ABSTRACT
Different methods of data analysis are used in many aspects of forensic science. Mathematical techniques of dealing with experimental data and specifically statistical and probabilistic analysis of data are frequently used in the interpretation of scientific results produced in a forensic context, including those resulting from glass evidence. This may be to measure associations between two variables, to determine standard deviations, calculate confidence intervals, generate accurate calibration curves, or calculate relative risks and probabilities. In applications to forensic science, usually a mixture of both the classical statistical approach and a probabilistic or Bayesian approach involving likelihood ratios is taken. Other types of data interpretation are also used, including techniques such as factor and cluster analysis. The type of analysis applied depends upon various factors such as:
• the type of data generated (for example, refractive index, elemental concentration)
• the questions being asked (for example, how many fragments should be analysed; do two fragments have a common origin?)
• other relevant background information which may have an influence on the evaluation of the scientific results (for example, persistence and distribution of fragments on clothing).
