U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Statistical Criterion for the Value of Evidence - Application to the Evaluation of the Results of Paint Spectral Analysis

NCJ Number
77749
Journal
Forensic Science International Volume: 17 Issue: 2 Dated: (March/April 1981) Pages: 101-108
Author(s)
J Zieba; A Pomianowski
Date Published
1981
Length
8 pages
Annotation
This paper describes a method for the quantitative comparison of slightly differing infrared spectra which may be used for the analysis of car paint specimens.
Abstract
Criminal investigations of automobile paints through infrared spectroscopy include a comparison of sample taken from the scene of a crime with a control to establish if the samples came from the same source. However, the interpretation of results may be difficult if the spectra are not different enough, a situation common among car paints since most use similar polymer vehicles which differ only slightly from one another. While particular paints may be different in respect of the sort of pigments, modifiers, and other improvers used, these components do not exceed concentrations of a few percent and are not visible as separate peaks in the spectrum of whole paint since their characteristic absorption bands are masked by the main ingredients. However, they may increase the intensities of some bands of the main components, and it is on this effect and on the effects resulting from resin aging processes that analysis must focus. The method of analysis described here is based on a comparison of the differences in absorbance of some selected band pairs which represent identification features of infrared spectra. An optimal combination of band pairs was chosen using a graphical method based on the validity of Beer's law and a purely statistical method. The spectra characterized by the selected features were compared and the significance of differences between them was tested using the mathematical Hotelling's T-square test. It was found that the method developed enables the quantitative and objective evaluation of the statistical significance of differences between compared spectra. Illustrations, data tables, and a 10-item reference list are included. (Author abstract modified)