Department of Statistics Seminar
North Carolina State University

presents

Bruce S. Weir

North Carolina State University

"Can a DNA Profile Be Considered Unique?"

ABSTRACT

The use of DNA profiles for human identification rests on the comparison of the profile from a biological sample (e.g. a stain on clothing) and the profile from a person suspected of having provided that sample. When the profiles are different, the person is excluded as being a contributor to the sample. When the profiles match, however, it has been the practice to calculate the probability of the match being coincidental. In October 1997, the FBI announced that if the population proportion of the matching profile was estimated to be less than 1 in 260 billion then the profile could be regarded as unique.

The question of uniqueness in a forensic context has been considered previously in the statistical literature (e.g. Kingston, JASA 60:70-80, 1965) and was taken in a 1996 report of the United States National Research Council to mean there is a small probability of a DNA profile being present more than once in a population. The accompanying formulation ignored the fact that DNA profiles are genetic, and so are affected by evolutionary history. When two people belong to the same population, their profiles are not independent because of their shared history, and the NRC formulation is overly simplistic.

A statistical genetic treatment of the notion of uniqueness of DNA profiles will be given.

Friday, September 4, 1998

3:35 - 4:35 pm

206 Cox Hall

Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm.