Fuente: JUSTICE & SCIENCE Nº 50 Marzo 2010
R.D. Stoel, E. Van den Heuvel, L.C. Alewijnse
Structural Equation Modelling (SEM; Jöreskog, 1973; Bollen, 1989) has been extensively used and discussed in the behavioural sciences. Roots of this technique date back to the work on path analysis, factor analysis and their combination. The attractiveness of SEM is largely due to its flexibility in specifying and testing complex models concerning the relations among both observed and latent variables. Advantages are certainly not limited to the behavioural sciences, and although much can be achieved by using these techniques in a forensic context their application in this field remains scarce (see for instance Meehan, 2007). In this presentation we will introduce SEM, and illustrate what may be gained by the forensic scientist in applying these techniques, especially in fields that rely on more subjective measurements where reliability and validity is not self-evident. The data of a study performed by Alewijnse et al., (2008) on the subjective complexity judgement of signatures by forensic handwriting examiners will be used to illustrate several applications of SEM. In this study, objective measures of signature complexity (Found and Rogers, 1996) were related to the subjective judgements on signature complexity by the forensic handwriting experts. The presentation starts with a short overview of the substantive and methodological background of the study by Alewijnse et al. Subsequently, we will illustrate how reliability and validity of the complexity measures may be investigated by means of SEM. The presentation will end with a discussion of opportunities for future research in modelling subjective measures, especially with respect to research on the effects of cognitive bias, and methods for minimising them.