Pollwatch: Conservative lead narrowed, but two polls don’t make a trend (14 April 2010)
The latest findings from Populus and ComRes may be just a statistical quirk
Two polls today, by Populus for the Times and ComRes for ITN and the Independent, narrowed the Conservative lead, dangerously so for David Cameron’s hopes of becoming prime minister. It remains to be seen whether these are the start of a trend or a statistical quirk.
Across the board, the polls barely shifted in the first week of the campaign. The Conservatives remain within a point or two of 38% support, and Labour within a similar distance of 31% in nearly all the polls. Perhaps this is not very surprising, because the first week was not particularly engaging and the big setpiece events of the campaign – the manifesto launches and the leader debates – were all yet to take place.
We can expect more change and volatility if and when the events of the campaign start to excite the public. But if the polls remain constant at this level, the end of the campaign and the result will be pretty exciting stuff, because this is probably on the cusp of what the Conservatives need for an overall majority.
Translating polls and vote shares into who will win how many seats at the election is an imprecise business, and depends on a lot of assumptions. The simplest is “uniform national swing”, which takes the change in each party’s vote since 2005 and then applies that change in every constituency. If voters behaved this way, the Tories would need a lead of about 11 points to win a majority.
But there are good reasons for assuming that swing will not be uniform. In local elections and polls concentrating on marginal seats the Tories seem to be outperforming the national swing by a point or two. Their strategy since 2005 has involved relentless targeting of money, promises, messages and campaigning on the marginal seats to achieve this, and it would seem to be working.
It is also to do with political strategy: Tony Blair’s rapport with swing voters in marginal seats long outlasted his honeymoon with metropolitan liberal opinion. Cameron, as shown in the detail of the Guardian’s latest ICM poll, seems to be going over well with the same voters (in social classes C1 and C2) that Blair targeted so successfully. The swing in the marginals is assisted, although not by as much as could have been expected last week, by people who had previously voted Labour for tactical reasons abandoning the party for the Lib Dems or Greens.
Another factor that seems to be helping the Conservatives is the way changes in voting intention vary by region. According to YouGov’s combined data (reported in PoliticsHome), the swing is strong in several regions that are rich in Labour-held marginals (north-west, west Midlands and east Midlands), adequate to win most of the targets in other regions (south-east) and low in the regions where there is little room for Conservative gains from Labour (Scotland, south-west). However, local sub-samples can be unreliable and there is room for more polling to be done to sample what is going on in the English regions and Wales in particular (polling in Scotland is pretty consistent).
Against these factors that are helping the Conservatives, there are a couple of adjustments that point the other way. Uniform swing assumptions are bad at predicting how the Lib Dems will do, because strong local campaigns and popular incumbents can resist national swings. Uniform swing would see the Conservatives winning large numbers of Lib Dem seats, but this is almost certainly not going to happen.
Then there is turnout. A large part of the electoral system’s bias against the Conservatives stems from the tendency for the Tories to pile up large numbers of votes in their safe seats because turnout is relatively high, and for safe Labour seats to have low turnout. If a close election brings out Tory voters in their droves in the countryside and the suburbs, as it did in 1992, but does not cause Labour turnout in safe seats to rise much, it will not help the Tories win more seats.
All this considered, the lead the Conservatives need falls to around 7% or so, similar to 1992 when a Conservative lead of 7.5% was enough for a majority of 21 seats. However, one cannot be certain. If they are lucky with turnout and varying swing, they might squeak across the winning post with a lead of 5%; if the cards fall badly for them, and Labour grassroots campaigning counteracts the Ashcroft marginal strategy, they could fall just short with a lead of 9%.