The actual election outcome is shown above side-by-side with the final Electoral Calculus prediction. The Speaker is taken as Labour. Note that vote shares are given on a GB basis (excluding Northern Ireland). On a UK-wide basis, the Conservatives got 43.6% and Labour got 32.1%.
The polling error was relatively small, with the estimated Conservative lead of 9.4pc being close to, but lower than, the actual lead of 11.7pc.
A rule of thumb is that a prediction is pretty acceptable if the seat error for the major parties is within twenty seats. For the Conservatives, the prediction error was 14 seats and for Labour it was 21 seats, which seems a fair result.
This seems a good practical illustration of the ability of regression-based polling analysis, which can be effectively done with moderate sample sizes, and which Electoral Calculus has pioneered.
It also provides a welcome relief to the polling industry overall, which had a relatively good performance, after several years of disappointing results.
More details and analysis to follow.
Posted 13 December 2019
The Electoral Calculus voter migration graphic has been updated for the recent election. An important feature this year were the voters who abstained from voting, particular Labour leavers. Also notable was the success of the Conservatives in keeping their voters loyal, even the Remainers. Read details.
Read the full analysis only on Electoral Calculus.
Posted 16 January 2020
Much of the post-election analysis has focused on the perceived defects of the Labour campaign and Jeremy Corbyn personally. Less remarked upon has been the remarkable popularity of Boris Johnson. Johnson significantly outscored Corbyn in polling on who would be the best Prime Minister. But he also achieved an instant and significant turnaround in the fortunes of the Conservative party.
The graph below shows the monthly Electoral Calculus seat predictions since the previous election in 2017. Under Theresa May, the Conservatives were never predicted to win a parliamentary majority (with one narrow exception in February 2019). Once Boris Johnson took over on 24 July, the Conservatives were never predicted to be short of a majority.
We can explore the details of what happened in this eventful year, by looking at some key political dates. The table below shows those key dates, along with the opinion poll lead of the Conservatives over Labour, the Electoral Calculus seat prediction, and the predicted Conservative majority.
|Date||Event||Con-Lab lead||Con seats||Con Maj|
|03 Jun 2019||Euro Election aftermath||-4%||54||n/a|
|12 Jun 2019||ComRes/Telegraph hypothetical poll||15%||395||140|
|02 Sep 2019||Polling after Boris Johnson becomes PM||8%||356||62|
|30 Oct 2019||Start of campaign||10%||363||76|
|30 Nov 2019||Campaign high for Labour as Lib Dems squeezed||8.5%||336||22|
|12 Dec 2019||Final Electoral Calculus prediction||9.5%||351||52|
|12 Dec 2019||General election outcome||11.5%||365||80|
Following the Euro elections in May, the Conservatives polling at the end of the month was dreadful. They were on 19pc, while Nigel Farage and the Brexit party were ahead of them on 24pc. An election at that moment could have seen the Conservatives almost wiped out and reduced to around 50 seats in the House of Commons.
Whilst the Conservative leadership campaign was in full swing, there was a ComRes/Daily Telegraph poll on 12 June which asked the hypothetical question of how people would vote if Boris Johnson became Prime Minister. The poll showed a Conservative lead of 15pc which Electoral Calculus converted into a Conservative majority of 140 seats. This analysis was widely dismissed as unrealistic at the time by a professor of politics ("bulls••t"), the Guardian ("Theresa May was going to get a landslide too...") and the BBC's Laura Kuenssberg ("treat with... a bucket of salt"). Although the poll was not a precise prediction of events to take place six months later, it was broadly accurate in forecasting a complete turnaround in Conservative fortunes and a large Conservative majority under Johnson.
Once Johnson became Prime Minister, the next Electoral Calculus poll-of-polls at the end of August had the Conservatives 8pc ahead of Labour, and predicted to get a majority of 62 seats. This was the first time since the previous election that the Conservatives had been predicted to get a working majority from voting intention polling.
Two months later, at the start of the election campaign at the end of October, the Electoral Calculus poll-of-polls showed a Conservative lead of 10pc over Labour and a majority of 76 seats. This followed the Conservatives successful squeeze on Brexit party supporters as Boris Johnson indicated he was serious about achieving a Brexit deal.
