MDS use with crime scene data replicates poorly: response to Goodwill, Alison, and Humann (this issue) and Davis (this issue)

Abstract

In a previous article (Sturidsson et al., Psychology, Crime & Law, 12, 221–230, 2006), we failed to replicate a study utilizing multidimensional scaling (MDS) methodology with crime scene data from cases of assault rape with unknown offenders (Canter & Heritage, Journal of Forensic Psychiatry, 1, 185–212, 1990). Our study attracted critique related to the choice of MDS methodology. This response contains additional results from reanalyses of our data, obtained by carefully following the suggestions from the critics. However, the recommended reanalyses did not yield results substantially better than those from our prior attempt to replicate the findings by Center and Heritage (1990). We conclude that replication this far has failed, and our concerns regarding the uncritical, widespread use of MDS methodology for forensic psychology applications remain.

Keywords:

• multidimensional scaling
• offender profiling
• PROXSCAL
• ALSCAL
• MDS