Naming Names: The Impact of Supreme Court Opinion Attribution on Citizen Assessment of Policy Outcomes

Abstract

The manner in which political institutions convey their policy outcomes can have important implications for how the public views institutions' policy decisions. This paper explores whether the way in which the U.S. Supreme Court communicates its policy decrees affects how favorably members of the public assess its decisions. Specifically, we investigate whether attributing a decision to the nation's High Court or to an individual justice influences the public's agreement with the Court's rulings. Using an experimental design, we find that when a Supreme Court outcome is ascribed to the institution as a whole, rather than to a particular justice, people are more apt to agree with the policy decision. We also find that identifying the gender of the opinion author affects public agreement under certain conditions. Our findings have important implications for how public support for institutional policymaking operates, as well as the dynamics of how the Supreme Court manages to accumulate and maintain public goodwill.

The manner in which institutions convey policy choices to the rest of the world has long interested students of politics and public policy and represents a core concern of governance dynamics (e.g., Druckman 2001; Easton 1965; Estlund 2007; Mondak 1992). In the case of the U.S. Supreme Court, the American convention of attributing court decisions to justices in signed opinions involves a careful balancing of important and sometimes competing concerns, including judicial independence, accountability, and how the opinions will be perceived by external political actors and the public. Legal scholars have advanced strong opinions on whether the U.S. Supreme Court's approach to transmitting its case decisions is optimal or might be better handled in another manner. Fiss (1983) argues that the current method of individually signed majority opinions best serves the Court's institutional legitimacy while also promoting justices' accountability for their decisions. On the other hand, some suggest that the Court's institutional credibility and robustness would be better served through the use of anonymous opinions—similar to the practice employed in civil law countries (Bozzo 2015; Markham 2006).

In this paper, we address a puzzle that persists regarding how the U.S. High Court fosters and preserves legitimacy and support among the public: can the manner in which the Court communicates its policy decrees affect whether members of the public react favorably to its legal decisions? Specifically, we are interested in whether citizens' agreement with a decision by the Court is affected by source cues tied to the identity of the majority opinion author. We argue that majority opinions attributed to the U.S. Supreme Court as a whole should enjoy higher levels of agreement than those attributed to particular justices because the Court-attributed opinions will connote neutrality, credibility, and institutional legitimacy (Bartels and Mutz 2009; Gibson et al. 2014; Hoekstra 1995).

Using an experimental design with approximately 1200 respondents, we find support for our primary thesis that legal decisions attributed to the U.S. Supreme Court—as opposed to a specific justice—enjoy higher levels of agreement. Our results also confirm that this relationship is conditioned on citizens' ideological identity. In extended analyses, we find that environmental cultural influences help determine whether and how the gender of the attributed justice has an impact on citizen agreement with U.S. Supreme Court decisions.

For the U.S. Supreme Court, these considerations are especially significant given that the Court lacks the means possessed by Congress or the executive to enforce and implement its judgments. The Court is an unusually vulnerable policy-making institution that relies on public support to maintain its policy viability and protect it from institutional encroachment (e.g., Caldeira 1986). As Gibson (2012, 2015) and others have noted, “legitimacy is for losers.” In other words, a person's assessment of the legitimacy of an institution and the legitimacy of its policy decisions quite often turns on whether they agree with them. The U.S. Supreme Court typically only has to draw from its “reservoir of good will” when citizens disagree with its verdicts. Hence, making policy decisions in a manner that is more palatable to a wider portion of the public, all else being equal, goes a long way toward an institution being able to maintain long-term diffuse support from the public (i.e., legitimacy) and enjoy its governing benefits (2015: 82–84).

The paper proceeds as follows. Section 1 lays out the theoretical framework for understanding agreement with U.S. Supreme Court decisions, and Section 2 describes our experimental design. In Section 3, we discuss the results from our main models and then in Section 5 test whether our findings on justice attribution are affected by the gender of the opinion author. In the final section of our paper, we discuss the implications of our findings for American politics and legal policymaking and suggest potential paths for future research on this subject.

Theoretical Framework

Agreement and Supreme Court Decisions

Because the U.S. Supreme Court is dependent on public support to maintain its legitimacy and to ensure the implementation of its rulings, scholars have focused a great deal of their attention on its public perception (Gibson et al. 2003; Gibson and Caldeira 1992; Scheb and Lyons 2000). Notable in this literature is the conclusion that the American High Court benefits from a robust sense of legitimacy not enjoyed by other high courts (Gibson et al. 1998) and that this diffuse support does not turn on affection for its decisions (Gibson and Caldeira 2009). In recent years, others have argued that assessments of legitimacy are at least in part a function of ideological agreement with the Court, with the implication being that the U.S. Supreme Court's legitimacy may not be so stable after all (Bartels and Johnston 2013; Christenson and Glick 2015; but see Gibson and Nelson 2015). Central to this debate is the question of how the public actually processes output from the Court (Johnston et al. 2014). Given that legitimacy is most important for those who do not agree with the Court's decisions (Gibson 2012; Gibson et al. 2014) and can impact implementation (Canon and Johnson 1984), it becomes important to understand what factors influence public agreement with the specific decisions of the Court.

