Alright, who wants some more analysis of the @BestForBritain data? 😬 This time I'm going to look at their recommendations to vote Lib / Lab (or both). I'll exclude Greens / PC. How do the recommendations correspond to 2019 BfB polling and 2017 GE performance. Pretty well... 1/n
@BestForBritain What I have done is run 2 simple logit models - both predict BfB recommending voting Lib Dem tactically (over Lab). The 1st model bases predictions on 2017 GE vote share of Cons, Lab, Lib, UKIP and the 2nd on 2019 BfB poll estimates. Both include 2016 referendum vote estimate 2/n
@BestForBritain What we can then do is compare the probability of BfB recommending voting LD to (a) these 2017 or 2019 vote shares / estimates and (b) to who is predicted to win the seat in 2019. Let's begin... 3/n
@BestForBritain Let's start with the easiest figures. Here we plot probability of recommending voting LD against (a) how well LDs did in 2017, and (b) how they are performing in 2019 in BfB polling. Y/N corresponds to actual BfB recommendation, colour to predicted winner. 4/n 
@BestForBritain What we see is that where LDs did well in 2017 or are predicted to do well in 2019, the predicted recommendations are to vote LD. There's nowhere where LDs did better than ~22% in 2017 that they don't get the nod. Though there are some +30% LD 2019 estimates where Lab do. 5/n
@BestForBritain Here are the predicted recommendations vs Labour vote in 2017 / estimates in 2019. Again follows patterns very closely. Nowhere where Labour got >45% in 2017 that LDs get nod. And I think nowhere where Lab poll >25% now. Pretty favorable to Labour 6/n
@BestForBritain So where are all the 178 LD tactical vote recommendations coming from? Mostly from Conservative voting areas. The probability of recommending voting LD goes above 50% where Tories won > 50% of vote in 2017 or where they poll over 38% in 2019. LDs are the weapon used vs Cons 7/n
@BestForBritain The other way to look at it is to compare to @chrishanretty's 2016 Referendum estimates. Here we see that the LDs are most likely to be recommended as the tactical vote by BfB in constituencies that voted around 50% Leave. NOT necessarily in Remainy areas. True in both models 8/n
@BestForBritain @chrishanretty In sum, BfB recommendations seem fair enough to Libs and Lab compared to each other's vote share and polling. Where the Libs are being favored is as a tactical voting vanguard in (a) Cons voting seats and (b) seats that were around the national average in Brexit support. 9/n
@BestForBritain @chrishanretty That strategy does seem to fit more with the Jo Swinson tactic of peeling of Conservatives (hence reluctance to ally with Labour). But the referendum swing seat issue is more of a surprise given LDs have much more Remainy policy than Labour. 10/n.
@BestForBritain @chrishanretty As ever I'll put data up at github.com/benwansell/Gen… and thanks to data magician @chrishanretty for assembling BfB data. n/n.
