“I’m so bored. I hate my life.” - Britney Spears

Das Langweilige ist interessant geworden, weil das Interessante angefangen hat langweilig zu werden. – Thomas Mann

"Never for money/always for love" - The Talking Heads

Thursday, August 20, 2015

Comres' fishy poll: you just ask for the results, and we will deliver them!

One of the key tools of contemporary politics is the gamed poll - the poll that shows results satisfactory to those who commissioned it. These polls wear their disnonesty in their footnotes. With that said, one should look at Comres's internet poll that shows Jeremy Corbyn as a crater for the next election, as compared to the ever electable, ever conservative David Millibrand.

It looks bad for Corbyn until you read how the results were filtered. Because of course, you don't want to just accept the voices of your complete set of respondents - you want to filter them just right. http://www.independent.co.uk/news/uk/politics/jeremy-corbyn-would-reduce-labours-chances-of-winning-the-next-election-poll-reveals-10457458.html

Here is the revealing footnote: "ComRes interviewed 2,035 GB adults online between 12th and 13th August 2015. Data were weighted to be demographically representative of all GB adults. Data were also weighted by past vote recall. Voting intention figures are calculated using the ComRes Voter Turnout Model. ComRes is a member of the British Polling Council and abides by its rules."

Now, what does past voter recall and the Comres Voter Turnout Model mean? It means that Comres decided that the trouble with polling that showed Labour leading in the last election relied too much on the responses of young voters. This, in combination with voters who haven't voted recently, means that basically, the poll was skewed towards just the demographic that would have voted for the most conservative labour candidates. What Comres doesn't say is how they tested their conclusion. If we transpose this model to the American election of 2012, for instance, excluding black and young voters, Romney would clearly have been first in the polls - as indeed he was in the Fox News poll and in the Gallup poll. Intellectual honesty would demand, I think, that Comres publish the results without applying their "comres" model alongside the results of applying their model. But don't hold your breath for that to happen. After all, this poll is commissioned to get the Comres model results.
Which will then be twittered about by the usual Blairite suspects.

No comments: