A Sample of Witness, Crime, and Perpetrator Characteristics Affecting Eyewitness Identification Accuracy

• Author: Brian L. Cutler
• Position: Professor and Chair of the Department of Psychology
• Pages: 327-340

Page 327

Brian L. Cutler, Ph.D., is Professor and Chair of the Department of Psychology at the University of North Carolina at Charlotte. He is the author of Mistaken Identification: The Eyewitness, Psychology, and the Law (with Steven Penrod, 1995, Cambridge University Press) and numerous book chapters and research articles on the psychology of eyewitness memory. He is also Editor-in-Chief of the journal Law and Human Behavior. Professor Cutler has been qualified as an expert witness in several states and federal courts and has served as a consultant or expert witness in about 100 cases.

Overview

Archival and laboratory research on eyewitness memory has clearly established that eyewitness identification errors occur and have led to erroneous convictions.¹ As of the time of this writing, the Innocence Project²has uncovered 159 cases of erroneous conviction with mistaken identification as the leading precursor. To psychologists, it comes as no surprise that mistaken identification occurs, for over 100 years of research on human memory has revealed many of its limitations.

In their attempts to better understand errors in eyewitness identification, many researchers have devoted their efforts to uncovering the factors that increase or decrease the likelihood of mistaken identification. Much of the recent research has focused on system variables-factors that are under the control of investigators-such as how photo arrays are composed and presented to eyewitnesses.³System variables are of interest because they can be used to increase the likelihood ofPage 328 correct identification and decrease the likelihood of false identification.⁴Other research, however, has examined the roles of estimator variables- factors associated with the eyewitness, the perpetrator, or the eyewit-nessed event that are not under the control of investigators-such as the conditions under which the eyewitness viewed the perpetrator.⁵Estimator variables are of interest because knowing how estimator variables influence identification accuracy gives us some basis for gauging the reliability of eyewitness identifications. The better we are able to assess identification accuracy, the better we can assess defendant culpability and ultimately determine the best way to resolve a criminal charge.

Considerable psychological research has addressed the question of whether there are reliable associations between estimator variables and eyewitness identification accuracy, and, if so, which specific estimator variables predict identification accuracy. The purpose of this paper is to summarize some of the main findings from this research and to direct the interested reader to relevant summary papers concerning these and other estimator variables.⁶

In this paper, I will summarize six specific estimator variables. These were chosen from a more exhaustive set⁷for the following reasons: (1) research findings point to clear conclusions regarding their effects; (2) a survey of eyewitness researchers revealed substantial agreement about (most of) their effects; (3) my personal experience in about 100 cases in which I served as a consultant or expert witness reveals that these variables are common; and (4) research shows that most of the variables are not merely a matter of commonsense-a justification often given by judges for excluding expert psychological testimony about estimator variables. An additional estimator variable, eyewitness confidence, also meets these criteria but will not be discussed in this article.

My discussion will proceed as follows: First, I will present a research summary for each factor. Next, I will summarize the results from the survey of eyewitness researchers. Finally, I will present research addressing the extent to which these factors are a matter of common sense to jurors.Page 329

I Estimator Variables

Where possible, I rely on conclusions from meta-analyses to explain the effects of estimator variables. Meta-analysis is a technique for summarizing the results of multiple-sometimes conflicting-studies to arrive at a general conclusion regarding the effects of a variable on identification accuracy. For example, meta-analysis can be used to average the results across numerous studies to see if, across studies, the effect of a given factor is positive, negative, or inconsequential. Meta-analysis can also be used to identify factors that qualify the effect of a given estimator variable on identification accuracy. For example, one estimator variable might have a large effect on identification accuracy in a group of studies that share a common feature but a small effect in another group of studies that do not share that feature. More specific examples are given below in the descriptions of the study findings. It is important to note that meta-analysis is a commonly accepted technique for reviewing a research literature both within and outside the field of psychology.

Another common feature of the research cited below is the use of target-present and target-absent lineups or photo arrays. The target is the person to be identified, such as the robber in a simulated robbery. Target-present lineups are those in which the to-be-identified target appears in the lineup, thus simulating the situation in which the suspect is the perpetrator. Modern research also commonly uses target-absent lineups, in which the target is replaced with someone who looks like him. Target-absent lineups simulate the condition in which the suspect is not the perpetrator. Normally, each eyewitness is shown either a target-present or target-absent lineup, not both. Use of both target-present and target-absent lineups enables the researcher to independently examine the influence of each factor on the likelihood of correct identifications (i.e., from target-present lineups) and false identifications (i.e., from target-absent lineups).

A Own-Race Bias

Own-race bias refers to the common finding that eyewitnesses are more accurate at identifying perpetrators of their own race than perpetrators of other races.⁸A recent meta-analysis reviewed the results of thirty-nine separate studies involving ninety-one separate experimentalPage 330 tests of own-versus same-race identifications.⁹In total, nearly 5,000 eyewitnesses participated in these studies.¹⁰Across all studies, witnesses were 1.4 times more likely to correctly identify a previously viewed own-race face when compared with performance for other-race faces.¹¹Witnesses were 1.56 times more likely to falsely identify a novel other-race face when compared with own-race faces.¹²White participants demonstrated a significantly larger own-race bias when compared with Black participants, but only with respect to false identifications.¹³With respect to correct identifications, Whites and Blacks showed the same own-race bias.¹⁴The own-race bias effect was larger with shorter exposure durations (i.e., the amount of time for which the eyewitness was able to view the perpetrator).¹⁵This meta-analysis also examined recognition of other races (e.g., Hispanic, Asian), but there were many fewer studies using these groups.¹⁶

The own-race bias is nicely illustrated in a study by Platz and Hosch (1988).¹⁷They conducted a field study in which White, Black or Hispanic customers visited and interacted with ninety White, Black, or Hispanic convenience clerks.¹⁸Two to three hours after each visit, an investigator asked the clerk to attempt an identification from photo arrays containing the customers.¹⁹White clerks were more likely to correctly recognize White customers (53.2%) than Black (40.4%) or Mexican (34%) customers.²⁰Black clerks were more likely to correctly recognize Black (63.6%) than White (54.6) or Hispanic (45.4%) customers.²¹Hispanic were more likely to correctly recognize Hispanic (53.6%) than White (35.7%) or Black (25%) customers.²²Page 331

B Exposure Duration

Exposure duration refers to the amount of time for which an eyewitness is able to view a perpetrator at the time of the crime. Some crimes occur in a matter of seconds or minutes (in such cases the eyewitness's opportunity to view the perpetrator is very brief), while others unfold over hours or-such as in the case of kidnapping-days. Not surprisingly, the more time an eyewitness has to view a perpetrator, the more time she has to encode the perpetrator's characteristics into memory, and the more likely she is to make a correct identification at a later time. In a meta-analysis of 128 studies involving nearly 17,000 participant-witnesses, exposure duration significantly predicted identification accuracy.²³

To illustrate, Memon, Hope, and Bull exposed sixty-four young adults (ages 17-25) and older adults (ages 59-81) to a videotaped reconstruction of a robbery in which the perpetrator's face could be seen for either twelve seconds or forty-five seconds.²⁴Each witness then attempted to identify the robber from a target-present or target-absent photo array.²⁵Exposure duration had a significant impact on identification accuracy;²⁶95% of the young adults and 85% of the older adults made correct identifications from the target-present photo arrays when the robber was exposed for forty-five seconds,²⁷but only 29% of the young adults and 35% of the older adults made correct identifications when the robber was exposed for twelve seconds.²⁸Similarly, 41% of the younger adults and 50% of the older adults made false identifications from the target-absent photo arrays when the target was exposed for forty-five seconds,²⁹but 90% of the younger adults and 80% of the older adults made false identifications when the robber was exposed for only twelve seconds.³⁰Page 332

It is not uncommon for crime perpetrators to...