The Electoral Calculus prediction I find dubious.
It has this for Brighton Pavillion as a description for the seat as a whole:
15% Leave, 24% Left and 31% International.
(Chris Hanaretty's estimate of the leave vote is different. He suggests 25%, which makes electoral calculus's prediction seem even more crackers).
Lucas got 22,000 votes there. UKIP ARE standing but even with the UKIP and Conservative vote added together you get a mighty 15,000 odd votes. Labour are also standing and got 15,000 in 2015. The LDs are not standing.
So I'm very intrigued as to how they have got a CON win there.
Corbyn's Lab are in much the same part of the political compass. Are they really expecting a massive swing of Labour voters in such a remain area to the Cons?!!!
I can tell you exactly what they have done and way its wrong. They are applying national - or regional swings - to a single constituency without any consideration for local nuisances.
This is why Ashcroft whose data is based off actual canvassing is probably different.
The other flaw from the electoral calculus prediction is that its off the 2011 Census data. It now being 6 years later, there may have been a few changes to areas in that time!
What I take from that is this though: why is electoral calculus being so conservative about its prediction about the Green, LD and Labour seats compared to Ashcroft and Elections Etc.
The answer is very simple and I'm seeing it on a very localised level.
Where have the young and mobile professionals been moving and congregating in the last 6 years since the Census? Anyone know how they know where people who have turned 18 since then, now live? Or anyone who has moved into a home and started a new family? Cos I don't....
Its a fairly major flaw.
But the data is still useful. To a point.