Polling Problems
"Number Cruncher Politics" has a good piece at
http://www.ncpolitics.uk/2015/06/as-the-investigation-begins-a-summary-of-pollsters-views-on-the-failure.html/
which gives early reports on the excuses, sorry explanations, coming out of various polling companies' inquests on why they got the 2015 election so utterly wrong.
A short summary
* Survation and BES think there was a very late swing to the Conservatives
* YouGov blames "shy tories"
* Ipsos Mori and ComRes think their figures were thrown out by differential turnout (82% of people polled by the former company said they were "certain to vote" which is 16% higher than the share of the electorate who actually did.)
* ICM think the mix of voters in their sample had an imbalance of certain types of voter (C1 and C2s) and that they weighted them wrongly
I suspect all these factors may have had an influence.
I suspect everyone will also remember next time that polls are only a snapshot, not a prediction and that they can be wrong.
http://www.ncpolitics.uk/2015/06/as-the-investigation-begins-a-summary-of-pollsters-views-on-the-failure.html/
which gives early reports on the excuses, sorry explanations, coming out of various polling companies' inquests on why they got the 2015 election so utterly wrong.
A short summary
* Survation and BES think there was a very late swing to the Conservatives
* YouGov blames "shy tories"
* Ipsos Mori and ComRes think their figures were thrown out by differential turnout (82% of people polled by the former company said they were "certain to vote" which is 16% higher than the share of the electorate who actually did.)
* ICM think the mix of voters in their sample had an imbalance of certain types of voter (C1 and C2s) and that they weighted them wrongly
I suspect all these factors may have had an influence.
I suspect everyone will also remember next time that polls are only a snapshot, not a prediction and that they can be wrong.
Comments