Purpose. Previous experiments have demonstrated asymmetrical scepticism in investigators' judgments of criminal evidence – evidence inconsistent (vs. consistent) with the dominant hypothesis about a case is judged as less reliable. In addition, some types of evidence (e.g., witness testimony) are more susceptible to asymmetrical scepticism than others (e.g., DNA evidence), indicating varying degrees of elasticity. This article proposes that inconsistent evidence arouses cognitive dissonance, and that the dissonance can be reduced through either asymmetrical scepticism (for high-elasticity evidence) or belief change (for low-elasticity evidence). The hypotheses are tested in two experiments.

Methods. In both experiments, law students made a preliminary judgment about the guilt of a suspect in a homicide case, and subsequently received a piece of DNA or witness evidence which was either consistent or inconsistent with the preliminary judgment. The extent to which participants changed their guilt judgments, judged the additional evidence as reliable, and felt dissonance served as the main dependent variables.

Results. Inconsistent (vs. consistent) evidence did arouse stronger dissonance, but only for witness (and not DNA) evidence. Experienced dissonance (Experiment 1) and dissonance reduction (Experiment 2) accounted for the effect of the evidence on changes in guilt judgments, but not for the effect on reliability judgments. The greatest dissonance reduction was observed among participants who received inconsistent witness evidence but did not change their guilt judgments accordingly.

Conclusions. It appears that dissonance plays a significant, although complex, role in investigative judgments of guilt and reliability. Alternative dissonance-reducing mechanisms that can account for the findings and practical implications are discussed.