Latest voting intention and leadership ratings opinion polls

Not so: the evidence is that they are still pretty good – and haven’t got worse. Here’s the data that does that myth-busting. Here too is how the polls did in the 2024 general election.

A rough idea of the likely margin of error in any one opinion poll is to think that it’s pretty likely to be within 3 percentage points of the correct result. Anthony Wells explains here in more detail what this margin of error calculation means, and why it does not strictly apply to modern polls.

Based on the historic record of polls, the British Polling Council requires its members to use this explanation of the margin of error: “All polls are subject to a wide range of potential sources of error. On the basis of the historical record of the polls at recent general elections, there is a 9 in 10 chance that the true value of a party’s support lies within 4 points of the estimates provided by this poll, and a 2 in 3 chance that they lie within 2 points.”

It can appear baffling that a poll of only 1,000 people is meant to be enough to reveal the mood of a nation of tens of millions of people. But 1,000 or so is indeed enough, as I explain here.

The table above includes the latest UK or British voting intention poll from each of the currently active reputable pollsters.

If a company isn’t listed this is because it has not carried out a recent poll, it isn’t reputable or I’ve made an error. Please get in touch if you suspect it’s the latter. ‘Reputable’ usually means being a member of the polling industry regulatory body, the British Polling Council (BPC). I occasionally make exceptions, such as for a new polling firm with a good pedigree which hasn’t yet had its BPC membership approved.

MRP polls are a way of getting give individual constituency results without having to do a full poll in each constituency. Because they have larger sample sizes, they should also be less subject to ‘noise’, i.e. support fluctuating because of who happened to be in the sample. But they also involve significant modelling of the electorate, which if done well produces good results but which also leaves plenty of ways it can go wrong.

I’ve written a guide to how MRP polls work and whether they’re likely to be right here.

There are many good quality polling averages which are produced by others. But there are significant problems with even the best polling averages.

When looking through the polling figures, remember the much ignored but still very relevant warning about individual polls.

To put the voting intention numbers above into longer context, take a look at PollBase, my database of general election voting intention figures from opinion polls going back to 1945. It is updated quarterly.

Separate figures are not given for the SNP and Plaid because the relative size of Scotland and Wales means that the percentage vote share for each of the across Great Britain is too low for variations to mean much. (For example, at the 2017 general election, the SNP scored 3% of the total vote across Great Britain. A fall to 2% would be a move that is well within the margin of errors on polls yet also, if accurate, would be a massive hammering in the constituencies it contests.)

I include Restore (and also Your Party) in the quarterly PollBase updates.

However, I don’t currently include Restore in the summary table on this page because of the low level of support found for them. It was the same with Greens and Reform, who only started appearing in the summary table when their support started being consistently higher than that. If Restore, or another party starts consistently scoring higher support, then I’ll add them in too.

Most of these polls are for Great Britain, i.e. excluding Northern Ireland but including both Scotland and Wales. General election voting intention polls conducted over a smaller area, such as London only, are excluded.

A plausible-sounding critique of voting intention opinion polls is over the choice of parties to ask about. These polls list some parties up front and then give an ‘other’ option, behind which sits other, much smaller parties. Labour, for example, will be in the first category but the Women’s Equality Party in the second. Which often leads people to complain that a poll is biased against party X because it is listed in the other section rather than in the main party listing. That sounds plausible, but the evidence is that this doesn’t unfairly depress the support for other parties.

Yes, by the Market Research Society and also by the British Polling Council, which all the reputable political polling firms are members of. The BPC’s rules include requiring pollsters to publish in full the exact questions asked for their polls, protecting against leading questions being secretly asked.

The regulation could, though, be improved.

You do now.

And for an explanation of why it’s common for people not to have been polled, see this.

There’s a chapter on the subject in my book, Bad News, or for a whole book about polls, see Polling UnPacked.

No.