52-48 for Remain

That’s my story and I’m sticking to it. It’s a significant improvement from my freaking out a week ago today, and I’ve further downgraded the threat level from cautiously pessimistic to cautiously optimistic. It’s also going well out on a limb with a chainsaw considering the volatility in polling estimates and the lack of a consensus across the various polling houses.

Four new polls were released overnight, and the spread is to be expected of this entire frustrating campaign. The distribution in support for Remain is +8%, +2%, -1% and -2%.  (I’m ignoring the two updated polls by the houses not affiliated with the British Polling Council, SurveyMonkey and Qriously, which likewise vary).  NCP is currently forecasting 53-47 for remain with a 75% probability of a remain result.

The source of my unexpected cautious optimism is twofold, based on both variables I discussed a week ago. First, the share that are undecided is still stubbornly impressive. While two of the new polls do not top line undecideds, the two suggesting a narrow leave win both do: 9% and 16%. NCP estimates an undecided rate at 10%. Some will, of course, not vote. Those who are still undecided this late in the game that do vote will skew significantly for remain.  The final poll tracker on the Scottish referendum in 2014 from the BBC estimated 50% no, 45% yes, 6% undecided, the day before polling. The result was 55.3% no, 44.7% yes.

Incidentally, the NCP poll tracker at the top of this page appears similar in pattern to my recollections of the Scottish referendum, although only two polls suggested independence, whereas numerous polls have recently predicted a Brexit victory.

The second source of this (very) cautious optimism is the turnout. Voter registration hit a new record high for the referendum. Ordinarily this is to be expected with each passing year; however as I’ve written elsewhere, roughly 18 months ago the system of voter registration changed in a manner that would inordinately hit the younger cohorts.  Indeed, I saw evidence a couple days ago that suggests a significant increase in self-reported probability of voting amongst the 18-29 year old cohort in the past two to four weeks (this is from memory). Writing this just after 0900 BST, anecdotal evidence, both locally and nationally, is already reporting queues at polling stations.  Recall, the strongest systematic relationship in estimated support for remain is age: the younger are significantly more likely to support remain, the over 55s leave.

One hypothesis, indeed my hypothesis, combines the undecideds with (what might be) higher than expected turnout, the probability for a remain result increases.

For balance, however, while my take on the record registration implies higher than expected levels of the young turning out to vote, I can readily imagine an alternative explanation. A strong runner-up in the various estimated systematic relationships in support for one side or the other in this referendum is social class. The British have a measure for this: A, B, C1, C2, D, E.  C1 and above are significantly for remain, C2 and below for leave. Of course, a turnout relationship also exists here: higher socio-economic groups are significantly more likely to vote. Thus, the increased registration could be a function of new registrants from the lower echelons of this scale mobilised to vote in order to finally “take back control”.

So there’s the hedge. That said, I’m sticking to my prediction above, and warily eyeing this chainsaw I’ve taken with me out on this uncomfortable limb . . .

EDIT: How could I forget the bookies?  Here’s one, 76% for remain, and a tracker.