OnPolicing Blog

Eyewitness identification remains one of the most powerful—and most scrutinized—forms of evidence in the criminal justice system. From photo arrays to live showups, identification procedures play a pivotal role in investigations, prosecutions, and court outcomes. Yet despite decades of scientific research, law enforcement policies and field practices, and courts have not always been aligned with the best available evidence.

New findings from the Eyewitness Identification Research Project (2025) shed critical light on how law enforcement agencies (LEAs) across the United States are conducting eyewitness identification procedures, where knowledge gaps persist, and when witness confidence—captured exclusively at the first viewing of a photo lineup or showup—is associated with high levels of accuracy.

This article translates those findings into actionable insights for practitioners, trainers, and decision-makers seeking to strengthen investigative integrity, reduce wrongful convictions, and enhance public trust. Importantly, most prosecutors know that it is unwise for police or prosecutors to rely solely on eyewitness evidence in establishing guilt or innocence. Even with the best safeguards in place, people make mistakes, and sometimes do not have sufficient opportunity or time to encode and retain important details of criminal perpetrators. Similarly, just because a person fails to make an identification, does not mean the person was not the perpetrator. Corroborating evidence and the reliance on multiple witnesses’ accounts (when available) are essential.

Why Eyewitness Identification Still Matters

Eyewitness testimony has long been a cornerstone of criminal investigations. However, it has also been a leading contributing factor in wrongful convictions, particularly when identification procedures are improperly administered or biased.

Scientific research over the last 30 years has demonstrated that:

• Obtaining confidence statements other than during the initial identification is considerably less reliable and should be avoided
• Bias can be reduced through a variety of practices and safeguards
• Providing clear and cautionary instructions to witnesses improves accuracy

Despite this, implementation of evidence-based practices across law enforcement has been inconsistent. The NPI’s 2025 research confirms that the research–practice gap remains a central challenge.

How Law Enforcement Agencies Use Identification Procedures Today

The national survey of law enforcement agencies reveals clear patterns in current practice:

Photo Arrays vs. Showups

• 86% of LEAs (local police and sheriffs, as well as state agencies) conduct photo arrays
• 64% conduct live showups, though usage rises to 70% when considering local agencies alone

Photo arrays are more widely used, likely due to their structured nature and longstanding acceptance. Showups—where a single suspect is presented to a witness shortly after an incident—are often viewed as more suggestive, and even discouraged by some, which is likely due to both a lack of understanding about how showups are conducted in the field and the fact that suggestiveness can easily be reduced through proper administration and instruction as it is in photo lineups.

Initial Confidence Statements: A Mixed Picture

• 84% of agencies collect confidence statements for photo arrays
• Only 72% do so for showups, a notable decline from PERF’s 2013 study

This gap is concerning, given strong scientific evidence that initial witness confidence is one of the most powerful predictors of identification accuracy, regardless of procedure type, though understandable given the prior lack of research on showup procdures.

Confidence Statements: The Most Underutilized Tool in Eyewitness Identification

One of the most important findings from the research is that more than half of responding agencies were unaware of the strong predictive relationship between initial confidence and accuracy, and that was even lower for state law enforcement agencies (27%), and mid- to large-sized agencies (less than 30%) as compare to small agencies (54%).

Why Initial Confidence Matters

Scientific studies consistently show that:

• Witness confidence recorded immediately after identification is strongly predictive of identification accuracy.
• Initial identifications made with high-confidence (possibly even moderate confidence, as we found) are more likely to be accurate.
• Later expressions of confidence can be inflated by media and social media coverage, newly learned details about the suspect, repeated discussion of the identification, and feedback suggesting they picked the right person, and therefore should be avoided at all costs.

Sequential vs. Simultaneous Photo Arrays: Rethinking Old Assumptions

For years, the sequential photo array (viewing photos one at a time) was widely promoted as a superior method to reduce false identifications. However, the science has evolved and this is no longer considered the best method.

