Friday links: Epstein fallout continues at Harvard, memes vs. intro biostats, and more (includes quick poll) (UPDATED)

Also this week: is ecological integrity a thing, when experts lie to the public, Stephen Hawking as self-promoter, is the American Sociological Association collapsing, many analysts vs. one dataset, Alan Turing vs. money, and more.

From Jeremy:

Harvard University sanctions evolutionary biologist Martin Nowak over his ties to convicted sex offender Jeffrey Epstein. Nowak cannot serve as a PI on any grants or contracts, or take on any new advisees, for at least two years. The Program on Evolutionary Dynamics, which Nowak directs, will be shut down. Nowak can resume teaching undergraduate courses. He has been on paid administrative leave for the past year, in the wake of a Harvard investigation into his contacts with Epstein.

The American Sociological Association is collapsing, according to one of its elected officers.

Another many analysts vs. one dataset project, this one from microeconomics. This one finds huge variation among analysts in data preparation and analysis decisions, many of which wouldn’t ordinarily be reported in the methods section. No two analysts reported the same sample size (!), and not only the statistical significance but even the sign of the focal parameter estimate varied among analysts. As the authors note, this isn’t just a “p-hacking” or “garden of forking paths” issue. The issue is that different, reasonable-seeming analyses lead to different conclusions. That would still be true even if all analyses were preregistered. I think we’re just beginning to find out how big of an issue this is.

Which brings me to the reason experts should be more reluctant to lie to the public: They aren’t experts on the topic of when to lie.” Also addresses the argument that experts should refrain from publicly criticizing the work of others because their criticisms will be “weaponized” by bad actors. See also this related old post from Brian.

Straight into my veins. See also.

In a new paper, Rohwer and Marris argue that there’s no such thing as “ecological integrity”. I agree. Do you? Take the poll:

Writing in Science, Declan Fahy reviews a new, and critical, biography of Stephen Hawking.

Also writing in Science, W. Patrick McCray reviews a new documentary about Stewart Brand.

Using machine learning and data compression algorithms to identify novelist Elena Ferrante. I don’t know anything about this topic or about Elena Ferrante. So I found this intriguing, but have no idea what to think about it.

Pandemic scientists fight burnout. I was talking recently to someone who switched to working on Covid-19 about one factor (among many) driving the burnout: all the time and effort it takes to sort through Covid-19 preprints. This isn’t a criticism of preprints, it’s just an observation.

Xiaoyang Song and Richard Corlett have retracted their meta-analysis of density-dependent mortality in tree seeds and seedlings, after they were made aware of a statistical error that altered the conclusions of the paper. Kudos to the authors for doing the right thing. Mistakes happen in science.

Proceedings B has published Expressions of Concern for two papers by ecologist Denon Start. Concerns have been raised about the data underpinning the papers and investigations are underway. See here and here for the EoCs, and here for a PubPeer thread on one of the papers (there is no PubPeer thread on the other). Denon Start recently had another paper corrected (it’s a rather unusual correction), and had two early career awards rescinded for reasons that have not been made public.

UPDATE: Just found this and decided not to wait until next week with it. Denon Start also has three more new EoCs. They’re all from Am Nat. They’re for failure to comply with the journal’s mandatory data sharing policy, and for apparent statistical and graphing mistakes that Denon Start has yet to correct.

The new UK 50 pound note features Alan Turing and a bunch of cool details.

Rutgers University will require students to get vaccinated for Covid. I wonder how many universities will follow their lead, either in the US or elsewhere.

I mostly avoid linking to or commenting on controversies like the recent one about John Ioannidis. But in the unlikely event that you want to read something further about it, Tal Yarkoni’s remarks are thoughtful.

A famous remark of Fisher’s, now in convenient meme form. 🙂

Speaking of stats memes: here’s intro biostats in meme form. These are great. 🙂

8 thoughts on “Friday links: Epstein fallout continues at Harvard, memes vs. intro biostats, and more (includes quick poll) (UPDATED)

  1. Thanks for putting up that link to the hypothesis paper – I’m seeing lots of implications of the lack of training in strong inference thinking where I work at the intersection between environmental impact assessments and marine ecology.

  2. Only 31 votes in the poll so far, but “no there’s no such thing as ecological integrity” has 19 of them–pretty solid majority. “yes there is such a thing as ecological integrity” only has 5 votes.

    Not a random sample of ecologists, of course, or even of our readers. And it’s possible that readers of this blog are particularly likely to share my skepticism of vaguely defined concepts like “ecological integrity” (or “ecosystem health”). I’d be very curious to see a much bigger poll of a more representative sample of ecologists.

    Would also be curious what ecologists *think* the responses to that bigger poll would be. Are ecologists who believe that “ecological integrity” is a thing aware that many other ecologists disagree?

  3. For those less familiar with Ioannidis’ work (like me) one could replace ‘Ioannidis’ with ‘Dawkins’ and Tal Yarkoni’s Twitter thread would be just as good.

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.