The Labour party hit their high point around the middle of the campaign in late November by squeezing Liberal Democrat supporters. The Electoral Calculus poll-of-polls at the end of November showed a Conservative lead of 8.5pc over Labour and a smaller Conservative majority of 22 seats.
The final campaign prediction from Electoral Calculus was published early on election day. Using our latest regression-based seat-by-seat predictions, our poll-of-polls showed a Conservative lead of 9.5pc giving them a majority of 52 seats. The actual election result had a Conservative lead which was 2pc higher at 11.5pc, and with a few more seats as a consequence. If the pollsters' accuracy had been completely precise, rather than just very good, then our seat predictor would have predicted a majority of 74 seats.
Before this year, Boris Johnson had built up a reputation as an election winner. Winning twice as Mayor of London in a Labour-leaning city was proof that he could reach more parts of the electorate than Conservatives normally could. In the leadership contest, Conservative MPs and members judged that he could do the same at the national level. That judgement has been vindicated by all the polling since his selection and especially by the general election result.
Posted 20 December 2019
The Electoral Calculus has now been updated with detailed data of the 2019 election result. Available features include:
|Seat details pages now show the actual election result compared with our prediction, so you can see how
accurate we were in each seat (spoiler: we got the winner right in 605 out of 650 seats).
It also shows the party winner at the the last three elections, and gives a map of the seat with the estimated winner in each ward at the 2019 election.
|Results have been estimated at ward and street level using our proven regression-based techniques. See how your own neighbourhood voted by entering your postcode.|
|The browsable seat map has been updated to show the new results, giving
a graphical view of the election result.
And you can click on any seat to view the seat in detail.
|Our data map has also been updated to show the 2019 result, as well as the 2017 and 2015
results. Clicking from one view to another gives a way of seeing how political opinion has moved over the last four
Also newly available are estimates of social attitudes (from liberal to conservative), and the Electoral Calculus tribe identity.
The map can show results at seat level, ward level, or street (output area) level.
Posted 19 December 2019
Electoral Calculus has projected the result of the 2019 election onto the new constituency boundaries proposed by the Boundary Commissions in 2018.
election result 2019
|Implied result at 2019|
under new boundaries
The projection gives the Conservatives a majority of 104 in the new 600-seat parliament, with the party winning 352 seats. The Labour party is projected to win fewer than half that number, only winning 174 seats. Many of the smaller parties lose a few seats with the Liberal Democrats predicted to gain as many seats as the DUP and Sinn Fein.
Note that the government may or may not press ahead with new constituency boundaries. If they did, they might either use the 2018 proposals, or they might ask the Boundary Commissions to refresh the proposals. They might continue the plan to shrink parliament to 600 seats, or they might retain the current 650 seats. In any event, it should be expected that the Conservatives would benefit from a boundary review since the smaller seats tend to be in Wales and the North where Labour is stronger.
See also today's story in the Daily Telegraph.
Posted 19 December 2019
The final Electoral Calculus prediction is made up of two components: the poll of opinion polls and the seat predictor. Both parts performed well.
The poll-of-polls showed a Conservative lead over Labour of 9.4pc compared with an actual lead of 11.7pc. The difference between those, an error of 2.3pc, is low by historical standards. The equivalent lead errors were significantly higher in recent elections, with an error of 4.3pc in 2017 and a large error of 6.3pc in 2015.
Even with the poll error, the seat predictor performed very acceptably. The two major parties were within the margins of error, and the Conservative prediction of 351 seats was only 14 seats too low. Of all major UK seat predictors in 2019, the Electoral Calculus final prediction came closest to the actual result. Other predictors generally had the Conservative seat prediction too low, such as FocalData (337 seats), YouGov (339 seats) and Datapraxis (344 seats).
Of course, there is more to prediction accuracy than the headline seat count, since every seat is predicted individually which provides 650 opportunities to be right or wrong. Looking at the final Electoral Calculus seat-by-seat prediction, the winner was predicted correctly in 605 seats and incorrectly in 45 seats. That is a success rate of 93pc, which is the same as the high success rate of YouGov's final large-scale MRP in 2017. About a third of the incorrect seats were marginals which could have gone either way, leaving 31 seats which were mis-predicted.
