Saturday 9 May 2015

How did the polls get it so wrong?

How did the polls get it so wrong? The question becomes all the more telling when we remember that all the parties fine-tuned their campaigns based upon those polling figures. The simple fact that they got it so badly wrong could mean that policies were altered based upon erroneous information. That means we either stop even taking an interest in polls and sail by the wind, or the pollsters are going to have to go back to the drawing board just as they did after 1992. I am not qualified to go into the finer points of modern political polling, but I will make one point about it which may account, at least in part, for the errors.

Modern polling is either online or via the telephone, with very few organisations being able to commission polls that are conducted by sending an army of pollsters out to knock on doors to collect results. I know nothing about telephone polls, but I am a member of the panels for both YouGov and Panalbase.

I joined YouGov about three years ago thanks to a campaign that they ran to increase the membership of their panel. I quite liked the idea of having my views taken into account, and I liked even more the promise of a few bob now and again for giving my opinions to the world and his wife. I will never get rich out of this since from YouGov I get rewarded with about fifty quid every 18 months or so, and from Panelbase roughly a tenner comes my way every other month. Still, it is good fun to be asked about things and be paid for giving my answers.

The question is how representative are the people who join these panels to the population at large? The panels are massive, and growing all the time, but they are still made up of people who elect to join them. In the case of YouGov especially, that may mean people who have an interest in politics to begin with rather than those who regard it as background noise.

Panelbase is more concerned with consumer opinions so it may be more representative of the population, but they still got it wrong as well, so clearly something is wrong with British polling other than just the self-selecting panels.

It should be noted that even the much talked about exit poll was actually out by quite a bit. That poll led us all to believe that it was still a hung parliament, albeit with the Tories on 316 seats and Labour getting 239. In actual fact the Tories got an overall majority of 331 seats and Labour trailed badly behind with 232.

It is as if people who vote Tory either do not join opinion panels, lie to the pollsters even in exit polls, or a combination of both.

How the polling companies manage to tweak their systems to allow for people lying about their intention to vote Tory is a matter for them, but they had better come up with something fast.

No comments:

Post a Comment

Views Themes -->