PUBLIC SUPPORT FOR THE SURPEME COURT AND ITS LEGITIMACY

Historically, the US Supreme Court has enjoyed higher levels of support and approval from the American public than other institutions of American government, particularly Congress (Gibson and Nelson Reference Gibson and Nelson2015).Footnote ¹³ Scholars have long theorized that this greater public support stems from the way the Supreme Court and other courts present themselves to the public. This positivity theory argues that the legitimizing symbols of law and courts, such as black robes, gavels, and the use of legal language in opinions, teaches the lesson that courts are different from political institutions like Congress and the presidency, and, therefore, are credible and worthy of respect (Gibson, Caldeira, and Spence Reference Gibson, Caldeira and Kenyatta Spence2003; Gibson and Caldeira Reference Gibson and Caldeira2009). According to positivity theory, when the public is exposed to these symbols they are met with an image of courts as impartial, fair, and neutral arbiters of legal questions, not as political policy makers, which serves to enhance the Court’s legitimacy among the public (Tyler Reference Tyler1990; Scheb and Lyons Reference Scheb and Lyons2000; Gibson Reference Gibson2007). Thus, when the public subscribes to this “myth of legality,” the Court’s legitimacy is sustained even when it issues decisions that subsets of the public disagree with. In sum, the apolitical presentation of the Court emphasizing its supposed neutrality in decision making explains the broad diffuse support the public gives to the Supreme Court.

However, recent research has developed additional ways to understand the source of public support for the Supreme Court and its legitimacy. A number of studies have found that people believe that justices’ personal politics and ideologies influence their decisions to at least some extent (Scheb and Lyons Reference Scheb and Lyons2001; Baird and Gangl Reference Baird and Gangl2006; Gibson and Caldeira Reference Gibson and Caldeira2011). That the public views the Court as political is perhaps unsurprising. The confirmation process for Supreme Court nominees is mired in overt partisanship, and elected officials as well as interest groups frame the Court’s decision in political terms. The assumption that people see the Court at least partly in political terms underlies the more recent second strain of research on public support for the Court’s legitimacy. This work finds that support for the Court and its legitimacy is tied to whether members of the public demonstrate ideological or partisan agreement with the Court’s decisions, its perceived ideology, and the elite political actors praising or criticizing the Court (Bartels and Johnston Reference Bartels and Johnston2013; Nicholson and Hansford Reference Nicholson and Hansford2014; Christensen and Glick Reference Christenson and Glick2015; Clark and Kastellec Reference Clark and Kastellec2015; Strother and Gadarian Reference Strother and Gadarian2019). According to this research, support for the Supreme Court mirrors how the public assesses other political institutions: through the lenses of ideology and partisanship. Moreover, media coverage of the Supreme Court has an important role in increasing the ideological or partisan considerations the public uses to evaluate the Court. Research shows that media coverage of the Court has become increasingly partisan and increasingly frames the Court’s decisions in terms of which political party wins and which loses in the case, thus encouraging viewers to see the Court in partisan or ideological terms (Linos and Twist Reference Linos and Twist2016; Hitt and Searles Reference Hitt and Searles2018; Bass Reference Bass2018). Taken together, this recent research on explaining the foundations of support for the Court’s legitimacy suggests important ideological considerations on the part of the public. The public’s view of the Court as partisan and ideological influences how they evaluate the institution.

Importantly, this spatial/party cue understanding of public support for the Court is not diametrically opposed to positivity theory, as the public’s perception of the Court likely reflects elements of both. Indeed, the public’s view of the Court can be effectively described as “half-politics-half-law” (Bybee Reference Bybee2010, 4). Hansford, Intawan, and Nicholson (Reference Hansford, Intawan and Nicholson2018) echo this characterization in finding that the public perceives the Court to be less political than the overtly partisan Congress but more political than a traffic court.Footnote ¹⁴ Given this depiction of the Court as “somewhat political,” both positivity theory and spatial/party cue inform our expectations regarding how the public will evaluate the Supreme Court in light of the Court’s attempts at institutional maintenance.

