Screen test

One of the most important recent developments in the UK legal system has been in the manner in which evidence is gathered from eyewitnesses. Live parades have now been largely replaced by video parades, an innovation that has been made possible by development of a searchable database of standardised moving video clips.

Video parades are considered by policymakers to be an efficient and cost effective alternative to live parades, reducing the number of cancelled parades and possibly reducing witness anxiety by avoiding the witness having to confront the suspect. The size of the database also means police can construct parades for relatively distinct looking people.

Two different systems are in widespread use in British police forces: VIPER (Video Identification Parade Electronic Recording) and PROMAT (PROfile MATching). Identification parades consist of 15 second clips of each person shown one after another. The sequence starts with a head and shoulders shot of the person looking directly at the camera. The person then slowly turns their head to right and left profile. The administrator enters descriptive information about the suspect and the system finds a set of people who match this description from a large database of standardised high quality images. The video parade has to be viewed twice before the witness can make a decision.

The Police and Criminal Evidence Act (PACE) codes of practice provide a detailed mandate for how identification evidence must be collected in England & Wales (though not legally binding in Scotland, Scottish procedure shares many commonalities with procedures in England, Wales and Northern Ireland). The relevant guidance in Scotland is the Lord Advocate's Guidelines (2007).

Scottish and English studies

In the real world an identification parade is put together with a police suspect who may or may not be the actual culprit. Laboratory research has repeatedly shown that eyewitnesses make errors (Wells, Memon and Penrod, 2006), but what do we know about current procedures? Evaluation of video identification procedures was undertaken by Memon and colleagues in Scotland (Memon, Havard, Clifford, Gabbert & Watt, 2011) and in England & Wales (Horry, Memon, Wright and Milne, in press). In Scotland, data was obtained from 1718 real witnesses who viewed video parades in 2008-09. The responses were recorded by identification officers using a one page questionnaire.

The major variables of interest were: characteristics of the witness, characteristics of the suspect, offence type, delay since incident, and aspects of the VIPER procedure. All police forces in Scotland participated in this study. Of the 1,044 witnesses who did not know the suspect, the rate of suspect identification was 44% and the rate of volunteer identifications was 42%, with the remainder of witnesses declining to make a choice. The suspect identification rates are comparable with earlier field studies conducted in the UK, but the volunteer identification rates are twice as high as those obtained with live parades (e.g. Valentine et al, 2003). However, it is possible that investigators may have confused the negative and no identification categories, marking a lineup rejection (no choice) as a negative identification (choice of volunteer) in some cases. The questionnaire was reworded for the study conducted in England.

The study in England (Horry et al, in press) sampled witnesses from five police forces in England. Within each suite, one questionnaire was completed for every video lineup conducted over a three month period. The data collection began in September 2009 and ended in April 2010. The catchment areas for these suites included inner city and rural communities which varied widely in socio-economic demographics and ethnicity. Data was obtained from 1,359 witnesses and 783 distinctive crime events. The suspect image was chosen in 539 cases (40%) and a volunteer image was chosen in 361 cases (26%). Thus video technology, which is less costly and more efficient, can yield comparable levels of performance to live parades.

This study was part of a larger initiative to engage in knowledge transfer activities with police officers. We examined all protocols and policies concerning identification procedures from the 43 forces that make up the police service in the UK. In addition, members of the research team attended user group meetings in order to discuss issues surrounding video identification at a national level. All police forces adhere to the guidelines set out in PACE, but each force has come up with its own in-house procedure based on the codes of practice. Our goal was to standardise the procedures to ensure consistency in practice as well as to see if we could suggest any changes in policy and practice based on research.

The changes suggested in our guidance document did not require changes to existing PACE codes, but were considered to be improvements in practice based on research. A workshop was also held with the associated identification inspectors responsible for all identification across the UK, Association of Chief Police Officers (ACPO) representatives, and Home Office officials. The workshop provided a forum for discussing the issues that arose from the analysis of UK police procedures and was also used to agree a national standard for the conduct of video parades. In this paper we summarise the issues that arose in our discussion with police officers, and some recommended changes to practice that arise from research and consultation with identification officers.

Recommendation 1

Identification procedures should only include one suspect. In cases with multiple suspects, each suspect should be placed in a different procedure with unique volunteers.

If there are two suspects of roughly similar appearance, the PACE codes allow both to be placed in a single video identification along with at least 12 volunteers. This can create several problems. First, how does one determine whether two suspects are "roughly similar" enough to be included in the same procedure? Secondly, how does one select appropriate volunteers for two different suspects? To match all of the volunteers to one of the suspects would risk making the other suspect stand out. To match some volunteers to each of the suspects may reduce the number of plausible volunteers for each suspect. Thirdly, including more than one suspect in a lineup increases the probabilities of an eyewitness making a harmful error. The computer-based system used in the UK, coupled with the large database of volunteers, means that it is not difficult to create separate VIPs for each suspect, thereby reducing the risk of false identification (Wells and Turtle, 1986).

Recommendation 2

A standard set of proforma (instructions) is needed across UK police forces to ensure consistency in what witnesses are told. For example, the term "positive identification" is used to denote the identification of a suspect. Hughes (2005) interviewed 30 identification officers in England & Wales about the meaning of this term and noted a large disparity in responses, with every officer having a different understanding of the term. It seems that if police officers are confused about the term "identification", then eyewitnesses too will be confused. We recommend that instructions to witnesses be clarified to explain the purpose of the identification procedure, and that the word "positive" should be removed from PACE Code D. Witnesses should simply be asked if they have made an identification of the person they saw at the scene of the crime.

Recommendation 3

Identification officers should ask whether there were any other eyewitnesses to the crime. If yes, officers should ask (i) whether the eyewitness discussed the crime with any other eyewitnesses; and (ii) what was discussed. Relationships between eyewitnesses should also be recorded as well as any discussions or interactions on multi-media sites.

Many crimes are witnessed by multiple eyewitnesses, and eyewitnesses who discuss an event may influence each other’s memory. This is called memory conformity (Wright et al, 2009). These effects occur because memory is reconstructive. Every time we remember something, we reconstruct that memory, and we may change it in some small (or large) way to incorporate information from other sources. Opportunities for eyewitnesses to discuss their memories are plentiful with increased use of social media. While it is not possible for identification officers to prevent communication between eyewitnesses, they can ask eyewitnesses about the extent of their contact with other eyewitnesses involved in the case. We recommend that such questions become standard procedure in all cases.

Recommendation 4

The identification officer responsible for conducting the procedure should be unaware of the position of the suspect within the procedure, and the eyewitness should be told this. A move towards fully automated administration of procedures would be desirable.

Social psychologists have long acknowledged that researchers may inadvertently influence the outcome of their experiments by passing on cues to their participants. Such expectancy effects can occur in a wide range of domains, both inside and outside of the laboratory. There is a simple way to eliminate expectancy effects: neither the experimenter nor the participant should know what the expected outcome is. This is called a double-blind design. The use of specialist identification suites in the UK offers ample opportunity for double-blind video procedures at very little cost.

Recommendation 5

Any requests to view whole or part of the procedure again should be accurately recorded, including the numbers of any images shown and any comments made by the eyewitness. At the end of the second viewing of the procedure, all eyewitnesses should be reminded that the culprit may or may not be present in the procedure.

The Scottish Guidelines (Lord Advocate's Guidelines, 2007), as well as PACE, require each witness to view the entire parade twice before making a decision. In our discussions with police officers it came to light that witnesses were reluctant to wait until they had seen both viewings before speaking. Eyewitnesses who requested additional viewings made more errors (volunteer identifications) than eyewitnesses who did not (Memon et al, 2011; Horry et al, in press). In fact, Horry et al found that eyewitnesses who requested an additional viewing were just as likely to identify a volunteer as they were to identify the suspect. Eyewitnesses who did not request an additional viewing, on the other hand, were about 2.5 times more likely to identify the suspect than a volunteer. Further research on this issue is warranted.

Recommendation 6

A statement of confidence must be taken immediately after the decision, before the eyewitness has any chance to learn whether they identified the suspect.

Psychologists have attempted to find markers of eyewitness accuracy that will help investigators and prosecutors to assess whether an eyewitness’s decision is likely to be correct or incorrect. One such marker is the confidence with which the decision is made. Recent research has shown that identifications made with higher confidence are generally more likely to be accurate, although overconfidence can occasionally result in eyewitnesses who are certain but wrong (Brewer and Wells, 2006).

One of the reasons for this is that the confidence expressed by an eyewitness can be inflated by exposure to information that implies that the decision was correct. As well as increasing confidence, feedback can also change eyewitnesses’ reports of how much attention they paid to the culprit, how good their view was, how easy they found the identification task, and so on (Douglass et al, 2010).

Our discussions with officers across Britain revealed that eyewitnesses in some suites are routinely asked for confidence statements, but not in others. Furthermore, in some suites, eyewitnesses are routinely given feedback about their decision. Failing to record confidence at the time of the decision and providing feedback is a particularly dangerous combination that can make an unreliable eyewitness appear confident and persuasive in front of a jury.

Summary and conclusions

We have revised some of the factors that can influence video identification performance in field settings. Video was introduced to use resources more efficiently. Live parades were difficult, costly and frequently cancelled due to lack of suitable volunteers. Our results show that suspect identification rates remain unchanged with the use of video. Video parades can be put together within 24 hours, and our data suggest the speed with which a video identification parade can be put together is likely to impact suspect identification rates. The longer term effect on prosecution of crimes is something that should be examined. The shorter delays also have a further beneficial impact for the police by increasing the chances that when victims and witnesses come forward an identification parade can be put forward in a timely fashion and before a witness/victim changes their mind about helping the police.

This paper focused on factors that are directly relevant to the conduct of an identification procedure. There are many other factors that can influence the reliability of an eyewitness – for example, the viewing conditions at the time of the crime, the age of the eyewitness – that are outside the scope of this paper. Advances in technology are presenting police with a wide new range of issues to consider. It is essential that the police and practitioners keep abreast of the relevant research and that academics work together with them to improve policy and practice.

References

Brewer, N, and Wells, G L, 2006: “The confidence-accuracy relationship in eyewitness identification: Effects of lineup instructions, foil similarity, and target-absent base rates”, Journal of Experimental Psychology: Applied, 12 (1), 11-30.

Douglass, A B, Brewer, N, and Semmler, C, 2010: “Moderators of post-identification feedback effects on eyewitnesses’ memory reports”, Legal and Criminological Psychology, 15 (2), 279-292.

Horry, R, Memon, A, et al, forthcoming: “Predictors of eyewitness identification decisions from video lineups in England: A field study”, Law and Human Behavior (published online 19 April 2011).

Hughes, C (2005): Making the best use of video film identification parades and providing a fair and consistent approach to witnesses, unpublished Masters thesis, University of Portsmouth.

Lord Advocate’s Guidelines (2007): www.copfs.gov.uk/sites/default/files/Publications/Resource/Doc/13547/0000269.pdf

Memon, A, et al, 2011: “A field evaluation of the VIPER system: A new technique for eliciting eyewitness evidence”, Psychology, Crime, and Law, 17 (8), 711-729.

Valentine, T, Pickering, A, and Darling, S, 2003: “Characteristics of eyewitness identification that predict the outcome of real lineups”, Applied Cognitive Psychology, 17 (8), 969-993.

Wells, G L, et al, 1998: "Eyewitness identification procedures: Recommendations for lineups and photospreads", Law and Human Behavior, 22 (6), 1-39.

Wells, G L, and Turtle, J .W, 1986. Eyewitness identification: the importance of lineup models. Psychological Bulletin, 99 (3), 320-329.

Wells, G L, Memon, A, and Penrod, S D, 2006: “Eyewitness evidence: Improving its probative value”, Psychological Science in the Public Interest, 7 (2), 45-75.

Wright, D B, et al, 2009: “When eyewitnesses talk”, Current Directions in Psychological Science, 18 (3), 174-178.