Last night some over-excited bloggers said YouGov was conducting an exit poll for the BBC. This was a case of putting two and two together and getting ten. Yes, YouGov’s Anthony Wells is helping the BBC analyse yesterday’s votes; and, yes, YouGov did return yesterday to more than 3,000 respondents we had previously questioned during the campaign and asked them how they had voted. As always we conducted our poll online, so it wasn’t a true exit poll – we didn’t have fieldworkers standing outside polling stations asking people how they had just voted.
But, no, our figures were never destined for the BBC results programme. We intended this as an internal exercise. Our Swedish company has a good record in conducting on-the-day polls to predict election results; we were seeing if the same methods could work in Britain.
However, given the evident interest in what we were doing, the clear signs of late movement and the fact that the referendum votes have not yet been counted, we have decided to disclose our results. (Our sample sizes are far less than those for a conventional exit poll, so the risks of sampling error are greater; but for the sake of transparency, and also because of the clear evidence of late movement, we felt that the data contributes to our understanding of what happened yesterday.)
First, the referendum. Our final survey for the Sun predicted that No would win by 60-40%. Our voting-day survey indicates a further widening of the “No” lead; 62% voted yesterday to keep first-past-the-post, while only 38% supported the Alternative Vote.
As for Scotland, YouGov’s campaign polls indicated great volatility, with the SNP taking the lead in mid-April and enjoying fluctuating leads since then. Our latest poll detects an election-day surge for the SNP. These are our figures
As those figures show, the SNP lead widened sharply in the final hours of the campaign, especially in the second, regional list, vote. What happened, above all, is that tens of thousands of Labour supporters decided at the last minute to vote for the SNP. In our final pre-election poll, 17% of those who said they would vote Labour in a general election said they would vote for the SNP in the constituency vote, and 14% would back the SNP in the regional, list vote. Our election-day poll indicates that these figures jumped to 23% and 25% respectively.
The 11-point surge in the list vote (from 14% to 25% of Labour supporters voting SNP) is especially significant. This 11% is equivalent to around 80,000 voters. They are people who have delivered Alex Salmond the real chance of an overall majority in the new Scottish Parliament.
In Wales, there was also late movement, though on a more modest scale:
The late shifts in Wales were on too modest a scale to upset our election-eve prediction that Labour could just secure an overall majority – at worst, it would be just one or two seats short.
