Statistical modelling of experts’ perceptions of the ease of signature simulation

Abstract

The perceived complexity of handwriting traces by forensic experts is a critical element in the process by which opinions regarding the authorship of handwriting are formed. Variations in experts’ perceptions of how complex an image is can significantly impact on the appropriate administration of social justice . There currently exists no test which is available in forensic science which provides a guide for the expert. This study used discriminate function analysis to construct a model which can be used for such a test. The model is based on 13 government forensic experts’ perceptions of how easy or difficult it would be to successfully simulate each of 300 signatures. The variables used by the model to classify these signatures into three complexity groupings were ‘number of turning points’ and ‘number of intersections and retraces’. The test was validated by comparing the model’s calculation of complexity grouping versus fourteen forensic experts’ groupings of an additional 197 signatures. Although substantial variation was found between the experts’ perceptions overall, up to 72.9 % of their perceptions of complexity could be predicted by the model. Misclassification rates were found to be highest when discriminating between signatures where a qualified opinion in the direction of identification would be expressed versus those where a full opinion would be expressed. There was no misclassification associated with signatures where a full opinion would be expressed versus those for which no opinion would be expressed. This test can now be trialed in routine forensic casework and should provide forensic experts with a guide to signature complexity. Research should now be focused on validating the expert perceptions outlined in this paper. Purchase Article - $10