We can also separate out the errors caused by polling and by the seat predictor. If we feed the actual national vote shares for GB and Scotland into the Electoral Calculus seat predictor we get the following result:
You can see this for yourself by running the seat predictor with the correct national and Scottish vote shares: run predictor.
This is quite an accurate performance, with all parties correctly predicted to within just three seats. As it happens, this is as accurate as the TV Exit poll.
Of the two causes of error: polling error and predictor error, the polling error was actually the larger component, even though it was fairly small. The error in the prediction model was quite small indeed.
Posted 16 December 2019
Many people are interested in a single file which contains all the constituency results in one place. This is now available for download as an Electoral Calculus flat file (53k).
The data format is explained on the Historical Data page.
Please note this is a first draft which has been prepared rapidly and not cross-checked. Please let us know if you see any errors or omissions.
Posted 16 December 2019
|Party||2017 Votes||2017 Seats||Pred Votes||Low Seats||Pred Seats||High Seats|
Prediction based on opinion polls from 04 Dec 2019 to 11 Dec 2019, sampling 23,869 people.
|No overall control|
The future is never certain. But using our advanced modelling techniques, we can estimate the probability of the various possible outcomes at the next general election. Minority government probabilities assume possible alliances between the Conservative and Brexit parties, and between Labour, Liberal Democrats, SNP and Plaid Cymru.
In the headline prediction table, the columns 'Low Seats' and 'High Seats' give the 90 per cent confidence interval for the number of seats won.
This revised version of the battleground (originally from 11 December) has now been amended to show the actual election result of Conservative 44.7pc and Labour 33.0pc. The actual result is marked by a yellow dot.
The actual result is well within the 90pc confidence area (red oval), which shows that the polling industry had a fairly accurate performance this year, with many pollsters being quite close to the actual result.
Posted 13 December 2019
This graphic shows the electoral battleground as a two-dimensional map. The horizontal x-coordinate, or longitude, is the possible Conservative vote share running from 25pc to 50pc, and the vertical y-coordinate, or latitude, is the possible Labour vote share which goes from 20pc to 45pc.
Using the Electoral Calculus seat predictor, we can convert the Conservative and Labour vote shares into seats (assuming the other parties gain and lose votes in proportion to their current vote shares). These seat predictions can then be categorised into zones depending on the overall parliamentary result. The main categories are: Conservative majority government, Labour majority, Lib Dem majority, Labour minority government, or a hung parliament in which there is no obvious majority coalition. The chance of a minority Conservative government is vanishingly small because the Brexit party has faded and the DUP seem unlikely to reforge an alliance with the Conservatives.
The current Electoral Calculus poll-of-polls is marked with a red dot labelled "We are here", and various recent polls are marked by green dots. The slanted red oval, which is actually an ellipse, is the 90pc confidence area for the actual election result. In other words, it is pretty likely that the actual election result will be somewhere inside the red oval. The oval is centred on the poll-of-polls but has sufficient size to include all the various pollsters, plus a little extra. Experience has shown that the pollsters are not always correct, so margin has been left for polling error.
Although the central case is for a moderate Conservative victory, other outcomes are possible. The oval clearly extends into the hung parliament zone and nearly touches the edge of the Labour minority government. At its other extreme, the confidence oval reaches deep into Conservative majority territory and includes a landslide Conservative majority of 224 seats.
The main outcomes which look unlikely to happen are any majority government by a non-Conservative party. But otherwise, the other options are all (just) possible.
Posted 11 December 2019
A major new piece of analysis by Electoral Calculus and pollster ComRes creates a three-dimensional landscape of British politics and identifies the seven political tribes which occupy it.
Electoral Calculus has moved beyond a simplistic one-dimensional view of politics, and even gone further than our own pioneering two-dimensional view, to create a three key dimensions of political attitudes:
Using major polling work by the British Election Study, we can give a political three-D position for each poll respondent identifying their political position on each of the three dimensions.
Groups of like-minded individuals can be spotted to identify seven political tribes of people with similar political attitudes, and gain insight about their demographics and voting behaviour.
You can also find out your own 3D political position, and that of your neighbourhood, by taking our short 3D survey.
Posted 20 April 2019