NCJ Number
245540
Journal
Forensic Science International: Genetics Volume: 9 Dated: March 2014 Pages: 93-101
Date Published
March 2014
Length
9 pages
Annotation
This report presents an efficient algorithm for computing an exact distribution of likelihood ratios (LRs) that can be applied to any LR-based model.
Abstract
The distribution of LRs is used in this report to compute a p-value that corresponds to the observed LR. The p-value is the probability that a DNA profile would attain a likelihood ratio that is at least the same magnitude for an unknown person as for the suspect/defendant. The p-value can be viewed as a scaled version of the likelihood ratio, providing a quantitative measure of the strength of the evidence relative to the specified hypotheses and the model used for the analysis. The algorithm is demonstrated with examples based on both simulated and real data. The one example with simulated data illustrates some basic concepts of the algorithm and how the p-value is computed, both with and without consideration of drop-in and drop-out, as well as when accounting for kinship between questioned contributors. One of two examples with real data computes the p-value for a large dataset with 15 loci. The second example with real data compares the exact p-value with results from a simulation-based performance test. Suggestions are offered for future work to improve the algorithm's efficiency. 8 tables and 17 references