Double-Blind Lineups and Confidence Statements: Why Procedure Matters
Author: Ukrainian Psychological Hub · Published: October 11, 2026 · Editorial Policy
Ukrainian Psychological Hub | October 11, 2026 | Psychology for the AI Era.
A double-blind lineup protects an eyewitness identification from information the administrator already possesses. An immediate confidence statement preserves what the witness actually believed at the first identification decision, before subsequent feedback, discussion, and legal proceedings can reshape that belief. Together, these safeguards improve the conditions under which an identification is obtained, recorded, and later interpreted. They do not turn a memory judgment into proof of guilt. Their value is in protecting an evidentiary process from avoidable influence and preserving its original informational content. The National Academies' eyewitness-identification report and the American Psychology-Law Society scientific review place both safeguards at the center of recommended identification practice.
What Is a Double-Blind Lineup?
A lineup is a procedure in which a witness is asked whether a person who committed an event is present among several people or photographs. The suspect is the person investigators are considering; the fillers are people known not to be the suspect in that particular lineup. In a double-blind lineup, the administrator who presents the array does not know which member is the suspect. This protects the witness from administrator-origin cues, including unintentional signals about where investigators expect the witness to look, whom to choose, or whether a particular response is approved. A live lineup and a photographic array can both be administered blind. The National Research Council's detailed recommendations explicitly cover both.
The expression double-blind comes from experimental methods. In eyewitness work, the essential operational fact is that the person administering the identification cannot identify the suspect during the procedure. The witness ordinarily also does not know which person is the suspect; that feature alone is insufficient, because a knowledgeable officer can still unintentionally affect the result. A blinded alternative is a carefully designed mechanism that prevents an administrator who knows the suspect's identity from knowing which photograph the witness is viewing. Some computer presentations and properly implemented folder methods can approximate the protection of a genuinely blind administrator. These alternatives require documentation and validation of the actual workflow; the label 'blind' cannot substitute for examining what the administrator could see or infer.
Double-blind, single-blind, and blinded alternatives
In a single-blind administration, only the witness lacks the identity of the suspect. The administrator may know it. In a double-blind administration, both the witness and the administrator lack that information during the test. A blinded presentation is a procedural workaround in which a knowledgeable administrator is prevented from seeing the order or identity of the photographs being evaluated. Each arrangement exposes different channels of potential influence. Kovera and Evelo's research review explains why nonblind administrators can both signal expectations and interpret tentative witness answers in a manner that favors suspect selections.
What Is an Immediate Confidence Statement?
An immediate confidence statement is the witness's own assessment, documented when the first lineup decision is made, of how certain the witness is about the identification. Its timing is crucial. It is an account of the witness's subjective certainty at that particular moment, not a psychometric measurement that proves the person selected committed the crime. Investigators should record the precise words used, the question that elicited them, the response, the timing, and any contextual information needed to understand the answer. The National Academies recommend verbatim documentation after the first identification and before confirmatory feedback.
A spontaneous statement such as 'I think it is number three, but I am uncertain' conveys different information from a later recollection that the witness 'always knew.' A statement like 'I'm 80 percent sure' has to be preserved as the witness's own phrase, rather than converted by an investigator into a claimed 80 percent probability that the suspect is guilty. Verbal descriptions can also be ambiguous across people and contexts. A 2024 set of experiments on verbal eyewitness confidence documents substantial variability in the language witnesses naturally use; preservation of original wording therefore matters for interpretation.
Why Procedure Matters: The Psychology of Influence
Administrator knowledge can change behavior
Investigators conducting an identification procedure may know a suspect's name, appearance, or position. Even a conscientious professional can reveal expectations through changes in pacing, attention, tone of voice, facial expression, gestures, or follow-up questions. An administrator may linger on the suspect's photograph, respond more positively to a tentative suspect identification, or treat a similar response toward a filler as a mistake. These behaviors are mechanisms of information leakage, often operating without a deliberate effort to influence the witness. Blind administration removes the administrator's access to the critical information and therefore blocks a direct source of these cues. Experimental work synthesized by Kovera and Evelo (2020) supports this rationale.
