At the start of the 21st century, lower-court judges in Spain were granting about 80% of restraining orders filed in court by victims of intimate-partner violence (IPV), and only a few among the 1,000-plus judges who heard these cases granted orders significantly above or below that average. In this area of law, therefore, differences in decisions among judges (judicial inequality) were rare. In 15 years, however, judicial inequality has doubled.

The one single factor that has shaped change in the structure of the Spanish judiciary over the last two decades is the steady feminization of the bench. While women represented only a tiny fraction of the judiciary in 1980, they now account for over 70% of judges in Spain’s lower courts. The growth of judicial inequality might be explained because male and female judges judge differently, with the latter approving a higher percentage of restraining orders. However, official records show that, over this period, the rate of approval in Spain has decreased by 5 points.

This is the main puzzle that this article intends to address: whether, and in what ways, female and male lower-court judges decide cases differently, and how these differences interact with institutional factors to shape judicial inequality in Spain.

There is a large amount of empirical evidence showing that the legal facts of the case are just one among many other factors that influence judicial decisions (Dyevre, 2010). Like doctors, fire workers and pilots, judges can be seen as expert, professional decisionmakers whose decisions depend on certain incentives and constraints (Epstein et al., 2013). Incentives include genuine love for the work, job tenure, promotion, social prestige, or the advancement of group-based interests shaped by ethnicity, class, or gender. These identity-based factors are especially relevant when dealing with cases where these factors are salient, when judges tend to decide in favour of plaintiffs who match their own identity (Segal, 2000; Boyd et al., 2010; Boyd, 2016).

Thus, when deciding IPV cases, Spanish lower-court judges’ actions are shaped by two different factors. Firstly, the law in Spain defines intimate-partner violence (also called “gender-based violence”) as “violence directed against women because they are women”, which means that courts distinguish IPV from general domestic violence cases. Therefore, in IPV cases, the stakes are well defined on gender lines — plaintiffs are always women who plead their right to security and safety, and defendants are always men who plead their right to move freely — and thus, gender might work as an identity-based mechanism that influences judges’ decisions.

Secondly, on the costs side of the equation, judicial decisions are usually constrained by the uncertainty surrounding the problem at hand, and by the information costs of making good decisions. As experts, judges become better professional decision-makers by identifying patterns of events from past decisions that can be used as templates to make better and more efficient decisions in the future (Vallbé, 2015). Therefore, as they learn from experience, judges need less time to gather information. However, quality decisions require time, and judges have to make a large number of quick decisions, in part because judges typically must attend to several cases at once, each one requiring swift and parallel judgements. But more importantly, in cases where fundamental rights are at stake (victims’ safety and defendants’ freedom of movement), the law compels courts to expedite procedures to prevent unnecessary harm to plaintiffs’ rights. In that sense, when judges must attend to a large court caseload, they tend to resort to cognitive shortcuts or heuristics that help them make quicker decisions and avoid information costs (Gigerenzer and Engel, 2006; Vallbé et al. 2019). Furthermore, a large caseload also makes judges less creative in their decisions (Epstein et al. 2013) and, on average, produces worse judicial outcomes for plaintiffs (Engel and Weinshall, 2020). In this sense, while experience helps judges to become better decision-makers, excessive caseload seriously harms the quality of their decisions.

To test the effects of such factors on IPV cases, we have analysed lower-court decisions on restraining order petitions filed to on-duty courts by victims of intimate partner violence (IPV). Typically taken as a preliminary action within a criminal procedure against an alleged IPV abuser, a restraining order commands an (alleged or confirmed) abuser to stay at a certain distance from the victim. In Spain, restraining orders may be heard by ordinary or on-duty lower-court judges. If petitions are filed with court within office hours, they will be heard by judges who have the exclusive competence to hear IPV cases. If an IPV case is brought to court during off hours, a judge on duty will process the incoming case and will decide on the restraining order petition. By nature, petitions heard by on-duty courts refer to critical cases, and this explains why on-duty courts approve 10% more restraining orders than ordinary courts in Spain. Because judges in the same district are on duty according to a random rotation system, we focus on IPV cases heard by on-duty courts, because the assignment of IPV cases to on-duty judges is as good as random, which we need to estimate causal effects (Angrist and Pischke, 2014).

Our data set contains 44,335 decisions on all restraining orders from IPV cases decided by 1,063 Spanish pretrial on-duty courts for the years 2011 to 2018. The data set is the result of processing large amounts of already existing official, public data sources. For each decision, we have information about court location, court caseload, gender of judge, judge’s level of experience, and whether the restraining order was approved or not. Due to data protection issues, we do not have information about the facts of each case. This is potentially challenging for our analysis because certain facts of cases (e.g., the extent of physical violence used against the victim) may seriously influence judges’ decisions. However, the random rotation system to assign cases to on-duty judges means precisely that judges are assigned cases regardless of the facts, thus minimizing the risk of selection bias. 

Petitions for restraining orders were heard by 1,528 individual judges, of which 63 % were female and 37 % male. On average, each judge heard 13 petitions per year, although annual caseload varies significantly across courts. In addition, the average level of experience among our judges is 12 years on the bench, although variation is also common in this regard. Due to their later incorporation to the bench, female judges tend to be slightly less experienced and face a marginally lighter caseload than male judges.

