Week 2 Graham Davies Eyewitness Identification and Composite

Week 2 Graham Davies Eyewitness Identification and Composite Production

MISTAKEN IDENTIFICATION • In English law, one positive identification by a confident witness is sufficient to convict an accused • English criminal history is littered with cases where miscarriages of justice have occurred because of honest but mistaken identification • One of the most astonishing and instructive cases is that of a Hungarian émigré called Laslo Virag…

THE DEVLIN REPORT (1976) • Devlin Recommendations Caution to witnesses Judges’ summing up No ‘identification only’ cases • ‘Turnbull’ Judgements (1977) ‘Good’ versus ‘Poor’ quality identifications • ‘PACE’ Guidelines

IDENTIFICATION PARADES TODAY • - Styles of Identification permitted in UK A parade Group identification Confrontation - Video identification Frequency and costs of parades (£ 14 M) - Delay and exposure time (Slater, 1994) - Outcome by witness age (Pike et al, 2002)

RATES OF IDENTIFICATION IN LONDON IN 1992 (Wright & Mc. Daid, 1996) Picked suspect 611 (39%) Picked foil 310 (20%) No ID 640 (41%) TOTAL 1561 % of Violent Crimes 71% % of White Suspects 57%

RESEARCH ON IDENTIFICATION PARADES (OR ‘LINE-UPS’) “Research should be directed to establishing ways in which the insights of psychology can be brought to bear on the conduct of parades and the practice of the courts” (Devlin, 1976) • Wells on ‘System’ versus ‘Estimator’ variables

‘ESTIMATOR’ VARIABLES • Impact of delay (Shepherd et al. 1982) - The Demjanjuk case (Wagenaar, 1988) • Cross-Race Identification (Lindsay & Wells, 1983)

‘SYSTEM’ VARIABLES • Exposure to photographs (Gorenstein & Ellsworth, 1980) - The George Ince case • Concept of ‘Functional Size’ (Wells et al. 1994) • Positive feedback on performance (Wells & Bradfield, 1997)

CAN ACCURACY BE IMPROVED ? • Confidence not a reliable guide (Penrod et al. , 1982) • Simultaneous versus successive line-ups (Wells et al. , 2000) - Absolute vs relational judgements (Stern & Dunning, 1994) • VIPER parades (Valentine, 2002) - 51% of all live parades cancelled - 39% selection from VIPER; 35% live

COMPOSITES • Old technology – Identi. Kit – Photofit – Police Artists • New technology – E-fit – Comp. Photofit • But how accurate?

EXPERIMENTAL STUDY OF PHOTOFIT (Christie et al. 1981) From View From Memory Identified 37% 23% Type likeness 54% 48%

FIELD STUDY OF PHOTOFIT (Home Office, 1976) ‘Entirely responsible’ 5% ‘Very useful’ 17% ‘Useful’ 33% ‘Not very useful’ 20% ‘No use at all’ 25%

WHY ARE TRADITIONAL COMPOSITES SUCH A POOR GUIDE TO LIKENESS? • Selective attention (Ellis et al. 1975) • Limited range of features (Shepherd et al. 1980) • Compatibility with encoding method (Davies & Christie, 1982)

ARE THERE BETTER METHODS ? • Computer-based systems - Lots of features - Whole-face construction - ‘Lifelike’ appearance • But only more accurate when composite made in the presence of the target (Davies et al. , 2000) • ‘Breeding’ Faces –Evo. FIT (Frowd et al. , 2005) • Machine Recognition: the Newham ‘Experiment’