Back to Presentations
Forensic match information: exact calculation and applications
M.W. Perlin, "Forensic match information: exact calculation and applications", International Conference on Forensic Inference and Statistics, Minneapolis, MN, 7-Sep-2017.
PowerPoint presentation with live audio recording of Dr. Perlin's talk.
The log likelihood ratio (LLR) summarizes the weight of evidence for one person. An LLR can be compared with the LLRs of all people. Measure theory reduces the people set to a distribution.
The contributor distribution (CD) shows LLRs for people who contributed to the evidence. Average CD value expresses evidence inclusionary power. LLR position in the CD reflects contributor probability. A left CD tail gives false exclusion probability.
The noncontributor distribution (ND) describes LLR values for the population. Average ND value expresses evidence exclusionary power. A right ND tail is the false match probability of inclusionary error, useful to jurors.
The exact LLR distribution is rapidly constructed from independent evidence tests. Information theory permits fast CD or ND determination for one test. Convolution of logarithmic tests yields the joint LLR distribution.
We describe rapid calculation of exact LLR distributions. A DNA case example illustrates forensic applications.