James Austin: With a week to go Delta, Yougov, Survation and Ipsos were closest. Putting aside yougov doing performance art with their polling... those 23/23/24% leads respectively.

The professional negligence election continues to deliver, with the latest salvo being “crowdfunding enough to hire an extra organiser for Jeremy Corbyn to instead commission a (notoriously unreliable!) constituency poll that then shows Jeremy Corbyn losing”.

Serious question. Do you think the avg. of 41/20/12/15/5 is now baked in (if not the batter ready to go into the oven) or are the polls monumentally wrong?

The wider data strongly supports the polls not being wrong: party behaviour, constituency polls, random sampling polls etc. + they've had a very good record through this Parliament including in the locals 2 months ago. So unlikely they're wrong.

Gold standard Ispos just come in with that

Yeah, IPSOS and survation are arguably the two with the best track record - and using phone polling. And they've come in with 23% leads.

I also note that Deltapoll was ominously stable in 2019. And has been this election as well.

If you go on who and what methods were accurate in 2019 (which doesn't always carry over) its properly death zone for the Tories.

With a week to go Delta, Yougov, Survation and Ipsos were closest. Putting aside yougov doing performance art with their polling... those 23/23/24% leads respectively.

I think YouGov don't know how to weight Tory yeses against the backdrop of Farage Ltd being a national challenger for vote share, if not seats.

Post