The Gender Gap in Supreme Court Legitimacy

Representation

There is strong evidence that women (and others) support institutions that look like them (descriptive representation) and act in their interests (substantive representation) (Mansbridge, 1999; Hayes & Hibbing, 2017).

Descriptive representation matters across different types of institutions. Despite the uniquely non-representative nature of the judicial role, scholars have found repeatedly that descriptive representation influences public evaluations of courts and that gender and racial diversity among judges affects judicial decision-making. Scherer and Curry (2010) show that increasing the (perceived) aggregate racial diversity within the courts (via experiment) increases the legitimacy of the courts among members of the underrepresented class. Similar relationships show up in observational studies. Badas and Stauffer (2018) demonstrate, for example, that women on a survey who identified as conservative or very conservative were more supportive of the confirmation of Justice Kagan (a woman) than men who identified as conservative or very conservative.

The presence of a minority judge on a judicial panel clearly affects decision-making. Black judges are more likely to uphold affirmative action programs than nonblack judges. What is more, random assignment of a black judge to a three-judge panel increases the likelihood that the panel upholds affirmative action programs by about 20 percentage points (Kastellec 2013). The same can be true for the presence (or lack thereof) of a woman on a judge panel. Boyd et al. (2010) use advanced matching techniques to show that women are 10 percentage points more likely to vote in favor of the plaintiff in a sex discrimination case than men. Furthermore, the likelihood that a man rules in favor of the plaintiff increases by 12–14 percentage points when a woman sits on his panel in those same cases.

We argue that evidence of greater representation (whether descriptive or substantive) will increase the legitimacy of the Court among women. A lack of representation will be associated with lesser legitimacy. The Supreme Court has long been dominated by men descriptively and substantively. Descriptively, not a single woman served on the Court until 1981. ⁹ And only recently have even three women served on the Court at the same time. Never have a majority of justices on the Court been women, despite women making up a majority of the US population and a majority (55%) of students in ABA-approved law schools. ¹⁰ This matters greatly. Studies are clear that descriptive representation on courts influences perceived legitimacy (Scherer & Curry, 2010), support for judges (Badas & Stauffer, 2018) as well as actual decision-making (Boyd et al., 2010).

Substantively, the Court’s history is replete with examples of support for legislation treating women as second-class citizens, both politically and legally. For example, it was only after the elected representatives overrode the Court’s unanimous Minor v. Happersett (1875) precedent in 1920 by passing the 19th Amendment that women were guaranteed the right to vote. The 14th Amendment’s equal protection clause was not applied to protect women from gender discrimination until 1971, in Reed v. Reed. As recently as 1961, the Court allowed states to exclude women from jury service unless they made an affirmative request to serve in this capacity (Hoyt v. Florida 1961). While women enjoy greater rights and protections today than in the past, women may be hesitant to extend legitimacy to an institution with this history of negative treatment.

To test our theory that descriptive and substantive representation drive the gender gap, we measure perceptions of gender representation on the Court using two items. The first item asks, “What do you think of the level of female representation on the Supreme Court today?” Response options ranged from well represented to poorly represented on a four-point scale. We also asked respondents, “How important do you feel it is for women to serve on the Supreme Court?” Response options ranged from not important to very important on a four-point scale. ¹¹ We combine the two measures using an additive scale. We expect that women will be most likely to say women are poorly represented and that it is important for women to serve on the Court. Differences in perceptions of representation will correlate with differences in perceptions of the Court’s legitimacy. ¹²

Fairness

The public expects the President and Congress to represent particular viewpoints, for that is the purpose of elections in a democracy. The Supreme Court, on the other hand, is typically viewed as distinct in this sense. Rather than “representing” a party or a subset of people, the justices collectively are expected to act neutrally in interpreting and enforcing the laws. It is in this sense that Alexander Hamilton argued in Federalist #78 that the Supreme Court would exercise “neither FORCE nor WILL, but merely judgment.” Because they are unelected, hold life tenure, and are virtually unaccountable to the public, legitimacy comes when the justices are perceived as resolving disputes fairly and neutrally.

To strengthen this sense of fairness, justices emphasize legal text, precedent, and canons of legal interpretation when justifying their decisions. A vast literature argues that these references to law create perceptions of procedural fairness and perhaps a legitimacy-reinforcing “myth of legality.” Tyler (2006, 375) argues that perceptions of procedural fairness lead people to “feel that they ought to defer to decisions and rules, following them voluntarily out of obligation rather than out of fear or punishment or anticipation of reward.”

