This week, as postal vote papers drop on doormats, people have begun voting for their next MPs. Since Rishi Sunak called the election, nearly 70 opinion polls have surveyed more than a quarter of a million people. But despite all the campaigning activity, the results have remained stable. Almost all predict a Labour lead of around 20%. But the size of the parliamentary majority depends on the outcomes in individual constituencies. So what can we tell about the future political map in our region?
Why do we need MRP polls?
Conventional opinion polls sample opinion across the whole country. But the final result in a general election depends, not on national vote share, but on what happens in each of the 650 individual constituencies. This is why the new tool – the MRP poll – is important. This predicts the result in each constituency, by matching the known characteristics of the population (like people’s age and education levels) in each, with national polling evidence on the voting intentions of those groups.
A number of pollsters have carried these out, but sometimes with quite different results. Why is this, and what conclusions can we draw about future MPs in the East of England?
MRP polls are better, but not perfect
Because MRP polls take account of the history of a constituency, the sort of people living there and their past voting patterns, they have proved reasonably accurate in the past. Certainly, they are a better guide to the final seat tally in Westminster than simple national vote share.
But they can’t reflect very local factors, especially the strength of the parties on the ground, which varies hugely between constituencies. Nor can they reliably reflect the strategies of the smaller parties, which usually focus all their campaigning in a few target constituencies.
Since the beginning of June there have been a series of MRP polls. Although they all agree about the national vote share, they disagree, sometimes quite a lot, about specific constituencies.
There are a number of reasons for this. A key factor is the pollster’s ‘model’. MRP polling is built on complex mathematical models, which determine what factors will be taken into account, in addition to the obvious ones like age and education. The model determines how much weight to give to factors like past voting patterns and previous election results, to people who say ‘don’t know’, and to predicted turnout. Each pollster’s model is unique, and unlike polling questionnaires, which are always published, these models are too complex (and commercially sensitive) to publish, so observers can only guess at the reasons for disagreements in results.
A further factor is timing: all MRP polls require a very large sample (10,000 or more) and some pollsters take longer to achieve this. If public opinion is changing, polls conducted over a week or more will not reflect the most recent patterns. This particular issue in the current election is the position of Reform UK. All polls agree that the late decision of Nigel Farage to stand for Reform in Clacton has boosted their vote share, but too late for all the polls to pick it up.
A lot of marginal seats
This year’s general election is very unusual, but even so, some seats are safe for a particular party, and all polls will agree on the likely result. At the other end of the scale there are seats where the result is too close to call. Here, the National Polling Council warns that,
All polls are subject to a wide range of potential sources of error. On the basis of the historical record of the polls at recent general elections, there is a 9 in 10 chance that the true value of a party’s support lies within 4 points of the estimates provided by this poll, and a 2 in 3 chance that they lie within two points.
So it would be wise to discount any apparent polling lead of less than 4%. But in the East of England current polling suggests that a quarter of all seats fall into this category. But no journalist ever made a reputation with a headline ‘No change in constituency X’. So there is always a temptation to ignore the warnings about margins. And of course, sometimes a small change is the beginning of a larger swing.
This is particularly important this time, because there seems to be a very large rise in the number of marginal seats. James Kangasooriam, of Focal Data, has used the results of last May’s local elections to predict the likely size of majorities in the related constituencies this year. He has then compared this with the size of the majority in each seat in the 2019 general election. As the chart shows, in 2019, half of all seats had majorities of over 30 points, but now it looks as if half have majorities of under 15. This means that quite small shifts of opinion can make a big impact on the numbers of seats each paty wins in Parliament, and may explain why the national polls differ so much on their estimates.
Mapping the East
To try to overcome some of the weaknesses of individual polls, we have compared the results in the East of England, of six MRP polls conducted since the beginning of June. They come from Electoral Calculus, Survation, More in Common, Ipsos, and two from YouGov.
The map shows the result. We have classified seats as:
- Safe: where at least four of the six polls agree on the winner, with a majority of over 4% for that party.
- Marginal: where two or three polls agree on a winner with a majority of over 4%, but other polls disagree.
- Tied: where the polls disagree, giving no parties majorities of over 4%
As the map shows, the 61 seats in the East of England are distributed as follows:
The 14 tied seats, where there is no consistent agreement between the polls, or they report leads of below 4%, are:
Labour clearly will make major gains on their 2019 position, especially in the urban areas, and across Hertfordshire commuterland, with some very large majorities in their traditional strongholds. Conservatives still hold on to seats in Essex and across some rural areas, but with much smaller majorities. The Liberal Democrats have two ‘safe’ seats – St Albans (which they already hold) and the new constituency of South Cambridgeshire – and three marginals: Chelmsford, St Neots & Mid Cambridgeshire and Harpenden & Berkhamsted.
Clacton is the only seat where Reform seem the likely winner. There, Ipsos, which is the most recent poll, shows them on 52%, with Labour second on 24%. But there are four more seats where Ipsos shows Reform with over 25% of the vote, and the potential to challenge, since their overall vote share seems to be still rising. These seats are South West Norfolk, Great Yarmouth, and North East Cambridgeshire.
The fourth is Waveney Valley, where Ipsos has Greens leading (the only pollster to predict this so far), with Reform in second place, only five points behind. However, Waveney Valley is the only constituency to have a conventional telephone poll carried out in the constituency itself (normally this is too expensive, which is why we have MRP). A poll of 500 people, carried out there by WeThink polling on behalf of the Green Party shows the Greens comfortably in the lead with 37%, and with the Conservatives second on 24%. But tactical voting is clearly in play here. When they asked how people would vote if they knew the leaders were Conservatives and Greens, the Green vote share rose to 46%.
Change over time
YouGov have carried out two MRP polls since the election was called, which makes it possible to consider changes during the campaign. They show ten seats where the prediction has changed. Three which were too close to call in their first poll now show secure Labour majorities: Broadland & Fakenham, Hertford & Stortford and Huntingdon. Clacton has switched from a 15% Conservative majority to a 20% Reform one.
To conclude
All the polling evidence points to a clear Labour victory on 4 July, barring some unexpected major event. And postal ballots are going out now, so as many as one in five votes could be cast well before then.
However, what that means in our region is much less clear. These polls will not be ‘wrong’ if the Labour seat tally is between 20 and 42 seats, and the Conservatives between 7 and 30. Anything within those ranges will be a dramatic change on the current position, where the Conservatives hold 52 of the 59 seats in the region.
But this is not really much of a prediction. On the night, the real results may come down in some places to a handful of votes. We may well be seeing multiple recounts. It could be a long night.
