Courts and Issue Attention in Canada

Objective: To inform international discourse about judicial countermajoritarianism, we assess whether decisions enhancing gay rights by Canadian courts increase the media’s attention to homosexuality and related topics. Methods: We ﬁrst collect a dataset of monthly counts of relevant articles published in two prominent Canadian newspapers and then estimate Markov-switching models to evaluate whether increases in media attention to homosexuality are coincident with judicial decisions enhancing gay rights in Canada. Results: Each of ﬁve landmark gay rights decisions is coincident with a period of heightened media attention to homosexuality. The data show that Canadian newspapers publish nearly twice as many relevant stories during these “active” regimes compared to “inactive” periods. Conclusions: Canadian courts can increase attention to issues in the national media. This result supports a dynamic view of the interaction between courts and democratic majorities in place of the static view of democracy endemic to normative discourse about judicial countermajoritarianism. ﬁrst attention to the issue spaces implicated in courts’ actions. Although Canadians more rapidly assimilated sexual minorities equal rights claims than Americans, Canadian courts’ decisions were among the political and social events and inﬂuences that drew national media attention to issues of homosexuality. This result demonstrates that Canadian courts can inﬂuence national politics by elevating the presence of issues in the national media that were previously excluded from political debate by established constellations of power, creating opportunities for further political change. The results also show that judicial inﬂuence on media issue attention extends beyond the United States to another advanced democracy, suggesting that constitutional courts may more generally play a role in elevating the salience of minority rights claims. Our ﬁndings demonstrate how acts of judicial countermajoritarianism can disrupt existing policy equilibria that disadvantage political minorities, activating new dimensions of policy conﬂict, and creating conditions for the emergence of new political majorities with different orientations toward rights claims. inference about the inﬂuence of constitutional courts in Canada on the nation’s issue agenda. However, these results support additional insights into the linkages between judicial decisions 77). predicts time point will We these results explain the basic pattern of Canadian media attention to homosexuality and related topics over the last three decades. In general, homosexuality is relatively low on the issue agenda (i.e., in the inactive state). However, media attention periodically shifts into a relatively heightened, active state. This issue attention framework is consistent with the theoretical expectations of Downs’s (1972) issue-attention cycle and United able to inﬂuence the by attention to issues related to its decisions, there are no prior assessments of the ability of other countries’ judiciaries role in shaping systemic issue attention. on the costs and of judicial review view of majoritarianism the critique of judicial countermajoritarianism. we taken some ﬁrst steps to assessing whether constitutional inﬂuence the media’s attention to issues related their decisions. In we have applied a research design used to evaluate the Supreme Court’s inﬂuence on the American media to the Canadian case. key and to gay or marriage equality shows each of ﬁve gay rights decisions signiﬁcantly associated (active regime) media attention to these issues. This result shows that constitutional courts Canada can motivate systematic media attention to an issue related to their decisions. Of course, we only considered a single issue analyzed linkages between media attention and public components the estimates of which regime is in existence at any given time point. In all of our estimations, the two-regime models were favored over higher-ordered regime models. Speciﬁcally, the three-regime models we tested clearly could not distinguish between which regime existed at various time points. In other words, the three-regime models had overﬁt the data. As a result, we report the more parsimonious two-regime models in this paper.

