Statistical considerations when determining measures of lineup size and lineup bias

Much eyewitness research has centred on the development and evaluation of measures of lineup properties. These measures have proved useful to researchers and to expert witnesses who testify on eyewitness testimony. However, the inferential statistical properties of the measures themselves are rarely taken into account, despite the fact that most of them are derived from the mock witness task, which relies on an implicit probability model. This is a failing, and it is argued that estimates of lineup properties reported without inferential considerations may be misleading. Methods are suggested here for reasoning inferentially about lineup measures, and for planning studies to ensure adequate statistical power. Suggestions made by eyewitness researchers regarding the practical interpretation of lineup measures are critically evaluated, as are recommendations for new lineup measures, as suggested by Lindsay et al. (this issue).