I’m still mad about this. The results of that paper matter to me, it affects some decisions Open Philanthropy (on the margin, but still). If economics as a discipline wants people to listen to evidence, it needs to hold up its side of the bargain and update or retract major errors.

As I've said before I think many psychologists too often make universalist claims without taking time to evaluate the ethnohistorical evidence, but this seems like the opposite sort of error of substantially over-extrapolating from ethnographic tidbits taken out proper context.

Now that I no longer have the option of logging in thru a university's library, I feel non-academics' pain more fully and am *begging you all* to please post non-paywalled versions of your academic papers on your websites.

Interesting result finding that you basically can’t measure the statistical effects of policy differences across US states unless they have huge effects: 50 states just aren’t enough.

My blog post critical of this re-analysis is up: matthewbjane.com/blog-posts/blo… The re-analysis conducted by Rausch & Haidt does not follow a principled statistical approach. I re-do their re-analysis with appropriate statistical methods

Introduction: I conclusively show Body: I tentatively explore Conclusion : I apologise

Great news that the RSV vaccine is being rolled out for pregnant women and the elderly. For context, RSV is responsible for around 20,000 infant hospitalizations in England per year. Vaccine efficacy in reducing severe RSV is around 80%. bbc.co.uk/news/articles/…

Re-reading Lundberg et al. which I always assign in the first week of intro methods. It's so polite, but its core point amounts to the most savage of critiques: studies should be very precise about the question they are trying to answer, but that usually doesn't happen.

Submit your abstracts to the 7th Perspectives on Scientific Error conference, and join us in Bern on February 5-7, 2025! Keynotes by Saloni, Stephanie M. Lee, and Liam Bright

TBC, if you are using data to make a claim you are making a statistical inference. The fact that you haven’t calculated anything formally, just makes it more likely it’s a bad one.

It’s kind of wild to me that, as of two weeks ago, there are now 3 large cluster randomized trials across five countries in Sub-Saharan Africa showing that giving every kid antibiotics reduces child mortality by ~14% and I don’t see anyone talking about it!

OK, well I guess it's as good a time as any to announce I'm looking for a postdoc at Princeton -- areas are politics, machine learning and statistics. Please share widely! puwebp.princeton.edu/AcadHire/apply…

Literal footage of a random effects meta-analysis

Berlin tweeps: anyone know good places to get an A0 poster printed? Preferably near the Mitte area!