Adam Boulton: The polling industry should be shut down for being useless if the Tories win general election

Have you ever taken part in an opinion poll? I was about to say that I haven't and then a dim memory surfaced from decades ago of a nice lady with a clipboard asking me questions in Pangbourne, Berkshire.

Whether your opinions have been sampled or not, you will have found it impossible not to see coverage of polls during this election campaign.

There have been more opinion polls on voter opinion published than ever before - not least because technology such as the internet has made them easier and cheaper to carry out.

Reporting of opinion polls has been such a feature of the current race that some are complaining about it. The controversy has been enflamed following the dramatic results shown in recent MRP polls, such as this week's YouGov survey for Sky News.

Critics range from those who argue polls are one-day wonders which take coverage away from the "real issues" to others grumbling that polls distort how people vote, even though they may turn out to be wrong.

It is certainly true that tactical voting - switching from your personal preference to the party most likely to defeat the party you most oppose - is only practical if you have an accurate up-to-date idea of how the competing parties stand in the constituency where you are casting your vote.

There are bans on the publishing of polls at election time in other countries. The blackout period varies. It can be total, several weeks long or even just on polling days, as in France.

Bhutan, Brazil, Canada, Mexico and Venezuela are among countries imposing restrictions on polling as well as Greece, Norway, Poland and a dozen other EU member states.

Back in 1945 there was only one pre-election pollster in the UK, George Gallup. Competition between half a dozen companies only hotted up in the 1970s.

In this election there are at least 18 separate companies publishing voting intention polls. They are busier than ever now the campaign is under way.

In the four weeks since Rishi Sunak called the election on 23 May, 98 reputable national opinion polls were published. That amounts to an average of 25 per week, or just under four a day. Each one is zealously promoted by the client who commissioned it.

Polling in the 21 weeks this year before the announcement was heavy but nothing like at the rate during the campaign. Between January and 23 May there were 175 separate opinion polls, about eight per week.

Politicians have a standard comment dismissing polls when they are asked about them. "The only poll that matters", they like to say, "is the real one when all the votes are cast on election day".

Remarkably Conservative candidates have stopped using this excuse in this campaign.

Their pessimism must be based on what they are taking from those darned opinion polls - both those published and available to all of us and private polls carried out for interested parties.

Cabinet ministers including Chancellor Jeremy Hunt and Defence Secretary Grant Shapps have all taken the unusual step of admitting that a Labour victory is the likely outcome.

With less fanfare, Conservative headquarters is redirecting its resources and its online campaign towards safe seats to try to save as many MPs as they can from being buried in the expected Labour landslide.

👉 Click here to follow the Sky News Daily wherever you get your podcasts👈

Were it not for the data generated by opinion polls, nobody would have any useful information to go on other than hearsay and gut feeling. Instead we can see that Labour has been consistently ahead of the Conservatives since the winter of 2021, around the time news emerged of Boris Johnson's Partygate scandal.

That lead widened following Liz Truss's brief premiership in the autumn of 2022 and has stayed at about 20% ever since. Sustained leads of this kind in aggregated "polls of polls" are unprecedented and record-breaking.

The advice from polling experts is always don't look at "the figures" of an individual poll, look at the "trends" in support for the parties.

By that standard if the Conservatives win this election the entire political polling industry should be shut down for being wrong and useless.

Tory 'catastrophe' predicted

Polls used to be conducted face-to-face. As more people got phones at home, many companies switched to the telephone because they could now reach a representative sample of voters, rich or poor, by dialling.

This has become less reliable since many people switched to mobile phones instead of a phone number listed in the local telephone directory.

Since 2000 the internet has become increasingly dominant for pollsters.

In its early years YouGov offered to pay punters a small sum for taking part as it tried to build a panel that reflected the demographic make-up of the nation. A quarter of a century later it is one of the leading and most respected pollsters in this election.

Read more:
General Election 2024 poll tracker
What the polls tell us about what will happen on 4 July
What are in the party manifestos?

The pollsters have developed new techniques as they try to become more accurate. During this election there have been many more so-called MRP polls. MRP stands for multilevel regression and post-stratification.

In essence, pollsters gather the opinions of a much bigger sample and then try to feed this into what they know about the voter make-up in individual constituencies. The other method is "uniform swing", which assumes the changes in the voting pattern will be similar across the country.

Frankly the maths for both polling systems is beyond me. But MRP is much more complicated. It depends on the accuracy of the statistical models built for each of 650 UK constituencies and the quality of the profile of the original large number of voters questioned.

In this election, MRP polls have shown the Conservatives doing catastrophically badly and losing more seats than with simpler uniform polls.

These calculations are very delicate and there are wide variations between companies.

This week, MoreinCommon, YouGov and Savanta published MRPs on the same day. The number of Conservative MPs predicted was 155, 108 and 53 respectively.

Sir Keir Starmer's Labour majority ran from 165 to 200 and 382. These are enormous and consequential differences but they all point to a massive win for Labour - like most of the simpler polls.

All eyes will be on the broadcasters' exit poll published at 10pm as voting ends on polling day on Sky, BBC and ITV.

This predicts the make-up of the next parliament and is also an MRP-type poll, but with "interviewers" asking voters to complete a secret ballot identical to the official one they just completed at the polling station.

Accuracy has greatly improved since the broadcasters stopped competing with each other and pooled their resources, including the expertise of Sir John Curtice and Professor Michael Thrasher.

You may be lucky enough to play your part in the exit poll's moment of history if you are coming out of one of more than 100 polling stations selected around the UK, and a nice lady or gentleman with a clipboard asks you how you have voted.