Writing about Vote Leave’s targeted digital campaign, Carole Cadwalladr (a writer for The Guardian and The Observer) claimed on Twitter:
On June 19, Britain would have voted to Remain. What changed Britain so directly, so dramatically, so invisibly?
This claim is unlikely, given most public opinion polls of European Union referendum voting intention in the preceding week showed Leave ahead.
‘Britain would have voted to Remain’: This is unlikely, as most estimates of public opinion placed Leave ahead in the week before.
Uncertainty in public opinion: Polls are samples of public opinion. Referendums can be difficult to poll right.
Late Remain shift: There was a shift towards Remain in the final week, partially driven by methodological changes.
What the polls said
In the week preceding 19th June, there were 15 public opinion polls conducted by members of the British Polling Council.
In this period, the unweighted average vote intention estimate for the EU referendum (excluding those who say they do not know) was 51%.
It is unlikely that Remain “would” have triumphed on 19th June, given Leave were typically shown ahead.
Uncertainty and Methodology
There was a great amount of uncertainty in the EU referendum.
Despite assured beliefs that Remain would win, more polls during the campaign showed Leave ahead than found Remain leads.
There were also methodological issues, distinguished by vote intention disparities when polling by telephone or through internet panels.
What mattered when seeking a representative sample in this referendum?
How would those saying in surveys they don’t know eventually vote, if at all?
There was a polling shift towards Remain in the final week, but this was partially driven by methodological changes, as opinion polling companies responded to these challenges.
Dr Jennings (University of Southampton, NCRM) compiled these methodological changes, which all favoured Remain, potentially driven by expectations.
Polling showed it was going to be close, and so it proved.