Around the world, courts play an important role in protecting minority rights in other-wise majoritarian political systems. Yet, there is a tension between majority rule and judicial review-the authority of courts to invalidate or veto legislation or executive actions inconsistent with fundamental law. This conflict is most famously articulated by Alexander Bickel (1986), who argued that "morally supportable . . . government is possible only on the basis of consent" (1986,20; emphasis in original); yet, when a court "declares unconstitutional a legislative act . . . it thwarts the will of representatives of the actual people of the here and now" (17). On the other hand, judicial review is a principal safeguard for minority rights in otherwise majoritarian systems (U.S. v. Carolene Products Co., n.4, 1938). Debate over this "countermajoritarian difficulty" is increasingly as prominent in international discourse about democratic constitutionalism (Hirschl, 2004) as it has been in American constitutional theory for the last half-century (Friedman, 2009).
Yet, theories of political communication and issue attention in political systems show concerns about judicial countermajoritarianism to be misdirected or overly simplistic. Concerns judicial countermajoritarianism proceed from a narrow view of democracy and majority rule. In particular, normative theoretical debates about the costs and benefits of judicial countermajoritarianism are typically cast as a contest between the interests of a static majority and the rights of a persistent minority. However, agenda-setting theory posits that there are often multiple possible, overlapping majorities (and minorities) surrounding in any political conflict that are activated (or not) by the specific definition and framing of political alternatives at any particular point in time (Bachrach and Baratz, 1962;McKelvey, 1976). The power to define problems, that is, the power to set the agenda, is the power to activate these latent majorities and reorient the cleavages of political conflict (Schattschneider, 1960).
A court that can influence its country's policy agenda, then, can escape the countermajoritarian difficulty by activating latent majorities in support of decisions that invalidate a law created by another majority (Ura 2014a, see also Ura 2014b).
Prior empirical research has shown that the Supreme Court of the United States has, in a few salient instances, made decisions that significantly increased national media attention to the issues involved in those cases (Flemming and Wood, 1997;Flemming and Bohte, 1999;Ura, 2009Ura, , 2014b. In particular, these studies have found that the Supreme Court raised the prominence of civil rights, free speech, religious liberty, and gay rights in the national media in recent decades. In turn, the prominence of these issues in the media has been reflected in the policy agendas of Congress, the president, and the states, with significant consequences for the development of partisan and electoral politics, particularly with respect to issues of race and civil rights (e.g. Carmines and Stimson, 1989;Hetherington, 2009;Klarman, 2004;Poole and Rosenthal, 1997;Perlstein, 2001;McMahon, 2011).
Although the U.S. Supreme Court decisions have played an important part in the ascension of some minority rights claims in the American media, the availability of new resources for collecting data on international media now makes it possible to evaluate whether judicial decisions amplify political attention to issues in national settings outside the United States. As a first step to assessing constitutional courts' agenda-setting power in a comparative context, we evaluate the dynamic association between Canadian constitutional courts' decisions affecting gay rights and media attention to homosexuality, gay rights, and marriage equality. While widespread interpersonal and institutionalized discrimination and violence against homosexuals and other sexual minorities persist around the world, in Canada and the United States there have been substantial advances in the legal and political standing of gay men and women in recent decades, including reductions in systemic harassment by police, repeal, or invalidation, of laws against same-sex sexual activity, legal recognition for same-sex partnerships, and permitting same-sex couples to marry. While the political dynamics that supported these advances varied substantially between the two countries, constitutional courts were (temporally) leaders in acknowledging and supporting rights claimed by sexual minorities in both Canada and the United States. Testing for media agenda-setting effects from Canada's constitutional courts extends empirical scrutiny of the theoretical claims of judicial influence on issue attention to a new national case that is similar to the United States in important ways (mostly shared language, common law legal heritage) to make meaningful comparisons, while also being different enough (organizational structure of constitutional judiciary, political culture) to provide an independent test of the generalizability of political dynamics first observed in the United States.
To test for judicial agenda-setting effects in Canada, we collect an original data set of monthly counts of articles published in two prominent Canadian newspapers mentioning key words and phrases indicating attention to these issues. We next aggregate counts across search terms and newspapers to generate monthly counts of published articles related to homosexuality, gay rights, or marriage equality. We then estimate a series of Markov-switching models to evaluate whether significant increases in media attention to homosexuality correspond to the resolution of constitutional cases or claims dealing with gay rights by Canada's national and provincial constitutional courts. The data support our theoretical expectations: significant increases in print media attention to gay rights and homosexuality are temporally associated with legally salient gay rights decisions by Canadian constitutional courts.
These results have several implications. They show first that Canadian judicial decisions that rearrange the prior distribution of material or symbolic benefits draw the media's attention to the issue spaces implicated in courts' actions. Although Canadians more rapidly assimilated sexual minorities equal rights claims than Americans, Canadian courts' decisions were among the political and social events and influences that drew national media attention to issues of homosexuality. This result demonstrates that Canadian courts can influence national politics by elevating the presence of issues in the national media that were previously excluded from political debate by established constellations of power, creating opportunities for further political change. The results also show that judicial influence on media issue attention extends beyond the United States to another advanced democracy, suggesting that constitutional courts may more generally play a role in elevating the salience of minority rights claims. Our findings demonstrate how acts of judicial countermajoritarianism can disrupt existing policy equilibria that disadvantage political minorities, activating new dimensions of policy conflict, and creating conditions for the emergence of new political majorities with different orientations toward rights claims.