Feedback can alter later confidence and memory reports
When a witness selects a person, comments such as 'You picked the person we suspected' can retrospectively strengthen confidence. The witness may later remember having been more certain during the original selection than the contemporaneous record actually shows. Experimental studies also find changes in remembered viewing opportunity, attention, and ease of identification after feedback. In a classic controlled investigation, Wells and Bradfield (1999) examined these postidentification-feedback effects; related experimental evidence from Bradfield, Wells, and Olson (2002) examines how confirmatory feedback damages the relationship between expressed certainty and accuracy. The practical safeguard is to capture the initial decision and confidence before any informative reaction.
A lineup is an evidence-gathering test
An identification procedure tests recognition under structured conditions. Its result depends on what the witness remembers, how that memory was encoded, what information the witness has encountered afterward, how fillers were chosen, how instructions were framed, and how the response was obtained. These are different sources of variation. Scientists commonly distinguish estimator variables—circumstances investigators cannot retrospectively control, such as lighting or exposure duration—from system variables, which authorities can control, such as instructions, filler selection, blinding, and documentation. The National Research Council (2014) emphasizes that systems can improve the integrity of evidence collection even though witnessing conditions themselves cannot be rewound or repaired.
How a Properly Documented Procedure Works
1. Record the witness account before constructing the lineup
A sound workflow begins with an open, nonleading interview and a detailed record of the witness's first description, the circumstances of observation, and any prior exposure to suspect images, social media, news, discussions with other witnesses, or investigative information. Investigators should have an evidence-based reason for including a particular suspect. This baseline enables a later reviewer to ask whether the lineup was built around the original description or whether the process introduced features absent from it. The 2020 AP-LS recommendations expressly include a prelineup interview and evidence-based suspicion among the nine practices aimed at improving evidence quality.
2. Build a fair lineup
An array should normally contain one suspect and fillers who match the witness's description of the perpetrator rather than artificially highlighting the suspect. Distinctive features may need to be replicated or otherwise neutralized according to policy. The witness should have a meaningful opportunity to reject the entire array: neither an obviously different set of fillers nor a suggestion that someone must be selected produces a fair memory test. The actual composition, photo sources, number and order of positions, any alterations to images, and identity of the person who prepared the lineup should be preserved. The National Academies' guidance treats lineup composition and recording as necessary components of reliable practice.
3. Assign a genuinely blind administrator
The administrator should be selected so that investigative records, suspect labels, photo metadata, officer communications, or other cues do not disclose who the suspect is. The administrator should also avoid discussing the case facts in ways that might cue the witness. If a small department uses an alternative blinded method, its design must prevent exposure to the suspect's position at the moment the witness makes the choice. North Carolina's Eyewitness Identification Reform Act explicitly provides for independent administrators and particular neutral alternatives. This is an example of one state's statutory model, not a universal rule governing every jurisdiction.
4. Deliver standardized, comprehensible instructions
Before displaying photographs or participants, explain that the perpetrator may or may not be present, that the witness is free to reject the lineup, that the investigation will continue regardless of the outcome, and that the witness should avoid guessing merely to be helpful. Instructions must be understood by the particular witness, with competent language access or other appropriate accommodation where needed. An instruction delivered formally but not understood fails to create the intended psychological conditions. Agencies should preserve the exact script and record any departure from it. Standardized witness instructions are a core National Academies recommendation.
5. Capture the decision without steering it
The witness may identify the suspect, choose a filler, reject the lineup, express uncertainty, decline to choose, or describe partial resemblance without making an unequivocal identification. Each result matters and should be documented without converting a tentative observation into a categorical choice. A blind administrator can still ask neutral clarifying questions when needed, but should not pressure a witness to select or revise an answer. The result must preserve the exact response, its sequence and any hesitation or qualification. Procedural recording is especially useful when competing later accounts describe the same decision differently.
6. Elicit and preserve the initial confidence statement
Immediately after a selection, the administrator should obtain the witness's confidence in neutral terms and record it in the witness's own words. Ideally, the full encounter is video-recorded with clear audio; where recording is unavailable, the written account should identify exactly what question was asked and what words were spoken. Investigators must not say whether the chosen person is the suspect, congratulate the witness, imply confirmation, or allow others to provide information before the statement is preserved. These precautions derive from the 2020 policy review, which also recommends video-recording the entire procedure.
7. Preserve all results and subsequent contacts
The file should preserve original arrays, instructions, timestamps, witness responses, administrator identities, recording files, any interruptions, deviations and later witness communications. Nonidentifications and filler selections should be recorded along with suspect identifications. Investigators should document whether subsequent interviews, media exposure, co-witness discussions, or courtroom preparation occurred, because a later report of certainty is not the same evidentiary event as the initial selection. The integrity of identification evidence depends partly on being able to reconstruct exactly what happened, not merely on receiving a summary that 'the witness identified the suspect.' Wells and colleagues (2020) recommend avoiding repeat identification attempts involving the same suspect and witness.
