Where the polls went wrong

The relationship between the averaged data points (shown on the scatter plots) gives an indication of the net flow of voters between the parties. The steeper the line, the more voters are switching. (For the mathematically inclined: where the fit wasn’t linear, I just used basic calculus to work out the gradient at each point and then took an average).

Why the polls were wrong

What it suggests is that for extra each point gained by the Conservatives, the Lib Dems lose an extra 0.49 points. For each extra point gained by Labour, the Lib Dems lose an another 0.43. Contrary to the received wisdom, the election results suggest that the Lib Dem collapse wasn’t a massive net benefit to Labour – the Conservatives shared the benefit.

Why the polls were wrong

The plots for the two main parties against UKIP show an equally striking picture. Every point worse the Tories do corresponds to UKIP doing 0.40 points better, while every point worse Labour does is matched by a further 0.45 point gain for UKIP. Again, far closer than the polls had suggested.

Why the polls were wrong

When I first discussed this train of thought in the summer, many of the people I spoke to simply couldn’t believe it. So ingrained was the idea that UKIP was splitting the right and ex-Lib Dem voters reuniting the left – most of whom, the thinking went, were left-leaning Iraq war rebels infuriated by tuition fees – that anything close to the picture painted by this analysis was really quite hard to fathom. It was almost like before the election when a Tory majority was “impossible”.

Why the polls were wrong

But to me it far seemed more plausible that the scenario was right. If it were wrong, how could the swing to Labour in England and Wales be just 1%? There was very little sign of an offsetting flow of voters directly from Labour to the Conservatives and turnout patterns didn’t seem to suggest a major shift, even though that appears to have been a contributory factor.

(Continued, please navigate using numbered tabs…)





About The Author

Related Posts