Tamworth and Mid-Beds - a quick dissection
The headline results for Labour are fantastic, but are we seeing enough evidence of direct switching?
Yesterday, I called for a ‘show me the money’ moment from Labour regarding their ability to perform well in Middle England - particularly in Tamworth.
This morning, two wins and some massive swings later, you could very well argue that Labour emphatically answered that call.
Tamworth and its predecessors have been bellwether seats since October 1974, and Labour haven’t really been competitive at all there since 2010. Until last night, when they took it on a 24pt swing.
In Mid Bedfordshire, Labour overcame a tricky three-party contest dynamic in which the Liberal Democrats went hard early at the seat. Starmer’s party took former Culture Secretary Nadine Dorries’ seat on a 21pt swing, and while their 3pt margin of victory was smaller than the 5pts they managed in Tamworth, given the circumstances it was an equally impressive result.
In their ‘just for fun’ graphics this morning, Election Maps UK reminded us that if a Tamworth-style swing was repeated at the next general election, the Conservatives would be reduced to around 30 seats.
![Image Image](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F648fbb2c-fc04-4eec-9049-689eed4a1428_652x1200.jpeg)
Now of course, this graphic is indeed ‘just for fun’. Everyone is aware that this is a by-election, and there is no direct read across from by-election voting behaviour and results to upcoming general elections.
But what do those swings really tell us?
One thing I want to immediately throw caution on is the way that ‘swing’ is being interpreted in these by-election contests.
Swing in the context of sharp turnout decline (as we see in by-elections, where typically only 2/3 of those who vote in a general election will turn out) needs to be understood very differently to how it is assessed in general elections.
Typically, we think of two-party swing in a national vote contest as symbolic of the general percentage of the electorate who moved from one party to another (sometimes directly, sometimes not, depending on how strong competition is from third, forth, fifth, parties).
In by-elections, we have to factor in that turnout itself swings, and that the vote swing we see between two parties (in this case, 21 points in Mid Beds and 24 points in Tamworth) does not at all necessarily mean direct switching from one part to the other.
And that is very important; as well as putting on big swings in Middle England heartlands, the other thing I wanted to see from Labour vis-à-vis assessing their chances of going from their worst ever election result since the 1930s to building a House of Commons majority in 2024 (No, don’t even think about reminding me of the latest possible election date. Don’t do it, stop it) was evidence that they were successfully winning over 2019 Conservative voters to their cause.
The latest YouGov polling suggests that around one-in-ten (12%) of those who voted for the Conservatives in 2019 are currently planning to vote Labour at the next election.
That may sound like a small number, but it most certainly isn’t in the context of winning elections - carving off 12% of your opponent’s vote from one election to the next is not a feat we often see. It eclipses the rate of direct switchers that Johnson achieved in 2019 (around 7-8%).
But it is overshadowed by the ~16% of the Conservative 1992 vote that Blair peeled off in 1997 (according to figures from the British Election study).
Remember - Starmer’s Labour have to outperform Blair’s 1997 win in terms of vote swing achieved and number of constituencies flipped in order to get over the majority line.
Winning ~12% of Conservative 2019 voters is good for Labour, but unless that switching is coupled with an extremely efficient geographical concentration (as Johnson’s ~7-8% was in 2019), it is tricky to see how it gives Starmer a parliamentary majority.
Further, that direct switching figure is dwarfed by the number of 2019 Conservative voters currently telling us they either would not vote if an election were being held tomorrow (6%) or that they don’t know what they would do (21%).
This group, currently without a party vote intention, is a far bigger contributor of Labour’s headline vote intention lead than any combination of switchers.
And this big heap of former Conservative voters is the focus of a huge amount of energy and focus for both parties. For the Tories, it’s about winning them back. For Labour, it’s about convincing them to trust Starmer and trust that the party offer a viable, credible alternative.
They are also a big part of the story last night…
What do the by-election numbers tell us?
Mid-term, the best possible source of information we have for how the various parties are doing with the electorate are of course vote intention polls. But by-elections (and local authority elections) provide us with a great opportunity to check that the story we are telling in the polling is consistent with observable voter behaviour on the ground.
