by Stephen Fisher

It has been reported that David Cameron called for tactical voting in The Sunday Times today (e.g. here and here). He didn’t. What he said was:

“So if you are considering voting Ukip or Lib Dem, I urge you to think of the chaos of a weak Ed Miliband, propped up by Nicola Sturgeon demanding ever more borrowing and more taxes. Only a Conservative vote in your local constituency will keep Ed Miliband and the SNP out and secure Britain’s future.”

This is a blanket call for all UKIP and Lib Dem supporters to vote Tory. Not only is there nothing tactical in this, but the final claim is both false and contrary to the logic of traditional tactical voting.

For those living in constituencies where Labour and the Liberal Democrats are expected to share first and second place, tactical voting by Conservative supporters in favour of the Lib Dems increases the chances that another Con-Lib coalition would have a majority.

By contrast Polly Toynbee in the Guardian on Thursday made a well argued case for classic anti-Tory tactical voting by Green and Labour supporters. She rightly points out that tactical voting by Green supporters for Labour in Con-Lab and Lab-Con marginals could be critical to the outcome. Similarly she advocates Labour and Green supporters both vote Liberal Democrat in contests dominated by the Conservatives and Liberal Democrats.

Traditional tactical voting in our first-past-the-post system involves supporters of parties coming third or lower in their constituency trying to better influence the outcome by voting for one of the top two candidates. Since 1992 about 8% of voters have been doing this kind of tactical voting. This constitutes about 21% of all third and lower placed party supporters (see my working paper for more details.)

Tactical voting might go up at this election since it looks like there will be more people who support a party coming third or lower in their constituency.

On the other hand it is not clear that third-party supporters see as much difference between the top two as they used to. I argued here that Conservative and Lib Dem supporters in Scotland probably would not often vote tactically for Labour because they are not much more fond of Labour than they are of the SNP.

A recent Channel 4 News/YouGov poll suggested otherwise, reporting that potentially 9 Labour and 2 Lib Dem seats could saved from the SNP by tactical voting between unionist parties. The poll used questions that strongly prompted tactical voting, such as the following:

“Imagine that in your constituency at the general election only Labour and the Scottish National Party had a realistic chance of winning. How would you then vote?”

Questions like this force respondents to assume they know for sure what the strategic situation is in their constituency when in fact many are unclear.

By contrast Lord Ashcroft’s constituency polls prompt people to think about the candidates in their constituency without informing them of the likely order of finish. These polls typically do not show much sign of switching between general and constituency specific vote intention, except in Liberal Democrat seats where it is unclear whether the switching is tactical or due to the personal popularity of the sitting MP. But while Ashcroft polls provide some succour for the Liberal Democrats in England, they do not suggest that tactical voting in Scotland will necessarily save Labour or the Liberal Democrats any seats.

The methodological differences between the two types of polls illustrates an important substantive point. Uncertainty over the expected order of finish of the candidates in the constituency is important in real world tactical decisions.

Uncertainty leads to lots of mistakes: tactical voting away from one of the top two candidates in favour of a lower placed one.

Also, uncertainty over local candidate order dampens the overall level of tactical voting. Sensibly, people are less likely to vote tactically if they are not sure who will be first and second in their constituency. Given the huge changes in overall levels of party support for the Liberal Democrats, SNP and UKIP since the last election, voters are likely to be much more unsure of the parties relative standings in their constituency. Some may be aware of constituency polls and forecasts, but local parties often muddy the waters.

Since some factors suggest an increase and some a decline, we will have to wait for the post election surveys to know how tactical voting changes at this election.

Finally, there is another kind of strategic voting known as strategic coalition voting (or coalition-directed voting). It is well known in multi-party proportional representation systems and highly relevant to this election but not much discussed and evaluated. I’m planning to write about that soon.