The Politics of Polling (Liaison Committee Report)

Lord Rennard Excerpts

Thursday 19th May 2022 Lords Chamber

[Lord Rennard (LD)]

My Lords, I too pay tribute to the noble Lord, Lord Lipsey, for his tenacity in trying to insist on statistical accuracy.

It was a pleasure to serve as one of two Liberal Democrat Members on the Select Committee on Political Polling and Digital Media, which met in the 2017-19 Session, and also to provide evidence to the Liaison Committee, which reviewed its work in 2020. One of the main issues we addressed was whether, as in some countries, there should be an attempt to ban the publication of opinion polls for some days before polling day. Polls are, of course, sometimes responsible for guiding public opinion, rather than merely reflecting it. However, I think the Select Committee helped to dismiss any suggestion that polling companies generally set out to fix the outcomes of their polls in order to influence elections and referendums.

We discussed how a final weekend poll in the Scottish independence referendum in 2014 caused considerable controversy. I remember reading the Sunday Times with some shock when it reported a poll with “yes” to independence leading by 51% to 49%. It was later suggested that this must have been inaccurate because independence was rejected in the votes cast a few days later by 55% to 45%. However, I do not think it was fair to attack the Sunday Times for publishing a poll which did not appear to be an accurate forecast. My own sources within the Better Together campaign suggested that “yes” may well have moved into a narrow lead at that point and that far from leading to a further surge in favour of independence, the poll served as a wake-up call to those opposed to it and led to some dramatic interventions, most notably by Gordon Brown, and plans to give people in Scotland what most of them wanted—which was more power to their Parliament while remaining part of the UK.

Of course, polls should be seen as only snapshots of opinion, and it has always been irresponsible to simply suggest that they are forecasts of what will happen on polling day. In many of the famous by-election victories in which I was involved, my party was often second in the polls, but I knew that we would win, because momentum and tactical voting would see us in the lead on polling day—the day that really mattered. I notice the noble Lord, Lord Hayward, smiling, recalling no doubt the 1993 Christchurch by-election, where I think his party never actually had a lead—if memory does not serve me correctly, no doubt he will correct me.

Knowing what other people think is an important factor in deciding how to vote, and an entirely legitimate one. But the danger is that polls can be inaccurate, and the reporting of them can be highly misleading. I think our Select Committee was on the whole supportive of the polling industry’s significant attempts to improve the accuracy of polls and the way in which it is helping to discourage misleading reporting of them. That is why there was little support for banning the publication of polls immediately prior to polling day. We felt that banning publication of properly conducted opinion polls might give even greater power to unscrupulous people with too much control over much of the media.

The potential power for the polling companies to inform the electorate was shown when they conducted snapshot polls at the end of the first leaders’ televised debates in the 2010 general election. Some 20 minutes before the conclusion of the first debate, I sent the late and much-missed noble Lord, Lord Ashdown, into the media room to claim a knockout victory for Nick Clegg. My view of what the public perception would be was quickly confirmed at the conclusion of the debate by the publication of instant polls. Typical of the polls was Populus for the Times. It found that Nick Clegg was declared by the public to be the overwhelming winner, with 61% saying that he had won the debate, compared with Cameron and Brown, who were trailing on 22% and 17% respectively. The broadcast news programmes accurately led with reports of these polls, but some of the tabloids struggled with the news. In the media room at the debating centre in Manchester, I watched George Osborne and the Sun writing team looking agonisingly at draft front pages on their computers as they struggled to try to suggest in any way that the debate had been a win for David Cameron.

There was general agreement in our committee’s deliberations that, if there are problems with the use of opinion polls in elections and referendums, the problems are generally with how they are reported rather than with the polling methodologies themselves. I will cite four examples. First, the noble Lord, Lord Lipsey, has already referred to the infamous Chicago Daily Tribune headline, “Dewey Defeats Truman”, in 1948. The headline was based on polls, rather than many of the votes actually being counted.

Secondly, and more recently, the polls appear to have been inaccurate in our 1992 general election when John Major won what was called a “Lazarus-type” victory after trailing in the polls. My own view is that this was explained in part by a late swing against the prospect of a Kinnock-led coalition.

Thirdly, the opinion polls during the 2017 general election led canvassers such as me to personally assure voters I was speaking to that there was no danger of Jeremy Corbyn getting in. He did rather better than the polls had indicated, but perhaps that was because some people felt that they could vote for him as a protest—perhaps against the incompetence of the Conservative campaign at the time—without any prospect of him becoming Prime Minister.

Finally, the polls suggesting victory for remain over leave in the 2016 EU referendum by a margin of 52 to 48 were contradicted by actual votes being cast in reverse proportions. But who knows what the effect was of 150,000 Russian bots mobilised in the last few days of the campaign to spread Putin’s propaganda in favour of Brexit? The Minister shakes his head at that, but his Government refused specifically to investigate that, although we now know that 150,000 Russian bots were mobilised on that campaign. Some people certainly voted leave as a protest, believing from the polls that leave would not win. Perhaps some of its principal proponents campaigned for leave only because they trusted in polls showing that we would remain but that they could become Prime Minister by supporting leave.

As a committee, we certainly sympathised with the polling industry struggling to refine techniques on which their professional reputations rested. I highlighted to the Liaison Committee one of the problems for the polling companies: the disconnect between those over 18 legally entitled to vote and those who make it on to the electoral register. When people are asked whether they are on the electoral register, they often do not know, and the Government do not make it easy for them to check.

The Electoral Commission has estimated that around 9 million people who are entitled to vote are either missing from the registers or inaccurately included on them—so the pollsters’ task is not an easy one. Despite this, the evidence is that polls were rather more accurate in the 2019 general election than they had been in the two previous ones. Efforts have certainly been made by the British Polling Council, the Market Research Society and others to encourage better and fairer reporting of opinion polls. This reporting is important because it does have real influence on the outcome of elections and referendums. The best, fairest and most accurate reporting of polls comes often, although not exclusively, from public service broadcasters, and some of the worst comes from what some of us still call the tabloids.

I will conclude by saying that the constant threats to the independence of the BBC by reducing its income and promising to end the licence fee, together with the planned privatisation of Channel 4, are significant threats to the health of our democracy