Best Practices in Identification Tests

Perhaps the ultimate form of eyewitness evidence is the identification of a suspect from a live or photo lineup, as opposed to more general information provided by a witness, such as a verbal description of an event. Best-practice recommendations in this area are based on a combination of some good procedures used by law enforcement for decades, sound logic and probability theory, basic psychological principles, and dedicated psychology-law research. The primary goal of a good identification procedure is to let the witness’s memory be the basis of his or her decision, rather than any implicit or explicit influences that derive from either the procedure used or the nature of the lineup itself. And, of course, the desired outcome of a good procedure is to secure either an accurate identification of a guilty suspect or a “Don’t know” or “Not there” response if the actual offender is not in the lineup.

There are at least four techniques for obtaining an identification from an eyewitness, and most of the best-practice procedural recommendations apply in all of them (as opposed to filler-selection issues, for example, which don’t apply for at least one of the techniques). The two techniques that have received the most attention by researchers and the legal community, and are the primary focus of this entry, are live lineups (also known as identification parades in the United Kingdom and some other countries) and photo lineups (also known as photo arrays or photo spreads and sometimes called a “6-pack” in the United States, in reference to the most common number of photos used). The other two procedures are the field identification procedure (often called a “showup”), in which just one individual is shown to a witness, usually soon after an event has occurred and within close proximity to the scene, and the in situ procedure, in which a witness is asked to view a group of individuals in a relatively informal setting, such as the lobby area of a police station or a public location that a suspect is known to frequent, such as a bar or a place of employment.

The showup is thought by most eyewitness researchers, and some courts, to be “inherently suggestive,” and few researchers would recommend it as a best-practice technique. The two most obvious potential advantages of a showup are that a potentially dangerous person could be detained on the basis of a positive identification, often with the aim of protecting a person who might be revictimized otherwise, and that an innocent suspect could be quickly exonerated. The showup procedure is included as a legitimate option in the U.S. National Institute of Justice (NIJ) document “Eyewitness Evidence: A Guide for Law Enforcement,” published in 1999, and the Wisconsin Department of Justice’s “Model Policy and Procedure for Eyewitness Identification,” released in 2005. Despite the situations in which the potential advantages of a showup might outweigh the otherwise prudent decision to conduct either a live or a photo lineup with nonsuspect fillers, the best practice-recommendation is to think of a lineup as the default procedure. It is not unreasonable, for example, to expect that law enforcement can use current and near-future technology to construct an electronic photo lineup at a crime scene using a digital image of a suspect who was found in the vicinity and benefit from the immediacy associated with the showup and the safeguards associated with a lineup.

The relatively informal in situ identification procedure tries to take advantage of a naturally occurring situation where a witness gets a chance to observe a suspect when the suspect does not know that he or she is being observed for that purpose. As mentioned, this often happens when a suspect is casually waiting in a police station regarding an incident, usually with a number of other people, and a witness to the incident is asked to look through a door or window to see if he or she recognizes anyone. Although this technique has received little attention in the eyewitness research literature, the legal community typically accepts the results, as long as the basic principles of the more common and formalized procedures discussed below are included. Also, it is common that a more formal identification procedure will follow if an identification is made.

Viewing mug shots could also be construed as an identification procedure, but the phrase identification test implies that the police have a suspect in mind and that they are “testing” their hypothesis that the suspect is the offender (as opposed to testing the witness’s memory per se). A mug shot viewing is typically used when the police do not yet have a suspect, so they show potentially hundreds of photos of people who have been arrested in the past for a similar crime and/or who match a general description of the offender. The result of a mug shot viewing could be a relatively positive identification from an eyewitness, but typically, the procedure just helps to narrow down what the witness remembers about an offender (e.g., “He had beady eyes like Number 55, long blond hair like Number 132,” etc.).

Finally, a witness could be asked to work with a sketch artist to create a likeness of the offender, or use a facial composite system, but these procedures are more in the realm of witness recall as opposed to an identification test.

Constructing the Lineup

This aspect of identification procedures has received the most attention from eyewitness researchers, and several of the issues are addressed in detail in separate entries. A prototypic lineup consists of one suspect and at least five known-innocent fillers (variously called distracters, foils, stand-ins, shills, and other terms). Known innocent means that the police have no reason to believe that the person could have committed the crime in question, as opposed to meaning that the person has never committed a crime in general. The logical power of this single-suspect lineup is that the identification of anyone other than the suspect is a “known error,” because that person could not have committed the crime. At that point, the police need to consider several possibilities—the suspect is not the offender, the witness’s memory is not sufficient to recognize the offender if he or she is present, the offender’s appearance is sufficiently different from the way he or she appeared at the time of the crime so that the witness cannot recognize the offender, or perhaps the witness is reluctant to identify the offender even if he or she is present.

