The Politics of Polling (Political Polling and Digital Media Committee Report)

Lord Lipsey Excerpts

Tuesday 3rd July 2018 Lords Chamber

Moved by

[Lord Lipsey]

That this House takes note of the Report from the Select Committee on Political Polling and Digital Media The politics of polling (HL Paper 106).

[Lord Lipsey (Non-Afl)]

My Lords, I understand that our debate this afternoon is not time limited, so it may give comfort to noble Lords if I assure them that I intend to resume my seat in time for anybody who wishes to catch the kick-off tonight.

It was a real pleasure and privilege to chair this committee. The self-congratulatory phrase “the House of Lords at its best” is grossly overused but I am going to use it once more, because I think if anyone attended our evidence sessions and the grillings that we gave to our witnesses, they would feel that sentiment was justified. We were wonderfully well served by our two clerks—Helena Peacock, until she left for the BBC, then Sarah Jones—and our peerless policy analyst Beth Hooper. We were also well guided by our specialist adviser, Patrick Sturgis, of the University of Southampton, who has also served us subsequently by convening a conference of polling’s good and great to discuss our report.

It was a pleasure to chair this group of people, but it was not easy. At the beginning, the committee members had a range of instincts, from one who was in favour of strict regulation of polling to one who thought that everything was fine and dandy as it was. They were, incidentally, both members of the same political party.

My confidence as chairman was not exactly boosted when, at the end of each of our sessions, the noble Baroness, Lady Jay, would mutter: “I can’t see how we are ever going to get agreement on this”. Little did she know that, as we entered the last stages of our drafting, I would be borne off to Guy’s and St Thomas’. Thanks to them and the NHS, I am standing here this afternoon. That left her to persuade the committee that it would agree on something. I am very grateful for her efforts in that regard.

Most Select Committees of this House are about agreeing a set of recommendations to government for changes in policy and the law. Ours was, I think, slightly different. Three successive votes—the 2015 general election, the 2016 referendum and the 2017 general election—had produced results entirely contrary to what the polls had led people to expect. To misquote Oscar Wilde, to get one election wrong may be regarded as a misfortune, to get two wrong looks like carelessness, and to get three wrong suggests that something somewhere has gone horribly amiss. So our first and fundamental task was to assemble evidence on whether these were blips or might go on happening. Was polling no longer fit for purpose, serving only to mislead voters as to what they collectively thought?

The good news from polling’s point of view is the research by Will Jennings of the University of Southampton, which we quote and shows that internationally there is not much evidence of a decline in polling accuracy over the years. Of course, that could mean that it has not got worse but it could also mean that it was always pretty bad and continues to be bad. That is a matter of opinion. However, these three successive setbacks will have pollsters on tenterhooks about the results they will achieve in the 2022 general election. Even if they get it right, it must be said that, as happened in 2010, they sometimes get it right because their errors cancel each other out.

There is good reason to be cautious in trusting the polls. First, polling is, by common consent, getting more difficult. Our report highlights two main reasons. One is non-response rates. A pollster might approach upwards of 10 people to get one who is prepared to join in and answer the questions, whereas far fewer used to be required. That creates a bias towards those interested in politics, who are much more likely to say yes than someone who knows nothing about it.

The other reason is the decline of social class as an indicator of voting intention. Once upon a time, as long as pollsters got the right proportions of working and middle-class people, they were all right. All the middle-class people voted Tory, all the working-class people voted Labour and pollsters would get the result right. That is not the situation today. Today, Labour gets more middle-class votes than working-class votes. No doubt that makes it much more difficult for pollsters to know whether their samples are right.

Beyond that, there is the separate question of the margin of error. Strictly speaking, there is no scientific way of measuring the margin of error for non-random polls, which all pollsters use today, except one or two state-backed pollsters. Since we reported, however, the British Polling Council has put the margin of error at 4%, based on past poll errors. Let us be clear about what this means: it is not a measure of the margin of error in the total lead of a party. It does not mean the Tories are on 42% and Labour is on 38% within the margin. It is the measure of the error in each party’s share. Say you have a poll that tells you Labour and the Tories are both on 40%. Within the margin of error, that could mean that Labour is on 44% and the Tories are on 36%, or it could mean that the Tories are on 44% and Labour on 36%. If you see a 40:40 poll, is your immediate assumption that either party could be well in the lead or that they are level-pegging? Not many people realise the margin of error—certainly not the hedge funds, which apparently pay huge sums for sophisticated polling that still gives them no more insight into the true state of the parties than anyone else.

