You might have seen references in the press recently to the the latest findings of a Survey undertaken by NatCen and the subsequent report written by Sir John Curtice (HERE). It has been given wide coverage including by the BBC (HERE). The headline finding is that in any new referendum 55% would vote to remain and 45% to leave (Table 14). This isn't really news since it follows the vast majority of recent polls. The other main finding, again hardly a revelation, is that about 80% of both remain and leave voters think the government is doing a bad job of negotiating Brexit (Table 7). But the really surprising thing is in a small detail and not mentioned at all by the media.
The survey was of people who were all part of previous BSA (British Social Attitudes) surveys and all selected at random. The report (page 2) describes the cohort like this:
In this report, we focus on the findings of the latest wave of interviewing, which was conducted between 24 January and 17 February 2019, that is, after the House of Commons initially voted against acceptance of the deal. During this period, we interviewed either online or by telephone 2,654 people, representing 57% of those who were invited to complete the questionnaire. All the respondents are people who were originally selected (at random) and subsequently interviewed as part of the fieldwork for a previous British Social Attitudes survey, and who then have agreed to undertake short follow-up interviews.
In other words they had all been interviewed before. The original interview showed they were a perfectly representative sample of the population in that in 2016 they voted 51% to leave the EU and 49% to remain.
But now note this comment on page 15:
At this point, though, we should enter a note of caution about our evidence. When we first conducted our surveys, the way in which our respondents said that they had voted in the EU referendum was little different from the actual outcome. In our September 2016 survey, for example, 49% said that they had voted Remain, while 51% indicated that they had backed Leave, figures that are only one point adrift of the actual outcome. However, the proportion who report having voted Remain has gradually increased during the course of the last two years, and in our most recent survey, 54% said they had voted Remain, while only 46% said that they had backed Leave. This apparent bias should be borne in mind in evaluating the implications of what our respondents say that they would do if another referendum were held now.
Professor Curtice doesn't explain how, or even if, this impacts the present 55% to 45% majority for remain and perhaps it doesn't. He only observes that it's happening and calls it an 'apparent bias' that has 'gradually' crept in during the last two years.
The discrepancy surely cannot be explained by people forgetting how they voted since it would presumably apply to both leavers and remainers equally. It seems more likely (in my opinion) that some regretful leave voters don't want to admit - even to themselves perhaps - their own small part in the chaotic mess that Brexit has resulted in.
Now this 'bias' has only come to light because the cohort in this particular survey has remained the same. In most polling the pollsters rely on the respondents to self-declare how they themselves voted in 2016 so the bias goes undetected.
Does it mean more people have actually changed their mind than admit doing so? I don't know.
Curtice often says there is little evidence of voters changing their opinion yet this is precisely what seems to be happening, albeit in a subliminal way, and note the bias is consistently towards remain and not the other way.
I expect the number of people thinking and claiming they voted remain will continue to rise. After the war (and here I am not suggesting for one second any link or similarity whatsoever between leave voters and Nazis) it is claimed you couldn't find anybody in Germany who voted for Hitler in August 1934. It had become socially unacceptable to admit to it.
One day Brexit will be the same.
