ABSTRACT
So far, the study has been conducted at an international level, and all countries in the world have constituted units of analysis. When conducting empirical research, the choice of units of analysis often constitutes a stumbling-block for many social scientists. A common mistake is to choose as objects of research countries that are geographically, culturally or mentally close to the researcher. Instead of allowing the research problem to guide the choice of units of analysis, the units of analysis are chosen prior to formulating the research problem. A familiar problem in comparative research is that we have a lot of potential explanatory factors to control for, at the same time as we are operating with a limited number of units of analysis. As a “solution” to this problem, Arend Lijphart (1971: 686–690) has suggested that we should:
increase the number of cases as much as possible;
reduce the “property space” of the analysis, i.e. combine variables that express the same phenomenon;
focus the comparative analysis on “comparable” cases, i.e. variables ought to be dissimilar concerning the phenomenon we want to study and similar in those respects that we want to keep constant, and
focus the comparative analysis on the “key” variables, i.e. variables that are theoretically motivated.