In the judicial context, fair procedures can even be more important than favorable outcomes. For example, Hollander-Blumoff & Tyler (2008) studied whether objectively good outcomes were as important as subjectively fair negotiation proceedings between two parties. When negotiations between attorneys were perceived as fair, people reported more positive feelings and greater willingness to accept the outcome, independent of whether the outcome was objectively good. Tyler (2021, 736) argues that “people will focus on how decisions are made, not [just] on the decisions themselves, when making evaluations of fairness.”

Work more directly tied to the federal judiciary produces similar findings. For example, Krewson & Owens (2022) demonstrate that people evaluate federal decisions based on their agreement with outputs as well as the judicial philosophy employed in a case. Using a person’s less preferred judicial philosophy caused them to be less trusting of a judge and less willing to accept a judge’s decision, compared to their trust and acceptance for an identical decision using their preferred philosophy. In other words, liberal (conservative) respondents reacted more negatively to a liberal (conservative) decision that was decided in the “wrong way” compared to the same decision decided the “right” way. Baird & Gangl (2006) demonstrate that the use of legal framing to describe a decision of the Supreme Court strengthened perceptions of procedural fairness. Opinion rationale and other characteristics of decisions independent of the policy content of the outcomes alter perceptions of fairness and shape views of courts (see Farganis, 2012; Zink et al., 2009).

Perceptions of procedural fairness vary by collective experience. Indeed, recent research has heavily emphasized the extent to which group identity shapes perceptions of the judiciary’s fairness. Ono & Zilis (2022) demonstrate that perceptions of impropriety in the rule of law and bias depend on judge characteristics, political identity, and gender. Gibson & Caldeira (1992) demonstrated as early as 1987 that diffuse support among African Americans was “decidedly less positive” than it was among whites, largely because African Americans felt the Court had no longer treated blacks fairly since the Warren Court era.

More recently, Gibson & Nelson (2018) argued that legal symbols—traditionally understood as enhancing Court legitimacy—may work differently among marginalized groups. For example, an image of a gavel may to some indicate rule of law and order, while to others it may represent repression and abuse—see Gibson & Nelson (2018, xix-xx) for enlightening comments on this theme. Gibson et al. (2014) include and exclude legal symbols as part of an experimental manipulation; they find that black students attached negative affect to the symbols more strongly than white students reading their vignette. Other research suggests that legal symbolism works differently between men and women as well. Symbolically, many core concepts surrounding the Court (such as whether its members are acting in a “judicious” manner) are considered to be gendered masculine (Chen & Bryan, 2021), which may naturally lead men to show greater support for the Court. ¹³

Accordingly, we argue that women will be less likely than men to have confidence in the Court’s fairness. This is largely due to the differences in perceptions driven by asymmetrical effects of legal symbols of procedural fairness, as well as derivative effects related to the Court’s representation of women. ¹⁴ If women are consistently less confident that the Supreme Court will make fair decisions, then that view may explain their lower level of support for the Court. To measure perceptions of fairness, we ask “How much confidence do you have in the Supreme Court to make fair decisions?” Responses are coded from “none” to “a great deal” of confidence on a 5-point scale.

Bias

We next consider group affect. It is well known that Republicans and Democrats in the United States increasingly dislike and distrust members of the other political party: “Democrats and Republicans both say that the other party’s members are hypocritical, selfish, and closed-minded, and they are unwilling to socialize across party lines”; this “animosity” towards one another is termed affective polarization and is based on “the power of partisanship as a social identity” (Iyengar et al. 2019, 129). The effects of in-group bias associated with affective polarization are driven by identity rather than driven by differences in policy positions.

Likewise, in-group and out-group orientations may impact Court support. We’ve already shown that the gender gap in diffuse support is independent of political identity (such as ideology and partisanship). We consider now whether women are more likely to see the Court as less legitimate because women identify with other marginalized groups whom they perceive to be mistreated historically by the Court as an institution. Instead of focusing specifically on representation and fairness, here we consider whether people who generally see the Court as biased towards marginalized groups perceive the Court as less legitimate.

As a marginalized group, women may sympathize with other marginalized groups historically disfavored by the Court, including racial and ethnic minorities, LGBTQ individuals, and criminal defendants. This is a reasonable theory to consider. A belief that the Court is biased towards some groups over others has long been acknowledged in the literature as an important predictor of legitimacy (Gibson et al., 2003).

