Practice with national-level exam (FACT, FACT Plus, NET, CUET, etc.) mocks, learn from structured notes, and get your doubts solved in one place.
The system-variable reforms that improve eyewitness-identification accuracy: sequential vs simultaneous lineup (Lindsay + Wells 1985 sequential-superiority effect and its contested replication record); double-blind administration to remove unintentional administrator cues; immediate post-identification confidence statements; the NIJ 1999 + 2017 *Eyewitness Evidence: A Guide for Law Enforcement* recommendations; the UK PACE 1984 Code D lineup procedure; the Indian Identification Parades under BNSS 2023 § 54 (replacing CrPC § 9) and the *R v. Turnbull* 1976 ABC warning still cited by Indian High Courts; the post-2018 ABA Eyewitness Identification Standards reform.
Last updated:
The identification parade, the police lineup, the video identity procedure: the procedure by which a witness is asked to identify a suspect from a group of people is among the most consequential and most error-prone steps in the criminal process. Decades of experimental research and a growing body of data from wrongful-conviction cases have established that the way a lineup is constructed and administered substantially affects the probability of a correct identification and the probability of a false one.
This is not primarily a question of witness honesty. A witness who falsely identifies an innocent suspect is not, in the vast majority of cases, lying. They are responding to an ambiguous cognitive situation in a way that the procedure itself encourages. Change the procedure and the error rate changes with it. That basic insight, demonstrated first in controlled laboratory conditions and then progressively validated in field research, is the intellectual foundation of the lineup-reform movement.
The practical stakes are clear. The US Innocence Project data establish that eyewitness misidentification contributed to roughly 69 percent of DNA-based wrongful convictions. Many of those misidentifications were shaped by lineups that used simultaneous presentation, were administered by officers who knew which person was the suspect, gave witnesses implicit cues that a choice was expected, and then provided feedback confirming the selection. Reforming any one of these features improves accuracy. Reforming all of them substantially does.
This topic covers the experimental evidence for each major reform, the contested replication record of the sequential-superiority effect, the cross-jurisdictional implementation record, and the current state of procedure in the United States, England and Wales, India, and Canada.
*In a simultaneous lineup, the witness asks: which person looks most like the perpetrator? In a sequential lineup, the question becomes: is this person the perpetrator?*
The traditional lineup is simultaneous: the witness views all persons or photographs at once and selects the person most resembling the perpetrator from the array. Roy Lindsay and Gary Wells published the experimental case against this format in 1985 in the Journal of Applied Psychology.
The simultaneous format encourages relative judgement. A witness who is uncertain will scan the lineup members and select the person who most closely resembles their memory of the perpetrator, even when the actual perpetrator is not present. The result is that the simultaneous format produces both high correct-identification rates (when the actual perpetrator is in the lineup) and high false-identification rates (when the perpetrator is absent and an innocent person resembles the perpetrator). From a justice system perspective, this is a dangerous combination: the procedure is impressively accurate in true-positive cases and impressively inaccurate in false-positive cases.
The sequential format. Lindsay and Wells (1985) proposed showing lineup members one at a time and requiring the witness to make a yes/no decision on each before seeing the next member. A critical feature: the witness does not know how many members are in the lineup, removing the ability to compare across members and select the best match. The procedure forces an absolute judgement: does this person match my memory of the perpetrator?
The 1985 results were dramatic: the sequential format reduced false identifications from 43 percent to 17 percent while reducing correct identifications only from 58 percent to 50 percent. The diagnostic ratio (correct identifications divided by false identifications, a measure of the procedure's ability to discriminate guilty from innocent suspects) improved substantially.
The replication controversy. The sequential-superiority effect was replicated in numerous laboratory studies through the 1990s. Field research was more mixed. A large randomised controlled trial conducted across four US police departments (Mecklenburg et al. for the Illinois State Police, 2006) found no consistent sequential advantage and in some conditions found the simultaneous procedure performed better. The methodology of that study was subsequently criticised (Wells et al., 2011) for its non-blinded administration: officers administering the sequential lineups knew who the suspect was, which potentially undermined the sequential procedure's advantage. The same study has been characterised as a case in which a reform was tested under conditions that guaranteed it would fail.
