Bias in Legal Judgments: Evidence from a Survey Experiment among Law Students
About this Session
Time
Thu. 16.04. 15:55
Room
Lobby
Speaker
by Lennart Goldemann and Maike Schlosser
Democracies typically guarantee citizens the right to a fair trial (e.g. EU Charter of Fundamental Rights, Article 47). A fair trial requires, among other things, judicial impartiality: judges and juries must evaluate cases based on evidence and the governing law rather than on socio-demographic characteristics of the parties involved. Yet, existing research indicates that this ideal is not always realized in practice (e.g. McConnell et al. 2024; Choi et al. 2022). Differential treatment in courts is particularly striking given that judges and legal professionals undergo extensive training of objective, evidence-based reasoning. Therefore, this paper addresses critical questions: Is the bias observed in the legal system already present in law school? Do law students exhibit bias from the outset, and how does it develop over the course of law school? We study judicial decision-making of roughly 3.000 law students from 19 different German universities. We design a survey experiment that mirrors legal decision-making in law school and develop a set of five criminal-law case vignettes, each mapping to a different offense. The treatment variation consists of randomizing the name of the main subject to manipulate perceived ethnicity or gender while holding all other details of the cases constant for each case separately. Each respondent evaluates all five cases. Students evaluate cases on a four-point scale ranging from a lenient to a harsh evaluation. In our main specification, we pool together the three ethnicity cases and the two gender cases separately. We estimate linear probability models with the probability of giving a harsh evaluation as the dependent variable. Law students are, on average, 3.4 percentage points more likely to pass harsh judgements when evaluating subjects with an ethnic minority name compared to subjects with German-sounding names. By contrast, female-named defendants are roughly 9 percentage points less likely to receive a harsh evaluation compared to their male counterparts. Suprisingly, we find that the documented treatment effects persist over the course of law school. First year students display the same bias as students in higher years. Our contribution is threefold. First, we shift focus to law students and document that bias exists and persists already during legal training, offering insights for early interventions. Second, we introduce vignette-based experiments to the law literature, combining clean identification with realistic legal settings. Third, we provide novel evidence from the German context, where court-level data are scarce and direct evidence on judicial bias is largely absent.
Democracies typically guarantee citizens the right to a fair trial (e.g. EU Charter of Fundamental Rights, Article 47). A fair trial requires, among other things, judicial impartiality: judges and juries must evaluate cases based on evidence and the governing law rather than on socio-demographic characteristics of the parties involved. Yet, existing research indicates that this ideal is not always realized in practice (e.g. McConnell et al. 2024; Choi et al. 2022). Differential treatment in courts is particularly striking given that judges and legal professionals undergo extensive training of objective, evidence-based reasoning. Therefore, this paper addresses critical questions: Is the bias observed in the legal system already present in law school? Do law students exhibit bias from the outset, and how does it develop over the course of law school? We study judicial decision-making of roughly 3.000 law students from 19 different German universities. We design a survey experiment that mirrors legal decision-making in law school and develop a set of five criminal-law case vignettes, each mapping to a different offense. The treatment variation consists of randomizing the name of the main subject to manipulate perceived ethnicity or gender while holding all other details of the cases constant for each case separately. Each respondent evaluates all five cases. Students evaluate cases on a four-point scale ranging from a lenient to a harsh evaluation. In our main specification, we pool together the three ethnicity cases and the two gender cases separately. We estimate linear probability models with the probability of giving a harsh evaluation as the dependent variable. Law students are, on average, 3.4 percentage points more likely to pass harsh judgements when evaluating subjects with an ethnic minority name compared to subjects with German-sounding names. By contrast, female-named defendants are roughly 9 percentage points less likely to receive a harsh evaluation compared to their male counterparts. Suprisingly, we find that the documented treatment effects persist over the course of law school. First year students display the same bias as students in higher years. Our contribution is threefold. First, we shift focus to law students and document that bias exists and persists already during legal training, offering insights for early interventions. Second, we introduce vignette-based experiments to the law literature, combining clean identification with realistic legal settings. Third, we provide novel evidence from the German context, where court-level data are scarce and direct evidence on judicial bias is largely absent.