Abstract

The present thesis attempts to explain why ``extremist" candidates sometimes win elections against centrist opponents. The thesis proposes two different theoretical models to explain this phenomenon. The first model is a basic Downsian model except voters are uncertain about candidate platforms. This modification makes it possible for voters to choose extreme candidates over more moderate candidates to whom their preferred policy point is closer because of differences in certainty. Interestingly voters may choose extreme candidates either because the extreme candidates position is more or less certain than the moderate to whom they are closer. The second model modifies the basic Downs model to include voter preference over the candidate type, insider or outsider, which candidates can signal by taking more extreme policy positions. I find that when the voter preference for outsiders is either too high or too low both candidates select the median position. For intermediate values they form a separating equilibrium in which outsider type candidates claim an extreme position and insider candidates choose the preferred position of the median voter.