Does the level of turnout matter?

"...does it matter for the prospects of the parties how many people will flock to the polls? "

It is generally expected that turnout in the general election will be somewhat higher than in 2005, when it reached 61%. It is unlikely, though, to reach levels in excess of 70% which were common in the 1990s and earlier.

An increase is expected because the feeling that the election will be one of the closest since a long time, and that there is still everything to play for. The expectation that there is much ‘at stake’ possibly in the form of unprecedented political change will also contribute to a likely increase in turnout.

But does it matter for the prospects of the parties how many people will flock to the polls?

If the tendency to abstain is not the same for the potential supporters of the different parties, the level of turnout will affect election results, in terms of vote shares and possibly also in terms of seats. Politicians and political advisors (and journalists too) are fully aware of the possibility of turnout affecting the outcome.

In many elections we see that politicians who did poorly at the polls point to low turnout amongst their supporters to ‘explain’ the disappointing results. It almost seems as if they derive some solace from this interpretation that says that their supporters stayed home, but did not ‘defect’. But do such explanations hold water? Can the level of turnout be blamed for any particular party’s disappointing electoral showing?

Assessing the consequences of abstentions for the overall result is difficult, because it is impossible to ‘rerun’ reality with different numbers of people going to the polls. Researchers have therefore invented a number of different ways to estimate what would have happened if turnout had been lower than it actually was, or higher.

Reviewing a large number of such analyses Lutz and Marsh conclude that “The main finding ... is that turnout does not matter a great deal, no matter what method, dataset or period of time the authors apply”.

In our own research we asked non-voters what their choice had been had they voted after all. We did so in the context of last year’s European Parliament elections, which are a excellent occasion for such studies as turnout was only 34% (in the UK) as compared to 61% in the 2005 general election.

This 27% difference in turnout should give an excellent opportunity to find whatever effects exist on vote shares. If in the European elections of 2009 the same level of turnout would have been obtained as in the general election of 2005, the vote shares of the various parties would hardly have been different, for each less than half of 1 percent.

Obviously, the effects on vote shares would have been smaller yet for smaller differences in turnout. This finding replicates what had been found earlier for other elections in the UK (see, e.g., van der Eijk and van Egmond, 2007), and for other elections in the western world.

Low levels of participation may be bad for all sorts of things –bad for the development of a feeling of citizenship, bad for the inclusiveness of the political system, and bad for the development and strengthening of partisanship.

It does not mean, however, that the composition of Parliament would be different or the colour of the ensuing government, with different policy majorities, if only more citizens would have been motivated to turn out and vote.

Cees van der Eijk and Eliyahu Sapir