Does filler database size influence identification accuracy?

Area of Science:

• Forensic Psychology
• Eyewitness Identification Research

Background:

• Police increasingly use facial recognition software and large photo databases for lineup construction.
• The impact of large filler databases on eyewitness accuracy remains largely unexplored in research.

Purpose of the Study:

• To investigate the effect of filler database size on eyewitness identification accuracy.
• To examine the relationship between algorithmic similarity, database size, and identification outcomes.

Main Methods:

• Conducted a facial recognition experiment using filler databases of varying sizes (5,000, 25,000, 125,000 photos).
• Lineup fillers were selected using facial matching software in a legally realistic manner.
• Assessed objective similarity between suspects and fillers and analyzed identification accuracy rates.

Main Results:

• Larger databases were associated with higher suspect-filler similarity and lower overall identification accuracy.
• In target-present lineups, larger databases reduced correct identifications and increased innocent filler selections.
• In target-absent lineups, larger databases led to fewer correct rejections and more filler identifications.

Conclusions:

• Facial matching software used with large photo databases may reduce eyewitness identification accuracy.
• Filler database size is a significant system variable influencing eyewitness reliability in lineups.