Estimating Party Unity Scores with Formal Support: A New Gamson-Cox Approach

• In Event: Novel Methodologies for Electoral Systems

Abstract

Party unity – broadly defined as a party’s capacity to act as a unified entity – has been key to understanding variations in electoral and legislative outcomes across countries. However, existing measures of party unity are often limited by issues like poor data availability, narrow scope coverage, and/or weak theoretical support. In this paper, I develop a new, improved measure of party unity, building on a novel modified Gamson model of coalition formation (Cox 2021) and using only basic information about an election. Specifically, I use a convex optimization approach and model a party’s cabinet portfolio share as a convex combination of its seat share in the legislature and the size of the cabinet, the weights of which are determined by and thus exploited to estimate the individual unity of each coalition partner. The resulting measure shows good face validity as well as variability within and between countries. I also propose an “honest” interval estimation method to quantify estimation uncertainty, using a procedure that resembles fixed-X regression bootstrap. I apply the recipe to calculate party unity scores with 95% standard intervals for more than 100 parties in 14 European democracies over 60 years in the postwar era. Using the new estimates, I also look at two important questions concerning party unity: the effect of electoral system design on party unity (as outcome), and the effect of party unity on party voting record in the legislature (as factor).

Author

• Haoyu Zhai, NYU Abu Dhabi