“It would be interesting to know what has happened to the Labour Muslim vote”, wrote Neal Lawson at the Guardian’s Comment is Free yesterday. Ask and ye shall recieve Neal – the rather surprising answer is that appears to be alive and well.
Accurate polls of British Muslims are actually rather hard to do. For phone polls there is no nice list of only Muslim phone numbers to randomly select from, and the census data from 2001 is the only available data for coming up with weighting targets for the demographic make up of the Muslim population – hence many polls of Muslims are unweighted. Added to this is that, presumably for cultural reasons, polls of British Muslims encounter a very high refusal rate from women.
ICM have tried to get round the problem of sampling Muslims by ringing back Muslims they’d identified in their normal surveys, and increasing the sample size by asking people interviewed if they have contact details of any other Muslims they think would agree to be interviewed. This isn’t perfect of course, since in theory people may well be more likely to give details of people with views similar to their own.
The recent NOP poll for Channel 4 on the other hand, used normal random telephone dialing, but only in places where more than 5% of the population were Muslim (presumably even in areas with more than 5% Muslims it must have taken a huge number of unsuccessful phone calls to houses that turned out not to contain anyone Muslim in order to get a sample of 1,000 people). This is obviously a trade-off between the absurdly prohibitive cost of doing random dialing across the whole country and discarding 97% of your calls, and excluding Muslims who live in areas where less than 5% of the population is Muslim. NOP also did 7% of their interviews in languages other than English, while I believe that ICM only interviewed English speaking British Muslims.
What this all means, is that polls of Muslim voting intention aren’t 100% reliable, and aren’t easily compared to each other. Add to that the fact that ICM’s polls of British Muslims back in 2004 were not weighted because of a lack of authoritative demographics, while their more recent poll in February 2006 used the 2001 census data for weighting purposes (as did NOP). Still, with those caveats out of the way…what has happened to the Labour Muslim vote?
Now, you need to remember all the caveats above – NOP used a very different sampling technique from ICM, and it might have been more favourable to Labour; ICM’s newest poll is weighted, when the older ones were not – that said, on first appearance it certainly seems as though Muslims voters have drifted back towards Labour since the last election.