Reassessing the Supreme Court: How Decisions and Negativity Bias Affect Legitimacy

Abstract

While the Supreme Court’s legitimacy is generally considered essential to its influence, scholars continue to debate whether the Court’s decisions affect individuals’ assessments of it. The last week of the 2013 term provides an unusual opportunity to evaluate these issues because the Court made a conservative decision concerning the Voting Rights Act (VRA) only one day before it made a liberal one about same-sex marriage. We use original panel data of individuals’ views throughout this period, including a wave collected on the day between the two decisions, to investigate the links among decisions and legitimacy. We find that diffuse support for the Court is sensitive to decisions in these two salient cases conditional on individuals’ ideological distance to the Court and their policy support. Moreover, the negative effects of an unfavorable decision are stronger than the positive effects of a favorable one.

Keywords

public opinion judicial processes and institutions legitimacy gay marriage voting rights

Political observers have long emphasized the importance of perceptions of institutional legitimacy to the Supreme Court. It is generally thought that legitimacy allows the Court the flexibility to make unpopular decisions without fear of losing its influence. While partisanship and ideology have been shown to be powerful explanations of everything from positions on political figures, institutions, and policies to established facts in the political behavior and psychology literatures (e.g., Campbell et al. 1960; Christenson and Kriner 2017; Jerit and Barabas 2012; Taber and Lodge 2006; Zaller 1992), similar findings that identification or attitudes condition perceptions of the Court are relatively scarce. Indeed, the prevailing view in the scholarly literature has been that diffuse support for the Court is not materially affected by ideological agreement with the Court’s decisions (e.g., Gibson 2007). Consistent with the prominence of partisanship and ideology broadly in American politics, this established view of stable Court legitimacy in the face of salient and contentious decisions has been challenged recently (e.g., Bartels and Johnston 2013; Christenson and Glick 2015a). While both sides of the debate offer nuanced arguments (see, for example, Gibson 2015; Gibson and Nelson 2015), the basic questions of whether and how diffuse support responds to the Court’s decisions remain.

Despite great attention to the Court’s legitimacy, few works have endeavored to compare the effects of different cases. Changes over time in the composition of the Court, in public opinion toward political institutions, and in the salience of cases make arriving at a general understanding of the micro-foundations of legitimacy extremely challenging. As a result, a number of questions have been left unanswered pertaining to the conditions under which we should expect to see a change in public opinion around case decisions.

In particular, is ideology the sole moderating factor of case decisions on legitimacy? Or does the public update assessments of the Court in ways consistent with its policy views, and regardless of the issue? Finally, are pleasing and displeasing decisions equally effective? In short, how do various underlying attitudes interact with Court decisions to impact the public’s perceptions of the Court’s legitimacy?

Questions pertaining to legitimacy around the Court’s decisions were particularly prevalent in June of 2013, when the Court struck down important components of both the Voting Rights and Defense of Marriage Acts. Not only did this week feature two salient decisions from the same Court on two different days, but it also featured one decision that upset liberals (voting rights) and one that upset conservatives (marriage equality). Thus, this period provides a rich opportunity to test whether and how members of the public update their assessments of the Court around decisions.

We capitalize on the fact that the Court rendered these decisions in sequence while we were collecting panel data to address the fundamental questions of the legitimacy debate. Our original dataset includes multiple waves before the decision week, multiple waves after it, and, most importantly, a wave collected in the twenty-four hours between the two key decisions. Together, the research design and the particular nature of these cases offer the ability to observe the causal effects of decisions at the individual level with substantial external validity. Indeed, the data provide an unusual opportunity to test whether individuals’ assessments of the Court’s legitimacy wax and wane following qualitatively different decisions.

The Legitimacy Debate

Going back at least a couple of decades, a cadre of scholars have argued that the Supreme Court’s legitimacy is high and stable (see, for example, Caldeira and Gibson 1992; Gibson 2007; Gibson, Caldeira, and Baird 1998). According to this view, the Court can rely on a reservoir of diffuse support that insulates it from unpopular decisions. Moreover, according to Positivity Theory, the underlying behavioral model, the public is exposed to decisions alongside symbols and reminders that the Court is a legalistic institution and, therefore, “different” from the other branches. This exposure means that even unwanted decisions can reinforce the Court’s legitimacy, though they may temporarily challenge people’s confidence or “specific support” (Gibson 2007; Gibson & Caldeira 2009a, 2009b, 2011). Thus, the dominant view in the literature has been that legitimacy is stable even in the face of controversial salient decisions. Perhaps the strongest illustration of this claim is that even Bush v. Gore (2000) did not appear to affect diffuse support (Gibson, Caldeira, and Spence 2003). Importantly, the claim is not that diffuse support for the Court can never change, but that only a run of unpopular decisions, and not one case, can affect it (Gibson and Caldeira 2009a).

Recent work has found evidence that diffuse support can be affected by ideological disagreement with the Court, calling into question the sources of legitimacy. While some of this work uses dependent variables that may be closer to specific rather than diffuse support (Durr, Martin, and Wolbrecht 2000; Egan and Citrin 2011; Hetherington and Smith 2007; Hoekstra 2000), others challenge Positivity Theory using essentially the same measures of legitimacy (Bartels and Johnston 2013; Christenson and Glick 2015a). Durr, Martin, and Wolbrecht (2000) show that divergence between the ideological content of the Supreme Court’s outputs and aggregate ideology affects support for the Court. Bartels and Johnston (2013) focus on the proximity of individual subjective assessments of the Court’s ideology and find strong evidence that diffuse support is a function of individual-level ideological congruence. Christenson and Glick (2015a) find that diffuse support decreases among individuals whose updated assessments of the Court’s ideology move away from their own ideology after observing a decision, and that it increases among those whose assessments of the Court’s ideology move toward their own. Bolstering these challenges, Sen (2015) reevaluates legitimacy theory in the judicial nominations context (see, for example, Gibson and Caldeira 2009b) and provides additional evidence that ideology rather than factors such as qualifications affect perceptions of potential justices.

Gibson and Nelson (2015, 34) offer a number of critiques of key variables and of the experimental manipulations used in the most recent studies, concluding that legitimacy is hardly, if at all, affected by ideological disagreement: “The Court’s legitimacy seems not to be grounded in policy agreement with its decisions, nor is it connected to the ideological and partisan cross-currents that so wrack contemporary American politics.” In sum, the current state of the literature is unclear, given two perspectives that yield opposing predictions about diffuse support when the Court makes important decisions. Positivity Theory predicts individual-level stability while its challengers predict systematic change.

We hypothesize that the evidence from current and real decisions will support the latter perspective. That is, we expect (Hypothesis 1; H1) that legitimacy will be sensitive to outputs, such that individuals’ assessments of the Court’s legitimacy will change based on the Court’s decisions and conditional on their underlying views. This expectation would square the legitimacy literature with other work in political behavior and psychology. More specifically, we focus on political attitudes moderating the impact of observable events (i.e., Court decisions) on deeper views about the institution. We draw parallels to the motivated reasoning literature, which shows how political attitudes condition the revelation of new facts (e.g., Schaffner and Roche 2016; Taber and Lodge 2006), and to very recent work showing that partisan associations also condition views about institutions and their use of policy tools (Christenson and Kriner 2017).

More specifically, we consider three manifestations of the sensitivity to outputs expectation that are not mutually exclusive. Conceptually, we focus on attitudes moderating the effect of the immediately observable Supreme Court action (the decision) on long-run legitimacy assessments. The first (H1A) is partisanship. Partisanship has been shown to shape understandings of and reactions to political events (e.g., Bartels 2002), and, as such, we expect it to moderate the effect of Court decisions as well. The second (H1B) is ideological distance. Building on the findings in Bartels and Johnston (2013), partisanship may matter less than how far one perceives the Court’s ideology relative to oneself—for example, two strong Republicans may react differently to the same Court if one perceives the Court to be relatively liberal and the other perceives it to be moderately conservative. In this model, new decisions provide information for people to update their perceptions of the Court and legitimacy assessments follow (Christenson and Glick 2015a). It is potentially a more forward-looking mechanism as people use individual decisions to assess whether the Court will be with them or against them in the future. The third closely related mechanism (H1C) is issue-specific attitudes: people’s case-specific policy outcome preferences condition changes in their legitimacy assessments. In some instances, issue-specific preferences may so closely align with partisanship or ideology as to be virtually indistinguishable. In other cases—such as marriage equality, where Republicans’ views, in particular, are less uniform (Kelly 2014)—the potential moderators of a decision’s impact may be distinct.

Related work shows that issue opinions can update with individual Supreme Court decisions (e.g., Bartels and Mutz 2009; Christenson and Glick 2015b; Franklin and Kosaki 1989; Mondak 1994; Stoutenborough, Haider-Markel, and Allen 2006) and after repeated decisions in the same area (Brickman and Peterson 2006). Likewise, there is empirical (Christenson and Glick 2015a) and theoretical (Mondak and Smithey 1997) work that distinguishes individual from aggregate sensitivity, and salient cases from routine ones, to suggest ways to reconcile the contradictory findings in the literature. While our expectations side with the revisionists who find a link between outputs and legitimacy, a strong body of work (above) offers a compelling null hypothesis. This alternative prediction emphasizes stability, such that individuals’ assessments of the Court’s legitimacy will be unaffected by individual decisions.

Assuming that individual decisions affect legitimacy, it is important to consider whether the effects of pleasing and displeasing outputs are symmetrical. In many contexts, people are more sensitive to negative information (Fiske 1980; Lau 1985; Pratto and John 1991) and/or more upset about losses than pleased about gains (Kahneman and Tversky 1984). Diffuse support generally refers to the Court’s ability to make displeasing decisions rather than its ability to benefit from pleasing ones. Gibson and Nelson (2015) identify the question of whether “good decisions balance out bad decisions” as an important, broad, and understudied issue. The extant literature concerning the Court broadly suggests that “bad” decisions will have larger negative effects than “good” decisions will have positive effects (Gibson and Nelson 2015; Grosskopf and Mondak 1998; Mondak and Smithey 1997). Our expectations are consistent with the broader literatures concerning the psychology of negativity biases and loss aversion. Thus, our second hypothesis (Hypothesis 2; H2) is that there exists a negativity bias in response to decisions, such that disfavored decisions will have larger negative effects on legitimacy than favored decisions will have positive effects.

Data and Methodology

During one week in June of 2013, the Court struck down the pre-clearance provisions of the Voting Rights Act (VRA) in Shelby County v. Holder and the Defense of Marriage Act (DOMA) in U.S. v. Windsor. The voting rights decision was widely condemned by liberals, and the same-sex marriage decision, released one day later, by social conservatives. This roller coaster of salient and polarizing decisions provided each ideological side with a win and a loss. It did so within the same week, and, therefore, factors such as the composition of the Court and the political/economic context were constant. These events provide a valuable opportunity to evaluate the effect of the Court’s outputs on diffuse support. Windsor was followed a year later by Obergefell v. Hodges, which struck down state-level marriage bans, while Shelby County has been untouched. Nevertheless, at the time we collected our data, both were brand-new, marked major changes, and could stand alone as treatments.

Combined with the opposing ideological decisions the Court made in the last week of the term, our dataset provides leverage to address the hypotheses above. We are able to take full advantage of these ideologically divergent decisions because we collected individual-level survey responses about the Supreme Court, political attitudes, and issue support before, throughout, and after the decisions. Moreover, we did so using a panel of respondents. Panel data are helpful in observing individual-level changes through time, and offer many advantages over other data, yet, with few exceptions (e.g., Christenson and Glick 2015a), they are “woefully scarce” in the literature (Gibson and Caldeira 2009a, 5). Specifically, the data enable us to estimate the causal effects of decisions at the micro-level by treating the decisions as exogenous events that interact with preexisting attitudes. Indeed, these decisions are similar to the experimental treatments in various studies (e.g., Bartels and Johnston 2013), but with greater external validity as these are real decisions occurring in real time. Because coverage of the Courts and the decisions ostensibly provide strong signals about the Court, we explore whether individuals in the panel change their evaluations of it as a result. Of course, all designs have limitations. In our case, we recognize the potential that repeated surveys can affect attitudes (see Bartels 1999; Warren and Halpern-Manners 2012) and that despite the advantages of having two salient cases separated by a day, it is also possible that such concentrated Supreme Court activity could also confound our results in unanticipated ways. Finally, we need a data source that provides speed and flexibility, which meant relying on an online convenience sample (see below).

Starting three weeks before the decisions, we launched an Internet panel survey. We recruited and then recontacted (retained without replacement) participants using Amazon’s Mechanical Turk (MTurk) crowdsourcing marketplace. The survey comprised a variety of questions about politics, the Supreme Court, and the major issues on its agenda including same-sex marriage and voting rights. We completed six total waves including one in each of the three weeks before the decisions. The timing of the survey waves is especially important. One advantage of MTurk is the speed with which researchers can collect responses from those in the panel (Christenson and Glick 2013). This attribute was especially useful because we were able to collect a wave of responses in the twenty-four hours between the VRA decision and the DOMA one. We also collected data immediately after the DOMA decision, and again five weeks after.¹

While the MTurk sample is not perfectly representative of the nation, it allowed us to rapidly collect panel data, despite the Court’s unpredictable schedule. The ability to time our data collection around and, importantly, between the Court’s decisions makes this study unique and required the flexibility and response rates afforded by an online convenience sample.² As with other MTurk samples, our participants are younger and more liberal than they are in the gold standard national probability samples, but our demographics skew considerably less than they do in other convenience samples.³ In all, we believe that the boost in external and internal validity offered by the panel responses gathered tightly around real decisions is a worthwhile trade-off for the slight nonrepresentativeness of the sample, especially as we control for a host of factors in the multivariate analyses that follow.

