How did the polls do at the general election? Round 2

Welcome to the 124th edition of The Week in Polls (TWIP), which returns to the question of how the national voting intention polls did at the general election.

Then it’s a summary of the latest national voting intention polls and a round-up of party leader ratings, followed by, for paid-for subscribers, 10 insights from the last week’s polling and analysis. (If you’re a free subscriber, sign up for a free trial here to see what you’re missing.)

First, a gentle sigh of disappointment directed at Huffington Post for reporting a poll about whether people think Labour is ‘most’ interested in helping itself and its allies as whether people think it ‘only’ wants to do so. The graphic from the pollster had it right but the story’s first sentence went for the more dramatic, sexier and, er…, wrong formulation.

---

Been forwarded this email by someone else? Sign up to get your own copy here.

Want to know more about political polling? Get my book Polling UnPacked: the history, uses and abuses of political opinion polling.

---

Polling records under the microscope

Last week Conservative peer Robert Hayward held an event in Parliament about how the polls did at the general election. The tone of Robert’s comments, and those of the two pollsters taking part (Martin Boon and Will Jennings), were rather more negative about how the voting intention polls did than my own post-election write-up, though I was more critical of how some MRPs performed.

But once you scratch under the surface, I’m not sure that our views are that much different. More a case of whether you choose to describe a glass as half full or half empty.

Martin is, as his social media postings often show, very much in the ‘the glass is half empty and people aren’t worrying enough about that’ camp. For example:

So let’s dive into the cases that Martin and Will made, with thanks to them and Robert for letting me have a copy of the slides to use.¹

Let’s start with some polling data from Will:

That is, I think it is fair to say, a pretty darn good picture over the decades of political polling: on average, the polls get closer to the actual result the closer you get to polling day, and by that day, they are on average only a couple of points out.

Remember too that we’re talking about 100 points overall (i.e. 100%). Being on average a couple out on that sort of scale is pretty good. Show me any other source of information about party popularity that can get close to to consistently matching that over decades and with that sort of precision.²

But pretty good isn’t the same as spot on, and in particular at the 2024 general election there was a consistent over-estimate of the Labour vote as another of Will’s graphs shows:

Some of that may have been due to last minute switches away from Labour, such as tactical voting leading people to switch to the Lib Dems or perhaps people deciding Labour was going to win and so they did not need to vote. But the fieldwork for those polls included plenty of sampling right up to just before polling day. So it’s unlikely that ‘the polls were right but the public then changed their minds’ works as much of an explanation.

Moreover, there is an apparent pattern here:

It looks like there is a pattern here of the polls, most of the time, over-estimating Labour’s position relative to the Conservatives. (Though beware, with the relatively limited number of general elections we do not have many data points. Even so, this pattern extends to pre-1979 polls too, making it likely to be real.)

Enter now the more critical comments from Martin, emphasising that pattern of the polls over-estimating the Labour position and adding that this is an international problem:

His point is that the errors being consistently in the same direction indicates a problem. If there was one simple, known explanation for this, pollsters would have long since fixed the issue. It’s likely that the explanation is, at least in part, due to those who pollsters can get hold of (i.e. people who can be reached and who are willing to participate in a poll) being skewed towards those characteristics that also make people a bit more likely to have voting intentions on the right.

Martin’s highlighting of this trend is a powerful point, but also one that I think leaves political polling intact as a useful source of information, including on voting intention. (And indeed his own company does voting intention polls here in the UK.)

That is because although the consistency of the error indicates a systematic flaw, the size of errors overall are sufficiently small as to make polling a pretty good guide to what is going on.

The flaw should be fixed, but while fixing it political punditry based on polling should still be given more attention than that based on flaunting disregard of it.

---

Voting intentions and leadership ratings

A thin table so far as I have reset it to include only post-general election voting intention polls:

Here too is a summary of the the leadership ratings, which vary a lot depending on the wording but overall patterns are consistent:

For more details, and updates as each new poll comes out, see my regularly updated tables here.

For the historic figures, including Parliamentary by-election polls, see PollBase.

---

Last week’s edition

Is Kemi Badenoch leading in the Conservative leadership contest?

---

My privacy policy and related legal information is available here. Links to purchase books online are usually affiliate links which pay a commission for each sale. Please note that if you are subscribed to other email lists of mine, unsubscribing from this list will not automatically remove you from the other lists. If you wish to be removed from all lists, simply hit reply and let me know.

---

Should the Conservatives merge with Reform?, and other polling news

The following 10 findings from the most recent polls and analysis are for paying subscribers only, but you can sign up for a free trial to read them straight away.