Mayor elections polling bonanza

Welcome to the 105th edition of The Week in Polls, which divers in to the latest Mayor polling including the bounty published this morning by More in Common.

Then it’s a look at the latest voting intention polls 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.)

But first, this week’s eyebrow of doubt is raised by Labour turned Conservative turned Reform politician Lee Anderson who tweeted “polling” showing him cruising to a landslide victory in his constituency. But didn’t include any details about who did the poll, when they did the poll, what questions were asked in the poll, how the poll was weighted or any evidence that it was a poll in the proper meaning of the word.

---

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.

---

A burst of Mayor polling

West Midlands

Thankfully it’s been a busy week for polling on the Mayor contests coming up on Thursday. The paucity of polling until this week has been a shame both because of the importance of the contests in their own rights and also because we have quite a few new pollsters on the scene. Seeing how those who have put their reputations on the line with Mayor polls compare with reality and with more established pollsters is going to be of particular interest.

First, though, an important update to the West Midlands polling covered last time. Savanta was showing a 2 point Conservative lead (fieldwork 11-17 April), in contrast to Redfield & Wilton’s 14 point Labour lead (fieldwork 10-14 April). When writing up those polls I commented that Savanta had a higher proportion of people recalling that they voted Conservative in 2019 than R&W. But I missed the real story which @ElectionMapsUK spotted, namely that Savanta had done their past vote weighting wrong by confusing the West Midlands (county) and the West Midlands (Combined Authority), thereby weighting their figures based on the wrong previous election shares.

Not what a pollster wants to discover the day after running a marathon. Kudos to Savanta for responding quickly, acknowledging and correcting the error. Their new figures gave a 3 point Labour lead rather than the previous 2 point Conservative lead. Closer to R&W’s figures, though still a 3 point lead rather than R&W’s 14 point Labour lead.

The big difference is a 7 point lower Reform share (6% with Savanta, 13% with R&W) and a 10 point higher Conservative share (38% vs 28%). Labour’s share is almost identical (41% vs 42%).

Since then we’ve had a new Redfield and Wilton poll (fieldwork 22-24 April). which showed a sharp increase in the Conservative vote (+9), Labour’s vote static (+1) and Reform’s vote down sharply (-9). That brings R+W much closer to Savanta’s earlier corrected poll figures, but there is also a More in Common poll just out which paints a different picture with Andy Street ahead:

Although More in Common has the Conservatives ahead, the vote share figures are not that different from Savanta and the latest R&W. It’s possible to imagine an actual result which makes all three of those polls look reasonably accurate.

The story between the two R&W polls appears to be of Reform being squeezed, and indeed on Westminster general election vote share there’s a much smaller fall in Reform’s support between the two polls (down 3 points) than the 9 point fall for Mayor vote share. But with only R&W having done an earlier poll, even if their final poll turns out right we’ll not have enough data to know for sure if that Reform squeeze was genuine (and so an interesting sign for the general election) or a sign of their first poll being a bit off.

That debate will be of interest with an eye on the general election. But this isn’t the only contest where we’ll have a polls versus results comparison to pick over.

Note: of those pollsters, only Redfield & Wilton polled this race last time around. Their poll had it 46% Conservative / 37% Labour compared with a result on first preferences of 48.7% / 39.7% and so they were spot on with a 9% lead.

London

In London we also have a new Savanta poll, which comes with significant change in their figures brought on by a deliberate methodology change:

In essence, this latest poll includes a weight to the results of the 2021 London Mayoral election, whereas previous polls did not. From a brief audit of other pollsters, some do weight to this variable, others don't, and there are valid reasons on both sides.

But we've made the change because our two previous polls for @centreforlondon both seemed to suffer from false recall. We had higher than accurate rates of respondents telling us they voted for Sadiq Khan [last time] and not enough telling us they voted [for Conservative candidate] Shaun Bailey with their 1st pref.