Majority Rule, Minority Rights, and Judicial Review
Judicial review of national laws is implicit in the United States Constitution ratified in 1789 and part of American political practice since 1803 (Marbury v. Madison). Yet, national constitutions that permit courts to invalidate laws contrary to constitutional limits on legislative authority or grants of individual liberty were uncommon until the second half of the twentieth century. As late as 1942, only the United States and Norway authorized judicial bodies to declare national laws void for unconstitutionality (Guarneri and Pederzoli, 2002). By 2011, though, "83% of the world's constitutions [gave] courts the power to supervise implementation of the constitution and to set aside legislation for constitutional incompatibility" (Ginsburg 2014 p. 587; see also Hirschl (2004); Gardbaum (2009)).
Despite the spread of judicial review, empowering unelected judges to nullify laws enacted by elected legislators remains one of the most controversial institutional arrangements in modern constitutional theory. American legal scholar Alexander Bickel famously articulated the principal normative critique of judicial review: "The root difficulty is that judicial review is. . . counter-majoritarian," and, as a result, a "deviant institution in. . . [a] democracy" (1986,(16)(17)(18). This normative dilemma has been as a focal point for American constitutional theory for the last half-century (Friedman, 2009;Keck, 2007) and a growing concern for scholars of comparative constitutionalism over the last two decades (Guarneri and Pederzoli, 2002;Hirschl, 2004;Stone Sweet, 2000;Vanberg, 2000).
Although some scholars dispute characterizing judicial review as a countermajoritarian institution (Dahl, 1957;Hall and Ura, 2015;Klarman, 1997;Rogers, 2001Rogers, , 2009), many American constitutional theorists and legal scholars argue that courts' ability to act contrary to the will of the majority is a virtue, a feature and not a bug (e.g. Chemerinsky, 2004;Dworkin, 1989;Ely, 1980). From this view, judicial review provides a mechanism for remedying some public inequalities that disadvantage persistent minorities. In general, allowing courts to set aside laws that violate minorities' constitutionally protected rights or liberties, judicial review can subvert a "tyranny of the majority" (Guinier, 1994).
The rapid diffusion of judicial review around the world has globalized intellectual concern over judicial countermajoritarianism (e.g., Hirschl, 2004). Although normative debates about the tensions between majority review have been most developed in the context of American constitutional theory, the global spread of judicial review invites new normative and positive research into the political dynamics that connect and balance judicial behavior with majoritarian politics outside of the United States. Next, we consider how Canada's constitutional courts' influence on political communication can resolve countermajoritarian tensions by reorienting politic discourse to reveal latent political majorities.
We direct our attention to Canada for two reasons. First, it is geographically proximate and sociopolitically similar to the United States, a common law heritage, meaningful federalism, and majority English-speaking populations with substantial linguistic minorities. Second, despite these similarities, Canada's politics and its judiciary are sufficiently different from the United States to provide an independent test of the generalizability of political dynamics first observed in the United States. In short, Canada is similar enough to the United States to be comparable in the first place, but different enough from the United States to make comparisons between the two interesting.

Which Majority? Courts and Agenda Setting
Debate over the tension between majority rule and judicial protection of minorities' rights claims is as old as the the institution of judicial review itself. Yet, the formulation of the countermajoritarian difficulty in judicial review as a contest between the normative merit of the fixed political will of a static majority and the rights of a permanent political minority is at odds with positive political scientists' claims demonstrating the inherent instability of majority political coalitions and the impermanence of political cleavages (e.g. McKelvey, 1976;Schattschneider, 1960). This alternative view shows that there are multiple possible, overlapping majorities (and minorities) surrounding in any political conflict that are activated (or not) by definition and framing of political choices.
Time and attention are scarce resources, and a political system cannot engage all of the potential conflicts or concerns arising from society. Instead, out of the "billions of potential conflicts in any modern society, but only a few become significant" (Schattschneider, 1960, 66). Choices about which issues to consider have powerful implications. Moving an issue onto a political system's agenda is a necessary (but not sufficient) condition for policy change. Issues excluded from a systemic agenda will, by default, continue to be structured by prior policy decisions (or nondecisions) (Downs, 1972;Kingdon, 1984).
Individuals and groups vested in a standing policy arrangement may serve their interests by devoting energies to creating or reinforcing social and political values and institutional practices that limit the scope of the political process to consideration of only those issues which are comparatively innocuous" (Bachrach and Baratz, 1962, 948). Likewise, governing regimes may prevent changes to their preferred policies and institutional arrangements by suppressing the consideration of issues that might lead to the redistribution of symbolic or tangible values, including changes that would have been adopted in place of the status quo were they considered (Bachrach and Baratz, 1962;Baumgartner and Jones, 2009;Downs, 1972;Schattschneider, 1960). An important implication of this perspective is that formulations of democratic political conflicts as contests between majority interests and minority rights, such as those that animate normative debates over judicial countermajoritarianism, are incomplete.
The composition of a political majority depends on what questions the polity is asked to consider and how those choices are framed. An institution that can influence the composition of its system's issue agendas has at least some power to define the cleavages that divide majorities from minorities in the first place. Courts that can influence the content or framing of policy agendas may be able to escape the countermajoritarian difficulty by activating a previously latent support for the policy implications of a decision to invalidate a law created by a prior majority (Ura, 2014b).
This alternative view of the relationship between judicial review and majoritarian politics raises an important empirical question: Can courts influence the composition of their systems' issue agendas? Evidence from the United States suggests that the answer is conditionally "yes." In particular, political scientists have found that US Supreme Court decisions that disrupt established policy regimes draw the national media's attention to the issues involved in those cases (Flemming, Bohte, and Wood 1997;Flemming, Wood, and Bohte 1999;Ura 2009Ura , 2014; but see Rosenberg 1991. These studies find that some salient Supreme Court decisions significantly raised the prominence of civil rights, civil liberties, free speech, religious liberty, and gay rights in the American national media during the last half century. For example, Brown v. Board of Education increased media attention to racial segregation in schools (Flemming and Wood, 1997) and The New York Times and USA Today persistently increased publication of stories related to homosexuality by more than one third following Lawrence v. Texas in 2003(Ura, 2009. While the Supreme Court has merely one voice among many in shaping the American media's issue agenda, there is convincing evidence that the Court has played an important role in increasing media attention to some important issues in recent American history.
Yet, it remains unknown whether constitutional courts outside the United States exercise similar influence over systemic issue attention. As courts continue to play a growing role in national politics around the world, it is increasingly important to evaluate whether courts in other national contexts can shape public agendas. Here, we take a first step toward addressing this important problem by evaluating the dynamic association between Canadian constitutional courts' decisions affecting gay rights and media attention to homosexuality, gay rights, and marriage equality. In particular, we test whether Canadian constitutional court decisions that expanded the scope of gay rights increased Canadian media attention to homosexuality, gay rights, and related topics.
We focus on issues related to homosexuality and the political and social rights of homosexuals and same-sex couples in Canada for several reasons. Over the last few decades, Canada and the United States have substantially enhanced the legal standing of gays and lesbians, include reductions in systemic harassment by police, repeal or invalidation of laws against same-sex sexual activity, providing legal recognition to same-sex partnerships, and permitting same-sex marriage. Although these developments generally occurred earlier in Canada than in the United States, constitutional courts at both the national and subnational levels were (temporally) leaders in acknowledging and supporting rights claims by sexual minorities in both countries. Given established evidence of the U.S. Supreme Court's influence on media attention to homosexuality (Ura, 2009), a parallel analysis of Canadian politics permits a test of theories of judicial agenda-setting effects in a comparative context, while preserving the ability to make meaningful comparisons with the United States.

Assessment
We implement a test of judicial influence on issue attention in the Canadian media using the same basic research design utilized throughout the literature evaluating the U.S. Supreme Court's influence on issue attention in American national media (e.g., Flemming, Bohte, and Wood, 1997;Flemming, Dan Wood, and Bohte, 1999;Rosenberg, 1991;Ura, 2009Ura, , 2014b. First, we identify a set of legally salient Canadian court decisions involving the legal rights of homosexuals or same-sex couples. Then, we measure the level of media attention to issues related to homosexuality, gay rights generally, or same-sex marriage in particular in nationally visible daily newspapers with large circulations as a proxy for the larger media space. Finally, we estimate changes in the level of media attention to homosexuality and related topics at the time of the identified judicial decisions. Evidence of significant increases in media attention to homosexuality coincident with salient court decisions is taken as evidence of a court's influence on systemic issue attention.

Identifying Relevant Cases
To identify relevant cases, we reviewed several prominent accounts of the historical evolution of gay rights in Canada (Herman, 1994;Pierceson, 2005;Smith, 1999Smith, , 2008. These studies highlight a common set of five landmark decisions that expanded the scope of homosexual individuals and couples' material or symbolic rights under the law. These are: Egan v. Canada (May 1995), Vriend v. Alberta

Egan v. Canada (1995)
In 1986, James Egan applied to receive old age security pension funds as the spouse of his long-term same-sex partner, John Nesbit. Upon application for the entitlement, the Canadian government denied the entitlement on the basis that Egan did not qualify as a "spouse," as he was in a same-sex relationship. The claim progressed to the Supreme Court of Canada. A divided bench dismissed Egan's appeal. However, a majority of the Court did find that while the definition of "spouse" in the Old Age Security Act was constitutional, sexual orientation was a prohibited ground of discrimination pursuant to the Canadian Charter of Rights and Freedoms. As a result, though James Egan his specific case, the Canadian Supreme Court for the first time found that sexual orientation was a protected status under the law.

Vriend v. Alberta (1998)
In 1991, King's College, a private religious college in Alberta, fired Delwin Vriend for being in a same-sex relationship. After termination, Vriend attempted to submit a complaint for discrimination to the Alberta Human Rights Commission, however the governmental agency refused to investigate the matter. The agency contended that the relevant Alberta law did not encompass protections on the basis of sexual orientation. In April of 1998, ruled that the exclusion and legislative omission of sexual orientation as a protected status under provincial law violated the Canadian Charter of Rights and Freedoms.

M v. H (1999)
In this case, a lesbian couple who had been together for a decade separated. One of the individuals in the relationship requested a courtordered partition and sale of the joint assets, including a home. This individual also asked the court to award the equivalent of spousal support. With no formal marriage, the individual seeking support claimed a common law relationship, thereby serving as the basis for the "spousal support." After appeals, the Supreme Court of Canada ultimately found that the exclusion of same-sex couples from the Ontario Family Law Act's definition of "common-law spouse" constituted a violation of the Canadian Charter of Rights and Freedoms. The ruling itself did not alter the legal definition of "marriage." Rather, it extended legal rules about the division of property and partner-support to common law same-sex cohabitating partners. Seven gay and lesbian couples applied for marriage licenses with the Clerk of the City of Toronto. Ultimately, the lower court determined the couples were entitled to be married and prohibition of such constituted a violation of their rights under the Canadian Charter of Rights and Freedoms. On June 10, 2003, the Ontario Court of Appeals found the common law definition of marriage, which was defined as between one man and one woman, violated the Canadian Charter of Rights and Freedom.

Reference Re Same Sex Marriage (2004)
By 2003,the Liberal government in Canada had moved to make same-sex marriage legal by statutory enactment. Ultimately, the Canadian Parliament prepared a federal law extending the definition of marriage to include same-sex couples. Unlike the U.S. Supreme Court, Canada's Supreme Court can issue advisory opinions on the constitutionality of legislation in advance of its enactment and effect. The Canadian government submitted the proposed bill to the Canadian Supreme Court to determine if it was consistent with the Canadian Charter of Rights and Freedoms. On December 9, 2004, the Canadian Supreme Court determined that the proposed law was permissible. Thereafter, the Canadian Parliament enacted the legislation extending marriage to same-sex couples.