A natural concern with panel data is attrition, but we do not find evidence of systematic panel attrition or learning in our data. Not only do the demographics of the sample look stable throughout the six waves and multiple months of this study (e.g., percent white moves less than a point, while average age moves up only two years), but so do the variables more substantively related to legitimacy. The range of wave averages on partisanship and ideology, along with interest in politics and information about the Supreme Court, are all extremely stable. This stability is particularly important as it helps us discount plausible scenarios in which only people who were (un)interested in the Supreme Court or only those that received an ideologically (un)favorable decision remained in the panel, thereby inflating the potential for attitude change across the waves. In all, we did not find evidence to suggest any systematic change in the sample across the panel waves.⁴

Figure 1 depicts the research design by showing the timing of the panel waves around the two key decisions. The timing of these waves offers a number of advantages. First, having three pre-waves enables us to differentiate the effects of the decisions from random noise in legitimacy assessments. Second, because the cases were highly anticipated in the weeks before their release, our pre-waves also help distinguish salience effects from the Court’s actual outputs. Finally, having a sixth wave many weeks later enables us to check for the durability of any effects we observe immediately around the decisions.

Figure 1.

The timing of panel waves around the decisions.

Panel data around the decisions allows for externally valid causal estimates of the decisions’ effects because they capture the impact of learning about the Court’s outputs in natural ways. While, at times, researchers want to control the information on which respondents rely, in this case, the fact that people will have learned (or not) about the decisions through the sources of their choice is a feature, not a bug. Externally valid estimates of the Court’s effect on the public should include the fact that people selectively learn about the Court, and that they likely hear more than the basic content of a decision when they hear about the Court’s outputs.

Macro-Stability and Micro-Change in Legitimacy

Our surveys included a variety of typical political and demographic questions, as well as items focused on the issues and the cases. Our dependent variable is a respondent’s legitimacy score, which is based on an index of several measures of diffuse support for the Supreme Court (e.g., Gibson, Caldeira, and Spence 2003). We use the respondents’ levels of agreement with five related statements concerning whether disagreeing with Court rulings leads respondents to “favor doing away with the Court”; the Court is “too political”; the Court “favors some groups”; the Court can be “trusted to make decisions that are right for the country” and in the “best interest of the American people” (see also Bartels and Johnston 2013; Gibson, Caldeira, and Spence 2003). Each of these is measured on a 4-point ordinal scale on which respondents indicate their agreement with the statements. The five responses are summed, resulting in values that can range from 0 = minimum agreement and low legitimacy to 15.

Figure 2 plots the average legitimacy score for the sample in each wave. On average, legitimacy appears both high and stable across the waves. The score stays between 8 and 8.5 (on a 15-point scale) throughout this eventful period. If one were to draw a conclusion based on the aggregate longitudinal data, one would conclude in favor of stability and, thus, Positivity Theory. However, this aggregate stability can be misleading, as it can obscure substantial individual-level change. This limitation is especially relevant when the Court decides on issues that are overtly political or politicized. In these kinds of cases, it is plausible that those on opposite sides may respond to decisions by altering their perspective of the Court in opposite directions and in roughly equal number, producing aggregate stability despite real individual-level change (Christenson and Glick 2015a).

Figure 2.

Mean legitimacy across waves.

We focus on three sets of independent variables that capture the major theoretical sources of legitimacy. Our first set of key independent variables captures subjects’ political attitudes. The primary concern in the current state of the literature is ideology (see, for example, Bartels and Johnston 2013; Gibson and Nelson 2015). The left panel of Figure 3 again plots average legitimacy but, this time, subset by respondent ideology. Even here, at the aggregate level, there appears to be a relationship between legitimacy and ideology. Liberals’ legitimacy scores are generally higher than moderates’ or conservatives’. In addition, both liberals’ and conservatives’ legitimacy evaluations appear to have bounced around following the decisions in their expected directions.

Figure 3.

Mean legitimacy by ideology and party identification.

In the subsequent analyses, we focus on subjects’ relative ideological proximity to the Court that may change as they observe the decisions. We follow Bartels and Johnston (2013) by specifying a measure comprising one’s own ideology and one’s subjective assessment of the Court’s. We measure this ideological disagreement as the difference between a respondent’s self-assessment of her own ideology (from Wave 1) less the respondent’s perception of the Court’s ideology given its “recent decisions” on the same scale (from each wave). For example, a participant might say that she is “somewhat liberal” (2) and the Court is “on the conservative side” (5) for a 3-point gap. Our ideology measures differ in small but important ways from previous studies of the Court. The key difference is that we use branching questions (see also Christenson and Glick 2015a), which follow recommended practice in survey research (Aldrich et al. 1982) and avoid the debate over the middle response category (see Bartels and Johnston 2013; Gibson and Nelson 2015). Also different from existing studies that use snapshot measures of ideology or proxies for it, our research design allows individuals’ assessments of the Court’s ideology to change, which captures the reality that they may update their views in the periods surrounding the decisions.

Party identification is likely correlated with issue support and ideological distance to the Court. Moreover, especially in times of salient and politicized decisions, party identification has the potential to be associated with evaluations of the Court. The right panel of Figure 3 shows that party identifiers had different dynamics and responded in foreseeable ways following the decisions.

On average, and as expected, the Court lost legitimacy in the eyes of Democrats after the VRA, but gained after the DOMA decision. The dynamics for the Republicans were in the opposite direction and even more pronounced, with the largest change—that is, loss—across the waves after DOMA. The independents appear to mirror the Democrats, though less emphatically. This is true even though same-sex marriage may not always map onto conventional partisan divides, especially for younger citizens. Thus, the subsequent analyses also explore the strength of party identification (on a 7-point scale, strong democrat = 1) as a potential moderator of the decisions’ effects on legitimacy.

Our second set of key independent variables captures respondents’ support for the issues underlying the two policies about which the Court issued rulings. While most of the literature uses partisanship or a general ideology measure as a proxy for attitudes related to a Court case, we posit here that there is merit in relying specifically on policy questions that tap the attitudes about the issues on which the Court is ruling. We, therefore, test the factors of legitimacy using a more nuanced attitudinal variable. Doing so is especially important in this instance as preferences about same-sex marriage, in particular, do not always fall along partisan or ideological lines.

To measure attitudes relevant to the DOMA case, we constructed an index of support for same-sex marriage. This index comprises three items. Two tap into general attitudes about same-sex marriage by asking whether individuals favor or oppose “allowing gay and lesbian couples to marry legally” and whether “legal marriage,” “civil unions without marriage,” or “no legal recognition” comes closest to the respondent’s position. The third item concerns the more specific DOMA issue by asking whether the respondent believes that the federal government should be able to define “marriage as being only between a man and a woman.” On this index, higher scores indicate greater support for same-sex marriage and, thus, opposition to the law (DOMA) before the Court. This index intentionally taps the broader equality issues underlying the DOMA case rather than only measuring attitudes about DOMA itself.

