Friday links: evidence vs. scientific reforms, p-values vs. speed limits, Delft Daphnia, and more

Also this week: off-label stats, absence of evidence = evidence of absence, and more.

From Jeremy:

When does absence of evidence constitute evidence of absence? The answer is straightforward, I think, but I still found the illustrative examples interesting.

Statistician Daniel Lakens doesn’t expect to be convinced by data that scientific reform is working (and is fine with that). That is, he thinks that some scientific reforms are justified by “logical principles,” and don’t need any empirical evidence that they work. Statistician Jessica Hullman disagrees.

An arbitrary-but-widely-agreed threshold for statistical significance is useful for the same reason that speed limits on roads are useful. I agree with this and have been teaching this point to undergrads for years, though the analogy to speed limits hadn’t occurred to me.

The US ORI has updated its research misconduct policy for the first time in 20 years.

The inflation-adjusted net cost of US higher education (so, tuition+fees+room+board, minus grants, scholarships, and other discounts) has gone down 11-13% over the past five years.

How to use the Kolmogorov-Smirnov test to bound the fraction of subjects in a between-subject experimental design who exhibited a response to the treatment. This was interesting and new to me.

A theory that the optimal level of shitposting on social media is low but not zero. I am not on social media, and my choice not to be is independent of the amount of shitposting, so I leave this theory to others to evaluate.

Philosophizing about mortality is mostly just cope, says philosopher Helen de Cruz. I’m increasingly coming round to the view that many things are mostly just cope…

The peer-reviewed literature on “degrowth” almost entirely comprises opinion papers that lack any data analysis or theoretical modeling. The linked piece goes on to argue the few non-opinion papers aren’t very good. I found it pretty convincing, but then I’m not a degrowther. As best I can tell from my very limited exposure, the literature on degrowth does seem to be aimed at reviewers and readers who are already degrowthers.

And finally, here’s a company that lets you use AI to generate Delft-style ceramic tiles, which they will make and ship to you. Now I know what to get Meghan for Christmas:

Delft-style ceramic tile design in response to the AI image generator prompt “Daphnia”.

🙂

Hot Topics

Related Articles