In fact, recent research places beliefs in how the Court treats other groups as crucial to our understanding of public evaluations of the Court. Zilis (2020, 183) argues that “citizens see a court as less legitimate when its decisions benefit groups that they dislike” and he finds support for this argument when analyzing the legitimacy of the Court as a function of affective attitudes towards conservative and liberal groups benefited by its rulings. Consistent with his theoretical arguments and empirical findings, Zilis concludes that court evaluations related to legitimacy and other rule-of-law concepts stem in part “from the psychological tendency to evaluate political outcomes based upon the groups that benefit” (180).

But Court evaluations based on orientations towards out-groups go far beyond “political” groups. For example, Zilis (2021) analyzes observational data to show that antipathy towards gays and immigrants was associated with lower levels of Court legitimacy following same-sex marriage and immigration rulings favoring these groups. He finds similar results using experimental data as they relate to views of other social groups. We think that women are more likely than men to sympathize with marginalized groups; and believing the Court is biased against marginalized groups may be associated with less Supreme Court legitimacy. To test whether the gender gap is driven by a belief that the Court is biased against marginalized groups generally (rather than against women specifically due to a lack of gender representation or perceived unfairness against women), we stated to survey takers that “Some people believe the Supreme Court is biased, while others believe it treats all groups equally.” Then, focusing first explicitly on women, we asked “Thinking about women as a group, do you think Supreme Court decisions tend to favor men or women?” After they answered this question, individuals were provided a likert scale in which, for each of (1) racial and ethnic minorities, (2) LGBTQ individuals, and (3) criminal defendants, they also indicated “the degree to which Supreme Court decisions are also biased against the group.” We combined responses related to each of the four groups using an additive scale.

The Conflictual Nature of the Supreme Court

Finally, we briefly consider whether the “conflictual nature” of the Supreme Court dampens Supreme Court legitimacy for women. On average, perhaps women may have lower levels of tolerance for conflict and disagreement than do men, but a large literature has debated whether this explains gender gaps in political participation and political attitudes (Ulbig & Funk, 1999; Schneider et al., 2016; Wolak, 2020), with evidence available to support both sides of the debate. Judicial scholarship has found that divisiveness on the Court can influence views of the Court (Zink et al., 2009), but no work we know of has tested whether divisiveness on the Court is of particular salience to women relative to men.

It is possible that conflict aversion among women may explain their greater tendency to attribute less legitimacy to the Court. The Court is by definition conflictual. Its job is to pick sides in the context of a case and the judicial role is one of justice (rightness) rather than mercy (compromise). The judges are divided when they resolve salient issues and the Court often entrenches division as it chooses sides in the most contentious political issues of the day (such as on issues surrounding gun rights, abortion, free speech, etc.). Still, little consensus regarding the influence of conflict on gender gaps in the judicial and broader literatures leaves us skeptical of its explanatory power. Given the prominence of conflict-avoidance as an explanation for gender gaps, though, we test its effect here in the judicial context.

To test this theory, we asked individuals first whether they “think the justices on the Supreme Court are divided or in agreement when resolving important cases before them?” Then, we followed up with asking “Do you think Supreme Court decisions tend to increase or decrease conflict in society?” The first question analyzes the internal divisiveness of the Court and the second considers whether the Court’s actions cause conflict in society. Again, we combined the responses to the two questions using an additive scale.

The 2021 Gender Gap in Diffuse Support

The Cooperative Election Study (formerly the Cooperative Congressional Election Study) is a large national stratified sample survey. Half of the questionnaire consists of “Team Content,” in which groups can include questions asked to subsets of 1,000 people per group. The CES survey was in the field from November 2 to December 9 of 2021. This was immediately following the 2021 elections. We measured legitimacy using the same measure as employed in TAPS. This battery of questions continued to perform well, with each of the questions loading on the same factor (each loading greater than .76 on the first factor) and obtaining a cronbach’s alpha of .91. We ordered the responses to the legitimacy battery questions so that larger values indicated greater institutional support. Also as before, we average responses to these legitimacy questions. The resulting legitimacy variable has a range of 1–5.

First, we confirm that the gender gap in legitimacy we observed in previous surveys existed in November-December of 2021. We also want to compare this gender gap to partisan and racial gaps. Figure 3 plots the differences in legitimacy scores reflecting each of these gaps along with the 95 percent confidence intervals of the gap estimates. As before, the gender gap remains, with women feeling the Court is less legitimate. The racial and partisan gaps are larger. The slightly larger racial gap is consistent with our TAPS analysis. Typically, the partisan gap is not this large. It is possible that the solidified conservative Court majority created with the appointment of Justice Barrett or the prominence of the Court as 2021 election issue enhanced the partisan gap in diffuse support.OPEN IN VIEWER

Across an extensive set of high-quality survey waves, we have found that women consistently view the Court as less legitimate than do men. Why? We now answer this question, based on additional and novel questions appended to the 2021 CES module.

Testing Explanations for the 2021 Gender Gap in Diffuse Support

As previously discussed, we employed nine questions to measure perceptions of (1) women’s representation on the Court, (2) the Court’s fairness, (3) its bias against marginalized groups, and (4) the conflictual nature of the Court.

We now test how these four variables explain variation in diffuse support. We first regress our measure of diffuse support on the four measures. In Figure 4, we provide a coefficient plot of our results. Horizontal lines indicate 95 percent confidence intervals. Any horizontal line that intersects with the vertical dashed line indicates statistical insignificance for the associated coefficient. We estimated our model on all the data (filled circles), for women only (filled squares), and for men only (filled triangles).OPEN IN VIEWER

For most of the variables (excluding perceptions of a Biased Court, for men), we find statistically significant relationships between our measures and diffuse support. This is true regardless of the gender of respondent. That is, for each, the estimated effects do not depend on the gender of the respondent. ¹⁵ The direction of the relationships are as expected for most variables. The stronger the perception that the Court is unrepresentative, unfair, or conflictual, the less legitimate people perceived the Court to be. Interestingly, the effect of perceived bias against marginalized groups is in the wrong direction; the more one perceives the Court as biased against marginalized groups, the more legitimate they perceive the Court to be. Finally, the differences across intercepts reflect the residual gender effect: all else equal, men are more likely to see the Court as legitimate.

Clearly, differences in coefficient effect sizes do not explain the gender gap in diffuse support. For example, men and women who see the Court as unfair are equally likely see the Court as less legitimate. Logically, if these viewpoints explain the gender gap it won’t be because of effect size but it may be due to overall differences in the viewpoints themselves. For instance, if women tend to see the Court as unfair and men tend to see the Court as fair, then the coefficient estimates combined with those baseline differences in viewpoints would predict the gender gap we have uncovered over many waves from 2012–2021. That is, legitimacy would be positively correlated with men because they tend to see the Court as fair and negatively correlated with women because they tend to see the Court as unfair.

Thus far, our analytical method has not allowed the strength of viewpoints to differ by gender. We use the lavaan package in R to estimate a structure of equations that tests for such mediation effects. we regress each of our viewpoint measures related to representation, fairness, bias, and conflict on gender, in order to account for overall differences in viewpoints by gender. We also regress legitimacy on each of these viewpoint measures and on gender itself. Doing so allows us to separate the direct effect of gender on legitimacy from its mediated effect. That is, the model separates the effect of gender into its direct effect on legitimacy and its indirect effects through perceptions of Court representativeness, fairness, bias, and conflict. ¹⁶

Figure 5 visualizes our simplified theoretical model, where respondent gender directly impacts Court legitimacy but also indirectly impacts legitimacy through its effect on measures of views of the Court. We already determined that the gender gap in legitimacy is not a result of an interaction between gender and viewpoints related to things like representation (see Figure 4). So, if the gender gap is driven instead by baseline differences in these viewpoints, then we should observe a strong direct connection between Respondent Gender and Perceptions of unfairness and other viewpoints, as well as strong connections between those viewpoints and Court Legitimacy; the direct effect of Respondent Gender on Court Legitimacy should attenuate or disappear.OPEN IN VIEWER

The analysis results are in Table 2. In the system of equations, an exogenous variable is one that predicts other variables but is not predicted itself by other variables. Endogenous variables (predicted by other variables) can themselves explain other endogenous variables. The exogenous variable in our model is an indicator for women. Gender directly predicts one set of endogenous variables (each of the particular views related to the Court). Those endogenous variables predict Court Legitimacy, after accounting for the direct effect of gender itself on legitimacy. The coefficients within the system of equations are estimated simultaneously.

OPEN IN VIEWER

Table 2.

A mediation analysis using a structural equation model, in which respondent gender directly explains legitimacy and indirectly explains legitimacy through its effect on perceptions of gender representation, fairness, bias, and Court conflict. All coefficients estimated in system of equations (sem) model using the lavaan package in R. The coefficient estimates in the Table are standardized for ease of comparability.

