Paper is a collab effort w/ awesome @scotvoting.bsky.social team; @robjohns75.bsky.social @frasmcm.bsky.social @ailsahenderson.bsky.social & @cjcarman.bsky.social . Thanks also to UK and Scot Govs for providing high profile scandals while we were able to collect data

Diolch Thom - you can get a general idea of it here and see individual constituency estimates. This isn’t exactly how i calculate mine but it does a good job jaclarner.github.io/senedd_ethol...

Here you go yougov.co.uk/politics/art...

This should be Reform on 37 not 35!

The Westminster implications are staggering too. On these numbers, Labour would likely be reduced to just 3 Welsh seats, Plaid would win around 6, and Reform could take most of the remaining 23 seats out of Wales's 32 constituencies 🧵 End

Reform's success creates its own problems for government formation. By cannibalising Conservative support, they may find themselves without natural coalition partners, while their vote could be inefficiently distributed across constituencies 🧵

Coalition arithmetic looking a bit complicated - on these numbers only current realistic possibility is some sort of Plaid and Lab agreement (perhaps with LD and Green support). However, now in territory where more Labour losses put this at risk 🧵

The seat projections are pretty dramatic. Allowing for movement of a few points in many directions: Plaid Cymru ~38 seats, Reform UK ~35 seats, Labour ~11 seats, Conservatives ~6 seats, Lib Dems ~3 seats, Greens ~ 1 🧵

Probably would! www.nber.org/papers/w33767

Yes exactly - it calculates swing from 2021 estimates so Greens would win in Cardiff South as they already did well there previously

Finally it gives us information about coalition possibilities and proportionality of the results. I've written up an explainer here that outlines how it works, what it does and doesn't do, and also plans for improvement. blogs.cardiff.ac.uk/thinking-wal...

It also visualises for seats are actually allocated in different constituencies and tells us how tight the margins are.

First, and the obvious one, is that it can turn vote shares into estimates of seats in the new expanded Senedd using either uniform or proportional swing. Users can either enter their own numbers or choose from presets of different scenarios

This makes complete sense given who voters do and don't like! Here is a plot of mean 0-10 liveability scores 2024 voters gave to other parties. Labour and Plaid voters like each other quite a lot, whereas Reform voters don't particularly like anyone!