An expected cost model of eyewitness identification

Yueran Yang, Laura Smalarz, Sarah A. Moody, Jean J. Cabell, Christopher J. Copp

Research output: Contribution to journalArticlepeer-review

12 Scopus citations


This article presents an expected cost model for evaluating and comparing the performance of eyewitness identification procedures. The model estimates the expected cost of an identification procedure in order to quantify how well the procedure helps the police achieve the investigation goal of identifying and incriminating the culprit. We first apply the expected cost model to analyze five major procedural reforms, including showups versus lineups, filler similarity, administrator influence, lineup instruction, and presentation format. Our analysis reveals that when there is a trade-off between accurate and mistaken identifications, conclusions about procedural superiority depend on the prior probability of guilt and relative costs of different identification outcomes. We then conduct an additional analysis based on a simultaneous consideration of all identification outcomes (i.e., suspect identifications, filler identifications, and rejections). Our analysis shows that assuming different costs for filler identifications and rejections can change conclusions about procedural superiority. We conclude by discussing insights provided by the expected cost model regarding how the legal system can reduce expected costs of eyewitness identification-by changing the conditional probabilities, by reducing the costs of identification outcomes, or by increasing the prior probability of guilt.

Original languageEnglish (US)
Pages (from-to)205-219
Number of pages15
JournalLaw and human behavior
Issue number3
StatePublished - Jun 2019
Externally publishedYes


  • Expected cost
  • Eyewitness identification
  • Prior probability of guilt

ASJC Scopus subject areas

  • Arts and Humanities (miscellaneous)
  • General Psychology
  • Psychiatry and Mental health
  • Law


Dive into the research topics of 'An expected cost model of eyewitness identification'. Together they form a unique fingerprint.

Cite this