OFF-BENCH COMMENTS AND THE COURT’S LEGITIMACY

As demonstrated by the various examples of justices’ recent public rhetoric presented above, the Supreme Court is not a bystander in how the Court’s public perception is shaped. Previous research demonstrates that justices take both policy and nonpolicy actions out of consideration for their standing with other institutions, such as Congress, and their support among the public (Mark and Zilis Reference Mark and Zilis2018; Williams and Smith Reference Williams and Smith2018). In their public rhetoric, justices actively frame their role in the political system as apolitical interpreters of law, emphasizing a nonpartisan professionalism to their process. As Posner (Reference Posner2008, 3) notes, justices consciously foster an aura of legality that “exaggerates not only the professional’s skills but also his disinterest” in politics. By projecting an image of themselves as politically neutral, the justices ask the public to evaluate them on criteria rooted in law.

This deliberate attempt at image shaping seems to have grown, as in recent years, Supreme Court justices have increasingly appeared in public (Hasen Reference Hasen2015; Krewson Reference Krewson2019; Glennon and Strother Reference Glennon and Strother2019). For example, using crowd-sourced data on public appearances from scotusmap.com, we find that justices together averaged about 144 public appearances a year from the 2014 through 2019 terms.Footnote ¹⁵ In fact, such is the growth in justice travel that scholars find the amount of such travel to be comparable to trips taken by members of Congress. Despite not needing to court voters for reelection, “on average, Supreme Court justices travel more domestically each year than a member of Congress travels domestically and abroad” (Black, Owens, and Armaly Reference Black, Owens and Armaly2016, 372). These off-bench appearances frequently feature justices giving interviews, appearing on late-night talk shows (for example, Justice Breyer’s appearance on The Late Show with Stephen Colbert to promote his new book), and speeches to law schools, bar associations, and other groups. A look at the total number of public appearances by justice is presented in Figure 1.Footnote ¹⁶

FIGURE 1. Supreme Court Justices’ Public Appearances.

Importantly, scholars argue that the purpose of these trips is often tied to shoring up the Court’s legitimacy (Davis Reference Davis2011; Schmidt Reference Schmidt2013; Black, Owens, and Armaly Reference Black, Owens and Armaly2016; Glennon and Strother Reference Glennon and Strother2019). Black, Owens, and Armaly (Reference Black, Owens and Armaly2016, 378) argue: “By making public speeches or other appearances that accompany travel, justices might enhance public awareness of and support for the Court.” Moreover, in their study of justices’ televised media interviews from 1998 to 2014, Glennon and Strother (Reference Glennon and Strother2019) find that the justices’ primary focus is on reinforcing the Court’s legitimacy.

The question then is whether justices’ goals of preserving or increasing the Court’s legitimacy with their public rhetoric is effective. Two recent studies have begun investigating this question (Krewson Reference Krewson2019; Strother and Glennon Reference Strother and Glennon2021). However, with respect to enhancing legitimacy, these studies present mixed findings, with Strother and Glennon (Reference Strother and Glennon2021) finding more consistent evidence that off-bench judicial rhetoric increased legitimacy. In keeping with recent literature pushing back against positivity theory as the primary source of the Court’s diffuse support, Strother and Glennon (Reference Strother and Glennon2021) find that respondents who disagreed with the Court’s decision were not affected by the legitimizing rhetoric.Footnote ¹⁷

Our article makes two important contributions to this literature. First, neither of these articles test whether characteristics of the particular justice who is delivering the message influence the extent to which legitimizing rhetoric impacts perceptions that the Court is legitimate (and affects the sense that it is a legal or political body). However, in this hyperpolarized world, with several recent high-profile nomination hearings during Trump’s presidency, it is likely that perceived ideological characteristics of the justices delivering this legitimizing rhetoric influence the extent to which it is effective.

Second, we also expect that the gender identity of the public could influence the extent to which individuals are convinced by their off-bench comments.Footnote ¹⁸ Thus, we test whether the strength of respondents’ gender identities moderates the degree to which they are affected by justices’ off-bench legitimizing rhetoric. Recent research finds mixed results regarding the impact of gender on approval of Supreme Court justices. This research focuses primarily on Supreme Court nominees. Badas and Stauffer (Reference Badas and Stauffer2018) find that sharing a gender identity with a Supreme Court nominee led to increased support for that nominee even among ideological opponents, as conservative women increased their support for Elena Kagan. However, Kaslovsky, Rogowski, and Stone (Reference Kaslovsky, Rogowski and Stone2021) find no evidence that respondents increase their support of judicial nominees that share their same gender identity. Nevertheless, we expect that women with strong gender identities will be more likely to be persuaded by the copartisan female justices whom they are exposed to, as shared social identities tend to lead to increased levels of support and sense of legitimacy among the public (Mansbridge Reference Mansbridge1999; Sapiro Reference Sapiro1981; Plutzer and Zipp Reference Plutzer and Zipp1996; Fox Reference Fox1997; Sanbonmatsu Reference Sanbonmatsu2002; Brians Reference Brians2005; Scherer and Curry Reference Scherer and Curry2010; Badas and Stauffer Reference Badas and Stauffer2018).

