Credibility in Workplace Discrimination Claims: Evidence from MCAD Hearing Decisions

• In Event: Sociology of Law Section Roundtables Session
  In Refereed Roundtable (60 minute): Sociology of Law Section Roundtables: Table 5 — Courts, Credibility, and Legal Decision-Making

Abstract

Although thousands of workplace discrimination complaints are filed each year, only a small portion reach the formal hearing stage. At this stage, complainants—often at a structural disadvantage—must confront both formal laws and their employer’s counter-claims. During the hearing process, their narratives are translated by investigators and hearing officers into the formal language of law and subsequently evaluated by hearing officers. While previous studies mainly focus on the barriers that prevent complainants from gaining credibility within the legal system, this study argues that certain narrative patterns are more acceptable and recognized, making them more credible. Through the triadic framework of symbolic, relational, and discursive power, this study analyzes officers’ narratives in workplace discrimination claims to examine how claim framing influences complainant credibility evaluations.
I examine Massachusetts Commission Against Discrimination (MCAD) hearing decisions using Large Language Models to classify claims and linear probability models to evaluate which types of claims are persuasive before the law. Comparing the varying probabilities that complainants are considered credible based on specific narrative patterns, this study demonstrates several positive narrative frameworks that are more likely to be successful within formal hearings. In particular, claims including hostile work environment issues, retaliation, and the combination of race discrimination and terms and conditions issues are treated by hearing officers as more credible than other claims.
Empirically, this study extends the use of Large Language Models to the sociology of law, showing how they can extract themes and disposition from administrative legal texts with accuracy and efficiency. The findings also identify which discrimination narratives are more likely to be legitimized and credited within the hearing process, offering perspectives for understanding patterns in anti-discrimination practices.

Author

• Chenxi Yuan, University of Massachusetts-Amherst