A substantial base of laboratory research is now available to aid our understanding of eyewitness identification processes and to support recommendations for lineup reform. However, there are also a limited number of peer-reviewed, published studies that measure eyewitness responses in real police cases. Although few, the studies include large-scale investigations involving a sizable combined sample of eyewitnesses (4000+). The traditional simultaneous lineup format in these studies produces a modal suspect identification (SI) rate of around 40% to 50% and a filler selection rate of approximately 20%.
Field studies bring unique strengths and weaknesses to research efforts, capturing eyewitness decisions in the most forensically relevant settings but under circumstances that lack the control and precision of the lab. Existing field studies—archival summaries of police reports and descriptive data from pilot research—effectively augment laboratory findings.
Each witness decision for a field lineup falls into one of three response categories: (1) an SI, (2) a filler identification, or (3) no choice from the lineup. A challenge for eyewitness field research is that an unknown percentage of real-world lineups do not include the perpetrator. Suspect selections cannot be directly equated with accurate identifications, because any false identification of an innocent suspect is contained within the SI category. Filler selections (foils [innocent persons] or false alarms) are known errors and signal investigators that the witness has a poor memory or is uncooperative, or that the filler is a better match to the offender than is the suspect. “No choice” responses (a lineup rejection) include witnesses unable or unwilling to make a lineup selection. These limitations of data interpretation must be kept in mind as the following field studies are examined.
Archival field studies provide baseline data regarding eyewitness responses under traditional lineup practice—a simultaneous display of lineup members administered by an investigator who knows the identity of the suspect. Some field information is also available for showups—a single-member lineup.
An early examination of 224 identifications made by eyewitnesses to real crimes in California revealed an SI rate of 56% and a showup SI rate of 22%. A year later, a 1994 study in Vancouver, Canada, detailed 170 identification attempts, 90% from simultaneous photo lineups. The authors reported SI rates for robbery victims (46%) and witnesses (33%) and for fraud victims (25%).
A larger sample of police files was reviewed in 2001 for 689 California identification attempts following crimes ranging from homicide to theft. Similar rates of SI were found for 284 simultaneous photo lineups (48% SI) and 58 live simultaneous lineups (50% SI). Live lineup decisions produced 24% false alarms and 26% lineup rejections. (Researchers do not always separate filler and “no choice” decisions, often because police reports do not provide this level of detail.) Showup identification rates were similar whether live (258) or photo (18)—76% and 83% SI, respectively—and significantly higher than rates for the full array. Of particular interest were 66 lineup identifications by eyewitnesses who had made an earlier identification of the same suspect. Significantly more SIs were made in later attempts (62%) compared with witnesses attempting a single identification (45%). A 2005 update of the California simultaneous lineup data, including overlap with the earlier data set, produced an SI rate of 52% for photo and 46% for live lineups; filler picks were at 15% for the overall group.
Additional archival summaries come from researchers in England. These include 2,200 witness identifications for 930 live, simultaneous identity parades. Outcome similarities across studies are evident (also including an unpublished third study of 843 witnesses and 302 lineups by the London police): When the offender was not known previously to the witness, approximately 40% of witnesses identified the suspect, 20% chose a filler, and 40% made no choice from the lineup. When the perpetrator was previously known, not surprisingly, SI was more likely.
Along with recent reforms in lineup practice, data are emerging that capture eyewitness responses under double-blind sequential lineup practice—a one-at-a-time presentation of lineup members, administered by an investigator who does not know the identity of the suspect. A 2006 Minnesota pilot project generated SI rates of 54%, fillers 8%, and “no choice” 38%. This field study also showed that repeated viewing of a lineup by the witness was associated with a reduction in SIs and rising filler selections.
Some of these descriptive studies have also attempted to examine the impact on witness decisions of crime-incident features, such as weapon presence. The researchers are careful to point out the dangers of comparing pseudo-experimental conditions. For example, weapon absence may be confounded with crime type (fraud vs. robbery) and, therefore, also with differential witness attention, quality of culprit description, and delay prior to lineup. While substantial support has been found in controlled laboratory tests for the negative impact of factors such as weapon presence, delay, and cross-race identification, field studies present inconsistent results. The difficulty of interpreting study results following nonrandom assignment is illustrated by a London research team, comparing a “lineup suite” with a standard police-station setting. The researchers noted that lineups assigned to the suite differed in important ways from those assigned to ordinary police stations: time lapsed since the crime event, race of the suspect, and crime violence. Lineup setting was confounded with other critical factors.
Finally, an ancillary line of hybrid lab-field research has developed around testing for fairness of real lineups. A mock witness procedure requires lab participants, who have not seen the crime and are armed only with the culprit description provided by the real witness, to identify the suspect from the lineup. This procedure is typically used to evaluate individual lineups suspected of biased structure. An emerging use of this paradigm is to analyze a sample of lineups from a jurisdiction of interest. Lineup fairness was tested in England using this procedure, demonstrating video lineups to be fairer than photos. In the Minnesota pilot of double-blind sequential lineups, a mock witness procedure confirmed fair lineup construction through a sample of field lineups.
As we look to the future, there is great potential for information gain in well-designed experimental field tests that include methodological necessities such as random assignment and double-blind administration, but data from such tests are not yet available.
- Klobuchar, A., Steblay, N., & Caligiuri, H. (2006). Improving eyewitness identifications: Hennepin County’s Blind Sequential Lineup Pilot Project. Cardozo Public Law, Policy & Ethics Journal, 4, 381—H3.
- Valentine, T., Pickering, A., & Darling, S. (2003). Characteristics of eyewitness identification that predict the outcome of real lineups. Applied Cognitive Psychology, 17, 969-993.