A new YouGov poll has just been published which puts the Yes campaign marginally ahead for the first time with a 51/49 split. Instead of talking about the possible effects of this poll I want to chat about its mechanics, as I was a part of the YouGov panel which gave that result.
The problem with modern polls is that the companies cannot afford an army of people to go round knocking on doors trying to get a representative sample of the population together. Instead they rely in a panel of volunteers who are paid about 50p for every survey that they complete. YouGov does not ask its full panel to complete every survey as that would rather defeat the object of trying to get a panel together that reflects the population being surveyed by demographic, gender, age and class. For that reason the YouGov IndyRef survey was restricted to about a thousand people who live in Scotland in the first place, and then further restricted by the other criteria. I was obviously selected as representing the late middle aged, disabled, Englishmen who live in Scotland - or something.
Normally I can complete these surveys in a few minutes, but this one was ferociously complicated. That's not to say that it was intellectually challenging, just time consuming as it asked questions that obviously aimed at discovering how likely we were to vote. For instance we were asked not just how we planned to vote, but how we had voted in the recent European elections and the previous 2010 general election.
All in all it took about twenty minutes to complete, and I filled the survey in with total honesty. If everyone else did the same then we should have a pretty good snapshot of opinion in Scotland taken during the first week of September.