Our article adds to a growing literature assessing whether this increased use of legitimacy-reinforcing rhetoric fulfills its purpose of increasing the public’s support for the Court’s legitimacy, which if pundits are to be believed regarding a “legitimacy crisis,” is more important than ever for the future viability of the Supreme Court. Importantly, our article contributes to recent research showing an ideological underpinning for the foundation of institutional support for the Supreme Court.

HYPOTHESES

All hypotheses were preregistered with the Open Science Framework (see Appendix C available in the supplementary material).Footnote ¹⁹ In line with Strother and Glennon (Reference Strother and Glennon2021), we expect that legitimizing rhetoric from Supreme Court justices will have the effect of reducing people’s view of the Court as a political institution and strengthening their view of the Court as legal and their support for the Court’s legitimacy. Such a finding would be consistent with positivity theory, as much of the justices’ rhetoric in their public appearances is geared toward presenting the Court as apolitical and as a legal institution. For positivity theory, “to know courts is to love them” (Gibson, Caldeira, and Baird Reference Gibson, Caldeira and Baird1998, 345), meaning that any exposure to law and courts should cause individuals to see the Court as apolitical and thus improve legitimacy support. This should be especially true for exposure to legitimizing rhetoric. Thus, we expect:

However, in line with recent research on ideological and party cues’ impacts on public support for the Court, we expect that the strength of individuals’ perceptions of the Court’s legal and political nature and legitimacy will depend on individuals’ ideological preferences and the identity of the justice providing rhetoric. Conservatives will be more likely to have higher ratings of Court legitimacy and to describe the Court as legal when they are met with legitimizing language from an ideologically similar justice. Conservatives will also have lower ratings of the Court as political. We expect the same for liberal respondents. This expectation echoes recent and developing research on the role of individuals’ policy preferences in driving public attitudes toward the Supreme Court. For example, Gadarian and Strother (Reference Gadarian and Strother2023) find “strong evidence that the Court’s hybrid legal-political nature enables a unique form of policy-motivated reasoning: respondents who agree with Court outputs view the Court and its work as more ‘legal’ in nature, while those who disagree view both as more ‘political’” (1). This leads to our second set of hypotheses:

Our final hypotheses consider the role of gender identity in moderating public perception of the Court and support for its legitimacy. For women, high levels of identifying as a woman will lead to a greater effect of the justices’ legitimizing rhetoric on these women’s perceptions of the legal and political nature of the Court and belief in the Court’s legitimacy, but only within partisanship/ideological groups, leading to our third set of hypotheses:

DATA AND METHODS

In order to evaluate these hypotheses, we fielded an original survey experiment through the firm Lucid in March 2022 (N = 1,351). While Lucid makes use of an online (opt-in) panel of respondents, like Qualtrics, Survey Sampling International (SSI), or mTurk, it has been demonstrated to track well with US national benchmarks and to be suitable for experimental research (Coppock and McClellan Reference Coppock and McClellan2019). Respondents were recruited to match national distributions on gender, race, ethnicity, region, and age (see Appendix A for descriptive statistics). After consenting to the survey, respondents were asked a series of demographic and ideological questions pretreatment. Here, they reported their ideology (with the standard seven-point scale) and the extent to which they identified with their gender group (respondents were asked: How important is being [male/female] to your identity?).Footnote ²⁰ Respondents also answered two attention-check items immediately before the treatment. Following Berinsky, Margolis, and Sances (Reference Berinsky, Margolis and Sances2014), we did not exclude respondents who failed the attention checks from the data, but we retained them so that we could perform analyses on the full data set as well as on those respondents who passed at least one attention check.

