There were a LOT of polls, including my own, that failed to notice. 2/?
— Martin Boon (@martinboon) April 26, 2019
How much can we rely on polling for the Euros?
The above was the start of a Twitter thread over the weekend from Martin Boon who has been a major figure in he British polling industry for several decades. He was with ICM and is now part of DeltaPoll.
His Tweets raised questions on the mechanics of polling which given how the surveys were five years ago is relevant today.
The first thing online pollsters seek to do is to establish the broad demographic profile of those responding to an invite email to assess whether they match their requirements for an overall balanced sample. Some segments will fill up faster than others. His observation about “bed-blocking” suggests that certain groups that responded to the invitation email almost instantly could distort the overall pattern. Thus once the, say, males aged 45-60 cohort is full then no more respondees in that group would take part.
A problem arises for a pollster can arise if the instant responders are not representative of the cohort as a whole. Clearly firms send out far more survey invites than the total required for the poll and some pollsters now have mechanism to try to deal with this effect.
The Tweets above were followed by these from Boon:-
UKIP were overstated in many polls, which may or may not have been down to a phenom like this.
The Brexit Party are now going ballistic in latest polling. Coincidence? I’ve noticed that in recent polls past vote recall of UKIP is materially higher than their actual vote share.
UKIP 2017 quota cells are, you guessed it, filling up in quick-step time, alongside Brexit Party ceiling breaking.
No doubt we are going through a period of unprecedented political change and the conditions are bang on for the Brexit Party.BUT, just remember that overstating of a anti-European right of centre party in poll after poll is not unheard of, nor misunderstood as to why it might be the case.
In 2014 Farage’s UKIP touched 38% in the Euros polling more more twenty points ahead of the Tories. In fact it ended up with UKIP on 27% just above LAB and 3.6% above the Tories.
There’s little doubt that The Brexit Party is going to do very well but maybe not as well as some of its best polls.