From The Guardian: #IndyRef Three reasons polls might be right – and one why they could still be wrong Tom Clark is gearing up to provide analysis of the results as they pile in through the night, and he’s flexing his psephology muscles early with some thoughts on the polls we’ve seen so far: All the final polls have converged neatly around 48% Yes, 52% No. Even if the surveys were perfectly designed, you’d expect more random variation. The suspicious mind wonders whether nervous pollsters have sprinkled a little magic dust, judging it’s better to hang together than hang apart if things go wrong. The world will soon forget if they underestimate the No lead, but not if Yes comes out on top. There are three decent reasons to think No looks a reasonably safe bet: 1) Safety in numbers: Scores of surveys have now been done using very different methods – online, telephone and face-to-face, with all sorts of different methods – and yet only three at any stage have shown Yes ahead. While any one poll might have a 3-percentage point margin of error, the chances of so many being out in a particular direction are dramatically reduced. 2) The swing back to the status quo: governments often pick up at the 11th hour as voters cling to ‘nurse for fear of something worse’. If voting for independence is the risky option, you’d expect a swing back to No in the privacy of the ballot box. 3) The great theory on the Yes side, that super-high turnout will confound the pollsters, looks optimistic: if interest is high, you’d expect high turnout on both sides, even if one side is more engaged than the other. If it averaged (say) 85% overall, then – if Yes are more motivated – they might get (say) 88% out, compared with (say) 82% among Nos. That sort of differential is only going to swing the overall outcome if the race is extremely tight, within the four point gap that the polls predict.
Posted on: Thu, 18 Sep 2014 20:39:03 +0000
Trending Topics
Recently Viewed Topics
© 2015