The Politics of Polling (Liaison Committee Report)

Lord Lipsey Excerpts

Thursday 19th May 2022 Lords Chamber

Moved by

[Lord Lipsey]

That this House takes note of the Report from the Liaison Committee The Politics of Polling: an update (8th Report, Session 2019–21, HL Paper 197).

[Lord Lipsey (Lab)]

My Lords, I start with the history of this matter. I proposed and was asked to chair the original Select Committee on opinion polls, and chaired it for most of its history, until I had an unexpected engagement with the grim reaper across the river. My noble friend Lady Jay kindly took over and, with her usual supreme competence, finished off the report. I am not saying this to get out of any of the things the committee then decided; I am sure they are all perfect.

That committee reported in April 2018 and there was a debate in the House in July 2018. Under the new procedure of the House, the Liaison Committee set up a follow-up committee to see how its recommendations fared. It reported in December 2020 and, after the usual inexcusable lag, this is the report we are debating today. In that context, I particularly thank that wonderful servant of the House, Michael Collon, whose work as clerk to the follow-up committee—I think his last for this House—showed why he was a legend in his lifetime and a remarkable man.

As the follow-up report points out, the Liaison Committee report differs from most of those that are considered by this House, as did the original report, in that its recommendations were mostly targeted not at the Government but at the polling industry. We therefore do not have to focus on the Government’s reply—the kind of stuff that greets Select Committee report after Select Committee report, and reflects some civil servant’s attempt to disguise the Government’s refusal to seriously consider the central recommendations in warm guff. The quality of our Select Committees is reduced in its impact by the Government’s refusal to take them seriously.

Even in this case, the Government turned down one recommendation—an enhanced role for the Electoral Commission. That comes as no surprise now, since the Government have shown through the Elections Act that the idea of an independent elections commission is foreign to their nature.

However—and I say this with great satisfaction—the industry’s response has been very different. In particular, I thank Sir John Curtice, the head of the British Polling Council, and Jane Frost, of the Market Research Society, for what they have achieved in bringing polling into the present century.

I think that we can claim that we had some influence on this as our recommendations helped the progressives in polling to push ahead against some resistance. I am particularly surprised that good progress has been made on the problems of polling and, at least as important, on the reporting of polling, which was so evident following the polling disaster of the 2017 general election—do we all remember Theresa May’s landslide that was not?

Polling is now more transparent, the technical issues surrounding it are more understood, its spokespeople are more measured in their claims and, most of all, reporting is now, with rare exceptions, less misleading. The days when you had stories saying, “Disastrous poll for Labour”, when all that had happened was that Labour had gone down 1% in its poll rating—well within the margin of error, as our committee pointed out—have mostly gone, thank the Lord.

I cannot say, however, that I am 100% confident that the polls will not get it wrong again. It has happened intermittently throughout the history of polling. Mark Pack, who produces an invaluable polls blog, reminds us of this in his new book, Polling UnPacked. As the first editions of the Chicago Tribune appeared in 1948 with the headline, “Dewey Defeats Truman”, Pack reports that pioneer pollster George Gallup turned to his staff and said,

“Boys, I think we’re in trouble”.


This happened in this country too, in 1974, 1992, 2015 and 2017. I remember the report of Professor Patrick Sturgis, now of LSE, on the 2015 failings. He was, incidentally, a distinguished specialist adviser to our committee.

Talking of errors, I remember most vividly and very personally the American election of 2019. I am generally only a small gambler—£5 each way is a lot for me—but in that particular case the polls were so overwhelmingly certain that Biden was going to win that I had, I am afraid, a four-figure sum on his victory. I went to bed that night—before midnight—cheerful. I woke in the middle of the night to the BBC reporting, “Well, it’s looking as if Trump has won.” I did not easily get back to sleep. We now have the report of the American Association for Public Opinion Research on that poll, though I am afraid that it concludes that it was impossible to identify the source of the error, which does not give total confidence that it will get it right in future. Anyway, as a result of that very poor night’s sleep, I am finding it possible to resist the temptation to back Labour to be the largest party after the next election in Britain, though I find the current price of 6/5 against unbelievably tempting.

As poor reporting lies at the heart of so much of perceived poll failures, I start with that. Better practice was first encouraged by ESOMAR, the European polling association, which published an excellent guide written by the doyenne of British polling journalists and a great friend of this House, Peter Kellner. More recently, the National Council for the Training of Journalists produced a professional training course for journalists, which was developed with the BPC and MRS. I was really pleased to be present at the launch of its short version of that guidance, which is great.

