Thread by @jonworth

Just checking how well my #tacticalvoting recommendations did when compared to reality

In the end I recommended explicitly in 187 constituencies

I was wrong in just
Eddisbury (rec Lib Dem, should have been Labour)

Isle of Wight (rec Green, should have been Labour)

Kirkcaldy and Cowdenbeath (couldn't recommend SNP due to candidate suspended for anti-semitism, recommended Labour, and the ex-SNP candidate won)

Maidstone and The Weald (rec Lib Dem, should have been Labour)

New Forest East (rec Lib Dem, should have been Labour)

New Forest West (rec Labour, should have been Lib Dem)

North East Somerset (rec Lib Dem, should have been Labour)

Solihull (rec Lib Dem, should have been Labour)

South East Cornwall (rec Lib Dem, should have been Labour)

9 wrong, out of 187

How many of those would tactical voting have worked in?
Eddisbury ❌ (Tories >50%)
Isle of Wight ❌ (Tories >50%)
Kirkcaldy and Cowdenbeath (special case - see above)
Maidstone and The Weald ❌ (Tories >50%)
New Forest East ❌ (Tories >50%)
New Forest West ❌ (Tories >50%)

North East Somerset ❌ (Tories >50%)
Solihull ❌ (Tories >50%)
South East Cornwall ❌ (Tories >50%)

So everywhere where it could have made a different to recommend between among Remain parties I got it right

How does that stack up against other tactical voting recommendation tools?

@tacticalvote didn't call a few dozen of the complex ones, and also got Isle of Wight and South East Cornwall wrong, but made no serious wrong recommendations

@BestForBritain got Broxtowe, Bury St Edmunds, Eddisbury, Isle of Wight, Luton South, Maidstone and the Weald, South East Cornwall, Sheffield Hallam, Solihull wrong

(not checked them all, but at least Broxtowe did not have > 50% for the Tory)

@votetools got Cities of London and Westminster, East Lothian, Epson and Ewell, Isle of Wight, Kirkcaldy and Cowdenbeath, Mid Sussex, Midlothian, SE Cornwall, Tunbridge Wells, Wantage, Wimbledon, Woking wrong

(serious wrong calls on Cities of London and Westminster, Wimbledon)

@remainutd got Arfon, Ceredigion, Cities of London and Westminster, Epsom and Ewell, Kensington, Kirkcaldy and Cowdenbeath, Mid Sussex, Sheffield Hallam, Solihull, South East Cornwall, Surrey Heath, Tunbridge Wells, Wantage wrong

(serious wrong call on Cities of L&W, Kensington)

@peoplesvote_uk did not call loads of Scottish seats, and got Broxtowe, Eddisbury, Isle of Wight, Kensington, Maidstone and The Weald, New Forest East, North East Somerset, Sheffield Hallam, Solihull, South East Cornwall wrong

(Broxtowe, Kensington problematic)

I need to do some more analysis on this, and will then publish my tables. But for now I am pretty reassured - where it mattered I got these calls right.

Who needs MRP models, eh? 😜

Or probably I was just lucky! 😇

/ends

Constituencies that fell to the Tories where I did not consider them close enough to make videos 😢

Rother Valley
Burnley
Durham North West
Blyth Valley
Redcar
Hyndburn
Bury North
Bolton North East
Norfolk North
Leigh

Constituencies that need a swing of 4% or less next time, where I did not make videos, but tactical voting may have helped defend them

Coventry North West
Coventry South
Wansbeck
Stockton North
Normanton, Pontefract and Castleford
Hemsworth
Chesterfield
Warrington North

Cont.

Oldham E & Saddleworth
Wolverhampton SE
Hull East
Carmarthen E & Dinefwr
Wentworth & Dearne
Lancaster & Fleetwood
Doncaster Cent
Halifax
Doncaster N
Batley & Spen
Sunderland Central
Stalybridge and Hyde
Wirral West
Worsley & Eccles South
Eltham
Houghton & Sunderland S

Seats that changed hands in Scotland where I had made no recommendation

Rutherglen and Hamilton West
Renfrewshire East
Coatbridge, Chryston & Bellshill
Dumbartonshire East
Glasgow North East
Fife North East