Four polling mysteries

Alastair Meeks
8 min readMar 24, 2024

There’s something odd about current polls in the UK. They don’t completely stack up with either anecdotal experience or with empirical evidence. There are four mysteries that need resolving.

What is happening with Reform?

Every new poll seems to show Reform climbing just a little bit higher. Their average rating in the polls is now nearly 12%. A YouGov poll had them at 15%, within 4% of the Conservatives. The only problem is that there is absolutely no evidence outside polling to support such findings.

I’ve written about this before. Since then, there have been limited signs that Reform are indeed improving. They tallied 13% in the Wellingborough by-election and 10% in the Kingswood by-election. They managed only 6% in the Rochdale by-election, however (less than they managed in 2019). Since all three of these were heavily Leave-voting constituencies, they weren’t exactly setting the Thames on fire.

I suggest that their by-election performance in February was consistent with an underlying vote share of 6–7% at that time. If you want to bid me up to 8%, I’ll live with that though I will feel charitable doing so. I struggle to go beyond that though.

Now since then Lee Anderson has defected to Reform and the Conservatives have continued to flounder, so I accept that it is possible that Reform’s support has risen. Possible, but it’s not a particularly obvious assumption.

I stress that I can easily envisage circumstances where Reform achieves lift-off (most of them involve Nigel Farage taking over from Richard Tice, who is a stuffed shirt incarnate). I just don’t believe that it’s yet happened at anything like the levels that some pollsters are finding. They seem to be overrated by something like 4–5% on average. In other words, nearly half of the Reform support at present seems phantom.

If Reform are overstated, where’s that other 4–5% of the electorate to be found?

This is potentially a really important question. For starters, 4–5% could make all the difference between a Conservative wipeout and a respectable seat tally. And as I shall come onto, the problem may be a lot bigger than that.

There are two basic possibilities. The first is that the pollsters are picking up people who say they’re going to vote Reform but in fact do something else. The second is that the pollsters are picking up too many people who support Reform, thus crowding out other respondents who would do something different.

Cards on the table, I’m a firm believer in option two. I believe that pollsters are finding it way too easy to find GB News Man.

But we should consider all possibilities. If this is a vent to pollsters, what are these people going to do instead? Handily, we do have polling on what Reform supporters might do if they don’t vote for Reform. For example, in a poll from last year, only 31% of those who intend to vote Reform UK said they would back the Tories if Reform UK did not stand a candidate in their constituency. Almost as many wouldn’t vote. Only 12% didn’t know what they would do. Taking this polling at face value, this isn’t going to be much of a secret stash of voters for the Conservatives to draw on, even if this is the cause of overstated Reform polling. It might make the difference of a percent or so at the margins.

If I’m right, however, and what we are looking at is simply oversampling of Reform supporters, there are two possibilities. The first is simply to discount the Reform polling appropriately, then scale up all other parties in proportion to the oversampling. That would marginally increase Labour’s apparent lead over the Conservatives.

I worry that’s too simplistic. My worry is that what we are seeing with Reform may be a particularly clear artefact of the problem that people who respond to pollsters may not behave or think in the same way as people who don’t.

Think about the type of person who is inclined to respond to polls. I guess we’d call him a geek. The loudest thing on two feet. He’s proud as he can be of his ideology, he’s going to give us a peek.

These are not going to be properly representative of the wider population. A lot of the time, the polls are going to be close enough to the underlying reality. But not necessarily. So I worry that the polls have an obvious problem with them, a problem of which Reform merely illustrates, and may be particularly weak evidence this time around. If nearly half of just one party’s support seems to be phantom, what other problems lurk beneath the surface?

To be clear, it’s obvious from all extrinsic evidence that Labour have a big lead over the Conservatives. In that sense, this question of reliability of polling doesn’t matter. Still, try not to get too fixated on exact numbers. We’ve got a pretty big hint right now that they’re not to be blindly trusted.

How are the Lib Dems going to do?

The Lib Dems have fought four successive elections by failing to target rigorously and as a result seriously underperforming when it comes to winning seats in Parliament. All the signs are that they risk making it five in a row.

In 2019 they got 12% and 11 seats. On the new boundaries, they notionally hold just eight seats. The Lib Dems are polling in single digits at present. The contours of their own position don’t exactly look promising.

Yet Lib Dems are giddy with excitement. You don’t have to look far to find Lib Dems making predictions of 50+ seats after the next election. They seem to have been huffing sandal.

The cause of their excitement is the collapse in Conservative support. Correctly, they note that in 1997 they made large numbers of gains even as their vote share went down because of the relative swing from Conservative to Lib Dem. They fancy the same thing will happen again.

I expect it will to some extent. But the extent of the possibilities are being seriously overstated in my view.

Let’s charitably assume that the Lib Dems get 14% of the vote share at the next election. That’s more than they have achieved in any election since 2010 and there’s no sign of them reaching such levels in the polls at present, but I’m a charitable kind of guy. Let’s assume also that the Conservatives get 25% of the vote share at the next election. That’s above the Conservatives’ current average polling levels but it would still be an unprecedentedly awful result for them and I still think it more likely than not that they’ll eventually start steadying the ship to some extent (for further details, see below).

This would represent a 10% swing from the Conservatives to the Lib Dems. That is eye-popping. But a 10% swing would, on a uniform national swing, net the Lib Dems just 24 Conservative seats. Labour will be serious challengers in at least three of those (Wimbledon, Finchley & Golders Green and Cities of London & Westminster).

“Aha”, Lib Dems say, “but uniform national swing isn’t very useful when looking at our prospects. We shall be targeting only the seats we have a chance in”. Sure. But there are limits, limits that the Lib Dems rudely came up against in 2010, 2015, 2017 and 2019. You’d think they’d learn.

Their vote is already very concentrated in a relatively few seats. There are only so many miracles that can be worked. The Lib Dems need to work out where they are going to eke out those miracles.

The Lib Dems might also argue that 25% is not a floor for the Conservatives. And I’d agree. But if the Lib Dems aren’t scooping up extra vote share, they’re going to see their lunch eaten by Labour in seat after seat. And there’s no guarantee at all that the Lib Dems will get as much as 14% of the vote share. Right now just holding steady at 12% looks more likely. If so, the Lib Dems will effectively be looking to redistribute its vote share between seats.

The Lib Dems have a few chances to win seats off the SNP and will even be hoping to win Sheffield Hallam back from Labour. Unless they can increase their actual vote share substantially, however, 35 seats looks like their absolute maximum for the next Parliament.

To be clear, they should be very excited about such a result, which represents a step change up from where they are now. They risk not achieving that, however, by putting effort into targeting remote prospects. They need to focus.

What is going to happen with 2019 Conservative voters who are still saying “don’t know”?

YouGov’s recent poll showing the Conservatives on 19% of the vote got a lot of attention. Some of the detail in that poll deserved attention too. For example, 22% of 2019 Conservative voters are still saying that they don’t know who they’re going to vote for (two thirds of the number who are saying they are going to vote Conservative). That’s nearly 10% of the electorate.

Now it should be noted that 8% of 2019 Labour voters are also saying that they don’t know who they’re going to vote for. However, this group is just over a quarter the size of 2019 Tory Don’t Knows because (a) far more 2019 Labour voters have made up their minds and (b) there were far more 2019 Conservative voters in the first place.

So what this group decides to do is potentially important. There is a widespread assumption that they will eventually, grudgingly, return to the Conservative fold. This indeed is behind the methodology of pollsters such as Opinium that treat them as Conservative voters when reporting their results.

The longer, however, that they remain Don’t Knows, the less likely that becomes. And indeed there has been some evidence that they are slowly turning their backs on the Conservatives. The Opinium polling series was recording the Conservatives at 28 or 29% in late September and October last year. This had drifted down to 26 or 27% at the turn of the year. The last three Opinium polls have had the Conservatives on 25%. Given Opinium’s methodology, this should be especially alarming for the Conservatives.

We know these voters don’t like either Rishi Sunak or Sir Keir Starmer, and are fed up with the Conservatives without particularly trusting Labour. Many of them may ultimately still vote Conservative but it seems increasingly likely to me that quite a lot of them will simply not bother to vote.

For me, the important fact about these voters is that they represent a source of votes for the Conservatives to lose. They haven’t lost them yet but they are currently losing a lot of them. Can they stop that process? Right now it’s very unclear to me that they can.