The APLS white-paper response. The American Psychology-Law Society issued a white paper in 2020 re-examining the evidence base. The consensus position is that the sequential format, when properly administered by a blind administrator, produces a better diagnostic ratio than the simultaneous format, primarily by reducing false identifications at a modest cost to correct identifications. Several states had by that point already adopted sequential procedures. The debate continues, but the direction of evidence supports the sequential-superiority position when confounds are controlled.
*An officer who knows which person is the suspect communicates that knowledge to the witness, without lying, without intending to, and without being aware of doing so.*
Even a technically well-constructed lineup can be corrupted by the administrator. If the officer conducting the lineup knows which person is the suspect, they can unconsciously influence the witness through gaze, posture, tone of voice, and response timing. This is not a claim about misconduct. The effect operates below conscious awareness on both sides.
The evidence comes initially from analogy. Robert Rosenthal's experimenter-expectancy work in the 1960s demonstrated that experimenters who knew the expected result of a study influenced their subjects' behaviour toward that result, even in carefully controlled conditions. The mechanism is primarily non-verbal: gaze direction, response latency, and subtle affirmation cues. If the effect is powerful enough to shape rat maze-running results in a controlled experiment, it is powerful enough to shape a witness's lineup decision in a highly charged real-world setting.
The Wells (1988) blind-lineup proposal. Gary Wells formally proposed double-blind lineup administration in 1988. The term "double-blind" here mirrors its use in clinical-trial design: neither the witness nor the administrator knows which lineup member is the suspect. In practice, this means the lineup is administered by an officer who is not involved in the investigation and does not know the identity of the suspect.
Field implementation challenges. Police departments, especially smaller ones, often argue that they cannot spare a second officer to administer a lineup. The practical solution is the "folder shuffle" method developed by Natalie Seva for smaller agencies: the administrator shuffles the photo lineup face-down and shows each photo to the witness in random order, not seeing which photo is being shown. The administrator therefore cannot know when the witness is viewing the suspect photo. A second practical solution is electronic administration: the witness views photos on a laptop or tablet without an officer present, with responses recorded automatically. Electronic administration achieves double-blind conditions without requiring a second officer.
Post-Innocence Project legislative response. By 2023, more than twenty US states had enacted mandatory double-blind lineup requirements or adopted mandatory best-practice guidelines. North Carolina was among the first, enacting the Eyewitness Identification Reform Act in 2007 following a series of exonerations linked to identification procedures. Texas (2011), Ohio (2010), and New Jersey (post-Henderson 2011) followed with mandatory blind-administration requirements.
In England and Wales, the Police and Criminal Evidence Act 1984 Code of Practice D addresses identification procedures. The default method since 2003 has been the VIPER (Video Identification Parade Electronic Recording) system, in which witnesses view recorded video clips of the suspect and eight or more non-suspect individuals. VIPER lineups are administered by a dedicated identification officer who is not involved in the case, satisfying the blind-administration requirement by institutional design rather than ad hoc procedure.
In India, BNSS 2023 § 54 requires identification parades to be conducted before a magistrate. The magistrate's presence provides some degree of independent oversight, but the procedure does not include an explicit requirement for administrator blindness in the sense used in psychological research. The investigating officer is typically present and the witness is asked to identify the suspect from among a group of similar-looking persons assembled by the officer. This format remains vulnerable to administrator-cue effects.
*If the witness believes the perpetrator must be in the lineup, they will find one.*
The instruction given to a witness before a lineup significantly affects their behaviour during it. The key issue is the "must-be-there" bias: if a witness believes, or is led to believe, that the perpetrator is definitely present in the lineup, they will feel pressure to make an identification even when uncertain.
The seminal Malpass and Devine (1981) study. Participants witnessed a staged act of vandalism and then viewed a lineup. Half received biased instructions ("We have a suspect; please identify him") and half received unbiased instructions ("The perpetrator may or may not be in the lineup; it is just as important to clear an innocent person as to identify the guilty one"). When the actual vandal was in the lineup, instruction condition had no effect on correct-identification rates. When the actual vandal was absent (a target-absent lineup), the biased instruction produced false identifications in 78 percent of participants; the unbiased instruction reduced this to 33 percent. The biased instruction dramatically inflated false positive rates in the most dangerous case: when the perpetrator is absent but the witness makes an identification anyway.
Mandated unbiased instructions. The NIJ 1999 Eyewitness Evidence guidelines made unbiased pre-lineup instructions one of the four core recommendations, alongside double-blind administration, lineup composition, and immediate confidence recording. The specific language recommended: "The person who committed the crime may or may not be in this lineup. You should not feel compelled to make an identification. It is just as important to clear an innocent person as to identify the guilty one." This language, or close equivalents, is now required by statute or departmental policy in most US jurisdictions that have adopted lineup reforms.
PACE 1984 Code D in England and Wales similarly requires that witnesses be told before the procedure: "You do not have to say that any of the photos or any of the people is the suspect; it may be that the person you saw is not on display." The 2003 and 2017 Code D revisions maintained this requirement.
In Australia, the Australian and New Zealand Guidelines on Eyewitness Identification (ANZPAA, 2016) include equivalent pre-identification instructions as a recommended procedure, with particular emphasis on the message that "it is acceptable not to identify anyone."
In India, the BNSS and associated procedural guidance does not contain explicit mandated pre-parade instructions of the unbiased type. This gap is a recognised area of procedural vulnerability in cases where the identification parade evidence is challenged by defence counsel.
*A lineup with only one person matching the witness's description is not a six-person lineup; it is a one-person showup.*
The statistical fairness of a lineup depends on whether all members could have committed the crime based on the witness's prior description of the perpetrator. This property is called lineup composition or, in the statistical measurement literature, lineup functional size.
The filler selection problem. Lineup fillers (non-suspect members) are sometimes selected to match the suspect's appearance rather than the witness's description of the perpetrator. This sounds reasonable but creates a dangerous asymmetry: if the suspect is the only person who matches the witness's prior description, the witness can identify the suspect by description-matching rather than by genuine memory-based recognition. This is called the "description-match fallacy." Correct practice is to select fillers who match the witness's description of the perpetrator, not necessarily the suspect's actual appearance.
Wells, Rydell, and Seelau (1993) demonstrated this point experimentally: lineups where the innocent suspect was the best description-match produced very high false-identification rates, because witnesses with uncertain memories could simply select the best match.
Functional size. The functional size of a lineup is the number of members who plausibly could be the perpetrator based on the prior description. A lineup with six members but only one who matches the description has a functional size of one, regardless of its nominal size of six. Computing functional size requires having the witness's prior description and assessing how many lineup members fit it. Research by Malpass and colleagues in the 1990s developed the "propitious heterogeneity" standard: each filler should be plausible given the description, but collectively, the lineup should not be so homogeneous that identification from description alone becomes trivial.
PACE Code D requirements in England and Wales. Code D paragraph 3.2 requires that the lineup should consist of at least eight persons in addition to the suspect for a live lineup, or at least eight images in addition to the suspect for a video parade (VIPER). The persons or images must bear "a general likeness to the suspect," which has been interpreted in practice as match-to-description rather than match-to-suspect-appearance. The VIPER database contains thousands of video clips specifically collected to provide diverse filler pools.
Indian practice under BNSS 2023 § 54. The statute requires that the suspect be placed among persons of "similar appearance." The operational standard for selecting these persons is not defined in the statute with the specificity of Code D, and the composition of identification parades in Indian practice has been litigated in numerous High Court cases, typically on the ground that the parade was constituted with persons obviously different in age, build, or complexion from the suspect.
*England moved the identification parade from a police station yard to a laptop screen, and the logic of the procedure changed with it.*
The Police and Criminal Evidence Act 1984 (PACE), and in particular Code of Practice D made under it, is the statutory framework governing eyewitness identification procedures in England and Wales. Code D has been revised several times, with significant updates in 2003 and 2017 addressing digital procedures.