Our measure of support for the VRA similarly taps a mix of attitudes about the narrow provision at stake in the case and broader attitudes about race and the legitimacy of action to promote racial equality. We use two Likert-type scale items about “affirmative action programs” designed to “increase the number of black and minority students’ college campuses” and “help blacks and minorities get better jobs and employment opportunities” to capture underlying attitudes about racial equality and government action. In addition, paralleling our more specific DOMA question in the marriage index, the third item in our VRA support index captures responses to the statement that “states with a history of racial discrimination at the polls should have to get approval from the federal government to change their election procedures.” Just as in the DOMA case, these three questions allow broad access to relevant attitudes by combining views about the somewhat obscure issue of pre-clearance with views about race and policy broadly. We think people were generally more likely to interpret the VRA case from the latter perspective, and unlikely to have preexisting attitudes about the former especially as most of the portrayal of the Court’s decision was often broadly framed in terms of racial equality and as a defeat for minority interests. The two issue-support indices have the virtues of capturing relevant attitudes from a couple of different perspectives and paralleling each other as nearly as possible. The key difference between them is that higher scores on the gay marriage support indicate opposition to the law before the Court while higher scores on the VRA index indicate support for it. Thus, higher scores on each are consistent with more liberal attitudes.⁵

Figure 4 hones the previous aggregate diffuse support graph by grouping respondents according to their level of support (the top and bottom thirds of the support scale) for each issue. Differences in the dynamics across the groups and waves, therefore, suggest that issue support may affect legitimacy when the Court delivers a favorable or unfavorable verdict. In the left graph, we present average legitimacy by those with different levels of support (prior to the decision) for government action to promote minority interests (VRA support). While the wave-to-wave changes are certainly not large here, the direction of change is exactly as we would expect. Overall, those who scored high on the index dropped their legitimacy ratings when the Court struck down the VRA while those who were more opposed to voting rights increased them.

Figure 4.

Mean legitimacy by prior issue support.

On the right side of Figure 4, we present a similar graph for those with different levels of support for same-sex marriage. The changes here are steeper than when grouping by the voting rights support index, particularly among those who are more opposed to gay marriage. While supporters’ average legitimacy gradually increases across the period, the major change occurs when opponents reduce their legitimacy assessments following the DOMA decision. This is an apparently long-lasting change that does not rebound weeks later.

Last, in the models below, we control for a series of potential confounders, including typical demographic variables, such as gender, race, age, income, and education. We also control for other variables that have been shown to correlate with legitimacy in static models (e.g., Bartels and Johnston 2013; Gibson and Nelson 2015), including knowledge about the Supreme Court, media exposure, a general political trust measure, and a measure of political tolerance based on a question about whether there are normally two justifiable sides to big issues.

In sum, this section’s initial graphic (Figure 2) illustrates that even across a roller-coaster week of decisions, the public, in the aggregate, consistently holds the Court in high esteem. However, the subsequent graphs of legitimacy decomposed by attitudinal groups suggest that a macro-perspective may be obscuring more micro-level changes among segments of the population. This is consistent with other recent findings (see, for example, Christenson and Glick 2015a). Of course, these graphs are merely descriptive; to test these relationships, we now proceed to more rigorous multivariate models.

How Case Decisions Affect Legitimacy

The coefficients in Table 1, with standard errors in parentheses, are from linear mixed models with random group intercepts for individuals to account for the error correlation in the repeated measurements (see Galwey 2007; Gelman and Hill 2007; Goldstein 2011; Raudenbush and Bryk 2002). The random effect model is a partial pooling model, estimating both the overall mean response as well as the deviation in each individual. That is, we assume that an individual in a wave of the panel shares a common mean effect with herself in the other waves. We allow the individual’s effect in each wave to deviate from the common effect by a random variable that follows a Gaussian distribution. The motivation here is that we avoid estimating an effect by pooling all individuals, which would mask variation in individuals’ repeated presence across the waves, and avoid estimating an effect for all individuals separately, which would give poor estimates for low-sample individuals.⁶ The p values are calculated based on normal approximations, but are virtually identical to those based on Satterthwaite’s adjustments (see Schaalje, McBride, and Fellingham 2002).

Table 1.

Linear Mixed Model Results.

	Supreme Court Legitimacy
	1	2	3	4	5	6
Age	0.013 (0.009)	−0.001 (0.009)	−0.00003 (0.009)	−0.002 (0.009)	0.009 (0.009)	0.013 (0.009)
Black	−0.100 (0.374)	−0.661* (0.375)	−0.599 (0.369)	−0.556 (0.382)	−0.270 (0.370)	−0.112 (0.374)
Education	0.059 (0.123)	0.056 (0.128)	0.091 (0.126)	0.079 (0.129)	0.058 (0.125)	0.060 (0.123)
Female	0.263 (0.185)	0.248 (0.194)	0.286 (0.191)	0.267 (0.195)	0.233 (0.189)	0.265 (0.185)
Income	0.201** (0.091)	0.272*** (0.095)	0.246*** (0.093)	0.259*** (0.095)	0.227** (0.093)	0.205** (0.091)
Supreme Court Info.	0.201** (0.088)	0.166* (0.091)	0.193** (0.090)	0.170* (0.092)	0.154* (0.089)	0.195** (0.088)
Media Exposure	0.059 (0.070)	0.045 (0.074)	0.043 (0.072)	0.042 (0.074)	0.054 (0.072)	0.058 (0.070)
Political Trust	1.565*** (0.128)	1.528*** (0.132)	1.589*** (0.126)	1.631*** (0.132)	1.538*** (0.126)	1.568*** (0.128)
Political Tolerance	0.093 (0.127)	0.070 (0.133)	−0.0001 (0.130)	0.008 (0.132)	0.111 (0.129)	0.093 (0.127)
Gay Family Member	−0.074 (0.183)	−0.045 (0.190)	0.040 (0.187)	0.019 (0.192)	−0.153 (0.187)	−0.085 (0.184)
Wave 2	−0.154** (0.070)	−0.071 (0.142)	−0.054 (0.121)	−0.052 (0.168)	−0.103 (0.183)	0.454 (0.380)
Wave 3	0.027 (0.073)	0.131 (0.147)	0.084 (0.124)	0.112 (0.175)	−0.199 (0.188)	−0.024 (0.398)
Wave 4	−0.137* (0.082)	−0.301* (0.164)	0.168 (0.144)	0.306 (0.200)	0.074 (0.209)	0.591 (0.440)
Wave 5	−0.103 (0.075)	0.083 (0.153)	0.392*** (0.131)	0.040 (0.183)	−0.843*** (0.195)	−0.279 (0.412)
Wave 6	−0.185** (0.081)	−0.003 (0.164)	0.079 (0.142)	0.142 (0.196)	−0.879*** (0.203)	−0.249 (0.427)
Party ID Strength	−0.029 (0.064)	−0.135** (0.060)				−0.013 (0.070)
Wave 2 × Party Strength		−0.026 (0.040)				−0.067 (0.050)
Wave 3 × Party Strength		−0.027 (0.041)				−0.010 (0.052)
Wave 4 × Party Strength		0.052 (0.046)				−0.007 (0.058)
Wave 5 × Party Strength		−0.059 (0.043)				0.021 (0.054)
Wave 6 × Party Strength		−0.061 (0.046)				−0.008 (0.057)
Ideological Distance	−0.126*** (0.025)		−0.058 (0.036)			−0.053 (0.036)
Wave 2 × Ideol. Distance			−0.042 (0.043)			−0.045 (0.043)
Wave 3 × Ideol. Distance			−0.022 (0.044)			−0.014 (0.045)
Wave 4 × Ideol. Distance			−0.129*** (0.049)			−0.117** (0.050)
Wave 5 × Ideol. Distance			−0.213*** (0.046)			−0.184*** (0.047)
Wave 6 × Ideol. Distance			−0.112** (0.049)			−0.072 (0.049)
Voting Rights Support	−1.056** (0.427)			−0.001 (0.441)		−0.726 (0.465)
Wave 2 × VRA Support				−0.185 (0.288)		−0.259 (0.319)
Wave 3 × VRA Support				−0.123 (0.299)		−0.306 (0.334)
Wave 4 × VRA Support				−0.832** (0.342)		−0.583 (0.387)
Wave 5 × VRA Support				−0.263 (0.310)		−0.365 (0.350)
Wave 6 × VRA Support				−0.622* (0.331)		−1.032*** (0.369)
Gay Marriage Support	1.931*** (0.345)				1.553*** (0.321)	1.690*** (0.372)
Wave 2 × GM Support					−0.062 (0.216)	−0.203 (0.257)
Wave 3 × GM Support					0.314 (0.223)	0.366 (0.269)
Wave 4 × GM Support					−0.277 (0.247)	−0.152 (0.301)
Wave 5 × GM Support					0.947*** (0.230)	0.937*** (0.280)
Wave 6 × GM Support					0.886*** (0.241)	1.055*** (0.288)
Constant	1.960** (0.891)	3.614*** (0.807)	3.030*** (0.763)	3.011*** (0.790)	1.698** (0.802)	1.784* (0.915)
Random Effect
# of Groups	650	650	650	650	650	650
Group Standard Deviation	2.145	2.25	2.215	2.266	2.197	2.15
Observations	2,936	2,936	2,936	2,936	2,936	2,936
Log Likelihood	−5,552.55	−5,587.57	−5,568.74	5,578.88	−5,546.45	−5,542.71
Akaike Inf. Crit.	11,149.11	11,223.14	11,185.48	11,205.77	11,140.90	11,169.42
Bayesian Inf. Crit.	11,280.77	11,366.77	11,329.11	11,349.40	11,284.53	11,420.79

