footnote 6 By contrast, a central tenet of right-wing economic ideology is that scarce resources require social choices, that is, that one cannot have it all. Thus, it makes sense that those on the right preferred a definition of open government that emphasizes the importance of trade-offs.
Second, clear differences emerged in accordance with our generational politics hypothesis. Regarding the former, the likelihood of preferring information, officials and social to debate all significantly declined with age. The first two were compatible with a “decline of deference” thesis, whereby elder respondents were less likely to be enamoured of participatory measures that might undermine the autonomy of legitimated representatives to act. Such hesitancy to embrace social media is possibly reflective of different comfort levels with technology between generations. Third and finally, remarkable differences emerged between French and English Canadians. French Canadians were more likely than English Canadians to rank every definition, except for responsive, higher than debate.
To assess whether these differences were the result of different journalistic cultures, as documented by Pritchard et al., (Reference Pritchard, Brewer and Sauvageau2005) or whether this was attributable to the much more broadly unique political culture in Quebec, we fit two more models, one for journalists and one for nonjournalists. The results in Table 7 clearly suggest that francophone journalists were much more likely to prefer information and accessible than were anglophone journalists. By contrast, francophone politicians and bloggers did not rank these substantially differently than did anglophone politicians and bloggers. This finding suggests that there may be a uniquely francophone journalistic culture in the province of Quebec. Notably, however, Pritchard et al., (Reference Pritchard, Brewer and Sauvageau2005) argued in 2003 that anglophone journalists were be more likely to conform to a watchdog model of journalism. These data suggest the opposite.
Table 7: Comparing Ranked Definitions Of Open Government By Francophone and Anglophone Journalists vs. non-Journalists
Note: *p < 0.05**p < 0.01***p < 0.001
Discussion
The key finding of this article is that different political actors—namely journalists, government politicians, opposition politicians and political bloggers—define open government quite differently. Institutional theories of political behaviour offer a plausible explanation for why this is the case. Journalists, bloggers and politicians in government and the opposition have different objectives and are expected to perform different roles (or functions) in the political system. Thus, it makes sense that each of these sets of actors associates different meanings with the term open government based at least in part on occupational self-interest.
One of the classic tensions political actors face is the degree to which they are motivated by pursuing the public interest or their own self-interest. Open government is generally understood as fulfilling the public interest. However, it is evident that important political actors in Canada—particularly politicians and journalists—define the concept at least partially based on occupational self-interest. It is possible that this definitional pluralism could serve the public interest by generating robust debates about political and policy decisions about transparency. However, it is equally possible that this state of affairs creates situations that are at best complex and at worst negative. We suggest that the broader implication of such self-interested conceptions of open government is that they pose two important obstacles to creating more open governance.
The first obstacle to more open governance flows from journalists’ preferences for defining openness as access to information (information) and sources (accessible and officials). Such a definition is logical because it flows in part from journalists’ practical need for raw material for stories. Access offers such raw material. However, access also conforms to the idea of watchdog journalism: monitoring elite behaviour to look for wrongdoing or misbehaviour and then signalling to the public and other elites when misconduct has been found (Serrin and Bennett, Reference Serrin, Bennett, Graber and Graber2000). The growing importance of the watchdog function is one of the central features of Canadian journalism (Pritchard et al., Reference Pritchard, Brewer and Sauvageau2005; Taras, Reference Taras1990). However, given journalists’ proclivity to define it in ways that reflect their day-to-day interests, one is left to wonder whether journalists can effectively scrutinize proposals for open government made by opposition parties and governments. As we have discovered, open government is a multidimensional concept, and it can be addressed with different reforms. Each of these reforms brings a mix of costs and benefits. It is fair to ask to what extent journalists’ preferences for open government as access to sources hinders their ability to evaluate fairly the mix of costs and benefits. Effectively, journalists have a vested interest in this game. Are they able to scrutinize and evaluate the advantages and disadvantages of different ways in which government might be made more open, while they are invested in a particular set of policies?
For example, it is well known that governments invest heavily in news management to try to shape news coverage (Esser and Pfetsch, Reference Esser and Pfetsch2004; Kiss, Reference Kiss2014; Kozolanka, Reference Kozolanka2006; Roberts Reference Roberts2000, Reference Roberts2005). It is entirely possible that politicians provide journalists “access” to government documents comprising strategically crafted material intended to subtly shape news coverage to promote the government's preferred narrative about issues and events. If so, then what journalists consider access can contribute to government efforts to manage the news.
Moreover, the journalists’ definition of open government as access has a self-limiting impact on the effectiveness and impact of political journalism in Canada. Notably, Pritchard et al.’s (Reference Pritchard, Brewer and Sauvageau2005) analysis of journalists’ role conceptions demonstrated an increased commitment to the watchdog role. While the watchdog role is important, the single-minded definition journalists’ embrace with respect to openness perhaps closes off other important ways that journalists could contribute to more open government and politics in Canada. For example, notably missing is an appreciation of the potentially positive impact of defining open government as a discussion of policy trade-offs. All public policies have consequences, and they usually involve trade-offs between or within target populations. While it might be too much to expect politicians seeking re-election to openly discuss the trade-offs associated with particular policies, journalists face no such restriction. If journalists were more creative in their conception of open government and took the discussion of trade-offs as a serious component of promoting open government, citizens would benefit from a more thorough vetting of government proposals, which would, in turn, contribute to more informed and politically sophisticated public opinions.
The second obstacle to more open governance is more overtly political and flows from how politicians in opposition and government use the term open government for political purposes. In our study, government and opposition parliamentarians demonstrated profoundly different conceptualizations of open government, even controlling for ideology, age and language. Government parliamentarians were less likely to prefer information, accessible, officials and trade-offs than were opposition parliamentarians. These different definitions seem to be based on the respective interests of politicians occupying different positions in Canada's Westminster parliamentary system. Simply put, Canadian politics is organized as a daily contest between government and the opposition: the government governs while the opposition holds the government to account (Franks, Reference Franks1987). Politicians’ notions of open government are consistent with such roles. Government politicians in government shy away from definitions of openness that would imply that they should be held to account or that their officials should be easily accessible or that debate should involve hard conversations about trade-offs, while opposition politicians have the reverse set of interests and preferences (defining openness as access to government and emphasizing the costs of decisions) to facilitate questioning and criticism.
Notably, while parliamentarians in our sample tended to define open government in ways that assigned a role for public opinion, our definitions did not ask respondents to define open government in ways that envisioned more direct citizen control over policy. It is entirely possible that parliamentarians like to think of open government as involving a greater role for citizen participation, but only to the point where politicians do not have to give up actual control. Those actors interested in pursuing more open government defined as citizen control, rather than mere influence, may not find friends among parliamentarians or journalists.
Our findings suggest that the systemic or institutional forces that lead to competing definitions here may contribute to a predictable, cyclical pattern in the politics of open government. Opposition politicians and political parties have strong incentives to promise more open government. Indeed, one of the most common spectacles in Canadian politics can be found in the routine criticism made by opposition parties that governments are not being open or transparent. Such references fulfill an important opposition function because they criticize government performance and expose it to scrutiny. Such criticism can shine a spotlight on an important element of the machinery of government (decision-making processes and procedures) that frequently elude public vetting. Promising more openness also allows opposition parties to differentiate themselves from governing parties. Moreover, such promises are inexpensive to make. Promises to strengthen provisions or procedures around freedom of information legislation or increasing public availability of government data sets do not entail considerable reallocations of scarce resources. Thus, it seems predictable that opposition politicians and parties will press government to be more open and that they will promise more openness once they are in power. However, the data presented here suggest that it is equally predicable that, once in government, the same politicians or party that criticized previous governments as being closed and opaque will tend to show similar resistance to openness.
In Canadian politics, this process has occurred in recent political history. In opposition under Stephen Harper, the Conservatives pledged to be more open (in the sense of accountable) in government than the Liberals were. These pledges were made in response to the scandals that characterized the Liberals’ final years in power. Once in office, the Conservatives quickly enacted the Federal Accountability Act, expanded the scope of the Access to Information Act to include Crown corporations, added a Parliamentary Budget Officer charged with providing public, independent assessments of government estimates and the national economy, and made Deputy Minister accounting officers accountable to the House of Commons for the financial administration of their department. Yet, despite this opening move, the Harper government ultimately became highly managed and closed—the antithesis of open government.
Indeed, this very process seems to be playing out again with the Trudeau government. While in opposition, the Liberals criticized the Harper government for lacking sufficient openness and made promises to be more open and transparent if they formed the government. Yet the Liberal government has also exhibited increased secrecy and opacity when managing key policy files. For example, the Trudeau government has been unwilling to allow public scrutiny of an arms deal with Saudi Arabia, and former Parliamentary Budget Officer Kevin Page has noted that the Trudeau government is less transparent with its budgetary practices than the previous Conservative government (Chase, Reference Chase2016; Hall, Reference Hall2016b). Moreover, it was widely noted that the first Liberal party convention since getting into government was a closed and highly managed affair (Hall, Reference Hall2016a). Roy describes this as follows: “While the Liberal government campaigned heavily on themes of open government and democratic renewal, the first half of their first mandate has been marked by political fundraising and secrecy, an aborted electoral reform agenda and the absence of any new flagship digital government initiatives” (Reference Roy2017: 440).
Such a process of governments becoming more “closed” is an entirely logical and predictable outcome in Canadian politics. In the Westminster parliamentary system, governments are expected to govern, that is, to make decisions for which they are held accountable and responsible. At some point, governments need to do just that—take control of the policy process, push their agenda forward and make decisions. At some point, consultations need to end and governments need to make decisions about what course of action to follow and to communicate those decisions to the media and the public. Within this context, it is logical that governments might view openness and effective decision making as being incompatible. In a quip from the BBC series Yes, Minister a civil servant relates this point, stating, “You can have openness or you can have government” (Jay and Lynn, Reference Jay and Lynn1980).
However, this pattern of promising “open government” in opposition then offering some banal administrative reforms once in government may be facilitated by journalists’ commitments to defining open government in very limited ways, namely, in terms of access to information and sources. By equating open government with their own access to information and sources, journalists incentivize politicians—particularly opposition politicians—to cater to these demands while other, perhaps more creative or more meaningful reforms to make government more open go undiscussed.
This dynamic, whereby politicians make grand promises to increase open government in opposition while delivering limited reforms in terms of access to information once in government, may be highly problematic in two ways. First, in an era of widespread cynicism about politics, the prospect of having a political party say one thing in opposition and then do something completely different when it gets into power is likely to further fuel public cynicism about politics. Second, because of the ways in which open government can operate with competing conceptions, it is possible that the public may be denied a clear debate about the genuine possibilities for and limits to making government more open.
That said, our statistical analysis did identify a pattern whereby politicians of the left tended to prefer information as a definition of open government, while politicians of the right tended to prefer trade-offs. These ideological differences may apply cross pressures to and mediate the institutional imperatives discussed above, leading to genuine left–right variations in the extent to which open government can become a coherent set of reforms.
Conclusions and Future Research
The idea of open government is an important normative concept in democratic politics. Indeed, the expectation that government behaves in a transparent manner is a powerful assumption and may very well comprise a keystone for public expectations about politics and government in Canada and elsewhere. However, our findings suggest that key Canadian actors have very different things in mind when they speak of open government. Importantly, the evidence suggests that journalists, opposition politicians and government politicians prefer definitions of open government that are congenial to their interests. Notwithstanding the centrality of the concept, the lack of a common understanding (or definition) of what open government means in practice makes it susceptible to political manipulation and thus renders it less likely to serve the public interest.
In future research, the views of public servants regarding how they define and interpret open government should be investigated. While public servants obviously work at the crucial intersection of politics and administration, they also typically remain silent about their own preferences. Uncovering how they see these questions may help explain the success or failure of different attempts to institutionalize “open government.”
Acknowledgements
The authors gratefully acknowledge the advice of Dr. Shoja'eddin Chenouri (University of Waterloo), Dr. Neil Malhotra (Stanford University) and Dr. Jason Roy (Wilfrid Laurier University) for methodological advice and Dr. Rune Slothuus for comments at the 2016 meeting of the Midwest Political Science Association.
