Excerpted From: Jacqueline Katzman and Margaret Bull Kovera, Potential Causes of Racial Disparities in Wrongful Convictions Based on Mistaken Identifications: Own-race Bias and Differences in Evidence-based Suspicion, 47 Law and Human Behavior 23 (February 2023)(1 Footnote/References)(Full Document)


KatzmanKoveraThere are clear racial disparities in the number of wrongful convictions based on eyewitness misidentifications of Black versus White defendants. The National Registry of Exonerations has documented 825 of these cases in total, including 517 exonerations of misidentified Black defendants as opposed to 162 cases of misidentified White defendants. Similarly, the Innocence Project(2021) database--which tracks exonerations via DNA evidence--contains twice as many cases of misidentified Black defendants as misidentified White defendants. Given that Black people make up only 13% of the population, the overrepresentation of Black individuals in exoneration databases is alarming. Despite this large disparity, the lens that researchers have used to examine race effects in the eyewitness context, to the exclusion of almost any other, is own-race bias in the encoding and recognition of faces. Because of an own-race bias in facial recognition, witness performance suffers when a witness is asked to make an identification of a cross-race face, such as when a White witness is tasked with identifying a Black suspect as opposed to a White suspect. In signal detection terms, witnesses who are presented with a cross-race target should produce fewer hits(i.e., correctly state that they have previously seen a face before) and more false alarms(i.e., incorrectly state that they have previously seen a face) than participants who are presented with an own-race target. This pattern is consistent with witnesses being less able to discriminate among innocent and guilty suspects when the suspects are from a different racial group than their own.

Although the own-race bias likely results in increased misidentifications of Black suspects by White witnesses, it fails to sufficiently explain the racial disparities in exoneration data. To do so, rates of cross-race identifications of Black suspects would have to exceed the rates of same-race identifications. However, cross-race identifications are unlikely to be as frequent as own-race identifications given racial distribution of victims and perpetrators. According to the National Archive of Criminal Justice Data, within-race crime is more than five times as prevalent as cross-race crime. This pattern is evident for robberies, a case type for which prosecutors frequently rely on uncorroborated eyewitness evidence. Although witnesses to crimes are not necessarily the same race as the victims of those crimes, racial segregation still exists in residential patterns and friendships, increasing the likelihood that witnesses and victims will belong to the same racial groups. Thus, because most crime occurs within race, and because U.S. neighborhoods and friendships tend to be racially segregated, it is unlikely that the own-race bias in eyewitness identifications is the only cause of racial disparities in eyewitness misidentifications among exonerees.

[. . .]

In 76% of exoneration cases in which the original conviction was based on eyewitness evidence, the defendant was Black as opposed to White. Researchers have examined race in the eyewitness context through the own-race bias, even though most crime is perpetrated by individuals who are the same race as their victims. In the present study, the own-race bias failed to explain racial disparities in exonerations based on eyewitness misidentification on its own. We argue that racial disparities in exoneration data are more likely caused by social interactions that can influence the identifications made by witnesses, such as the interactions between police officers and suspects at the earliest stages of a crime investigation. Overall, our findings provide further support for the need to require officers to have evidence-based suspicion prior to conducting any identification procedure.


Jennifer S. Hunt served as Action Editor.

Jacqueline Katzman(iD) https://orcid.org/0000-0001-9967-8775