VRA = Voting Rights Act; GM = Gay Marriage.

p < .1. **p < .05. ***p < .01.

These models also contain fixed effects for the waves, which convey any overall change in a period relative to the first wave (i.e., pre-decision) baseline. The goal here is to capture the precise timing of changes in legitimacy, or a lack thereof, brought about by the exogenous case decisions occurring within the panel windows. Recall that the VRA decision occurred between the third and fourth waves while the DOMA decision occurred between the fourth and fifth waves. Thus, should a case decision affect legitimacy, the effect should show up beginning in the fourth and/or fifth wave dummies.⁷

Table 1 contains the results from a series of linear mixed models introducing additional terms in a stepwise fashion. The first model contains all of the lower order terms in the models, while the second through fifth introduce the interaction terms necessary to test our hypotheses. The sixth looks at the entire set of terms together. Before we turn to our variables of primary concern, we consider the direct effects of some of the controls and lower order terms presented in model 1. Turning first to the wave dummies, we find the same pattern as in Figure 2. There are significant drops in legitimacy from the Wave 1 baseline unpredicted by the other variables in the model at Waves 2, 4, and 6. However—again, as the figure suggested—the change is substantively quite small. Thus, a first look at the data suggests support for Positivity Theory. While we see changes in legitimacy across the panel, it remains high and fairly stable, at least in the aggregate.

In line with previous studies (Bartels and Johnston 2013; Christenson and Glick 2015a), we find that ideological distance (a dynamic variable that allows the subjective perception of the Court to vary at each wave) from the Court is negatively related to legitimacy. The further one perceives the Court from oneself, the lower one’s legitimacy score. Even with a rather comprehensive set of explanatory variables, many of which are moderately correlated with ideology, one’s proximity to the Court remains a significant predictor of legitimacy. While similarly signed, party identification strength (a static variable measured in Wave 1), however, is not significantly related to legitimacy.

The model also includes the two measures of issue support (static variables measured in Wave 1), the estimates of which are signed in opposite directions. We find that voting rights support is negatively related to legitimacy across the panel. That is, support for voting rights is associated with lower evaluations of legitimacy in this period. In contrast, support for gay marriage is positively related to legitimacy. That is, greater support for gay marriage is associated with higher evaluations of legitimacy. That policy support is associated with diffuse support for the Court, even when controlling for ideological distance, is somewhat surprising given the generally close relationship between policy views and ideology. Moreover, given the directions of the decisions, the relationships with legitimacy are in line with our expectations from the sensitivity to outputs hypothesis, which we more directly address below.

Among the larger effects, and as previous literature has also suggested (Gibson and Nelson 2015), political trust is highly significant and positive throughout the specifications. The other democratic value, political tolerance, is similarly signed, though small, and does not reach conventional levels of significance. Having greater information about the Supreme Court correlates with higher institutional support for it (Gibson and Caldeira 2011). Likewise, income is shown to have a consistent association with legitimacy. Those with greater annual family incomes perceive the Court as more legitimate than their less well-off counterparts.

The first model largely confirms a host of aforementioned findings in the literature on the correlates of legitimacy. Our primary interest, however, has to do with the ability of case decisions to affect legitimacy. More precisely, our hypotheses expect a number of factors—party identification, ideology, and policy support—to condition the effect of decisions on legitimacy. That is, despite stability in legitimacy overall, significant change may be occurring for particular portions of the public (Christenson and Glick 2015a). To test these hypotheses, beginning with model 2, we introduce interactions of these factors with the wave dummies.

In this panel framework, the pertinent wave variables act as indications of when respondents (could have) received the “treatment” (i.e., news of the decision) via the various channels through which they learn about the news. Because the wave dummies capture unobserved departures from the pre-decision (Wave 1) baseline, the interaction variables (e.g., wave × factor) provide the primary test of the hypotheses. That is, the model specifications position case decisions as exogenous treatments that interact with various key factors of interest to affect legitimacy evaluations. As the VRA decision was handed down on the day before Wave 4, and the DOMA decision on the day before Wave 5, should a factor condition the relationship of the VRA decision on legitimacy, we should see a significant effect in the Wave 4 interaction. Should a factor condition the effect of the DOMA decision on legitimacy, we should see a significant effect in the Wave 5 interaction. Significant interaction effects in the waves following decisions would indicate support for the sensitivity to outputs hypothesis (H1) generally, while null findings would suggest support for the stability hypothesis. Differences in the interaction effects among supporters and opponents of the policies—such that those who are unhappy with decisions drop legitimacy more than those who are happy with them raise legitimacy—would provide support for the negativity bias hypothesis (H2).⁸

We begin with an exploration of party identification (H1A) strength as a moderator of the effect of decisions on legitimacy. If party ID moderates the impact of these decisions on legitimacy, we should see a positive interaction between the 7-point party ID variable and Wave 4 (the VRA decision) because republicans would be higher on both scales. Conversely, we should see a negative interaction between the party ID variable and the Wave 5 indicator. Looking at model 2, we find no evidence that the effect of these decisions is conditioned by party. At Wave 1, we find a significant negative relationship with legitimacy, suggesting that the more one identifies with the Republican Party, the less diffuse support she has for the Court at the start of the panel. That goes largely unchanged throughout the panel. Subsequent wave interactions are insignificant, suggesting little difference from Wave 1. As such, we do not find evidence that the effect of the decisions on legitimacy were conditioned by party. Likewise, in the fully specified model, model 6, party identification strength is entirely ineffectual.

The data provide evidence of sensitivity to outputs when focusing on the ideological distance moderator (H1B). Because this variable is nondirectional, we have the same expectation for each case. More distance should be associated with less legitimacy. Focusing on ideological distance, we find evidence directly in line with our hypotheses. Beginning in Wave 4, immediately following the VRA decision, and continuing through the next two waves of the panel, the ideological distance interaction is significant and negative. That is, relative to the Wave 1 baseline, we find that ideological distance conditions the unobserved exogenous effects in Waves 4–6 in the expected direction. Even in the fully specified model, model 6, the results are nearly unchanged in the two waves immediately following the decisions, though the conditioning effect does not persist to the sixth wave when accounting for policy support.

We unpack the findings and illustrate the magnitude of effects in Figure 5. The figure is based on the results from the fully specified model 6 in Table 1. At each wave, we plot and connect the model predictions of the continuous ideological distance variable at five evenly spaced levels from minimum (same ideology as the Court) to maximum distance. Looking at legitimacy across the first three waves, we see that prior to the two decisions, legitimacy was fairly high and stable regardless of ideological distance to the Court. However, after the first decision on the VRA case, we begin to see substantial movement conditional on ideological proximity. Legitimacy is steady and high through the fifth wave for those who see the Court as ideologically close to them at each wave. Those who perceive the Court as more ideologically distant have lower support for the Court in the same period, and increasingly so as distance is extended. That is, those who see the Court as very far from them after a decision drop their legitimacy evaluations considerably; the effect of the decision being stronger for the ideologically distant than proximate. Specifically, the most distant group decreased legitimacy by nearly a full point through the panel, while those in the closest group increased it by about one-third of a point. Interestingly, the convergence of lines in the final wave shows that the conditioning effect of ideological distance diminishes several weeks after the last decision, though not quite to pre-decision levels.

Figure 5.

Effects of ideological distance by wave.

Finally, we turn to the question of whether prior support for the issues at stake in the cases conditions the effect of receiving information about the Court’s decisions (H1C). Here, the directional expectation is different for each case. Because of the way the cases came down, we expect supporters of voting rights and opponents of marriage equality to reduce legitimacy.

Looking first at the sets of interaction terms in separate models, models 4 and 5, we see that the interaction effects of the VRA decision (Wave 4) with voting rights support, as well as the DOMA decision (Wave 5) with support for gay marriage, are in the posited directions—negative and positive, respectively—and statistically significant. The timing is exactly as expected. At Wave 4, immediately after the VRA decision, the interaction with VRA support becomes statistically significant. Likewise, at Wave 5, immediately after the DOMA decision, we see a statistically significant interaction effect for gay marriage support. These findings provide support for the sensitivity to outputs hypothesis (H1). Upon receiving information about the Court’s position on these issues, people updated their legitimacy assessments in ways consistent with their policy attitudes. In other words, the effects of the Court’s outputs on legitimacy were magnified by issue support.

In the fully specified model 6, we find similar patterns, but one clear difference: when taking into account ideological distance and marriage equality support, the moderating effect of VRA support plays less of a role. While the direction of the effect is stable, the standard errors are larger, and the effect is only significant at the end of the panel. Notably, while both issue effects appear to be long-lasting—evident more than a month later in Wave 6—the effect is much stronger for the DOMA decision. The results suggest heterogeneous treatment effects; all cases are not equal in their ability to affect the public’s legitimacy evaluations.

To further explore the direction and magnitude of these relationships, we plot the interaction effects in Figure 6. The plots illustrate the interactions for both the voting rights and gay marriage support measures combined with the respective decisions, as captured by the appropriate wave dummies. We see most clearly from the figure that negativity bias (H2) is manifest in both interactions. Consider first the VRA case that was decided in a way contrary to preferences of those who indicated high support on our VRA index. Looking at the left graph, we see the greatest change in legitimacy following the VRA decision by those most supportive of voting rights. While these VRA supporters generally had lower legitimacy evaluations of the Court, they dropped their evaluations to a much greater extent (about half a point for the strongest supporters) than those who opposed the VRA raised their evaluations (none). Notably, the VRA issue support appears to have a cumulative effect on change in legitimacy evaluations, which includes a slight uptick in support among those seeing the Court move toward them at the end of the term. Despite small effects immediately after the decision, by the end of the panel, there is substantial divergence conditional on issue support, suggesting relatively long-lasting effects from the decision.

Figure 6.

Effects of issue support by wave.

Turning to the right panel in Figure 6, we find similar evidence of both sensitivity to outputs and negativity bias for the gay marriage issue. Greater support for gay marriage is associated with higher legitimacy rankings to begin with. However, those most supportive of gay marriage did not massively reward the Court (roughly a quarter point increase). Rather, the “losers,” or those that were most opposed to gay marriage, were more likely to punish the Court with lower legitimacy evaluations. That is, we see the greatest change in legitimacy, nearly a full point, immediately following the DOMA decision by those most opposed to gay marriage. The effect diminishes, and eventually flattens, as we move up in levels of gay marriage support. In both cases, exposure to the decisions resulted in a notably bigger drop among the “losers” and slight to no change among the “winners.” The fact that these effects manifest on the figure for voting rights supporters and same sex marriage opponents provides strong and consistent support for the sensitivity to outputs and negativity bias hypotheses.

Discussion

The Court’s two ideologically divergent decisions in landmark cases in a single week provide unprecedented leverage in the lively debate about the micro-foundations of legitimacy and its stability. We find that people updated their legitimacy assessments in response to the decisions’ alignments with their issue preferences and with perceived ideological distance. To be sure, they updated them to different degrees depending on the salience of the case, but our results clearly support the proposition that the Court’s legitimacy varies systematically with people’s feelings about its outputs. The evidence here provides a strong and unique challenge to the prevailing view that respondent-level legitimacy is stable and independent of individual case decisions.

