Supplemental data for this article are available online at https://doi.org/10.1080/13218719.2020.1733696.
Information pertaining to this study was placed on the Open Science Framework (https://osf.io/vbzrk/?view_only=340f0aaecc414e0e877b9b2ace4d77a1). We thank Martin Safer for his excellent suggestions for improving the article.
Copyright © 2020 The Australian and New Zealand Association of Psychiatry, Psychology and LawFor over 35 years, scholars have searched with little success for a legal safeguard that can sensitize jurors to eyewitness testimony. The present study explored whether expert testimony that uses the I-I-Eye method of analyzing eyewitness testimony can improve juror sensitivity to eyewitness evidence. Participants read a trial transcript with no expert testimony, standard expert testimony or expert testimony that used the I-I-Eye method. The two transcripts for the three expert groups had either strong or weak eyewitness testimony. Unlike the control participants, the I-I-Eye expert participants rendered significantly more guilty verdicts in the strong than in the weak case. The standard expert testimony did not affect verdicts even though it increased participants’ knowledge of the eyewitness factors. It appears that the I-I-Eye method improved sensitivity because it not only increased participants’ knowledge of eyewitness factors, but also explained how to use that knowledge in assessing eyewitness accuracy.
Keywords: adjudication, decision making, expert testimony, eyewitness, eyewitness interview, judges, juries, legal evidence, memory
Eyewitness error is the leading cause of wrongful convictions (Smith & Cutler, 2013; Technical Working Group for Eyewitness Evidence, 1999). Several reasons account for its prominent role in wrongful convictions. Research generally shows that jurors have limited knowledge of eyewitness factors (Benton, Ross, Bradshaw, Thomas, & Bradshaw, 2006; Desmarais & Read, 2011; Leippe & Eisenstadt, 2009). Attorneys, judges and law enforcement officers also have limited knowledge of eyewitness factors, which decreases their ability to help jurors assess eyewitness accuracy (Benton, McDonnell, Ross, Thomas, & Bradshaw, 2007).
Consequently, for over 35 years many scholars have recommended eyewitness experts as a safeguard against erroneous eyewitness testimony (Devenport & Cutler, 2004; Leippe & Eisenstadt, 2009; Penrod & Cutler, 1999). Eyewitness experts can provide jurors with current, detailed scientific information about human memory and eyewitness factors. They can also tailor their testimony to the specific eyewitness factors in a case. Therefore, unlike other legal safeguards, the effectiveness of eyewitness expert testimony does not depend on legal professionals and jurors’ ability to identify the relevant eyewitness factors in a case and to know how they affect eyewitness accuracy (Leippe & Eisenstadt, 2009). Furthermore, several U.S. appellate courts have held that when the prosecution’s primary evidence of guilt is eyewitness testimony, the exclusion of eyewitness expert testimony can constitute reversible error (Mnookin, 2015).
Cutler, Penrod, and Dexter (1989) stated that eyewitness experts can have three effects: (a) no effect because the expert confuses jurors or does not persuade them; (b) skepticism because the expert causes jurors to disbelieve the eyewitness even if the eyewitness is likely to be accurate; and (c) sensitivity because the expert helps jurors to identify the relevant eyewitness factors and to apply them to the case.
Eyewitness experts appear to increase jurors’ knowledge of eyewitness factors at least to some extent (Leippe & Eisenstadt, 2009). However, experts may still produce juror skepticism if jurors do not understand how to apply their increased knowledge to the case. For example, Magnussen, Melinder, Stridbeck, and Raja (2010) generally found no relationship between greater knowledge of eyewitness factors and improved assessments of eyewitness accuracy. Even experts have difficulty applying their knowledge to a case (Cutler & Penrod, 1995).
Schemata are mental representations of knowledge that serve as ‘scaffolding’. They help guide attention and interpretation, and individuals who lack the requisite schema for a particular type of information have difficulty understanding the information and using it (Bruning, Schraw, & Norby, 2010). In prior studies, the eyewitness experts at most explained how eyewitness factors individually likely affected eyewitness accuracy. The experts did not provide mock jurors with an analytical framework for assessing how all the eyewitness factors in the case, both positive and negative, collectively likely affected eyewitness accuracy. In short, to avoid juror skepticism, eyewitness experts may have to provide jurors with a schema for analyzing eyewitness testimony that helps them to apply their knowledge of eyewitness factors to the case.
Wise developed the interview–identification–eyewitness factor (I-I-Eye) method for analyzing eyewitness accuracy to help jurors and legal professionals assess eyewitness testimony (Wise, Fishman, & Safer, 2009). The I-I-Eye method consists of four steps: First, it is determined whether the eyewitness interviews were properly conducted by assessing whether law enforcement (a) obtained the maximum amount of information from the eyewitness; (b) contaminated the eyewitness’s memory of the crime or perpetrator with post-event information; or (c) increased the eyewitness’s confidence. The I-I-Eye method provides criteria for evaluating the eyewitness interviews (Technical Working Group for Eyewitness Evidence, 1999; Wise et al., 2009). For instance, did the interviewer establish rapport with the eyewitness, tailor questions to the eyewitness’s narrative and not interrupt the eyewitness?
Second, the quality of the identification procedures in the case is evaluated by determining whether proper procedures were followed in conducting them. For example, was the lineup double blind, was there only one suspect in the lineup, and were cautionary instructions given to the eyewitness? Third, assuming there was no substantial bias in how the eyewitness interviews and identification procedures were conducted, or if an exception applies (e.g. the viewing conditions were very good or other evidence corroborates the eyewitness testimony), the eyewitness factors at the crime scene are evaluated. For instance, did the perpetrator use a weapon or wear a disguise, or was a different race from that of the eyewitness?
Lastly, summary questions are answered to determine the likely accuracy of the eyewitness testimony. The summary questions include questions such as was the maximum amount of information obtained from the eyewitness during the interview? Is there a high, medium or low probability that the eyewitness identification was accurate (Wise et al., 2009)?
Pawlenko, Safer, Wise, and Holfeld (2013) investigated whether the I-I-Eye method could improve participants’ assessment of eyewitness accuracy. Participants read one of three teaching aids that were presented on 24 PowerPoint slides at the beginning of the study: (a) the Jury Duty aid (i.e. general jury instructions), (b) the Neil v Biggers aid (i.e. discussed the five eyewitness factors the U.S. Supreme Court mentioned in Neil v Biggers, 1972), or (c) the I-I-Eye method aid. Participants then read a trial transcript with either strong or weak eyewitness testimony. Only the I-I-Eye expert participants could differentiate between the strong and the weak case.
Safer et al. (2016) determined whether the I-I-Eye method would still be effective when other inculpatory evidence was added to a criminal case and whether it would produce skepticism of the other evidence. In the two experiments, only the I-I-Eye expert participants were sensitized to the eyewitness evidence, and it did not increase skepticism of the other evidence.
The I-I-Eye method of analyzing eyewitness testimony may improve the effectiveness of eyewitness expert testimony. It not only helps jurors identify and organize the different kinds of eyewitness factors in a case, but it also provides them with a framework for applying them to a case. It gives jurors standards for evaluating the eyewitness interviews, identification procedures and eyewitness factors at the crime scene (i.e., was there substantial bias in how the interviews or identification procedures were conducted? Were the eyewitness conditions at the crime scene poor?). The I-I-Eye method also specifies the order in which eyewitness factors are evaluated, and begins with the jurors evaluating the eyewitness interview and identification procedures.
The I-I-Eye method begins with an analysis of system variables in the case because the criminal justice system can control system variables and can usually create an objective record of them by videotaping them. In contrast, the criminal justice system cannot control estimator variables, and often there is no objective record of them (Wells, 1978). Consequently, jurors must frequently rely on an eyewitness’s subjective report about the estimator variables.
Moreover, post-event information provided during an eyewitness interview (Loftus, 1975, 1979), and post-identification feedback can alter an eyewitness’s retrospective report of estimator variables (Wells & Bradfield, 1998). Accordingly, if the eyewitness interview and identification procedures are properly conducted, the eyewitness’s report of the estimator variables is more likely to be accurate. Lastly, the expert’s discussion of the summary questions of the I-I-Eye method may help jurors make the correct conclusion about the likely accuracy of the eyewitness testimony.
In U.S. federal courts and most U.S. state courts, Daubert v Merrell Dow Pharmaceuticals, Inc. (1993) determines whether expert testimony is admissible. Under Daubert, expert testimony is admissible if it is based on reliable and valid scientific evidence, and will assist the trier of fact in understanding or determining a fact in issue. Although there is limited research on the I-I-Eye method, the extant research suggests that it meets the Daubert’s criteria of admissibility.
Prior studies show that the I-I-Eye method can sensitize participants to eyewitness testimony. However, judges are unlikely to permit the I-I-Eye method to be explained to jurors by a 24-slide PowerPoint presented at the beginning of the trial before the eyewitness has testified as was done in the prior studies. The present study tested the I-I-Eye method under circumstances where judges are more likely to permit its use (i.e. as part of an expert’s testimony). The study also differs from prior studies of the I-I-Eye method because it included jury instructions and varied not only the system variables but also the estimator variables in the cases. Lastly, we evaluated whether the I-I-Eye method can improve eyewitness expert testimony.
Participants read a trial transcript describing a convenience store robbery and the murder of the store clerk that a sole eyewitness observed. The transcripts varied how the eyewitness interview and lineup were conducted and the quality of the eyewitness conditions during the crime. In the strong case, the police followed more proper procedures for conducting the eyewitness interview and the lineup than in the weak case, and the eyewitness conditions during the crime were better (see Table 1 ). The transcripts had no expert testimony, standard expert testimony or expert testimony that used the I-I-Eye method.
