Surely I can’t be the only one perturbed by Trevor Phillips’ analysis of the now famous ICM poll which claimed to find out what British Muslims really think.
Four percent of the 0.04% of British Muslims polled said they had sympathy for terrorist acts. From this the broadcaster extrapolated that 100,000 British Muslims have sympathy for terrorist acts. But all we can really say for sure is that 0.00016% of them do.
How does this compare to the UK population in general? Well we don’t really know, because while the programme insisted on extrapolating percentages for Muslim participants, the pollsters forgot to apply that logic to the control group.
For rather than interviewing 0.04% of the UK general population — around 25,640 people — to make comparative judgements, they selected 0.001% instead. Yes, we had two groups of 1000 people: one representing 2.7 million people, the other representing 62 million.
Would it be fair to extrapolate findings from the control group and claim that 600,000 people in the wider UK population also had sympathy for terrorist acts? Maybe they do. And maybe that many Muslims do too. Whatever sympathy means. But is it fair to make such a claim, based upon how a tiny percentage of a tiny percentage of the population responded to a question?
Maybe it is. The British Crime Survey targets 35,000 households — 0.05% of the UK population — although they also analyse police recorded crime and a range of supplementary sources to try to build a more complete picture of crime. Something seriously lacking in the anecdotal pastiches stitched into the narration of last night’s television programme.
If we are going to make use of statistics to make bold claims, shouldn’t we at least be consistent? What is more, should we not review all of the figures and not just focus on those which serve a predefined political agenda? How exactly did the other 96% of those surveyed respond to that question? And why was that not significant?