If you've read a fair number of my blog posts, you'll be aware that I am always a little suspicious of statistics used in the news, and I think it's worth regularly digging a little deeper to see what lies beneath those numbers, and whether they really tell us what the media say they do.
A little over a week ago, there was a fair amount of coverage to say that for the first time, possibly influenced by the Paris attacks a few days before, there was a majority of voters in the UK in favour of leaving the EU. But was that really true?
After a bit of digging (none of the newspaper reports I looked at had a link to the data) I found the results from the ORB International survey. (It's on page 8, as this is a summary of a range of surveys.) The survey was of 2067 people and, indeed, 52% were in favour of leaving over 48% against. But there are two issues here. Can we be sure that such a tight margin is representative when we blow that 2067 up to the 45 million or so voting population, and what about the 'don't know's?
The survey doesn't provide error values (though some of the other surveys in the report do), but if they did, I would be very surprised if they didn't allow for a couple of percentage points in inaccuracy - so there could still be a majority the other way. The survey does tell that around half of the population would be likely to say that they would vote against, but it isn't accurate enough to tell us that we have crossed a line from 'most for staying in' to 'most for leaving'.
However, the more interesting aspect is the 'don't know's. The survey didn't give the option of 'don't know' or 'undecided.' I guess the argument was that this won't be an option on the ballot paper in 2017 (or whenever the real decision is made). But the trouble is, probably over a year out from the vote, when we haven't yet heard the full arguments from both sides, there will be a lot of floating voters. Previous surveys suggest that numbers are in the 15-25% range. So, in effect, there could be 500 or more of those votes cast in this poll that weren't correct. That's a huge error, and it really says that we just haven't got a clue.
Is the survey useless? No, though I wish they would include 'undecided'. We do get a feel from the series of polls that there has been a drift towards a 'get out of the EU' decision. But what certainly is the case is that this poll does not show, as the headlines claimed, that the majority of voters are now in favour of leaving the EU.