Opinion Polling

Lord McColl of Dulwich Excerpts

Thursday 18th June 2015 Grand Committee

[Lord McColl of Dulwich (Con)]

My Lords, I think it was generally accepted that the 2015 election was not the pollsters’ finest hour, although I have to admit that I, too, got it wrong: I predicted to my friends a Conservative majority of 20, so I was out by eight seats. The vast majority of companies in the industry predicted the vote shares of most of the parties correctly within the margin of error, and several were within the margin of error on all vote shares.

The British Polling Council and the Market Research Society were exceptionally fast in, first, recognising that things had not gone as well as they should have, and, secondly, setting up a comprehensive review under the auspices of Professor Patrick Sturgis at Southampton University. That review has yet to hear evidence—its first session is actually tomorrow—so we should not prejudge its outcome.

That said, the vast majority of polls during the campaign pointed to a much closer result than we eventually saw and it is right to consider the implications of that variance. Establishing a regulatory authority to regularise question wording and sample design, far from increasing the accuracy of polling, has the potential to compound the problem further. All scientific inquiry relies on a cycle of experiment, evaluation and modulation. Polling firms ought to be encouraged to experiment with novel methods and trial their own question wording and sample design, rather than being prevented from doing so. It is also in their commercial interests to try to out-compete their rivals and develop more accurate methods, and we ought to cheer them on as they do so.

With regard to a ban on publication of opinion polls before elections, we should remember that it is voters, not polls, which change election outcomes. A ban would not prevent such polls from being conducted; rather, it would lead to a small number of wealthy individuals and organisations being the only ones with privileged information, leaving the wider public at a disadvantage. Even if we wished to deny voters access to this type of information, in the 21st century it would be impossible to do so. The internet makes a mockery of enforced censorship. We could try to emulate the practice of North Korea and prevent our own citizens having information which is so freely available throughout the rest of the world, but it simply would not work. Political bloggers such as Guido Fawkes use servers in countries such as Ireland and the United States which the Government could not block even if they wished to. British voters would therefore be left open to the unchecked rumours and misinformation that would result from such a ban.

The idea of a voluntary agreement not to publish is interesting, but I suspect that it would ultimately be doomed to fail. The question of who enters into the agreement is fundamental. Even if all the established members of the British Polling Council agreed not to publish polling in the final week before an election, it would not prevent new entrants and overseas pollsters doing so.

The polling industry may be licking its wounds but I do not doubt its integrity in seeking honest answers from the post-election inquiry. We must be able to hear the results of that and understand what went wrong before leaping for the statute book. We are privileged in this country to have a polling profession which, through the good offices of self-regulatory bodies such the British Polling Council, is clearly committed to transparency. That is quite a rarity. In the United States, for example, voters have access to a far poorer quality and quantity of information about individual polls. The industry here ought to be commended for its transparency. Our role should be to recognise and encourage that and to find ways of enabling more and greater competition to encourage greater accuracy.

I am concerned about what has been said about the ComRes polling for the charity CARE in relation to the proposed three-parent, pronuclear and maternal spindle transfer techniques. In addressing this subject I should declare an interest: I have worked closely with CARE over the years. It has provided me with invaluable assistance, especially in relation to my human trafficking Bill. Criticism of this polling seems to hang very much on a report commissioned by the Wellcome Trust—one of the principal advocates for changing the law to permit the controversial procedures—to critique some of the polling in question. That report is deeply flawed in four respects.

First, and most importantly, the report criticises the polling against benchmarks that would be appropriate only if a deliberative public consultation was being discussed, in which lots of background information could be provided. Both more deliberative public consultation exercises and polling have their place and their respective strengths and weaknesses, but to critique one against the standards of the other is to make a very basic blunder. While deliberative public consultations can contain lots of information and nuance, and can in turn receive lots of information and nuance, polls cannot do so. The questions have to be relatively short, and as long as they can provide a range of responses other than simple yes or no answers, the responses are predetermined.

The CARE poll questions were designed very obviously to measure public opinion on pronuclear and maternal spindle transfer as these were defined in common parlance by the media, which spoke almost universally about “three-parent embryos” and “three-parent children”. Far from it being inappropriate to design questions reflecting that terminology, it would have been illogical and poor research practice not to do so. Moreover, as a matter of scientific fact, the designations “three-parent children” or “three-parent embryos”, while disliked by some, were not inaccurate, in that the resulting children will have DNA from three rather than two parents.

The aim of the questions has also been misrepresented. Their aim was first to track basic attitudes in response to the information gleaned by the public through the media, and, secondly, to test responses to a range of arguments both in favour of and against permitting the procedure. This is basic standard practice in any campaign. The purpose of testing arguments in favour of the procedure is to understand which of one’s opponent’s arguments are the strongest and which are the weakest. The purpose of testing arguments against the procedure is to recognise which arguments are the strongest for CARE itself to deploy. Publishing all the results was not an attempt to distort public discourse but rather to ensure that CARE and ComRes were compliant with the British Polling Council requirement for full transparency.

Secondly, the report objected to the polling on the basis that it had been commissioned by an organisation opposed to the introduction of PNT and MST. I found that rather odd. Most polls are commissioned by organisations that want to test public opinion to see whether it fits with their own position. To suggest that this is somehow inappropriate is to misunderstand polling completely. Indeed, the point has been made that the report commissioned by the Wellcome Trust has been commissioned by a body with a very clear agenda. The irony of its position, however, does not seem to have occurred to it. None of us approaches these things from a position of value neutrality. For the record, as a doctor, I am very much opposed to the use of pronuclear transfer but am more open to the use of maternal spindle transfer. Thirdly, the report seeks to criticise the polling on the basis of an assessment of the press coverage secured, which is entirely irrelevant to the efficacy of the polling in question. Fourthly, the report lays great stress on stakeholder interviews but is entirely lacking in transparency. Parliament has looked at these matters and decided what it has to do.

In conclusion, I state that ComRes did not threaten to sue the noble Lord, Lord Lipsey, but the noble Lord, Lord Lipsey, pointed out to it that he intended to have a debate in your Lordships’ House, which would, of course, be privileged.

[Lord Lipsey]

May I just correct the noble Lord? He said that ComRes did not threaten to sue. It said that it was taking legal advice and would take what legal action was recommended to it. If that is not a threat to sue, I don’t know what is.

[Lord McColl of Dulwich]

It said it would defend its position, which is not quite the same.