ABSTRACT
Although a winner–loser gap in political support is present in most democracies, in some countries losers are more content than in others. Earlier research has demonstrated that formal institutional arrangements may affect the size of the gap. In this article, we argue that the quality and performance of institutions are more important than type of institutional arrangements when it comes to the size of the winner–loser gap. A well-functioning democracy on the input-side of the political system and high quality of government on the output-side significantly narrows the gap in support between electoral winners and losers in European democracies.
Notes
1 A systematic comparative analysis based on status as old or new democracy is unfortunately not possible due to the sample of countries available for analysis. In our sample, we have only eight countries that could be classified as relatively new democracies (the post-communist countries in Central and Eastern Europe).
2 We exclude countries not considered “Free” by Freedom House at the time of the ESS fieldwork. Thus, we do not include Albania (“partly free”), Kosovo (“partly free”), Russia (“not free”) and Ukraine (“partly free”) in this study. See www.freedomhouse.org for ratings and more information about the country rating procedures.
3 We have also estimated the models also without the control for institutional trust. This does not alter the main results concerning the effects of quality of democracy and quality of government on the magnitude of the winner–loser gap, while the effect of winner/loser status on satisfaction with democracy is greater when institutional trust is omitted from the models.
4 The mean value on our measure of the rule of law is 6.4 for the new post-communist democracies compared to 7.8 for the established democracies. The corresponding figures for electoral processes are 7.7 (new) and 7.6 (old). Thus, new democracies show a very good performance in terms of quality of elections, but are lagging behind with regard to quality of government, measured as the rule of law.
5 More information about the system-level control variables included in the analyses can be found in the Appendix.
6 Here we decided to conduct sub-group regressions, such as that we are testing the interactions in two different sub-groups of a third variable, instead of including a three way-interaction term. The main reason behind this strategy is an attempt to decrease the degree of multi-collinearity in the models. We have tested both approaches, which do not affect the coefficients, but due to the very high VIF-values in the three-way interaction model, the standard errors are increasing dramatically.