Live lineups and the 2003 shift to VIPER. Before 2003, the standard England and Wales identification procedure was a live lineup conducted at a police station. The suspect stood among a group of persons selected by the officer. The witness viewed the lineup from behind a one-way screen and indicated any recognition. The problems with this format are familiar: it was resource-intensive, it created practical difficulties in assembling sufficiently similar-looking persons, and it did not guarantee blind administration.
The 2003 Code D changes introduced video identification as the default procedure. The VIPER system uses a standardised database of video clips of persons who have consented to inclusion, recorded under controlled lighting conditions. When a suspect is charged, a VIPER coordinator (an identification officer who has not been involved in the case) selects eight or more database clips of persons matching the suspect's description and combines them with a clip of the suspect. The witness views the sequence electronically, with the option to view it multiple times. The result is inherently double-blind: the VIPER coordinator does not know which clip is the suspect until after the procedure is complete.
Code D documentation requirements. The 2017 revision required that identification procedures be video-recorded in full, including the witness's words at identification, any expression of confidence, and any other behaviours. This mirrors the NIJ recommendation for recording the identification process and provides a contemporaneous record available for defence scrutiny at trial.
The Turnbull warning. Independent of the identification procedure itself, R v. Turnbull [1976] QB 224 requires the trial judge to warn the jury about the risks of eyewitness identification evidence. The ADVOKATE mnemonic covers: Amount of time the witness had to observe the perpetrator; Distance; Visibility conditions; Obstruction of view; whether the witness knew or had seen the perpetrator before; Any particular reason to remember the face; Time elapsed; Errors or material discrepancies in the account. The Turnbull warning is mandatory when the case turns substantially on identification evidence, and its omission or inadequacy is a ground of appeal.
*A 1976 English judgment about an armed robber is now routinely cited in Indian High Court identification cases.*
Indian appellate courts have cited R v. Turnbull [1976] QB 224 with regularity since the 1980s. The Supreme Court in Harchand Singh v. State of Haryana (1974) had established early that identification evidence must be treated with caution when it is the primary basis for conviction. The detailed Turnbull framework provided a structured analytical tool that Indian courts found useful.
Key Indian Supreme Court identification cases. Manohar Lal Sharma v. State of Uttar Pradesh (2019) reiterated that an identification parade must be held promptly after arrest, while the witness's memory is fresh, and that delay in holding the parade substantially reduces the evidentiary weight of any subsequent identification. The court emphasised that where the witness had already seen the accused during investigation, the parade loses its value.
State of U.P. v. Ram Babu Upadhyay (2001) established that if the accused refuses to participate in an identification parade, the court may draw adverse inference, but that refusal alone cannot substitute for a parade. Sk. Yusuf v. State of West Bengal (2011) held that a dock identification at trial, without a prior identification parade, carries little weight.
BNSS 2023 § 54 reform. The provision, replacing CrPC § 9, maintains the requirement that identification parades be conducted before a magistrate. The 2023 reform introduced expanded provisions for electronic identification processes, though the implementation guidelines for digital identification in India (analogous to VIPER) are still being developed at state level. The statutory framework does not yet incorporate the four-component NIJ best-practice framework (blind administration, unbiased instructions, immediate confidence recording, appropriate composition), all of which remain matters of departmental practice rather than statutory requirement.
The gap between Indian procedure and international best practice. The practical difference between BNSS § 54 identification parades and VIPER-based procedures is significant. BNSS parades are live; the suspect stands among assembled persons in a police station yard or room; the investigating officer is often present; there is no explicit unbiased-instruction requirement; and witness confidence is not routinely recorded immediately post-identification. Scholarly work by Indian criminologists and defence lawyers has identified these gaps. The direction of reform at state police level in several jurisdictions has been toward video-based identification procedures, but standardised implementation is not yet national.
Lindsay and Wells (1985) proposed the sequential lineup format primarily to address which cognitive process that distorts identifications in target-absent simultaneous lineups?
Test yourself on Forensic Psychology with free, timed mocks.
Practice Forensic Psychology questions