The number of fillers is not as important as the degree to which the fillers serve to make the lineup unbiased against the suspect, yet they should not be so similar to one another and the suspect that it becomes very unlikely for even a witness with a good memory of the offender to recognize the offender if he or she is present. The best-practice recommendation here is that fillers be chosen based on their match to the witness’s description of the offender, as opposed to their match with the suspect’s actual appearance, with certain logical qualifiers. Suppose, for example, that a witness describes an offender as White, male, 30-35 years old, and with a distinctive feature, such as an “insect tattoo on his right cheek.” Suppose further that the suspect in the case has a spider tattoo in the correct location. The other lineup members should also be White, male, 30 to 35 years old, but they need only have an “insect tattoo” on their right cheek (a bee, a scorpion, or even different kinds of spiders) in order to qualify as good fillers. Requiring the fillers to have the identical spider tattoo or covering up the suspect’s tattoo and the corresponding spot on the fillers’ faces serves only to remove a potentially useful memory cue for the witness to recognize the offender if he is in the lineup.

Instructions to the Witness

The most common instruction, recommended by eyewitness researchers since the early 1980s, is “The offender may or may not be present.” The logic of this instruction, supported by empirical research, is that it provides witnesses with a “None of the above” alternative and counters to some extent any tendency of the witness to assume that the police wouldn’t be bothering with an identification procedure if they didn’t have the “right” person. This tendency might lead to witnesses guessing or choosing someone they don’t feel very confident about, which is a primary concern because it is of course possible that the suspect is in fact not the offender. The NIJ Guide for Law Enforcement goes a step further on this specific point and includes the additional instruction, “It is just as important to clear innocent persons from suspicion as to identify guilty parties.”

At this point, it is important to discuss the potentially confusing use of the terms suspect and offender in this context. Many law enforcement agencies have actually changed their wording of the caution that the offender may or may not be present to read that the suspect may or may not be present, most often with a well-intentioned motive not to bias witnesses, but that wording change is not a legitimate substitution. It is just that suspect has become a more acceptable term to use for many people in law enforcement and the media when referring to offenders. Television viewers are exposed to this tendency on a regular basis on shows where a person is on videotape driving at well over the speed limit on the wrong side of the road, crashing into other cars and sometimes attempting to run down or shoot at police officers, all the while being referred to as “the suspect” by the narrator (just saying “the driver” would solve the problem). Granted, in some cases, the term refers to the fact that the driver is suspected of being involved in the bank robbery that initiated the chase, but it can blur the line between describing a person who by all reasonable standards is currently committing a crime (an offender) versus someone who has been apprehended after a crime has occurred as a possible candidate for the offender (a suspect). It also comes up when the media report that the “police are looking for two suspects in the case; the first suspect is described as male, White, average build,” when they really mean to say that the police are looking for two offenders, the first of whom matches that description. Suspect would only work in that example if there’s some reason to believe that the crime did not really occur. Of course many law enforcement policies—and the NIJ Guide, for example—do use an appropriate alternative for offender, such as “the person who committed the crime.”

There are other important points to include as instructions for witnesses prior to participating in an identification test, detailed in the NIJ Guide, the Wisconsin Model, and other sources. In some cases, the relevance of a particular instruction depends on the particular procedure used, but one that deserves special attention concerns witnesses’ confidence in their decision. Eyewitness researchers are in general agreement that it is crucial to get some expression of witness confidence at the time of the identification decision, instead of relying on what is provided at trial, often many months later. The concern is that a witnesses’ experiences after making an identification might influence their confidence, most likely in an upward direction. These experiences can range from the unavoidable implication that the witness has identified the suspect, when the witness is summoned to testify at the trial against that suspect (now the defendant), to something subtle, such as a smile or a nod, to something explicit, such as “Good, you’ve picked the right person!” Therefore, it is recommended that witnesses be told prior to viewing the lineup that they are expected to state, in their own words, and not necessarily on a scale of some kind, how certain they are of their decision and that this confidence rating be obtained prior to any kind of feedback from the person conducting the lineup, subtle or otherwise.

Conducting the Identification Procedure

Almost all the research on procedure concerns photo and live lineups. The instructions for the showup and in situ procedures are the most important part, assuming reasonable safeguards against influencing a witness to say “Yes” or “No” in the case of the showup and against choosing a particular person in the case of the in situ procedure. The most recommended safeguard for all identification tests, except the showup (where it is not possible), is the double-blind procedure, the rationale for which is to avoid unintentionally influencing the witness’s decision and, thereby, the outcome of the lineup procedure. In general, a double-blind lineup can be accomplished in one of at least two ways: The person conducting the lineup does not know who the suspect is and/or cannot see which photo or person the witness is viewing or discussing at any particular time. In practice, it can be difficult if not impossible to conduct a live lineup in such a way that the administrator cannot see the lineup members, so the only reasonable option is usually to have another person administer the lineup. With a photo lineup, however, there are at least two ways for a person who knows the suspect to conduct the procedure in a double-blind fashion (assuming a sequential presentation, as detailed below). The first, low-tech, approach is to randomize the position of the photos (but ensuring that the first photo is a filler) and then use folders or envelopes to conceal the photos until the witness views them, at an angle (or with some sort of small obstruction) that blocks the administrator’s view. This way, if the witness says something like “Number 4 looks a lot like the person I saw,” the lineup administrator does not know whether the witness is referring to the suspect or a filler photo. The relatively high-tech approach is to use a computer to administer the lineup, with built-in randomization and with the screen positioned in such a way that the administrator can’t see it. In fact, there are computer applications available to construct and present the lineup and record the procedure.

The other major recommendation is to present the members of a lineup one at a time (a sequential lineup), as opposed to all at once (a simultaneous lineup), and that the sequential presentation be done in a very particular manner (not just one at a time). The rationale for the sequential lineup is that it reduces the tendency of witnesses to choose the person from the lineup who looks most like the offender they saw (a relative judgment strategy), as opposed to choosing a lineup member only if he or she matches the witness’s memory trace for the offender beyond some threshold level (an absolute judgment strategy). Of course, if the suspect is the offender, then the outcome from both strategies should be the same, but in the case where the suspect is not the offender, the relative judgment strategy can increase the chance of that person being chosen.

There is some criticism about the recommendation for a sequential procedure, based largely on the concern that the rate of witnesses accurately identifying guilty suspects might be lower. There are, in fact, some data showing that the rates for choosing suspects can be lower overall with the sequential procedure, but it is difficult if not impossible to determine if that means guilty people are being identified less often—the lower rate might mean that fewer innocent suspects are being chosen. Also, it has been argued that some of the “accurate” choices of guilty suspects from simultaneous lineups are essentially lucky guesses, which are less likely to occur with the sequential technique, and that lucky guesses are not a legitimate route to justice. So the best-practice recommendation is to use a double-blind, sequential procedure. Step-by-step details are available in the NIJ Guide and the Wisconsin Model.

Recording the Identification Procedure

Ideally, the entire identification procedure would be video- and audiotaped. The camera(s) should be positioned such that the witness and the photos or persons in the lineup are viewable, and a microphone that can pick up any of the witness’s spontaneous utterances should be used. This recommendation is not made with the intent of monitoring the conduct of the person administering the lineup but to capture the procedure and outcome in a way that provides as much information as possible, especially the confidence statement. In fact, most eyewitness researchers consider what happens during the identification procedure as the only relevant information, as opposed to what the witness says about it at trial. As mentioned previously, computers are ideal for administering the double-blind photo lineup procedure, with digital cameras that record directly to a disc or a hard drive, and for monitoring decision times and the order in which the photos were displayed.


  1. Kassin, S. M. (1998). Eyewitness identification procedures: The fifth rule. Law and Human Behavior, 22, 649-653.
  2. Turtle, J., Lindsay, R. C. L., & Wells, G. L. (2003). Best practice recommendations for eyewitness evidence procedures: New ideas for the oldest way to solve a case. Canadian Journal of Police and Security Services, 1, 5-18.
  3. S. National Institute of Justice. (1999). Eyewitness evidence: A guide for law enforcement (NIJ 178240). Washington, DC: Author.
  4. Wells, G. L., & Luus, C. A. E. (1990). Police lineups as experiments: Social methodology as a framework for properly-conducted lineups. Personality and Social Psychology Bulletin, 16, 106-117.
  5. Wells, G. L., Malpass, R. S., Lindsay, R. C. L., Turtle, J. W., & Fulero, S. M. (2000). From the lab to the police station: A successful application of eyewitness research. American Psychologist, 55, 581-598.
  6. Wisconsin Department of Justice, Bureau of Training and Standards for Criminal Justice. (2005). Model Policy and Procedure for Eyewitness Identification. Retrieved from

See also: