When eyewitness testimony is used in the courtroom, it is seen as a compelling form of evidence. There are, however, many variables that can affect the accuracy of the testimony. The amount of time the witness saw the culprit, the amount of time between the crime and the viewing of the lineup, the presence of misleading information before viewing the lineup, and whether the witness and culprit are of the same race or different race, can all influence the accuracy of eyewitness testimony. Jurors often believe that the credibility of the testimony is dependent upon the confidence of the witnesses; however, researchers have found little or no relationship between the confidence of the witnesses and the accuracy of the identification. What, then, can be used as a determinant of accurate eyewitness testimony?

The researchers believe that the answer could be found in the response times of the witnesses making identifications; those making accurate identifications would do so in a shorter period of time than those not making accurate identifications. The researchers claim that these time differences are found because of the different cognitive processes that are used with accurate and inaccurate witnesses. Accurate witnesses are more likely to be making automatic decisions without conscious effort, strategy, or awareness. Inaccurate witnesses are more likely to deliberate using a conscious elimination process. Therefore, the question to be answered is… "Is there some point at which the investigator can be confident that a witness choosing someone in the lineup is likely to be accurate, or to be inaccurate?" (pg. 952).

To answer this question, Dunning and Peretta used four studies (two old, two new) to analyze response times and accuracy.

Method

Participants in Studies 1 & 2 watched a videotape about recruiting people to work in the childcare industry. At the end of the video they saw a man go up to a purse, take the money out of the wallet, and walk off. In Study 1, the participants were told to "think-out-loud" while they analyzed the line-up. The participants in Study 2 were not supposed to think out loud, but some of them were forced to make their identification using automatic processing (by rehearsing a 9-digit number during the task). Participants in Studies 3 & 4 viewed an actual scene in one of their classes and were then asked to come to the lab to identify the culprit.

Results

147 witnesses accurately identified the culprit, 116 mistakenly identified someone else. (Table 1, pg. 955)

Witnesses making accurate identifications were faster than those making inaccurate identifications across all four studies. (Table 2, pg. 955). (Study 1 took a considerably longer amount of time in both conditions, but this is in part due to the think-out-loud strategies that were employed.)

To determine a ‘time-boundary’ each study was examined separately. For each split (conducted in intervals of 1 second) the number of accurate and inaccurate witnesses within the boundary and outside of the boundary were counted. A chi-square analysis was then computed, with larger chi-square values indicating that the time-boundary defining the boundary differentiated between accuracy and inaccuracy to a greater degree. (Table 3, pg. 956). All studies showed strong peaks between 10 and 12 seconds (with all studies having smaller, secondary peaks at later time intervals). The same thing was done again, but this time a log odds ratio was computed. Again, a peak was found around 10 to 11 seconds.

When looking at the percent of accurate identifications in 10-second blocks, 87.1% of those witnesses making an identification in the first 10 seconds were accurate. This percentage fell to 52.5 % for 11-20 seconds, 51.4% for 21-30 seconds, and 40.7% for 31-40 seconds.

When analyzing the confidence of the participants, those making accurate identifications were significantly more confident than those witnesses making inaccurate identifications in Study 1. The same trend emerged in the subsequent studies (but significance was not reached).

Participants that made accurate identifications were more likely to claim that they made those decisions based on more automatic processing types of strategies, whereas participants making inaccurate identifications were more likely to claim that they had used process-of-elimination strategies. (This would help explain why accurate identifications take less time than inaccurate ones.)

General Discussion

Witnesses who made accurate identifications made faster decisions than those who made inaccurate identifications. Dunning and Peretta claim that the 10-12 second mark is the "time-boundary" with which most positive identifications are made. However, they do point out that we should not conclude that witnesses who take longer than 10-12 seconds are incorrect. They also point out that this finding is even stronger given the fact that they analyzed the response times of four different studies that all had different aspects associated with them (thinking out loud, rehearsing the 9-digit number, and actually witnessing the event as it was happening). It is also important to remember that, because correlation does not imply causation, forcing witnesses to make a quick judgment would not enhance accuracy.