Jennifer L. Short
University of Calgary
J. Thomas Dalby
University of Calgary
Abstract
Statement analysis has been used for years to determine the accuracy of statements. The Judgement of Memory Characteristics Questionnaire was revised in the current study to assess the accuracy of eyewitness identifications. Participants watched a video of a theft then identified the perpetrator from a line-up. Two statements were obtained: descriptions of the perpetrator and post-identification statements. The characteristics present in descriptions did not predict identification accuracy. However, analysis of the characteristics present in post-identification statements resulted in two predictive factors: Quality of Description and Amount of Detail. Statement analysis of post-identification statements resulted in a 70% correct classification rate of identifications. Further development of this measure and subsequent application of it to forensic investigations could help minimize the detrimental effects of inaccurate identifications.
Keywords: eyewitness identification, memory, accuracy, statement analysis
Eyewitness identification evidence in a criminal case may be perceived as overwhelming proof that a suspect committed the crime of which he/she is accused. When presented, these identifications are influential factors in the outcome of a case. Unfortunately, inaccurate eyewitness identifications have historically been found responsible for more wrongful convictions than all other factors combined (Huff, Rattner, & Sagarin, 1986; Wells et al., 1998). The number of exonerations in the United States that occurred when DNA analysis became available illustrates the poor accuracy of eyewitness identifications. Of the 181 convictions reported by the Innocence Project that were overturned based on DNA, more than 75% had been prosecuted based on inaccurate identifications (Innocence Project, n.d.). In one case, Kirk Bloodsworth was sentenced to death for rape and murder due to 5 eyewitness identifications and self-incriminating statements. The conviction was later overturned when DNA proved his innocence (Wells et al., 1998). Therefore, it is essential that the forensic community increasingly recognizes the fallibility of eyewitness memory and takes measures to minimize the consequences of inaccurate identifications.
Differences between Accurate and Inaccurate Identifications
Previous studies have examined the qualitative differences between accurate and inaccurate eyewitness identifications, including behavioural characteristics of witnesses and information provided by them. One consistent finding is that accurate witnesses report making more absolute judgments than inaccurate witnesses, indicating they immediately knew which photo depicted the perpetrator. Conversely, inaccurate witnesses tend to reach a decision regarding the identification through a process of eliminating the other photos (Dunning & Stern, 1994; Smith, Lindsay, & Pryke, 2000). As indicated by this finding, identification decisions made by accurate witnesses are affected by their own memory of the person while line-up photos affect inaccurate witnesses (Dunning & Stern, 1994).
In addition to the differences between accurate and inaccurate eyewitnesses in the process of identification, differences also exist in the characteristics of their memorial reports. Witnesses who make accurate identifications tend to report a clearer memory for the perpetrator’s face than inaccurate witnesses (Smith et al., 2000). Related to this clearer memory, accurate witnesses tend to report more distinct details when describing the perpetrator than inaccurate witnesses (Sporer, 1992). Finally, accurate witnesses are more confident in their identifications (Dunning & Stern, 1994; Smith et al., 2000; Sporer, 1992).
These reliable differences in the behaviours and memory reports of eyewitnesses may provide information that can be useful for predicting the accuracy of identifications. In one study, raters predicted the accuracy of identifications based upon various differences including type of identification (automatic vs. elimination), impact of photos, and impact of memory (Dunning & Stern, 1994). Raters informed of these differences were able to correctly classify the identifications as accurate or inaccurate 65.8% of the time. Rater training on these differences increased the rate of correctly classifying inaccurate identifications as compared to non-trained raters. However, this training did not change the correct classification rate of accurate identifications.
Statement Analysis
The ability to correctly assess the accuracy of identifications would provide an important investigative tool that could potentially decrease the rate of wrongful convictions based on eyewitness identifications. Although a measure for classifying identifications has not been developed thus far, similar procedures have been used for years to assess the veridicality of statements obtained in forensic settings. These statement analysis procedures are based upon the Undeutsch hypothesis, which indicates that statements based on experience differ both in content and quality from statements based on invention or fantasy (Vrij, Edward, Roberts, & Bull, 2000). Statement analysis involves examining the qualitative characteristics of statements and determining the veridicality of the statement based on the presence/absence of these characteristics. For example, accurate statements tend to contain more sensory information, time information, clarity/vividness, semantic information, spatial information, emotions/feelings, reconstructability of the story, and realism. Inaccurate statements tend to contain more cognitive operations (Sporer, 2004).
The first of these measures to be developed was Criteria-Based Content Analysis (CBCA), which was originally intended for use with children recalling sexual abuse (Steller & Koehnken, 1989). Despite the original intent, CBCA has been used and validated with adult populations. Another form of statement analysis, Reality Monitoring (RM), was later developed (see Masip, Sporer, Garrido, & Herrero, 2005). RM allowed for increased ease of use, required less training, and had better inter-rater reliability compared to CBCA. Most recently, a form of RM labelled the Judgment of Memory Characteristics Questionnaire (JMCQ; Sporer, 2004) was developed based on the original Memory Characteristics Questionnaire (Johnson, Foley, Suengas, & Raye, 1988). The JMCQ provides specific questions related to 43 items. For example, “are details described only superficially or very precisely?” (Sporer, 2004: p. 99). The statement being analysed will be rated on each of these items using a 7-point Likert scale. As it is the form of statement analysis that requires the least amount of training, the JMCQ may be the most practical version on which to train forensic investigators. However, all the incarnations of statement analysis have maintained the goal of veridicality assessment as determined by the examination of characteristics present in the statements.
The application of statement analysis to true and fabricated statements has resulted in varied but good correct classification rates. The use of CBCA in lab settings has resulted in correct classification of the veridicality of statements ranging from 65 to 73% (Masip et al., 2005). Overall correct classification of statements as accurate or inaccurate using RM range from 64 to 86% (Masip et al., 2005; Sporer, 2004). Although these measures do not allow for perfect accuracy when assessing the veridicality of statements, the correct classification rates indicate valid instruments that can provide important information during forensic investigations.
Similarly, the application of statement analysis to statements relevant to eyewitness identifications could provide insightful information for investigations. When a witness is questioned and later identifies a person presented in a line-up, two statements may be obtained that can be analysed. First, the witness provides a description of the person who committed the witnessed crime. Second, Canadian guidelines recommend that witnesses explain what characteristics or overall impressions led them to identify the person chosen (Brooks, 1983). The presence of previously discussed differences between accurate and inaccurate identifications (Dunning & Stern, 1994; Smith et al., 2000; Sporer, 1992) indicates that statement analysis procedures may be applicable to the assessment of identification accuracy. Currently developed procedures contain some items that would be appropriate for use with identification statements, such as quantity of visual details and clarity of the memory (Sporer, 2004). However, other characteristics that are assessed using statement analysis would not be present in statements related to the appearance of a person. For example, CBCA examines contextual embedding (Stellar & Koehnken, 1989) and the JMCQ examines temporal details such as day and season (Sporer, 2004). Therefore, adaptation of current statement analysis procedures may be a useful investigative tool for assessing the accuracy of eyewitness identifications.
Current Study
The current study was a preliminary investigation into the efficacy of statement analysis for determining the accuracy of eyewitness identification. The JMCQ was adapted to become appropriate for analysis of eyewitness descriptions and post-identification statements. Specifically, 8 of the original 43 JMCQ items were applicable to these statement types: clarity, colours, quantity of visual details, vividness, precision of details, thoughts, quality of remembering, and doubts about accuracy of remembering. In addition to the 8 original JMCQ items, statements were analysed on characteristics examined in previous eyewitness identification studies. The pre-identification confidence provided by the eyewitnesses was analysed with the description (Sporer, 1992). The post-identification statements were examined on type of identification (absolute vs. relative: Dunning & Stern, 1994; Smith et al., 2000). The types of characteristics (dynamic vs. static) portrayed in the descriptions and post-identification statements were also examined. This item is included due to the potential impact of relying upon dynamic, changing characteristics when the identifications may occur after a considerable time lapse in a forensic situation. Therefore, the primary goal of the current study was to determine if the development of a statement analysis procedure is appropriate for assessing the accuracy of eyewitness identifications.
Method
Participants
Participants were 113 undergraduate students at the University of Calgary who participated in exchange for .5 credits toward a psychology course of their choice. Seven participants were excluded because they did not choose a person from the photo line-up. A total of 106 participants completed the study.