Respondents were then randomly assigned to one of three conditions. In the first two conditions, respondents read a news article, presented in Figure 2, about the ways in which the Supreme Court is a legal, not political, body.Footnote ²¹ In the first treatment, the justice interviewed was Elena Kagan and in the second, it was Amy Coney Barrett. The respondents were also informed of the perceived ideology of the justices (i.e., “Kagan is seen as a liberal justice” and “Barrett is seen as a conservative justice”) and given a picture of the justice in her robes. Respondents in the control condition did not read a news article, following Strother and Glennon (Reference Strother and Glennon2021). Note that between the two treatments, other than the shift in justice, the only manipulation is a single word (i.e., “liberal” or “conservative”). Otherwise, the text is identical.

FIGURE 2. Treatment Text.

We intentionally chose Justices Kagan and Barrett in order to keep the gender and race of the featured justice constant, while only varying the perceived ideology. We made the choice to focus on white women given the constraints of the identities of sitting justices. While at the time of fielding the experiment, we could have used two white men with varying ideologies, it was already known that Justice Breyer, the white male on the Court that leaned liberal, would soon retire. We were unable to produce a pair of justices for which race could be held constant without using white justices (as, at the time, there was only one Latina and one Black male justice). Thus, we settled on using two white women, which the makeup of the Court allowed, to focus on varying only the perceived ideology of the justice. With that being said, it should be noted that Justice Kagan and Justice Barrett are by no means identical, even apart from their ideological differences. Justice Kagan is Jewish and from the North while Justice Barrett is Catholic and from the South. Nevertheless, sharing racial and gender identities is an important feature—as these identities are the most readily apparent and visible categories that others notice (Brewer and Lui Reference Brewer and Lui1989; Devine Reference Devine1989; Devine and Elliot Reference Devine and Elliot1995). Thus, while there are certainly differences between Kagan and Barrett, they share two key identities while differing on ideology. We certainly recognize the potentially powerful role that other identities such as those relating to race and religion may play in impacting public perceptions of the Supreme Court’s legitimacy. However, our focus on gender identity reflects the constraints described above.

Further, focusing on women in particular is potentially problematic, as the presence of women and female leaders in institutions tends to lead to increased perceptions of legitimacy (Scherer and Curry Reference Scherer and Curry2010; Stauffer Reference Stauffer2021) and less of a sense that an institution is corrupt (Armstrong et al. Reference Armstrong, Barnes, O’Brien and Taylor-Robinson2022). However, recall that we are building on previous work by Strother and Glennon (Reference Strother and Glennon2021) and Krewson (Reference Krewson2019), who both demonstrate that legitimizing rhetoric does increase perceptions of legitimacy (Strother and Glennon Reference Strother and Glennon2021) and institutional loyalty (Krewson Reference Krewson2019) for the Supreme Court when delivered by both female and male justices. Strother and Glennon (Reference Strother and Glennon2021) demonstrate in multiple experiments that legitimizing rhetoric attributed to Chief Justice John Roberts, a man, effectively increased the extent to which the Court was perceived as legitimate by respondents. Krewson (Reference Krewson2019) demonstrated the same effect, for language attributed to either Justice Sotomayor, a woman, or Justice Alito, a man. Thus, we build on previous research that demonstrates that legitimizing rhetoric from justices (both male and female) increases perceived legitimacy of the Court by uncovering the contours of these effects. We want to understand whether this connection varies based on the identity of the justice who is delivering the legitimizing message.

After the treatment, respondents answered a series of questions about the Supreme Court. First, respondents were asked: “How well does the word ‘political’ describe the Supreme Court?” (five-point scale ranging from “not well at all” to “extremely well”). Then, they were asked: “How well does the word ‘legal’ describe the Supreme Court?” with the same answer options. These items follow Strother and Glennon (Reference Strother and Glennon2021), who use them to evaluate the extent to which respondents see the Court as a political or legal body—that is, the extent to which they were convinced by the legitimizing rhetoric of the justice.

