Abstract
Previous research, using standard regression analyses, suggests that national context affects citizens’ satisfaction with democracy. We note that these standard techniques are inappropriate for examining the interaction between national and individual-level data, and we show empirically that these techniques have in fact led to severely biased findings in past studies. We therefore utilize a multi-level model to reassess existing theoretical claims. We find support for the claim that a consensual political system reduces the difference in democratic satisfaction between winners and losers. We do not find support, however, for existing arguments that citizen satisfaction is influenced by economic growth, corruption, democratic longevity, or changes in political freedoms. More generally, we argue that the bias produced by traditional statistical techniques is so severe that scholars must discontinue their use in future research.
Get full access to this article
View all access options for this article.