h1

Maybe the whole idea of national voting intention polls has had its day

December 29th, 2014

The one thing you can say as we enter election year is that there’s no consistent picture

I’ve broken the final polls of 2014 into five broad areas for comparison.

CON to LAB swing The most important measure and where there’s a huge variation from the Ipsos-MORI 2.65% swing to the TNS-BMRB and Opinium 7.15%. If it was the latter on May 7th then LAB would probably secure an overall majority even if it lost all but a handful of its 41 Scottish seats.

The CON+LAB aggregate This shows up the very big difference between the phone pollsters, who are all on 61%, and the online ones which are much higher. I put this down to the way samples are assembled.

The LD share As can be seen this ranges from 5% to 14% which is down to different methodologies. If ranges on this scale continue some pollsters will have egg on their faces.

The UKIP share Not quite as much variation as the LDs but still big differences measuring a segment of the electorate that could be decisive.

The GRN share This moves from 2% with ComRes online to 9% with Ipsos-MORI. I can’t explain it.

It used to be that you’d take the national shares and input them into one of the online seat calculators to get an ides of what the numbers would mean in terms of party MPs totals. That form of calculation isn’t really possible any more because of Scotland and the UKIP surge.

I’m beginning to wonder – what’s the point.

Anyway less than four and a half months to wait.

Mike Smithson

2004-2014: The view from OUTSIDE the Westminster bubble