The evaluation of evidence for microspectrophotometry data using functional data analysis

Colin Aitken; Ya-Ting Chang; Patrick Buzzini; Grzegorz Zadora; Genevieve Massonnet

doi:10.1016/j.forsciint.2019.110007

The evaluation of evidence for microspectrophotometry data using functional data analysis

Forensic Sci Int. 2019 Dec:305:110007. doi: 10.1016/j.forsciint.2019.110007. Epub 2019 Oct 28.

Authors

Colin Aitken¹, Ya-Ting Chang², Patrick Buzzini³, Grzegorz Zadora⁴, Genevieve Massonnet⁵

Affiliations

¹ School of Mathematics and Maxwell Institute, University of Edinburgh, Peter Guthrie Tait Road, Edinburgh EH9 3FD, UK. Electronic address: c.g.g.aitken@ed.ac.uk.
² School of Mathematics and Maxwell Institute, University of Edinburgh, Peter Guthrie Tait Road, Edinburgh EH9 3FD, UK. Electronic address: ytchang.sabrina@ed.ac.uk.
³ Department of Forensic Science, Sam Houston State University, Huntsville, TX, USA. Electronic address: patrick.buzzini@shsu.edu.
⁴ Institute of Forensic Research, Krakow, Poland; Institute of Chemistry, The University of Silesia in Katowice, Katowice, Poland. Electronic address: gzadora@ies.krakow.pl.
⁵ School of Criminal Justice, University of Lausanne, Switzerland. Electronic address: genevieve.massonet@unil.ch.

PMID: 31756683
DOI: 10.1016/j.forsciint.2019.110007

Abstract

Microspectrophotometry data arise in the study of many forensically applicable situations. The situations here are those of ink and fibres. In a criminal investigation, data associated with a crime scene are compared with data associated with a person of interest. Methods based on the likelihood ratio are often used to evaluate such evidence. A technique known as functional data analysis for determining likelihood ratios using the full spectrum is described. It provides support comparing a proposition of common source with a proposition of different sources for data from the crime scene and from the person of interest. Data are available from ink, woollen and cotton fibres. The effectiveness of the method is assessed using false positive and false negative rates and Tippett plots in the comparison of data from known sources.

Keywords: Evidence evaluation; Functional data analysis; Likelihood ratio; Microspectrophotometry.