iiinova — where your digital self keeps evolving

Reposted by Jon Mellon

WeRateDAGs @weratedags.com · 10/09/2026

Economists would declare that rainfall is already random, then go play outside their discipline. Which is a great moment to give 13/10 to the DAG o'Mellon for its child-friendly iv-assumptions & huge cast of possible treatment variables From @jonmellon.bsky.social (2024) 'Rain, Rain, Go away' AJPS

Rainfall: the uncaused cause the Medievals were looking for. An instrument for the ages.

021

Reposted by Jon Mellon

John V. Kane @uptonorwell.bsky.social · 31/08/2026

Compared to "significant effects," null results lead citizens to: --View study as less important --Ask for less info --Be less willing to share it --Think journalists would be less likely to cover the study --See the study as being lower quality & researchers as less competent (😬)

284

Reposted by Jon Mellon

John V. Kane @uptonorwell.bsky.social · 31/08/2026

🧐 What do citizens think about "null results" in scientific research? To find out, @johnholbein1.bsky.social & I fielded a large-scale online experiment. We gave respondents straightforward info about a study, asked them some Qs, showed them the results, then asked them Qs again.

53617

Reposted by Jon Mellon

The British Election Study @britishelectionstudy.com · 20/08/2026

📢 New Data 📢 We are pleased to announce the release of Wave 31 of the British Election Study Internet Panel. You can find out more information, including where to download the data, here: www.britishelectionstudy.com/uncategorize...

britishelectionstudy.com

Release note for Wave 31 of the British Election Study Internet Panel - The British Election Study

298

Jon Mellon @jonmellon.bsky.social · 15/08/2026

People have asked whether the 1.4 pt average overestimate of left wing parties that we find matters. Well that enough to move the democrats senate chances from realistic to long shot according to Nate Silver’s latest model

110

Reposted by Jon Mellon

Chris Prosser @caprosser.com · 05/08/2026

Lots more analysis in the paper, breaking it down over time and by country, so take a look! osf.io/preprints/so...

osf.io

OSF

121

Reposted by Jon Mellon

Chris Prosser @caprosser.com · 05/08/2026

Obviously, 'on average' doesn't mean polls are always-and-everywhere biased to the left - there is a lot of heterogeneity between elections - but elections are much more likely to have a leftward polling bias than a rightward one.

121

Reposted by Jon Mellon

Chris Prosser @caprosser.com · 05/08/2026

Using ~8.5k election-poll-party observations from 372 elections in 32 countries, we find a persistent partisan asymmetry: polls tend to overestimate the left relative to the right. On average across elections, polls the month before an election overstated the left by about 1.4 points

3128

Reposted by Jon Mellon

Chris Prosser @caprosser.com · 05/08/2026

New working paper from Stuart Perrett, @drjennings.bsky.social , @jonmellon.bsky.social , @cbwlezien.bsky.social and me: Do polls underestimate support for right-wing parties? Assessing variation in polling error by party family osf.io/preprints/so...

osf.io

OSF

33221

Reposted by Jon Mellon

Chris Prosser @caprosser.com · 29/07/2026

No doubt you’ll all be rushing to pre-order the hardcover for the bargain price of £95 (😬), but should that prove a little steep, we’re pleased to say that the electronic version will be available open access, i.e. free!

183

Reposted by Jon Mellon

Chris Prosser @caprosser.com · 29/07/2026

It’s not out for another six months or so, but *Electoral Realignment: How Brexit Reshaped British Voting Behaviour*, my book with Ed Fieldhouse, @profjanegreen.bsky.social , Geoff Evans, @jack-bailey.co.uk and @jonmellon.bsky.social, is up on the OUP website! global.oup.com/academic/pro...

Cover of *Electoral Realignment: How Brexit Reshaped British Voting Behaviour*
By Edward Fieldhouse, Jane Green, Christopher Prosser, Geoffrey Evans, Jack Bailey, and Jonathan Mellon

13013

Jon Mellon @jonmellon.bsky.social · 11/07/2026

Followup question from my 3 year old: how many windows are there in the world?

080

Jon Mellon @jonmellon.bsky.social · 06/07/2026

There’s going to be 10 minutes of additional time isn’t there?

210

Jon Mellon @jonmellon.bsky.social · 05/07/2026

This study only varies gender and not AI use. The headline results are compatible with a pure gender effect on competence judgments with no specific AI effect.

