Gender Bias in the Judiciary

Awarded Scholars:
Elliott Ash, ETH Zürich
Daniel L. Chen, Toulouse School of Economics
Arianna Ornaghi, University of Warwick
Project Date:
Dec 2019
Award Amount:
Project Programs:
Behavioral Economics

Economist Elliott Ash and colleagues will examine how and the extent to which attitudes toward gender affect decision-making in the judicial system. They will address the measurement challenge by exploiting the large corpus of written text for appellate judges, arguing that text can provide important insights into human social psychology. The PIs will proxy judges’ attitudes toward gender by measuring the degree of language slant that is displayed by their writing. In order to construct their measure, the PIs will use word embeddings, a language modeling technique from natural language processing. It relies on word co-occurrence statistics to obtain a semantic vector space model that conveys meaningful relations between words. They will use the measure to explore two main research questions. First, they examine the extent to which gender lexical bias impacts judges’ decisions in both gender and non-gender related cases. If lexical slant did not correlate with gender preferences, or judges were able to correct for it when voting on cases, it should have no impact on their judicial decisions. The second research question is the extent to which lexical slant manifests itself in how judges respond to female judges. This will allow the PIs to understand to what extent gender discrimination exists in the policy-relevant setting of Appellate Courts. They will focus on dimensions that are relevant to a judge’s career: the extent to which female judges are assigned the writing of majority opinions, citations, and reversals of lower court decisions. The principal investigators will then extend the methodology to legislators in the U.S. Senate and House. They have access to the digitized Congressional Record, which consists of transcripts of the speeches given by members of Congress from 1870 through the present. Each speech is tagged to a congress member, for whom they have a range of metadata on personal characteristics. They will link these data to voting record scores compiled by advocacy groups and roll-call votes on gender-related issues such as women’s rights.


RSF: The Russell Sage Foundation Journal of the Social Sciences is a peer-reviewed, open-access journal of original empirical research articles by both established and emerging scholars.


The Russell Sage Foundation offers grants and positions in our Visiting Scholars program for research.


Join our mailing list for email updates.