Do Resemblance Ratings Measure the Accuracy of Facial Approximations?

• NCJ Number: 195241
• Journal: Journal of Forensic Sciences Volume: 47 Issue: 2 Dated: March 2002 Pages: 239-242
• Author(s): Carl N. Stephan
• Date Published: March 2002
• Length: 4 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America
• Note: Portions of this work have been presented at the 9th Biennial Scientific Meeting of the International Association for Craniofacial Identification.

Annotation

This experiment tested the validity of using resemblance ratings to assess the accuracy of forensic facial approximation (FFA).

Abstract

Since forensic facial approximations are used to promote recognition of a deceased person, an accurate FFA should be easily recognized as the person to whom the skull belonged (target individual); however the accuracy of FFAs has been previously assessed by the direct comparison of an FFA to the corresponding target individual for similarity (i.e., a resemblance rating). Resemblance ratings may not indicate a facial approximation's accuracy, since the resemblance of non-target individuals is not taken into account. The current study tested this limitation by determining whether facial approximations correctly identified as the target individual received a higher resemblance rating than those facial approximations that were incorrectly identified as the target individual. Four skulls were approximated with four different techniques of facial approximation: a 3D American sculpting method, a 3D combination sculpting method, a 2D FACE assisted computer method, and a 2D American drawing method. Thirty-seven assessors, with a background in the medical sciences, attempted to identify target individuals from a face pool for each facial approximation. Face pools consisted of 10 photographs. Antemortem photographs were used of the target individuals. Non-target faces in the face pools were of the same sex and approximate age as the target individual. The study found that there was no statistically significant difference between resemblance ratings of FFA's to target individuals and resemblance ratings of FFA's to individuals incorrectly identified as the target individual. The study concluded that it is not possible from resemblance ratings to determine the accuracy and/or quality of a facial approximation, since a non-target individual may receive a resemblance rating equal to, or higher than, the target individual. 1 table, 2 figures, and 21 references