Some aspect of plitical participation in Japan is the subject of this paper. Three methods are used to analyse categorical data : D-systems analysis, path analysis using coded “values”, and odds ratios with hierarchical modeling. The first two construct causal systems and pretend no interaction. The last one does not assume causal order or does not mind interactions at all. Women in Japan have disadvantage in socio-economic status, which increases the inherently negative attitude toward political interest. Therefore, political participation is lower on the part of women. The interrelationship among political interest, electoral activities, and party support is shown. That questions posed and answered are of multivariate nature implies the effectiveness of the methods.