The main question of the thesis is why some people abstain from voting. Meters of literature has been produced on this subject, with problem descriptions similar to the classical why some people vote and others do not . The central place of elections in democracy s continuation and survival defends this interest in turnout.
The thesis describes the central theoretical approaches rational choice and cross- pressure, in addition to main findings from the previous research. The statistical tool used in these analyses is mostly logistic regression analysis. I also include the results from research based on another statistical technique, cluster analysis, which tries to find groups of nonvoters.
From this literature study I make a research design with two separate analyses. Firstly, a logistic regression analysis is conducted, with turnout as dependent variable. The independent variables of the analysis are if the respondent thinks it matter who is governing, attention to campaign news, whether the incumbent government has influenced ones personal economy positively or negatively, whether one thinks there is sufficient attention to the regional composition of the Dutch Parliament, political interest, external and internal political efficacy, political knowledge, political cynicism, party identification and feeling that it is a duty to vote. The logistic regression analysis provides few new insights, most variables are significant, but the results remain vague. Secondly, the same variables are used in a cluster analysis, a technique correlating similar units (in this case nonvoters) rather than similar variables as techniques like regression and factor analysis do. The procedure begins by correlating the two persons with the most similar answers, and continues this procedure until all persons are merged into one cluster. The researcher decides how many clusters to describe in her results. In this thesis, a solution with four clusters was judged the most convenient. Study of the four clusters showed one group of nonvoters characterised by negative views of politics and politicians, no party attachments and little knowledge, interest or self esteem regarding politics. The second group is characterised mainly by the opposite, interest in politics, they are positive about politics and politicians, find it a duty to vote etc. This group to a large extent states lack of time or health problems as their reason not to vote. The third group is characterised by trust in the system and the politicians, but low involvement. The fourth group consisted of only five persons, and was difficult to describe.
Looking at the results provided by the two analyses, I find that the cluster analysis is better suited for studying nonvoting. I think the cluster analysis, looking for patterns among nonvoters is better suited to illuminate electoral abstention than is the logistic regression analysis that only looks for differences between all voters and all nonvoters. For finding such differences it is an excellent method, but such differences do not tell us why some people vote and others do not. Different groups of nonvoters might indicate different ways to the decision not to vote, and thus do tell us why some people do not vote. I recommend the use of that technique in further research, and suggest a nonvoter typology based upon previous studies and my own analysis.