An investigation of description quality from real-life interviews

Abstract

The quantity and quality of descriptive information about perpetrators makes an important impact on criminal investigations. For over a decade, the Cognitive Interview (CI) technique has received increasing research attention as a method to improve the accuracy and completeness of eyewitness recall, and such research has led to its incorporation into police training programmes. Although the CI has heen reported to improve both the quantity and quality of resultant information, none of the previous studies has measured information quality from a police perspective. The study reported here is an extension of that of George (1991) and investigates the perceived quality of the descriptions of perpetrators from George's study. The quality of such descriptions, as judged by police officers, is compared across four interview groups before and after training. These interview groups are: untrained (control), CI, Conversation Management (CM) and CI + CM. Three measures of quality are investigated, two rating measures and one ranking measure. Unlike the results of George's (1991) quantitative study, the results of this study do not strongly favour the CI. However, results are complicated by the nature of the field study which included certain uncontrollable variables. In particular, differences in the number of perpetrators described had an impact on this study's results. The reasons for the differences between this study's findings and those of quantitative studies are addressed, as well as important issues concerned with field experimentation which are highlighted by the present results.