The forensic sciences have recently been criticized, most notably by the National Academy of Sciences, due to the alarming frequency of forensic science errors in known DNA exoneration cases. Empirical data now suggests that confirmation bias -- i.e., the tendency to interpret new evidence in ways that validate one's pre-existing beliefs -- can produce such errors. That is to say, forensic examiners' objectivity and accuracy can be undermined by their a priori beliefs regarding a suspect's guilt or innocence. The current studies will enhance our knowledge of factors that exacerbate and mitigate confirmation biases in judgments of forensic evidence, in hopes of guiding reforms to minimize the risk of costly miscarriages of justice.
In each study, mock forensic examiners will assist in a simulated criminal investigation by comparing handwriting samples from the perpetrator and suspect. Study 1 will test whether mock examiners who are led to believe that the suspect is guilty will consequently perceive exculpatory handwriting evidence as more incriminating than those who are not. In addition, Study 1 will explore whether this bias is exacerbated when the handwriting samples are highly similar and/or examiners receive simulated negative feedback from police investigators.
Study 2 will provide the first empirical test of "evidence lineups," a procedural reform often proposed as a means of combating bias, but whose efficacy remains unknown. Having been led to expect the suspect's guilt, innocence, or neither, some examiners will receive one handwriting sample each from the perpetrator and suspect -- a method which some claim is inherently suggestive of guilt. The remaining examiners will utilize an "evidence lineup," which forces them to compare three different suspect samples against the perpetrator's sample and indicate which, if any, matches it. Study 2 will thus provide the first empirical data to inform the longstanding debate over the potential value of evidence lineups.
To date, 317 wrongfully-convicted individuals have been exonerated by DNA testing, and forensic science errors have been found in approximately half of these cases. In the hopes of avoiding similar miscarriages of justice in the future, the National Academy of Sciences recently called for additional research on psychological factors that increase the risk of forensic science error. One such factor is confirmation bias, i.e., the tendency for people to judge new evidence as consistent with their pre-existing beliefs. When forensic scientists are exposed to information that implies the guilt or innocence of a criminal suspect, this expectation may lead them to unconsciously evaluate the forensic evidence in a biased way, which can increase the risk of making an erroneous judgment. The aims of this project were to explore how confirmation bias affects judgments of forensic evidence, and to test methods of counteracting these effects. In Study 1, participants read a summary of an ongoing bank robbery investigation, and then compared handwriting samples from the robber and the suspect. Our results showed that participants who were told that the suspect had confessed to the robbery judged the same handwriting samples as more similar, and were more likely to conclude that they were written by the same person when in fact they were not, compared to participants who were told that the suspect had maintained his innocence. In other words, the content of the summary influenced participants' subsequent evaluations of the handwriting evidence, such that they became more likely to mistakenly implicate an innocent suspect. To reduce the effects of bias, some researchers have argued that forensic scientises should make their judgments using evidence lineups. That is to say, instead of analyzing only two samples (one from the crime scene and one from the suspect), forensic scientists compare one sample against an array of several comparison samples -- similar to an eyewitness lineup. However, there is virtually no existing research on evidence lineups, and thus we set out to test their effectiveness in Studies 2 and 3. In Study 2, participants read a summary of a robbery investigation, and were then given either a single pair of handwriting samples (i.e., an evidence showup, as in Study 1) or an array of handwriting samples (i.e., an evidence lineup). Although showups produced a higher percentage of accurate judgments than lineups, our results suggested that showups were also more susceptible to confirmation bias than were lineups. In Study 3, we compared two methods of presenting an evidence lineup: simultaneously (i.e., all of the comparison samples are shown at the same time, as in Study 2) and sequentially (i.e., comparison samples are shown one-at-a-time). We found that, relative to simultaneous lineups, sequential lineups reduced the amount of incorrect judgments but did not reduce the amount of correct judgments. This finding is noteworthy, given that this same phenomenon has been found in eyewitness lineups as well. Taken together, the results of Studies 2 and 3 suggest that the processes of eyewitness identification and forensic identification may be similar. As such, the vast body of existing research on eyewitness lineups may be able to help improve the quality of forensic science judgments. It is our hope that the findings of the current study will stimulate additional research on this topic, which will ultimately be used to promote reforms that will reduce the incidence of wrongful convictions in the criminal justice system.
