Here we go again ...
It's a big year for elections - and I am starting a Substack!
In the end I lost my bet.
I argued that there were good reasons to think the election would be in May. My sister favoured November. Since we agreed that one month either way would count as a win, I lost out by four days: the election has been announced for July 4th.
Why did I favour an early election? There were a few reasons. A big one is that while pollsters are united in saying that the Labour Party presently have a large lead, this doesn’t mean it can’t get worse for the Conservatives. In fact, there are good reasons to think it would get worse; the longer they are seen to be holding on to power as long as they can, the harder it is to claim they have the interests of the country—not the party—at heart.
What about the size of that lead? Some pollsters think it will be 26% (Conservative) vs 41% (Labour); that implies a big Labour win, a humiliation for the Conservatives. But others—equally respectable—have it at 20% vs 45% instead. If that’s borne out then the governing party is perilously close to annihilation.
These differences are much larger than the inescapable statistical margin of error. They exist because there are a large number of wavering voters; pollsters vary in how they treat this group. Some essentially assume these fence-sitters won’t vote. Others ask follow-up questions that force a decision. This tends to lead to smaller Labour leads.
You can guess what that means: a lot of Conservative-leaning voters are undecided.
A foregone conclusion?
Onto the big question: will Labour win?
If I learnt anything from the 2017 election, it’s to avoid certainty (there’s a reason I called my newsletter ‘The Limits of Data’)! In that election something unprecedented happened: an 18-point lead at that start of the campaign was all but eliminated by the end. So in principle we know that it’s uncommon but not impossible for public opinion to shift by that much during an election campaign.
It’s also worth reminding ourselves how that happened. At least some of it was due to consolidation of the left-wing vote; as I said at the time, voters aren’t stupid—and they can count.
Is an analogous option open to the Conservatives? They may think so. Most pollsters put the Reform party (the artist formerly known as ‘the Brexit party’, if you remember those far-off days) at around 12%1. You may recall that they more-or-less stood aside at the last moment in 2019, sealing the result.
So now the bones of the Conservative strategy emerge. Mr Sunak strikes me as a numbers guy; he knows all of what I’ve just said. If he is to succeed2, he must
Convince the waverers. If they vote Conservative after all (as some pollsters think they will), the remaining gap is 15%.
Consolidate the right-wing vote. The highest vote share recorded by any pollster for Reform is over 15%.
It’s not clear to me that Reform voters are amenable to turning Conservative; I suspect that ship has sailed. In any case, even achieving a hung parliament seems like a reach for the Conservatives. But I suppose it’s better to have a plan than not to have one at all.
Let me take one moment to point something out: by 22nd May (the date the Prime Minister stepped out into the pouring rain and announced an election in the unusual month of July), the polls seemed to indicate that Reform had peaked …
So what will happen?
None of that implies it’s likely that this plan will work. Early signs don’t suggest this campaign is going well for the Conservatives. But time will tell.
Let’s finish with some indicative results. When public opinion swings so much between elections, you cannot use traditional ‘swing based’ techniques to figure out how many seats each party will win.
The best method at present seems to be the ‘MRP model’, a sophisticated approach with a good track record of success in UK elections. It uses a very large poll to figure out how different demographics are likely to vote, then projects these onto constituencies based on their population statistics.
The latest YouGov MRP model gives a headline vote share of 41% Labour vs 24% Conservative; one of the smaller Labour leads relative to what other pollsters are reporting.
Here’s how that translates into seats:
A 154 seat Labour majority3 is substantial. For reference, in 1997—a ‘landslide’ election—Labour had a majority of 179, which is comparable.
And yet: consider the headline vote share. This is one of the better results for the Conservatives compared to other pollsters. If the actual split is closer to the extreme of what other pollsters (and YouGov themselves, in their more traditional polls!), we can surmise that the Labour majority could be bigger still.
Ultimately, there are weeks until polling day … and as the 20th century Prime Minister Harold Wilson famously said, even one of those is a long time in politics. However, I hope this post helps you get a sense of what might plausibly happen a month from now.
Why Substack? Well, social media is not what it was. We don’t all use Facebook every day like we did in 2014. The race-to-the-bottom contest between apps as they chase maximum engagement has killed quality.
I’ve been enjoying Substack as a way to find decent things to read. If you think what I write is interesting then I’d love it if you subscribed!
The Reform vote share doesn’t really depend on how each pollster treat undecided voters, by the way.
Here I’m defining ‘success’ as avoiding a Labour majority, although more traditionally he would be aiming for a majority of his own …
As a quick reminder, a government’s ‘majority’ is the number of surplus MPs they have relative to the number they need to equal all the other MPs combined. There will be 650 MPs elected in July. If you have 326, you have a majority of two (two more MPs than all the rest put together). If you have 403 MPs, that leaves 247 MPs you don’t control. Even if they all oppose you, you can match their votes and still have 156 MPs to spare (I am not sure why YouGov say 154 in their graphic).