Seth Masket wonders whether Duverger got the 2010 UK elections right. Matt Singer, an assistant professor at the University of Connecticut, sent me the following analysis that answers this question with a resounding NO.
Duverger famously predicted that plurality rule elections in single member districts will generate two-party competition (aka Duverger’s law). There is an ongoing debate, however, over whether this law accurately describes election outcomes at the constituency level. For example, Chibber and Kollman (2004 chapter 2) argue that the central tendency in the UK, US, India, and Canada is two-party competition while Gaines (1999) and Diwaker (2007) document numerous divergences from two-party competition in Canada and India respectively (see also Grofman et al 2009). In a broader sample of countries using plurality rule, Laura Stephenson and I (2009 Electoral Studies) find that the average effective number of parties getting votes in plurality districts is right around 2 after controlling for demographic and institutional factors.
The 2010 UK elections, however, represent a divergence from Duvergerian competition even at the district level. Below I plot a histogram of the effective number of parties getting votes (ENPV) for the 649 constituencies contested last week, using data compiled by the Guardian. The average district saw an effective number of parties of 2.98. Only 12% of districts had less than even 2.5 effective parties. Electoral fragmentation was slightly higher outside of England (the average districts in Scotland, Wales, and Northern Ireland had 3.04, 3.33, and 3.23 effective parties respectively) than within it but even in English districts the average outcome was 2.92 effective parties. These data thus show that the 2010 elections strongly contradict Duverger’s law.
The 2010 elections continue a recent trend of electoral fragmentation in Britain. If we consider elections since 1918, there has been a general increase in electoral fragmentation that begins with the elections that led to the last hung parliament and has accelerated over the previous 3 election cycles (data compiled from CLEA, Iain Outlaw, and the Electoral Commission).
The result is that if we consider the overall distribution of plurality district competition in the UK over the 1918-2010 period, there is a bimodal distribution, with one mode centered at 2 parties and the other at around 2.6 parties (please feel free to email me if we you want more charts). The latter mode represents the central tendency since 1974. Thus the United Kingdom seems to be another case where plurality rules are not generating sufficient strategic voting to yield local two-party competition.
Matt Singer is an assistant professor of political science at the University of Connecticut.
ps. Henry of course wrote a post with almost the same title a few days back referring to Matthew Shugart’s post.