ICM/NotW and would UKIP cost the Conservative 50 seats?

There is an ICM snap poll on the BNP in today's News of the World, but what's online really isn't enough to do a proper analysis - we really need to see what was asked. From first sight, while it's reported as a "shock horror people agree with BNP poll!", a glance at the article suggests it actually shows small majorities of people don't want gay sex taught in schools and think Islam has a poor record on women's rights, which really isn't the sort of thing I'd find particularly surprising. Only 10% think there should be a total stop to immigration.

The one surprising finding from the News of the World article is that ICM apparently found a third people agreeing with stopping benefits to British-born people from ethnic minorities to pay for them to leave the country. That seems counter-intuitive, after all, if only 10% of people want a stop to immigration, stripping benefits from British people based on their skin colour and paying them to leave would normally be regarded as a lot more extreme, but the News of the World claim ICM found it was three times more popular. I think we'd better wait and see ICM's tables, rather than the News of the World's interpretation of them.

Moving on, the Observer has an article saying that Labour strategists think that UKIP could cost the Conservatives 50 seats at the next election. Quite frankly, while a good performance by UKIP might well disadvantage the Conservatives, the sums here don't even begin to add up.

If we take the Conservative's 60 most winnable seats, the majorities range from 31 to 2686 (0.1% to 5.7%). Of course in reality these aren't the seats that are under question, the Conservatives are going to win these anyway. Different levels of support will give us different distributions of required swing, but since we don't know what the actual swing will be at the next election, these make as a good an illustration of how the majorities in seats are distributed as any.

If one assumes an even level of support for UKIP across these seats, UKIP would have to take an extra 5.7% of the vote, above and beyond the 2% they got at the last election, and take it all from people who would otherwise have voted Conservative in order to prevent the Conservatives winning all 60 seats. UKIP getting 8% at the next election seems pretty damn unlikely to start with, given they are currently on 4% and have been in decline since their boost at the European elections. In fact, even if they did, it wouldn't cost the Conservatives 60 seats.

The Observer's article has that old canard that UKIP cost the Conservatives 27 seats at the last election, if one starts from that basis, 50 seats sounds entirely plausible with a higher level of UKIP support. Unfortunately, it's rubbish. Firstly, it's only actually 24 seats where the majority over the Conservatives in 2005 was smaller than the UKIP vote (the 27 comes from lumping Veritas in with them). Secondly, for the Conservatives to have won all of them they would have had to win every single vote that went to UKIP.

In reality, some UKIP voters are people who would otherwise vote Labour or Lib Dem. A large chunk of UKIP's voters are people who probably wouldn't vote at all in the absense of UKIP or an alternate fringe party to cast a protest vote for. I'm pretty certain that UKIP take more people who would otherwise vote Conservative than people who would otherwise vote Labour, but once you also take out people who wouldn't vote, or would vote for another minor party, the difference won't be massive.

In the Observer's report, they are assuming that UKIP takes about two-thirds of its support from Conservative minded voters and a third of their support from Labour minded voters - so every 3 votes for UKIP reduces the Conservative performance relative to Labour by just 1. On that basis, UKIP cost the Conservatives all of 10 seats at the last election, and for UKIP to prevent the Conservatives winning 50 seats, would take a performance in marginal seats equivalent to them winning 19% nationally. On that front, we really are into fantasy land.