Measuring Issue Attention
Testing our theory requires longitudinal data on the volume of media attention to topics relating to homosexuality, gay rights generally, and same-sex marriage in Canadian newspapers. We build on Ura's (2009)  . To identify pertinent articles, we utilized Python and the extension selenium to automate the submission of Boolean search queries to Lexis-Nexis for a series of topical keywords: gay, gays, gay marriage, gay rights, homosexual, homosexuals, homosexual rights, lesbian, lesbians, same sex marriage, and marriage equality. We then generate discrete monthly counts of articles containing each search term for each newspaper. Finally, we generate an aggregate measure of attention to homosexuality, gay rights, and same-sex marriage by summing the monthly counts for each newspaper. Figure 1 shows the monthly article count for each newspaper, and Figure 2 shows the final, combined monthly article count series. Table 1 reports summary statistics for the Toronto Star, Globe & Mail, and the combined newspaper counts. 1 1 The aggregated monthly counts (of both Canadian newspapers collapsed into a single measure) provide a macroview of the Canadian media environment's attention to homosexuality and related issues. One limitation of our measurement approach, though, is that individual articles with multiple keywords will be included in each keyword's article count. For example, an article that includes the terms "gays" and "lesbians" will count as one article in each of those keyword's series.
Our view is that this is the most appropriate macro-level measure of media issue attention, since an article with multiple keywords represents broader engagement of homosexuality or related topics than an article with only one keyword. However, this choice "averages away" variance in emphasis on different aspects of homosexuality as a political and social topic; for example, attention to "gay rights" generally as opposed to "same sex marriage" in particular. In order to evaluate the consequences of our decision to analyze the media count data at the highest level of aggregation, we also perform a principle-components factor analysis with varimax rotation on the search-term series. This analysis reveals five factors with Eigenvalues greater than one: (1) a factor that corresponds to search terms from the Globe & Mail, (2) a factor associated with search terms from the Toronto Star, (3) a factor that loads highly counts of articles related to marriage, (4) a factor that related to general, cultural attention to homosexuality, and (5) a factor that corresponds to terms invoking rights. (The factor loadings are reported in the Appendix.) Since the factor analysis indicates that our aggregate measure combines several distinct dimensions of attention to homosexuality, we repeat the Markov-switching dynamic regression (MSDR) analysis we describe below for each of the retained factors. The MSDR estimates on these more-focused issue attention measures support the same substantive inference about the influence of constitutional courts in Canada on the nation's issue agenda. However, these results support additional insights into the linkages between judicial decisions

Estimation
Our aim is to determine if the Canadian courts acted as policy agenda-setters when they issued their substantive rulings on gay rights and same-sex marriage. To do so, we employ a statistical method that can identify policy issues flow in an "attention cycle" (Downs, 1972), a process of punctuated equilibrium in which "inactive" issues become "activated" (Baumgartner and Jones, 2009). Markov-switching dynamic regression (MSDR) models are an appropriate modeling strategy for identifying shifts between active and inactive issue agenda states in the media. MSDR models permit the usage of time-varying parameters contingent on the existence of a particular (unobserved) latent state of existence (or, "regime"). In the context of our study, we expect that if courts act as agenda-setters, we should observe the media respond to court decisions with increased attention.
For Markov-switching models, consider a time series, y t , where t = 1, 2, ..., T and is characterized by two (simple) 2 states: State 1: y t = µ 1 + t State 2: y t = µ 2 + t where µ 1 and µ 2 are the intercept terms in state 1 and state 2, respectively. t is white noise error with variance σ 2 . In the above equation, the two-state model shifts intercepts and is expressed as: where s t is 1 if in State 1; 0 otherwise. Because we do not know with certainty which state ("regime") we are in, we can calculate the probabilities of being in a particular state (see, e.g., Hamilton (1989)) and update the likelihood for each period (with a non-linear algorithm).
For Markov-switching models, key to testing our concept of oscillating "regimes" is the ability to transition into, and out of, these distinct latent states. The probability that s t is equal to j ∈ (1, ..., h) depends on the most recent realization of the latent state, s t−1 . Thus: In a Markov-switching model, the Transition Matrix (Q) is structured to permit movements throughout the distinct regimes, such that: Here, we utilize shifting intercepts and a basic model, though more complicated estimations with covariates and time-varying parameters can be utilized The innovation for Markov-switching models to agenda-setting and issue-attention studies is that it permits us to evaluate whether oscillations in issue attention at particular points in time represent significant shifts between regime states. Those oscillations permit movement through various "regimes," as well as reversions back to previous states of existence following the decline of an activation mechanism. With respect to our study, this method will we identify any movement from a "low regime" (where the media is generally not focused on gay rights) into a "high regime" (where the media focuses on gay rights when courts issue rulings). We estimate a series of MSDRs (with switching-intercepts) for the relevant Canadian gay rights court cases on the aggregate Canadian newspaper series. 3

