Which Elections Can Be Lost?

TLDR

Electoral competition is a key concept in political science, but how to measure it has received limited direct attention. The authors argue that to determine which elections can be lost and when parties face electoral accountability, scholars should consider the full range of elections where competition is possible, revisiting Sartori’s distinction between competition as a rule and competitiveness as an outcome. They compile a dataset of all national elections from 1945 to 2006, define a minimal set of conditions that make competition possible, critique existing measures for bias, and illustrate how their new metric differs from prior approaches.

Abstract

The concept of electoral competition is relevant to a variety of research agendas in political science, yet the question of how to measure electoral competition has received little direct attention. We revisit the distinction proposed by Giovanni Sartori between competition as a structure or rule of the game and competitiveness as an outcome of that game and argue that to understand which elections can be lost (and therefore when parties and leaders are potentially threatened by electoral accountability), scholars may be better off considering the full range of elections where competition is allowed. We provide a data set of all national elections between 1945 and 2006 and a measure of whether each election event is structured such that the competition is possible. We outline the pitfalls of other measures used by scholars to define the potential for electoral competition and show that such methods can lead to biased or incomplete findings. The new global data on elections and the minimal conditions necessary for electoral competition are introduced, followed by an empirical illustration of the differences between the proposed measure of competition and existing methods used to infer the existence of competition.