Eyewitness factors in the strong and weak cases.
| Eyewitness factors a | Strong case | Weak case |
|---|---|---|
| 1. Asked leading questions. [I] | no | yes |
| 2. Interrupted the eyewitness. [I] | no | yes |
| 3. Eyewitness told not to guess & to indicate uncertainty. [I] | yes | no |
| 4. Told the eyewitness to mentally recreate the crime. [I] | yes | no |
| 5. Distractions in interview room. [I] | no | no |
| 6. When interview conducted after crime. [I] | one week | one week |
| 7. Retention interval for lineup. [L] | one week | one week |
| 8. Lineup size. [L] | 12 | 6 |
| 9. Type of lineup. [L] | sequential | simultaneous |
| 10. Double blind lineup. [L] | yes | no |
| 11. Cautionary lineup instruction. [L] | yes | no |
| 12. Only one suspect. [L] | yes | yes |
| 13. Foils match eyewitness description. [L] | yes | yes |
| 14. Perpetrator wore hat. [C] | yes | yes |
| 15. Weapon used. [C] | yes, but not visible | yes |
| 16. Reported crime duration in minutes. [C] | 3 | 1 |
| 17. Estimated distance in feet from perpetrator. [C] | 10–15 | 20–15 |
| 18. Lighting good at crime scene. [C] | yes | yes |
| 19. No cross-racial identification. [C] | yes | yes |
| 20. Eyewitness has good vision. [C] | yes | yes |
| 21. High stress. [C] | yes | yes |
Note: [I] = interview factors; [L] = lineup factors; [C] = crime scene factors.
a Eyewitness factors that differed between the strong and weak transcripts are in bold.The following hypotheses were tested in the study:
Hypothesis I: The I-I-Eye expert participants will know more about eyewitness factors in the cases than the no expert and standard expert participants.
Hypothesis II: The I-I-Eye expert participants will be significantly more likely than the no expert and standard expert participants to render a guilty verdict in the strong eyewitness case and a not guilty verdict in the weak eyewitness case.
Hypothesis III: The I-I-Eye participants will be able to distinguish between the strength of the prosecution and the defense case in the strong and weak eyewitness case but not the no expert and standard expert participants.
Hypothesis IV: The I-I-Eye expert participants will rate the eyewitness interview, lineup and crime scene factors as significantly better in the strong than in the weak eyewitness case but not the no expert and standard expert participants.
For the current study an a priori power analysis conducted with G*Power 3.1.9 (Faul, Erdfelder, Buchner, & Lang, 2009) determined that at least 158 mock jurors (27 per condition) were needed to detect a medium-sized effect with 80% certainty. Participants were 230 (Mage = 20.38 years, SD = 3.36) undergraduate students from a medium-sized Midwestern university in the US. There were 160 female students and 66 male students (four participants did not indicate their gender). The sample was predominantly Caucasian (90.7%), and participants earned extra credit for their participation. Informed consent was obtained from all participants in the study, and the university’s institutional review board approved the study.
A 2 (case strength: strong vs. weak) × 3 (expert type: no expert vs. standard expert vs. I-I-Eye expert) between-participant factorial design was employed. Participants sat at individual cubicles and were randomly assigned to read one of six transcripts. Because the transcripts without an expert were shorter than the transcripts with the standard expert and the I-I-Eye expert (20 pages vs. 30 pages vs. 32 pages, respectively), the no expert participants completed a brief filler task (i.e. solved Sudoku puzzles) prior to reading the transcript. After reading the transcript, participants completed a questionnaire and were debriefed.
The transcripts concerned a convenience store robbery and murder of the store clerk that a sole eyewitness observed. In all the transcripts, the eyewitness and the detective in the case testified for the prosecution. The defendant’s girlfriend provided an alibi for the defendant. The defendant did not testify. If there was an expert, he testified for the defense. All the transcripts contained opening and closing arguments, direct and cross-examinations of witnesses and in some instances redirect and re-cross-examinations of witnesses, and jury instructions.
The transcripts included 21 eyewitness factors. Of the six interview eyewitness factors, four of those factors differed in the strong and weak cases (if leading questions were asked, if the eyewitness was interrupted, if the eyewitness was told not to guess and to indicate uncertainty in her answers, and if the eyewitness was told to mentally recreate the crime), and two factors were the same (no distractions and interview conducted one week after crime; see Table 1 ).
Of the seven lineup factors, four differed in the strong and weak cases (lineup size, type of lineup, if the lineup was double blind, and if cautionary instructions were given), and three factors were the same (retention interval, only one suspect in the lineup, and foils matched the eyewitness’ description of the perpetrator). Of the eight crime scene factors, three differed in the strong and weak cases (if the gun was visible, crime duration, and distance), and five factors were the same (perpetrator wore a hat, good lighting conditions, no cross-race identification, eyewitness had good vision, and high stress). In sum, 11 of the 21 eyewitness factors differed in the strong and weak cases. In both cases, the expert testified that the eyewitness likely significantly overestimated the time she viewed the crime because of stress (see Table 1 ).
