Our last article broke down the data we have collected over 3 years, and allowed us to refine our method of converting constituency-level bookmaker odds into a full General Election model.
With that out of the way, it's time to unveil our initial findings!
With that out of the way, it's time to unveil our initial findings!
The data
The odds we use are an average of odds offered by a variety of bookies. At the time of writing there are six constituencies that currently don't have a market (see footnote [1]).
One aspect of this election which is different to the three previous ones we have modelled is that the constituencies have changed. Whilst there are still 650 seats the boundaries are changed, with many seats having changed names, some new ones appearing and some disappearing. For each seat if it shares at least 50% of its area with an old constituency then we consider that constituency's MP the incumbent.
To pick a completely random example, here is the map of the new constituency Fylde (taken from the House of Commons Library page on boundary changes):
The odds we use are an average of odds offered by a variety of bookies. At the time of writing there are six constituencies that currently don't have a market (see footnote [1]).
One aspect of this election which is different to the three previous ones we have modelled is that the constituencies have changed. Whilst there are still 650 seats the boundaries are changed, with many seats having changed names, some new ones appearing and some disappearing. For each seat if it shares at least 50% of its area with an old constituency then we consider that constituency's MP the incumbent.
To pick a completely random example, here is the map of the new constituency Fylde (taken from the House of Commons Library page on boundary changes):
This constituency is shown in purple. It is made up of parts of two old constituency (green border): Fylde (left) and Preston (right). The new Fylde constituency is 93% old Fylde and 7% old Preston. So the Fylde MP (Mark Menzies) is considered the incumbent for Fylde. So this seat is considered to be held by independents. Should, say, the Conservatives win this seat it would be a gain.
The data in this article was collected on 4th of June, just before the leader's debate.
With the housekeeping out of the way, let's get onto what you're here for!
The TGIAF 2024 General Election projections
The metric we use is 'Expected seats'. This is found by summing each party's total probability across every constituency. This is explained, among other places, here.
Here are the results:
The data in this article was collected on 4th of June, just before the leader's debate.
With the housekeeping out of the way, let's get onto what you're here for!
The TGIAF 2024 General Election projections
The metric we use is 'Expected seats'. This is found by summing each party's total probability across every constituency. This is explained, among other places, here.
Here are the results:
There are two parties included in our data absent in the table above: Alba and the Workers Party. In both cases this is because they have no odds listed for any seat (see footnote [2]).
Anyway, onto the headline results:
A reminder that this is with six seats missing, so the final numbers will slightly higher for some parties.
Seat favourites
Another way to look at our data is to just see which party is favourite in each seat and count the totals. This is ultimately the method used by pollsters because it is simpler.
The benefit of this method is that we can declare individual seats and therefore see where parties are expected to gain or lose seats.
Using this method, here are the results:
Anyway, onto the headline results:
- Our seat estimate has Labour earning a majority of 184, with 418 seats in total. This would be the largest majority since the Conservatives earned a 210 majority exactly one century ago.
- The Conservatives would end up on 133. This is 232 lower than in the 2019 election.
- The Liberal Democrats would overtake the SNP to become the country's third-biggest party again, with their increase from 15 to 47 seats almost a mirror of the SNP going from 41 to 18.
- The smaller parties see very little change in their seat counts. The biggest gains could be made by Reform UK, who are estimated to gain four seats.
- Northern Irish parties also see little change. The DUP and Sinn Féin will share the bulk of the seats, with the SDLP, Alliance and Ulster Unionists taking the remainder.
A reminder that this is with six seats missing, so the final numbers will slightly higher for some parties.
Seat favourites
Another way to look at our data is to just see which party is favourite in each seat and count the totals. This is ultimately the method used by pollsters because it is simpler.
The benefit of this method is that we can declare individual seats and therefore see where parties are expected to gain or lose seats.
Using this method, here are the results:
As well as the six missing constituencies, three more are missing from the table above, as they are toss-ups - that is, seats where the odds have two parties dead level. They are:
Anyway, the take-aways from this new table are:
Summary
Time will tell as to which of the two methods we've shown will prove the more accurate. What is abundantly clear is that our model is projecting a huge Labour majority. If the second scenario comes to pass it would be a monumental result.
The case of Reform and the Greens is an interesting one, as our probabilistic expected seats model has Reform quite a way higher, with 4.7 seats to the Greens' 1.9. But as Reform are actually only favourites in one seat and the Greens in two, the second model has the Greens higher. The difference is that the second model has Reform as roughly 10-20% likely to win dozens of seats, and calculates that they will pick up at least a few of these.
We will be producing more articles in the run-up to the election, and soon will have a tool so that you can look at the individual predictions for each seat.
- Bicester & Woodstock (new seat; projection: CON/LIB)
- Maidstone & Malling (incumbent: CON; projection: CON/LAB)
- Sutton & Cheam (incumbent: CON; projection CON/LIB)
Anyway, the take-aways from this new table are:
- Labour would have an even bigger majority, up to 214 which would make it the largest majority in British history.
- The Conservatives' total of 116 would be their lowest seat total ever, going back to the party's formation in 1835. This would still be the case were they to win the nine missing constituencies.
- Compared to the expected seats method, Reform UK would just stay on 1 seat.
Summary
Time will tell as to which of the two methods we've shown will prove the more accurate. What is abundantly clear is that our model is projecting a huge Labour majority. If the second scenario comes to pass it would be a monumental result.
The case of Reform and the Greens is an interesting one, as our probabilistic expected seats model has Reform quite a way higher, with 4.7 seats to the Greens' 1.9. But as Reform are actually only favourites in one seat and the Greens in two, the second model has the Greens higher. The difference is that the second model has Reform as roughly 10-20% likely to win dozens of seats, and calculates that they will pick up at least a few of these.
We will be producing more articles in the run-up to the election, and soon will have a tool so that you can look at the individual predictions for each seat.
Footnotes
[1]
|
The six missing constituencies are Alloa & Grangemouth, Chorley, Maldon, Rochdale, Stirling & Strathallan and West Suffolk. Chorley is the seat of the Speaker (Lindsay Hoyle), who traditionally isn't challenged by any of the main parties. So this seat is very unlikely to matter in terms of the electorical calculations.
|
[2]
|
Alba currently hold two seats (Cowdenbeath & Kirkcaldy and Lothian East), the former of which is being contested by the incumbent MP (Neale Hanvey). It is not clear whether they are absent from the odds because the bookmakers have written them off, or for some other reason.
The Workers Party hold one seat (Rochdale). This is one of the six missing constituencies. |