Three essays on measuring political behaviour

Thesis

Fundamental issues of political representation persist in modern democratic states. Policies favoured by large majorities of the population are not acted upon by elected representatives. Direct forms of democratic policymaking are increasingly used to settle policy disputes. And the amount of money in politics continues to grow. At the same time, new sources of data allow us to test the political behaviour of democratic actors in novel ways. Across three papers, this thesis explores how representation and policymaking operate in modern democratic systems and how researchers should measure political behaviour in these contexts.

In Chapter 2, I ask why popular legislation is passed by direct democratic means instead of by the legislature? I use a combination of district-level voting data, ideological estimates of financial donors, and a survey of state legislators to test competing theories of ballot initiative success in the United States. I argue that direct democratic policymaking is necessary when issues lie outside mainstream policy networks. Legislators appear averse to policies that are novel, untested, or go against the preferences of their financial supporters. In Chapter 3, I show how experimentalists can make valid statistical inferences when their data contains group-level structures. I illustrate this guidance using both Monte Carlo simulations and a review of experimental publications over the past three years. In Chapter 4, I examine how campaign finance disclosure affects vote choice in both representative and direct democratic contexts. Applying the lessons of the previous chapter, I use conjoint experiments to show that while voters can use disclosure information to inform vote choice, these effects are crowded out once partisan, ideological and issue cues are made explicit. Across all three chapters, I develop inferential techniques designed to account for limitations of observed data and, in particular, incomplete data.

Authors: Robinson, T

• DOI: 10.5287/ora-zbmdnwn9w
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: quantitative; experiments; reponsiveness; variance; campaign finance; measurement; representation
• Subjects: Political Science