ABSTRACT
Abstract. In this article, attention is focused on the analysis of categorical data that is cross-classified; i.e., on the analysis of contingency tables. Measures, models, and graphical displays are considered here for the analysis of nonindependence (or association) in the contingency table. Some of the work done on this topic during the twentieth century is discussed, including work done very early in the century and related work done near the middle and near the end of the century. Brief comments are also included here on the prehistory of this topic (on related work on this topic done in the nineteenth century) and on its future in the twentyfirst century. In addition, a new measure of association is described in this article, and some new methods pertaining to graphical displays are also considered. These new methods can serve, for example, as a necessary correction to methods commonly used now pertaining to graphical displays in correspondence analysis.
