Abstract
An archival study was conducted using offender descriptions reported to the police by witnesses (N=29) of the murder of Swedish Foreign Minister Anna Lindh in 2003. All descriptions had been collected within a month after the attack, and each witness had been interviewed between one and five times. Description accuracy was established using photographs of the perpetrator, captured by CCTV cameras minutes before the attack. Contrasting previous archival studies, offender descriptions were quite unreliable (42% of reported attributes were incorrect), and this pattern held for both basic features (e.g., height, age) and more detailed attributes (e.g., clothes). The completeness and accuracy of descriptions increased after (vs. before) images of the perpetrator had been published in the media, but only with regard to the perpetrators' clothes. We acknowledge the potential effects of co-witness influence and post-event information.
Notes
1. Note that the crime took place at the women's clothing section of the department store.
2. Some witnesses were fairly detailed in their descriptions, and a number of categories, thus, contain several details described by a single witness. For example, the attribute ‘trousers’ could contain details of size, color and placement of pockets.
3. Because the frequency data were skewed, nonparametric analyses were performed on the differences between the frequency of reported details before the publication and the cumulative frequency. Parametric analyses (ANOVA) yielded identical results.