The IndyRef maybe over but a fierce flare-up is going on over the polling
The Mail on Sunday this morning
Mail on Sunday asks if the YouGov 2% YES poll was a rogue & cost UK £45bn
http://t.co/CQ393tPpzF pic.twitter.com/sYWq9Yqhk2
— Mike Smithson (@MSmithsonPB) September 21, 2014
YouGov’s Kellner responds
War of words between developing Mail/S & YouGov's Kellner http://t.co/La4AQIJlkv over that YES 2% poll 2 wks ago pic.twitter.com/EVGeHaFsBq
— Mike Smithson (@MSmithsonPB) September 21, 2014
Some polling lessons for an EU referendum?
Two things gave that 2% YES lead its greater potency: the advance hyping on Twitter by Rupert Murdoch and the fact that until a few weeks beforehand YouGov had been the firm showing NO with 19-20% leads. Those earlier polls came at a time when others firms like ICM, Panelbase and Survation had the battle much closer.
Thus on that weekend a fortnight ago Panelbase was reporting almost no change.
A driver of the movement at YouGov, I’d suggest, was the big methodology change in mid-August. Until then, surprisingly, YouGov had not included 16-17 years olds in its samples. Also it introduced a new weighting based on whether those polled had been born in Scotland or not. All the evidence suggested that they were big opponents of change and the YouGov measure meant that their views could be scaled back if there was a disproportionate number of them in a sample.
A feature of the 2% lead survey was that there was significant oversampling in this “born elsewhere in the UK” segment. This moved from 174 unweighted to 108 weighted which is quite some movement. Without that there probably would not have been a YES lead.
Overall in this controversial poll YouGov had an unweighted 475 YES and 538 for NO.
It is important after massive events like the IndyRef to look at the polling because so much can be learned and applied later.
Maybe we’ll be reviewing these old polls if ever there is an EU referendum?