Materials
Theft videotape. A 45 second video depicting a theft was shown to all participants on a 27-inch (68.58 centimetre) colour monitor. In the video, a man enters a hair salon and sits on a couch next to a woman reading a magazine. The woman forgets her purse on the couch when she leaves the scene. The man then looks around nervously and steals the purse. The video ends when the man leaves the scene with the woman’s purse.
Distractor task. Participants were provided with a 2-minute distractor task in which they looked for the 18 differences between two cartoon pictures. This task was presented to
allow a brief time lapse between the participant providing a description of the perpetrator and identifying the suspect in the line-up. A Find the Differences puzzle was chosen to focus the participants’ attention on contextual information irrelevant to the scene presented in the video.
Photographic line-up. Although current guidelines recommend the use of sequential line-ups (FPT Heads of Prosecutions Committee Working Group, 2004; Manitoba Justice, 2001), only 5 of 10 surveyed Canadian police agencies surveyed use this procedure. The remaining 5 police agencies continue to use simultaneous line-ups (FPT Heads of Prosecutions Committee Working Group, 2004). Due to only partial adoption of the guidelines regarding line-up type, participants were presented with the more traditional simultaneous photographic line-up depicting 8 men. Photos were 2.6 by 2 inch (6.8 by 5 centimetre) colour photographs labeled with the numbers 1 through 8. Photographs showed a front view of the men’s heads and necks. The perpetrator was consistently shown in position number 7. The remaining seven photographs showed foils that had similar characteristics as the perpetrator. For example, foils and the perpetrator all had brown hair, had similar hairstyles, were Caucasian, and were of a similar age.
Procedure
Participants viewed the theft videotape at the beginning of the test session. Immediately after watching the video, participants were interviewed by the first author. This interview began by the participants describing in detail the person who stole the purse. Following the free narrative description, mental images described by participants were probed until their recall of that image was exhausted (Fisher & Geiselman, 1992). Each mental image described in the free narrative was similarly probed. If participants did not describe anything about the perpetrators hair, build, or face, the interviewer asked if they had noticed anything about that feature. All questions and probes were worded in a neutral manner that did not introduce misinformation and were not misleading. Participants were then asked how confident they were that they could accurately identify the perpetrator if asked to do so. As recommended in the Thomas Sophonow inquiry (Manitoba Justice, 2001), this description was audio-taped and later transcribed for analysis.
Following this interview, participants were provided with a distractor task in which they were instructed to look for the 18 differences between two cartoon pictures. The interviewer left the room during the two minutes provided for the Find the Differences puzzle. After the two minutes allotted for the distractor task had elapsed, the interviewer returned with a photo line-up. Participants were instructed to take their time to think back to the scene they had watched on the videotape, carefully look at each of the photographs shown in the line-up, and let the interviewer know when they were ready to identify the purse thief. Participants who asked if the perpetrator was in the line-up or indicated they did not think he was in the line-up were told they could respond that the perpetrator was not present.
Consistent with Canadian guidelines (Brooks, 1983), participants who made identifications were asked what characteristics or overall impressions led them to choose the person indicated. This question and the interviewer’s response to identifications was worded to minimize post-identification feedback to participants, responding to the identification as follows:
“Can you tell me what characteristics or overall impressions led you to choose number X”
As with the initial description, each mental image described in the post-identification statement was probed for further details (Fisher & Geiselman, 1992). If participants responded vaguely by naming a feature (for example, the hair), the interviewer asked what about that feature had led them to choose him. This post-identification statement was audio-taped (Manitoba Justice, 2001) and later transcribed for analysis.
Through discussion with the researcher, two independent raters were trained on the items present in the statement analysis used when analysing both statement types. They rated a practice description and post-interview statement on each of the relevant items using a 7-point Likert scale. Both types of statements were rated on clarity, colours, quantity of visual details, vividness, precision of details, thoughts, quality of remembering, doubts about the accuracy of remembering, and description of features. Descriptions were also rated on pre-identification confidence. Post-identification statements were also rated on recognition type. Discrepancies were resolved through discussion. Any confusion about the items following the practice ratings was also resolved. Following this practice and discussion, both raters expressed confidence in their ability to accurately rate the statements. Each rater independently analysed the statements provided by participants. After completing the ratings, discrepancies were resolved by the two raters through discussion.
