Political polling – a reason to be cheerful

Political polling – a reason to be cheerful

In the interim the industry has worked tirelessly with the British Polling Council and the Market Research Society to find a way forward. Much hard work lies ahead.

Last week’s polling success for the local elections is obviously welcome news for the entire market research industry. Quite a number of polls were broadly accurate – especially in London – and have hopefully restored a little faith in this important part of our industry.

 

It’s also interesting that there were no differences between phone and online surveys. The only phone poll released (?by Survation?) showed the same result as a YouGov online poll that took place at the same time although, as we saw last year, a lack of any modal difference is no guarantee of accuracy.

Without detracting from my team’s glory this past week, where they accurately predicted the results for the London Mayoral Election for the second time in a row, there were (?and I assume this was the case for other agencies too?) several factors making it easier for us to get an accurate result:

We had a general election last year which meant that we had firm and up to date levels of party support that helped our sampling and weighting.

We had a relatively straightforward two-horse race in which minor parties didn’t play a significant role, if you compare Khan and Goldsmith’s 2nd round vote share with the share of just Labour and Conservative voters in the first round, they’re remarkably consistent. Sadiq Khan got 57% of the votes in the second round but 56% of the Labour-Conservative vote in the first round.

Compared with previous London elections which were dominated by big personalities, this was a relatively party-line vote. Labour got 43.7% of the vote in London in 2015 and Sadiq Khan got 44.2% in 2016. The Conservatives got 34.9% last year and Zac Goldsmith got 35.0% last week. So as long as the proportions of supporters of each party were correct in your sample then you should get broadly the correct result.

Of course the real test for political polling will be the fast approaching EU referendum where we have support for each side cutting across traditional party lines and demographics, differing levels of enthusiasm for each side and, above all, difficult to predict turnout which is likely to prove to be the decisive factor.

So political polling has had a good week and cleared an important hurdle but now it’s time to put our noses to the grindstone to meet the much more difficult challenge coming in June.