The results plotted in figure 1 show that when restraining orders are requested from female judges, the probability of these being granted increases compared to male judges. Although the effect persists even when controlling for caseload and professional experience, it is very small.

This suggests that gender alone explains only part of the decisions taken, because its effect is greatly conditional to information costs and uncertainty. On the one hand, results show that judges’ caseloads are critical to decisions concerning IPV cases. Controlling for gender, the average probability of approving a restraining order for judges with a low caseload (5 IPV cases per year) is around 82%, while this probability decreases to 74% when their caseload is a little higher (15 cases) and to 69% at high caseload levels (50 cases per year). This suggests that, as information costs increase, judges tend to favour the defendants’ rights to move freely above the victims’ right to be protected, and thus IPV victims have lower chances of getting their petitions approved.

In contrast, as judges gain professional experience, the rate of approval for restraining orders can increase by up to 10 points. However, and more interestingly, figure 2 shows that the effect of experience is stronger among female judges than among male judges — the model predicts that the difference in approval rates between junior and senior male judges is of just 4 points, while the difference among female judges is up to 15 points.

Finally, this distinct effect of constraints on decisions among male and female judges also appears when we include the three main factors explored in one single model. Our objective here is to test whether the distinct behaviour of male and female judges across professional experience is also observed at different levels of caseload, and figure 3 shows that this is not the case. The first panel of the figure shows that when both male and female judges have low caseloads, the effect of professional experience is positive for both. However, the other two panels show that, as judges face higher caseloads, the effect of professional experience is markedly different for male and female judges, to the point of only being positive for female judges.

The empirical analysis of judicial behaviour in IPV cases in Spain yields four main findings. First, gender alone is not enough to account for major differences between male and female judges when deciding on IPV. Our second finding is that caseload is a major determinant of judicial behaviour, to the point of significantly decreasing the probability of judges approving restraining orders to protect victims. Thirdly, our results show that, as judges gain experience, they tend to approve restraining orders at higher rates. Finally, we have also observed that the effects of both caseload and professional experience are conditional on judges’ gender. Results show that professional experience has positive effects among female judges regardless of caseload, but that when male judges face higher caseloads, the positive effects of professional experience vanish. Therefore, when dealing with IPV cases, the gender of judges is critical for both male and female judges as it interacts with other factors.

The major sources of judicial inequality for IPV victims are both endogenous and exogenous. On the one hand, judges’ gender becomes relevant when dealing with cases where fundamental rights are at stake, and female judges are systematically more ready to protect victims than male judges. This effect gets stronger as judges become expert decision-makers through experience. On the other hand, an exogenous factor, such as court caseload, shapes the conditions under which judges must make their decisions, to the point of systematically reducing the chances of IPV victims being granted restraining orders. Since unequal court decisions on IPV cases have critical consequences on the everyday lives of millions of women in Spain, they may have an impact on the overall legitimacy and impartiality of the justice system.

Article based on the project directed by Joan-Josep Vallbé “The Political Economy of Gender and Inequality in the Spanish Judiciary”, which received funding from the Social Observatory of the ”la Caixa” Foundation under code LCF/PR/SR19/52540006, GENEQUAL.

ANGRIST, J. D. and J.-S. PISCHKE (2014): Mastering ‘Metrics: The Path from Cause to Effect, Princeton (NJ): Princeton University Press.

BOYD, CHRISTINA L. (2016): “Representation on the Courts? The Effects of Trial Jueces’ Sex and Race.” Political Research Quarterly, 69(4). 

BOYD, CHRISTINA, LEE EPSTEIN and ANDREW MARTIN (2010): “Untangling the Causal Effects of Sex on Judging.” American Journal of Political Science, 54.

DYEVRE, ARTHUR (2010): “Unifying the field of comparative judicial politics: towards a general theory of judicial behaviour.” European Political Science Review, 2(2).

ENGEL, CHRISTOPH and KEREN WEINSHALL (2020): “Manna from Heaven for Jueces: Jueces’ Reaction to a Quasi-Random Reduction in Caseload”, Journal of Empirical Legal Studies, 17(4).

EPSTEIN, LEE, WILLIAM M. LANDES and RICHARD A. Posner (2013): The Behavior of Federal Jueces: A Theoretical and Empirical Study of Rational Choice. Harvard University Press. 

GIGERENZER, G and E. ENGEL, eds. (2006): Heuristics and the Law. Cambridge, MA: The MIT Press.

SEGAL, JENNIFER A. (2000): “Representative Decision Making on the Federal Bench: Clinton’s District Court Appointees”, Political Research Quarterly, 53(1).

VALLBÉ, J-J. (2015): Frameworks for Modeling Cognition and Decisions in Institutional Environments. {A} Data-driven Approach. Nº 21 en “Law, Governance and Technology”, Springer.

VALLBÉ, J.-J., M. GONZÁLEZ-BEILFUSS and BARAK Kalir (2019): “Across the Sloping Meadow Floor: An empirical analysis of pre-removal detention of deportable non-citizens”, Law & Society Review, 53(3).