A new algorithm for estimating voter turnout when the number of ballot papers issued is unknown

Abstract

Methods for estimating turnout in multi-member plurality (MMP) elections where the number of ballot papers issued is unknown tend to focus upon the distribution of votes between parties. This paper shows that existing methods may lead to significant errors in turnout estimates for English local government elections. We introduce a new method of calculating voter turnout that uses not only the total number of votes cast, but also information relating to the presence or absence of candidates and the distribution of votes between candidates of the same party. This method not only appears to improve substantially upon previously used algorithms, but also demonstrates that partial abstention is more likely when voters are presented with fewer of their chosen party's candidates than the number of seats at stake. We believe that this new algorithm may have general application to MMP elections.