One month ago, shortly after the General Elections were announced, I shared this update.
At the time, Conservatives were around 20%, Labour 39%, Reform 9%, with about 14% Undecided.
The Conservative leader (currently Prime Minister) did some unpopular things, and Reform - the party ideologically closest to Conservative - has well-known figure (and former Conservative) Nigel Farage playing a more prominent role.
In short, if you're trying to forecast, figuring out how much support the major party (Conservatives) have, and how much support the conservative-leaning third party (Reform, in this case) truly have is a tough job.
Pricing in some strategic voting (Reform defecting to Conservative) plus accounting for the conservative lean of undecideds allowed me to have a very accurate forecast for the London Mayoral Election.
Indeed, that was how I believe things were shaping up for the national elections, until recent events.
In short, Labour 39%, Con + Reform 29%, IDK 14% one month ago.
Here's an updated poll average.
Labour seems to have shed a little support (now around 36.5%)
Con + Reform around 32%
IDK around 13%
But in elections, Con+ Reform don't get to combine. Splitting their support - which is looking increasingly likely to happen - will likely cost them many seats.
There is still a very good possibility support consolidates somewhat behind one of these parties, and Labour does not achieve what is viewed as a consensus blowout right now. An easy majority (420+ seats) remains the most likely outcome for sure, but it would be very short-sighted to assume that's a given.
To illustrate, I want to provide something of a visual:
The “Simultaneous Census” standard
Based on the poll averages, here is what I believe the range of possible CURRENT states.
In other words, if instead of a random sample, a census of likely voters (as qualified by each poll's methodology) were asked right now, the results would probably fall within this range:
You'll note, this average includes a lot of undecideds. That's because, if you asked everyone who they would vote for right now, some of them would say they're undecided.
This isn't a problem, but it is a limitation of poll data that must be accounted for. Not providing a future value should be a given, but due to the misunderstandings within the field, I have to repeat it.
How we allocate/model undecideds (and even if some voters will change their minds between now and the election) are subject to errors. Those errors are not poll errors.
My first and second forecasts maintained a wide range of outcomes. That wide range reflected what has proven to be a justifiable amount of uncertainty.
As the election approaches, there is still room for things to change - and all forecasts will have uncertainty - but that should decrease.
Here is my first four-party forecast.
Some notes, according to my forecast:
There's about a 1/5 chance that Reform gets more votes nationwide than Conservatives.
There is a roughly equal chance that Labour wins by a margin of 15-20 (25%) as a margin of 25+ (25%).
The most likely outcome currently is Labour by 20-25 (40%).
The odds that the election is closer than almost anyone expects - a margin of victory under 15 points - is about 8%. This would make things very interesting for seat shares.
A single digit margin of victory has under a 1% chance, largely attributable to the Con/Ref vote not consolidating. This number was close to 2% last month.
A Conservative (or Reform, I guess) win looks to be about 0 right now.
As of June 23:
50% chance Labour finish with 41%-45%
50% chance Conservatives finish with 18.5%-23.5%
50% chance Reform finish with 14%-18%
50% chance Lib Dems finish with 10%-13%
80% chance Labour finishes with 39%-46.5%
80% chance Conservatives finish with 17%-26.5%
80% chance Reform finish with 11.5%-20%
80% chance Lib Dems finish with 9%-14%
Updated seat forecast will be added here soon.