OK, so curious what @patricksturg.bsky.social thinks of this. In the raw F2F @britishelectionstudy.com from 2024, the raw (unweighted) vote share is 44% Labour, 20% Conservatives. This seems a big over-statement of the actual lead?
Replies
Is random probability still 'the gold standard'? Discuss.
sep 3, 2025, 12:49 am • 0 0 • view
(Normally when I find something like this I spend about an hour worrying my code is wrong...)
sep 3, 2025, 1:03 am • 3 0 • view
I don’t know much about the BES sample design, but would you expect it to yield accurate results unweighted? I believe the BES releases the design weight prior to any nonresponse adjustment. I’d want to look at that. Might still show the same thing, but at least you’d be able to rule out the design.
sep 3, 2025, 1:19 am • 2 0 • view
Word
sep 3, 2025, 1:48 am • 1 0 • view
I looked at the design weights and I think they are pretty similar numbers (just away from my computer now).
sep 3, 2025, 1:40 am • 1 0 • view