GENDER, RACE, AND SELF-ASSESSMENT

A large body of research in the social sciences has documented that individuals' confidence in their abilities and intellect is heavily influenced by the structural hierarchies that assign meaning to race and gender. Social forces like family, mass media, teachers, and other cultural institutions shape the expectations that individuals have for themselves and for others (Reference Bian, Leslie and CimpianBian et al., 2017; Reference Furnham and RawlesFurnham & Rawles, 1995; Reference PajaresPajares, 2002).

With respect to gender, children internalize messages that relay different expectations for boys and girls, as well as assertions that particular domains are masculine (e.g., math and science) or feminine (e.g., language arts).Footnote ³ As a result of these and other processes, researchers consistently find that “boys are more likely to express confidence in skills they may not possess and to express overconfidence in skills they do possess” (Reference PajaresPajares, 2002, p. 118). Regardless of objective indicators of ability, research confirms that girls are more prone than boys to experience internal distress, evaluate themselves negatively, and worry about their academic performance (Reference Pomerantz, Altermatt and SaxonPomerantz et al., 2002; Reference Silverman, La Greca and WassersteinSilverman et al., 1995). These confidence differences persist into adulthood (Reference Kling, Hyde, Showers and BuswellKling et al., 1999) and have been documented even among highly accomplished individuals such as medical professionals (Reference Blanch, Hall, Roter and FrankelBlanch et al., 2008). When tasks are masculine-typed, women are more likely than men to display inaccurately low self-perceptions of their performance (Reference Beyer and BowdenBeyer & Bowden, 1997) and to underestimate their intelligence relative to men (Reference Furnham and RawlesFurnham & Rawles, 1995).

With respect to race, research points to a similar mechanism whereby negative societal expectations about race contribute to low self-esteem and reduced performance among racial minorities. From this perspective, harmful stereotypes associated with racial identity (e.g., African Americans are not as intelligent as whites) can create an unconscious fear of confirming the negative stereotype (Reference SteeleSteele, 1997). Empirical evidence suggests that this stereotype threat emerges early on in life and undermines self-confidence, resulting in reduced academic performance among minority children and young adults (Reference Vecci and ZelinskyVecci & Zelinsky, 2019; Reference WasserbergWasserberg, 2014). Findings from multiple experimental studies of adults consistently document underperformance by African Americans who were primed by a negative stereotype; in the absence of the primed stereotype, performance was consistent with ability (Reference Steele, Aronson and SpencerSteele et al., 2007). Outside of the lab setting, research indicates that those who were the only people of color in their office experienced a greater degree of stereotype threat, which subsequently affected how those workers interpreted feedback from peers and supervisors (Reference Roberson, Deitch, Brief and BlockRoberson et al., 2003). Scholars have also observed the consequences of stereotype threat for professionals in higher education, including the legal academy (Reference ManningManning, 2018; Reference McClainMcClain, 2016; Reference WhaleyWhaley, 2018).Footnote ⁴

Differences in confidence may also manifest themselves in a psychological phenomenon known as impostor syndrome (Reference Clance and ImesClance & Imes, 1978). Impostors believe that they are unworthy of success and fear that they will be revealed as frauds, which drives them “to work even harder to keep up the charade” (Reference Ewing, Richardson, James-Myers and RussellEwing et al., 1996, p. 54). As such, impostorism is often characterized by extreme overpreparation, a desire to be special relative to peers, and perfectionistic tendencies (Reference Sakulku and AlexanderSakulku & Alexander, 2011) that drive self-perceived impostors to “hold themselves to exceptionally high standards” (Reference Cokley, McClain, Enciso and MartinezCokley et al., 2013, p. 85). Research has documented impostorism in a variety of male-dominated professional contexts for women (Reference Henning, Ey and ShawHenning et al., 1998; Reference Vaughn, Taasoobshirazi and JohnsonVaughn et al., 2019), as well as among racial and ethnic minorities in predominantly white settings (Reference Ewing, Richardson, James-Myers and RussellEwing et al., 1996; Reference Lige, Peteet and BrownLige et al., 2017; Reference Peteet, Montgomery and WeekesPeteet et al., 2015).

Next, we explore how these feelings of otherness may be fueled in law school and in subsequent experiences in the legal profession, ultimately affecting the behavior of the elite few who attain judicial office.

EXPERIENCES IN LEGAL EDUCATION AND THE PATH TO THE BENCH

Through most of the 20th century, law schools were almost exclusively white and male in their student composition (Reference TelesTeles, 2012), and the few women and racial minorities who pursued a legal education encountered outright hostility and explicit discrimination from professors and classmates alike (Reference Haire and MoyerHaire & Moyer, 2015). By the end of 2000, however, women had made significant inroads, making up just under half of the student population (Reference Moyer and HaireMoyer & Haire, 2015). With respect to race, one report noted that from 1993 to 2013, the percent of law school graduates who were racial minorities increased from 14% to 25% (NALP, 2014).

In spite of the trend toward a somewhat more diverse law student population, law schools have been the subject of critiques about the ways in which the climate and pedagogical approach undermines the self-confidence of women and people of color and fosters feelings of alienation (Reference Weiss and MellingWeiss & Melling, 1988). One study found that women and racial minorities were significantly more likely than white men to agree with the statement, “Before law school I thought of myself as intelligent and articulate, but I often don't feel that way about myself now” (Reference FischerFischer, 1996; Reference Guinier, Fine and BalinGuinier et al., 1994; Reference KrauskopfKrauskopf, 1994). Another account from first-year law students noted that,

The homogeneity of law school faculty certainly plays a role in these critiques (Reference FischerFischer, 1996), particularly during the period of time that present-day judges attended law school. Indeed, half of US Courts of Appeals judges actively serving between 2008 and 2018 graduated from law school by 1972, and 75% earned law degrees by 1982 (FJC, 2020). In the mid-70s, faculty were almost entirely white (96%) and male (93%) (Reference FossumFossum, 1983). In contrast, by 2002, two-thirds of tenured law faculty were white men (Reference Mertz, Frances, Katherine, Wamucii, Molly and JoanneMertz et al., 2012). The numbers of female law faculty have risen, but they are still less likely to earn tenure (Reference NeumannNeumann, 2000) and are overrepresented in lower paid, nontenure track positions teaching legal writing. The authors of a 2001 report criticized this practice, noting that while “virtually all lawyers and judges acknowledge that legal writing is the single most important course in law school and agree that this course provides the fundamental underpinnings of law practice,… the overwhelmingly male power structure in law schools disdainfully treats teaching this course as ‘women's work’” (Reference Stanchi and LevineStanchi & Levine, 2001, p. 5). Women of color who teach legal writing describe how they are “presumed incompetent” and report the need to continually demonstrate their legitimacy to their fellow colleagues (Reference BannaiBannai, 2014). One recent book documents substantial, widespread professional and personal challenges faced by women of color law faculty that extend to all levels of the legal academy, reinforcing the idea that non-white women are “outsiders” or even “intruders” in the white, male legal profession (Reference DeoDeo, 2019, p. 41).

Beyond law school, women and racial minorities also deal with occupational stereotyping that undermines their sense of belonging in the legal profession. Surveys of attorneys reveal that lawyers who are not white males often deal with embarrassing and distressing instances of “mistaken identity” from peers, clients, and even judges (e.g., Reference EpnerEpner, 2006). For example, black male attorneys report being confused with criminal defendants, while Latinx attorneys are confused with interpreters and women of color assumed to be administrative staff (Reference RhodeRhode, 1994; Reference Garcia-LopezGarcia-Lopez, 2008). White women, women of color, and—to a lesser extent—non-white men are more likely than white male attorneys to have experienced a situation where a client requested a new attorney (Reference Nelson, Sendroiu, Dinovitzer and DaweNelson et al., 2019), and black judges are more likely to be asked to recuse themselves in cases involving race (Reference Means and UnahMeans & Unah, 2020). In the courtroom and in interactions with other attorneys, including their own co-workers, female attorneys report demeaning behavior, such as inappropriate comments about their appearance (Reference Collins, Dumas and MoyerCollins et al., 2017, Reference Collins, Dumas and Moyer2018). These experiences illustrate and reinforce perceptions that women and people of color do not belong in prestigious legal occupations, like that of attorney or judge.

In addition to occupational stereotyping that links white maleness with being a judge, women and people of color must also contend with negative stereotypes related to their gender and race (and the intersection of the two). As noted above, negative stereotypes about the competence and personality traits of particular groups can create anxiety for individuals in those groups. This anxiety is heightened with awareness that their behavior could be judged as “stereotypical” by outside groups. This, in turn, impacts the performance of achievement-oriented tasks (Reference Davis and SilverDavis & Silver, 2003). For example, Reference EpnerEpner (2006, p. 10) quotes a non-white, female attorney who described the demeaning comments she received because of her gender, race, and age: “When I first start practicing, it would make me incensed to the point where I would lose my concentration and focus and I would not be as good of an advocate as I would have been if it weren't an issue.” A black female attorney observed, “White associates are not expected to be perfect. Black associates. .. have one chance and if you mess up that chance, look out. There is no room for error” (Reference EpnerEpner, 2006, p. 25). Among attorneys, one study concluded that women of color “felt a need to continually establish their competence to professors, peers, and judges” (Reference EpnerEpner, 2006, p. 2).

Although negative stereotypes may ultimately fuel anxiety and result in under-performance by women and non-white lawyers, these negative stereotypes can also produce positive responses (Reference Hoyt and BlascovichHoyt & Blascovich, 2007). Described by Reference BrehmBrehm (1966) as psychological reactance theory, this perspective suggests that individuals respond to stereotype threat by engaging in behavior that is contrary to a negative stereotype about one's race or gender. For members of racial and ethnic minority groups, there is an additional motivation to counter negative expectations: linked fate. In his observations about race, and the legal profession, Wilkins noted that the “actions of individual blacks will affect the fate of the black community as a whole, and that ties the opportunities available to any individual black to the progress of the group” (Reference WilkinsWilkins, 1994, p. 1041).

Empirical support for reactance theory can also be found in research on legislative bodies where women's lower self-assessments of ability translate into more productive legislative behavior when they are elected to office. According to Reference Lazarus and SteigerwaltLazarus and Steigerwalt (2018), women in Congress overestimate their electoral vulnerability in part because of higher levels of self-doubt. As a result, they work harder than their male peers in a variety of ways, including increasing communication with constituents, garnering more federal spending for the district, and more effectively representing constituent preferences. Reference Thomsen and SandersThomsen and Sanders (2020) find similar results among state legislators. A recent study of judicial elections found that female lawyers who wish to become state supreme court justices are risk averse and underestimate their abilities; female prospective candidates wait until a seat is open before pursuing an office and delay running until they are certain they will have the “best” qualifications (Reference NguyenNguyen, 2019; but see Reference WilliamsWilliams, 2008). This literature suggests that female judges may be inclined to underestimate their abilities and believe they must overprepare in achievement-related tasks like writing majority opinions. For example, a female attorney, specifically commenting on gender stereotypes in the legal profession, said, “I show up extremely prepared, professional, articulate, knowledgeable about my case, friendly and authentic. I've found that it works in gaining the opposing side's respect and it helps educate them so they don't disrespect other women lawyers or me again” (Reference WeissWeiss, 2018).

The pressure to overwork may also be fueled by contextual cues that signal to “outsiders” that they will be held to a higher standard of performance (e.g., Reference HengelHengel, 2017). Reference Lazarus and SteigerwaltLazarus and Steigerwalt (2018) note that the overworking behavior of women in Congress reflects the reality that female incumbents are more likely than their male counterparts to draw a challenger. In fact, women in Congress have to “introduce twice as much legislation as men to see the probability of challenger emergence decrease to a level that is indistinguishable from that of men” (Reference Atkinson and WindettAtkinson & Windett, 2019, p. 769). Reference Gill, Lazos and WatersGill et al. (2011) also note that external assessments of judicial performance are biased against minority and female judges.

Pressures toward overpreparation may also be shaped by the desire to persuade key audiences. As Reference Lazarus and SteigerwaltLazarus and Steigerwalt (2018) find, female legislators react to their gendered vulnerability by working harder to appeal to constituents, an audience that is central to legislators' ability to stay in office. Judges are also mindful of their audiences (Reference BaumBaum, 2006) and seek to persuade those audiences that their decisions are legally sound (Reference KleinKlein, 2002). For federal appellate judges, key audiences for a majority opinion author include their colleagues on the panel and circuit, the litigants, and the Supreme Court (Reference MoyerMoyer, 2008). Consistent with Reference ChaikenChaiken (1980) and Reference O'KeefeO'Keefe (1998), women and non-white judges may feel that they must work harder than their white male colleagues to persuade these audiences that their decisions are legally correct. Along these lines, Judge Patricia Wald of the DC Circuit observed that opinion authors write “to persuade their colleagues and the public they are moving law in the right direction” (Reference WaldWald, 1995, p. 1372).

Because white men have authored the vast majority of court opinions in the American legal system, we argue that this identity is implicitly linked with judicial opinion writing (see also Reference KirkpatrickKirkpatrick, 2020). Ifill (2011, p. 441) similarly argues that “society builds the values embodied in our laws, for example, upon a set of accepted and legitimized narratives … In general, the dominant community's narratives form the basis of our approach to legal doctrine, theory, and practice. When white men were the only group permitted to offer and legitimate narratives in the legal process, master narratives could function virtually undisturbed as unassailable truths.”

Taken together, these mechanisms all point in a similar direction: women and people of color will feel additional pressures to succeed in an environment that has long been the domain of white men. Both individual assessments of ability related to stereotypes and socializing experiences in the law will drive these judges, while well-credentialed, to feel that they must work harder than their white male peers to demonstrate their competence. As a result, we argue that “outsider” judges will strive for perfection in their work, including in the most visible and enduring aspect of their position as an appellate judge: authoring an opinion for the court.

On the federal appeals courts, rotating three-judge panels must quickly come to a collective resolution of a case. The opinion author plays a prominent role in shaping the output of the panel's deliberations (Reference HinkleHinkle, 2017). With help from clerks, the majority opinion assignee writes an initial opinion draft, which is then circulated to the other panelists. Commonly, the other judges on the panel will suggest changes to the draft, which are usually incorporated into the opinion (Reference Bowie, Songer and SzmerBowie et al., 2014). Majority opinion authors have strong incentives to accommodate their colleagues' preferences (Reference HinkleHinkle, 2017) because doing so can help insulate the opinion from en banc and Supreme Court review (Reference Beim, Hirsch and KastellecBeim et al., 2016).Footnote ⁵ Norms of consensus also motivate the opinion writer to incorporate her colleagues' perspectives (Reference Bowie, Songer and SzmerBowie et al., 2014, p. 97).

Extant research suggests that “outsider” opinion authors may more thoroughly explain the reasons behind their rulings than do white male opinion authors. Reference BoydAmong federal district court judges, Boyd (2015) finds that female and non-white judges tend to write lengthier opinions, perhaps to better justify the outcome to litigants. Judges who are racial minorities are more thorough in drawing upon due process jurisprudence when adjudicating sentencing guidelines (Reference Sisk, Heise and MorrissSisk et al., 1998) and have been found to write longer opinions in other appellate contexts (Reference Budziak, Hitt and LempertBudziak et al., 2019). This leads us to our first hypothesis:

In addition, over-preparation may also manifest itself through a greater identification of, reliance on, and incorporation of legal authorities in the majority opinion. While citing legal authorities in a majority opinion is linked to self-presentation motivations among all judges (Reference BaumBaum, 2006; Reference Choi and GulatiChoi & Gulati, 2007; Reference HumeHume, 2009), we argue that female judges and judges of color will be especially driven to justify their rulings by incorporating more legal authorities compared to their white male colleagues. This leads to the following two hypotheses:

DATA AND MEASURES

Using an original sample of signed, published opinions from the US Courts of Appeals (2008–2016), we merged case data with information on judicial backgrounds acquired from the Federal Judicial Center.Footnote ⁶ The analysis contains three dependent variables, each of which captures an aspect of opinion thoroughness: page length (Opinion Length), the total number of precedents cited in the opinion (Total Authorities), and the number of cited cases discussed in depth (Deep Cites).Footnote ⁷ Because the variables are counts, and because concrete evidence shows overdispersion,Footnote ⁸ we use negative binomial regression to estimate the model parameters. Additionally, given potentially correlated errors within circuits and over time, as well as across opinions authored by the same judge, we include unconditional circuit and year fixed effects along with White-Huber robust standard errors clustered by opinion author.Footnote ⁹

Our primary variables of interest reflect the race and gender of the judge who authored the opinion, Female Judge (1 if female, 0 if male) and Non-white Judge (1 if non-white, 0 if white).Footnote ¹⁰ We expect that the coefficients for these variables will be positive in all models.

We also include controls for characteristics of the opinion author that could influence the opinion length and sourcing: Law Professor (coded 1 if the judge was previously a tenure-track law school professor and 0 otherwise), Elite Law School (coded 1 if the judge attended an elite law school and 0 if not),Footnote ¹¹ and Judge Tenure (the number of years since the judge was commissioned to serve as a federal appeals court judge, logged due to skewness and potential diminishing marginal returns). Because tenure-track law school professors are socialized to write lengthy, well-cited texts due to research expectations within the field, we expect to find a positive Law Professor coefficient. Elite Law School should be positive to the extent that it provides a rough measure of the ability of the judge. Conversely, we posit a negative coefficient for Judge Tenure, as early-career judges may feel the need to write longer opinions with arguments justified by more sources.

Judges should generally pen longer, more thoroughly reasoned opinions in salient cases (Reference Bowie, Songer and SzmerBowie et al., 2014). Here, we incorporate three indirect measures of case salience: whether the opinion involved constitutional interpretation (Constitutional Issue), whether an amicus was filed (Amicus Curiae), and whether a prior published decision was associated with the case (Prior Publication). We follow the coding protocols developed by Reference SongerSonger (1997); information on coding procedures related to measures for these case characteristics can be found in the codebook that accompanies the Multi-User Database. We expect the estimated coefficients for all three to be positive in all models.

Closely related to these indicators of case salience is a dummy variable, Affirmed, coded 1 if the court affirmed the lower court or agency decision and 0 if the court completely or partially reversed the prior decision. Courts are more likely to affirm decisions of the lower court or agency when the issues are “easy.” In these routine cases, judges are more likely to write shorter opinions with fewer quoted citations. Consistent with this causal mechanism, one prior study concluded that Courts of Appeals judges tended to write longer opinions when the panel reversed the lower-court decision (Reference Epstein, Landes and PosnerEpstein et al., 2013), while another found that federal appellate judges typically wrote shorter opinions when the panel affirmed the decision below (Reference Haire and MoyerHaire & Moyer, 2015). Based on the theory and prior evidence, we expect the coefficients will be negative.

Opinions also may be longer and devote greater attention to precedents as a result of the number of lawyers in an appeal. Evidence suggests that Courts of Appeals judges are less likely to suppress issues raised in briefs by larger litigation teams; moreover, judges tend to side with litigants represented by more attorneys (Reference Haire and MoyerHaire & Moyer, 2008). Together, this suggests that larger litigation teams will present more persuasive and informative briefs—which can facilitate the work of the opinion author. Because the variable is positively skewed, and existing work finds diminishing marginal returns as the number of lawyers increases (Reference Haire and MoyerHaire & Moyer, 2008), we calculate the natural log of the total number of attorneys who are listed as counsel representing all appellant(s) and respondent(s) in the case (Number of Attorneys). We expect the estimated coefficient will be positive in all models.

We also include controls that account for the role of the majority opinion author's policy preferences and potential ideological differences among panel members. Outcome Congruence is coded 1 if the judges' presumed ideology (based on the party of the appointing president) is congruent with the ideological direction of the court's decision (i.e., a Democrat [Republican] supporting a liberal [conservative] outcome); otherwise, it is coded 0. We expect that judges will be inclined to write longer, more thoroughly cited opinions when the outcome is consistent with their policy predisposition and shorter, less thoroughly cited opinions when their policy views are inconsistent with the outcome. Split Panel is coded 1 if the panel included judges who were appointed by presidents from different political parties. Circuit court norms strongly encourage the judges to incorporate colleagues' suggestions when writing the majority opinion (Reference Bowie, Songer and SzmerBowie et al., 2014). Judges who are like-minded are more likely to converge quickly on the reasoning, resulting in an opinion that is shorter and less thorough in its reasoning. Based on these expectations, the coefficient for Split Panel should be positive.

We control for those cases in which conflict on the panel resulted in publishing a dissenting (Dissent) or concurring opinion (Concurrence). Separate opinions are more likely to occur in salient and legally ambiguous cases (Reference Hettinger, Lindquist and MartinekHettinger et al., 2006). In addition, majority opinion authors should respond to the dissenter's counterarguments. Supreme Court justices, Courts of Appeals judges, and judges on state courts of last resort tend to write longer majority opinions in cases with separate opinions (Reference Bowie, Songer and SzmerBowie et al., 2014; Reference Epstein, Landes and PosnerEpstein et al., 2013; Reference Leonard and RossLeonard & Ross, 2016). Therefore, the coefficients should be positive, indicating a tendency to write longer opinions with more directly quoted, and thoroughly cited legal precedents.

Finally, we estimate circuit caseload (Caseload) using the Judicial Caseload Profile measure of the US Courts' Federal Court Management Statistics. This profile measures the number of terminated cases per active judge during the year the case was decided.Footnote ¹² Reference Epstein, Landes and PosnerEpstein et al. (2013) posit that judicial behavior balances costs and benefits. When judges have higher workloads, the opportunity costs of writing longer and more diligently sourced opinions increases. Consistent with this expectation, researchers found that higher circuit workload (Reference Epstein, Landes and PosnerEpstein et al., 2013) and opinion-writer workload (Reference Bowie, Songer and SzmerBowie et al., 2014) tended to reduce the length of US Courts of Appeals majority opinions. With that in mind, the coefficient should be negative.

ANALYSIS

Based on multiple mechanisms, including stereotyping, impostorism, and socialization forces, we expect women and people of color, as “outsiders” on the prestigious federal appellate bench, to experience higher pressures to be successful in the status quo environment dominated by “insider” white male judges. Since our hypotheses generally focus on the effects of outsider status, we begin by comparing white male judges to all other judges.

Figure 1 displays evidence from the negative binomial regression model that confirms our general expectations: minority and female judges write longer opinions with more sources, cites, and deep cites. The plots show expected counts of the dependent variables (number of pages, total citations, and deep citations) for judges who are white men (“insiders”) and those who are not white men (“outsiders”), averaged across observed combinations of the other covariates.Footnote ¹³ The estimates are circles, and the horizontal lines are confidence interval ranges. On average, white men write opinions that are roughly two-thirds of a page shorter, with almost three fewer cites, and one fewer deep cite. We also generated percent changes to get a better sense of the relative effects of “outsider” performance pressures on each of the dependent variables. We observed the largest gaps in citation patterns. Opinions written by white males, on average, contain 10.7% fewer citations to precedent and 16.6% fewer citations discussed with more than a paragraph of text. Conversely, their opinions are 5.26% shorter than those written by women and non-white judges.

Figure 1. Marginal effects of white male versus other judges, US Courts of Appeals cases, 2008–2016. Note: N = 2906. Confidence intervals shown are for two-tailed tests (p < 0.05). Full results for these models appear in Table A1

The following analyses unpack the previous findings with respect to each of our hypotheses. Hypothesis 1 predicts that women and non-white judges would author lengthier opinions than their white, male counterparts. We include a control for Total Authorities, which allows us to evaluate the effect of our variable of interest after taking into account the number of citations on opinion length.Footnote ¹⁴ Figure 2 displays the results of the negative binomial regression on the number of pages in the majority opinion using average marginal effectsFootnote ¹⁵ plots, with the independent variables on the Y axis.Footnote ¹⁶ As the figure shows, there is a positive and statistically significant effect for the Female Judge variable. On average, women write opinions that are approximately two-thirds of a page longer than their male peers. Put another way, women write opinions that are roughly 5.6% longer, on average. Although the relationship is in the predicted direction, non-white judges do not author significantly longer opinions than their white colleagues to a statistically significant degree.

Figure 2. Marginal effects of expected count of opinion length, US Courts of Appeals cases, 2008–2016. Note: N = 2906. Unconditional circuit and year fixed effects not shown. Confidence intervals shown are for two-tailed tests (p < 0.05). Full results shown in Table A2

Moving to control variables, the plot also shows that former law professors author longer opinions than judges who lack experience in legal academia. Case characteristics such as the presence of a constitutional issue, amicus curiae, or a dissenting opinion tend to increase the number of pages in the majority opinion, as expected.

Figure 3 shows the average marginal effects generated from the negative binomial regression on the total number of authorities cited in the majority opinion, testing Hypothesis 2. Consistent with our hypothesis, female and minority opinion writers include more citations in majority opinions than their white male colleagues. Opinions of the court written by women contain, on average, 6.7% more citations than those authored by male peers. The results for Non-white Judge indicate that non-white judges write opinions that have 11.2% more citations than those written by white jurists, all else equal.

Figure 3. Marginal effects of expected count of the total number of authorities cited in majority opinions, US Courts of Appeals cases, 2008–2016. Note: N = 2906. Unconditional circuit and year fixed effects not shown. Confidence intervals shown are for two-tailed tests (p < 0.05). Full results shown in Table A2

Among the other covariates, Judge Tenure, Amicus Curiae, Number of Attorneys, Affirmed, Prior Publication, Dissent, and Constitutional Issue are statistically significant—and all in the expected direction.Footnote ¹⁷ The effects for cases that raise a constitutional issue are the most striking. On average, judges include more than eight additional citations in constitutional cases than in cases that merely interpret statutes or administrative regulations.

To evaluate our third hypothesis, Figure 4 shows the average marginal effects generated from the negative binomial regression on the number of cases discussed in depth (Deep Cites) in the majority opinion. Again, the marginal effects of Female Judge and Non-white Judge are positive and statistically significant. Consistent with expectations, on average, female judges extensively discuss about one-half more cited cases than do their male counterparts. When converted to percent changes, we see that, on average, women incorporate 11.3% more deep cites compared to male opinion authors. Also, as with the previous model, Non-white Judge is positive and significant; opinions authored by nonwhite judges discuss about one more cited authority in depth compared to opinions authored by white judges. Expressed using percent changes, appellate judges of color write majority opinions that contain, on average, 19.5% more deep citations than their white peers.

Figure 4. Marginal effects of expected count of the deep cites in majority opinions, US Courts of Appeals cases, 2008–2016. Note: N = 2906. Unconditional circuit and year fixed effects not shown. Confidence intervals shown are for two-tailed tests (p < 0.05). Full results shown in Table A2

The findings for most control variables support our expectations about the role of case-level factors in shaping the depth of discussion surrounding a cited precedent. Constitutional Issue, Amicus Curiae, Affirmed, Prior Publication, Number of Attorneys, and Concurrence are all statistically significant in the posited direction. In particular, judges seem to examine precedents more deeply when their majority opinion is challenged by a dissent. On average, the majority opinion includes 3.2 more deep cites when there is a dissent than in unanimously decided cases.

Of course, women of color in the legal profession experience multiple facets of discrimination that reflect the intersecting nature of their identities (Reference Collins, Dumas and MoyerCollins et al., 2017; Reference Collins and MoyerCollins & Moyer, 2008; Reference DeoDeo, 2019) and that could drive impostorism fears that affect opinion writing. We estimated a series of additional models that looked at the effect of race and gender together. In each model, the results failed to identify intersectional differences (see Supporting Information for full models).Footnote ¹⁸

DISCUSSION

As the ranks of the federal judiciary continue to include judges from a broader range of backgrounds, it is increasingly important to understand how social forces condition the ways that individuals assess their abilities and determine requirements for career success in male-dominated professions like politics and law. Even those who have reached the elite ranks of their profession are not immune to self-doubt and feelings of impostorism (Reference Blanch, Hall, Roter and FrankelBlanch et al., 2008; Reference Lazarus and SteigerwaltLazarus & Steigerwalt, 2018). While much of the extant work on gender and race in the judiciary focuses on the voting behavior of judges, another profitable avenue for inquiry focuses on the micro-foundations of racialized and gendered institutions, identifying ways in which actors and institutional rules or norms interact to shape expectations and how such interactions shape the distribution of power within an institution (Reference LowndesLowndes, 2020, p. 544).

Building from this body of work, our central argument posits that high-achieving women and people of color in the legal profession have adopted strategies for dealing with pressure and societal expectations and that they carry these strategies with them onto the bench. Consistent with our expectations, the evidence suggests that female and minority appellate judges are more likely to overprepare in achievement-related tasks like writing majority opinions, spending more effort than their male colleagues to justify to key audiences (litigants, other judges, and the Supreme Court) that their decision is legally correct. Majority opinions written by women and minorities include more citations to prior cases, and they are more likely to devote more attention to discussing specific precedents. Even after controlling for the tendency to include more citations, our study shows that women still write longer opinions than men do, consistent with findings from other, earlier studies of trial courts (Reference BoydBoyd, 2015). Moreover, these differences in opinion writing do not appear to vary based on the length of time a judge has served on the bench.Footnote ¹⁹

Our results have important implications for scholars of courts and other legal and political institutions. First, our findings suggest that standard measures of workload, like cases per judge, may obscure systematic differences in effort. To put this into context, the median number of written opinions per year for a judge in our sample is 44, consistent with what circuit judges themselves describe (Reference WaldWald, 1995). If women write about two-thirds of a page more in each opinion, this would translate to an estimated additional 26 pages a year for each woman. Over a female judge's career, this would total hundreds of additional pages of writing from her majority opinions alone. Of course, this is likely a conservative estimate, given studies that find that female judges are disproportionately assigned to write more opinions than are men on other courts with discretionary opinion-assignment practices (Reference Christensen, Szmer and StritchChristensen et al., 2012). In addition, the cumulative effect of additional work, particularly over time, could hasten work-related burnout and stress, which some studies have found to affect more female judges than male judges (Reference Lustig, Delucchi, Tennakoon, Kaul, Marks and SlavinLustig et al., 2008; Reference Flores, Miller, Chamberlain, Richardson and BornsteinFlores et al., 2008, Flores et al., 2009; but see Reference Miller, Reichert, Bornstein and ShulmanMiller et al., 2018). Future research should continue to explore whether and in what ways this gendered pattern of additional effort extends to other high-achieving female political elites, such as members of Congress and state legislators (Reference Dolan and FordDolan & Ford, 1995), as well how it may extend to the experiences of racial minorities in political institutions.

This study also contributes to the burgeoning judicial-politics literature on opinion content and the use of precedent in the federal judiciary (Reference Black, Owens, Wedeking and WohlfarthBlack et al., 2016; Reference Corley, Collins and CalvinCorley et al., 2011; Reference Nelson and HinkleNelson & Hinkle, 2018). Opinion content is an important focus for inquiry because it is “the opinions themselves, not who won or lost … which provide the constraining directions to the public and private decision makers who determine the 99 percent of conduct that never reaches the courts” (Reference ShapiroShapiro, 1968, p. 39). The US Courts of Appeals, in particular, are crucial institutions to examine because they establish the bulk of federal precedent due to the relatively small docket of the Supreme Court, which effectively renders most federal appellate decisions the final word (Reference HinkleHinkle, 2015).

There are, of course, limitations to our study that point to areas where more research is needed. Law clerks undoubtedly play a central role in the work of federal appellate judges; a 2013 survey of federal appellate judges found that more than 90 percent used their law clerks to do legal research and help them draft opinions (Reference Peppers, Giles and Tainer-ParkinsPeppers et al., 2014). Unfortunately, law-clerk information for each judge's chambers is not publicly available, so we are unable to control for the demographic makeup of each judge's chambers and cannot systematically evaluate how law clerks affected the content of particular opinions.Footnote ²⁰ It is quite likely, however, that men are overrepresented as law clerks for the judges in our sample; national data from the law school class of 2016 reveals that more than half of all federal clerkships went to men, a number that has stayed consistent since 2006 (NALP, 2017; see also Reference Szmer, Kaheny and ChristensenSzmer et al., 2014).Footnote ²¹ Future surveys will be an important vehicle to investigate whether the addition of more female and non-white judges has had the effect of socializing future judges into different norms of opinion writing than those more commonly adopted by white male judges. More generally, while we note that gender and race-based differences are suggestive of labor disparities—in terms of both observable work product and the psychological toll that may be associated with dynamics such as impostorism and stereotype threat—it is also possible that the contributions of “outsider” judges could raise expectations for all judges, in the long run, about what a majority opinion should be.

With respect to intersectional identities, studies of the legal profession have identified particular ways in which women of color (Reference Collins, Dumas and MoyerCollins et al., 2017; Reference DeoDeo, 2019; Reference EpnerEpner, 2006; Reference Haire and MoyerHaire & Moyer, 2015) experience distinct kinds of discrimination compared to white women and minority men. We fail to find evidence of different opinion-writing approaches by women of color, though our analysis is hampered by the relatively small number of majority opinions authored by this group in our sample. For this reason, quantitative approaches may not always be ideal for untangling the ways in which socialization and stereotypes play out in opinion writing for particular intersectional cohorts. Future work should use interviews and archival materials to explore how these judges negotiate the pressures and expectations associated with the prestigious posts they occupy. Moreover, researchers might profitably explore deeper narratives that might, for example, identify whether female judges and judges of color feel impostorism and stereotype threat pressures differently and under what circumstances.

Taken together, our findings suggest that norms about a central activity of judging, crafting opinions, are gendered and racialized in ways that create higher workloads for “outsider” judges. As scholars explore other facets of diversifying legal institutions, it will be important to assess whether other norms operate in the same fashion and, if so, what changes in organizational practice can create more equitable outcomes.

ACKNOWLEDGMENTS

The authors gratefully acknowledge that data for this project were collected with support from the National Science Foundation (NSF SES #1655159, 1654614, 1654559, 1654697). Any opinions, findings, and conclusions or recommendations expressed are those of the authors and do not necessarily reflect the views of the National Science Foundation. The authors also would like to thank Barry Edwards, Ethan Boldt, Alyson Hendricks-Benton, Madison Eckels, and Samuel LaVergne. Please direct all correspondence to Laura Moyer, 205 Ford Hall, Louisville, KY 40292, [email protected].