Why Initial Confidence Can Be Informative—and Why It Can Mislead
Research has significantly refined earlier sweeping claims that eyewitness confidence is unrelated to accuracy. Under well-controlled initial identification conditions, higher confidence in a suspect choice can correspond to higher accuracy at the group level. Wixted and Wells's 2017 synthesis argues that initial confidence obtained under strong procedural safeguards can be informative; a Houston Police Department field study offers field-based evidence that should be interpreted alongside its particular procedures and measurement design. What matters is the conjunction of timing, fairness, independence from suggestion, and the actual identification decision.
This confidence–accuracy relationship is neither perfect nor context-free. High confidence does not eliminate mistaken identification; moderate confidence does not establish innocence. A witness could be certain despite poor viewing conditions, prelineup misinformation, prior exposure to the suspect, an unfair array, or suspect-specific bias. The appropriate statistical question is how identification outcomes and confidence levels relate to correctness in comparable procedures and populations, not whether a witness's personal certainty directly measures the legal truth of a case. Moore and colleagues (2024) emphasize that even seemingly pristine procedures do not produce infallible certainty judgments.
Calibration, diagnosticity, and the role of base rates
Researchers assess confidence using different statistical methods, including confidence–accuracy characteristic curves, calibration analyses, and measures of discriminability. Some methods describe the proportion of correct suspect identifications among responses at a particular confidence level in a study; others quantify how well identification decisions separate target-present from target-absent conditions. These numbers depend on lineup construction, who enters the study, and the proportion of target-present trials. A real investigation has an unknown probability that the police suspect is the perpetrator, and an officer's decision to include a suspect is not random assignment. Consequently, a laboratory percentage cannot simply be imported as the probability of guilt for an identified defendant. The 2017 confidence–accuracy synthesis explains why analytic methods and sampling assumptions matter.
False positives, false negatives, and alternative outcomes
A false positive in the identification context occurs when an innocent suspect is identified. A false negative can occur when the perpetrator is present but the witness fails to identify that person. Filler identifications and correct rejections provide other important information about how the test operated. A particular reform may reduce one type of error while affecting other outcomes, and different evaluation methods may rank procedures differently. The goal is to collect and interpret the full distribution of responses, not to maximize suspect selections. Newer signal-detection analyses of lineup decisions examine how instructions, response thresholds and alternative formats influence the information gained from identifications and rejections.
Confidence judgments in the witness's own language
Statements such as 'fairly sure,' 'I recognize the eyes,' or 'I am sure that is him' may express different degrees or types of recognition. Numerical ratings seem precise but can obscure differences in how witnesses use scales. A witness may also spontaneously include qualifications relating to clothing, hairstyle, lighting, or a partial view. Recording a single investigator-assigned category such as 'high confidence' discards potentially relevant information. The 2024 verbal-confidence study illustrates heterogeneity even within common confidence phrases. A verbatim record and a video of the exchange allow analysts to examine what the statement meant in its original context.
What the Research Supports: Laboratory, Field, and Professional Evidence
Controlled experiments are powerful for identifying causal effects of administrator knowledge and confirmatory feedback because researchers can vary procedures while holding other conditions constant. Their simulated crimes and participant samples also differ from many real investigations, where stress, delay, repeated questioning, publicity, and consequential stakes vary. Field studies provide data from actual eyewitness decisions but can be affected by incomplete ground truth, case selection, local protocols, and the difficulty of establishing whether a suspect identification was correct independently of the identification itself. Evidence appraisal needs both designs rather than treating either as a complete substitute for the other.
The strongest policy basis is convergence across experimental findings, applied research, and consensus reviews. The National Academies' 2014 report recommends double-blind or blinded administration, standardized instructions, and verbatim initial confidence documentation. The American Psychology-Law Society's 2020 scientific review integrates those practices into nine recommendations covering prelineup interviewing, reasonable suspicion, fillers, recording, and avoiding problematic repeated tests. These recommendations are professional guidance about collecting evidence; their adoption as enforceable rules depends on the relevant law and agency policy.