Results
The data show, first, that there are, in fact, two significantly different issue attention regimes for homosexuality in the Canadian media, a less attentive state we will refer to as the inactive regime and a more attentive state we will call the active regime ( Table 2). The model predicts that our two national print media outlets will publish just over six hundred articles featuring keywords related to homosexuality, gay rights, or marriage equality each month when they are in the inactive regime. In contrast, during an active regime, the model predicts that the two papers will generate roughly twelve-hundred articles per month. A shift from the inactive regime to the active regime, therefore, effectively doubles the Canadian media's attention to homosexuality and related topics.
We can further extract the probabilities of transitioning from the inactive regimes to the active regime from the Q-Transition Matrix. In our aggregate issue attention data, the inactive regime is dominant (p 11 = 0.94). Specifically, the model predicts that there is a 0.94 probability that a time point in the inactive regime will be followed by another time point in the inactive regime. However, the active regime is also observed with some frequency and, when it emerges, shows persistence as well (p 22 = 0.77). The model predicts that a time point in the active state will be followed by another time point in the active state with probability 0.77. We can further calculate the expected duration of an instance of each regime across all time points. 4 The expected duration of an inactive regime is 16.42 months, and the expected duration of an active regime is 4.39 months.
Together, these results explain the basic pattern of Canadian media attention to homosexuality and related topics over the last three decades. In general, homosexuality is relatively low on the issue agenda (i.e., in the inactive state). However, media attention periodically shifts into a relatively heightened, active state. This issue attention framework is consistent with the theoretical expectations of Downs's (1972) issue-attention cycle and 3 We report smoothed-probabilities; see Appendix for a discussion of smoothed and filtered probabilities. 4 See the Appendix for discussion of this calculation. So, do shifts from the inactive regime to the active regime correspond to Canadian courts' gay rights decisions? Can courts influence the the composition of their systems' issue agendas? Our MSDR estimates allow us to generate predicted probabilities that media attention to homosexuality and related topics is in the inactive state for each observed time period. (Because there are only two regime states, the probability that the time period is in the active regime is the complement of the inactive state probability.) These predicted probabilities are illustrated in Figure  Out of 364 observed months, the model predicts that only 78 months (21.43 percent) are in the active state. Five of these months are those in which a Canadian court decided one of the identified gay rights cases. In other words, 6.41 percent of all months in which the Canadian media's attention to homosexuality was in an active state were months in which one of the five legally salient cases we identified were decided, and 100.00 percent of the months in which a legally salient gay rights case was decided by a Canadian court were months in which the media's attention to homosexuality was in an active state. Recalling that the Canadian media generates roughly double the number of relevant articles in the active regime compared to the inactive regime, these estimates indicate that Canadian courts have statistically and substantively significant influence on the country's media agenda. When courts in Canada have acted to expand the scope of gay men's and lesbians' symbolic or material rights under Canada's constitution or the Charter of Rights and Freedoms, the nation's media have responded by turning their attention to the relevant issue space.
The data show that constitutional courts in Canada can increase their national media's engagement of issues addressed by the judiciary in a way that parallels the influence of the United States Supreme Court's influence on American print media. Each of the five landmark gay rights cases decided by Canadian courts that we identified corresponds to a period of significantly greater publication of articles involving homosexuality, gay rights, or marriage equality in the two most widely circulating newspapers in Canada. These active issue attention regimes represent periods in which the number of articles containing relevant key words is roughly double the number typically published in the baseline, inactive regime. Although we consider only a single salient issue domain here, this result indicates that constitutional courts in Canada may be able to exercise some control over the systemic political agenda in which they operate, providing a mechanism to escape the dilemma of judicial countermajoritarianism by reshaping the issue agenda that structures public reactions to their decisions.

