Witnessing-condition information differentially affects evaluations of high- and moderate-confidence eyewitness identifications

Recent research on the eyewitness confidence-accuracy relationship reveals that confidence can be highly diagnostic of accuracy when the identification evidence is collected using pristine procedures (Wixted & Wells, 2017) and in the absence of suspect bias (Smalarz, 2021). Some researchers have further argued that eyewitnesses who make high-confidence suspect identifications are highly likely to be accurate even if they experienced suboptimal witnessing conditions (Semmler et al., 2018). The current research examined evaluations of eyewitness identification evidence in cases involving suboptimal witnessing conditions. Students (Experiments 1 & 2) and community members (Experiment 3) read eight crime vignettes involving an eyewitness's identification. We manipulated information about poor witnessing conditions (present vs. absent), the eyewitness's confidence level (high vs. moderate), and the format of the confidence statement (verbal vs. numeric) and measured evaluations of eyewitness-identification accuracy. Across all three experiments, information about suboptimal witnessing conditions disproportionately reduced evaluators' belief of highly confident compared to moderately confident eyewitnesses. This differential-discrediting pattern occurred for both numeric and verbal confidence-statement formats. Expert testimony describing the imperviousness of high-confidence suspect-identification accuracy to suboptimal witnessing conditions reduced, but did not eliminate, the differential-discrediting effect. Given that crime eyewitnesses frequently experience suboptimal witnessing conditions (e.g., Behrman & Davey, 2001; Wright & McDaid, 1996), the current findings have widespread implications for the capacity of the legal system to correctly classify suspects as guilty or innocent based on eyewitness identification testimony.