Scientific debate continues over how strongly initial confidence predicts correctness in different conditions, the optimal way of asking for confidence, and how lineup format changes diagnostic value. A 2025 expert-opinion survey reports broad support for blind administration and more nuanced views than earlier surveys on confidence and sequential versus simultaneous presentation. A 2024 review on limits of eyewitness confidence calls for particular care in transferring laboratory patterns to applied cases. Professional agreement supports procedural safeguards without eliminating scientific uncertainty about every outcome.
What Double-Blind Procedures Do Not Solve
Blinding does not reconstruct a face that was poorly seen, resolve obstruction or darkness, restore information forgotten over time, or erase exposure to suggestive material before a lineup. It does not guarantee that fillers were appropriate, that the suspect did not stand out, or that an eyewitness understood the instructions. A witness may have viewed a social-media photograph, heard a name from police, or discussed a possible suspect before any formal array. Those exposures remain relevant even when the eventual administrator is perfectly blind. The National Academies' review treats eyewitness reliability as a product of the full sequence of witnessing and identification conditions.
Nor does prompt confidence collection make every statement equally meaningful. A first statement can still reflect mistaken recognition, an unfair array, or information already acquired before the test. Conversely, delayed confidence can be affected by later experiences without implying intentional dishonesty. The evidentiary question is what information existed at the moment of the initial decision and how the procedure may have shaped that information. A procedurally sound identification can still be wrong, while an identification collected under a flawed protocol may require careful fact-specific examination rather than a simplistic categorical judgment.
Lineup Format Is a Separate Question
Simultaneous lineups show members together; sequential lineups present them one by one. This distinction concerns comparison and decision processes, while double-blind administration concerns the administrator's knowledge and potential influence. Either display format can be conducted blind or nonblind. Evidence about which format is preferable has changed over time and depends on how researchers measure discriminability, decision threshold, and error rates. The National Academies (2014) did not claim that one format had definitively won the comparison, and the 2025 expert survey documents continued differences in expert judgments. The full simultaneous-versus-sequential debate deserves its own analysis; it should not be confused with the separate need to prevent administrator cues.
Law, Police Policy, and Jurisdictional Differences
Psychological research asks whether a procedure reduces bias and improves the informational value of an identification. Law asks additional questions: what agencies must do, whether a violation requires a particular remedy, what evidence a judge may admit, what arguments a party may raise, and what instructions a jury receives. There is no single universal legal rule governing photo arrays and confidence statements across every US state, federal investigation, or national legal system. The procedure's scientific status, an agency's policy obligations, and the court's ultimate ruling must be documented separately.
An example of federal agency policy
A 2017 US Department of Justice announcement described department-wide photo-array procedures instructing federal agents, barring exceptional circumstances, to use blind or blinded administration and to document initial self-reported confidence through recording or near-verbatim transcription. That publication is an archival record of a policy announcement; its existence alone does not prove the current operative version of every component policy in 2026. Researchers and practitioners should consult the applicable agency's present rules when evaluating a contemporary case.
An example of state legislation
North Carolina General Statutes § 15A-284.52 requires specified safeguards for state and local lineup procedures, including independent administration or authorized neutral alternatives, witness instructions, and a confidence statement documented in the witness's own words at identification. The statute also addresses records of identifications and nonidentifications, witness separation, and possible legal consequences for noncompliance. These are jurisdiction-specific legal requirements. They illustrate how scientific recommendations may be converted into operational duties without establishing a universal standard for all courts.
Evidence admissibility and the court's decision
A judge may evaluate suggestiveness, legal admissibility and available remedies using constitutional law, applicable statutes, precedent, and local procedural rules. The jury or other fact-finder may later determine how much weight to give the identification in light of the complete record. Scientific quality does not automatically determine admissibility; admissibility does not certify accuracy; and a credible witness's sincerity does not establish that memory is correct. Forensic testimony should identify the evidence base, procedural deviations, and limitations rather than assign guilt or replace the legal fact-finding function.
How Investigators, Lawyers, and Courts Should Evaluate a Lineup Record
A complete audit asks who selected the suspect, what independent basis existed, what the witness said before the test, how photos and fillers were chosen, who administered the array, what that administrator knew, how instructions were delivered, what the witness saw, what decisions occurred, and exactly when confidence was documented. It also asks about audio-video recording, prior suspect exposure, co-witness contact, later feedback, repeat procedures, and any difference between contemporaneous records and courtroom testimony. These questions form an evidence-review framework, not a mechanical test that yields a legal verdict.
