YouGov President, Peter Kellner, on Woolwich and polling on sensitive issues
From time to time we receive complaints about some of our questions. These complaints usually come from people who disagree with the majority opinion. They would rather we asked the question in a different way (‘you should ask whether voters side with the honest people who wear lilac trousers or the crooks who wear yellow’), or inserted a key fact into the introduction (‘surely you would get different answers if people knew that yellow trousers cause cancer’).
Want to receive Peter Kellner's commentaries by email? Subscribe here
Normally such comments cause us little bother: apart from the odd occasions on which we, like any human institution, slip up, we are able to defend our practice of asking questions in a neutral manner, and seeking to measure opinion as it is, rather than as some people would like it to be if voters were ‘better informed’. (Occasionally we do add information to see what impact it has, but by publishing full results from all relevant questions, we make this clear.)
In the past few days, we have received more complaints than usual. They have been sparked by two surveys that explored public attitudes following the murder last week of Drummer Lee Rigby in Woolwich. One survey was for the Sunday Times, the other for Dr Matthew Goodwin of Nottingham University.
The complainants have made three broad points about this appalling event:
a) To ask questions about attitudes to Islam and terrorism at all just now is inflammatory and irresponsible (example: ‘I am shocked by your questions which fan the flames’)
b) We asked loaded questions (‘substitute “Jew” for “Muslim” and see how acceptable you find them’)
c) In our questions about the English Defence League, we failed to make clear the character of the organisation. (‘where are value Qs explicitly condemning EDL as racist and provocateurs of social tension?’)
Let us take these criticisms in turn.
a) Forty years ago, as a young journalist, I learned that ‘there are no indiscreet questions, only indiscreet answers’. In general, the same applies to polling. We do not create public opinion; we measure it. That opinion may give us pleasure or pain; it might reassure us or frighten us. The issue is whether it is better to know how people feel or to remain in ignorance. My strong belief, as a recovering-journalist-turned-pollster, is that, with extremely rare exceptions, knowledge is not only preferable to ignorance in principle; it also makes for a healthier society in practice.
In theory there may be times when, exceptionally, a poll finding might inflame a lynch-mob mentality and make matters worse. In practice, I cannot think of any event in Britain’s recent history when that danger existed at all – and certainly not to an extent to justify self-censorship.
Does the Woolwich murder provide one of the rare exceptions? Some complainants disliked the following question, which we asked for the Sunday Times:
| Which of the following statements about British Muslims comes closest to your own view?Practically all British Muslims are peaceful, law-abiding citizens who deplore Wednesday's killing as much as everyone elseThe great majority are peaceful and law-abiding citizens but there is a dangerous minority who feel no sense of loyalty to this country and are prepared to carry out, or at least to condone, acts of terrorismA large proportion of British Muslims feel no sense of loyalty to this country and are prepared to condone or even carry out acts of terrorism |
|---|
Complainants disliked the final option. Example: ‘Despite universal condemnation from every major Muslim organisation, the poll wrongfully gives you the option to state otherwise.’
The logic there is interesting. We are asked to believe that the unanimity of ‘every major Muslim organisation’ puts the matter beyond debate. The communities’ leaders have spoken; therefore it is impossible to suppose that British people can think ordinary Muslims hold a different view; therefore Islamophobia cannot exist; therefore we were wrong to offer an option that suggests it might.
The issue is not whether the opinion is offensive or incorrect (and I personally think it is both), but whether its prevalence should be measured. We found that 14% hold this view about Britain’s Muslims, up from 10% when we first asked it, in the aftermath of the London bombings in July 2005. We can debate whether this number is reassuringly low or worryingly high; but better to know than not to do so. To omit or soften the ‘no sense of loyalty’ option would be to restrict respondents to approved answers. Now that WOULD be wrong.
Of course, were we to have good reason to suppose that lives would be in jeopardy if we asked a specific question, we wouldn’t ask it. Nobody in their right mind would argue that the right to conduct controversial opinion polls trumps the right to life. But if opinion poll findings really did increase the risk of violence on our streets, then regulating YouGov would, I fancy, be the least of the Government’s problems. For the time being, I reckon that the fanciful risk of poll-induced terror does not begin to justify the evil of self-censorship.
b) The bulk of our questions for Dr Goodwin asked people whether they agreed or disagreed with a number of statements. Some of these were certainly one-sided – such as ‘British Muslims pose a serious threat to democracy’; and ‘There will be a clash of civilizations' between British Muslims and native white Britons’.
However, for every ‘anti-Muslim’ statement, we posed a ‘pro-Muslim’ statement – such as: ‘Muslims make an important contribution to British society’ and ‘The vast majority of Muslims are good British citizens’.
In constructing their questionnaire we followed our normal policy of testing statements on different sides of an argument. Had we posed statements from only one side, we would certainly have been open to the charge of bias. That is precisely why we ensured that a wide range of viewpoints was reflected in our agree/disagree statements.
As for the charge that we are reluctant to ask about Jews – well, were anti-semitism a significant issue, we certainly would. Back in 2009, when the British National Party won seats in the European Parliament, YouGov conducted a large-scale poll, part of whose purpose was to examine the views of BNP voters. Then we did test the extent of Holocaust denial among British voters, and how many people thought there was ‘an international conspiracy led by Jews and Communists’ against western Christian values’. See the survey results.
We found that 20% of BNP voters thought the Holocaust either did not happen at all (2%) or that the figure of six million deaths was exaggerated (18%); and 9% of BNP voters regarded allegations of a Jewish / Communist conspiracy ‘completely true’. I reckon these figures were worth knowing and actually rather reassuring. Have we become so touchy these days that we need to declare that to test such views is not to endorse them?
c) True, we did not set out the views of the English Defence League. Plainly, the responses we obtained were based on the level of knowledge people have. If people knew more, their views might well change. That is a task for those engaged in the debate, not for pollsters. Were we on some future occasion to add some information prior to asking questions, this might tell us the direction in which opinion might change were the public better informed. But that is not the same thing as measuring opinion as it is today, on the basis of the information voters have, or lack, today.
I doubt whether what I have written here will satisfy everybody. These kinds of controversies are bound to erupt from time to time, and views will vary. But YouGov will continue to explore what people think, and that includes testing views that offend some people including, sometimes, me.
Want to receive Peter Kellner's commentaries by email? Subscribe here
See the full YouGov / Sunday Times survey results
See the full survey results - YouGov / Dr Matthew Goodwin, University of Nottingham
See the full survey results from YouGov's 2009 large-scale poll