Next, respondents indicated their level of agreement with a series of items measuring legitimacy (Gibson, Caldeira, and Spence Reference Gibson, Caldeira and Kenyatta Spence2003). This battery is the most common measure that is used to estimate the extent to which individuals see the Supreme Court as a legitimate body. This scale is composed of the following items, to which respondents indicate their level of agreement:

1. 1. If the U.S. Supreme Court started making a lot of decisions that most people disagree with, it might be better to do away with the Supreme Court altogether.*

2. 2. The right of the Supreme Court to decide certain types of controversial issues should be reduced.*

3. 3. The Supreme Court can usually be trusted to make decisions that are right for the country as a whole.

4. 4. The decisions of the U.S. Supreme Court favor some groups more than others.*

5. 5. The Supreme Court can usually be trusted to make decisions in the best interest of the American people.

The first, second, and fourth items were recoded so that for every item, higher levels of agreement meant greater perceptions of legitimacy. Then, all items were combined (linearly) into a scale, following previous work. Responses range from 1 to 6, with a mean of 3.69.

After respondents completed the dependent variables, they completed a manipulation check to ensure that they understood and retained the information presented in the article they read. Then, they were debriefed about the article—learning that it was a fabricated article but that it pulled information from real sources.

ANALYSES

Before getting to the main analyses, note that balance tests revealed that randomization was successful and manipulation tests confirmed that respondents understood and retained all information from the treatments. These results can be found in Appendix A in the supplementary material.

To evaluate our first set of hypotheses, which followed previous research in expecting that reading either article containing legitimizing rhetoric from a justice would lessen perceptions that the Court is political (Hypothesis 1a) and lead to higher perceptions that the Court is legal (Hypothesis 1b) and legitimate (Hypothesis 1c), we regressed each of these dependent variables on treatment indicators using OLS regressions. Controls are not needed in experiments to estimate causal effects, as the randomization process ensures that groups are equal on both observable and unobservable characteristics (nevertheless, we include models with control variables in Appendix B in the supplementary material; results are robust to this specification for all models presented in the body of this article). The results for respondents who passed at least one attention check are presented in Table 1. Compared to the control condition, respondents in both treatment conditions were significantly less likely to say that the Court is political, significantly more likely to say that it is legal, and significantly more likely to see the Court as legitimate. Thus, this lends evidence for the first set of hypotheses and replicates previous findings: off-bench comments increase the perceived legitimacy of the Supreme Court.

TABLE 1. Average Treatment Effects

Note: *p<0.1, **p<0.05, ***p<0.01; includes respondents who passed at least one attention check. Models are OLS regression models. Models that include control variables presented in Appendix B in the supplementary material; results robust.

To evaluate whether ideology plays a role in the extent to which a justice’s rhetoric can preserve legitimacy and avoid perceptions of politicization of the Supreme Court, we turn to evaluating the second set of hypotheses. We expect that individuals’ own ideological beliefs will likely moderate the extent to which the justice’s off-bench comments deter perceptions of politics (Hypothesis 2a), strengthen belief in the Court’s legality (Hypothesis 2b), and impact legitimacy (Hypothesis 2c). Table 2 presents the results from sets of OLS models predicting each of the dependent variables with the indicator for the treatment condition, within ideological groups. In the supplementary material, Table B1 presents the results from three OLS regressions in which the treatment condition is interacted with a naive measure of ideology that breaks respondents into the categories of liberal, moderate, and conservative, for interpretability (for use in generating Figure 3).

TABLE 2. Average Treatment Effects by Ideological Group

Note: *p<0.1, **p<0.05, ***p<0.01; includes respondents who passed at least one attention check. Models are OLS regressions specified separately for liberals and conservatives; see Appendix B for OLS regression models that include moderates as well, which are used to make Figure 3 but are omitted here for ease of interpretation. Models that include control variables presented in Appendix B; results robust.

FIGURE 3. Predicted Perceptions that the Court Is (a) Political and (b) Legitimate, by Respondent Ideology.

Table 2 demonstrates that liberals were less likely to see the Court as political after being exposed to legitimizing rhetoric. In both the Kagan and Barrett treatment conditions, the legitimizing rhetoric makes liberals see the Court as less political, compared to the control condition. However, note that this effect is almost double the size in the Kagan condition (the OLS regression coefficient, beta, is −0.82) compared to the Barrett condition (beta = −0.48). This difference in effect size between the Kagan and Barrett conditions is statistically significant. This means that liberals were more convinced that the Court is not political when exposed to Kagan than they were when exposed to Barrett. Similarly, liberals were more likely to see the Court as legitimate when exposed to Kagan’s rhetoric, but there was no corresponding effect for exposure to Barrett’s rhetoric, compared to the control (again, the difference between the Kagan and Barrett treatments is statistically significant when the control condition is omitted from the analysis). There are no effects for the extent to which liberals see the Court as legal.

On the whole, this lends support for Hypotheses 2a and 2c for liberals. Respondents in this ideological group were less likely to see the Court as political and more likely to see it as legitimate when exposed to Elena Kagan’s legitimizing rhetoric, compared to when they were exposed to Barrett’s. Indeed, reading a news article that includes Justice Barrett’s contention that the Court is legitimate had absolutely no effect on the extent to which liberals believe the Court is legitimate.

Turning to the second set of models, we see that conservatives were swayed by both justices. Exposure to legitimizing rhetoric from Kagan and Barrett increased conservatives’ agreement that the Court is not political and that it is a legal institution and legitimate. While the coefficient on the Barrett condition is often larger than that on the Kagan condition, the differences are not statistically significant between the two treatment conditions for conservatives like they were for liberals. This indicates that conservatives were similarly swayed by the legitimizing rhetoric, regardless of the source.

In order to better interpret these findings, in Figure 3 we plot the predicted values for the first and third interaction models from the interaction model (Table B1 in the supplementary material) by ideology and treatment condition. Figure 3(a) shows that respondents in both conditions, regardless of ideology, reported lower levels of believing that the Supreme Court is political, relative to the control condition. However, it is also clear that respondents in the Barrett condition were more divided than they were in other conditions. Conservatives reported even lower levels of seeing the Court as political when they read Barrett’s remarks, while liberals were less persuaded by Barrett than they were by Kagan.

Similarly, there was ideological divergence in the extent to which respondents saw the Court as legitimate, based on treatment condition. Figure 3(b) demonstrates that conservative respondents were more likely to see the Court as legitimate after reading either article, but they were most likely to see the Court as legitimate when they read the Barrett article (though recall that the difference between the Barrett and Kagan conditions is not statistically significant for conservatives). Liberal respondents, on the other hand, were more likely to see the Court as legitimate when they read the Kagan article (compared to the control), but they were no more likely (compared to the control) to see the Court as legitimate when they read the Barrett article. For liberals, the Kagan article persuaded them that the Court was legitimate more than the Barrett article did (and this difference between treatment conditions is statistically significant).

This incongruity could be due to differences between liberals and conservatives, but there may be other factors at play as well. Barrett was confirmed more recently than Kagan, with a hotly contested nomination by former President Trump, a particularly polarizing figure. So, Barrett may have elicited the strongest reactions because she was at the forefront of respondents’ minds. Finally, and perhaps more likely as it requires respondents to possess less knowledge about the Supreme Court, conservatives may perceive the Court as more legitimate because it is currently dominated by conservatives, resulting in conservative rulings. Thus, even if they are unaware of the particular justices or ideological makeup, they may still hear about rulings from the Court that are in their (and their ideological group’s) favor.

Thus, we find mixed support for our second set of hypotheses. Liberals were more persuaded when exposed to the liberal justice than when exposed to the conservative justice that the Court is legitimate, but conservatives were persuaded by both justices. There are stark ideological differences in the effect that legitimizing rhetoric can have on perceptions of legitimacy by ideology, but these differences are largely contained to when Justice Amy Coney Barrett makes the off-bench comments. As a note, we also analyzed whether the treatment effects were moderated by partisanship, as partisanship and ideology are closely aligned. These models are presented in Appendix B in the supplementary material. For the most part, they corroborate the findings here. Democrats are more persuaded by Kagan than Barrett (though these differences are not statistically distinct), and Republicans are more persuaded by Barrett than Kagan (these differences are statistically distinct). We foreground the ideology models rather than the partisanship models as they more directly correspond to the manipulation in the treatment (i.e., saying that the justice is “liberal” or “conservative”). However, the additional partisan models lend credibility to the broad contention we put forth here: that the degree to which people are persuaded by off-bench comments depends on their own political views.

