How Prof Steve Fisher and Nate Silver provided the pointers
On Wednesday, alongside Keiran Pedley and other leading pollsters, I took part in a post GE2017 conference organised by the University of Loughborough at its London campus in the Olympic park. It was a good event and I’m looking forward to some of the serious studies, including the post-election BES analysis that will be published.
My contribution started by asserting that the whole polling CON landslide narrative during the campaign had been fueled by the May 4th local and mayoral elections where the Tories enjoyed their best night for ten years and UKIP had been all but wiped out.
The purples lost all the 145 council seats they were defending which was offset by just one gain. This fueled the widespread assumption that the Tories would pick up the lion’s share of UKIP’s GE2015 votes on June 8th and, accordingly, we could assume blue gains in some very unlikely constituencies.
One bit of analysis that got widely ignored was from Professor Steve Fisher of Oxford University and a leading member of the exit poll team. He examined what had happened in council seats which UKIP had contested four years earlier but didn’t put up a candidate in May. The conclusion from the real election data was that the Tories had benefited the most but didn’t do substantially better than LAB.
So the simplistic maths that many were doing of simply adding a large part of the UKIP GE2015 share in a seat to the CON total was probably overstating the blues.
A second theme that ran very strongly was, based on GE2015 experience, that the polls always understated the Tories. So if there was a range of GE2017 polls with similar fieldwork dates then the ones with the best CON position should be given the most credence. The polls that had it much closer like Survation and the YouGov model were dismissed as outliers.
I was very struck at the time by the US analyst, Nate Silver, who made an interesting observation on that which proved to be spot on.
Yes you can point to an historical trend for the Tories to be understated at general elections but this effect rarely occurred when the blues were enjoying clear polling leads.
The main exception to this was GE1987. Thus it has been LAB that has been understated in two of the past three general elections.
So my view and betting as we waited for exit poll on June 8th was that there would be no Tory landslide but I certainly wasn’t expecting a hung parliament.
With Commons seats spread betting the more you are right the more you win and as I reported here at the time I “sold” CON seats at 393. They ended up with 318 so I won 75 times my stake level.
This proved to be a very profitable night – a big contrast to two years earlier.