Would you be more or less likely to vote if...
It sounds like a obvious point, but when people answer survey questions, they know what you're up to: coming round here, asking all your questions about voting intention and elections tomorrow, you're just going to add up the figures, weight them and publish them in the newspapers. I know your game matey.
This may seem a statement of the bleeding obvious, but it does have an important impact in how you should interpret poll findings, particularly on one of my personal bug-bears, the "would X make you more or less likely to vote Y" question.
Imagine yourself in the shoes of, say, a staunch Conservative supporter who likes fox hunting. The man from ICM rings you up and asks "would you be more or less likely to vote Conservative if they promised to bring back fox hunting". Now, you'll vote Conservative anyway so really you should say no difference, but since you like fox hunting, you want the poll to show that fox hunting is a popular policy, and you can genuinely say that you'd scamper along to the polling station with more of a spring in your step with that policy in that manifesto, so you say yes.
You see the problem. Questions like this are liable to be used by people to give a preference on the policy being asked about, without them really considering whether it would change their vote. To their credit YouGov normally ask the question with two no difference options, one saying "no difference - I would vote for party X anyway", one saying "no difference - I wouldn't vote for party X anyway" to try and push people towards really thinking about whether it would actually change their voting intention, but I'm sure a lot of people are still likely to use it to indicate approval or disapproval of a policy.
Migration watch have one today - it has 33% of people saying they would be more likely to vote Tory if they promised to limit immigration to the level of emigration, with only 5% saying it would make them less likely. This certainly suggests it would be a popular policy, but it probably overstates its electoral effect: look at the actual figures and you'll see over half of that 33% are people who already say they are voting Tory anyway. True, they may be indicating that that it would firm up their intention, but they could also just be Conservatives who want to indicate support for a policy they like.
Questions like this probably over-egg the importance of any policy they ask about anyway. Firstly, "more likely" is a long way short of actually changing your vote, secondly it draws undue attention to a specific policy which would probably be a minor factor in actual voting intentions. My view remains that broad party image is the important factor, specific policies are important only in how they affect that. A hardline anti-immigration policy may make people who support that policy think of a party that adopts it as being in touch with their feelings, prepared to stand up for people like them... but also risks being seen as bigoted, negative, spiteful and so on.
Immigration, as it happens, is an interesting example. Polls always show that people are negative on immigration and that they support harsh policies on it. Yet parties generally don't play on it because it is perceived as actually being electorally unpopular to do so. At the last election the Conservative ran very heavily on immigration, and it was seen as a flop.
The exhaustive polling that Michael Ashcroft commissioned privately during the election campaign shows that the messages that people overwhelmingly recalled from the Conservative campaign were those on immigration and travellers (despite the fact that as the campaign progressed the Conservative's didn't actually concentrate on them that much). The same evidence tells us that the Conservative campaign didn't work - only 20% said afterwards that the campaign made them more positive about the party, 36% less positive. 49% thought it was "mean, negative and nasty". 41% characterised it as negative, it was also seen as aggressive and depressing.
That said, polls also showed that the immigration policy itself was popular. This seems strange, but consider that even after a campaign where one party, deliberately or not, made it the perceived main plank of their campaign, only 12% said it was the most important issue for them in deciding their vote, and some of those voted Labour or Lib Dem. For those who immigration policy was the deciding factor, it presumably helped the Conservatives. For the other 88% of people it probably contributed to the party being seen as negative and mean.
To return to those "more or less likely" questions, I suppose they have their uses, especially if they are carefully filtered so you can see the answers of actual swing voters - people who say they are undecided or likely to change their vote - and exclude people who will definitely vote for their party anyway. It is often a lot more complicated than policy that seems popular in polls equals electoral success though.
(Danny Finkelstein offered his own take on what he called the immigration mystery a couple of years back here, which incidentally includes a fantastically boneheaded comment from a would be constituent on the doorstep)