Sympathising a little bit with my old profession, I have to say that good reporting is not quite as easy as you might think. I reported on polls for the Sunday Times. There was a good deal of cross-pressure on me as a journalist. On the one hand was the sacred duty to report accurately and without hyperbole, and on the other hand was the news desk wanting the biggest possible splash. You might have thought that the pollsters would be the allies of the careful journalists. Certainly, MORI, which did our polling, wanted sight of the copy as we submitted it, but it, too, wanted a big splash; indeed, I think there were more occasions on which Bob Worcester, the redoubtable boss of MORI, pushed me to puff up the findings than occasions when he asked me not to go too far. One still gets the odd distortion—I had to take the Express to IPSO recently for presenting a poll of its readers as if it were a proper poll, because the idea that Express readers are representative of the nation as a whole seems bizarre—but such errors are much less frequent, and we are better off for it.

As for transparency, the BPC has certainly tightened its practices, with details of the methodology appearing on its website. It has taken on board a huge increase in the number of polling companies—I think that 28-odd are in business now, and all those are now carefully scrutinised.

After correspondence with Full Fact—I declare an interest as a past vice-chair of that organisation—the BPC has introduced a new rule that lays an obligation on members to check whether the accuracy of the figures quoted in any initial publication of a poll is justified, so that things can be put right. So things have moved forward, and I would claim on behalf of my excellent committee some share of the credit.

However, there is always a danger that things will go wrong again. It is true that, on average, polls have not got worse over the years—research by Will Jennings of Southampton University shows that—but, in reality, it is not the average error that matters; it is the disasters that everyone remembers and that can have a consequential effect on politics, which is why, for example, the French do not allow publication of polls during election campaigns. I therefore remind the polling industry of what our committee said. We came down against a ban on polling during the elections—we were not copying the French—and we came down against heavy-handed regulation, but we also said that if the polls messed up by too much too often, such issues would need to be revisited. It is vital that they keep their eye on the ball and keep their standards up; otherwise, this Parliament will have to make up for their failings.

To be fair to the industry, polling is a lot more difficult than it used to be when I started in the trade. A voting intention poll was quite easy in those days. You just had to get the right number of working-class people in your sample. Working-class people all voted Labour, so if you got the right number of them in your sample, you would get the Labour vote about right. That was before the days of the blue wall seats. The industry is adapting. For example, it now turns out that an important factor in voting intention is education, particularly university education, with graduates being much more likely to vote Labour than people who have not been to university—I could perhaps carry that as an accolade for my party; at the same time, if we do not get the working class back, we are going to be in trouble.

Techniques are still evolving. On balance, it is best for government to keep its distance and allow experimentation to proceed. Nobody has yet supplanted the basic principles of opinion polling, but there are interesting methodological developments that need to be tried and tested. That said, polls are too important to be left entirely to pollsters and, more crucial still, too important to be left entirely to journalists. The price of democratic liberty is eternal vigilance, and that includes, as the Liaison Committee report shows, eternal vigilance over opinion polls and their reporting. I beg to move.

--- Later in debate ---

[Lord Lipsey (Lab)]

My Lords, I thank the Minister for that thoughtful reply. I particularly thank noble Lords for the extraordinarily kind remarks they made about me personally, especially when, if I had popped it, they might not be kept in this House on a Thursday afternoon when the last summer sun is shining outside. I have always sworn never to use the phrase “the House of Lords at its best”, but I tiptoe near that by saying that this has been like an excellent webinar on the present state of polling, and I appreciate it.

To correct any misapprehension, I should say that I am not at all Pollyannaish about polls. I worry about their vulnerabilities, and in particular, to introduce one final new point, I am worried about their dirty little secret: essentially, the percentage of the public who agree to respond to pollsters’ questions keeps on falling, and as you get dodgy social media the whole time, people are more and more fearful of committing their views, and that will make it much more difficult to predict in the future. This is a concern that I know everybody in the industry shares, that everyone reading polls should share, and which we need to keep a careful eye on.

This is not the last word. We are not in a bad place today; we are in a better place than we would have been without these reports and the consequent action. However, we have not yet reached the sunny uplands for ever, and we need to keep our eye on what is going on and be ready to act if things start to go wrong in a way that genuinely endangers our democracy.

Motion agreed.