Education not taken into account sufficiently by polls

The report, led by Professor John Curtice senior research fellow at NatCen, examines several theories as to why there is a divergence in referendum voting intentions reported between the two polling methods.

His conclusion is that the difference is more likely to be down to the kind of people interviewed rather than the way the polls are administered.

Because surveys such as NatCen’s British Social Attitudes have shown that graduates are much more likely than non-graduates to be in favour of Remain, Curtice argues it is essential that all polls should identify how many graduates and non-graduates there are in their samples. At the moment they are not regularly doing so.

As a result, it is difficult to be sure that one of the reasons that it has been suggested why phone polls tend to have a higher proportion of Remain supporters than internet polls – that they contain more graduates – is correct or not. Such limited evidence as there is appears to be inconsistent.

Curtice said: “From the evidence available we are still unable to give a definitive answer as to why there is a difference between the polls conducted online and over the phone. It seems more likely, however, that the explanation lies in the composition of the samples that they obtain rather than the way in which they ask their questions.

“The failure of most polls to take into account the educational background of their respondents means that we do not know enough about how well either form of polling is adequately reflecting one of the key demographic differences in this referendum.’

The report looked a number of other possible explanations including ‘don’t knows’, ‘shy leavers’ and political commitment.

Some have posited that reticent Remain voters find it easier to say ‘don’t know’ on the internet than they do over the phone. However, online polls still secure lower levels of support for Remain than phone polls even when they do not offer ‘don’t know as an option.

One theory is that Leave supporters are less inclined to say to a phone poll which way they will vote because they think it might be a less socially acceptable answer. However, there is no sign that this is the case when those who say they ‘don’t know’ are pressed for an answer.

Online polls have been criticised for having too many politically engaged people, who are therefore more likely to vote Leave. However, it could be the case that online samples have both more Remain and Leave voters who are politically engaged than phone polls.