Pueschel, John; (2002) The application of Bayesian hierarchical models to heterogeneous DNA profiling data. Doctoral thesis (Ph.D), UCL (University College London).

Text The_application_of_Bayesian_hi.pdf Download (4MB)

Abstract

A situation is considered in which a suspect has been found whose DNA profile matches that of a sample, assumed to originate from the offender, found at the scene of a crime being investigated. The way in which this evidence should be used is reviewed, highlighting the role of the match probability, the probability of a particular individual having the profile in question given the suspects possession of the profile, and a database of individual profiles. The value of this probability is affected by the heterogeneity of the population, and failure to take account of this could result in a false conviction. A Bayesian hierarchical model designed to represent population substructure is presented. Parameters are clearly defined at each level within a model displaying justifiable conditional independence properties. This model is then used as a basis for inference about the required match probabilities, highlighting errors in previous approaches. As the match probability calculations described are impossible analytically, we use MCMC methods to analyse a UK database of DNA profiles. A comparison of results with those of previous methods highlights the practical importance of a clearly defined hierarchy and conducting the correct conditioning upon the database.

Download activity - last month

Export as XMLJSONCSV

Download activity - last 12 months

Export as XMLJSONCSV

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item