Conclusions
Since the 1940s, judicial review has grown from an institutional oddity to a ubiquitous feature of contemporary constitutional government. The globalization of judicial politics has spread an institutional arrangement that can protect minority rights under reasonable political circumstances (Clark, 2009;Ura and Wohlfarth, 2010;Whittington, 2005). It has also introduced an institution facially inimical to the principle of majority rule that is a cornerstone of democratic theory (Bickel 1986). Although this tension is often dormant in applied political cases, countermajoritarian judicial decisions can inspire intense political backlash among supporters of the losing side. These dynamics are evident, for example, in intense, often violent, reactions among white majorities in southern states where laws requiring racial segregation in public education were invalidated by Brown v. Board of Education (1954;Klarman 2004).
Yet, judicial countermajoritarianism is not an unescapable trap. A classic body of theory in political science demonstrates that the composition of political majorities and minorities depends on the nature and framing of issues under consideration by a political system (Bachrach and Baratz, 1962;McKelvey, 1976;Schattschneider, 1960). Majorities can be reshaped by redefining the scope and nature of political conflict, and institutions that can influence the content of systemic issue agendas may be able to support minority rights claims by activating latent majority support for them. Dynamic agenda-setting theory provides a positive theoretical challenge to understanding political conflict as a contest between a stable majority and a persistent minority, which is an implicit premise of normative theories of a judicial countermajoritarian difficulty.
Although there is evidence that the Supreme Court of the United States is able to influence the media by drawing attention to issues related to its decisions, there are no prior assessments of the ability of other countries' judiciaries role in shaping systemic issue attention. This ties the increasingly international debate on the costs and benefits of judicial review to the narrow view of majoritarianism endemic to the critique of judicial countermajoritarianism. In this article, we have taken some first steps to assessing whether constitutional courts outside of the United States can influence the media's attention to issues related to their decisions. In particular, we have applied a research design used to evaluate the U.S. Supreme Court's influence on the American media to the Canadian case.
Our analysis of an original data set of monthly counts of articles published in two prominent Canadian newspapers mentioning key words and phrases related to homosexuality, gay rights, or marriage equality shows that each of five legally salient gay rights decisions is significantly associated with heightened (active regime) media attention to these issues. This result shows that constitutional courts in Canada can motivate systematic media attention to an issue related to their decisions. Of course, we have only considered a single issue and have not analyzed linkages between media attention and public opinion in this article.
Moreover, our data show that courts are likely not the only influence on issue attention in Canada. (A majority of time active agenda months do not correspond to the timing of Canadian courts' gay rights decisions). However, these findings show the dynamic view of issue conflicts surrounding the United States Supreme Court and its decisions applies to Canada as well (e.g. Flemming and Wood, 1997;Flemming and Bohte, 1999;Ura, 2009).
Our analysis also suggests several fruitful avenues for future research. First, evidence that Canada's constitutional courts increased media attention to homosexuality and related topics invites inquiry into whether Canadian courts have influenced the media's attention to other issues. Likewise, our data collection and analytical methods can be used to collect issue attention data in any country where national media content is systematically archived online. This will enable studies of judicial influence on issue attention in other countries as well as genuinely comparative research that evaluates variance in the magnitude of courts' agenda-setting influence as a function of institutional arrangements, media structure, political culture, and other factors.
Lastly, our results suggest additional research on the role of media in linking courts to dynamics in public opinion (e.g. Marshall, 1989;Johnson and Martin, 1998;Brickman and Peterson, 2006;Ura, 2014a;Manzano and Ura, 2013;Stoutenborough, Haider-Markel, and Allen, 2006;Franklin and Kosaki, 1989). Finally, comparative studies of judicial politics may help identify the pathways over which judicial influence on the media passes: Are constitutional courts convenient focal points for media coverage in some political cultures, or does the structure of some judicial institutions provide some structured influence over attention to issues in national political systems? Cross-national variance in institutional design, political and legal cultures, and media environments may provide important analytical leverage for answering these and other important questions about links between courts and political communication around the world.

Appendix A: Factor Analysis Indices
As we mention in the article, one concern with our aggregated article count measure of media attention to homosexuality, gay rights, and marriage equality is that our counting procedure will count an article separately for each keyword it contains. For instance, an article that contains the words "gays" and "lesbians" is counted as one article in each keyword's total for the month it is published. We believe that this is most appropriate macro-level measure of media issue attention, since an article with multiple keywords represents broader engagement homosexuality or related topics than an article with only one keyword.
However, this choice "averages away" variance in emphasis on different aspects of homosexuality as a political and social topic; for example, attention to "gay rights" generally as opposed to "same sex marriage" in particular. In order to evaluate the consequences of our decision to analyze the media count data at the highest level of aggregation, we also perform a principle-components factor analysis with varimax rotation on the search-term series. Doing so yields five retained factors (Table 3): (1) Globe & Mail-specific factor, (2) Toronto Star-specific factor, (3) marriage equality factor, (4) sexuality factor, and (5) rights and equality factor. In order to determine if the Canadian courts have influence within these distinct factors, we performed MSDR estimations (with shifting intercepts) on each of the three retained substantive factor series, i.e. those associated with particular dimensions of the global homosexuality topic rather than a specific publication. Below we first report and describe the results of our principal components factor analysis and then MSDR estimates for the three disaggregated issue series.

Factor Analysis and Factor Loadings
We employed a principal-components factor analysis with varimax rotation on the search-term count series to create factor indices. In the unrotated factor analysis returns the eigenvalues reported in Table A1. Table A2 lists the unrotated factor loadings for each of the search term series from both the Globe & Mail (GM) and Toronto Star (TS). Table A3 lists the rotated factor loadings for the various indices.
We also computed Cronbach's α for the resulting factor indices. Those scores are contained in Table A4. Table A5 contains the summary statistics for the indices.
In Figures A1-A3 we present the time series plots for each of our substantive issue-area indices (we exclude the newspaper-specific factors).

Marriage Index
The MSDR estimation of the Marriage Index again reveals the existence of two regimes (Inactive and Active regimes). The results are contained in Tables A6 and A7.        We obtain similar results with the Marriage Index that we did on the aggregated newspapers series. The Inactive Regime lasts longer on average than the Active Regime. Indeed, the expected durations are significantly farther apart than on the aggregated series. Plotting the smoothed probabilities with the Canadian cases provides us with the result in Figure A4. In Figure A4, we see that our model of the Marriage Index identifies two regime switches consistent with Canadian court cases. These switches correspond to the Halpern and Reference Re Same Sex Marriage decisions. The results have intuitive appeal: these two cases were the most centered on the question of same-sex marriage. Accordingly, the fact that our model identifies these regime switches on the Marriage Index is precisely what one would expect, as the prior court cases were not as intimately associated with the question of   "marriage." Again, our model correctly identifies a shift in media attention (in our indice) and coordinately a shift in policy discussion through the mechanism of court decisions.

Homosexuality Index
The MSDR estimation of the Homosexuality Index again reveals the existence of two regimes (Inactive and Active regimes). The results are cloudier, however, for this index. Results are contained in Tables A8 and A9.   Figure A5.
In Figure A5, we see that our model of the Homosexuals Index identifies two regime switches. These switches correspond again to the Halpern and Reference Re Same Sex Marriage decisions. While these results are consistent with what we would expect, the other switches are not consistent with the cases, which we would have generally expected to an extent. Indeed, the smoothed probabilities indicate that this time series actually begins in the Active Regime, something we had heretofore not seen. Accordingly, while the results confirming a switch in regimes to the Active state for Halpern and Reference Re Same Sex Marriage decisions are consistent with our expectations, it is possible that the Homosexuality Index is a bit cloudier given that it appears to pick up on cultural matters that could be driving some of the non-results in the earlier portion of the time series. Nonetheless, the results do signify that courts can lead the media's issue agenda along with other important political, social, and cultural events.  With respect to the Rights/Equality Index, we see the Inactive Regime again lasts longer on average than the Active Regime. With this series, however, the expected duration of the Inactive Regime is on average approximately 17 times longer than the Active Regime. In other words, the movement to the Active Regime is truly punctuated, with quick reversion to a lower information state of existence. We see an interesting result, however, when we plot the smoothed probabilities against the Canadian court cases (as shown in Figure A6). In Figure A6, we see that our model of the Rights/Equality Index identifies one distinct regime switch. This switch corresponds to the Egan decision. Recall that Egan was the first time a Canadian appellate court found that sexual orientation was a prohibited basis of discrimination. Very clearly, Egan was, at its core, a statement about gay and lesbian rights and equality in Canada. Accordingly, we would expect that the Rights/Equality Index would capture the shift in media attention due to Egan. Additionally, we see slight movement away from the Inactive Regime and toward a heightened level of attention to gay rights and equality with the Vriend decision. This case also addressed the protected status of sexual orientation against discrimination. It is certainly the case that Vriend did not effectuate what would appropriately be called a regime switch (the probability of being in the Inactive Regime does not cross p = .50). However, it does suggest that, speaking probabilistically, there was an increased likelihood of the media environment shifting the policy discussion to that of rights and equality.

Concluding Thoughts on the Indices
What is clear from the MSDR estimations on the individual substantive indices is that even if we move to a distinct measurement of media attention (and away from our aggregated counts measure), we still find that Canadian courts (like their American counterparts) can act as agenda-setters. Indeed, our estimations find regime switches occurring consistent with the Canadian court decisions. But there is more. Putting the indices together also reveals an interesting change in the Canadian media's narrative concerning gay rights and same-sex marriage. Initially the media centered its attention on themes of rights and equality for gays and lesbians (as seen in the Rights/Equality Index). However, as time progressed, this narrative morphed to general considerations of homosexuality (including in a cultural sense), and finally onto the specific discussion of "marriage." Tracking the movements of the narratives from each index across time suggests that not only can courts inject a general policy discussion into the media environment, but additionally that courts can actually frame that discussion with the nature of their decisions. The evolving nature of cases that originally centered on employment discrimination and benefits/entitlements for gays and lesbians gave way to a frame about marriage itself. We do not suggest that courts frame through sheer tyranny of will. Clearly, the cases that come before the courts are driven in large measure by the strategic choices litigants make. But, the frames employed by the courts in their decisionmaking can influence the nature of the media's attention to the policy discussion, and thereby cause an agenda-setting at least sculpted by judicial institutions. models between "smoothed" and "filtered" probabilities is small, though it bears mentioning why we illustrate the smoothed probabilities rather than the filtered probabilities for our graphical displays. The difference between the filtered and smoothed probabilities is that the former represent the latent state of existence (s t ; or "regime") up to time period t, conditional on the data from the start of the time series up to time t. The smoothed probabilities, however, represent the probability of being in a particular latent state (s t ), conditional on all data points in the series from t n -t 1 . Our estimations (as employed by STATA 14), rely on Kim's (1994) innovation of backwards sampling of the forward-filtered s t probabilities. The advantage of this computation of s t is that it uses all available information within the series and then works backwards to create smoothed probability calculations using every data point from t n through t 1 . For a discussion of the methods employed by Stata 14, see http://www.stata.com/manuals14/tsmswitch.pdf#tsmswitchMethodsandformulas.

Appendix C. Expected Duration Calculations
To calculate the expected durations of the regime lengths, averaged across the time points, the formula for doing so for each distinct regime 5 is: