Statistical methods and software for risk assessment: applications to a neurophysiological data set

Different statistical methods for the assessment of potential risk factors are discussed, in the case of a complex neurophysiological system, involving binary observations. The first approach describes the use of non-parametric methods, which are based on the cross-product ratio (CPR). The estimates of the CPR are given both in time and frequency domains and their asymptotic distributions are discussed. The second approach is a parametric one and is based on the formulation of a logistic regression model. The estimated parameters and the corresponding odds ratio (OR) can be used to evaluate the behaviour of the system. These methods can be implemented using different statistical software, e.g. S-PLUS, GENSTAT, SPSS, LOGXACT, GLIM. Some of the computational procedures are provided, and the results obtained are clearly displayed. A discussion about the statistical packages is presented. © 2004 Elsevier B.V. All rights reserved.