While considerably less attention has been paid to studying citizens' agreement with the U.S. Supreme Court's decisions than citizens' perceptions of the legitimacy of such decisions, the existing research points to the importance of the content of Court opinions (Baas and Thomas 1984; Mondak 1994; Zink et al. 2009), as well as attributes of the Court (Boddery and Yates 2014; Zink et al. 2009) and the framing of the decision by the media (Mondak 1994; Zilis 2015). One especially promising avenue for unpacking agreement dynamics focuses on the use of source cues as heuristics in public opinion about the U.S. Supreme Court. Because U.S. Supreme Court rulings are often complicated and can be difficult for a lay person or even journalists (Slotnick and Segal 1994) to understand easily, relying on heuristics can ease the cognitive burden for those trying to understand and evaluate the Court's decision in a case. Such heuristic cue-following allows people to make quick assessments on complex matters in order to help make sense of the world around them in an efficient and largely effective manner (e.g., Kahneman 2011; Nicholson and Hansford 2014; Salamone 2014). In experimental settings, researchers have found that the public is responsive to partisan source cues about the Court (Clark and Kastellec 2015; Nicholson and Hansford 2014) and that the ideology of the opinion author in particular serves as a source cue that conditions individual agreement with court decisions—even those that run counter to an individual's expressed policy preferences (Boddery and Yates 2014). In the section that follows, we build on these insights to lay out an account that describes how the attribution of a U.S. Supreme Court decision on a case affects the public's feelings about that decision.

The Power of Attribution

While opinion writing on the U.S. Supreme Court typically involves contributions and negotiations by other members of the Court, the majority opinion writer represents the agreed upon edict and rational of the collective. Thus, as one scholar observed, “even though a single [j]ustice signs the opinion by name, the text insists throughout on its shared provenance as the voice not just of its author but of all those who have voted to join it” (Ray 2000: 518). Indeed, Justice Breyer faced a firestorm when he inadvertently used the pronoun “I” in a 1999 majority opinion (Bozzo 2015; Markham 2006; Mauro 1999; Ray 2000).

The tradition of the U.S. Supreme Court issuing a single signed opinion has been credited to Chief Justice John Marshall, who prevailed on the matter in a less famous disagreement with Thomas Jefferson, who favored the use of seriatim opinions (Ginsburg 1990: 138). But even the Court itself does not always follow this American convention. The U.S. High Court regularly issues unsigned “per curiam” opinions to convey its case decisions, and state high courts and the U.S. Courts of Appeals also follow this practice. While per curiam decrees are traditionally associated with routine low-salience cases, critics of the practice point to numerous U.S. Supreme Court case opinions that have been delivered in this manner that addressed important policy and political issues (e.g., Bush v. Gore 2000). Most recently, in the wake of Justice Scalia's death, the eight-member Roberts Court handed down a per curiam unanimous decision in a controversial case challenging the contraception mandate in President Obama's Affordable Care Act (Zubik v. Burwell 2016).

As U.S. Supreme Court scholars have documented, the choice of which justice will be assigned the majority opinion often reflects considerations about the symbolic value of particular justices and their ability to “help make policy more palatable to external actors, including the other political institutions and the public” (Epstein and Knight 1998: 127). For instance, Chief Justice Harlan Stone assigned the majority opinion in Korematsu v. United States (1944) to Hugo Black, precisely because of his reputation as a civil libertarian (Epstein and Knight 1998: 127). Similarly, in United States v. Virginia (1996), Chief Justice Rehnquist assigned the majority opinion striking down Virginia Military Institute's male-only admissions policy to Justice Ginsburg, as a nod to her expertise (Maltzman and Wahlbeck 1996) and lengthy experience litigating sex discrimination cases prior to becoming a federal judge. These examples suggest that members of the U.S. Supreme Court recognize the importance of opinion assignment and act strategically when selecting opinion authors because they believe it will have an impact on how the policy edict of the Court will be perceived.¹

Norms about how to attribute judicial decisions can vary a great deal across different court settings, but on the U.S. Supreme Court, the majority opinion author is identified in the case as delivering the “opinion of the Court.” This practice melds together both individual and institutional attributions to a decision, though of course, there are a number of exceptions to this practice throughout the Court's history (see Markham 2006). However, the public generally learns of U.S. Supreme Court decisions not directly from Court opinions, but from media outlets (Davis 2014; Johnston and Bartels 2010), and the way that a ruling is framed has been shown to have an impact on citizen assessments of the Court and its decisions (Baird and Gangl 2006; Clawson and Waltenburg 2003; Zilis 2015).² Of course, media framing related to attribution of Court opinions is malleable; the media can make meaningful and consequential discretionary decisions on how it chooses to portray Court decisions. Nonetheless, the U.S. Supreme Court ultimately has the final say on how a decision is attributed.³

But less is known about how specific attribution choices made by the U.S. Supreme Court affect public agreement with the Court's decisions. One attribution approach is to characterize the decision as merely coming “from the Supreme Court” without providing additional context about which particular justice authored the opinion. By crediting the institution with the decision this approach would seem to garner greater acceptance of the decision by members of the public by connoting neutrality, credibility, clarity, legality, and even institutional legitimacy (Baird 2001; Bartels and Mutz 2009; Brigham 1987; Gibson et al. 2014; Hoekstra 1995; Mondak 1992). This general deference to and reverence of the U.S. High Court (and its actions) has been described as “positivity theory” (e.g., Gibson 2007). In other words, by disentangling the identity of a policy edict from a specific justice (who may be associated with past voting choices or his or her nominating president) and attributing it to the nation's highest legal institution—cloaked in the accouterments of stylized symbols of justice—a verdict that could otherwise appear politically driven is made more palatable, all else being equal. Certainly, the Court has used unattributed per curiam opinions to convey its judgments in particularly volatile cases dealing with politically charged issues involving freedom of the press (New York Times Co. v. United States 1971), capital punishment (Furman v. Georgia 1972) and national elections (Bush v. Gore 2000) among others.

