Institutional Determinants of Parliamentary Party Unity: A Multi-Level Analysis

For too long, political science scholars saw parliamentary parties as unitary actors. That has started to change: recent contributions explain how parliamentary systems maintain overall high voting unity but occasionally feature dissent. Now, the field of study faces the next challenge: parliaments are treated as homogeneous environments, in which parties behave the same, regardless of differing institutional contexts. These differences, and their effect on party unity, have not been systematically explored yet. Cross-national comparisons of legislatures are still in their infancy; previous work has mostly focussed on either highly aggregate, system-level differences between presidential and parliamentary systems, or dived into highly specific contexts whose conclusions cannot easily be generalised to the whole set of parliamentary democracies. This poses two problems. Firstly, other levels of analysis that are not accounted for will bias the results, as their context is not controlled for in a single-level model. Secondly, the different levels of research did not concentrate on differences between parliamentary systems. Consequently, our systematic understanding of how parliaments work and operate, and how differences in so doing affect legislative behaviour at various levels, is limited. In addition, data on legislative voting outside the United States was, until now, hard to come by. The purpose of this thesis, thus, is to answer the question: to what extent do differences in institutions and rules amongst parliaments affect party unity, compared across systems, parties, votes, and individual legislators? In other words, what particular environment incentivises or discourages a legislator from dissenting from the rest of his parliamentary party? The use of a multi-level model will reduce the risk of an atomistic or environmental fallacy, and enable the thesis to understand institutional effects on party unity across different levels of comparison. This can range from how powerful committees are in the legislative process (a system-level explanation), through whether a party is in opposition or not (party-level), down to how senior a legislator is within the party and what prospects for promotion she has (individual-level). Using advanced web-scraping techniques, the thesis will collate a rich dataset of legislators' recorded votes from a variety of parliamentary systems. This dataset will be extended with political and personal data across all levels through careful analysis of parliamentary processes, party structures and individual biographies; it shall then be used to test hypotheses about the effect of institutions at all four levels, as generated by the expectations of the theoretical framework. The unique contribution of this thesis, therefore, lies in multiple fields: it creates a new, comprehensive data set, as well as a framework of parliamentary behaviour that embeds past work into a multilevel theory.
The proposed D.Phil. thesis will examine the effects of legislative institutions on legislators' voting behaviour. It intends to contribute to the field of legislative studies in multiple ways. First, it will develop a new, multi-level framework that takes into account how differences between parliamentary systems, parties, votes and individuals matter for a legislator's probability to vote against her own party. This framework will fill a substantial gap in comparative studies of parliaments. Second, it will test this demanding framework with a new, large dataset using the advent of recent open parliamentary data resources and advanced data scraping techniques. The framework will further be defended against two rival explanations, which deem only party cohesion and re-election chances relevant.