Do Double-Blind Administrator Behaviors Exacerbate Suspect Bias?

Project: Research project

Project Details

Description

Do Double-Blind Administrator Behaviors Exacerbate Suspect Bias? Do Double-Blind Administrator Behaviors Exacerbate Suspect Bias? Statement of the Problem DNA exonerations of innocent individuals over the last three decades have consistently implicated eyewitness misidentification as a leading cause of wrongful conviction (innocenceproject.org). A key factor producing these eyewitness misidentifications is social influence, or the process by which social interactions, such as those with a police investigator, influence an eyewitnesss memory, identification decision, or testimony about a witnessed event (Kovera & Evelo, 2017). In an effort to reduce social influence effects in eyewitness identification, researchers have recommended the implementation of double-blind lineups, in which the person administering the lineup does not know which lineup member is the police suspect and which lineup members are innocent fillers. The double-blind lineup procedure is intended to prevent the lineup administrator from influencing the eyewitness through verbal and/or nonverbal cues that could communicate to the eyewitness which lineup member is the suspect. Although double-blind lineup procedures are effective at preventing social influence directed at the suspect, a neglected observation from research on double-blind lineup administration is that double-blind administrators emit a variety of behavioral cues that could still influence eyewitness decision-making. For example, double-blind lineup administrators have been found to instruct witnesses to take their time, ask witnesses if they are sure following a non-identification, and smile prior to an affirmative identification (Greathouse & Kovera, 2009; Zimmerman et al., 2017). Research suggests that even seemingly helpful behaviors such as these increase witnesses choosing from the lineup (Clark et al., 2009). The objective of the proposed research is to test the possibility that behaviors emitted by double-blind administrators that encourage eyewitnesses to choose from a lineup interact with suspect bias factors present in the lineup to increase the risk of mistaken identification. Overview of the Relation of the Problem to the State of the Field Applied lineup theory, the only comprehensive theory of eyewitness identification in that it encompasses both social and cognitive factors influencing eyewitness accuracy, posits that there are two categories of factors that influence eyewitness accuracy: general impairment factors and suspect bias factors (Wells & Olson, 2001; Wells & Loftus, 2003). General impairment factors impair eyewitness accuracy by increasing the rate of errors generally rather than increasing mistaken identifications of the suspect specifically. For example, behaviors from a double-blind lineup administrator that encourage eyewitnesses to choose might increase false positive errors overall but should not increase identifications of the suspect specifically. Suspect bias factors, in contrast, encourage witnesses to choose the suspect rather than one of the lineup fillers. One such suspect bias factor is repeated exposure to the suspect. If the police conduct multiple lineup procedures with the same eyewitness in which the only person in common between the procedures is the suspect, the eyewitness might choose the suspect because he looks familiar (i.e., from the previous lineup) even if he is actually innocent (e.g., Hinz & Pezdek, 2001; Steblay & Dysart, 2016). According to applied lineup theory, general impairment factors exacerbate the effects of suspect bias factors (Charman & Wells, 2007; Brewer & Wells, 2011), yet scant research has investigated this theoretical proposition. The current research will test the possibility that social influence from a blind lineup administrator (a general impairment factor) can magnify the effect of a suspect bias factor (repeated exposure to the suspect) to compound the risk of mistaken identification. Repeated exposure to the suspect has already been shown to increase the risk of mistaken identification: Innocent-suspect identifications increased between 10% and 28% as a result of presenting the suspect repeatedly to the eyewitness (Hinz & Pezdek, 2001; Steblay & Dysart, 2016). Yet, these increases in the rate of innocent Do Double-Blind Administrator Behaviors Exacerbate Suspect Bias? Project Description 2 suspect misidentifications have been documented under conditions devoid of the social influences that could encourage witness choosing. As a result, the fields current best estimates of the impact of suspect bias in the form of repeated exposure to the suspect likely underestimate the risk posed by this practice to innocent suspects. Despite being proscribed by eyewitness scientists (Wells et al., 2020), police sometimes conduct multiple identification procedures in which the eyewitness is repeatedly presented with the same suspect. There are two scenarios in which police are likely to conduct multiple lineup procedures with the same eyewitness and suspect. One scenario is when the police seek to confirm an eyewitnesss initial identification decision. For many years in the state of New York, for example, police routinely conducted in-person lineup procedures following a witnesss identification of the suspect from a photographic lineup because photo identification evidence was inadmissible in New York courts (Fitzgerald et al., 2018). The second scenario is if the eyewitness fails to identify the police suspect or makes only a tentative identification of the police suspect from the initial lineup. Following an unsuccessful initial lineup, the police might conduct another lineup with the same suspect (or they might pursue a different potential suspect whom they present to the eyewitness in a subsequent lineup, a practice generally permitted by eyewitness scientists). Data from real cases involving eyewitness identification suggest that eyewitnesses fail to identify the police suspect in approximately 60% of police lineups (Wells et al., 2020), rendering the use of multiple lineups potentially quite common. How might eyewitnesses respond to administrator influence across multiple lineup procedures? We theorize that eyewitnesses may be differentially susceptible to administrator influence depending on their response to an initial identification procedure. Our focus in the current proposal is on the second scenario described above, in which an eyewitness fails to identify the police suspect from an initial lineup and later attempts an identification of either the same suspect or a new suspect from a subsequent lineup. We are particularly interested in this situation because of how frequently eyewitnesses fail to identify the police suspect in real lineups (Wells et al., 2020) and the relative dearth of research on eyewitness identification performance from multiple lineups (Smalarz et al., 2019). An eyewitness can fail to identify the police suspect in one of two ways: by identifying a known-innocent lineup member (i.e., filler) from the lineup or by rejecting the lineup outright (i.e., identify no one). Archival analyses of real eyewitnesses identification decisions suggest that eyewitnesses identify a known-innocent lineup filler approximately 25% of the time and reject the lineup approximately 36% of the time (Wells et al., 2020). We theorize that witnesses who pick a filler and witnesses who reject an initial lineup will differ in their susceptibility to social influence effects during a subsequent identification procedure. Specifically, we predict that eyewitnesses who pick a filler from an initial lineup will be especially vulnerable to social influence effects during a subsequent identification procedure because making an inaccurate identification from an initial lineup engenders memory distrust, which in turn increases susceptibility to social influence. Memory distrust is a phenomenon in which people come to doubt the accuracy of their own memories (Gudjonsson, 2003; Gudjonsson & MacKeith, 1982). Memory distrust has typically been discussed in the context of internalized false confessions, in which innocent suspects can be led to internalize guilt for a crime as a result of manipulative interrogation tactics that lead them to doubt their own recollections (Gudjonsson, 1993). However, memory distrust can also arise in other contexts. For example, memory distrust can occur in response to completing a difficult memory task (Saucier & Gaudette, 2001) or as a result of receiving negative feedback about ones memory (Gudjonsson, 1984; Linton & Sheehan, 1994). A central premise of the phenomenology of memory distrust is that memory distrust increases susceptibility to social influence because it makes people more willing to rely on external cues and Do Double-Blind Administrator Behaviors Exacerbate Suspect Bias? Project Description 3 suggestions rather their own internal frame of reference (Gudjonsson, 2003). Consistent with this idea, participants who were given negative feedback about their performance on a prior memory test were more susceptible to suggestive questioning than were participants who did not receive negative feedback (Linton & Sheehan, 1994). Moreover, eyewitnesses who were less confident in their ability to identify the culprit were more susceptible to administrator influence effects than were eyewitnesses who were more confident in their ability to identify the culprit (Clark et al., 2013). Hence, theory and data suggest that witnesses who make an initial inaccurate identification (by picking a filler) will be more susceptible to social influence during a subsequent identification procedure than witnesses who either rejected or were not exposed to an initial lineup. Indeed, witnesses who pick fillers tend to do so with low confidence (Smalarz & Wells, 2015) and are likely to receive negative feedback following their identification of a filler. Sometimes lineup administrators tell witnesses outright that they picked a filler (Wells et al., 2015). But even in the absence of explicit negative feedback, eyewitnesses are likely to infer that their prior identification was inaccurate when the police ask them to view another lineup. Even though it has generally been assumed that witnesses who pick fillers should not be relied on to provide identification evidence in a subsequent identification procedure, there are numerous examples of cases in which eyewitnesses who picked a filler went on to identify and testify against a different individual later (see Garrett, 2011; Smalarz et al., 2020). To summarize, we predict that eyewitnesses susceptibility to administrator influence in a subsequent identification procedure is determined in part by their response to the previous lineup, with witnesses who pick fillers being at increased susceptibility to social influence effects during subsequent identification procedures. Specifically, we hypothesize that the effect of administrators behavioral cues on eyewitness choosing behavior and the effect of suspect bias on identifications of the suspect will be stronger for eyewitnesses who picked a filler from a previous lineup than for eyewitnesses who rejected a previous lineup or were not exposed to a previous lineup. Project Method We will conduct two experiments to test whether administrator behaviors that encourage witnesses to choose from a lineup interact with repeated exposure to the suspect to disproportionately increase mistaken identifications. We will further test whether witnesses who made an inaccurate identification from a previous lineup are at increased vulnerability to social influence effects during a subsequent lineup. Experiment 1 will be conducted in person, and Experiment 2 will be an online replication of the in-person experiment, enabling us to collect data from a much larger sample to allow for the calculation of ROC curves and CAC curves, which require at least 100 data points per cell (Metz, 1978). Both experiments will be pre-registered on OSF and all of the materials and data from the research will be made publicly available one year after the conclusion of the research. Design. Both experiments will use a 2 (Prior lineup: No prior lineup vs. Prior innocent-suspect lineup) x 2 (Final lineup: Repeated suspect vs. New suspect) x 2 (Administrator behavior: Pressure to choose vs. No pressure to choose) factorial design. Participants. For Experiment 1, 640 participants will be recruited from the greater metropolitan area and the research participation pool where the research will be conducted. GPower suggests a sample size of 640 is required to detect small effect (OR = 1.68; Chen, Cohen, & Chen, 2010) with power of .80 and alpha of .05. We will recruit approximately twice as many participants into the prior innocent-suspect lineup conditions (n = 430) than into the no prior lineup condition (n = 210) to ensure that there is an adequate sample of participants who chose a filler and who rejected the lineup in the prior lineup conditions. For Experiment 2, 1,500 participants will be recruited on Amazons Mechanical Turk crowd- Do Double-Blind Administrator Behaviors Exacerbate Suspect Bias? Project Description 4 sourcing website, resulting in 125 data points per cell of the design to allow for the creation of ROC curves and CAC curves. Materials. We will create three videotapes of the same mock-theft, each with a different perpetrator. The videos will be scripted so that they are comparable across actors. Using multiple perpetrators is critical for stimulus sampling purposes (Wells & Windschitl, 1999) and will allow us to test the generalizability of the effects under investigation. Because we are interested in examining the effects of administrator influence on mistaken identifications, all photo lineups will be culprit-absent. We will pilot test the initial innocent-suspect lineup to ensure that we are obtaining similar rates of affirmative identifications and rejections to ensure sufficient sample sizes for our statistical comparisons of these two groups susceptibility to social influence during the subsequent lineup. Procedure. Participants will be paired with a confederate acting in the role of a fellow participant. A rigged random-assignment procedure will assign the research confederate to the role of lineup administrator and the participant to the role of eyewitness. Depending on the witnesss prior-lineup condition, the administrator will administer either one or two lineups to the eyewitness. The administrator will not emit any biasing behaviors during the initial lineup (i.e., in the prior-lineup conditions). The administrator will simply record the witnesss decision from the initial lineup (affirmative identification or rejection) and level of confidence in that decision. The administrator will then leave the room ostensibly to report the witnesss decision to the experimenter and will return to the room shortly thereafter, explaining that the witness will be given another opportunity to attempt an identification from a new lineup. If the witness made a previous identification, the administrator will tell the witness that the person the eyewitness previously identified was not the culprit. The administrator will then show the witness a new lineup that either includes one of the individuals from the first lineup (along with all new filler photos; repeated suspect condition) or includes all new photos (new suspect condition). During this lineup, which will be the first lineup for participants in the no-initial lineup condition, the administrator will emit behavioral cues that encourage the witness to choose (in the pressure to choose conditions) or will not emit any behavioral cues (in the no pressure to choose conditions). The behavioral cues will be behaviors that have been observed among double-blind lineup administrators in prior research (e.g., telling witnesses to take their time; asking witnesses to think about the perpetrator from another angle; Greathouse & Kovera, 2009, Zimmerman et al., 2017). Administrators will record witnesss identification decision and level of confidence. The procedure for Experiment 2 will use an online paradigm that will replicate the procedure for Experiment 1 as closely as possible. Rather than having a confederate administer the manipulation of pressure to choose, however, prompts will be delivered by the computer during the identification procedure and in response to eyewitnesses identification decisions. For example, if the eyewitness takes more than 10 seconds to make a decision, the computer will prompt the eyewitness to consider the perpetrator from another angle. If the eyewitness rejects the lineup, the computer will prompt the eyewitness to look again and to take their time. In the no-pressure-to-choose condition, these prompts will not be used. The negative feedback following an initial filler identification for participants in the prior innocent-suspect lineup condition will likewise be administered by the computer (see Smalarz et al., 2019, for a description of a similar online multiple-lineup paradigm). Hypotheses and Analyses. We will test three primary hypotheses. First, we predict that pressure from the lineup administrator will increase the rate of affirmative identifications from the lineup (H1). We will test this hypothesis using a logistic regression analysis in which in which prior lineup condition (no prior lineup; prior rejection; prior identification), final lineup condition (repeated suspect; new suspect) and administrator behavior (pressure to choose vs. no pressure to choose) and their interactions are included Do Double-Blind Administrator Behaviors Exacerbate Suspect Bias? Project Description 5 as predictors of affirmative identification decisions. Second, we predict that pressure from the lineup administrator will increase the rate of innocent suspect identifications when the suspect is repeated across identification procedures compared to when there is a new suspect in the final identification procedure (H2). We will test this hypothesis using a logistic regression analysis in which all experimental factors (prior lineup, final lineup, administrator behavior) and their interactions are included as predictors of innocent suspect identifications. Third, we predict that witnesses who made an inaccurate identification from the initial lineup will be particularly susceptible to social influence effects during the final lineup. We will test this hypothesis using the same analyses for testing Hypothesis 1 and Hypothesis 2. We predict that the tendency for administrator pressure to increase affirmative identifications (H1) will be stronger for witnesses who made a prior inaccurate identification than for witnesses who rejected the initial lineup or saw no initial lineup (H3a). We also predict that the tendency for a repeated suspect to draw more innocent suspect identifications than a new suspect (H2) will be particularly pronounced for witnesses who made a prior inaccurate identification than for witnesses who rejected the initial lineup or saw no initial lineup (H3b). Finally, the data from Experiment 2 will provide a replication test of these three hypotheses and will allow us to conduct ROC and CAC analyses, which measure the diagnostic accuracy of eyewitnesses decisions and the reliability of suspect identifications as a function of confidence. Anticipated Contribution In 2014, the National Research Council called on eyewitness scientists to investigate the interactive effects of variables affecting eyewitness accuracy (NRC, 2014), yet scant research has tested the theoretical proposition that general impairment factors exacerbate the influence of suspect bias factors on eyewitness accuracy. The proposed research marks a key advance toward that end by testing the effects of social influence from a blind lineup administrator on a critical suspect bias variable and by examining the additional role played by eyewitnesses individual vulnerabilities to social influence (by virtue of their prior lineup decision). Moreover, this work will be the first of its kind to use ROC curves and CAC curves to examine social influence effects on the diagnostic value of eyewitness identification evidence. The proposed research makes three additional transformative contributions to the field of psychology and law. First, scholars have heretofore clearly categorized non-blind administration as a suspect-bias factor, the effects of which can be eliminated by double-blind administration. We are not questioning that conclusion. What we are suggesting is that because research on double-blind procedures to date has almost exclusively been conducted with other fair procedures (e.g., single rather than repeated identification attempts of a suspect), this literature does not shed light on whether the behaviors from blind administrators might exacerbate the effects of suspect bias factors (or vice versa) to increase eyewitness misidentifications. Second, extant research on social influence in eyewitness identification has examined social-influence effects during a single identification procedure. The proposed experiments will expand the empirical landscape of administrator influence effects by examining individual eyewitness vulnerabilities to social influence and social influence carry-overs across multiple lineup procedures. Third, a long-term goal of the proposed work is to facilitate the continued development of more effective policy reforms that may reduce the likelihood of misidentification of innocent suspects and their subsequent wrongful conviction. To the extent that behaviors emitted by double-blind administrators increase the risk of mistaken identification, computerized identification procedures may be a preferable alternative to officer-administered lineups. Computerized procedures would not only prevent social influence effects but also allow for the standardization of lineup procedures and contemporaneously record and archive all aspects of the procedure. The proposed research has the potential to provide critical evidence in support of the adoption of fully-computerized lineup administration, which will definitively prevent social influence from contaminating eyewitness identification evidence.
StatusActive
Effective start/end date8/1/217/31/24

Funding

  • American Psychological Association (APA): $5,000.00

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.