Results
Inter-rater Reliability
Inter-rater reliability was determined by correlating the ratings assigned by the two raters on each item. The confidence provided by each participant was excluded from the analysis because that rating was a direct conversion from the percentage provided to a rating on the 7-point Likert scale. The remaining 19 items were included. The analysis revealed moderate inter-rater reliability, r = .578, p < .001.
Accurate Identification Rate
Of the 106 participants, 34 (32%) correctly identified the perpetrator in the photo line-up.
Differences between Accurate and Inaccurate Identifications
Statement ratings of accurate and inaccurate statements were compared for each of the 20 items analysed. The eyewitnesses’ thoughts were described more in the description for accurate identifications (M = 3.00) than inaccurate identifications (M = 2.25), t (104) = 2.04, p = .044. There was a marginally significant difference in the eyewitnesses’ confidence in their ability to identify the suspect. Higher confidence ratings were present with an accurate identification (M = 5.24) than when the identification was inaccurate (M = 4.91), t (88.55) = 1.97, p = .052. The following items were not significantly different for accurate and inaccurate identifications in the description: clarity, description of colours, quantity of visual details, vividness, precision of details, quality of remembering, doubts about accuracy of remembering, and description of features.
The post-identification statement was rated higher on clarity when the identifications were accurate (M = 4.24) than when they were inaccurate (M = 3.82), t (104) = 2.03, p = .045. Similarly, the quality of remembering in the post-identification statement was rated higher for accurate identifications (M = 4.00) than inaccurate ones (M = 3.54), t (104) = 2.65, p = .009. The following items were not significantly different for accurate and inaccurate identifications in the post-identification statement: description of colours, quantity of visual details, vividness, precision of details, thoughts, doubts about accuracy of remembering, description of features, and type of recognition.
Prediction of Identification Accuracy
Common factor analyses using varimax rotations were conducted on the items that were used to analyse the statements. These determined the subscales that could be used to assess the accuracy of eyewitness identifications, as shown in Table 1.
Factor analyses were conducted on: 1) the description items; 2) the post-identification statement items; 3) all the items used to assess both the descriptions and post-identification statements. Scores were then calculated for each of the extracted factors by averaging the scores of the contributing items. Logistic regression was conducted on factor scores to determine the extent to which the factors obtained could correctly classify the identifications as accurate or inaccurate. Ideally, both types of statements would be obtained in an investigation. However, analysing the items for each statement type separately provides information about the utility of these items when only one statement is available for assessment.
Description. Using an eigenvalue cut-off point of 1.0, factor analysis of the 10 items used to analyse the descriptions revealed 3 factors: Quality of Description, Confidence, and Description of Colours. The communality cut-off point for inclusion of items in a factor was .30. Therefore, 6 items loaded substantively on Quality of Description: clarity, quantity of visual details, vividness, precision of details, quality of remembering, and description of features. This factor accounts for 38% of the variability in the original 10 items. The second factor, Confidence, had 2 items load substantively: doubts about accuracy of remembering and confidence. The Confidence factor accounts for 14% of the variability in the items. Finally, one item loaded substantively on the Description of Colours factor: colours. This factor accounts for 12% of the variability in the original items. Overall, the three factors extracted for analysis of the description account for 64% of the variability in identification accuracy. There is not a significant amount of unaccounted for variance remaining with this solution, χ2 (18) = 13.47, p = .76.
When the three factors extracted from the description items are used to analyse the description, the overall rate of correctly classifying identifications as accurate or inaccurate was predicted at 68%. The predicted correct classification rate of accurate identifications was 0% and was 100% for inaccurate identifications. Therefore, all identifications would be classified as inaccurate when analysing the description using Quality of Description, Confidence, and Colours. None of these subscales are significant predictors of identification accuracy, zs < 1.00, p > .05.
Post-Identification Statement. Factor analysis of the items used to analyse the postidentification statement revealed three factors: Quality of Description, Amount of Detail, and Confidence. Six items loaded substantively onto Quality of Description: clarity, quantity of visual detail, vividness, precision of detail, thoughts, and quality of remembering. This first factor accounted for 28% of the variability in the original 10 items.
The second factor, Amount of Detail, was composed of three of the original items: quantity of detail, vividness, and precision of detail. This factor accounted for 15% of the items analysed. Two items loaded substantively onto the Confidence factor: doubts about accuracy of remembering and recognition type. Confidence accounts for 14% of the original items. Overall, the three factors extracted in this factor analysis accounts for 57% of the variability in identification accuracy. There is not a significant amount of unaccounted for variance remaining with this solution, χ2 (18) = 19.70, p = .35.
When analysing the post-identification statements using the three factors extracted, an overall correct classification rate of 70% is predicted. Unlike the predictions made when analysing the description, these factors can be used to correctly classify both accurate and inaccurate identifications. The logistic regression predicts that accurate identifications will be correctly classified 15% of the time and the correct classification rate of inaccurate ones is 96%. Quality of Description significantly predicts the accuracy of identifications, z = 2.37, p = .018. With every one-point increase of this factor on the Likert scale, it is 5.8 times more likely that the identification is inaccurate. Therefore, eyewitnesses who make inaccurate identifications will also provide post-identification statements that received a higher rating on Quality of Description. The predictive ability of Amount of Detail is marginally significant, z = 1.92, p = .055. With every one-point increase on this factor, it is 3.0 times more likely that the identification is accurate. The final factor, Confidence, does not significantly predict accuracy, z = .74, p > .05. Therefore, a high rating on Amount of Detail and a low rating on Quality of Description will likely characterize the post-identification statement accompanying an accurate identification.
Description and Post-Identification Statement. Six factors were extracted when the items used to analyse the descriptions and post-identification statements were analysed together: Overall Quality, Quality of the Explanation, Identification Confidence, Description Confidence, Description of Thoughts, and Description of Features. Ten items loaded substantively onto Overall Quality, eight of which are used to analyse the description: clarity, colours, quantity of visual details, vividness, precision of details, quality of remembering, doubts about accuracy of remembering, and description of features. The remaining two items that loaded onto this factor are used to analyse the post-identification statements: clarity and thoughts. Overall Quality accounts for 20% of the variability in the original 20 items. All seven of the items that loaded substantively onto the second factor, Quality of Explanation, are used to analyse the post-identification statements: clarity, colours, quantity of visual details, vividness, precision of details, thoughts, and quality of remembering. This factor accounts for 19% of the variability in the items. The Identification Confidence factor has four items load onto it, all of which analyse the post-identification statement: clarity, quality of remembering, doubts about accuracy of remembering, and recognition type. Identification Confidence accounts for 7% of the item variability. Description Confidence has two items loading onto it, both of which are used to analyse the description: doubts about accuracy of remembering and confidence. This factor accounts for 6% of the variability in the original items. The fifth factor, Description of Thoughts, has the thoughts items for both the descriptions and post-identification statements loading onto it. This accounts for 6% of the variability. Finally, the description of features item from both statements load onto the Description of Features factor. This factor accounts for 6% of the variability in the original items. Overall, the six factors extracted account for 64% of the variability in the 20 items analysed. There is not a significant amount of unaccounted for variance remaining with this solution, χ2 (85) = 66.20, p = .935.
Analysing both the description and the post-identification statement using the 6 factors extracted from the original 20 items results in an overall correct classification of 70% predicted. The correct classification rate is predicted to be 21% for accurate identifications and 93% for inaccurate ones. However, none of the subscales significantly predict the accuracy of identifications, zs < 1.50, p > .05.
Discussion
The correct identification rate obtained in the current study was particularly low (32%), potentially because of the temporal arrangement of the tasks. According to the verbal overshadowing literature (see Schooler, 2002), providing a description of the perpetrator would have activated verbal processes for the participants. In turn, the nonverbal processes required for accurate identification may have been inhibited, resulting in poor accuracy when identifying the perpetrator. This verbal overshadowing effect has previously been found to affect facial recognition (Clare & Lewandowsky, 2004; Schooler & Engstler-Schooler, 1990) and may have had a similar detrimental effect in the current study.
Although prior research has found that certain factors are predictive of identification accuracy (Smith et al., 2000; Sporer, 1992), an instrument had not been developed for this purpose. The present study attempted to revise a previously developed statement analysis procedure, the JMCQ (Sporer, 2004), for use with statements relevant to eyewitness identification. The eight appropriate JMCQ items and two additional items related to the descriptions formed three factors potentially useful for determining the accuracy of identifications: Quality of Description, Confidence, and Description of Colours. Unfortunately, none of these three factors were predictive of identification accuracy. Using them to classify identifications would likely result in all of them being classified as inaccurate. Although incorrect eyewitness identification may occur often, as evidenced by a 32% correct identification rate in the current study, they are certainly not all inaccurate. Therefore, the revised JMCQ developed for analysis of descriptions is not useful for predicting the accuracy of identifications.
These results are consistent with previous research that failed to find a relationship between descriptions and identification accuracy. For example, the accuracy of a perpetrator description is not related to the accuracy of later line-up identification (Pigott & Brigham, 1985; Yarmey, 2004). In addition, eyewitnesses are often inaccurate in their descriptions. Yarmey (2004) found that witnesses were very accurate in their recall of a person’s hairstyle and age with 92 and 97% correct recall respectively. However, they were quite inaccurate at recalling six other physical characteristics, such as eye colour and height. The correct recall rate for these characteristics ranged from 21 to 64%. Dependence on an initial description is often necessary for narrowing down possible suspects. The combination of these studies indicates that the relationship between descriptions and identifications should be interpreted with caution.
Analysis of the eight JMCQ items and two additional items used to analyse the post-identification statement revealed three factors: Quality of Description, Amount of Detail, and Confidence. The first two factors listed are predictive of identification accuracy. An accurate identification will likely be accompanied by a post-identification statement with a low rating on Quality of Description and a high rating on Amount of Detail. Application of these factor scores will result in correctly classifying identifications as accurate or inaccurate approximately 70% of the time. Inaccurate statements will be correctly classified most of the time (96%) while accurate statements will often be misclassified. This indicates that the revised JMCQ, the Judgement of Memory Characteristics Questionnaire – Eyewitness (JMCQ-E), has high positive predictive power. False negative classifications will result more often than false positives. Therefore, classification of identifications as accurate would be particularly strong evidence for their accuracy. Classification of identifications as inaccurate would necessitate the presence of strong corroborating evidence separate from eyewitness memory.
The eyewitness identification procedures recommended in Canada suggest that post-identification statements be obtained from witnesses, in which they indicate what led to their line-up choice (Brooks, 1983). However, the standards developed in the United States do not make this recommendation (Technical Working Group for Eyewitness Evidence, 1990). The potential benefits of analysing post-identification statements to determine identification accuracy suggests law enforcement agencies should endeavour to obtain these statements. They will provide additional evidence that will not only be a self-report measure but can be analysed systematically to provide insight into the identification itself.
The benefits of analysing the post-identification statement disappear when the description is assessed concurrently. Analysis of the 10 description items and 10 post-identification statement items together result in six factors: Overall Quality, Quality of Explanation, Identification Confidence, Description Confidence, Description of Thoughts, and Description of Features. However, none of these factors predict the accuracy of identifications. Therefore, prediction of identification accuracy is impaired by analysis of descriptions both alone and in combination with the post-identification statement.
Overall, the results of this study indicate that statement analysis can be beneficial in investigations relying on eyewitness identification. Further development of this procedure, the JMCQ-E, can help investigators detect reliable identifications. The classification of identifications as accurate, in turn, may increase their influence on investigations due to the positive predictive power of the procedure. Statement analysis would also indicate the necessity of more non-eyewitness reliant evidence when identifications are classified as inaccurate. The evidence provided by the JMCQ-E would likely not be admissible in court because it goes to the question of guilt or innocence. However, the ability to assess the reliability of evidence will strengthen the investigations and potentially minimize the consequences of incorrect identifications.
The JMCQ-E is predictive of identification accuracy/inaccuracy. However, the overall correct classification rate of 70% indicates room for improvement. Addition of other potentially predictive items may improve the rate at which identifications are correctly classified using this measure. For example, the U.S. Supreme Court has recommended that eyewitness identifications be evaluated based on five criteria: (1) certainty in the accuracy of the identification; (2) quality of the eyewitness’ view of the perpetrator; (3) amount of attention the eyewitness paid to the perpetrator; (4) how well the description matches the suspect’s appearance; and (5) the amount of time elapsed between witnessing the event and providing an identification (Bradfield & Wells, 2000). Some potentially useful factors indicated by previous research are the time witnesses take to make the identification and post-identification confidence (Smith et al., 2000; Sporer, 1992). Addition of these factors to the JMCQ-E may increase the predictive utility of the measure when evaluating both descriptions and post-identification statements.
Development of the JMCQ-E in a lab setting requires that this measure be tested using statements obtained in a forensic setting. An instrument valid for classification of identifications in a lab setting is not necessarily valid in a forensic situation due to the different environments in which the scene is witnessed and the statements obtained. For example, the witness’s attention may not be as focused on the details of interest in a natural setting compared to a lab. Also, a more emotional response may result when a person witnesses an event in person rather than on a video. These disparities may result in differences encoding and recalling information about the experience. Therefore, the effects of environmental influences in a natural setting on the utility of the JMCQ-E need to be examined.
In a forensic setting, eyewitnesses may make a number of choices when presented with a line-up. The current study investigated the possibility of classifying two types of identifications using statement analysis: inaccurate and accurate. Inaccurate identifications were those in which the participant incorrectly identified one of the seven foils present in
the line-up. Accurate identifications were those in which the participant correctly chose the perpetrator from the line-up. However, eyewitnesses may choose not to identify any of the people presented in the line-up. Such non-identifications can be accurate (the perpetrator is not present in the line-up) or inaccurate (the perpetrator is present in the line-up). As non-identifications are also likely to occur in forensic situations, it is important to examine the utility of statement analysis when classifying these as accurate or inaccurate. Therefore, future research should be directed at validating the JMCQ-E when all four identification types are analysed: accurate identifications, inaccurate identifications, accurate nonidentifications, and inaccurate non-identifications.
Another concern about the utility of the JMCQ-E relates to line-up type. The presentation of photo line-ups can occur in two ways: simultaneous or sequential. The current study involved the use a simultaneous line-up. However, a shift to sequential lineups is occurring in some law enforcement agencies (Gronlund, 2004; Yarmey, 2003). This change is an attempt to improve accuracy of identifications by increasing the eyewitness’s dependence on absolute decision processes (Gronlund, 2004; Kneller, Memon, Stevenage, 2001). In addition to this shift in decision processes, differential presentation of photos in line-up procedures may affect other features of statements related to identification. For example, the use of an absolute decision strategy may result in the expression of more doubts about the accuracy of remembering. Therefore, the use of sequential line-ups could affect the validity of the JMCQ-E in assessing the accuracy of the identifications. Future research should explore the efficacy of this measure when used in concert with sequential line-ups.
In conclusion, the analysis of the JMCQ-E indicates that statement analysis is useful for assessing the accuracy of eyewitness identifications. However, this utility is only present when post-identification statements are analysed. As the current study is the first one of its kind, future research is necessary to ensure the measure continues to be valid when assessing statements obtained in forensic settings and when identifications are made using sequential line-ups. Development and use of the JMCQ-E in police investigations could provide insight into the reliability of eyewitness identifications obtained. Thus, this measure may be useful in minimizing the detrimental effects of inaccurate eyewitness identifications.
References
Bradfield, A. L., & Wells, G. L. (2000). The perceived validity of eyewitness identification testimony: A tests of the five Biggers criteria. Law and Human Behavior, 24, 581-594.
Brooks, N. (1983). Police guidelines: Pretrial eyewitness identification procedures. Ottawa: Law Reform Commission of Canada.
Clare, J., & Lewandowsky, S. (2004). Verbalizing facial memory: Criterion effects in verbal overshadowing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30, 739-755.
Dunning, D., & Stern, L. B. (1994). Distinguishing accurate from inaccurate eyewitness identifications via inquiries about decision processes. Journal of Personality and Social Psychology, 67, 818-835.
Fisher, R. P., & Geiselman, R. E. (1994). Memory-enhancing techniques for investigative interviewing: The cognitive interview. Springfield, Ill.: Charles C. Thomas.
FPT Heads of Prosecutions Committee Working Group (2004, September). Report on the Prevention of miscarriages of Justice. Retrieved July 10, 2006, from http://www.justice.gc.ca/en/dept/pub/hop/PreventionOfMiscarriagesOfJustice.pdf.
Gronlund, S. D. (2004). Sequential lineups: Shift in criterion or decision strategy? Journal of Applied Psychology, 89, 362-368.
Huff, C. R., Rattner, A., & Sagarin, E. (1986). Guilty until proven innocent: Wrongful conviction and public policy. Crime and Delinquency, 32, 518-544.
Innocence Project (n.d.). Facts on post-conviction DNA exonerations. Retrieved July 10, 2006, from http://www.innocenceproject.org/docs/DNAExonerationFacts_WEB.pdf.
Johnson, M. K., Foley, M. A., Suengas, A. G., & Raye, C. L. (1988). Phenomenological characteristics of memories for perceived and imagined autobiographical events. Journal of Experimental Psychology: General, 117, 371-376.
Kneller, W., Memon, A., & Stevenage, S. (2001). Simultaneous and sequential lineups: Decision processes of accurate and inaccurate eyewitnesses. Applied Cognitive Psychology, 15, 659-671.
Manitoba Justice (2001). Thomas Sophonow inquiry report. Retrieved July 10, 2006, from http://www.gov.mb.ca/justice/publications/sophonow.
Masip, J., Sporer, S. L., Garrido, E., & Herrero, C. (2005). The detection of deception with the reality monitoring approach: A review of the empirical evidence. Psychology, Crime, & Law, 11, 99-122.
Pigott, M., & Brigham, J. C. (1985). Relationship between accuracy of prior description and facial recognition. Journal of Applied Psychology, 70, 547-555.
Schooler, J. W. (2002). Verbalization produces a transfer inappropriate processing shift. Applied Cognitive Psychology, 16, 989-997.
Schooler, J. W., & Engstler-Schooler, T.Y. (1990). Verbal overshadowing of visual memories: Some things are better left unsaid. Cognitive Psychology, 17, 36-71.
Smith, S. M., Lindsay, R. C. L., & Pryke, S. (2000). Postdictors of eyewitness errors: Can false identifications be diagnosed? Journal of Applied Psychology, 85, 542-550.
Sporer, S. L. (1992). Post-dicting eyewitness accuracy: Confidence, decision-times and person descriptions of choosers and non-choosers. European Journal of Social Psychology, 22, 157-180.
Sporer, S. L. (2004). Reality monitoring and deception detection. In P. -A. Granhag & L. Stromwall (Eds.), Deception detection in forensic contexts (pp. 64-102). Cambridge: Cambridge University Press.
Steller, M., & Koehnken, G. (1989). Criteria-based content analysis. In J. C. Yuille (Ed.), Psychological methods in clinical investigation and evidence (pp. 217-245). New York: Springer.
Technical Working Group for Eyewitness Evidence (1999). Eyewitness evidence: A guide for law enforcement. Washington, D.C.: U.S. Department of Justice.
Vrij, A., Edward, K., Roberts, K. P., & Bull, R. (2000). Detecting deceit via analysis of verbal and nonverbal behavior. Journal of Nonverbal Behavior, 24, 239-263.
Wells, G. L., Small, M., Penrod, S. Malpass, R. S., Fulero, S. M., & Brimacombe, C. A. E. (1998). Eyewitness identification procedures: Recommendations for lineups and photospreads. Law and Human Behavior, 22, 603-647.
Yarmey, A. D. (2003). Eyewitness identification: Guidelines and recommendations for identification procedures in the United States and in Canada. Canadian Psychology, 44, 181-189.
Yarmey, A. D. (2004). Eyewitness recall and photo identification: A field experiment. Psychology: Crime & Law, 10, 53-68.
Biographical Notes
Correspondence to: Jennifer Short
39, 9912 – 106 St.
Edmonton, AB T5K 1C5
CANADA
phone: 780-428-9223
fax: 780-428-7061
e-mail: jlcolton@ucalgary.ca
J. Thomas Dalby is a forensic psychologist and neuropsychologist and is Adjunct Professor (Psychology and Clinical Psychology) and Adjunct Associate Professor (Psychiatry) at the University of Calgary. He has over 100 professional publications including Applications of Psychology in the Law Practice (American Bar Association 1997) and Mental Disease in History (Peter Lang Publishers, 1997).
Jennifer L. Short has conducted research in the areas of false memories, eyewitness identification, and deception detection. She is currently working in private practice, focusing on forensic and clinical psychology.