Hit & False Alarm Rates across Sequential, Simultaneous, and Novel Sequential-Presentation-Only Line-Ups

A University of Melbourne APA Lab Report Assignment

Under the 2nd-Year Sociology subject "Cognitive Psychology"

Passed with High Distinction (H1)

============ ============ ============

By Benjamin L., written during Semester 2, 2012

Abstract

Erroneous identifications in simultaneous line-ups have resulted in wrongful convictions. Compared to simultaneous line-ups, researchers generally concur that sequential line-ups are superior because they reduce false identifications more than they reduce hits. This study investigated if a novel sequential-presentation-only (SPO) line-up could combine the advantages of both line-ups and hypothesised that (a) a sequential line-up will produce fewer hits and false alarms than a simultaneous line-up and, (b) a sequential-presentation-only line-up will produce fewer false alarms than a simultaneous line-up without significantly reducing hits. 620 participants watched a video of a suspicious man and attempted to identify him in one of the three line-ups. Results did not support either hypothesis. After examining potential moderating factors, this study suggests that certain procedural manipulations can make simultaneous line-ups superior to sequential ones and that the SPO line-up is not a superior alternative to either.

  

Hit & False Alarm Rates across Sequential, Simultaneous, and Novel Sequential-Presentation-Only Line-Ups

            Criminal identification typically involves a simultaneous line-up where the suspect (target) is placed among a group of innocent others (foils) who resemble the suspect (Wells & Olsen, 2003). Witnesses either identify the suspect from the group, reject the whole line-up, or declare that they are unsure. A positive identification is interpreted as evidence of the suspect’s guilt (Dysart & Lindsay, 2001). However, more accurate line-ups are needed as this has wrongfully incriminated many innocent suspects (Dillon, McAllister, & Vernon, 2009).

            Sequential line-ups could be a better alternative. It presents a line-up of photos in sequence and witnesses must decide whether one photo matches the suspect before viewing the next (Palmer & Brewer, 2011)). Each photo should only be shown once, witnesses should not know how many photos they will view, and must not revise earlier decisions. Line-ups can involve either target-absent trials, where the target is present in the line-up, or target-absent trials, where only foils are present. Sequential line-ups are superior because of the “sequential-superiority effect” (Steblay, Dysart, Fulero, & Lindsay, 2001, p.460) that enhances overall accuracy by reducing false identifications (false alarms) in target-absent conditions far more than it reduces correct identifications (hits) in target-present conditions.

            This stems from witnesses’ adoption of different decision strategies (Gronlund, 2004). In simultaneous line-ups, witnesses use a relative judgment strategy that performs a side-by-side comparison and selects the foil that bears the closest resemblance to their memory of the perpetrator, even if the resemblance is vague (Steblay & Phillips, 2011). In target-present conditions, the closest match is the perpetrator and results in more hits (Steblay et al., 2001). In target-absent conditions, the closest-resembling foil is picked and creates more false alarms. In contrast, sequential line-ups require absolute judgments where witnesses compare each photo exclusively to memory (Steblay & Phillips, 2011). In target-absent conditions, this creates fewer false alarms because foils are unlikely to match the witness’s memory (Steblay et al., 2001). However, in target-present conditions, sequential line-ups produced lower hit rates compared to simultaneous line-ups.

            This reduced hit rate could be explained through signal detection theory (SDT) (Meissner, Tredoux, Parker, & MacLin, 2005). SDT holds that our ability to recognise and distinguish between familiar and novel stimuli rests on our response criterion and discrimination accuracy. Response criterion is the strength of evidence required before a signal (hit) is registered, while discrimination accuracy is an individual’s ability to discriminate between hits and false alarms. In line-ups, response criterions are based on familiarity: the awareness that a stimulus was previously encountered even if one cannot remember its details (Medina, 2008). A line-up photo that exceeds a familiarity threshold and matches the witness’s memory of the perpetrator produces a hit (Yonelinas, 2001), otherwise it is rejected (Gronlund, 2004).

            However, sequential line-ups typically conceal the number of photos that will be shown (McQuiston-Surret, Malpass, & Tredoux, 2006). This raises the criterion threshold and makes it harder for a photo to qualify as a hit, thus creating a higher chance of missing the target photo and reducing the hit rate. Also, since witnesses cannot retract their decisions if a better photo appears, they become reluctant to make a positive decision and opt for a conservative ‘unsure’ response instead that reduces the hit and false alarm rate (Meissner et al., 2005).

            The ideal line-up would optimise discrimination accuracy by maximising hit rates while minimising false alarms. If a modified sequential line-up could still induce absolute judgments without a criterion shift, we could achieve lower false alarm rates without a reduction in hits. This study examines if a novel ‘sequential-presentation-only’ (SPO) line-up can do so. In this line-up, witnesses are withhold their response until all the photos have been seen in sequence. The sequentially-presented format forces witnesses to make absolute judgments and retain the lower false alarm rates of sequential line-ups. In addition, since witnesses will have seen all the candidates before making a response, their response criterion and conservativeness should not be raised. This should produce hit rates comparable to those of a simultaneous line-up.

            Therefore this study hypothesises that, in target-present manipulations, (a) a sequential line-up will produce fewer hits and false alarms than a simultaneous line-up and (b) a sequential-presentation-only line-up will produce fewer false alarms than a simultaneous line-up without a significant reduction in hits.

Method

Participants

            Participants consisted of 620 sophomore undergraduate students. Their participation was part of their psychology subject.

Materials

            The perpetrator-target was shown in a 43-second video. It was recorded from a first-person perspective where the witness looks out an office window and sees the target holding a suspicious bomb-like object. The target looks at the witness, runs down an adjacent corridor, and looks at the witness again for approximately 500msec. Presenting the target twice improves encoding (Meissner et al., 2005) and minimises the risk of confounding encoding deficits with the retrieval effects of each line-up.

            The line-ups were presented with a PowerPoint presentation consisting of 5 numbered, uncoloured mug-shot photos: one of the target (hence target-present) and 4 similar foils. All five are male, bald, young, have fair skin-tone without facial hair or accessories, and wore a black t-shirt. If participants believed that the target was present, they were to write his number on a response sheet or “NONE” if absent. ‘Unsure’ responses were omitted: they suggest that the perpetrator is not positively present and imply a “NONE” response (Steblay & Phillips, 2011).

Procedure

            Participants viewed the video before class and were surreptitiously asked to record their emotional state every five seconds. Once it ended, they were told that they have witnessed a potential crime and must see if the perpetrator is in a line-up. In accordance with standard police practice (Wells & Olsen, 2003), participants were instructed that the suspect may or may not be present.

            Classes were randomly assigned to each line-up. In the simultaneous condition, all 5 photos were concurrently presented and participants made a response. In the sequential condition, participants viewed and decided on one photo at a time. In the SPO condition, the photos were presented sequentially and participants were told to withhold their responses until all the photos were seen.

Design & Analysis

            An independent groups design was used. Two separate one-way ANOVAs were used to compare hit and false alarm rates as a function of line-up condition. Each hypothesis was tested with post hoc analyses using the Bonferroni correction for multiple comparisons.

Results

            Figure 1 summarises the mean hit and false alarm rates as a function of line-up condition.

Figure 1. Mean hit rates and false alarms as a function of line-up condition

            A one-way ANOVA revealed a significant difference in hit rates between line-up conditions (F(2, 617) = 9.49, p < .001, η² = .03). According to Cohen (1988), this demonstrates a small effect of line-up condition on hit rates. A post hoc analysis using the Bonferroni correction for multiple comparisons showed a significant difference the SPO and simultaneous line-ups (Mean Difference = .10, p < .001), but not for the SPO and sequential line-ups (Mean Difference = .05, p = .104) or the sequential and simultaneous line-ups (Mean Difference = .05, p = .051).

            Another one-way ANOVA also revealed a significant different in false alarm rates between line-up conditions (F(2, 617) = 6.82, p = .001, η² = .02). Cohen (1988) suggests that this demonstrates a small effect of line-up procedure on false alarm rates. A post hoc analysis with the Bonferroni correction for multiple comparisons showed a significant difference between the simultaneous and sequential line-ups (Mean Difference = .09, p = .001), but not for the sequential and SPO line-ups (Mean Difference = .03, p = .485) or the SPO and simultaneous line-ups (Mean Difference = .05, p = .090).

Discussion

            This study investigated if the SPO line-up could achieve optimal discrimination accuracy. It was hypothesised that in target-present manipulations, (a) a sequential line-up will produce fewer hits and false alarms than a simultaneous line-up and (b) a sequential-presentation-only line-up will produce fewer false alarms than a simultaneous line-up without a significant reduction in hits. Surprisingly, neither hypothesis was supported. Results showed that the simultaneous condition produced fewer false alarms than the sequential condition while hit rates were comparable. This is not consistent with previous research (Steblay et al., 2001). As for the SPO line-up, it produced fewer hits than the simultaneous condition while false alarm rates were comparable.

Sequential and Simultaneous Line-Ups

            The decision strategies framework held that simultaneous line-ups encourage participants/witnesses to use relative judgments (Gronlund, 2004). In target-present line-ups like ours, the target photo should bear the greatest resemblance to witnesses’ memory of the perpetrator and produce a high hit rate. However, our results do not support this. Given that our line-up photos were very similar, it is possible that the target did not have a sufficient relative advantage in resemblance over the foils to produce the higher hit rates we expected.

            The lower false alarm rate in the simultaneous condition relative to the sequential condition was also unexpected. Sequential line-ups induce a criterion shift and make participants give conservative ‘unsure’ (“NONE”) responses that neither identify nor reject a photo, thus reducing both hit and false alarm rates (Meissner et al., 2005). However, results showed that the simultaneous line-up produced significantly fewer false alarms instead. This could be due to the instructions used in the present study which caution that the target may or may not be present. Such instructions discourages guessing when participants are uncertain (Steblay et al., 2001). Since simultaneous line-ups do not entail a criterion shift, participants are relatively less conservative and more likely to guess who the perpetrator is (Palmer & Brewer, 2011). Guessing will probably create false alarms because foils always outnumber the lone target, and cautionary instructions that reduce guessing have been shown to reliably reduce false alarm rates (Wells & Olsen, 2003). However, since the same instructions were given in both the simultaneous and sequential conditions, the difference in false alarm rates suggests that the simultaneous line-up may be more sensitive to the instruction effect than the sequential line-up. Further investigations are needed to verify this.

SPO and Simultaneous Line-ups

            Criterion shifts are caused by the inherent structure of sequential line-ups (McQuiston-Surrett et al., 2006) and encourage conservative responses where participants refrain from making an identification, which in turn reduces the overall hit rate (Meissner et al., 2005). The SPO line-up was intended to prevent this by postponing responses till after all the photos were seen. However, it still produced significantly fewer hits than the simultaneous condition, suggesting that the criterion shift is unaffected by when participants report their decisions.

            Regarding false alarms, results indicated that the difference between the SPO and simultaneous line-ups were not significant. The SPO line-up was intended to retain the lower false alarm rate of sequential line-ups by forcing witnesses to make absolute judgments that compared photos exclusively to memory (Gronlund, 2004). This would avoid the false alarms caused by relative judgments where the photo that best resembles the suspect is selected, even if the resemblance is vague. However, results showed that the simultaneous and SPO forms of presentation produced comparable false alarm rates. Again, this could be explained by the instruction effect where a cautionary instruction reduces false alarms in a simultaneous line-up but may not be as potent in a sequential-like presentation format.

Limitations and Future Research

            This study tested line-ups in target-present trials. In practice, investigators are unlikely to know if the perpetrator is present or absent – that is why they are using a line-up to identify him/her. In order to minimise misidentifications, our results and inferences should be validated in target-absent trials before being used in practice. This would ideally address the inadequacies of current theoretical explanations and connect line-up results to fundamental psychological concepts about memory and decision-making tasks (McQuiston-Surrett et al., 2001).

Conclusion

            In conclusion, this study finds that none of the three line-ups possess optimal discrimination accuracy in terms of producing fewer false alarms and more hits than the others. The simultaneous line-up produced significantly fewer false alarms than the sequential line-up, although hit rates were comparable. This contradicts previous research and contributes to existing knowledge by showing that sequential line-ups may not be superior to simultaneous line-ups. Attempts at validating a novel SPO line-up was also unsuccessful: while it matched the simultaneous line-up in terms of false alarms, it produced significantly fewer hits. These findings were made in target-present trials and should not be taken out of context without due care. Further research should focus on theory-building that connects line-up results to relevant psychological concepts regarding memory and decision-making.

References

Dillon, J.M., McAllister, H.A., & Vernon, L.L. (2009). The hybrid lineup combining sequential and simultaneous features: A first test. Applied Psychology in Criminal Justice, 5 (1), 90-108.

Dysart, J.E. & Lindsay, R.C.L. (2001). A preidentification questioning effect: Serendipitously increasing correct rejections. Law and Human Behavior, 25 (2), 155-165.

Gronlund, S.D. (2004). Sequential line-ups: Shift in criterion or decision strategy? Journal of Applied Psychology, 89 (2), 362-368.

McQuiston-Surrett, D., Malpass, R.S., & Tredoux, C.G. (2006). Sequential vs. simultaneous lineups: A review of methods, data, and theory. Psychology, Public Policy, and Law, 12 (2), 137-169.

Medina, J.J. (2008). The biology of recognition memory. Psychiatric Times, 25 (7), 13-16.

Meissner, C.A., Tredoux, C.G., Parker, J.F., & MacLin, O.H. (2005). Eyewitness decisions in simultaneous and sequential lineups: A dual-process signal detection theory analysis. Memory & Cognition, 33 (5), 783-792.

Steblay, N., Dysart, J., Fulero, S., & Lindsay, R.C.L. (2001). Eyewitness accuracy rates in sequential and simultaneous lineup presentations: A meta-analytic comparison. Law and Human Behavior, 25 (5), 459-473.

Palmer, M.A. & Brewer, N. (2011). Sequential lineup presentation promotes less-biased criterion setting but does not improve discriminability. Law and Human Behavior, 36 (3), 247-255.

Steblay, N.K. & Phillips, J.D. (2011). The not-sure response option in sequential lineup practice. Applied Cognitive Psychology, 25, 768-774.

Wells, G.L. & Olson, E.A. (2003). Eyewitness testimony. Annual Review of Psychology, 54, 277-295

Yonelinas, A.P. (2001). Components of episodic memory: The contribution of recollection and familiarity. Philosophical Transactions: Biological Sciences, 356 (1413), 1363-1374.