Bypassing the legislature: how direct democracy affects substantive and symbolic representation

This dissertation demonstrates that the presence of the initiative process alters voting behavior in the selection of candidates. By allowing constituents to bypass their elected officials to directly select policy, the availability of the initiative leads to substitution between voters' concern over substantive versus symbolic representation. In states with heavy use of the initiative, votes for candidates depend less on the policy positions of the candidates but more on personal characteristics like integrity, attractiveness and descriptive representation. Predictions are derived from a formalization of the theory and are empirically tested using diverse types of data. I demonstrate diminished concern over policy positions through survey data. As use of the initiative increases, a multi-level model demonstrates that votes in that state are less dependent upon the similarity between respondent and candidate policy positions. Increased concern over the integrity of candidates demonstrated through decreased conviction rates for political corruption by the Department of Justice, and increased concern for descriptive representation is demonstrated by a greater balance in the gender of legislators in initiative states. Finally, I find that the personal attractiveness of legislators has a greater effect on votes in initiative states. This theory of substitution depends upon direct democracy leading to more representative policy which is a highly contested claim. This dissertation supports the substitution claim by demonstrating that the initiative does improve representation. I demonstrate how representation should be measured conceptually and statistically, replicate previous models, and then test the theory on an extensive new dataset.