Why the difference between phone and online in EU referendum polling?

Anthony WellsHead of European Political and Social Research
February 23, 2016, 11:00 PM GMT+0

There’s a big difference between the online and telephone polls on the EU referendum – with online polls showing the sides neck-and neck and telephone polls showing about a 15% gap in favour of ‘remain’. Why?

It’s striking that both methodologies right across the different polling companies give about the same number to the ‘leave’ campaign, around 40%. The difference is in the ‘remain’ number, which is around 52% from the telephone polls but only 40% for online polls.

Telephone polls ask their respondents “How will you vote in the referendum?” People are assumed to have an opinion, and 90% of them give one. By contrast, online polls present people with options: remain, leave, won’t vote, don’t know – there is less assumption of an opinion, and 20% or more don’t offer one.

Intuitively, it is much more likely that 20% rather than 10% have no opinion – indeed one might think the real number of ‘don’t knows’ should be even higher. But in person-to-person dialogue people feel pressured by the expectation of an opinion, and then it’s much more likely to be for the status quo rather than change.

We believe this accounts for most of the difference between the methodologies. It doesn’t make one right or wrong. Indeed, judging how many people really have an opinion and will turn out to vote is the big question mark hanging over this campaign, and we expect to find out more and more as we get closer to the moment of decision.

There is the possibility that the different methodologies reach slightly different types of people, but the evidence of the last general election showed this was largely a red herring – the final eve-of-election polls made the same prediction regardless of whether they were conducted online or by phone, and they were just as wrong.

So what should you conclude from the current polls on the EU referendum? Well, if there are lessons to be learnt from the polls last May, it’s that they are not infallible, and sometimes it’s the underlying questions that are a more useful pointer to the final result. Historically polls three or four months ahead of a referendum have not been a good predictor at all – voting intention in referendums can move fast. However the underlying questions in our polling shows that ‘remain’ has some notable advantages – leaving the EU is seen as bad for jobs, bad for British influence, bad for the economy. Above all, remaining is seen as the safe option, leaving is seen as increasingly risky. And in our experiments we found a much higher ‘potential vote’ for remain than for leave. If we had to make a prediction based on our current data, we would expect the poll to end up moving towards ‘remain’.