|Populus without DK adjustment||Unadjusted||Reported|
|ICM without DK adjustment||Unadjusted||Reported|
Do the formulas work in a 3-way fight?
I’ve raised before here about the methodology that three of the phone pollsters use to deal with those in their polls who say they will be voting but won’t say who for.
ICM and Populus two allocate a proportion of the “don’t knows” in accordance with what they said they did last time. I wonder about the validity of this approach in a tripartite contest.
Just look at the tables above at the massive differences that this makes with the final polls from Populus and ICM.
Thus with Populus a 3 point Lib Dem lead over Labour became a one point deficit.
The pre-DK adjusted Populus Labour share is not that far off the 24% that Angus Reid was reporting last night.
If I’m right then both the Tories and Lib Dems look set to do better than the headline numbers suggest. We’ll see tomorrow when we have got the full result whether my concern is justified.
If I’m wrong I’ll be ready to swallow humble pie!