The CSI Effect and the impact of DNA evidence on mock jurors and jury deliberations

ABSTRACT

The present research examined the CSI Effect and the impact of DNA evidence on mock jurors and jury deliberations using a 3 (Crime Drama Viewing: low, moderate, high) × 3 (Evidence: DNA innocent, DNA guilty, no DNA control) design. A sample of 178 jury-eligible college students read a case of breaking and entering. Pre-deliberation, some support for a CSI Effect was found with high viewers’ extent of guilt ratings significantly lower than moderate and low viewers’ in the no DNA control and the DNA innocent conditions. This effect was not present for verdicts. Contrary to a CSI Effect, crime drama viewing was not related to guilt judgments with incriminating DNA evidence. A content analysis of comments made during deliberations found little support for the CSI Effect entering the jury room. Specifically, CSI Effect predictions were not supported when examining the discussion of DNA evidence, expressing DNA opinions, or mentioning missing evidence. Overall, the limited CSI Effect found for individuals was attenuated during deliberation. The alarm raised over a possible CSI Effect influencing jury decision making may be unwarranted.

KEYWORDS:

• Juror decision making
• CSI Effect
• DNA evidence
• jury deliberation

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 The case included two additional charges: assault with a dangerous weapon and intent to murder. Only the breaking and entering charge was proved by forensic evidence and, therefore, was the only charge analyzed in this study.

2 For the pre-deliberation outcomes, preliminary analyses were performed to determine whether it was necessary to model the nesting of participants within juries (n = 15), with two-level mixed-effects regression models used to estimate the proportion of total outcome variance attributable to participants versus juries. Given the modest number of juries, the models were implemented using Bayesian MCMC methods (Ozechowski, 2014; chain length = 100,000) in MLwiN software (Rasbash, Browne, Healy, Cameron, & Charlton, 2015). In fully unconditional models, 13% and 12% of the variance in extent of guilt and verdict, respectively, was attributable to nesting within juries. However, after controlling for the evidence manipulation, this decreased to less than 1% for each outcome. As such, for the pre-deliberation outcomes, it was not necessary to include a random effect for nesting within juries.