Confidence, Latency, and Accuracy in Eyewitness Identifications Made from Show-Ups: Evidence from the Lab, the Field, and Current Law Enforcement Practices

Project Summary  

Recent evidence has established a strong confidence-accuracy relationship for photo array identifications. However, the relationship between eyewitness confidence and accuracy has not been established for show-ups (a procedure in which a witness views a single suspect). Also, little is known about the role latency (time to make an identification) may play in predicting accuracy in show-up procedures.



This study consists of three phases:  Phase 1 is a replication from a prior lab study that examined the confidence-accuracy relationship in the laboratory. Phase 2 is a field study testing the same question as part of actual police eyewitness procedures. In field studies, we cannot know who the actual perpetrator is, so we will rely on two proxies for ground truth; a previously developed ‘strength of evidence’ scale (Amendola & Slipka, 2009) and a signal detection model for which there is considerable evidence.

Our final study (Phase 3) is a survey of current practices in eyewitness identification that will help update our knowledge of changes in the field related to eyewitness identification practices and the extent to which these are evidence-based. The last known national survey was conducted by the Police Executive Research Forum almost a decade ago. Principal Investigators: Dr. Karen L. Amendola (National Police Foundation) and Maria Valdovinos, MA (National Police Foundation), with input from fellow psychologists Dr. Scott Gronlund (University of Oklahoma) and Dr. Laura Mickes (University of Bristol, UK). All are established researchers in eyewitness memory and law enforcement eyewitness practices.


Analytical Approach

Descriptive statistics (correct IDs, false IDs, correct rejections, and misses) and demographic data will be reported for both the lab and field study. We will measure discriminability and will rely on calibration analysis (CA) and response time accuracy characteristic (RAC) analyses for the lab study. Logistic and multinomial regression models will be developed and applied to the field study data. Mean ratings of evidentiary strength and signal detection modeling will be conducted for each respective proxy for ground truth. Descriptive and bivariate analysis will be conducted on the survey item with weight assignments as necessary to estimate population parameters.



If confidence is predictive of eyewitness accuracy in field show-ups, this could establish the probative nature of confidence statements and strengthen cases while reducing the likelihood of wrongful identifications. Also, by learning more about field practices in eyewitness identification, we can begin to determine the extent to which agencies have assimilated the knowledge generated by scientists and employed evidence-based practices.



This study is being funded by the National Institute of Justice.


Point of Contact

Karen L. Amendola, Ph.D.
Chief Behavioral Scientist
National Police Foundation



eyewitness identification, show-up procedures, photographic lineups, best practices for eyewitness identification, strength of evidence