Nigel Farage, leader of the newly-formed Brexit Party, declared on Twitter, referring to Labour’s deputy leader Tom Watson MP:
The second referendum that [Watson] backs is a total insult to 5 million Brexit Labour voters.
The Twitter account for ‘Labour Leave’ made similar claims.
Are there really 5m Labour voters who backed Leave in the 2016 EU referendum? Survey projections suggest there are less than 4m such voters.
How did Farage get to 5m Labour Leave voters?
It is unclear what source the Brexit Party leader has used for this claim.
Following the 2016 EU referendum, there were multiple surveys which looked at each party’s voters in 2015, and what share backed Remain or Leave.
Excluding people who did not turn out to vote in 2016, YouGov’s post-referendum survey suggested 35% of 2015 Labour voters backed Leave. For the Lord Ashcroft-badged survey (which was conducted by an actual polling company), that figure was 37%.
One possible method is the 37% figure was multiplied by Labour’s 2017 vote count (about 12.9m), arriving at 4.8m, before rounding up to 5m.
Why is this wrong?
That calculation wrongly assumes Labour’s 2017 vote has the same Remain-Leave shares as their 2015 vote, ignoring people who did note vote in 2016.
There were major vote changes in between those two elections, as the two-party vote share went from 67.3% in 2015 to 82.4% in 2017. We need to understand how many 2016 Leave voters cast a Labour ballot in 2017.
According to the British Election Study internet panel (conducted by YouGov), about 23% of Leave voters chose Labour in 2017. This figure is among those who voted in both the 2015 and 2017 General Elections.
Ipsos MORI, based on their pre-election vote intention polling, suggested that 24% of Leave voters backed Labour (among those who turned out to vote).
If we apply 24% to 17.4m Leave voters, we get about 4.2m. A better estimate would need to account for 2017 non-voting by Leave voters. If 10% of Leave voters failed to cast ballots in 2017, then that estimate would be about 3.8m.
There are MRP estimates, by constituency, of 2017 party vote by 2016 EU referendum vote. This research used YouGov polling data and was commissioned by Best for Britain. MRP is short for ‘multi-level regression with post-stratification’. This technique builds a model for individual opinion, based on demographic characteristics and constituency, giving weighted estimates to reflect how constituencies differ by what people live in them.
Selected constituencies suggest 2017 non-voting by 2016 Leave voters was at 11% in Maidenhead, 15% in Lewisham East, and 17% in Newport West. The full published estimates suggest there are around 3.4m Labour Leave voters (and 3.5m Conservative Remain voters).
Additionally, the British Election Study’s post-election random probability survey in 2017 estimated 30% of Labour voters had chosen Leave in the 2016 EU referendum. The central projection is then 3.9m, or 30% of 12.9m. Due to margins of error: the plausible range lies between 3.4m (26%) and 4.4m (34%).
Central projections based on survey data should then suggest there are less than 4m (2017) Labour Leave voters, rather than 5m. Survey estimates are subject to multiple sources of potential error.