More specifically, we have shown that underlying political attitudes moderate the effect of the Court’s interjection. Both policy preferences specific to the case and ideological distance, which is more generally applicable and forward looking, interacted with the actual decisions to affect legitimacy assessments. While we found evidence of these effects in both cases, and more generally found evidence of legitimacy vacillating as the Court’s decisions in major cases were announced, the response was stronger and more consistent in the marriage equality case. The most likely explanation is a mix of media salience and better formulated underlying attitudes. While the VRA case has large consequences and was salient among elites, it did not receive the same level of popular attention as the DOMA decision before or after it was released. This reality speaks to, among other things, the importance of further understanding the conditions under which Court decisions affect its public standing.

The second major source of variation we provide evidence for concerns negativity bias. We find that disappointment is generally more powerful than delight. The dramatic drops among those with positions contra the Court’s decisions point to asymmetrical effects. The pronounced evidence of negativity bias for cases decided in opposite directions prompts a natural follow-up question: why does legitimacy not tend toward zero? We speculate that the mechanisms we provide evidence for likely only apply in salient and polarizing cases. These cases are critical in that they are the ones people are paying attention to, but they are also rare. It is likely that ideological mechanisms and those from Positivity Theory are both in play but at different times. In between salient decisions, subtle mechanisms central to Positivity Theory may gradually help the Court recoup legitimacy. While our panel data show that ideological and, in particular, issue-support effects persist, it is possible that a long sequence of routine days and/or decisions helps reinforce the Court’s legitimacy until the next big case in which the Court provides fodder for ideological legitimacy updating. These findings and mechanisms are consistent with prior work that similarly allows for support for the Court to vary with outputs while reverting toward a baseline in the long term (Durr, Martin, and Wolbrecht 2000; Mondak and Smithey 1997).

Finally, and in the spirit of reconciling competing streams of research, we return to the very general questions about the sources of Court legitimacy and whether diffuse support is stable or sensitive to outputs and attitudes. The fact that individuals’ assessments of the Court’s legitimacy waxed and waned as the Court made a pair of decisions one day apart from each other challenges the widely touted claim that diffuse support is stable and grounded in factors other than satisfaction with particular decisions. In contrast, the work, in part, buttresses a version of the stability argument. We have shown how offsetting individual changes can add up to overall stability. Aggregate diffuse support was largely unaffected by a week in which the Court was constantly in the news for making controversial decisions in landmark cases. Moreover, while we find updating at the individual level in response to the latest case, this moderated updating is different than finding that the average person fundamentally reassesses the Court after every decision. People update substantially and significantly, but they do so around a relatively stable prior. The significant effects we find are not trivial, but they are moderate. Thus, while the Court affects its own legitimacy with its salient decisions, it is also true that some of its legitimacy is rooted in something else. People, collectively and individually, are sensitive to individual case decisions, but they do not deviate from seeing the Court as fundamentally legitimate to fundamentally illegitimate on the basis of a single ruling. Change is real and long-lasting after particular decisions, but it is also relative to different priors.

Supplemental Material

Online_Appendix – Supplemental material for Reassessing the Supreme Court: How Decisions and Negativity Bias Affect Legitimacy

Supplemental material, Online_Appendix for Reassessing the Supreme Court: How Decisions and Negativity Bias Affect Legitimacy by Dino P. Christenson and David M. Glick in Political Research Quarterly

Footnotes

Acknowledgements

We would like to thank John Carey, Mark Miller, Doug Kriner, Maya Sen, Sara Benesh, Jim Gibson, Adam Berinsky, Brendan Nyhan, and Graham Wilson for helpful comments. Earlier versions of this paper were presented at the 2015 Annual Meetings of the Western Political Science Association and the Midwest Political Science Association, as well as the 2015 American Politics Conference at Massachusetts Institute of Technology (MIT). It was the winner of the 2015 Pi Sigma Alpha Award for the best paper presented at the Annual Meeting of the Western Political Science Association.

Authors’ Note

Authors’ names are listed alphabetically.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Supplemental Material

The data and code required to replicate the analyses in this article are available on the Harvard Dataverse Network at . Supplemental materials for this article are available with the manuscript on the Political Research Quarterly (PRQ) website.

References

Aldrich

John H.

Niemi

Richard G.

Rabinowitz

George

Rohde

David W.

1982. “The Measurement of Public Opinion about Public Policy: A Report on Some New Issue Question Formats.” American Journal of Political Science 26:391–414.

Bartels

Brandon L.

Johnston

Christopher D.

2013. “On the Ideological Foundations of Supreme Court Legitimacy in the American Public.” American Journal of Political Science 57 (1): 184–99.

Bartels

Brandon L.

Mutz

Diana C.

2009. “Explaining Processes of Institutional Opinion Leadership.” The Journal of Politics 71 (1): 249–61.

Bartels

Larry M.

1999. “Panel Effects in the American National Election Studies.” Political Analysis 8 (1): 1–20.

Bartels

Larry M.

2002. “Beyond the Running Tally: Partisan Bias in Political Perceptions.” Political Behavior 24 (2): 117–50.

Berinsky

Adam J.

Huber

Gregory A.

Lenz

Gabriel S.

2012. “Evaluating Online Labor Markets for Experimental Research: Amazon.com’s Mechanical Turk.” Political Analysis 20 (3): 351–68.

Brickman

Danette

Peterson

David A. M.

2006. “Public Opinion Reaction to Repeated Events: Citizen Response to Multiple Supreme Court Abortion Decisions.” Political Behavior 28 (1): 87–112.

Caldeira

Gregory A.

Gibson

James L.

1992. “The Etiology of Public Support for the Supreme Court.” American Journal of Political Science 36:635–64.

Campbell

Angus

Converse

Philip E.

Miller

Warren E.

Stokes

Donald E.

1960. The American Voter. New York: John Wiley.

10.

Christenson

Dino P.

Glick

David M.

2013. “Crowdsourcing Panel Studies and Real-Time Experiments in MTurk.” The Political Methodologist 20:27–32.

11.

Christenson

Dino P.

Glick

David M.

2015a. “Chief Justice Roberts’s Health Care Decision Disrobed: The Microfoundations of the Supreme Court’s Legitimacy.” American Journal of Political Science 59 (2): 403–18.

12.

Christenson

Dino P.

Glick

David M.

2015b. “Issue-Specific Opinion Change: The Supreme Court and Health Care Reform.” Public Opinion Quarterly 79 (4): 881–905.

13.

Christenson

Dino P.

Kriner

Douglas L.

2017. “Constitutional Qualms or Politics as Usual? The Factors Shaping Public Support for Unilateral Action.” American Journal of Political Science 61 (2): 335–49.

14.

Druckman

James N.

Fein

Jordan

Leeper

Thomas J.

2012. “A Source of Bias in Public Opinion Stability.” American Political Science Review 106 (2): 430–54.

15.

Durr

Robert H.

Martin

Andrew D.

Wolbrecht

Christina

. 2000. “Ideological Divergence and Public Support for the Supreme Court.” American Journal of Political Science 44:768–76.

16.

Egan

Patrick J.