Last, we turn to analyses of whether the extent to which women in the sample identify as women affects their perceptions of the Supreme Court and its legitimacy after hearing a female justice make off-bench comments. We expect that it will, but that the effects will be constrained to respondents who share the same ideology as the justice to whom they are exposed. To evaluate this, we specify a series of OLS regression models that regress each of the three dependent variables on the measure of female identity, within ideological groups. Table 3 reports these findings. The first set of models do not subset female respondents by ideology, and it is clear that there are no overall interaction effects—this follows previous work demonstrating that female identity does not unite women above and beyond partisanship (Klar Reference Klar2018). The second set of models only includes liberal respondents. Here, the interaction between the Kagan treatment and female identity is positive and statistically significant for the legitimacy battery. In the Kagan treatment condition, as female respondents’ level of gender identity increases, so too does their belief that the Supreme Court is legitimate, after reading an article in which Justice Kagan made such an argument. Similarly, the third set of models demonstrates that the interaction between the Barrett condition and gender identity is statistically significant and positive for conservative women, indicating that for conservative women, as the strength of their gender identity increases, they are more willing to adopt Barrett’s argument that the Supreme Court is legitimate. These effects do not hold for perceptions that the Court is political or legal (Hypotheses 3a and 3b), but instead are contained to the legitimacy battery (Hypothesis 3c).

TABLE 3. The Effect of Gender Identity on Perceptions of SCOTUS, by Treatment Condition and Respondent Ideology

Note: *p<0.1, **p<0.05, ***p<0.01; includes female respondents who passed at least one attention check. All models are OLS regressions. Models that include control variables presented in Appendix B; results robust.

In order to visualize these interaction effects, we plot the predicted values for the legitimacy battery as a function of the treatment condition and the strength of gender identity for women in the sample. These are presented in Figure 4. Figure 4(a) shows that for liberal women, as gender identity increases, so too does their evaluation of the Supreme Court as legitimate, but only when they are exposed to the Kagan condition. This effect does not hold for the Barrett condition, as expected. This effect is also not present in the control, which lends credibility to the finding that Kagan specifically elicited this response; there is no general link between gender identity and evaluation of the Supreme Court in the abstract. Figure 4(b) demonstrates that increases in gender identity led to increases in perceptions that the Supreme Court is legitimate for conservative women as well, but only in the Barrett condition, as expected. This link does not exist for conservative women exposed to the Kagan or control condition. Similarly, Amy Coney Barrett, a justice perceived as conservative, was able to activate conservative women’s gender identity for use in their evaluations of the Supreme Court.

FIGURE 4. Predicted Perceptions that the Supreme Court Is Legitimate Based on Treatment Condition and Levels of Female Identity for Women, by Ideology.

We also analyzed whether male respondents’ gender identity affected the extent to which they were persuaded by the off-bench comments from two female justices. These results are presented in the supplementary material in Appendix B. Across these models, we only see one significant interaction effect. Male respondents that strongly identify with their gender are less likely to say that the Court is political in the Kagan condition, compared to the control, and this effect is driven by conservative men. It is unclear exactly why this would be the case—it could be that these men are responding to another aspect of Kagan’s identity and, as a result, seeing her as more credible. However, to test this more rigorously, future work should vary the gender of the justice. Ideally, we would be able to examine how male identity impacts respondents’ reactions to justices when the justice is a male versus when the justice is a female. With the design of the experiment presented here, though, we are unable to do so.

On the whole, these findings illustrate that the extent to which off-bench comments can impact perceptions of the Supreme Court’s legitimacy does depend on the ideological and gender characteristics of the individuals exposed to the comments and the justices making the comments. Ideologically, liberals are more swayed by Kagan than they are by Barrett, while conservatives are swayed by both. Further, women’s sense of gender identity does matter when they hear female justices making off-bench comments—but only within ideological groups. A greater sense of female identity increases perceptions that the Court is legitimate, but only when the female justice conveying the message shares the same ideological identity. While all of these results are relatively minor shifts, this is common in experiments, especially one with such a minor intervention as this one (i.e., a short news article that only varies one word (“liberal” or “conservative”) between the treatments). Keep in mind that in the real world, media coverage would likely be more extensive in its coverage of the ideology of justices, which would likely produce stronger shifts in attitudes. Nevertheless, this minor intervention indicates that the effect of off-bench comments does depend on the ideology of the justice making the comment.