Is Labour’s support coming from the 2015 “Others”? 2

Chris Hanretty, Reader in Politics at UEA, wrote a piece yesterday on how the “2015 Other” voters (those who in 2015 didn’t vote for any of the main parties – Conservative, Labour, Liberal Democat, UKIP – or didn’t vote at all) are splitting for Labour in recent polls. He looked at the most recent YouGov poll and identified that from this “2015 Other” group Labour was picking up one and a half times more voters than the Conservatives. He comments:

This means that, compared to the Conservatives, Labour is drawing more than one and a half times more voters from this pool of smaller party voters and 2015 non-voters. … I think these disparities are worth noting, because the two features of this data (non-voters break heavily Labour, and non-voters vote at consequential rates) are slightly controversial. There’s a lot on the habitual character of voting which suggests that non-voters are unlikely to return to being voters. There’s also a fairly large literature which suggests that non-voters are not systematically different from voters in their beliefs. … Eighteen months ago, the answer to the question “Can non-voters win the election for Labour” would have been a resounding no. Some of the polls seem to suggest that non-voters can get Labour close. That, to me, sounds unlikely — but we’ll know within a week.

Here at Forecast UK we wanted to explore this idea and in particular, look at the way this might have changed through May as the Labour lead increased. Our analysis is in this spreadsheet and you can open it yourself to follow where the data takes us.

First a word as to how we have constructed the data. We have taken every YouGov poll in May and put the raw figures for the 2017 Support / 2015 Vote cross tab into a table in weighted and unweighted form. We have adjusted percentages to bring them to sums of 100 (YouGov presents only rounded data so sometimes it does not sum to 100). From this we can extrapolate how the “2015 Others” split into 2017 Voting Intention. Once we have this data we can create horizontal and cell percentages (for eg, what proportion of 2017 Labour voters were 2015 Other). We do this for each YouGov survey in May.

We then tested a number of hypotheses to explore whether changes in the behaviour of “2015 Other” voters were part of the boost in 2017 Labour support.

Analysis

Hypothesis 1 – Raw Labour Support Increases through May

This is a fairly simple hypothesis which most readers will instinctively assume is correct, but for the sake of thoroughness we regressed the stated raw vote share against the closing data of the data sample.

Clearly Labour Support has increased through May

Hypothesis 2 – The proportion of “2017 Labour” voters that were “2015 Other” increases through May

This hypothesis explores whether the percentage of “2015 Other” voters in the “2017 Labour” group increases through May. We show results below for the Weighted data.

The p-value wasn’t significant, but it was of a value that made us consider that a relationship may exist. We regressed again against a 2 value weighted average.

Now we have a much stronger relationship. It appears that is eveidnce that as the month progressed the proportion of 2017 Voters made up of 2015 Others increased.

Hypothesis 3 – The Proportion of “2015 Other” that are 2017 Labour voters increases through May

This hypothesis explores whether the “2015 Other” are increasing their support for Labour in 2017 as we go through May. We show results below for the Weighted data.

We accept the hypothesis. As we have gone through May 2017, 2015 “Other Voters” have increased their support for Labour.

Hypothesis 4 – Share of voters that are “2015 Other” AND “2017 Labour” Increases Through May

Given that the group of voters who are 2015 Other supporting (boolean AND) Labour in 2017 increases through May, we wanted to test that group as a percentage of the overall group (all voters).

We accept the hypothesis. The regression indicates that “2015 Other” AND “2017 Labour” voters are 5.3% of all voters at the start of May and and has increased to 9.1% of all voters at the end of May. This is about an 80% increase and represents around 4% of all voters, a number around half the increase in headline Labour support we have seen through May.

Hypothesis 5 – The way “2015 Other” voters split for 2017 Labour and 2017 Conservative changes through May

This is the crux of Chris Hanretty’s argument, that 2015 Other voters are splitting for Labour rather than the Conservatives. We wanted to test whether that split had changed through May, so we constructed the following table. Remember, these are “2015 Other” voters, not all voters

We accept the hypothesis. What the tables shows us as well is that the proportion of 2015 Other voters who are moving to vote Conservative in 2017 stays broadly the same, but the proportion moving to vote Labour increases dramatically through May.

Hypothesis 6 – The Conservative Lead is related to the Ratio calculated in Hypothesis 5

We accept the hypothesis. There is a clear correlation between the increase in the ratio between Labour and Conservative 2017 support amongst 2015 Other voters and the Conservative lead.

Summary

Chris Hanretty’s analysis is correct and we build upon it to show how the observation he notes (the ratio in the split of 2015 Other Voters for Conservatives and Labour in 2017) changes dramatically through May. At the start of the month, 2015 Other voters are splitting around 3 to 2 Labour to Conservative. By the end of the month it is around 5 to 2 and that change is driven by an almost doubling of Labour support in this group. When we examine this group further we see that the boost in Labour support from “2015 Other” voters represents around 4% of the total electorate and this would be enough to explain around half of the increase in headline Labour support in May (from 29% at the start of May to 39% at the end of May). Approximately another 3% comes from 2015 Labour voters stating that they will vote Labour at the end of the month when they stated at the start of the month that they would support another party / not vote / didn’t know how they would vote.

A few words of caution about this analysis.

• Although we have demonstrated that the increase in support for Labour amongst 2015 Other voters is a factor in the growth of Labour support through May, it is not necessarily the only reason. As we indicate above, this boost in support for Labour amongst “2015 Other” voters only explains about half the increase in Labour support during May.
• This analysis does not include any analysis of likelihood to vote. There is a major unknown in predicting voting patterns on the 8th of June around this area. We cannot dissect the “2015 Other” voters into those who did or didn’t vote in 2015, we cannot assess their stated likelihood to vote in 2017 and we cannot assess their actual likelihood to vote.

Despite these concerns, the picture we are building is of the increase in Labour support being driven mainly by “2015 Other” voters moving to them, with a secondary effect of 2015 Labour voters firming up their support.

Labour’s vote on the 8th of June depends chiefly on these switching voters actually turning out and supporting Labour, If they do (including all the 2015 non-voters, potentially over half of the overall “2015 Other” group) then Labour will do as well as the polls suggest. If they don’t then we could be back in Conservative Landslide territory.

Our latest #ge2017 forecast will be published later today