4387

Reposted by Jon Mellon

APSA Preprints @apsa-preprints.bsky.social · 29/05/2026

The APSA Pres Task Force on AI, Politics, & Political Science's report comes in the form of an edited volume identifying questions & establishing a foundation for the empirical study of how politics & governance are affected by AI. Check out these early chapter drafts: shorturl.at/cMZzI #polisky

42014

Jon Mellon @jonmellon.bsky.social · 24/05/2026

An idea I’ve been pondering is whether the academic writing community needs a pressure valve for AI: slopXiv. Institutions that presuppose human effort are in danger of getting overwhelmed but I think it’s naive to think there’s any stopping the use of LLMs

160

Jon Mellon @jonmellon.bsky.social · 15/05/2026

GPT giving model recommendations on the same scale as faculty recommendation letters

010

Reposted by Jon Mellon

James Breckwoldt @jamesbreckwoldt.bsky.social · 12/05/2026

New devastatingly incisive, ruthlessly evidence-based, exquisitely nuanced, intellectually fearless and strategically indispensable analysis from me on how mainstream parties can win back the voters they’ve lost. Different parties require different responses.

jamesbreckwoldt.substack.com

How mainstream parties can win back voters

Different parties require different responses

0121

Reposted by Jon Mellon

Ryan Briggs @ryancbriggs.net · 11/05/2026

We have a new version of this paper out. The headline results are the same—political science must filter results heavily for statistical significance—but we've added many extensions and rewritten much of it in response to feedback (thank you!). A quick thread on updates 👇

26819

Jon Mellon @jonmellon.bsky.social · 11/05/2026

This was a key point that came up in feedback. People have the intuition that nulls are a skill issue. This can be true but there’s better solutions than selecting on significance

110

Reposted by Jon Mellon

Ryan Briggs @ryancbriggs.net · 11/05/2026

3. People that did experiments also told us that they thought that null results were often indicative of a failure to manipulate. We worked out a little model with Bayesian updating to show that this point is mostly misguided. If you want evidence of dosage, you really should collect it directly.

We can now address the question that many researchers implicitly pose when faced with a null result:
does a small |𝑧| constitute evidence of failed delivery? The intuition that motivates this question is partly
correct. Indeed, a large |𝑧| in a plausible direction does shift posterior mass toward higher 𝑑, because
Pr(𝑧 ∣ 𝑇 = 1, 𝐷 = 𝑑) assigns more probability to extreme primary statistics when dose is high.
However, two structural features limit how far this update goes.
Consider the posterior probability that the dose was properly administered, when we have no auxiliary
evidence. That probability is proportional to the distribution of the test statistic 𝑧, weighted by the prior
probability of each dose level:
Pr(𝑑 ∣ 𝑧) ∝ Pr(𝑧 ∣ 𝐷 = 𝑑) ⋅ Pr(𝑑).
The challenge is that, since the treatment effect 𝑇 is unknown, the likelihood of 𝑧 is a mixture of two
components: with and without a true effect.
Pr(𝑧 ∣ 𝐷 = 𝑑) = Pr(𝑇 = 1) ⋅ Pr(𝑧 ∣ 𝑇 = 1, 𝐷 = 𝑑) + Pr(𝑇 = 0) ⋅ Pr(𝑧 ∣ 𝑇 = 0).
Neither component is very helpful for pinning down 𝑑. Under 𝑇 = 1, the likelihood does depend on 𝑑,
but only through the same product ambiguity noted above: any combination of effect size and delivery
yielding the same product is observationally equivalent. Under 𝑇 = 0, the likelihood does not depend
on 𝑑 at all, so it contributes nothing to dose inference. As a result, even extreme 𝑧 statistics can lead to
only modest updating about 𝑑. The intuition that “a null result implies weak delivery” is not wrong per
se, but it is prior-sensitive and noisy.

151

Jon Mellon @jonmellon.bsky.social · 09/05/2026

One thing I really like about agentic coding is how much it reduces the effort to make a silly idea reality. In my case, all the addition quizzes for preschoolers online are really scammy. So I vibe coded one www.jonathanmellon.com/addition-quiz/

3181

Jon Mellon @jonmellon.bsky.social · 06/05/2026

The reasoning traces of LLMs are sometimes a little eery

010

Jon Mellon @jonmellon.bsky.social · 03/05/2026

My 3 year old asked me “how many people are outside right now?” and was indignant when I said I couldn’t look up the answer

2211

Reposted by Jon Mellon

Greer Mellon @greermellon.bsky.social · 02/05/2026

2-K is coming to NYC! Here’s our entry to the #NYC2KJingle contest!

041

Reposted by Jon Mellon

Andy Lawton @politics.andylawton.com · 08/04/2026

I love the message of "just look with your robot eyes", the frustrations of dealing with LLMs.

121

Jon Mellon @jonmellon.bsky.social · 08/04/2026

A problem I've encountered a bunch of times is when geographic data (such as ward boundaries) is only presented in a pdf map. I got chatGPT to extract that into usable shapefiles. Chat log here if anyone wants to extract this into a systematic workflow chatgpt.com/share/69d654...

261

Jon Mellon @jonmellon.bsky.social · 08/04/2026

Reading an older baby naming book with some interesting suggestions

010

Reposted by Jon Mellon

Peter Tennant @pwgtennant.bsky.social · 05/04/2026

Academics should still be forced to explain what decision would theoretically be informed their estimand! The culture of producing meaningless associations under the guise of 'interest' is extremely wasteful.

351

Reposted by Jon Mellon

Jon Mellon @jonmellon.bsky.social · 05/04/2026

Whereas in academia every chain of questions like that ultimately grounds out in “because that’s an interesting thing to know about the world” which isn’t a motivation that narrows the estimands

111

Reposted by Jon Mellon

Jon Mellon @jonmellon.bsky.social · 05/04/2026

That answer implies that the original estimand was probably not right (what we’re actually wanting to do is a predictive exercise)

111

Reposted by Jon Mellon

Jon Mellon @jonmellon.bsky.social · 05/04/2026

Concrete example. Talked to someone who was wanting to estimate the ATE of a particular user action on long term engagement. I asked “why do you want to know this?” And they said “so that stakeholders know which short term metrics to try and move”

101

Jon Mellon @jonmellon.bsky.social · 05/04/2026

An interesting difference between academic and industry data science is that there is usually a correct answer to what to the estimand *should* be in industry. This is because industry analysis needs to ultimately drive a decision whereas academic analysis has to be interesting to a community

3183

Jon Mellon @jonmellon.bsky.social · 04/04/2026

Had an agent just start writing a CSV by hand today (unclear what relation the data had to reality) when the function it wrote didn’t work correctly in its environment. Stay safe out there!

151

Reposted by Jon Mellon

James Breckwoldt @jamesbreckwoldt.bsky.social · 30/03/2026

Why is use of AI and LLMs so divisive for academics? I think it's because ~half of them vote centre-left and ~half of them vote environmental-left When I did a nat rep survey in the UK, these two had opposite views of economic growth and tech optimism jamesbreckwoldt.substack.com/p/what-do-th...

3107

Reposted by Jon Mellon

Kevin Collins @kwcollins.bsky.social · 29/03/2026

A modest-sized but truly random sample of opinions is a better measure of opinions in the population than an enormous but self-selected sample of opinions from the same population

211618

Jon Mellon @jonmellon.bsky.social · 22/03/2026

Some of these are quite popular but * military analysis: Anders Puck Nielsen * astrophysics: Dr Becky, Cool Worlds * AI/machine learning: Welch Labs * environmental science: Simon Clark * beautiful coding projects: Sebastian Lague * piano tutorials: mangold project * NFL analysis: Brett Kollman

050

Reposted by Jon Mellon

Jon Mellon @jonmellon.bsky.social · 21/03/2026

This is very annoying as an American citizen as the US also gets grumpy about traveling through its border on a non-US passport. so I’ll have to figure out the right sequence of passports to show not to hurt any government’s feelings

102

Jon Mellon @jonmellon.bsky.social · 09/03/2026

Interesting that the policy ends up coming down to "at least paraphrase what the AI wrote". I think there's a lot to be said for this to make sure at least one human brain paid enough attention to rewrite what the AI said. But fascinating how quickly we're having to grapple with all of this

051

Reposted by Jon Mellon

EconStor @econstor.bsky.social · 05/03/2026

The most downloaded paper on EconStor in Feb. 2026 was: "Briggs, Ryan C.; Mellon, Jonathan; Arel-Bundock, Vincent (2026) : It must be very hard to publish null results, I4R Discussion Paper Series, No. 281, Institute for Replication (I4R), s.l." hdl.handle.net/10419/336819 @i4replication.bsky.social

0166

Reposted by Jon Mellon

Yamil Ricardo Velez @yamilrvelez.bsky.social · 09/03/2026

Conditionally accepted at the APSR (w/ @scottclifford.bsky.social & @patrickpliu.bsky.social): Why does political information so often change beliefs but NOT attitudes? We highlight the role of belief relevance, or the extent to which beliefs bear on attitudes.

512635

Reposted by Jon Mellon

Philip N Cohen @philipncohen.com · 07/03/2026

This proposes a way of using AI agents to produce research. Ok. But this bit is a pipe dream: "And human scientists should retain authority over — and responsibility for — framing the question, validating the path and signing off on conclusions." Here's why... /1

32912

Reposted by Jon Mellon

Ryan Briggs @ryancbriggs.net · 06/03/2026

this but for scientists who have p-hacked

0201

Jon Mellon @jonmellon.bsky.social · 05/03/2026

I really do mean every day too

2220

Reposted by Jon Mellon

Chris Prosser @caprosser.com · 05/03/2026

This undersells the fact that you would sometimes do zoom meetings during the kayak!

5232

Jon Mellon @jonmellon.bsky.social · 05/03/2026

I kayaked to work every day for 3 years

4914

Jon Mellon @jonmellon.bsky.social · 04/03/2026

The main thing I think AI is going to do is generically put pressure on all of these institutions. That opens up space for changes in general including ones that might not be directly tied to AI.

120

Jon Mellon @jonmellon.bsky.social · 28/02/2026

Something tells me this very legitimate agent may not have read my book

120