Thread by @robfordmancs

OK, seeing as the false argument "Labour needs Leave seats therefore Labour needs Leave votes" is now one getting presented in Shadow Cabinet, let's have yet another go at highlighting the problems with it 1/?

https://twitter.com/robfordmancs/status/1141331997870252032

Attached are estimates of 2016 and 2017 vote patterns in Bassetlaw, constituency of @JohnMannMP who has repeatedly asserted that most Labour voters in Leave seats are Leave voters. Bassetlaw is estimated by @chrishanretty to have voted 68% "Leave" in 2016.

First, a quick note on the source and methodology - these come from Multiple Regression and Stratification models on very large survey datasets conducted by @kevcunningham and @election_data

The logic of the method is this: we know how different characteristics influence how people vote overall, and we know roughly what mix of people with different characteristics live in each seat, so we can put these two sources together to estimate the balance of votes

This is the same methodology that YouGov and @benlauderdale used in their very successful seat predictions in the 2017 election - the only ones which saw Labour's surge in Remain seats like Canterbury coming

Like any method, it isn't perfect, but it is seen by plenty of researchers, including people much smarter than me, as a powerful technique for estimating local behaviour from national data. Its the best we can do with what we have.

Now, what does the model tell us? Firstly, that John Mann is mistaken in thinking that "most" of his Labour vote comes from working class Leave voters. 45% of his 2017 vote backed Leave in 2016. 43% backed Remain, and 12% didn't vote in 2016.

So, though 68% of voters in Mann's constituency voted to "Leave" considerably less than half of Mann's Labour votes came from people who backed Leave in 2016. Instead, his vote is split almost exactly evenly between Leave and Remain, with a large chunk of 2016 non voters.

A 68% Labour Leave seat can also be one where 55% of Labour voters did *NOT* vote Leave because voting patterns are complicated. Mann wins a massive share of the local Remain vote, while the Leave vote breaks heavily to Con but with a substantial Lab minority

People also overlook turnout. Lots of 2016 "Leave" voters simply stayed home in 2017 - whereas 2016 "Remain" voters were much more likely to vote in 2017 too. Lots of people who stayed home or couldn't vote in 2016 turned out in 2017.

Work since 2016 has shown that non-voters in 2016 have since broken heavily towards "Remain". So that non-voting group is v likely "Remain" skewed. It is also growing steadily, as every day new voters who couldn't vote in 2016 are turning 18 and joining the electoral rolls.

Nor should we assume that Mann's 2016 Leave voters have the same views now as then. Another MRP analysis this time by @chrishanretty and Survation in November 2018 showed a move to "Remain" concentrated in "Labour Leave" areas survation.com/wp-content/upl…

Survation estimated that Bassetlaw council area, where Mann's seat is based, had a 56% "Leave" vote in Nov 2018, down considerably on 2016:
survation.com/what-does-the-…

Given all this, what can we say about "Labour Leave" seats. Firstly, the Labour vote in such seats will usually be much more Remain than the overall 2016 local vote was, for all the reasons outlined above.

Secondly, this doesn't lessen the dilemma for MPs in these seats, in fact it worsens it. Life would be much easier for Mann & others if their vote skewed v heavily "Leave". Instead, it usually splits evenly, which means antagonising *either* "Leave" or "Remain" voters is risky

Thirdly, what this underlines is that it is absolutely daft to claim, as the shadow cabinet briefing did, that because a subset of seats voted to Leave in 2016, Labour has to focus only on Leave voters to win/retain these seats. This is the ecological fallacy.

Fourth, it is even more daft to treat these seats as if they are the main or only places Lab need to worry about. Lab also won or came close in many Southern "Remain" seats, and need to win in Scottish seats with big "Remain" electorates, to cite but two egs.

Fifth, it is also daft to treat the 2016 election as the Tablets of Sinai, an expression of opinion Fixed For All Time. People who voted in that election can, and have, changed their minds since. Even more so people who didn't, or couldn't. Those effects steadily grow over time.

Labour's voters are divided over Brexit, but the split is uneven overall (70-30 to Remain), unstable & complicated right down to the local level. No one is served by shoddy simplistic analysis treating large groups or areas as homogenously one thing or t'other, or fixed over time

Things have come full circle in 6 yrs. Back in 2012-13 I regularly argued Lab faced a risk from neglecting disaffected working class voters defecting to UKIP. Now the risk they face is focusing only on this group and neglecting everyone else. Politics, eh? Funny old game. /ends/