On the other hand, Nicholson and Hansford (2014) cast doubt on this view, as they find minimal evidence that attributing a decision as coming from the U.S. Supreme Court changes how heavily the public relies on partisan cues in its evaluations of a decision. While attributing the outcome of a case to the Court increased public acceptance of the decision in certain instances, the effect was quite small. Assaying a variety of heuristic cues concerning partisan attribution and other political considerations, they found that “public expectations for the Court and its decisions may be no different than expectations for the elected branches of government” (Nicholson and Hansford 2014: 15). Providing additional insight is the work of Zink et al. (2009) on the effect of majority coalition size on citizens' willingness to agree with and find acceptable a U.S. Supreme Court decision. They found that when the justices are unanimous on a case ruling, the public is more likely to agree with the case and find it acceptable than when it is a split majority coalition—even when the Court's decision is at odds with citizens' ideological preferences. This suggests that outcomes that are perceived by the public as not turning on ideological or political lines (i.e., they are agreed to by all justices, regardless of well-known ideological divisions) are more apt to be found satisfactory. In contrast, other studies have found that unanimous votes do not have a positive effect on public views about U.S. Supreme Court decisions (Gibson et al. 2005), and that split decisions may even enhance public views of the Court outcomes under certain conditions (Salamone 2014). These findings provide only indirect evidence for the premise that the Court as an institution can act as a heuristic cue because they focus on Court unanimity rather than attribution, but they do prompt us to think about the impact of different ways in which the U.S. Supreme Court presents policy announcements to the public.

We endeavor to address a closely related concern inspired by the scholarly arguments of Justice Ruth Bader Ginsburg, among others, regarding comparative approaches to how legal institutions convey legal policy decisions. As outlined earlier, nations' methods for delivering high court outcomes differ in interesting ways and even within the U.S. Supreme Court we see the use of both signed and unsigned per curiam legal opinions in cases of policy consequence. Our study focuses on how U.S. Supreme Court case outcomes are conveyed to the public—specifically whether the Court's opinion is attributed to the majority opinion writing justice or to the Court as a whole. We theorize that when the identity of the legal decision is tied to the institution, rather than to an identifiable justice, citizens will be more likely to agree with the Court's decision, all else being equal.

The impact of this institutional cue is likely to vary in interesting ways, depending upon the characteristics of the respondent. More precisely, we posit that the positive relationship between attributing the case outcome to the U.S. Supreme Court (rather than a specific justice) and respondent favorability (i.e., agreement) described above will be conditioned on respondents' more general orientations. First, we expect that this relationship will grow stronger as respondents' ideology becomes more conservative. This possibility is consistent with studies suggesting an alignment between conservative values and generalized reverence toward authority, order, and governing institutions (e.g., Arceneaux and Nicholson 2012; Jost et al. 2003, 2007).⁴ Second, this relationship between U.S. Supreme Court attribution and citizens' agreement with Court outcomes should grow stronger as respondents' general feelings regarding the legitimacy of the Court are more favorable—that is, as feelings of legitimacy toward an institution increase, such positivity is parlayed to appreciation of policy edicts that are attributed to the institution (e.g., Boddery and Yates 2014).⁵

Finally, we anticipate that another source cue may affect the relationship between attribution and agreement: the gender of the opinion author. As the number of women on the U.S. Supreme Court has increased, it is important to ascertain whether attributing a majority opinion to a female justice will impact agreement with the Court's decision, relative to other attribution options.

The literature on gender stereotypes and implicit bias suggests that male and female justices may not be evaluated similarly. Due to stereotypes that link the law with masculine-associated traits (Pierce 1995), studies find that implicit bias appears to undermine assessments of the competence and qualifications of female judges (Gill et al. 2011; Sen 2014) and female attorneys (Bogoch 1997; Brown and Campbell 1997). Recent scholarship on the U.S. Supreme Court also shows that female justices tend to be interrupted more often than male justices by lawyers and their male colleagues, reflecting ingrained power dynamics (Jacobi and Schweers 2017; Patton and Smith 2017). More broadly, women are usually not associated with traits considered to be desirable for political leaders (Huddy and Terkildsen 1993), and at least one study has found that written work authored by women is viewed as less credible than work authored by men (Armstrong and McAdams 2009).

On the other hand, there is some work that suggests that attributing a decision to a female justice might improve agreement or, at the very least, have a similar effect to a male-attributed opinion. Normative arguments about descriptive representation on the U.S. Supreme Court often promote a link between the identity of justices and enhanced institutional legitimacy (Myers 2009; Neff 1981), and there is some evidence supporting the general argument with respect to minority representation (Scherer and Curry 2010).⁶ That is, apart from the substance of the decision, an opinion attributed to a female justice could send the message that the U.S. Supreme Court reflects the diversity of American society, at least in terms of gender, and is thus more “fair” (National Women's Law Center 2016).

Alternatively, the public may be indifferent to the gender of the authoring justice when evaluating whether they agree with a decision. Schneider and Bos (2014) speculate that because the public has so little exposure to women in office, they cannot form clear coherent judgments about their stereotypical qualities. If this is the case, it is not unreasonable to assume that the same dynamic could apply to assessments of female judges, given their underrepresentation in more prestigious and high-profile courts such as the U.S. Supreme Court.⁷ Taken as a whole, then, the literature does not provide us with clear expectations for how the public will respond to female-attributed opinions relative to male-attributed or Court-attributed opinions.

However, a number of studies on gender and political culture (Hill 1981; Norrander and Wilcox 1998; Windett 2011) suggest that the effect of gender attribution cues may vary in systematic ways, reflecting the prevailing cultural environments in which respondents live. Exposure theories argue that individuals' exposure to experiences and socialization can lead to either more egalitarian views or more traditional views about gender, depending on the content of the experiences and the environment (Davis and Greenstein 2009). These attitudes then translate into assessments of political elites. For instance, the traditionalistic political culture of the South (Elazar 1974) has been linked with more traditional attitudes about gender roles (Powers et al. 2003) and poorer electoral prospects for women gubernatorial and legislative candidates (Norrander and Wilcox 1998; Windett 2011). Beyond region, Moore and Vanneman (2003) find that living in a state with a higher proportion of religious fundamentalists is associated with holding more traditionalist attitudes toward gender. Recently, a study using Windett's (2011) measure of female sociopolitical culture (FSC) found that states with higher FSC scores are significantly more likely than low FSC states to elect female attorneys general and to have female candidates for this position in the party primary (Gordon 2016).

Drawing from this work, we expect that respondents living in states with unfavorable political environments for women (Windett 2011) should be more apt to find the U.S. Supreme Court's decision unfavorable if it is attributed to a female justice, and that the reverse should be true for respondents from states with more favorable political environments for women.

Experimental Design

To gauge whether opinion attribution affects the level of agreement a U.S. Supreme Court case disposition yields, we constructed a survey experiment in which we randomly populated three groups and presented respondents with one of three possible treatment vignettes that summarized the holding of an actual Court case. (See Supporting Information S1 for a randomization check and Supporting Information S3 for the vignettes' language.) Prior to the vignettes, respondents were given a pretest that asked basic demographic questions, as well as questions to assess their knowledge of and general feelings toward the U.S. Supreme Court. Following the treatment vignettes, a posttest asked respondents to rate their level of agreement with the Court's decision on a 6-point scale, ranging from strongly disagree to strongly agree.⁸

Each vignette summarized the U.S. Supreme Court's holding in Hiibel v. Sixth Judicial District of Nevada (2004). Hiibel scored a two out of eight on Collins and Cooper's (2012) expanded salience index because the case received news coverage in both the New York Times and Washington Post the day after the Court handed down the decision (see Greenhouse 2004; Lane 2004). Hiibel is a criminal procedure case; this issue area is particularly useful given our research question because it maps consistently onto a traditional left–right spectrum. Within this issue area, conservative case dispositions rule in favor of the government, supporting law and order concerns, and liberal case outcomes favor individual liberties. Arguments advanced in this issue area by the government often advocate its right to perform some kind of police or surveillance tactic, whereas individuals seek to prevent the government from using those procedures typically invoking the protections of the Bill of Rights.

Hiibel involved Nevada's “stop and identify” law. This statute requires individuals to identify themselves to a police officer if asked. Larry Hiibel was convicted under the Nevada statute for failing to identify himself to a police officer upon request and appealed to the U.S. Supreme Court claiming, among other things, that the law violated his Fourth Amendment right against unreasonable searches. Justice Kennedy's majority opinion held that Nevada's governmental interest in ensuring that its police force is capable of identifying, and potentially clearing, individuals of suspicion was a minor intrusion and was not barred by Hiibel's protection against unreasonable searches. Because this case ruled in favor of the government's advocated position, it is regarded as a conservative decision. The justices voting in Hiibel split along traditional ideological lines.⁹ Chief Justice Rehnquist and Justices O'Connor, Scalia, and Thomas joined Justice Kennedy's majority opinion. Justices Souter and Ginsburg joined a dissent penned by Justice Breyer, and Justice Stevens wrote his own dissenting opinion.

The vignette read by Group 1 credited the case disposition simply to “the Supreme Court” and was accompanied by a photograph of the U.S. Supreme Court building's exterior; such per curiam attribution should trigger the mechanisms posited by positivity theory—namely that the Court is uniquely suited to invoke deferential reactions because of its perception as a learned and hallowed institution (Gibson 2007). In the two other treatment groups, respondents were informed that either a male or female justice authored the Court's decision and were prompted with a corresponding picture¹⁰ of the authoring justice as well as appropriate pronoun attribution. Specifically, respondents in Group 2 read a vignette that identified Justice Anthony Kennedy as the author of the majority's opinion, whereas Group 3's vignette identified Justice Sandra Day O'Connor as the majority opinion author.¹¹ The substantive content of the Court's holding was held constant among the treatment groups. Thus, the vignettes differed only with respect to the identity of the majority opinion author and the corresponding pronouns and photographs.

Our study entails 1287 respondents obtained via Amazon's Mechanical Turk (AMT) platform. AMT has proven to be a dependable data source for isolating treatment effects in experimental decision-making settings and has been used in a wide variety of political science and policy contexts (e.g., Bishin et al. 2015; Clifford et al. 2015; Grimmer et al. 2012; Ryan 2012). We recognize that this platform is not without its limitations, however. AMT offers a nonprobability sample of opt-in respondents who tend to be younger, more liberal, and more educated than the general American public (Berinsky et al. 2012; Huber et al. 2012). But compared to student convenience samples, AMT's respondent pool is more representative (Paolacci et al. 2010), and AMT respondents are also more engaged with survey-taking tasks compared to various other respondent pools (Weinberg et al. 2014). In a comparison of AMT with American-based population samples like the American National Election Survey (ANES), Levay et al. (2016) found that while differences between AMT and ANES respondents did exist, these differences did not stem from immeasurable considerations. They concluded that differences between AMT respondents and population-based respondents could be identified and their effects largely ameliorated by accounting for a number of political and demographic controls. Importantly, a recent exhaustive study performed 20 survey experiments using AMT respondents while simultaneously performing the identical 20 experiments using a nationally representative sample, finding virtually indistinguishable results (Mullinix et al. 2015: 122).

Still, Mullinix et al. (2015) and others caution against considering convenience samples such as AMT as unqualified substitutes for population samples (e.g., Huff and Tingley 2015; Krupnikov and Levine 2014). In Supporting Information S1, we address these concerns by providing distribution information for our AMT sample on an array of relevant demographic and political variables. We find that our sample's distribution largely mirrors the patterns in AMT samples identified by Levay et al. (2016) and others (e.g., Berensky et al. 2012). Similar to these studies, our AMT sample respondents are generally younger, lower earning, more liberal, and more likely to be white and male than those respondents typically found in population based samples.¹² Accordingly, we follow the lead of Levay et al. (2016) and include control variables to help address these differences in our regression models. We hasten to add that in our descriptive analyses (Figures 1 and 2), these control variables are not employed, meaning that agreement levels displayed do not necessarily approximate population levels. Rather, the focus for those analyses is on whether there are significant differences across treatment groups.

Figure 1. Agreement with Court's Decision, by Attribution. Notes: Entries in the top graph (mean agreement with decision) are mean agreement scores based on a 1–6 scale (1 = strongly disagree, 2 = disagree, 3 = slightly disagree, 4 = slightly agree, 5 = agree, and 6 = strongly agree). Analysis of variance produced a statistically significant effect (F = 16.84, p < .01) with one-tailed tests. Entries in the bottom graph (proportion agree with decision) are proportions based on a collapsed measure of the agreement scores (0 = disagree, 1 = agree). Difference of proportion analysis produced a statistically significant effect (Z = 3.24, p < .01) with one-tailed tests. For both analyses, N = 1287. The dashed line allows for easier comparison of the effect size across attribution conditions.

Figure 2. Agreement with Court's Decision, by Attribution and Respondent Ideology. Notes: Entries in the top graph (mean agreement with decision) are mean agreement scores based on a 1–6 scale (1 = strongly disagree, 2 = disagree, 3 = slightly disagree, 4 = slightly agree, 5 = agree, and 6 = strongly agree). Analysis of variance produced a statistically significant effect for conservative respondents (F = 18.56, p < .01) and for liberal respondents (F = 5.23, p < .01) with one-tailed tests. Entries in the bottom graph (proportion agree with decision) are proportions based on a collapsed measure of the agreement scores (0 = disagree, 1 = agree). Difference of proportion analysis produced a statistically significant effect for conservative respondents (Z = 3.58, p < .01) and for liberal respondents (Z = 1.70, p < .05) with one-tailed tests. For all analyses, N = 1287. The dashed line allows for easier comparison of the effect size across attribution conditions.

Results

Our survey respondents agreed with the case outcome in Hiibel more often than they disagreed—with 62.2 percent of our sample favoring the case outcome and 37.8 percent not agreeing with it. We also find interesting variation among subcategories of agreement–disagreement. Of those who agreed with the decision, 11.5 percent of respondents strongly agreed, 28.3 percent agreed, and 22.5 percent slightly agreed. Of those who disagreed with the decision, 7.9 percent strongly disagreed, 12.4 percent disagreed, and 17.6 percent slightly disagreed. As depicted in Figure 1, we find support for our primary thesis. Case outcomes attributed to the U.S. Supreme Court rather than individual justices are more likely favored, and this holds for both difference of proportion (collapsed agree/disagree) and difference of means (degree of agreement/disagreement) analyses.

Given that the Hiibel case's outcome is generally regarded as conservative, we are interested in how this consideration factors into our sample respondents' propensity to agree with the decision. We find that 66 percent of self-identified conservatives agree with the holding, whereas just 61 percent of self-identified liberals agree with it. While this gap is not extremely large, we do find that there is a statistically significant difference between liberals and conservatives in both difference of proportions and difference of means tests.¹³ This prompts us to question whether ideological considerations may cast an influence upon our primary proposition regarding attribution's effect on how legal outcomes are viewed. Figure 2 provides preliminary evidence that while a relationship exists between case outcome attribution (to the Court) and respondent agreement—both for self-identified liberals and conservatives—the relationship is stronger for conservatives.

We report ordinary least squares (OLS) regression results for our analyses in Table 1. The second column of the table (Model 1) displays the results of a simple bivariate analysis of our primary thesis, which is supported. When the decision is attributed to justices rather than the U.S. Supreme Court—the excluded reference category—respondents are less likely to agree with the outcome. The results in the third column (Model 2) demonstrate that the relationship is robust to the introduction of relevant controls, including the respondent's race, gender, ideology, knowledge of the Court, feelings regarding the Court's overall legitimacy, age, income, and level of education.¹⁴ (See Supporting Information S2 for variable descriptions.)

Table 1. OLS Estimates for Degree of Agreement/Disagreement with Supreme Court Decision

Note: Standard errors in parentheses.

^** p < .01.

^* p < .05 (two-tailed tests).

In the multivariate model, we see that when the decision was attributed to a justice, this yielded a 0.348 point decrease in citizen agreement (again, on a 1–6 scale from “disagree strongly” to “agree strongly”) relative to the situation in which the decision was attributed to the Court as a whole. We find that respondents' race, education, income level, and knowledge of the U.S. Supreme Court have no statistically significant effect on agreement outcomes. On the other hand, women appear to find the decision in Hiibel moderately more palatable than men (a 0.195 point difference), and the results show that conservatism and positive views on the U.S. Supreme Court's legitimacy are also positively associated with agreement with the decision. In contrast, relative youth (i.e., under 35 years of age) was negatively associated with decision agreement. We also assessed the relationship using a logit model with a collapsed dependent variable in which ranges of disagreement are recoded as 0 and the levels of agreement as 1 (see Table A1). Our findings using this approach largely mirror the results we obtained in our OLS estimation model and support our primary hypothesized relationship.¹⁵

We next revisit the interactive relationship regarding the potential conditioning effect of ideology on attribution of the decision to a justice versus the Court. The interaction results provided in our OLS analysis in Table 1 (Model 3) are depicted graphically in Figure 3 which displays the average marginal effects (Williams 2012) of our primary relationship between decision attribution and respondents' degree of agreement with the case outcome—as conditioned on the respondent's ideology (from extremely liberal [1] to extremely conservative [6]). The negative relationship between justice attribution and agreement attains statistical significance with liberal respondents and grows stronger with increasing levels of self-identified conservatism. In supplemental models using logistic regression, our results are comparable (see Appendix Table A1 and Figure A1). ¹⁶

Figure 3. The Average Marginal Effects of an Opinion Attributed to a Justice Rather than the Court on Case-Specific Agreement as Political Ideology Ranges from Liberal to Conservative. Note: This figure demonstrates an interactive relationship in the linear regression model by plotting the average marginal effect of a justice attributed opinion on the dependent variable, case-level agreement (Hiibel), as political ideology ranges from liberal to conservative.

These results confirm our expectation that respondents with conservative ideological leanings may be more sensitive to the heuristic cue of the nation's highest legal authority, though with some caveats that we will discuss in our concluding remarks. Finally, our statistically significant findings regarding respondents' feelings on the overall legitimacy of the U.S. Supreme Court prompt us to consider the possibility that such views may also act as a conditioning variable for the influence of case attribution. However, we find that the coefficients for the proposed interactive relationship do not reach conventional levels of statistical significance in either the OLS (Table 1, Model 4) or the logit models (Table A1, Model 4).

Source Cues and Opinion Author Gender

Having found evidence that attribution of the majority opinion to a justice reduces agreement compared to Court-attributed opinions we next turn to the question of whether agreement is affected when the gender of the opinion author can be explicitly identified. As the U.S. Supreme Court has grown from an all-male institution to one with multiple women on the bench, we know very little about what it means for an opinion to be attributed to a female justice versus a male justice. To the best of our knowledge, no systematic study to date has evaluated how opinion author gender affects agreement with decisions in issue areas that lack a salient gender dimension.¹⁷ As discussed earlier, criminal procedure cases provide a useful lens for assessing agreement with U.S. Supreme Court decisions because they are both common to the Court's docket and salient to the general public. As the majority of U.S. Supreme Court decisions concern issue areas without an explicit gender dimension, the results of this analysis should allow us to identify any general advantage or disadvantage for case agreement that varies by the gender of the opinion author.¹⁸ Moreover, the four women who have served as justices on the U.S. Supreme Court have been active in writing decisions about a wide range of issues. From 1981 to 2015, female justices penned a total of 549 majority opinions, with nearly one-third of those dealing with criminal procedure.¹⁹ In this assessment we split our prior “Justice attributed” variable into female justice (denoting O'Connor) and male justice (denoting Kennedy) attributed variables, and once again, the reference is the Court-attributed treatment.

We also wish to assess whether sociopolitical context affects agreement with female-attributed opinions. Accordingly, we utilize Windett's (2011) dynamic measure of FSC, updated through 2012, which captures aspects of gendered political culture, general culture, and gendered social culture.²⁰ In our study, high FSC states include Washington, Vermont, and Colorado while low FSC states include Oklahoma, South Carolina, and Louisiana. The former group scores slightly higher in education, but both groups are similar with respect to age. Where respondents live in environments where there are many well-educated, employed women, and women officeholders, this should condition them to hold more egalitarian attitudes about gender roles and thus assess female-attributed opinions more favorably than respondents in low FSC states. We first add the FSC score variable to our existing set of relevant controls (to ascertain its additive effect). We then include it in a set of interactive model specifications.

Another relevant consideration for our investigation is whether male and female respondents will differ systematically in their assessment of female-attributed opinions, compared to male-attributed ones. No studies of which we are aware have directly examined this question, but related research on legitimacy and fairness has produced mixed results (Bartels and Johnston 2013; Cann and Yates 2008; Gibson and Nelson 2015). To unpack this in the context of our research question, we include a set of interactions by respondent gender in which we separately assay our tests regarding how female justices are perceived as opinion writers.

Differences in respondent agreement with Hiibel by gender are only slight. When the decision is attributed to a male justice (Kennedy), 58.16 percent of respondents agreed, compared to 60.1 percent agreement when the female justice (O'Connor) was depicted as the opinion author. Moving beyond these descriptive differences, in Table 2, we present the results of the OLS models²¹ analyzing the impact of gender attribution on agreement with the Court's decision, employing the set of controls discussed previously, along with FSC, and using Court-attributed opinions as the reference category.²²

Table 2. OLS Estimates for Agreement with Supreme Court Decision, Including Justice Gender and Interaction with Respondents' Female Sociopolitical Culture (FSC)

Note: Standard errors in parentheses.

^** p < .01.

^* p < .05 (two-tailed tests).

The results (Model 1) indicate that both male (Justice Kennedy) and female (Justice O'Connor) attributed decisions fare worse than the Court attributed decision—as we might reasonably expect from our prior results. While the coefficient for Male justice is somewhat more negative than Female justice, relative to the Court attribution, the difference between the two was found to not be statistically significant in auxiliary analyses.²³ Control variables largely mirror our prior findings; however, we do find that there is a negative and statistically significant association between FSC and respondent agreement.

As discussed above, we have reason to believe that gender attribution dynamics could differ for respondents in states with lower levels of FSC, who should be less likely to agree with a female-attributed opinion. The third column of Table 2 displays the interactive results regarding opinion attribution (female and male, relative to Court), conditioned upon FSC. In Figure 4, we depict the average marginal effects of the conditional relationship between Female justice and FSC on agreement. The interactive relationship indicates a positive relationship between female justice attribution and agreement as FSC increases to nearly one s.d. above the mean. In sum, attributing a decision to a female justice in a higher FSC environment may augment public agreement with the decision relative to Court and male justice attribution—under limited conditions.

Figure 4. The Effect of a Female Opinion Author on Case-Specific Agreement as Female Sociopolitical Culture Varies. Note: This figure demonstrates an interactive relationship by plotting the average marginal effects of a female opinion author on the dependent variable, case-level agreement (Hiibel), as female sociopolitical culture varies.

We further investigate the effects of gender with regard to respondents in Model 3. Our interactive term (Female justice × Female respondent) indicates no statistically significant relationship. However, the component terms of our interaction are also informative. When female justice is equal to zero, the statistical significance for Female respondent indicates that women respondents favor attributions to Male justice and Court attribution over Female justice attribution. The statistically significant component term Female justice indicates the effect of this variable (Female justice) when the respondent is male (i.e., Female respondent = 0). Here, we see that the coefficient for this component is negative and significant, suggesting that male respondents are less favorable to U.S. High Court decisions that are attributed to female justices.

Finally, we consider the respondent gender dynamic in the context of different sociopolitical environments by employing a triple interaction term: FSC × Female justice × Female respondent. Model 4 (Table 2) displays the results for this analysis, and the average marginal effects of the interactive relationship are depicted in more detail in Figure 5. As both indicate, for female respondents, there is no statistically significant interactive relationship between Female justice attribution and FSC. For male respondents, we see that while they are generally negatively inclined toward the female-authored decision, agreement levels become more positive as FSC rises, although this effect disappears at the highest levels of FSC.

Figure 5. The Effect of a Female Opinion Author on Case-Specific Agreement as Female Sociopolitical Culture Varies among Male and Female Respondents. Note: This figure demonstrates a triple interactive relationship by plotting the average marginal effects of a female opinion author the dependent variable, case-level agreement, as female sociopolitical culture varies among male and female respondents.

Conclusion

In this paper, we have shown that the manner in which a political institution conveys a policy decision to the public has an important influence on how citizens appraise the policy. More specifically, when a U.S. Supreme Court outcome is ascribed to the institution as a whole, rather than to a particular justice, people are more apt to agree with the policy decision. As a number of prior studies have demonstrated, the agreement dynamic that we uncover has important implications for public legitimacy—both for individual policy outcomes and for the institution more generally (e.g., Gibson 2012, 2015).

Our analysis also sheds light on debates over the relative suitability of the U.S. High Court's (and lower courts') present use of a “hybrid” approach for communicating policy decisions (i.e., generally attributing Court opinions of the majority coalition to a specific justice, while allowing other justices to write dissenting or concurring opinions at their discretion). Our findings suggest that this method of policy conveyance may work to undermine potential agreement with legal edicts, notwithstanding legal scholars' arguments to the contrary (e.g., Robbins 2012). In contrast, the U.S. Supreme Court's use of unsigned per curiam opinions may actually serve to help maintain the public's “reservoir of good will” for both specific decisions and the Court more generally. This finding lends credence to calls by some legal academics to do away with individually signed opinions—which they believe lead to unhealthy “judicial individualism” and may ultimately be damaging to the U.S. Supreme Court's standing (Lerner and Lund 2010). In light of trends related to mass polarization and declining confidence in the U.S. Supreme Court, it could be advantageous for the Court's legitimacy to utilize more per curiam opinions in order to dampen ideological cues associated with particular justices. It is conceivable that there may be an upper limit to the use of per curiam opinions—at least in relation to the degree to which they can be used to encourage public agreement. If so, the Court might strategically employ per curiam opinions only for its most salient and potentially politically volatile decisions. Of course, we recognize the importance of replicating our findings in other issue domains to understand more fully the impact on legitimacy.

Furthermore, our findings run counter to prior studies showing that attribution of legal policy decisions to the U.S. Supreme Court has little influence on how the public assesses them (Nicholson and Hansford 2014). It remains to be seen whether our findings will apply in similar fashion to issue areas other than criminal procedure. However, given that this issue area occupies a significant portion of the Court's docket (relative to any other single issue area), the phenomenon we have revealed involves a sizeable share of the U.S. Supreme Court's policy output. It is also possible that such institutional cue-taking dynamics might not be applicable to lower federal or state courts in which the personalities and ideological proclivities of the judicial actors may not be as well-known as those of the justices of the U.S. High Court. Future research might explore this intriguing question as well as the possibility that similar institution-oriented heuristics might affect public views on the policy edicts of other branches of government.

Our paper also breaks new ground in identifying the role that gender plays in how the U.S. Supreme Court's policy decisions are interpreted by the public. In the overall analysis, we find that attributing a decision to the institution is associated with more agreement than attributing it to either a male or a female justice. However, the social, political, and economic environment for women in a state also affects how citizens assess gender cues in the attribution of a decision. In states where female candidates face better odds in winning elections, respondents afford more deference to female-attributed opinions relative to Court-attributed decisions, preferring both over those attributed to a male author. But in states with low levels of FSC, female-attributed opinions are viewed less favorably relative to those attributed to the U.S. Supreme Court. Our findings regarding female sociopolitical context appear to be driven primarily by male respondents; they are most affected by their environment, at least in this regard.

To the best of our knowledge, no existing study has uncovered a gender effect for authorship of U.S. Supreme Court decisions before, and it has important implications for the efficacy of the Court in garnering agreement with its policy pronouncements. While this provides some support for the perspective that having a U.S. Supreme Court that “looks like America” may improve public perceptions of the Court's decisions, it is clear that this legitimizing effect is not uniform. From the Court's perspective, our findings suggest that it may be advantageous to utilize per curiam decisions more often, so as to mitigate differences in public reactions to the gender of the opinion author.

The effects isolated in this study are noteworthy but, like all experimental effects, call for replication. Future research should explore whether opinion attribution effects also appear in more gender-salient issue areas, such as reproductive rights and employment discrimination. Furthermore, it may be advantageous to see if our findings persist for U.S. Supreme Court policy outcomes that are traditionally thought of as liberal in orientation. In the context of our conservative policy vignette, we find that that the Court attribution effect is stronger as respondent ideology grows more conservative. It remains to be seen whether this same conditioning effect (presumably based on conservative deference to authoritarian institutions) holds true for conservatives' favorability toward liberal U.S. Supreme Court outcomes or, conversely, if there is a mirrored effect for more liberal respondents (suggesting an alternative path of such conditioned influence).²⁴ Finally, our primary finding regarding gender and political environment is driven to some degree by male respondents. This finding may be a product of the issue area (criminal procedure) that we utilize or may be a more general phenomenon. Either possibility is intriguing and prompts us to consider the contextual nature of citizen contemplation of U.S. Supreme Court verdicts. Along these lines, additional research should explore the negative relationship between FSC and agreement, even after controlling for ideology. This too could be a function of the criminal context of our experiment, although we cannot say for certain.

Although AMT samples have consistently been shown to mirror results from studies based on probability samples (Mullinix et al. 2015), future studies should also survey a representative sample of individuals in real time, as the U.S. Supreme Court releases its written opinions. Lastly, a natural extension of this study could employ content analysis to examine how media reports treat per curiam opinions compared to attributed opinions handed down by the Court. Given the recent spotlight political actors displeased with certain legal holdings have placed on individual U.S. federal judges (Liptak 2016; Wang 2017), employing per curiam opinions to a greater extent is a reasonable and straightforward way judges may avoid individualized scrutiny while simultaneously protecting the public support that is so crucial to a judiciary's longevity.