The author notices an increased interest in models for qualitative data. This increase in interest is also found with regard to the analysis of traffic safety data. There are two major developments that can be mentioned. First of all, a breakthrough is found in the (almost) classical method of analysing qualitative data, i.e., the analysis of contingency tables. The main reason for this development is the representation of the data by means of the so-called log-linear model this representation results in an analysis of contingency tables similar to the 'analysis of variance' for classified quantitative data. Secondly, we find generalisations of techniques like principal component analysis (factor analysis) and canonical analysis, that makes these techniques applicable to qualitative data. In this first of two articles, the author describes these generalised techniques and gives an example of the application to traffic safety data. In a second article he will give a description of the essential aspects of the log-linear model and its applicability to traffic safety data.
Abstract