What Agencies Are Doing

• 61% of LEAs use sequential presentation
• 24% use simultaneous presentation
• 13% allow either method

What the Research Shows

The National Research Council (2014) and subsequent studies found:

• No advantage to sequential presentation
• A slight advantage for simultaneous photo arrays
• Both methods are valid when safeguards are in place

Reflecting this, the International Association of Chiefs of Police (IACP) updated its guidance to explicitly state that neither method is recommended over the other. This highlights a recurring issue: field practices often lag behind updated scientific evidence, even when national model policies change or new evidence emerges, as it has in this case.

Showups Are Not the Problem, Poor Safeguards Are

Showups have long been criticized as inherently suggestive. However, the 2025 research challenges that assumption.

Key Findings on Showups

• There is little empirical support for avoiding showups altogether
• Showups can be highly accurate when conducted properly
• Initial confidence statements in showups are strongly related to accuracy
• Assumptions about suggestiveness are likely due to misunderstanding about how showups are conducted and an assumption that a patrol officer would automatically assume they have found the “culprit” when responding to BOLO alerts.

As part of our project, both the laboratory experiment (conducted in the UK) and U.S. field studies across two jurisdictions, resulted in significant support for confidence–accuracy relationships in showups, something usually reported only for photo arrays. The takeaway is clear: suggestibility is not unique to showups. Any identification procedure can be biased if safeguards are absent.

Why Science Takes So Long to Reach the Field

If the research is so clear, why hasn’t it been universally adopted? The study identified several structural barriers:

1. The Research–Practice Time Lag

In many fields, including medicine, it takes 15 years or more for scientific findings to influence everyday practice.

2. Different Goals, Different Languages

• Researchers prioritize precision and theory
• Practitioners prioritize efficiency and case resolution
• Opportunities for meaningful interaction are limited

3. Knowledge and Experience Gaps

Law enforcement professionals may have limited exposure to peer-reviewed research, while scientists may lack understanding of operational realities.

4. Inconsistent Findings

Earlier studies sometimes produced conflicting results, often due to weaker methodologies or less precise measures. Together, these factors create a system where policy evolves slowly, even when evidence is strong or alters prior knowledge.

What This Means for Law Enforcement Leaders and Trainers

The implications of this research are practical, not theoretical.

Policy Recommendations

Agencies should:

• Require immediate confidence statements for all identification procedures
• Ensure confidence is recorded verbatim, without providing any feedback
• Consider showups as valid tools when proper safeguards are applied

Training Implications

Training should emphasize:

• The importance of initial confidence
• Proper documentation procedures
• Avoidance of suggestive language or cues
• Departmental policies (as they are often informed by best practices research, prosecutor and state attorney general guidelines)

Organizational Leadership

Leaders play a critical role by:

• Encouraging evidence-based policing
• Supporting ongoing professional education
• Bridging communication between researchers and practitioners

Moving Forward: Translating Research into Practice

Eyewitness identification will always involve human memory, and therefore, human fallibility. But this does not mean agencies are powerless or eyewitness identifications are not useful. The 2025 findings make it clear that small procedural improvements can have a significant impact on accuracy, fairness, and legitimacy.

By closing the gap between science and practice, law enforcement agencies can:

• Strengthen investigations
• Reduce wrongful convictions
• Improve courtroom outcomes
• Enhance community trust

This is the most up-to-date evidence. The challenge now is proper implementation.

Research contributors:
Maria Valdovinos Olson, PhD; Curt Carlson, PhD; Scott Grunland, PhD; Laura Mickes, PhD; Jie Gao, MS; Yukun Yang, Colby Dolly, PhD

This project was supported by Award No. 2019-75-CX-0009, awarded by the National Institute of Justice (NIJ) , Office of Justice Programs, U.S. Department of Justice to the National Policing Institute. The opinions, findings, and conclusions or recommendations expressed in this publication are those of the authors and do not necessarily reflect those of the Department of Justice of National Policing Institute. This report serves as a follow-up to the 2013 report by the Police Executive Research Forum (PERF) also funded by the NIJ, Grant No. 2010-IJ-CX0032.