Dr Luke Mansillo

I call bullshit

I am saying that we have no clue epistemically. I’ve done MRP quite a lot and very small model assumptions often change a lot of estimates especially when you’ve got multinominal outcomes where k>5, there are many geographic units j=150 and n is less than 20 for each k*j.

a cartoon character says " or you could trade it all in " ALT: a cartoon character says " or you could trade it all in "

We have no clue what went into the model. What was the rationale for its inclusion? How well the variables predict the dependent variable in the model?
These are very basic things for model based inference.

There are simulations for the credible ranges of primary votes and then there are credibility ranges for the preference flows. This is a lot of matrix multiplication upon models of unknown quality with data that one shows to not be of great quality with design-based inference

My point is about the difficulties in modelling the heterogeneity of preferences with the necessarily complex models and the fancifully small samples for the demands of such a complex model.
Why is no one reporting the credibility intervals of these estimates?

There is a real knack to doing it …

There are major assumptions that makes be made clear. How is the primary vote estimated then how is the preference flow estimated. And I highly doubt for 150 seats there is enough of sample used to credibly estimate these well with any robustness.

Yes but this is conditional on the quality of the MRP model which Demos has not disclosed. I am skeptical of any MRP model without specified demographics, their interactions and the geographic level predictor.
Model based inference needs more disclosure than the design based inference it rests on

Holy cow? That is not an Australian expression.

The death of Brigitte Bardot necessitated the update of this marvelous chart. Only three people mentioned in Billy Joel's banger "We Didn't Start The Fire" are still alive. Source: buff.ly/cWkphRB

Are there so few things to do at the telegraph this was a priority?

What is this cheapening of Isaiah Berlin?

Job Opportunity!

University of Zurich
Faculty of Arts and Social Sciences

Professorship for Africa (open rank)

Focus: Sub-Saharan African societies and their complex relations with Europe and the world, late 19th-early 21st century.

networks.h-net.org/jobs/69543/u...

An Audit of Social Science Survey Experiments Abstract. Survey experiments have become a popular methodology for causal inference across the social sciences. We study the efficacy of survey experiment

Needed - larger samples, more realism about (the lack of) heterogeneous treatment effects:
-"less than a third of proposed hypotheses were supported... the largest predictor of positive exp. results was sample size"
-"moderation hypotheses were rarely significant"
academic.oup.com/poq/advance-...

🎉 New publication 🎉 Why do #youngpeople vote for the #AfD? This question has kept our research project busy for quite some time, so I'm beyond excited that our first article - co-authored with @timonscheuer.bsky.social - is now out in #GermanPolitics! 🤩 www.tandfonline.com/doi/full/10....

Decades of doing politics without ideas, meta narratives etc …

NYT: “Epstein emails a distraction from the Republican shutdown victory”

WSJ: “We estimate the Epstein files to be 67% Trump by weight”

You could have said that since Francis Fukuyama wrote the End of History

"This assault on higher education is best understood as a means of destroying a locus of political opposition."

The European Far-Right Politics of Truth in Digital Spaces Joint Sessions of Workshops, 7 – 10 April 2026, University of Innsbruck

⚙️ #ecprjs26 Workshop directors 🎬 @beatrizlbuarque.bsky.social & @scopelliti.bsky.social
💡 Seeks Papers exploring the European far-right politics of truth in digital spaces buff.ly/7Zm46uu
✔️ @ecpr-ead.bsky.social endorsed
⌛ Submit by 10 Dec

#CallforPapers #AI #PoliticalParties #Policymaking

What is the most profitable industry in the world, this side of the law? Not oil, not IT, not pharma.

It's *scientific publishing*.

We call this the Drain of Scientific Publishing.

Paper: arxiv.org/abs/2511.04820
Background: doi.org/10.1162/qss_...

Thread @markhanson.fediscience.org.ap.brid.gy 👇

The race to churn out papers is a systemic problem.

Early career scholars are desperate to get more papers to compete in the academic job market. This can make it hard for faculty mentors hard to reduce their output unless they shrink their lab (which removes opportunities from next generation).

Strategies of Non-Populist Parties Why most of them are wrong and how they can do better

I don’t agree with everything, but this is a very thoughtful and, indeed, thought-provoking post
open.substack.com/pub/laurenzg...

But can you profit from it?

Anthropologists have discovered a lost tribe that uses the "one, two, many" system of counting.

A two-headed coin.
Isn’t it the perfect symbol of Trumpian politics?
Heads, I win. Heads, I win. Again.
Yes, the odds are forever stacked against you!

Here's a law professor lamenting the fads in legal theory popular among law professors that become popular for a decade or two and then fade away— written in 1950. Specifically, it's Roscoe Pound, reflecting on trends since he became a lawyer in 1890.
jle.aals.org/cgi/viewcont...

free markets, amirite?

I’m reminded of Converse … there is a lot of knowledge missing if Australia is an attractive model for the UK - what vast supply of minerals does the UK have?