An analysis of multiple choices in MSL lineups, and a comparison with simultaneous and sequential ones

Abstract

Simultaneous lineups allow witnesses to compare lineup members, causing excessive mistaken identifications. Levi (1998b) has tested MSL lineups: they are sequential, larger, and allow multiple choices. [The MSL lineup was originally termed a Modified Sequential Lineup (Levi, 1998b). However, there are other modified sequential lineups.]

Each factor decreases mistaken identifications. However, witnesses make fewer single choices of culprits. Sometimes witnesses choose suspects more confidently than any foil. This analysis examines such multiple choices in four experiments. They account for half of multiple choices with culprits. Few foils are chosen, and such responses are rare in culprit-absent lineups, no more than single choices. They are therefore identifications too.

An experiment comparing simultaneous, sequential, and MSL lineups is also reported. The culprit was identified more in simultaneous lineups than in sequential ones. The simultaneous lineup had more mistaken choices than sequential and MSL lineups, whose results were identical. The simultaneous and sequential lineups were equally diagnostic, while the MSL lineup, four times larger, was more than four times more reliable.

Studies 3 and 4 were supported by a grant from the Crime Group of the Department of Criminology, Faculty of Law, The Hebrew University.

Notes

1. Real lineups tend to be unfair because of inappropriate choice of foils (Levi, in press) that enables witnesses to ignore some of them. Thus, for example, if witnesses can ignore two foils in a six-person lineup, the actual lineup size if four, and the chance of choosing the innocent suspect is 1/4, not 1/6. This is still less than when the innocent suspect is chosen as the most similar to the culprit. There are of course lineups (The State of Israel v. Nachmias, 2000) so biased that witnesses can ignore all the foils, in which case we have the experimental situation we have discussed. This paper is based on what best practice can achieve, and lineups can be constructed that are fair.

2. The lack of an effect, the null hypothesis, can of course never be proven. If a difference exists, however, if lacks practical significance.

3. A fifth study was not used in the analysis, since the little relevant data did not seem to warrant the space required to describe it. The experiment compared 20- and 160-person MSL and comparison lineups that allowed multiple choices.