I understand why these disclaimers exist, but Para 5. is a bit pious. I've seen better disavowals...

Every time I see Sanchez speak/read him quoted, I think "you lead a minority government, you are supposed to be smol bean, why are you giving such big quotes"

It's not even scalable within model! "As LLMs are stochastic there is substantial dispersion of estimates across different instances of the same AI model, including differences in the sign"

thank you!

Do you know who wrote that, or what it was in response to? (Don't have a research professional subscription)

Got to push back on this. Intro to Politics at @rhulpir.bsky.social is taught by three full professors, in a dept. where 64% of our research is world-leading.

Politics & Policy | PSO Policy Research Journal | Wiley Online Library This paper analyzes how electoral system change affects the electoral performance of those parties that initiate these reforms. The hypothesis is that those who created the new system benefited from...

Parties that initiated electoral reforms gained no clear electoral advantage over others, suggesting electoral engineering is less common or less effective than often assumed doi.org/10.1111/polp...

a cartoon of a penguin with sweat coming out of its face Alt: a cartoon of a penguin with sweat coming out of its face

TFW when you're searching for some stats term related to your work and one of the top results is a blog post by Terence Tao

I did read "priests" as "people who are paid enough to live but no more", and interpreted it as a call for better pay.

Motivating Job Seekers. A Field Experiment* Bart Cockx, Johan Egebark, Greet Van Hoye, Emilie Videnord, Johan Vikström February 2026 Abstract: Reduced motivation among jobseekers over the unemployment spell may lead to declining job-finding rates. We report findings from a low-cost digital intervention with motivational emails aimed at enhancing and sustaining motivation and search effort among job seekers in Sweden. Using a randomized controlled trial that included 200,720 job seekers, we evaluate both carrot messages aimed at encouraging the pursuit of personal goals and intrinsic motivation and stick messages focusing on external pressure and constraints. A large share of job seekers opened the emails, and they triggered behavioral responses. Both types of messages backfired, reducing search effort and job-finding rates. The carrot messages reduced both the number of job applications and job finding, particularly among men. One likely explanation is that these messages signal to job seekers that the Public Employment Service was less controlling than initially perceived, prompting a reduction in effort. The stick messages backfired for job seekers who, at the onset of unemployment, reported that they were motivated by an inner drive rather than by constraints. These findings underscore the challenges of motivating job seekers to actively search for jobs and suggest that low-cost digital interventions, in isolation, are inadequate and may even be counterproductive.

Interesting workingpaper on effect of motivational emails on on job search effort. The Nudge hype is as dead as a dodo. rfberlin.us21.list-manage.com/track/click?...

Screen grab of an email from jobs.ac.uk saying there have been no new jobs in Politics in the last 3 months

C'mon jobs.ac.uk , I know things are bad but this is ridiculous

Is the internet a bit shit today? I mean the plumbing, not the content.

Nice to be able to compare the coefficients across models by plugging in the standard errors of ... uh, zero? It's not often I'm asking for more significant digits.

Some things are still better done IRL. I thought instructing a solicitor online because there are firms that can bulk process house purchases, but I've spent so much getting people to witness things / posting documents that it's been more trouble than it's worth

Interesting chatter about this www.reddit.com/r/math/comme...

Math community is often skeptical, and I’ve never seen them this stunned by a result. And by a public model too.

The Hungarian Election Polls, the Day After Hungarian Election polls, the Day After Prepared by Claire Durand (Université de Montréal) and Tamás Bodor (University of Wisconsin) ...

How did the polls do in the Hungarian elections? "The pollsters who were considered close to the opposition or independent produced very good estimates in general" - ahlessondages.blogspot.com/2026/04/the-...

Thank you, installed now!

Isn't this dealt with in the first two sentences of the abstract?

The Professor Will See You Now: Ageing Lessons in political science. This week: ageing

Me, in @thehousemag.bsky.social, on politicians ageing - they talk less about the future as they get older - and on the public's time horizons for change: this government‘s "short term" is coming to an end very soon.

www.politicshome.com/opinion/arti...

I'm surprised that the Lords was fine with the government being able to point to this or that group and say, "banned".

Labour Party — 4,900 seats contested (96.7%) Reform UK — 4,821 seats contested (95.2%) Conservative and Unionist Party — 4,771 seats contested (94.2%) Green Party — 4,505 seats contested (88.9%) Liberal Democrats — 3,949 seats contested (78.0%) Independent candidates — 794 seats contested Trade Unionist and Socialist Coalition — 286 seats contested (5.6%) Workers Party of Britain — 69 seats contested (1.4%) Newham Independents Party — 66 seats contested (1.3%) Social Democratic Party — 48 seats contested (0.9%) Aspire — 45 seats contested (0.9%) Tower Hamlets Independents — 43 seats contested (0.8%) Redbridge Independents — 42 seats contested (0.8%) Havering Residents Association — 35 seats contested (0.7%) Kingston Independent Residents Group — 34 seats contested (0.7%) Christian Peoples Alliance — 30 seats contested (0.6%) Your Bradford Independent group — 29 seats contested (0.6%) Ealing Community Independents Party — 26 seats contested (0.5%)

Democracy Club tell me they think this round of local elections is the first time EVER the Greens have run more candidates than the Lib Dems.

By quite some margin too:

Green Party — 4,505 seats contested (88.9%)
Lib Dems — 3,949 seats contested (78.0%)

Transformative Treatments Contemporary social-scientific research seeks to identify specific causal mechanisms for outcomes of theoretical interest. Experiments that randomize populations to treatment and control conditions a...

Although parenthood is the archetypal example of the transformative treatment doi.org/10.1111/nous...

aes(alpha = 1 / 1e6)

Sometimes it's because there simply aren't enough respondents to model those effects *well enough*. The classic example is Orkney. If you have a national sample of 20,000, fourteen *might* be from Orkney, so hard to say Orkney and Shetland is nailed on for the Lib Dems based on those 14

What they don't do, and shouldn't do, is say, "you're telling me right now that you're going to vote for this party, but past experience says you'll end up voting for this other party" (2/2)

That's not really true. Models include variables which tap constituency context, and will also have some constituency specific effect, modelling what's left over after demographics + context (1/2)

MRP accuracy is comparable to seat polls dx.doi.org/10.1017/psrm... ; any method which doesn't have systematic biases, incl lots of seat specific polls, will benefit from errors cancelling out

I think what people mean by “public good” is “it’s good for the public” but I’m sorry economics never evolved a term for that

👀🗳️ When polls close in Hungary at 7pm, there won’t be proper exit polls. Instead, some reliable independent pollsters will release embargoed surveys conducted just days before the vote. Their results are in line with their previous, already published findings (no surprises).

The problem is not primarily statistical. I believe these MRPs are good estimates of what you would get if you did a large random probability sample in each constituency. I also believe people's vote intention right now matters very little for constituency outcomes in years' time.