Measuring the goodness of lineups: parameter estimation, question effects, and limits to the mock witness paradigm

All is not as it seems: Avoidable pitfalls in the interpretation of lineup field data.

“…Study B demonstrated the relationship between perpetrator-description detail and measures of lineup fairness (see also, Mansour, Beaudry, Kalmet, Bertrand, & Lindsay, 2017). This finding is not unexpected, given that the “match-to-description” concept—that each lineup member should match the witness’s description of the culprit and that the culprit should not stand out—is the foundation for proper lineup structure and the basis for generating a lineup fairness index (Luus & Wells, 1991; Wells & Bradfield, 1999; Wells, Small, et al, 1998). The more startling aspect of these Chicago data is just how insidious lineup bias was in the live lineups, perhaps because distinctive suspect features are not easily replicated in live lineup fillers.…”

“…If maximum functional size is held to the nominal size of the lineup, the mean functional size drops to 3.10 ( SD = 1.38, MED = 2.75). Mean effective size ( Tredoux E ) of the lineups was 2.94 ( SD = .92), MED = 2.88 (see Wells & Bradfield, 1999 for a discussion of lineup bias measures).…”

Assessment of bias in police lineups.

“…Luus and Wells (1991) explained that a fair lineup can allow police to learn something more from the witness and specifically to test the police hypothesis that the suspect is the guilty perpetrator. Wells and Bradfield (1999) referred to this as the uncertainty-reduction function of a lineup. An eyewitness’s suspect selection from a fair lineup may increase investigator certainty about the offender’s identity, as with the 44% of suspect identifications in the Wells dataset.…”

An examination of criminal face bias in a random sample of police lineups

“…Clearly, replication is warranted with randomly selected police lineups, and we hope that researchers and law enforcement can work together in this regard. Previous studies that have assessed the fairness of police lineups from actual criminal cases in the United States have analysed 10 or fewer non‐randomly selected lineups (Buckhout, Rabinowitz, Alfonso, & Kanellis, 1988; Corey, Malpass, & McQuiston, 1999; Reed, 1984; Wells & Bradfield, 1999); one study analysed 19 non‐randomly selected photospreads (Brigham et al, 1999). Studies conducted in the United Kingdom have used larger samples of randomly selected lineups (Valentine, Harris, Colom Piera, & Darling, 2003; Valentine & Heaton, 1999).…”