7 June 2019

Puzzling polls are a symptom of Britain’s fragmented politics


I wrote last month that British politics was witnessing an extreme level of fragmentation. Little has happened since to change that assessment, and indeed the last 24 hours have seen two data points underline it.

The Peterborough by-election saw the Conservative and Labour combined share collapse from 95 per cent in 2017 to 52 per cent, with the Brexit Party coming from nowhere to within 700 votes of taking the seat, while the Liberal Democrats made solid progress on unfavourable terrain. Put another way, it was a 43 per cent swing from the traditional parties to the insurgents.

We also got the latest YouGov/Times voting intention poll. I have cautioned before about over-interpreting results in the aftermath of events. But we are now a week after the results of the European elections became known, and the Brexit Party and the Lib Dems are still both polling in the 20s. This does not feel like a blip.

So what, in broad terms, is actually going on?

A lot has been said and written in recent years about the increasing importance of the social liberalism versus social conservatism axis – highly if imperfectly correlated with attitudes to Brexit – as a complement to the traditional economic left-right axis.

This predated Brexit – since the shift of the UK, like the developed world generally, from industry to an information-based economy, the focus has shifted from occupational social class (which correlated with left-right attitudes) to education level (which correlates with cultural disposition).

In the last few years, Brexit (and its present impasse) has pushed the cultural axis to the fore, and conversely, the fact that the deficit and the economy in general have fallen off the political radar has made the left-right axis less important. And recent months have seen the reorientation of the party system to reflect this.

For this reason, what we now see is not as simple as purely a rage against the establishment. The two traditional main parties are indeed part of the establishment, but their current difficulties may also reflect that they represent the ends of a spectrum that sits at right angles to the biggest issue of the day. As such, many voters will find them an ineffective tool with which to express their view on it.

And they have plenty of alternatives. Of course, how much voters actually know about them and their policies is hard to gauge. Some parties seem to be doing better on this than others – it’s clear, for example, what the Brexit Party stands for, simply from its name.

Yet even if voters have a good understanding of party policies and positions, it doesn’t necessarily follow that a vote for a party is a vote for its position on a particular issue. Number Cruncher Politics polling of the European Parliament election found that half of voters were voting on something other than Brexit – a huge chunk given the extent to which Brexit dominated the coverage and campaign.

This, in turn complicates the debates with Labour and the Conservatives around Brexit, and which course of action is optimal. This debate is complicated enough already, because both “main” parties have gained and lost so many voters over the last few cycles that the voters they have lost recently, and the ones either would need to win back in order to win a majority, are quite different.

In reality, the conflicting data often leads to cherry-picking of the bits that suit a preferred narrative.

The current environment is also a big headache for pollsters. Getting representative samples of the population – widely acknowledged as the root of the industry-wide disaster in 2015 – matters less in stable times, when weighting respondents’ past vote to the election result keeps their current voting intention more or less in line.

But when there is a high degree of electoral flux, and past and present voting decouple from one another, sample accuracy returns to the fore. It is very likely that the huge differences between pollsters at the European elections reflects this.

It also raises another risk to polling accuracy – with more viable parties on offer, voters’ predictions of how they intend to vote almost inevitably become less solid. Polling on and after the day of the European elections from both Opinium and Lord Ashcroft found that about half of 2017 Tory and Labour voters who switched to one of the pro-Remain parties made up their minds in the closing days of the campaign. This, along with higher turnout among Remainers, very likely explains why the Lib Dems and Greens performed better than the average of polls had suggested.

And what of Westminster polls conducted since the European elections? There have been four, from three different pollsters, putting three different parties in the lead. How long this state of flux continues is hard to predict, but presumably at least until there is greater clarity on Brexit.

The next few months will be interesting, to say the least.

CapX depends on the generosity of its readers. If you value what we do, please consider making a donation.

Matt Singh is the founder of Number Cruncher Analytics