How not to win an argument
The Joseph Rowntree Reform Trust decided that they wanted an opinion poll which 'proved' that Teh People have turned against the idea of locking up terrorist suspects for six weeks.
So just to make sure that the people would give the correct answers, when they commissioned ICM to conduct polling, they got the interviewer to read out a paragraph summarising the arguments against this policy, then two other questions designed to associate the policy with locking up innocent people, and then to present five different options, with six weeks being the longest of the options on offer.
Mission accomplished, in one sense, in that only 38% picked the option of six weeks. Cue the press release saying that 60% support detaining terrorist suspects for no more than 28 days.
There's nothing wrong with using opinion surveys to test out different messages and arguments to see how they influence the decisions that people make (indeed, this is usually a much better use of opinion polling then trying to find out what public opinion is at any given moment). For example, the survey included a question which asked people who supported the government's policy whether they would change their minds in the knowledge that six weeks is as long as the prison sentence which someone convicted of burglary or assault serves. This fact on its own persuaded 35% of people to change their minds.
But what the Joseph Rowntree Reform Trust have tried to do is to claim that the public are already being persuaded by the arguments of opponents of six weeks detention without trial, rather than that it could be won given the right arguments. I agree with them on the issue, but finding out that as long as people just hear only our side of the argument, then a majority will agree with us isn't actually very helpful. When in a minority, it is better to find out how to win people over, rather than declare that you are the majority.
What would have been much more valuable would have been if they had started with a neutral question about the issue, and then tested out different arguments to see which the most persuasive were. It wouldn't have got as good a headline, but it would have been a much better contribution to the debate and a much better resource for campaigners for civil liberties.