Stephen Wild
@stephenjwild.bsky.social
I try to put straight lines through things but usually fail. Try to be Bayesian when I can. Views my own. RT/like != endorsement.
Completely agree with this post. When people "check" parallel trends, they are implicitly imposing a prior belief that parallel pre-period trends imply parallel post-period trends. Motivated by this insight we developed a Bayesian sensitivty analysis method: arxiv.org/pdf/2508.02970
Now *this* is fun methodsmatter.substack.com/p/why-are-yo...
Why are your pre-trends always parallel? A question for the economics profession
A pretty pre-trends graph may be consistent with a biased DID estimator. What do these graphs actually teach us, and what does requiring them for publication encourage DID researchers to do?
methodsmatter.substack.com
I think much of this applies more generally to many types of privileged information, though maybe the manipulation piece is more central to classified information specifically. I meet very few people who, for epistemic hygiene reasons, prefer not to gain access to privileged information. #linklog
Daniel Ellsberg on the Limits of Knowledge
Why the president doesn't necessarily know any more than you do.
motherjones.com
Sean Connery initially wore a wig with a pony tail when he started filming The Hunt For Red October.
"Contemporary Simpsons" is a myth made up to scare children. Everyone knows that after Phil Hartman's death just after recording season 10 episode 3, they stopped making the show as a sign of respect. Anyone claiming to have seen an episode made after that is just trying to hurt you.
Michaela Joffe@joffeorama.bsky.social · 19h ago"did you here about the thing that happened in contemporary simpsons?" no I was busy doing anything. sex, art, crying, something good, something bad, nothing, it doesn't matter. the important thing was it kept me from knowing about contemporary simpsons. I don't know why you'd want to rob me of that
Canadians: who are some good coffee microroasters I should try ordering from? Which coffee beans do you recommend/are your favorite?
(I use eg set.seed(2026-11-19) which also notes the date and shows I didn't try a lot of seeds to get the results I wanted)
Liam did a version of the XKCD comic (Original: xkcd.com/386/)
@lastpositivist.bsky.social · 2d agoSome people were making me mad on the internet so I wrote a blog post about it. On the different reasons for people have for being left wing and how people manage the psychological tensions of the fact that they come apart. sootyempiric.blogspot.com/2026/10/the-...
As we approach Halloween...it's time for some Treehouse of Horror
The Simpsons Treehouse of Horror: Join Us, Father
Alt: The Simpsons Treehouse of Horror: Join Us, Father
static.klipy.com
something i've found in my own work (manuscript out for review, naturally!) is that audience competition online inevitably turns users towards a kind of exclusionary petty extremism meant to shore up their uncertain social status in fragile online communities
SE Gyges@segyges.bsky.social · 4d agothe number of people i see expressing the desire that AI should be conscious so they can torture it convinces me that 1) they're all sheep who will swallow whatever swill jreg sells 2) they are deliberately performing evil behavior to establish ingroup solidarity, as is a classic in gangs
Me: "It can't be that ridiculous, can it?" Me, a short later: "of course it's that ridiculous."
New paper 🚨: Is there a democratic advantage in war? 30 years of research say yes. We reevaluate these claims with new data on wars and democracy and find very little evidence to support the claim that democracies win wars at (much) higher rates than non-democracies Paper: doi.org/10.31235/osf...
The World is Not Enough For Tax Purposes
Dan Davies@dsquareddigest.bsky.social · 5d agoYou only die twice for tax purposes
Crocodile Dundee for Tax Purposes
David K Butler@davidkbutler.bsky.social · 5d agoKnives Out for Tax Purposes
Fun with geo-located respondents. The lack of detail about the models and demographics--and the absence of robustness tests--makes me skeptical. Up next, 83 page Economist articles that are 5 years out of date
Matt Grossmann@mattgrossmann.bsky.social · 5d agoVoters living near data centers are moving further toward Democrats economist.com/graphic-deta...
"Protect your brain from too many words"! Not a lot of people I talk to actually acknowledge that reading a ton is very effortful!
Just get AI for the washing machine and let it reply to emails for you smh
Jess Calarco@jessicacalarco.com · 5d agoSo, if you're waiting on an email from me, blame the washing machine : )
Many such cases. See also the question "WhY dOn'T tHeY mAkE aNyThInG lIkE tHeY uSeD tO?"
Answer David, nerds! (More detail in the thread)
David Bartram@dbartram.bsky.social · 2w agoQuestion for stats/quants people, re bootstrap standard errors in the context of generalized synthetic control (Xu 2017). The approach is: parametric bootstrap via resampling the residuals (in a series of leave-one-out runs), to get the “conditional variance of the ATT estimator”.
