Handling finding counts in handwriting analysis - Avoiding the overrepresentation of unusual writing scenarios.

Abstract

In forensic handwriting analysis, it is crucial to understand the relative frequencies of findings relevant to the specific author, especially when using statistical methods. These are factored into the likelihoods used to determine the probabilities for the different authorship hypotheses. However, if ad hoc writings are included in the comparison materials, the representation of a comparison writer's habits can be distorted. An overrepresentation of certain creation time points can be avoided by treating ad hoc series of comparison writing samples as internally homogeneous agglomerates, incorporating only a single value per series into the average relative frequency of a given finding for a comparison writer. Additionally, the proposed approach produces finding counts largely independent of the length of the handwriting sample, which has a positive impact on the efficiency and cost-effectiveness of the expert evaluation.