Who is right about the combined Labour and Conservative share in polls?
Welcome to this week’s edition which this time takes a look at how the pollsters differ in how much they think public opinion is polarised around choosing between Labour and the Conservatives.
But first, this week’s quiet mutter of regret is directed at those reporting Frank Luntz’s latest polling without mentioning the, ahem, lively debates over the quality of his polling and focus groups. (And good luck making sense of slide 90 in his latest deck, “When you ‘decide’ you are making an active choice about what you want…”)
As ever, if you have any feedback or questions prompted by what follows, or spotted some other recent polling you’d like to see covered, just hit reply. I personally read every response.
Been forwarded this email by someone else? Sign up to get your own copy here.
If you’d like more news about the Lib Dems specifically, sign up for my monthly Liberal Democrat Newswire.
Different pollsters, different stories about the combined Labour and Conservative share
Questions about whether some pollsters in this Parliament are recording figures that are too high for minor parties has long been part of informed polling punditry, whether it is doubts over the figures for Reform found by many pollsters or pondering about who is getting the Green share most accurately.
In one sense, there is no ‘right’ answer for what pollsters should be finding for parties such as Reform and the Greens, as we don’t know how many candidates they’ll be standing at the general election. But we do know it’s very likely that at least one, and possibly both, will have a long way short of a full slate of candidates and therefore asking people across the country whether or not they will vote for them is likely exaggerating what would happen at a general election.
(There’s also a question about how to prompt for such parties in polls, as the more prominence they are giving in voting questions, the higher their results may be and that may exaggerate their support, especially compared with a general election result that occurs after a campaign in which media coverage has only briefly blessed1 them with attention. It’s a complicated question with rather counter-intuitive evidence, more on which you can find on p.100-102 of Polling UnPacked.)
One way of looking at this is to see how the combined Labour and Conservative vote share varies between pollsters. To my surprise, given the sorts of discussions there has been, the picture is mostly one of similarity, as this analysis of all the national2 voting intention polls from pollsters in 2023 shows:
As the difference between triumph and disaster, both for parties and for pollsters, can be only a single digit margin, this is an occasion when abandoning the zero value on the X-axis and zooming is useful:
This differences now look much clearer, though overall pollsters and those who rely on them can take comfort from it being a range of (only) seven points between PeoplePolling (who have now ceased regular public polling) and the trio with the highest Lab/Con totals.
What’s intriguing though is the make-up of the five pollsters who depart from the 74%-75% ‘consensus’ result. PeoplePolling, when it was polling, was the most controversial pollster (though mainly due to its question wording on policy issues). Find Out Now is a new pollster with the most innovative approach to sampling, with its routes in free online prize draws. Kantar and Opinium have impressive records of general election polling (especially Kantar, whose record through various name changes is much under-appreciated). And YouGov is the OG3 of online polling, with the best conspiracy theories but also the biggest data set and longest track record to calibrate its findings.
There’s no obvious winner to pick here even though the margins are significant. The difference between other parties getting 25% and 30% is a big enough difference to mask very different stories of success or failure for those other parties.
The Reform Party is a case in point, with big variations between pollsters this year in the average vote share for them:
It’s worth noting that those finding the lowest two-party share also tend to find the highest Reform ratings, but it’s not a uniform picture. Kantar is a good way down this chart, in particular, while Focaldata is high up in it.
My hunch is that the pollsters showing the highest vote shares for Reform are those who are wrong as there’s precious little evidence from Parliamentary by-elections, all out council elections or council by-elections4 of Reform having that level of popularity.
One conclusion is more certain, however. Which is the limitations of looking at polling averages. There’s much to commend averaging different polls together, and I do it myself. But it would only be luck if averaging, say, Ipsos and PeoplePolling gets the right number for Reform or if averaging Find Out Now and Survation gets the right total Conservative and Labour vote share.
Averaging is good as it dilutes the worst results with the best. But it’s also limited as it dilutes the best results with the worst. When you don’t yet know who the best and worst are, that’s not an awful approach but nor is it flawless.
Know other people interested in political polling?
National voting intention polls
Around half of pollsters with very recent polls have their latest one showing a big increase in Labour’s lead over the Conservatives but the others don’t show that. So while it looks likely there’s been real movement, we’ll need to see more polls to be sure if this is more than a brief moment of widespread excitement.
Here are the latest figures from each currently active pollster:
Last week’s edition
An usual glimmer of hope for the Conservatives in the polls, and other polling news
These weekly round-ups are for paying subscribers only, but you can sign up for a free trial to read them straight away.
Keep reading with a 7-day free trial
Subscribe to The Week in Polls to keep reading this post and get 7 days of free access to the full post archives.