So, do the Tamworth and Mid Bedfordshire results reflect what we are seeing in the polls? Can we see evidence of a mass ‘Conservative 2019 stay at home’ effect? And what would this mean for Labour and their 2024 (stop it) prospects?
In the case of by-elections, with turnout down 30 points in Mid Beds and 28 points in Tamworth, the raw vote tallies give us a better clue as to what might have gone on regarding party 2019 bases and any switching between them. The table below outlines how the total number of votes and those for each of Labour, the Conservatives, and the Liberal Democrats changed between 2019 and yesterday in both Tamworth and Mid Beds.
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1617f682-82d5-47f3-b6af-86d4edd2388f_627x535.png)
An immediate and crude assessment is this: the drop in the Conservative vote share in both constituencies can, purely mathematically speaking, be almost entirely explained by the drop in turnout.
The drop in terms of total votes for the Conservatives in both seats is almost exactly equal to the overall drop in the number of voters who turned out to the polling booths.
Labour, in both seats, while hugely up in terms of vote share swing, saw their total number of votes barely change between 2019 and last night.
In fact, there is much more evidence of direct Conservative-Liberal Democrat switching in Mid Bedfordshire than anything else.
In other words, it is mathematically completely plausible that almost no direct switching from the Conservatives to Labour happened last night.
Now, realistically, that isn’t a sensible conclusion to draw for two main reasons:
Without individual level voter data, we should never made firm conclusions about switching from one party to the other.
We have polling, anecdotal, and briefing evidence to suggest that direct switching is happening in the public at large, and did happen in both seats last night.
It is much more likely that there was indeed direct switching from Labour to the Conservatives - and indeed across the party space - to some degree. But that degree is much more in line with the ~10% figure we are seeing in the polls than reflective of the 20+ point swings on the night.
The story from the raw numbers, and that from the Conservatives’ internal briefing leaked to Sky’s Sam Coates this week, suggests that the outcomes last night were much more about Conservative 2019 voter abstention than they were a mass endorsement of Starmer’s Labour.
The Conservatives, it looks like, simply could not get their 2019 voter coalition to turn out for them.
Again I must express - this would be entirely in line with the polling, which suggests that Labour’s 20+ point lead is driven by a much greater degree by Conservative 2019 uncertainty than it is Labour winning votes.
Is that good news for the Conservatives?
Bad results driven more by voter abstention than voter switching might seem like something for the Conservatives to be happy about, particularly in the context of losing two safe seats on mega swings.
Surely then, you might argue, the Conservatives have a fairly straightforward task in terms of turning things around, avoiding future Tamworths and Mid Bedfordshires, and winning (or least denying Labour a majority at) the next general election: motivate and mobilise the base.
Get the 2019 voter coalition back out.
But a simple question quickly throws the coldest of water onto those electoral embers: what evidence is there to suggest that this great raft of ‘Conservative 19 - Unsure/abstaining now’ voters will come back?
No change of leader, no policy announcement, no U-turn or row back, no ‘policy week’, no conference speech, and no pledge has pulled any significant numbers of drifters back into the Conservative column so far.
So where is this recovery of the “don’t knows” coming from?
Looking across all the polling data I have the privilege of working with at YouGov, I can tell you that there is little indication across any metric that these voters look like they can be convinced to trust the Conservatives again.
So, while Labour’s vote intention lead - and, I would argue based on the collection of evidence I can see, their by-election successes last night - is mainly rooted in unstable soil, the Conservatives have, by comparison, already been sucked well into the quagmire.
And, the point?
To cut a long story short, my advice in interpreting these results would be to exercise even more caution about what they can tell us regarding the next election than would be normally suggested for by-elections.
Why? Because of this Conservative 2019 abstention and uncertainty effect. It’s in our polls, it’s in the canvassing data, and it certainly looks like it was in strong effect in Tamworth and Mid Bedfordshire last night.
Labour certainly answered the question of whether they can win in Middle-England last night, but don’t let the swings fool you - the jury is still out with a tremendous number of voters.