Labour is on course for a crushing landslide later this year… or are they? Many parallels with 1997 are being drawn at the moment but it’s too early to be making detailed forecasts. Instead, I want to lay the foundations for my election forecast
later this year by looking at what 100 years of UK electoral history is telling me. I tell this story by assigning probabilities to 10 specific outcomes of interest and finish by explaining why 2024 is more likely to be a repeat of 2001 than 1997.
I remind you I was deemed the most accurate forecaster of the 2019 general election. This article is the first in what I hope will be a series of articles which lead me to hold on to that accolade!
Data used in this article
All electoral data I display and use in this article comes from the House of Commons Research Library. All polling data I use comes from Mark Pack’s invaluable Pollbase.
1922 to 2019 – 27 UK General Elections
Nearly all political analysts I follow will be looking at charts like this. Their forecasts will be based on their estimates of what the vote shares for Conservatives, Labour, Liberal Democrat and all Other parties will be.
The UK uses First Past The Post (FPTP) as its election system. That makes the Conservative lead over Labour the most important statistic to estimate the outcome. These two parties have battled for 1st and 2nd place in all elections since 1922 which was the first election to be held without Irish seats following the Republic of Ireland’s independence in 1922. With the exception of 1923, 85% or more of all seats contested have been won by a Labour or Conservative candidate in all elections.
At the end of 2023, the average of opinion polls gave Labour a 17 percentage point lead over the Conservatives. If replicated at the general election, that would be Labour’s largest ever victory margin (in terms of vote share) over the Conservatives. The current record is 12.6% in 1997 closely followed by 11.9% in 1945. Excluding 1931 (which was exceptional), the current record lead for the Conservatives was 14.9% in 1983 so if the polls are right, this would be the 2nd largest margin of victory in any election since 1922.
Outcome 1 – Turnout will be lower than 2019 (probability 90%)
In 2019, 32 million people voted out of a total of 47.6 million registered voters. The turnout was 67%.
For background, the large increases in the electorate in 1929 and 1970 are due to significant changes in the franchise. In 1929, the age threshold of 21 was equalised for both men and women (previously only women over 30 could vote) and in 1970, the voting age was lowered to 18. The DNV column in the table on the right is the number of registered voters who did not vote.
If turnout in 2024 were to be the same as 2019 and assuming the growth in the electorate follows the trends of the last decade, I estimate unchanged turnout in 2024 would equate to 33 million voters out of 49.2 million registered voters.
If the polls are right (LAB 43.0%, CON 25.6%, LD 10.7%), then 14.2 million people will vote Labour, 8.5 million will vote Conservative, 3.6 million will vote Lib Dem and 6.0 million will vote for other parties. It’s rare to see vote estimates in numbers but in this instance, the implied numbers are instructive since based on the last 27 elections, 2024 will see –
- A record number of people (14.2 million) voting Labour beating the previous record of 13.8 million in 1951.
- The increase in the Labour vote by 4 million will be a new record beating the previous record of a 3.5 million increase in 2017 (from a low base in 2015).
- The drop in the Conservative vote of 5.5 million from 2019 will be a new record beating 1997.
- The Labour lead of 17.4% over the Conservatives will smash the previous Labour record of 12.6% in 1997.
- The swing to Labour from the Conservatives of 14.5% will smash both the previous record for Labour of 10.8% in 1945 and the 12.4% swing from Labour to the Conservatives in 1931.
I may not be obsessed with politics but I do not pick up vibes from Labour as led by Keir Starmer today that makes me feel they are a better proposition than Tony Blair’s Labour in 1997 and Clement Attlee’s Labour in 1945 capable of achieving these records. However, if turnout was lower than 2019, then points 1 and 2 will not come to pass.
There is a second reason why I think turnout will be lower than 2019. This can be seen by referring to the turnout table above for 1964, 1974 and 1997 which are the last 3 elections when Labour came into power. In 1964 and 1997, Labour won the election on a reduced turnout, whilst in 1974 turnout was higher in the February election but fell back to 1970 levels in the October election.
