This is my initial forecast, so I’m in search of a reasonable starting place. What is most interesting (for me) at the moment is that there’s a fairly large gap between what the betting market suggests is the right prior and what the polls suggest is the right prior. The betting market tells us that there’s a 74% chance that Britain will vote to stay in the EU, and a 26% that she will vote to leave it. But the polls are a different story. One source reports a poll of polls that shows that there is a 51% to 49% split for staying.
However, another sources report a different picture:
“Nearly two thirds of UK voters are sceptical about the European Union, but less than a third want to leave, according to new analysis”.
The same course looks at the development of attitudes over time suggests that support for remaining in the EU has been growing over the last few years and is now roughly as high as it was in 1997.
Betfair itself takes up the matter of the discrepancy between the market and the polls. Not surprisingly, it sides with the markets! But the discussion is worth reading and implies (if I understand it correctly) that, when Scottish independence was up for a vote in 2014, the polls initially suggested a fair amount of support for independence, but much of that support melted away as the date of the referendum drew nigh.
The build-up to the Scottish Independence referendum of 2014 was fairly low key for almost two years. It was only after the SNP had gained considerable momentum, and independence became a possibility, that the campaign sparked into life and the majority of Scots (and politicians from south of the border) engaged with the question. Ultimately, voters decided to stick with what they knew and keep Scotland in the union. When uncertainty kicks in closer to June 23, the likelihood is that the millions of undecided Brits, who don’t feel strongly about the EU either way, will choose to support Remain.
But I’m not sure that’s how I read the data.
Opposition to independence went from 61% in December 2013 to 55% in April 2014 to 51% in September 2014. So there wasn’t really a trend in the polling away from being favorable to independence. Rather, what’s relevant here is that the independence referendum was eventually defeated about 55% to 45%. So – if the 2014 referendum is relevant here – the take away is not so much that we should expect the polls to narrow as we approach the Brexit vote. It’s that we should expect the polls to indicate that there’s more support for Brexit than the election will reflect. And that might simply be because polling in Britain has not that predictive over the last decade or so.
At this point, I think it’s reasonable to afford the polls a 40% share of determining my prior and the betting market the rest, though I’ll be hunting for better predictive measures.
- Stay in the EU = (0.4)(0.51) + (0.6)(0.74) = 0.648
- Leave the EU = (0.4)(0.49) + (0.6)(0.26) = 0.352
 http://www.telegraph.co.uk/news/newstopics/eureferendum/11617702/poll.html – probably taken from http://whatukthinks.org/eu/opinion-polls/poll-of-polls/#