Working Paper: NBER ID: w25013
Authors: Mark Hoekstra; Brittany Street
Abstract: This paper examines the extent to which criminal conviction rates are affected by the similarity in gender of the defendant and jury. To identify effects, we exploit random variation in both the assignment to jury pools and the ordering of potential jurors. We do so using detailed administrative data on the juror selection process and trial proceedings for two large counties in Florida. Results indicate that own-gender juries result in significantly lower conviction rates on drug charges, though we find no evidence of effects for other charges. Estimates indicate that a one standard deviation increase in expected own-gender jurors (~10 percentage points) results in an 18 percentage point reduction in conviction rates on drug charges, which is highly significant even after adjusting for multiple comparisons. This results in a 13 percentage point decline in the likelihood of being sentenced to at least some jail time. These findings highlight how drawing an opposite-gender jury can impose significant costs on defendants, and demonstrate that own-gender bias can occur even in settings where the importance of being impartial is actively pressed on participants.
Keywords: gender bias; jury composition; conviction rates; sentencing outcomes
JEL Codes: J16; K41; K42
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| increased representation of owngender jurors (J79) | lower conviction rates (K14) |
| change in jury gender composition (J79) | decline in likelihood of being sentenced to at least some jail time (K40) |