Date of Award

Spring 5-15-2013

Author's School

Graduate School of Arts and Sciences

Author's Department

Political Science

Degree Name

Doctor of Philosophy (PhD)

Degree Type



In this dissertation, I develop and apply sophisticated Bayesian models to the analysis of institutional effects on electoral and legislative behavior in the policy making process. Leveraging the flexibility of Bayesian methods for statistical modeling, I deal with several methodological problems encountered by political scientists, and social scientists in general, in some established research agenda. This dissertation shows the improvement of the ability to evaluate the success of conflicting theories when these methodological issues are properly dealt with.

The consequences of political institutions are investigated at three different levels in this dissertation: countries, political parties, and individual legislators. First of all, at the country level, I investigate whether there is a difference between the performances of democratic and nondemocratic regimes in social provision policy in 18 Latin American countries by focusing on the rarely changing property of political regimes. An appropriate model for the dynamic nature of rarely changing variables is built to thoroughly explore how democratic institutions improve social welfare. Second, at the party level, I develop a Bayesian structural equation model to examine the interdependence between party policy strategies and party support in multiparty systems, in an effort to illustrate the endogenous dynamics of multiparty systems. The results show that party manifestos do not provide clear-cut division of party policy positions. Instead, party labels are more important information than changes in party manifestos to the electorate. Finally, at the level of legislators, I focus on the role of the president and political parties in Brazilian legislative process, in which political exchanges between the government and legislature is an essential feature. By recognizing the existence of the non-ideological effect on voting behavior, I develop a random item-difficulty ideal-point model implied by the spatial voting model to analyze the relationship between coalition dynamics and party-based voting behavior of legislators.


English (en)

Chair and Committee

Jeff Gill

Committee Members

Nan Lin, Tse-min Lin, Jacob Montgomery, Guillermo Rosas, Itai Sened


Permanent URL: