Description of original award (Fiscal Year 2017, $464,910)
As submitted by the proposer:
As with many of the forensic disciplines that rely on feature-comparison methods, there is no gold-standard against which to test accuracy of handwriting examination. The present study builds upon laboratory-based research on handwriting motor control to examine whether questioned handwriting with uncertain writership can be distinguished with high-degrees of accuracy by independent dynamic features characterizing handwriting strokes.
The proposed research is the first to cross-validate expert examiner determinations of writership using an established approach to extract specific geometric, spatial, and kinematic features of cursive handwriting and hand printing.
Work in the researchers laboratories has led to the development of an independent method of quantifying handwriting features that can now be used to validate writership conclusions reached by expert examiners. Handwriting and hand printing samples will be obtained from writers using a digitizing tablet and published laboratory practices to record dynamic features of handwriting strokes. Over 1,200 high-resolution samples will be subjected to an automated computer vision algorithm to identify pairs of phrases that are visually similar in form and style.
Expert document examiners will be asked to evaluate these pairs using a standard numeric scale of confidence with regard to two propositions: (1) the questioned and known samples were written by the same writer; and (2) the questioned and known samples were written by different writers.
Each handwriting stroke will be subjected to analysis of kinematic (e.g. velocity, amplitude, pen contact duration), geometric (e.g. slant, straightness) and pen pressure from which a feature difference score (feature for known feature for questioned sample) will be calculated. Lower scores reflect greater similarity in feature sets between the known and questioned sample. Analyses will model the relationships between the expert opinion of writership and the difference scores for cursive writing and printing.
It is hypothesized that there will be a significant negative relationship between feature difference scores and level of confidence expressed by expert examiners that two handwriting samples were authored by a single writer. Erroneous opinions should not show this relationship.
This study will provide empirical evidence to support the validity of expert opinions as required for admissibility in courts of law under Daubert.
Note: This project contains a research and/or development component, as defined in applicable law, and complies with Part 200 Uniform Requirements - 2 CFR 200.210(a)(14).