ABSTRACT
So far, I have concentrated on identifying bivariate relations between the independent and the dependent variables. Along the way, many significant associations between various independent variables and the dependent variable have been detected. However, it is hazardous to draw any far-reaching conclusions on the determinants of the death penalty on the basis of these bivariate analyses. Many of the independent variables dealt with so far can be expected to be interrelated and I shall therefore proceed by assessing the relative impact of the independent variables studied so far. This is done by means of multiple regression analyses. Since I have made use of a substantial number of explanatory variables, it is necessary to reduce the number of independent variables in order to avoid making the regression models unstable. Another reason for cutting down the number of independent variables is the expectation already put forward that quite a few of the variables interact.