Multiple eyewitness factors were included in the case to increase the ecological validity of the cases. Moreover, the weak case contained several eyewitness factors that were conducive to an accurate identification, and the strong case included two eyewitness factors at the crime scene that likely impaired accuracy (i.e. the perpetrator wore a hat, and the eyewitness experienced high stress). Nonetheless, the eyewitness factors in the strong case were substantially better than the eyewitness factors in the weak case. In addition, unlike most studies of expert testimony, the eyewitness factors in the cases pertained not only to the crime scene and the identification procedure but also the eyewitness interview. How the interview is conducted can affect eyewitness accuracy (e.g. it can alter an eyewitness’s memory of the crime and perpetrator; Loftus, 1975, 1979; Wise et al., 2009).
The defense attorney asked both the standard expert and the I-I-Eye expert to explain the nature of memory, its stages, and what eyewitness research has generally discovered about eyewitness memory. The defense attorney also asked both experts about the eyewitness factors in the case that were likely to decrease eyewitness accuracy. The prosecutor cross-examined both the standard and I-I-Eye experts about eyewitness factors in the case that were likely to increase accuracy. The same eyewitness factors were discussed in the strong and weak cases.
Participants first rendered their verdict in the case (i.e. guilty or not guilty). They used 9-point Likert scales with labels of 1 = Not at all, 5 = Neutral, and 9 = Very to rate: (a) their confidence in their verdict; (b) the influence of the different witnesses on their verdict; (c) the effectiveness of the expert’s testimony (only for the expert participants); (d) the effectiveness of the I-I-Eye method (only for the I-I-Eye expert participants); (e) the strength of the prosecution and defense evidence; (f) the suggestiveness of the interview and lineup; (g) the overall fairness of the eyewitness interview and lineup; and (h) the likely effect of the crime scene factors on the accuracy of the eyewitness.
Participants also answered questions about the eyewitness factors in the case and gave reasons for their verdicts. In rating the eyewitness factors, they used 9-point Likert scales with labels of −4 = Leads to False Testimony or Identification, 0 = Neutral, and 4 = Leads to Accurate Testimony or identification. However, because the experts did not specify the magnitude of the effect of the eyewitness factors, 1 point was given for a correct answer, −1 point for an incorrect answer and 0 points for a neutral answer to questions about the eyewitness factors. Lastly, the participants provided demographic information.
To assess the case strength manipulation, participants rated the suggestiveness of the eyewitness interview, identification procedures and the quality of the eyewitness factors at the crime scene in the strong and weak cases. To determine whether the participants’ knowledge of the case facts differed in the three expert groups, and whether they understood the facts, they answered nine true or false questions: for example, whether the perpetrator wore a hat, whether the time interval between the crime and lineup was one week, whether the detective interrupted the eyewitness during the interview, and whether the detective warned the eyewitness that the perpetrator may not be in the lineup.
Participants in the strong case rated the eyewitness interview as less suggestive than the participants in the weak case, t(206.20) = −6.14, p < .001, d = 0.81. They also rated the identification procedure as less suggestive than the participants in the weak case, t(203.20) = −5.169, p < .001, d = 0.76. Participants’ ratings of how good the eyewitness factors were at the crime scene in the strong and the weak case were marginally significant, U = 5677.00, z = −1.669, p = .095, r = .11 (see Table 2 ). In sum, the manipulation of case strength appeared to be successful.
Participants’ ratings of interview, lineup, crime scene, prosecutor case and defense case.
| Variable | Strong case | Weak case | ||
|---|---|---|---|---|
| M | SD | M | SD | |
| All participants’ ratings of suggestiveness of interview | 4.16 | 2.42a | 5.85 | 1.72b*** |
| All participants’ ratings of suggestiveness of lineup | 4.15 | 2.56a | 5.81 | 1.77b*** |
| All participants’ ratings of eyewitness conditions at crime scene | 6.30 | 2.03 | 5.89 | 2.03 |
| I-I-Eye participants’ ratings of prosecution case | 7.19 | 1.05a | 5.86 | 2.20b* |
| Standard expert participants’ ratings of prosecution case | 6.05 | 2.04 | 6.28 | 2.34 |
| No expert participants’ ratings of prosecution case | 5.69 | 5.65 | 6.00 | 2.10 |
| All participants’ ratings of defense case | 5.63 | 1.78a | 6.44 | 1.73b*** |
| I-I-Eye participants’ ratings of interview | 7.81 | 1.02a | 5.61 | 1.93b*** |
| Standard expert participants’ ratings of interview | 7.10 | 1.39a | 6.05 | 1.61b** |
| No expert participants’ ratings of interview | 6.33 | 2.07 | 5.69 | 1.95 |
| All participants’ ratings of the lineup | 7.07 | 1.73a | 5.54 | 1.83b*** |
| I-I-Eye expert participants’ ratings of crime scene | 6.84 | 1.40a | 5.77 | 2.04b* |
| Standard expert participants’ ratings of crime scene | 6.33 | 2.07 | 5.74 | 2.09 |
| No expert participants ratings’ of crime scene | 5.77 | 2.32 | 6.13 | 1.98 |
Note: Ratings were based on 9-point Likert scales with labels of 1 = not at all, 5 = neutral, and 9 = very. Means that share subscripts do not differ significantly from each other.
A multivariate analysis of variance (MANOVA) indicated that the participants’ knowledge of nine facts in the strong and weak cases did not significantly differ in the three expert conditions, V = .013, F(4, 444) = 0.699, p = .593 (see Table 3 ). The average number of correct answers (M = 8.27, SD = 1.21) to the nine fact questions indicates that the participants understood the facts of the cases.
Participants’ knowledge of case facts and eyewitness factors.
| Variable | No Expert Participants | Standard Expert Participants | I-I-E Expert Participants |
|---|---|---|---|
| Knowledge of 9 Case Facts in Strong Case | M = 8.08, SD = 1.15 | M = 8.13, SD = 1.26 | M = 8.17, SD = 1.34 |
| Knowledge of 9 Case Facts in Weak Case | M = 8.10, SD = 1.50 | M = 8.62, SD = .78 | M = 8.54, SD = 1.01 |
| Knowledge of 10 Eyewitness Factors in the Weak Case a | M = 1.08, SD = 2.92 a | M = 4.05, SD = 2.92 b *** | M = 4.74, SD = 3.21 b *** |
| Knowledge of 10 Eyewitness Factors in the Strong Case a | M = 3.15, SD = 3.19 a | M = 5.55, SD = 3.48 b ** | M = 6.38, SD = 2.56 b *** |
Note. Means that share subscripts do not differ significantly from each other.
a One point was given for a correct answer, -1 points for an incorrect answer, and 0 points for a neutral answer to the 10 questions about the eyewitness factors in the weak and strong case.
The I-I-Eye expert participants were hypothesized to know more about eyewitness factors in the cases than the control participants. Participants’ knowledge was evaluated for 10 eyewitness factors in the cases.1 The 10 eyewitness factors had a Cronbach alpha of .53 for the weak case.
Similar results were obtained for the strong case. The 10 eyewitness factors for the strong case had a Cronbach alpha of .57. There was a significant effect of expert type on participants’ knowledge, F(2, 111) = 11.13, p < .001, w = .40. Gabriel’s post hoc test revealed that both the I-I-Eye expert participants and the standard expert participants knew significantly (p < .001, p < .003) more about the 10 eyewitness factors in the strong case than the no expert participants (see Supplemental Table 2 on OSF). However, the I-I-Eye expert participants’ scores did not significantly differ from the scores of the standard expert participants (see Table 3 ).
It was predicted that the I-I-Eye expert participants would be significantly more likely than the control participants to render a guilty verdict in the strong eyewitness case and a not guilty verdict in the weak eyewitness case. The percentages of guilty verdicts for the strong and weak cases were: no expert, 44% for strong and 41% for weak; standard expert, 41% for strong and 41% for weak; and I-I-Eye expert, 68% for strong and 27% for weak.
Logistic regressions were conducted to predict the probability that a participant would render a guilty verdict. The logistic regressions were conducted so that all the expert types were compared to one another. Regression results indicated that the overall model of three predictors (case type, eyewitness condition, Case Type × Eyewitness Condition) was statistically significant in distinguishing the strong from the weak case (−2 log likelihood = 301.64), χ 2 (5) =13.29, p = .02. The model correctly classified 62.2% of the cases. When the I-I-Eye expert participants were compared to the no expert participants, the only significant predictor of guilty verdicts was the interaction of expert type and case strength, β = 1.62, exp (β) = 5.06, p = .02. These results indicate that the I-I-Eye participants were 5.06 more times to vote guilty in the strong than in the weak case compared to the no expert participants.
Likewise, when the I-I-Eye expert participants were compared to the standard expert participants, the only significant predictor was the interaction of expert type and eyewitness condition, β = 1.73, exp (β) = 5.63, p = .01. These results indicate that the I-I-Eye participants were 5.63 more times to vote guilty in the strong than in the weak case compared to the standard expert participants. However, when the standard expert participants were compared to the no expert participants, there were no significant predictors of guilty verdicts including the interaction of expert type and case strength, β = −0.11. exp (β) = 0.90, p = .87. Consequently, the standard expert participants were not more likely to vote guilty in the strong than in the weak case compared to the no expert participants (see Table 4 ).
Regression coefficients for logistic regression for verdicts.
| Expert condition | B | Wald | df | p | Odds ratio | 95% CI for odds ratio | |
|---|---|---|---|---|---|---|---|
| Lower | Upper | ||||||
| I-I-Eye expert vs. no expert | 1.622 | 5.59 | 1 | .018 | 5.064 | 1.32 | 19.43 |
| I-I-Eye expert vs. standard expert | 1.727 | 6.316 | 1 | .012 | 5.625 | 1.463 | 21.633 |
| Standard expert vs. no expert | −0.105 | .026 | 1 | .872 | .900 | .252 | 3.217 |
Note: CI = confidence interval.
Hypothesis III stated that the I-I-Eye participants could distinguish between the strength of the prosecution and the defense evidence in the strong and weak eyewitness case but not the no expert and standard expert participants. A factorial ANOVA showed that there was a significant interaction between expert type and eyewitness case for ratings of the strength of the prosecution evidence, F(2, 222) = 4.08, p = .018, η p 2 = .04. A simple effects analysis indicated that the I-I-Eye expert participants ratings of the strength of the prosecution evidence significantly differed in the strong and weak cases, F(1, 222) = 8.24, p = .004, r = .19. The standard expert participants’ ratings of the strength of the prosecution evidence did not significantly differ in the strong and weak cases, F(1, 222) = 0.27, p = .61, r = .03. The no expert participants’ ratings of the strength of the prosecution evidence also did not significantly differ in the strong and weak cases, F(1, 222) = 0.48, p = .49, r = − .05.
Next, the expert groups’ ratings of the defense evidence in the strong and weak eyewitness case were examined. A factorial ANOVA indicated that the interaction for expert type and eyewitness case for the strength of the defense evidence was non-significant, F(2, 223) = 1.07, p = .34, η p 2 = .01. There was also a non-significant main effect for expert type, F(2, 223) = 1.42, p = .24, η p 2 = .01. However, there was a significant effect for case strength, F(1, 223) = 12.65, p < .001, η p 2 = .05. The participants in the weak case rated the strength of the defense evidence significantly higher than the participants in the strong case (see Table 2 ).
It was predicted that the I-I-Eye expert participants, but not the control participants, could differentiate between the eyewitness interview, lineup and crime scene in the strong and weak cases. Accordingly, it was determined how expert type affected these evaluations. The alpha level was adjusted to p < .0166 to account for the multiple tests.
A factorial ANOVA indicated a significant interaction between case strength and expert type for the overall fairness of the eyewitness interview, F(2, 223) = 4.21, p = .016, η p 2 = .04. A simple effect analysis indicated that the I-I-Eye expert participants could distinguish between the eyewitness interview in the strong and the weak case, F(1, 223) = 30.45, p < .001, r = .35. The standard expert participants could also distinguish between the eyewitness factors in the strong and the weak case, F(1, 223) = 7.43, p = .007, r = .18. However, the no expert participants could not distinguish between the interviews in the strong and the weak case, F(1, 222) = 2.76, p = .098, r = .11 (see Table 2 ).
Next it was determined whether the participants could distinguish between the overall fairness of the lineup in the strong and the weak case. There was a non-significant interaction between the expert type and case strength for the overall fairness of the lineups, F(2, 222) = 2.85, p = .06, η p 2 = .03. There was also a non-significant main effect for expert type on the overall fairness of the lineups, F(2, 222) = 0.80, p = .45, η p 2 = .01. However, there was a significant main effect for eyewitness condition for the overall fairness of the lineup, F(1, 222) = 43.59, p < .001, η p 2 = .16. Accordingly, the participants in the strong case rated the overall fairness of the lineup significantly higher than the participants in the weak case (see Table 2 ).
Lastly, it was determined whether the participants could distinguish between the crime scene in the strong and weak cases. The main effect for expert type, F(2, 222) = 0.63, p = .53, η p 2 = .01, and the main effect for case strength for the crime scenes, F(1, 222) = 2.63, p = .11, were both non-significant. The interaction for expert type and case strength for the crime scenes was also non-significant, F(2, 222) = 2.47, p = .09, η p 2 = .01.
Although the interaction was non-significant, the power to detect an interaction for the crime scene was low (.49), and a line plot indicated an interaction. Simple effects for the interaction indicated that the I-I-Eye expert participants could distinguish between the crime scenes in the strong and weak cases, F(1, 222) = 4.77, p = .03, r = .14. The standard expert participants could not distinguish between the crime scene in the strong and weak cases, F(1, 222) = 1.68, p = .20, r = .09. The no expert participants also could not distinguish between the crime scenes in the strong and weak cases, F(1, 222) = 0.62, p = .43, r = .05. (See Table 2 .)
