Stability does not equal stalemate

Quite a lot has been made of the fact that since the start of the year, polls haven't swung back to the governing parties as much as they have historically. I'll take a look at the Lib Dems in a future post, because experience in other countries suggests that junior coalition partners can be something of a special case. But the Conservatives have been quite confident that the polls will move their way before the election.
There are lengthy debates to be had about the relevance or otherwise of the various historical precedents from general elections. But one relevant piece of history (if not a genuine precedent) that seems to have have been totally overlooked is 2014 and the elections to the European Parliament. To be sure, there are plenty of differences between general and European elections – voting patterns are different, turnout is far lower in the latter, some areas had local elections last year and others didn’t, while the polls in the closing stages were all online rather than a mixture of online and phone. And last year, Labour and the Conservatives were ultimately battling for second place behind UKIP (who I’ve excluded from the charts for clarity).
But 2014 is of at least some interest simply because it’s so recent – there has been no other Britain-wide election under a Tory (-led) government since 1997. What happened ought to suggest that we can’t assume the polls will stay where they are, no matter how stable things appear. The chart below takes the YouGov European polls during the runup to last year’s election. I’ve used YouGov because house effects make cross-comparisons misleading, and YouGov was the only pollster to poll regularly during the period in question. They were also the most accurate.
February 1974 or 1992? What about 2014?
What we see is relative stasis until early May (three weeks prior to the election), then a narrowing from 8 to 4 points in the space of a week. The final call poll (N=6,124) also showed Labour four points ahead of the Conservatives, but when the votes were counted, the spread was a shade under 1.5 points.
February 1974, 1992 or... 2014? (part II)
This isn’t to say that the incumbent is sure to gain 6 or 7 points in the last few weeks, for all the reasons already given. But the fact that this happened so recently should be a reminder that poll moves don’t always happen smoothly, can happen late, and that a long string of very similar snapshots still does not equal a prediction. There is all to play for in this election.

About The Author

Related Posts