The polls, therefore, are not very accurate. What if, via the commentators, the public believe what they say? Will that affect election results? Did Labour lose in 2015 because voters believed Britain was headed for what the Tories called a coalition of chaos? On what basis? On the basis of the polls. Did Jeremy Corbyn do so well in 2017 because the polls meant that no one thought that he had a chance in hell of winning?

We discuss the evidence in our report, and it is mixed. However, we do not recommend a ban on polling in the run-up to elections, such as is in place in 16 of the 28 EU countries. Nor do we recommend the statutory regulation of polls. We were not for a ban on polls largely because we thought that polls would be done anyway, probably overseas. Badly reported offshore polls would be even worse than well-reported onshore ones. We were not for state regulation because we felt it might inhibit innovation in polling; and because we did not think it would work. The example of the Commission des Sondages in France was not encouraging. However, if they get it wrong again in 2022, the question of banning or regulating will be revisited, and probably should be.

We made a more modest suggestion of a greater role for the Electoral Commission with regard to polling during elections. The Electoral Commission last week produced its own agenda for changes in its powers, and the Government are consulting on that—I hope in a more positive spirit than when they responded to this report. We did want increased regulation in the sense of increased self-regulation. When there is a choice, self-regulation is always better because it gets into the culture and changes how people behave, whereas regulation always seems imposed. We want the British Polling Council to take on new responsibilities, including holding a public inquiry into the performance of the polls after each election and providing an advisory service on poll questions.

We concluded that many of the problems with polls are down to media reporting of polls. There are some reporters and commentators who have a good grasp of what polling is about and its limitations. There are some who are less good. It is a perennial temptation in today’s competitive news environment to distort and exaggerate. “May soars”, when a poll shows a 1% increase in the Tory lead—well within the margin of error—makes a better story than “parties remain level pegging”. In my many years of reporting polling, I would not guarantee that I had made no such distortion. The media is also prone to report, as if they were polls, surveys carried out by pressure groups which are neither representative nor random. We look to IPSO to strengthen its efforts to crack down on those who seek to mislead the public about what the polls are saying.

Finally, and briefly, I will refer to digital. This was included in the committee’s remit, but the problems associated with it mushroomed during the committee’s lifetime, and we were forced to conclude that we could not do proper justice to it. We asked the Liaison Committee to set up a Select Committee specifically on the digital side of our work; sadly, it has not yet agreed. Time heals many things. We commended the Government’s digital charter and the work going on on it; and noted the work of the Commons Digital, Culture, Media and Sport Committee. I will just say this: if some of the fears that have been expressed about what is being done to our politics in the digital space are correct, the threat is surely graver that anything that arises from opinion polls misleading, as they may occasionally do.

I look forward to this debate. I trust that we have provided both a guide to those interested in where polling stands today—a reference volume of its strengths and weaknesses—and a road map of the direction in which it needs to go tomorrow if it is to retain any credibility within our democratic system. I commend our report to the House, and beg to move.

--- Later in debate ---

[Lord Lipsey]

I shall be very brief, or kick-off really will be threatened, certainly for participants in the next debate. I thank the Minister for his reply and in particular for the tone of that reply, which was in contrast to the official government response to the committee. I welcome in particular what he said about the British Polling Council, although I do not agree with all of it. Transparency on its own is not enough. That was the old BPC doctrine; the new doctrine goes further than that. For example, it staged an inquiry into the 2015 general election and, if self-regulation is to work, it must have an increased role. I take the Minister’s words to mean that the Government would in no way be opposed to that and, indeed, would welcome it, because that would secure the self-regulatory alternative that he and, on balance, I would prefer.

I thank all those who have participated in this debate and the kind words that were said about the committee and its work. It was comforting, as chairman, that most of the debate was about two issues—digital and social media and the Bloomberg affair—which were not covered by the committee. So I assume that we got the rest of it right.

Motion agreed.