One of the most important features of polling in the EU referendum to date has been a marked and consistent divergence between those polls conducted by phone and those undertaken via the internet.
The former have consistently suggested that Remain are well ahead; across all phone polls conducted since the beginning of this year Remain have on average been put on 59% (once Don’t Knows are excluded), while Leave have been credited with 41%.
Internet polls, in contrast, have consistently suggested that the referendum race is much closer, either putting Remain or Leave narrowly ahead. Since the New Year they have on average put Remain and Leave equal on 50% each.
This divergence was even in evidence when one company, Survation, switched from doing its referendum polling via the internet to doing so by phone. Survation’s last poll conducted over the internet (in mid-January) put Remain on 47% and Leave on 53%. In contrast, its phone poll, which was conducted immediately after David Cameron concluded the renegotiation of the UK’s terms of membership last month, put Remain on 59%, Leave on 41%, exactly in line with the average of all other phone polls.
However, a poll conducted by ORB and published in today’s Daily Telegraph, has broken this hitherto consistent pattern. Until now the company has conducted its referendum polling via the internet. Like other internet polls, these polls have put Remain and Leave neck and neck – indeed the two polls they had conducted since the turn of the year had on average put Remain on 50%, Leave on 50%.
But ORB’s poll today has been conducted by phone. Yet it fails to put Remain well ahead. Instead, once (a small proportion of) Don’t Knows are left to one side, Leave are narrowly ahead on 51%, while Remain are on 49%. Anyone who was hitherto inclined to believe that those polls conducted by phone were giving the more accurate picture of the state of the referendum race must now feel rather less certain about what the outcome will be.
Of course, a degree of caution is in order. Perhaps this poll is a ‘rogue’, its figures overestimating Leave’s strength simply as a result of the chance variation to which all polls are subject. And perhaps we should note that the sample for this poll, 823 voters, is somewhat on the small side, thus making it particularly vulnerable to chance variation. But the findings of future phone polls are clearly now going to be greeted with a renewed interest.
In his write up of this poll, Lynton Crosby, the man often regarded as the architect of the Conservatives’ success last May, focused not on the fact that the poll was the first phone poll to put Leave ahead, but rather on its evidence that Leave voters seem more inclined to go to the polls than Remain voters – and that consequently Leave’s position may be rather stronger than appears at first sight. In particular, he reports that amongst those who say they are certain to vote, Leave are ahead in the poll by 54% to 46% rather than by the 51% to 49% margin amongst all those who expressed a vote intention.
There is certainly one reason why we might expect those who are inclined to vote to Remain to be less likely to vote than those who wish to Leave – they are much more likely to be younger and, in general, younger voters are less likely to make it to the polls. At the same time, however, Remain voters are also more likely to be middle class ‘AB’ voters than working class ‘DE’ ones – and middle class voters are also more likely to vote. In other words, the demographics of the referendum vote do not all point in the same direction so far as the relative likelihood of Remain and Leave voters turning out to vote is concerned.
Still, there are some straws in the wind that suggest that Leave voters may be more determined than those who wish to Remain to make it to the polls. Apart from the evidence of today’s poll, ORB’s previous poll (conducted over the internet) found that a Leave vote of 52% amongst all respondents increased to one of 58% amongst those who said they were certain to vote. Meanwhile, in the last poll that they conducted over the internet, Survation found that 76% of Leave voters said they were certain to vote, compared with just 59% of Remain voters – though in this instance the poll’s headline figure (53% for Leave) took account of respondents’ reported likelihood of voting.
Not that what people say in polls about how likely they are to vote is necessarily a good guide to the prospects for differential turnout in either an election or a referendum. Polls find it easier to interview those who are likely to vote in the first place, and one of the patterns that undid them in last year’s general election was that the younger voters that they managed to contact were more likely to turnout than were younger voters in general. Still, that experience serves as a warning that estimating differences in turnout is essential if polling is to be accurate. At the moment, however, most polls of referendum voting intention are not making any attempt to estimate possible differences in the likelihood of voting in the referendum. Perhaps it is now time that they did so.