Stole this from a forum - puts a bit of perspective into the recent polls :- Lets remind ourselves of some basics. An opinion poll is not a predictor of the referendum result; its a snapshot of voting intentions by a representative sample of the electorate. Those intentions can change, and a lot can go wrong in how representative the sample is. So you need to look at polls collectively, follow the trend from one poll to another by each pollster, and collect meta-polling data, or polls-of-polls. (And look at other evidence too.) Last year, the American data analyst Nate Silver, who correctly predicted the result of every state in the US Presidential Election, looked at the data on our referendum and concluded that NO will win, the only question being by how much. If you told Silver about the latest YouGov poll, his response would be one poll? That doesnt make any difference. Those of us who are confident of a decisive NO vote have drawn on the overwhelming body of evidence of that outcome, including the most one-sided collection of opinion polls ever taken. In more than two and a half years, more than 80 polls, by 8 or 9 pollsters, have consistently and continuously shown a NO win. Now, for the first time, we have a poll showing a yes lead. I dont think that changes anything, and heres why. One pollster, YouGov, has found an almighty swing to yes. In the space of one month, a swing of 22 percentage points; across three weekly polls, swings of 7%, then another 7%, then a further 8%, all in the same direction. This projects to 11% of those who vote, or 330,000 people (on a 75% turnout) changing their minds from being definite NO to being definite yes voters. In three weeks. What could have caused this shift, the biggest shift in voter opinion found in this long campaign, and a shift almost without parallel in British electoral history? Well, we know Alex Salmond is widely regarded as having won the last leaders debate, and the week that followed was a good one for yes. But no other opinion pollster has any comparable findings: there has been some movement to yes, but the latest poll by the pollster usually most favourable to yes, Panelbase, finds ... no change. And remember, the past week, when some of this polling was conducted, has been regarded as better for NO, who have clearly won two televised debates. Before this poll, the What Scotland Thinks poll-of-polls stood at a 10% NO lead, the BBC poll-of-polls showed a 10% NO lead, and the FT poll-of-polls showed a 9% NO lead. YouGov have moved from showing double that lead to removing that lead altogether, and a step further to a yes lead found nowhere else. Remember the latest polls by TNS-BMRB and Ipsos Mori still show NO leads of 13 and 14%. YouGovs findings are simply not credible. The question is why. 1. YouGov have changed their methodology, and the shifts theyre found in voting intention follow that. I await Peter Kellners explanation with interest; two months after criticising Survations methodology, he appears to have adopted it and gone even further. Ian Smarts latest blog discusses clustering among pollsters fearful of getting the result wrong, but YouGov is now the biggest outlier in the field. 2. What was the weighting on the latest poll? I havent seen the tables yet, but Im certain the unweighted data shows a clear NO lead. The infographic posted by Agent P shows a hugely disproportionate sample of SNP voters has been taken. 3. What is the motivation of the commissioning client? Andrew Neil is among those speculating that Rupert Murdoch has a score to settle with the British establishment. Are YouGov allowing themselves to be used in this way? We dont yet have all the answers, but the YouGov poll flies in the face of all the other evidence. Its wrong. We need to see other polls. What happened to ICM, who were also supposed to publish today? And there are more expected this week. Alistair Darling is right to say NO campaigners should treat this as a wake up call, but dont be disheartened - we are winning and we will win. As long as we all do our bit, well win comfortably. 60-40
Posted on: Sun, 07 Sep 2014 08:04:14 +0000
Trending Topics
Recently Viewed Topics
© 2015