The most ambitious crossover in [political] history
A strategy in tatters
In my last post I argued that the Conservative strategy for this election campaign has two main strands:
bring back the wavering habitual Conservative voters
consolidate the right-wing vote, squeezing the Reform party’s share as much as possible
How’s that going? I can only summarise as follows: not well.
A number of important (or at least interesting) things happened during the first couple of campaigning weeks, but in this article I will limit myself to just one: the return of Nigel Farage as leader of the Reform party.
Pollsters disagree on the precise level of Reform’s support, recording numbers in the last few days that vary from 11% to 17%. This is way outside the margin of error and reflects different assumptions and techniques among pollsters. But they all agree on one thing: Reform saw an immediate polling bounce of about three percentage points, mostly at the expense of the Conservatives.
You can see the ‘Farage Effect’ quite clearly here1:
Some commentators felt that the surprise announcement of the election was in part a Conservative attempt to de-fang Reform—to catch them off-guard with little chance to find candidates to stand in many seats and (above all) to close off the possibility of Nigel Farage returning.
But the Reform party is a curious outfit. It’s structured (very unusually!) as a limited company in which Mr Farage owns a majority stake, allowing him to avoid such inconveniences as leadership elections. His decision to return to the limelight may prove highly consequential, hamstringing the Conservative attempt to squeeze the newer party.
The Crossover
To be clear: right now it doesn’t seem very likely to me that Reform will win many (any?) seats. However, with their vote share currently as high as 17% (and the Conservatives as low as 19%) it seems quite possible we will get a ‘crossover poll’ in the near future—due to natural random variation and the large number of polls being run during the campaign.
I’m resigned to the fact that people will get very excited about The Crossover if and when it happens ... but it may not mean much. Not unless it goes much further. The Conservatives are already on track to be reduced to a rump of ultra-safe seats; as yet we’ve seen no indication that Reform can seize this territory from them. Of the constituencies where Reform are projected to perform best, over 80% are expected to be safe seats for Labour.
For me, the more important consequence of this Reform surge is its corrosive action on the Conservative vote share, particularly in formerly-safe seats where they are now fighting tooth-and-nail against the other big parties.
I looked at the YouGov constituency predictions (based on polling conducted before 1st of June) and modelled how they would change if Reform were to gain three percentage points at the expense of the Conservatives. Here are the ones where the result changes:
That’s 35 seats where a result flips from the Conservative side of the knife-edge to the other, dropping their total (per the YouGov projection) from 140 to 105.
Give it another three point swing and … well, uniform swing probably stops being a sensible way to treat the data. Perhaps we’d better wait for the next MRP model to be published2.
But will Reform go higher? Or is this the high water mark? The answer depends on some of the other things that have been going on during the campaign. But this post is already long enough—so I will save those for later in the week.
Most polling averages won’t look like this. When pollsters disagree so much (and publish on different dates) the shares for some parties can see-saw up and down, despite this not reflecting any real changes.
For this chart I removed the average differences between pollsters for each party, which really emphasises big changes that all pollsters agree on. That’s at the expense of changes they don’t agree on—so I will likely go back to the usual method in later articles.
But for the record the result flips in 67 seats, leaving the Conservatives with 73, total.