Writer identification using hand-printed and non-hand-printed questioned documents

Several federal district court judges have recently referred to the purported lack of information on the proficiency of forensic document examiners (FDEs) in identifying writers of hand-printed documents. In order to provide the necessary information, we have re-analyzed data on writer identification that were collected in 1996 from 90 forensic document examiners and 34 laypersons. These data were used previously to assess the proficiency of FDEs using handwritten documents in several different types of writing. In the new analysis we separated data on hand-printed (HP) documents from data on non-hand-printed (NHP) documents and compiled error rates and statistics in each category. The main findings are: (1) whether or not the documents were hand-printed, the performance of FDEs was much better than that of laypersons; (2) statistical tests found no difference between the data provided by the FDEs in the HP and NHP categories; (3) statistical tests found no difference between the data provided by laypersons in the HP and NHP categories; and (4) statistical tests found differences between the data provided by the FDEs and the laypersons in both the HP and NHP categories. Similar results were obtained when hand-printed documents were compared to cursive documents and when cursive documents were compared to non-cursive documents. All the evidence indicates that in our proficiency test the performance of FDEs in writer identification was much better than the performance of laypersons in each one of the following document categories: (1) hand-printed; (2) non-hand-printed; (3) cursive; and (4) non-cursive.