Citation

Boyd, C. L., Epstein, L., & Martin, A. D. (2010). Untangling the causal effects of sex on judging. American Journal of Political Science, 54(2), 389–411. https://facultyweb.cortland.edu/ocallaghan/pdf/boydepsteinmartin.pdf

Research Question

To what extent, and in which areas of law, does a judge’s sex causally affect appellate case outcomes directly (individual effects) and indirectly through mixed-gender panel composition (panel effects)?

Key Takeaways

Gender rarely shifts outcomes on federal appellate panels, but in Title VII sex discrimination cases it materially changes both individual votes and panel dynamics; In sex discrimination appeals, having a female judge on the panel measurably increases the odds that male colleagues will side with the discrimination claimant, reshaping the effective risk profile; Once ideology and case mix are properly balanced, most purported gender effects disappear, showing how misleading naive regressions can be; For plaintiffs in employment sex discrimination cases, the presence of a female judge on an appellate panel meaningfully increases expected success and may justify pursuing or pressing appeals; Defendants in sex discrimination litigation must adjust exposure assessments and settlement strategies when facing mixed-gender appellate panels; The results support a domain-specific informational account of gender effects rather than broad “different voice” or purely organizational theories of judging.

Dataset Description

The study uses observational data on U.S. federal court of appeals decisions, drawn primarily from datasets originally compiled by Sunstein et al. and subsequently supplemented by the authors. It includes tens of thousands of judge-vote observations across 13 doctrinal areas (e.g., abortion, ADA, affirmative action, campaign finance, capital punishment, contract clause, EPA, federalism, piercing the corporate veil, Title VII sex discrimination, sexual harassment, takings, Title VII race) over varying periods from 1976 to 2002, depending on the doctrinal area. Votes are coded at the judge level (individual effects) and at the panel level (panel effects), with information on panel composition, judge ideology (Judicial Common Space scores), lower court direction, and case characteristics. The authors apply semiparametric matching (primarily nearest-neighbor propensity score matching) and subsequent weighted logistic regression to create balanced treatment and control groups of male and female judges (and all-male vs mixed-sex panels) within each legal domain.

Methodology

statistical/quantitative

Key Findings

Using a formal causal-inference framework and semiparametric matching, the authors find that a judge’s sex has no robust causal effect on outcomes in 12 of 13 legal domains once ideology and case characteristics are properly balanced. The sole, consistent exception is Title VII sex discrimination litigation: female judges are about 10 percentage points more likely than similarly situated male judges to vote for the party alleging discrimination, and the presence of a female judge on a three-judge panel increases male colleagues’ propensity to vote for the plaintiff by roughly 12–14 percentage points. For sex discrimination cases, the probability of a pro-plaintiff vote by a male judge on an all-male panel never exceeds about 0.20, whereas on a mixed-sex panel it never falls below about 0.20, with an increase of roughly 85 percent in the likelihood of a pro-plaintiff vote for ideologically “average” male judges. These effects are sizable in practical terms and align most closely with an informational account: women’s distinct professional experiences in sex discrimination contexts provide information that both shapes their own votes and shifts their male colleagues’ behavior. The findings run counter to broad “different voice” theories that predict gender effects across many areas, and to purely organizational accounts that predict none, while also showing that conventional regression on unmatched data can misstate or even miss such effects.

Summary

This article asks whether, and in what ways, the sex of U.S. federal appellate judges causally affects case outcomes. The authors distinguish two mechanisms: individual effects (whether male and female judges, holding everything else constant, vote differently) and panel effects (whether male judges vote differently when serving with female colleagues). They argue that much prior work conflates these mechanisms and relies on standard regression analyses that do not adequately address non-random assignment of cases to judges, particularly across ideologically charged issue areas.

To improve on this, the study applies a modern causal-inference approach using semiparametric matching. Within 13 doctrinal domains, from abortion and affirmative action to environmental law, federalism, capital punishment, and employment discrimination, the authors use propensity score matching to construct balanced samples of male and female judges, and of all-male versus mixed-sex panels, that are comparable on ideology (Judicial Common Space scores), lower court outcomes, temporal factors, and other observable case characteristics. They then estimate treatment effects (judge sex and panel composition) using weighted logistic regression on these matched datasets, enabling a more credible assessment of whether sex itself has a causal impact.

The main empirical finding is that judge sex does not exert a robust, independent causal effect in 12 of the 13 legal domains examined. Once the authors balance on ideology and case mix, men and women on the federal courts of appeals typically cast similar votes, and the mere presence of a woman on the panel does not systematically alter male voting behavior. This result undermines broad versions of “different voice” theories that predict pervasive gender differences in judging and shows how conventional regressions on unmatched data can generate misleading inferences about gender effects.

Title VII sex discrimination cases are the striking exception. In these cases, female judges are about 10 percentage points more likely than observationally similar male judges to vote for the party alleging discrimination. In addition, there is a strong panel effect: male judges serving on mixed-sex panels are 12–14 percentage points more likely to side with the discrimination claimant than male judges on all-male panels. For an ideologically average male judge, the move from an all-male to a mixed-sex panel translates into roughly an 85 percent increase in the predicted probability of a pro-plaintiff vote. These effects are substantial in practice and withstand multiple robustness checks.

The authors interpret this pattern as evidence for an informational account. Female judges, given their likely exposure to and understanding of workplace sex discrimination, bring specialized information and perspectives that shape not only their own votes but also the deliberative environment. Their presence appears to shift male colleagues’ priors about what constitutes discrimination and the credibility of plaintiffs’ claims. The absence of similar effects in other doctrinal areas suggests that gender operates in a domain-specific way, salient where lived experience and subject-matter expertise differ sharply by sex. For practitioners, the results imply that panel composition is especially consequential in Title VII sex discrimination appeals, where having at least one female judge can materially change outcome probabilities and thus the calculus of appeal, settlement, and risk assessment, while in most other areas, gender per se is a weak predictor once ideology and case characteristics are properly accounted for.

How the Study Advances Empirical Understanding of Legal Outcomes

The study finds that legal outcomes on federal appellate panels exhibit a measurable structure where the causal effect of gender is primarily localized to specific decision environments rather than manifesting as a broad institutional regularity. The results indicate that while sex-based differences in voting behavior are statistically absent across twelve of thirteen tested legal domains, Title VII sex discrimination litigation demonstrates a repeatable pattern of increased pro-plaintiff outcomes associated with female presence on a panel. The analysis shows that this domain-specific shift in outcome regularities arises from both individual voting choices and a panel effect that alters colleagues’ behavior, establishing an empirical case-based analysis of decision contexts as a qualified yet necessary component of strategic litigation practice at a high level.