Citrin

Jack

. 2011. The Limits of Judicial Persuasion and the Fragility of Judicial Legitimacy. Working paper. https://escholarship.org/uc/item/2gh262w3.

17.

Fiske

Susan T.

1980. “Attention and Weight in Person Perception: The Impact of Negative and Extreme Behavior.” Journal of Personality and Social Psychology 38 (6): 889–906.

18.

Franklin

Charles H.

Kosaki

Liane C.

1989. “Republican Schoolmaster: The U.S. Supreme Court, Public Opinion, and Abortion.” American Political Science Review 83:751–71.

19.

Gaines

Brian J.

Kuklinski

James H.

Quirk

Paul J.

Peyton

Buddy

Verkuilen

Jay

. 2007. “Same Facts, Different Interpretations: Partisan Motivation and Opinion on Iraq.” The Journal of Politics 69 (4): 957–74.

20.

Galwey

Nicholas W.

2007. Introduction to Mixed Modelling: Beyond Regression and Analysis of Variance. Chichester, West Sussex, PO19 8SQ, UK: John Wiley.

21.

Gelman

Andrew

Hill

Jennifer

. 2007. Data Analysis Using Regression and Multilevel/Hierarchical Models. Vol. 1. New York: Cambridge University Press.

22.

Gibson

James L.

2007. “The Legitimacy of the U.S. Supreme Court in a Polarized Policy.” Journal of Empirical Legal Studies 4:507–38.

23.

Gibson

James L.

2015. “Legitimacy Is for Losers: The Interconnections of Institutional Legitimacy, Performance Evaluations, and the Symbols of Judicial Authority.” In Motivating Cooperation and Compliance with Authority, edited by Bornstein

Brian H.

Tomkins

Alan J.

, 81–116. New York, NY: Springer.

24.

Gibson

James L.

Caldeira

Gregory A.

2009a. Citizens, Courts, and Confirmations: Positivity Theory and the Judgments of the American People. Princeton: Princeton University Press.

25.

Gibson

James L.

Caldeira

Gregory A.

2009b. “Knowing the Supreme Court? A Reconsideration of Public Ignorance of the High Court.” The Journal of Politics 71:421–41.

26.

Gibson

James L.

Caldeira

Gregory A.

2011. “Has Legal Realism Damaged the Legitimacy of the U.S. Supreme Court.” Law & Society Review 45 (1): 105–219.

27.

Gibson

James L.

Caldeira

Gregory A.

Baird

Vanessa A.

1998. “On the Legitimacy of National High Courts.” American Political Science Review 92 (2): 343–58.

28.

Gibson

James L.

Caldeira

Gregory A.

Spence.

Lester Kenyatta

2003. “The Supreme Court and the U.S. Presidential Election of 2000: Wounds, Self-Inflicted or Otherwise?” British Journal of Political Science 33:535–56.

29.

Gibson

James L.

Nelson

Michael J.

2015. “Is the U.S. Supreme Court’s Legitimacy Grounded in Performance Satisfaction and Ideology?” American Journal of Political Science 59 (1): 162–74.

30.

Goldstein

Harvey

. 2011. Multilevel Statistical Models. Vol. 922. Chichester, West Sussex, PO19 8SQ, UK: John Wiley.

31.

Grosskopf

Anke

Mondak

Jeffrey J.

1998. “Do Attitudes toward Specific Supreme Court Decisions Matter? The Impact of Webster and Texas v. Johnson on Public Confidence in the Supreme Court.” Political Research Quarterly 51 (3): 633–54.

32.

Hetherington

Mark J.

Smith

Joseph L.

2007. “Issue Preferences and Evaluations of the U.S. Supreme Court.” Public Opinion Quarterly 71 (1): 40–66.

33.

Hoekstra

Valerie J.

2000. “The Supreme Court and Local Public Opinion.” American Political Science Review 94:89–100.

34.

Jerit

Jennifer

Barabas

Jason

. 2012. “Partisan Perceptual Bias and the Information Environment.” The Journal of Politics 74 (3): 672–84.

35.

Kahneman

Daniel

Tversky

Amos

. 1984. “Choices, Values, and Frames.” American Psychologist 39 (4): 341–50.

36.

Kelly

Jocelyn

. 2014. 61% of Young Republicans Favor Same-Sex Marriage. Pew Research Center. Online report, March 10, 2014. http://www.pewresearch.org/fact-tank/2014/03/10/61-of-young-republicans-favor-same-sex-marriage/

37.

Lau

Richard R.

1985. “Two Explanations for Negativity Effects in Political Behavior.” American Journal of Political Science 29:119–38.

38.

Mondak

Jeffery J.

1994. “Policy Legitimacy and the Supreme Court: The Sources and Contexts of Legitimation.” Political Research Quarterly 47 (3): 675–92.

39.

Mondak

Jeffery J.

Smithey

Shannon Ishiyama

. 1997. “The Dynamics of Public Support for the Supreme Court.” The Journal of Politics 59 (4): 1114–42.

40.

Pratto

Felicia

John

Oliver P.

1991. “Automatic Vigilance: The Attention-Grabbing Power of Negative Social Information.” Journal of Personality and Social Psychology 61:380–91.

41.

Raudenbush

Stephen W.

Bryk

Anthony S.

2002. Hierarchical Linear Models: Applications and Data Analysis Methods. Vol. 1. Thousand Oaks: SAGE.

42.

Schaalje

G. Bruce

McBride

Justin B.

Fellingham

Gilbert W.

2002. “Adequacy of Approximations to Distributions of Test Statistics in Complex Mixed Linear Models.” Journal of Agricultural, Biological, and Environmental Statistics 7 (4): 512–24. https://www-jstor-org.web.bisu.edu.cn/stable/1400374.

43.

Schaffner

Brian F.

Roche

Cameron

. 2016. “Misinformation and Motivated Reasoning: Responses to Economic News in a Politicized Environment.” Public Opinion Quarterly 81:86–110.

44.

Sen

Maya

. 2015. How Political Signals Affect Public Support for Judicial Nominations: Evidence from a Conjoint Experiment. Working paper. http://scholar.harvard.edu/files/msen/files/conjoint-judicial-nominations.pdf.

45.

Stoutenborough

James W.

Haider-Markel

Donald P.

Allen

Mahalley D.

2006. “Reassessing the Impact of Supreme Court Decisions on Public Opinion: Gay Civil Rights Cases.” Political Research Quarterly 59 (3): 419–33.

46.

Taber

Charles S.

Lodge

Milton

. 2006. “Motivated Skepticism in the Evaluation of Political Beliefs.” American Journal of Political Science 50 (3): 755–69.

47.

Warren

John Robert

Halpern-Manners

Andrew

. 2012. “Panel Conditioning in Longitudinal Social Science Surveys.” Sociological Methods & Research 41 (4): 491–534. doi:10.1177/0049124112460374.

48.

Zaller

John R.

1992. The Nature and Origins of Mass Opinion. Cambridge, UK: Cambridge University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.32 MB