A defense lawyer may scrutinize whether the suspect stood out or an administrator already knew the desired selection; a prosecutor may seek to establish that safeguards were followed and the witness's first statement was reliably preserved; a judge may consider whether any procedural departure has a legally relevant remedy. A forensic psychologist can explain the empirical mechanisms and their applicability, but should not infer the accused person's guilt from a confidence phrase or describe a study's group-level accuracy rate as a calibrated probability for the individual case. The AP-LS scientific review provides a principled framework for examination of procedure.
Special Populations, Language Access, and Fairness
Lineup administration should accommodate witnesses' language comprehension, hearing, vision, age, and communication needs without converting accommodation into suggestion. A verbal confidence prompt and the meaning of numeric percentages may differ among individuals and across languages. Investigators should preserve the original wording and the interpretation process where an interpreter is used. Evidence concerning face recognition across racial or ethnic groups and about age-related variations in performance is relevant to generalization; no individual witness's accuracy can be determined from demographic membership. Selection of fillers and the avoidance of suggestive instructions remain important whatever the witness's background.
Independent oversight and transparency also matter for institutional fairness. A documented, reproducible procedure allows both prosecution and defense to examine how the evidence arose and offers a clearer basis for expert testimony. Training should explain cognitive mechanisms, common sources of contamination, and the reasons for every safeguard; merely checking a 'double-blind' box does not demonstrate competent implementation. Local validation, quality control, and regular review of recordings can identify preventable deviations. An appropriate protocol preserves witness dignity alongside procedural neutrality.
Common Misconceptions
A highly confident witness must be right
High initial confidence can be informative in fair, uncontaminated procedures, but mistaken high-confidence selections occur. The strength of the relationship varies with conditions and study methods. Neither a confidence number nor a witness's sincerity is a stand-alone probability of guilt. The Wixted–Wells synthesis and Moore and colleagues' cautionary analysis should be read together.
If the witness becomes more confident, the evidence has improved
An increase after police feedback, repeated identification, news exposure, or trial preparation can occur without any improvement in original face memory. Later confidence is a new psychological state and should be attributed to the time at which it was expressed. Preserving the initial confidence record protects against retrospective distortion documented in feedback experiments.
A blind administrator makes a lineup error-free
Double-blinding removes an important channel of expectation-based influence. It does not ensure favorable witnessing conditions, fair filler choice, a trustworthy prior description, or independence from earlier contamination. A valid procedural evaluation therefore examines the entire sequence of evidence collection rather than a single checklist item. The National Academies recommend multiple coordinated safeguards.
Any failure to follow best practice automatically makes the evidence inadmissible
Scientific recommendations and legal requirements have different sources of authority. In some jurisdictions particular procedures are required by statute; in others, policies, case law, and facts determine the available remedies. Noncompliance can matter to admissibility or weight, but legal effects are jurisdiction-specific and fact-sensitive. North Carolina's statute supplies one concrete example of specified legal consequences.
Frequently Asked Questions
Is a double-blind lineup the same as a fair lineup?
No. Double-blind refers to the administrator's lack of knowledge of who the suspect is during testing. Lineup fairness concerns suspect and filler composition, instructions, presentation, and the opportunity to reject the array. Strong practice requires attention to both.
When should police ask a witness how sure they are?
Immediately at the first identification decision, using a neutral prompt and documenting the response in the witness's own words before any feedback, confirmation, or discussion of the chosen person. The procedure and question should be preserved in an audio-video or sufficiently detailed contemporaneous record.
Should confidence be recorded as a percentage?
Witnesses sometimes use percentages voluntarily; others use natural-language qualifications. Policies differ on prompts and scales. The minimum evidentiary safeguard is to preserve exactly what was asked and answered, rather than rewriting the response into an investigator's preferred category. Numeric expression should not be mistaken for a validated probability of correctness.
Does double-blinding require a second police officer?
A separate independent administrator is one common implementation. Some policies authorize technically blinded computer or folder methods that prevent the administrator from knowing which picture is being examined. Whether such a method complies with law depends on the applicable jurisdiction and how the method is implemented. The operational requirement is actual protection from information leakage.
Can an eyewitness reject the entire lineup?
Yes. An adequately instructed witness must be able to say the perpetrator is absent or that no identification can be made. Rejection, hesitation, filler choice, and selection are different outcomes and should all be documented accurately.
Can police repeat the lineup if the witness is uncertain?