These two reasons are why I am confident in predicting lower turnout in 2024 and have put a probability of 90% on my prediction being correct.
Outcome 2 – Number of non-Voters will be new record (probability 80%)
In 2019, 15.6 million people did not vote even though they were entitled to. I will denote these non-voters as DNV in the charts and tables that follow.
The current DNV record is just over 18 million non-voters in 2001 and I am giving an 80% probability of this being beaten in 2024. If the electorate is 49.2 million in 2024 (based on recent trends) then a turnout of 63.25% will result in a new record for DNV. This will be an increase of 2.5 million non-voters from 2019 which is less than the 2.9 million extra DNV in 1997 and the whopping 5.5 million extra DNV in 2001 and matches the 2.5 million extra DNV in the 1974 October election.
Two of those elections mentioned are when Labour formed a majority government which is why I am confident of a new record for non-voters.
In the above graphic, the table on the right shows the change in number of voters as a percentage of the total number of voters in the previous election. I consider this a reasonable way of placing the numerical change in the context of a larger electorate.
Outcome 3 – Turnout will be lower than 2001 (probability 40%)
I am confident of lower turnout to the extent that we will see a record number of non-voters which will occur if turnout is around 63% or less. The lowest ever recorded turnout was in 2001 when only 59.4% of registered voters actually voted compared to 71.4% in 1997.
Could turnout in 2024 be lower than 2001? I am not confident of this hence why I only give a 40% chance but I am not ruling it out. If one believes in parallels between 2024 and 1997, then the 6.3 point fall in turnout from 77.7% in 1992 to 71.4% in 1997 would, if replicated, lead to a turnout of 60.9% in 2024, not far off from 2001. Alternatively, I note the total number of voters in 1997 of 31.3 million was 7% lower than in 1992 which, if replicated, would result in 60.6% turnout in 2024.
This is an outcome whose probability I will refine as I dig into the polls to see what they are saying about likelihood to vote. However, polls are generally not a good tool for estimating turnout.
By the way, an immediate implication of the turnout collapse in 2001 was that DNV won the plurality in 2001 and has held it ever since. Prior to that, either Labour or Conservatives had the plurality on number of votes hence why in the left hand table in the graphic above, the two major parties are strongly shaded prior to 2001 to denote the plurality but only lightly shaded since. To me, this explains the psychological importance of turnout exceeding 70% as that nearly always denies DNV the plurality and conveys more democratic legitimacy on the election winner in my view.
Outcome 4 – Lib Dems Votes & Share down on 2019 (Probability 75%)
I currently give a 3 in 4 chance of the Liberal Democrats ending up with fewer votes and vote share than 2019. The fewer votes is a direct consequence of my prediction that turnout will be lower but the lower vote share is a combination of two factors.
First, I am not seeing any significant trend towards the Lib Dems in the polls. This is
evident in the chart near the beginning of this article and in the table shown here. This shows the 9-poll moving average at the end of the months listed.
Second, another 1997 parallel has been the willingness in recent by elections to vote tactically in order to kick out the Conservatives. Sometimes, the Lib Dems are the beneficiaries but in the majority of seats, it is Labour who will be the beneficiaries. The effect of significant tactical voting will depress the Lib Dem vote share relative to the polls whilst delivering them extra seats which is what happened in 1997.
Outcome 5 – Others Vote Share will be > 20% (Probability 65%)
The following graphic shows a plausible scenario in 2024 compared with the last 40 years given the following –
- my prediction of lower turnout
- my prediction of record levels of non-voting
- my scepticism over Labour setting new electoral records
- my expectation of the Lib Dems being down on 2019 in terms of votes and share (but not seats).
I must make it clear this is not an official forecast by myself but simply a scenario which combines the above predictions.
If correct, then the vote share for all parties other than Conservatives, Labour and Liberal Democrats will exceed 20% for only the second time ever after the 24.9% in 2015. I am confident in this but not majorly so hence why I assign only a 65% probability to this outcome. This is a probability I expect to change as new polling data comes in. There are 3 elements I will keep an eye on.
The first is the performance of Reform, the successor to the Brexit Party of 2019 and UKIP before them on the populist right. After getting over 10% of the vote in recent by-elections in Wellingborough and Kingswood, it will be interesting to see if they can sustain this. An unnoticed point is the possibility that they are now starting to pick up voting intentions (not votes) from Labour. In the table I showed before (repeated here), since Sunak took over at the end of 2022, the Conservatives are broadly unchanged in voting intention, Labour is down 4 points whilst Reform is up 4 points.
What makes interpreting these polls more complicated is the second factor which are the SNP troubles in Scotland. They are another driver of the Other vote but I have heard Labour are starting to gain from the SNP in Scotland. If so, then the 4 point weakening in voting intention over the last year or so occurred whilst Labour were making gains in Scotland. That means more voters in England are switching their intentions.
The third element is how many other parties intend to stand in the general election that could offer an option to disaffected Conservative and Labour voters. A little known fact from the 1997 election is the question of who came 4th in 1997 in terms of number of votes with 811,000 to be exact? The answer is the Referendum Party who stood in 547 seats, retained 42 deposits and ended up with 2.6% of the national vote. When I look at recent by elections and the 2019 elections, the following minor parties might be able to garner enough votes to boost the overall Other vote share –
- Britain First & assorted other far right parties have averaged 1.4% in 6 by elections where they have stood.
- The Yorkshire Party has averaged 4% in the 3 Yorkshire by-elections & stood in half of the Yorkshire seats in 2019 where they averaged 2%. The majority of these seats are Conservative seats.
- The Workers Party had a good showing in Batley and Spen and are targeting Rochdale.
Outcome 6 – A major party polling error >4% will occur (Probability 65%)
The scenario I show in outcome 5 has Labour beating the Conservatives by 12.7% rather than the 17.4% given by the polls at the end of 2023. If correct, this would be a polling error of 4.7%. In 5 out of the last 8 elections, a major party (CON, LAB, LD) has ended up with a national vote share more than 4 points away from the average of all polls in the week before the election. This is why I give a 65% probability of this happening again.
The data for the chart below comes from my article “How accurate are UK voting intention polls?“. This again shows an error of 4% or more in the estimate of the Conservative lead over Labour in 5 out of the last 8 elections and would almost be 6 out of 8 when one looks at 2005. 2017 is the only one where the error favoured Labour, all others favoured the Conservatives which includes the 1997 election.
There is weak evidence that errors favouring the Conservatives are more likely to happen in elections where turnout is lower than the previous election. Since 1970, when the age threshold was lowered to 18, 5 elections saw a fall in turnout and 9 saw a rise in turnout. Of the 5 lower turnout elections, 4 had errors favouring the Conservatives (3 larger than 4%) and 1 had an error favouring Labour. Of the 9 higher turnout elections, 6 had errors favouring the Conservatives (3 larger than 4%) and 3 had errors favouring Labour. I’m still analysing this data for more insight.
Key Outcomes & Ancillary Outcomes
Outcomes 1 to 6 above are all what I call Ancillary Outcomes. None of those outcomes determine the key outcome of the election which is whether Labour or the Conservatives will form the next government. It is too early to be giving specific predictions for such Key Outcomes but outcomes 11 to 14 below are all relevant to predicting these.
Outcome 11 – 2 Party Vote Share will be < 70% (Probability 60%)
This is a direct result of the probabilities I have given for ancillary outcomes 4 (Lib Dems) and 5 (Others).
Outcome 12 – Conservative votes will be less than 2001 (Probability 50%)
This outcome depends on what the polls are saying but if turnout is down significantly and the Conservatives do not stage a recovery in the polls, then even with a favourable poll error, the total number of people voting Conservative run the risk of being less than the 8.36 million in 2001 or even the 8.23 million in 1929 when the voting age threshold was equalised at 21 for both men and women.
At the moment it is touch and go as is evident in this plausible scenario I showed earlier.