	Views of Court				Outcome
	Unrepresentative	Unfair	Biased	Conflictual	Legitimacy
Exogenous
Female Respondent	0.115∗ (0.017)	0.055∗ (0.017)	0.033∗ (0.013)	−0.007 (0.012)	−0.048∗ (0.016)
Endogenous
Unrepresentative					−0.245∗(0.028)
Unfair					−0.362∗ (0.030)
Biased					0.130∗ (0.039)
Conflictual					−0.175∗ (0.042)
Observations: 953BIC: −608.490

We standardized all estimates so they are comparable to one another. First, consider the direct relationship between gender and each of the Views of the Court. The strongest gender effect is on perceptions of the Court as unrepresentative. The effect size of gender on this variable is more than double the effect size of gender on the other viewpoints. The second largest effect of gender is on perceptions of unfairness, followed by perceptions of bias. Gender is not related to a view that the Court is conflictual.

Not only is there a gender difference in belief that the Court poorly represents women, but that variable also has a strong effect on legitimacy itself. Thus, representation (whether descriptive or substantive) is a strong reason for lower legitimacy among American adult women. The other Court-related viewpoint important in explaining the gender gap is perceptions of unfairness. Women are moderately more likely than men to perceive the Court as unfair, and unfairness has a strong effect on legitimacy in the expected direction.

Perceptions of bias towards marginalized groups and a belief that the Court is conflictual cannot explain the gender gap in diffuse support we have uncovered. Perceptions of conflict lead to less legitimacy, but men and women do not differ in their perceptions of the conflictual nature of the Court. As for perceptions of bias, women are slightly and significantly more likely to see the Court as biased against marginalized groups, but this viewpoint is positively related to views of legitimacy and thus cannot explain the gender gap we have uncovered. ¹⁷ Note that the direct effect of gender on legitimacy is still significant after having accounted for the mediated (or indirect) effects of gender. That is, the gender gap is not explained completely by perceptions of the Court’s gender representation and levels of fairness. But accounting for indirect effects is important. In fact, the combined indirect effect of gender on legitimacy through perceptions of both representation and fairness is as large as the estimated direct effect of gender on legitimacy of −.048. ¹⁸ That is, approximately 50% of the total effect of gender on legitimacy can be explained by differences in perceptions of representation and unfairness between men and women.

Exploring Partisans Differences in the Reasons for the Gender Gap in Diffuse Support

We aggregated over women of different partisan identities in the analysis of mediated effects. But Republican women and Democratic women may have different perceptions of the Court’s gender representation, fairness, bias, and conflictual nature. In other words, the reasons for the gender gap in diffuse support might vary between party, with Democratic women and Republican women viewing the Court as less legitimate than men but for different reasons.

To test this consideration, we re-estimated our Table 2 analysis on Republicans only and Democrats only. Table 3 shows the effects of gender (Woman = 1) on each of the measured viewpoints of the Court. The first row shows the estimated effect of gender on these views as already reported. The second row includes results for Republicans only and the third for Democrats only. We see that even within each partisan group women are more likely than men to say the Court poorly represents women. The gender effect on perceptions of representation is strongest within Republicans. For perceptions of Court unfairness, though, we observe the gender differences for Republicans but not for Democrats. The gender effect on perceptions of a biased or conflictual Court are meaningless, regardless of partisanship.

OPEN IN VIEWER

Table 3.

Estimated gender effects (Woman = 1) on perceptions of Court representation, fairness, bias, and conflictual nature, by respondent party identification. Democrat men and women do not differ in perceptions of unfairness. As within Republicans, Democrat men and women differ in their perceptions of gender representation on the Court.

Party	Gender Effect on Latent Variables
	Unrepresentative	Unfair	Biased	Conflictual
All	0.115* (0.017)	0.055∗ (0.017)	0.033∗ (0.013)	−0.007 (0.012)
Republicans	0.131∗ (0.029)	0.067∗ (0.030)	0.023 (0.017)	−0.001 (0.020)
Democrats	0.061∗ (0.018)	0.019 (0.021)	0.004 (0.018)	−0.021 (0.016)

In sum, the gender gap is explained largely by differences between men and women in their perceptions of representation on the Court. Another important predictor—though with no explanatory power for Democrats alone—is that women tend to see the Court as less fair to them. Even for perceptions of representation, the gender differences in viewpoint is twice as large for Republican respondents as it is for Democratic respondents.

For Democrats, the direct effect of gender on legitimacy is no longer statistically significant, meaning that perceptions of representation completely explain differences in legitimacy between Democratic men and women. For Republicans the direct effect of gender remains significant, suggesting that while differences in perceptions of representation and fairness between Republican men and women explain much of the gender gap in legitimacy, they do not explain away the gender gap completely.