This is reasonably common, but we felt it was important to try and address here, especially in part due to the underestimation of Conservative support by pollsters at the 2021 London Mayoral election.

The new Savanta poll with this new methodology gives Labour Mayor Sadiq Khan a 13 point lead over the Conservatives. That appears to be an 11 point fall in his lead from their previous earlier in April. But… Savanta have reworked their earlier poll with the same new past voting weighting and so comparing the new poll with their previous, now adjusted, poll it shows only a 2 point fall in Khan’s lead.

Those figures - 13 points now, 15 points earlier in the month - are also in the same ballpark as Redfield & Wilton, whose latest poll gives a 13 point lead for Khan. That 13 pointer is massively up on the slim 1 point lead for Khan which R&W found back in September, though that was the narrowest lead of any poll taken of this race.

Finally we have a new poll from YouGov, with a chunkier 19 point lead for Khan, which is down 6 points from their February poll.

With only three pollsters with recent polls, there aren’t enough data points to be sure we’re drawing conclusions rather than imagining mirages. Especially as one shows a big increase in Khan’s lead (12 points with R+W, though comparing with pre-Christmas), one shows a twitch downward in his lead (2 points with Savanta, well within the margin of error for being a mirrarge) and one a substantive decline in his lead (6 points with YouGov).

What we can say is that the polls show a large enough lead for Khan that even if there was as big a polling miss this time as there was last time (when Khan’s lead was higher in the polls than in the result)¹ then he’d still win. And it’s rare for there to be two big polling misses in a row, both in the same direction - because, as with Savanta’s change of its methodology this week, when pollsters get it wrong, they change what they do to try to fix things.

We’ll shortly see how much reality requires them to get fixing poor polling or get celebrating accurate polling.

Note: all three pollsters also polled the last London race.² In contrast to the West Midlands, R&W was the last accurate of the three giving it on first preferences as Labour 47 / Conservative 26 (lead 21 points) compared with reality of 40 / 35 / lead 5. Both YouGov and Savanta were close on the Labour vote share but underestimated the Conservative vote share, resulting in still being significantly out on the Labour lead. They both had a 12 point lead, well above the actual 5 point one. YouGov was 43/31 and Savanta was 41/29.

North East, Greater Manchester, East Midlands and Tees Valley

Despite the interest of the North East contest, with former Labour North of Tyne Mayor Jamie Driscoll running as an independent, we’ve only had the one poll:

A nice touch with this poll is the breakdown provided showing how the different stages of the pollster’s methodology change the figures between the original raw data and the published result:

That final step - modelling turnout - has been tricky for pollsters at times in the past, and the lower the turnout is in an election, the less accurate overall polls are. So it’s notable that it is apparent differential turnout rates which move the figures from a clear Labour lead of 11 points to a very close race with a Labour lead of only 2 points.

More in Common also gives us the first Greater Manchester and East Midlands Mayor polls:

If this trio of polling from More in Common is right, there could be an important story to tell about Reform’s performance in actual elections for a change. Or it could be a case of them underperforming their poll standings again. We will soon know.

Finally, as very briefly mentioned last time, we have the first proper³ Tees Valley poll, which comes from Redfield & Wilton and is for a contest without a Reform candidate:

So if the polls are right, two of the highest profile contests - and the two being most closely watched for general election straws in the wind - are both, in Tees Valley and in West Midlands, looking pretty close.

Let’s hope the importance of the contests and their apparent closeness brings us some more polls before Thursday.

---

National voting intention polls

Once again, little of drama to report with the last week’s polls though overall this year the Conservative Party’s support has slipped by a couple of points with Labour static. The next few polls will give us a clearer picture of whether news such as the Rwanda Bill passing and the defence spending announcement has turned around that pattern.

For more details and updates through the week, see my daily updated table here and for all the historic figures, including Parliamentary by-election polls, see PollBase.

---

Last week’s edition

Do the Conservatives really have a problem with support among young people?

---

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.

---

Do people want the Conservatives ever to win again?, 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.