Outcome 13 – Labour votes will be down on 2019 (Probability 25%)
This is what happened in 1964 when Labour won a majority despite getting slightly fewer votes than 1959 due to a 5.6% swing from the Conservatives to the Liberals. I see this as unlikely at the moment in 2024 but a combination of lower turnout than 1959 and larger polling errors than expected and poorer polling in general, then the probability of this outcome rises.
Outcome 14 – Labour lead over Conservatives will be new record (Probability 40%)
Given the likelihood of a polling error, which if it happens is more likely to favour the Conservatives according to history, the current Labour lead over the Conservatives in the polls is likely to be exaggerated. However, I can’t rule out a new record for Labour hence why I give it a 40% chance. This is an outcome though where I will be revising the probability as new data comes in.
Why 2001 was the most important election of all time
No it wasn’t I hear you cry! It was the most boring election of all time with John Prescott’s punch-up the only thing to enliven it!
I agree it was boring. My best friend and I planned to stay up the night but in the end, only 8(?) seats changed hands, the vote shares were similar to 1997 and we gave it up as a bad job around 3am.
However, history now tells us it was the most consequential election of all time. A minimum of 5.5 million people who had voted in 1997 chose not to vote in 2001 making them the largest block of voters to have visibly changed their voting behaviour in an election. This is apparent from this graphic which I showed earlier where the middle table shows the change in the number of people voting for Conservatives, Labour, Lib Dems and Others as well as the change in non-voters denoted by DNV.
As a percentage of the electorate for the previous election, the change in 2001 DNV was the 3rd largest on record at 12.5% of the 1997 electorate. It was beaten the increase 12.8% increase in the Labour vote in 1929, which was a franchise change year, and the 12.8% fall in the Liberal vote in 1931. The graphic below shows the top 15 voting blocs separately by changes in number of votes and changes in percentage of previous electorate since 1922. The final table on the right lists the 12 voting blocs that appear in both tables and ranks them by their average rank order as shown. Top of this list is 2001 DNV.
Third in the list is the collapse of the Conservative vote in 1997 when a minimum of 4.5 million people who voted in 1992 did not do so in 1997. On current polls, it is likely this will be exceeded and the possibility of exceeding the 2001 DNV cannot be ruled out if outcome 12 comes to pass.
By now, you should have seen that I have referenced 2001 as often as 1997 when seeking to tell the story of what I see in UK electoral history. This is why I said at the start of this article that 2024 may end up being more similar to 2001 than 1997.
List of GE24 Ancillary and Key Outcome Probabilities – 15th February 2024
I plan to update this article at the end of April before the local elections. I will add my revised probabilities to the table along with any new ancillary and key outcomes I think of.
If you would like to submit your own probabilities, I have set up a Google Forms poll here.
— Would you like to comment on this article? —-
Please do leave your comments on this X/Twitter thread.
— Subscribe to my newsletter to receive more articles like this one! —-
If you would like to receive notifications from me of news, articles and offers relating to Elections & Polling, please click here to go to my Newsletter Subscription page and tick the Elections and/or Surveys category and other categories that may be of interest to you. You will be able to unsubscribe at anytime.
— Read some of my other blog posts on elections —
- Jeremy Corbyn’s road to a majority in 2019 and the roadblocks they face – in a nutshell strategies that focus on marginals, Tory-Remain seats, young voters, non-voters are all capable of gaining the 30+ seats that Labour needs to be the largest party in parliament. But the 60+ seats that Labour needs to get a working majority requires Labour to make major inroads into either Conservative Leave marginals, Safe Conservative Remain seats or the SNP seats in Scotland and to do that without losing seats they already have. I believe that Corbyn is trying to sit on the fence on Brexit for as long as possible and try to have an election instead of a referendum so that other issues can take centre stage. Another reason why Labour’s road to a majority is now harder is because they no longer have the solid bank of 30+ seats in Scotland which were swept away in the 2015 Nationalist wave. There is no sign this is going to change soon and so I ask where in England can Labour replace these seats?
- I am the most accurate forecaster of the 2019 UK General Election
- How accurate are voting intention polls for UK General Elections?
More posts can be found by clicking on the Elections tab at the top of your screen.