Repeated identification attempts involving the same witness and suspect may introduce familiarity and source-confusion concerns. The AP-LS recommendations advise avoiding such repeats. Any later procedure must be carefully justified and documented; a fresh choice should not be treated as an independent replication of the original memory test.
Are lineups always conducted sequentially?
No. Simultaneous and sequential procedures are distinct formats, and jurisdictional rules vary. Scientific debates about these formats do not alter the rationale for neutral administration and initial confidence documentation.
Can a confident identification prove criminal guilt?
No. An eyewitness identification is one item of evidence, interpreted with other facts under the applicable legal standard. A confidence statement is a psychological report about a memory decision. Guilt is a legal determination reached through the appropriate justice-system process.
Practical Conclusion
A double-blind lineup and an immediate confidence statement address two linked vulnerabilities in eyewitness evidence: contamination while a decision is being made and distortion in how that decision is later remembered. A strong procedure keeps suspect information away from the administrator, presents a fair array with neutral instructions, accepts all response outcomes, records the witness's exact initial words, prevents confirmatory feedback, and preserves the full sequence for later scrutiny. These practices support better-informed investigation and legal decision-making while maintaining the crucial distinction between a scientifically evaluated identification and an adjudicated fact.
For broader context, see Eyewitness Testimony: How Memory Works in Legal Cases.
Related Articles
References
Bradfield, A. L., Wells, G. L., & Olson, E. A. (2002). The damaging effect of confirming feedback on the relation between eyewitness certainty and identification accuracy. Journal of Applied Psychology, 87(1), 112–120. https://doi.org/10.1037/0021-9010.87.1.112
Kovera, M. B., & Evelo, A. J. (2020). Improving eyewitness-identification evidence through double-blind lineup administration. Current Directions in Psychological Science, 29(6). https://doi.org/10.1177/0963721420969366
Moore, K. N., Pennekamp, P., Yu, C., & Zwemer, D. U. (2024). Eyewitness confidence does not necessarily indicate identification accuracy. Policy Insights from the Behavioral and Brain Sciences, 11(2), 149–155. https://doi.org/10.1177/23727322241268384
National Institute of Justice. (1999). Eyewitness evidence: A guide for law enforcement. US Department of Justice. https://nij.ojp.gov/library/publications/eyewitness-evidence-guide-law-enforcement
National Research Council. (2014). Identifying the culprit: Assessing eyewitness identification. National Academies Press. https://doi.org/10.17226/18891
North Carolina General Assembly. (Current statute). N.C. Gen. Stat. § 15A-284.52, Eyewitness identification reform. https://www3.ncleg.gov/EnactedLegislation/Statutes/HTML/BySection/Chapter_15A/GS_15A-284.52.html
Pennekamp, P., et al. (2024). Variability in verbal eyewitness confidence. Applied Cognitive Psychology, 38(2), e4190. https://doi.org/10.1002/acp.4190
Seale-Carlisle, T. M., Quigley-McBride, A., et al. (2025). New insights on expert opinion about eyewitness memory research. Perspectives on Psychological Science, 20, 903–924. https://doi.org/10.1177/17456916241234837
Tuttle, M. D., Starns, J. J., & Cohen, A. L. (2025). Protecting the innocent in eyewitness identification: An analysis of simultaneous and ranking lineups. Journal of Memory and Language, 140, 104581. https://doi.org/10.1016/j.jml.2024.104581
US Department of Justice. (2017). Justice Department announces department-wide procedures for eyewitness identification. Archival policy announcement. https://www.justice.gov/archives/opa/pr/justice-department-announces-department-wide-procedures-eyewitness-identification
Wells, G. L., & Bradfield, A. L. (1999). Distortions in eyewitnesses' recollections: Can the postidentification-feedback effect be moderated? Psychological Science, 10(2). https://doi.org/10.1111/1467-9280.00121
Wells, G. L., Kovera, M. B., Douglass, A. B., Brewer, N., Meissner, C. A., & Wixted, J. T. (2020). Policy and procedure recommendations for the collection and preservation of eyewitness identification evidence. Law and Human Behavior, 44(1), 3–36. https://doi.org/10.1037/lhb0000359
Wixted, J. T., & Wells, G. L. (2017). The relationship between eyewitness confidence and identification accuracy: A new synthesis. Psychological Science in the Public Interest, 18(1). https://doi.org/10.1177/1529100616686966