Interaction odds ratios indicated that the I-I-Eye expert participants were more than five times as likely to find the defendant guilty in the strong eyewitness case as in the weak eyewitness case compared to the standard expert and no expert participants. The I-I-Eye participants also could also distinguish between the strength of the prosecution evidence in the strong and weak cases. However, the verdicts of the standard expert participants did not significantly differ from the verdicts of the no expert participants in the strong and weak cases. The standard expert participants also could not distinguish between the strength of the prosecution evidence in the strong and weak cases. The standard expert participants could not make these distinctions even though the expert testimony in their cases was identical to the I-I-Eye expert testimony except it did not describe the I-I-Eye method. In contrast, the I-I-Eye expert participants could distinguish between the strong and weak cases, though the cases’ opening and closing arguments did not discuss the I-I-Eye method or how to apply it to the cases, the cases were not uniformly strong or weak, and they included a large number of diverse eyewitness factors.
Unlike the no expert participants, the I-I-Eye expert participants’ ratings of the strength of the prosecution evidence and their ratings of the interview differed in the strong and weak cases. The I-I-Eye expert participants also knew more about the eyewitness factors in the cases than the no expert participants. These differences may help explain why the I-I-Eye expert participants were more likely to vote guilty in the strong than in the weak case compared to the no expert participants.
However, both the I-I-Eye expert participants and the standard expert participants could distinguish between the interviews in the strong and weak cases, and their knowledge of the eyewitness factors in the cases did not significantly differ. In addition, the participants could distinguish between the lineups in the strong and weak cases. In retrospect, the lack of differences between the I-I-Eye expert participants and the standard expert participants in their ability to evaluate the eyewitness interview and lineup and their knowledge of eyewitness factors is not surprising. The expert testimony about the eyewitness factors in the cases, and how and why they likely affected eyewitness accuracy, was identical for the I-I-Eye expert participants and the standard expert participants.
So why did the standard expert participants’ ability to distinguish between the interviews and lineups in the strong and weak cases and their increased knowledge of eyewitness factors not affect their verdicts or enable them to distinguish between the strength of the prosecution evidence in the two cases? The standard expert participants may not have understood the importance of assessing whether the interview and lineup were properly conducted, and how the eyewitness factors at the crime scene likely affected eyewitness accuracy. Even if they had understood the importance of these assessments, they probably did not understand how to use these assessments when evaluating the likely accuracy of the eyewitness in the cases.
In contrast, the I-I-Eye method explicitly instructed participants to evaluate the interview, lineup and crime scene factors. The I-I-Eye method also explained how to use this information in assessing eyewitness accuracy. For example, it provided standards for evaluating the interview, lineup and crime scene factors, specified the order in which they should be evaluated and provided summary questions for making conclusions about the likely accuracy of the eyewitness testimony in the cases. Because the standard expert participants likely did not realize the importance of making these assessments or understand how to use them in evaluating eyewitness accuracy, they may have been unable to apply their increased knowledge of the eyewitness factors to the cases.
The present study did not determine whether it was the I-I-Eye method as a whole or only certain components of it that improved participants’ verdicts. Accordingly, I-I-Eye expert testimony may have been more effective in improving jurors’ verdicts simply because it, unlike the standard expert testimony, explicitly divided the eyewitness factors into three different types (i.e. eyewitness interview factors, identification procedures factors and crime scene factors).
However, this explanation for the greater efficacy of the I-I-Eye expert seems unlikely. In their closing arguments, both the prosecutor and defense attorney urged the participants to evaluate how the eyewitness interview and lineup were conducted. Both the I-I-Eye expert participants and the standard expert participants’ ratings of the interview and lineup indicated that they could distinguish between the strong and weak cases. Lastly, in the only two prior studies that the authors are aware of where the expert explicitly divided the eyewitness factors into different types, the expert did not improve participants’ verdicts (Jones et al., 2017; Leippe et al., 2004).
In short, it appears that the standard expert testimony was ineffective not because it failed to increase participants’ knowledge of eyewitness factors, but rather because it did not explain how to use that knowledge in assessing eyewitness accuracy. These results suggest that educating jurors about eyewitness factors may be insufficient to improve their evaluations of eyewitness accuracy because they may be unable to apply their knowledge to a case. In addition, the present study suggests that training for legal professionals should not only include education about eyewitness factors, but also instruction on how to apply eyewitness factors to a case.
The present study may also help explain why eyewitness expert testimony and eyewitness jury instructions have generally had at best limited effectiveness in prior studies (Devenport & Cutler, 2004; Devenport et al., 2002; Leippe, 1995; Leippe et al., 2004; Martire & Kemp, 2009). Expert testimony and jury instructions in prior studies did not provide participants with an analytical framework (i.e. a schema) for assessing how the eyewitness factors collectively affected eyewitness accuracy. At most, they explain how the eyewitness factors individually likely affected eyewitness accuracy. Accordingly, participants in these studies may not have known how to integrate the effects of the individual eyewitness factors to evaluate the collective effect that all the eyewitness factors in the case, both positive and negative, likely had on eyewitness accuracy. In contrast, the I-I-Eye method not only informs jurors how individual eyewitness factors likely affected accuracy, but it also provides them with a schema for assessing how the eyewitness factors collectively likely affected eyewitness accuracy. Of course, many other factors, such as the expert’s believability, credentials, likeability, confidence, and so on, influence the persuasiveness of expert testimony (Cramer, Brodsky, & DeCoster, 2009). Consequently, these factors also play an important role in determining the effectiveness of eyewitness expert testimony.
Lastly, future studies of experts should include eyewitness factors that affect the eyewitness interview. The eyewitness interview can have an important effect on eyewitness accuracy. The authors are aware of only one other study of eyewitness experts that examined mock jurors’ ability to evaluate how the eyewitness interview likely affected eyewitness accuracy (Buck, London, & Wright, 2011). Moreover, that study concerned a child eyewitness and did not include the other two types of eyewitness factors (i.e. identification procedures and crime scene factors).
A major problem that confronts the legal system in minimizing eyewitness error is that many law enforcement agencies fail to follow scientific procedures for conducting eyewitness interviews and identification procedures (Wise, Safer, & Maro, 2011). Moreover, even if all law enforcement agencies conducted proper identification procedures, eyewitness error may still frequently occur.
For instance, Wells, Steblay, and Dysart (2015) conducted a field study of 494 lineups in actual criminal cases that used the best procedures for conducting lineups. They found that among witnesses who made an identification, 41% of the eyewitnesses who viewed a simultaneous lineup identified a filler, and 32% of eyewitnesses who viewed a sequential lineup identified a filler. Wilford and Wells (2013) stated about the study: ‘…it is shocking that using all the best system-variable procedures still resulted in nearly one of every three actual eyewitnesses making a mistaken identification’ (p. 42).
In sum, what the legal system needs is a means to improve jurors’ and legal professionals’ ability to assess eyewitness accuracy. What appears critical in achieving this goal is that legal safeguards must not only explain how individual eyewitness factor affect accuracy, but also explain how to apply this knowledge to the facts of the case. Despite this need, there is limited research on how to improve eyewitness expert testimony and other legal safeguards. Eyewitness researchers should devote more time and effort to developing efficient, practical means of improving jurors’ and legal professionals’ ability to assess eyewitness accuracy. Achieving this goal appears essential to minimizing wrongful convictions from eyewitness error. The present study suggests that the efficacy of expert testimony can be improved so that it is an effective legal safeguard for eyewitness error.
There are some limitations to the present study. It is unknown whether the I-I-Eye method would be effective in an actual trial. However, the transcripts in the present cases contained all the elements of an actual trial. The present study used a transcript rather than a video of a trial; it used college students rather than community residents, and the participants did not deliberate prior to their verdicts. These factors may have affected the results (Wiener, Krauss, & Lieberman, 2011). College students tend to have a higher need for cognition than the general public, and therefore may have exerted greater cognitive effort to understand the I-I-Eye method than a community sample (McCabe, Krauss, & Lieberman, 2010). The I-I-Eye method may be easier to understand when it is presented in writing, as was done in the present study, rather than when it is presented aurally in a videotaped trial. It is unknown how jury deliberations would affect participants’ comprehension of the I-I-Eye method.
The I-I-Eye method also has some limitations. For instance, it does not inform jurors and legal professionals how to weight the various eyewitness factors (i.e. the magnitude of the effect sizes of the eyewitness factors). The I-I-Eye method does not include this information because there is limited research on weighting eyewitness factors. If there was sufficient research on these matters, this information could be incorporated into the I-I-Eye method. However, even if sufficient research existed, it is unknown whether the I-I-Eye method could effectively convey this information. Moreover, when effect sizes for eyewitness factors have been included in expert testimony and jury instructions, they have not helped participants assess eyewitness accuracy (Cutler et al., 1989; Jones et al., 2017). Despite these limitations, the studies of the I-I-Eye method suggest that it can improve jurors’ ability to assess eyewitness accuracy and that providing jurors with a schema for applying their knowledge of eyewitness factors to a case is necessary to improve jurors’ ability to assess eyewitness accuracy.
In the present study, the I-I-Eye method increased participants’ ability to differentiate the strong from the weak eyewitness case. On the other hand, standard expert testimony had no effect on participants’ verdicts even though it explained in detail the nature of memory, identified the relevant eyewitness factors, explained how and why they likely affected accuracy, and increased participants’ knowledge of the eyewitness factors. Lastly, the I-I-Eye method is inexpensive to use and appears relatively easy to learn. This latter characteristic is especially important because jurors are unlikely to comprehend a highly complex, abstruse method for analyzing eyewitness accuracy.
1 We intended to include 18 eyewitness factors in assessing how knowledgeable the expert groups were about eyewitness factors in the cases. However, 8 of the eyewitness factors were worded incorrectly for the strong and weak cases. Consequently, they were not included in the analysis.
Richard A. Wise has declared no conflicts of interest
Andre Kehn has declared no conflicts of interest
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee (The University of North Dakota Institutional Review Board) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent was obtained from all individual participants included in the study