Also this week: missing Jane Lubchenco, Gilbert & Sullivan vs. digital humanities, the mysterious behavioral ecology of joggers, nerd sniping Stephen Heard, and more.
The Trump administration is nominating AccuWeather CEO Barry Myers to run NOAA. Sigh. I remember when Jane Lubchenco ran NOAA. So does Jane Lubchenco.
The right way to argue to national governments that they should support basic research. Canadian, but I think it generalizes.
25% of US STEM grad students at R1 universities lose interest in an academic career during grad school; only 5% become more interested. Interestingly, people who lose interest mostly do not do so because of perceptions of lack of tenure-track job availability, even though awareness of the tight faculty job market increases during graduate school. Rather, they mostly lose interest because they become less interested in research, especially basic research. Another tidbit of interest: men are about 8-10 percentage points more likely than women to start grad school with an interest in an academic career, and that gap persists through grad school. Related: my old post on protecting grad students from their own optimism. (ht @noahpinion)
A very interesting remark from Andrew Gelman: pseudoscience (or bad science; call it what you will) is becoming harder to distinguish from actual science (or good science)–but only at the cost of addressing trivial topics rather than important ones with the potential to fundamentally alter our understanding of the universe. It used to be ESP and perpetual motion machines. Now, says Gelman, it’s bad social psychology. Like some of Gelman’s commenters, I wonder if his Gelman isn’t overgeneralizing from the bits of social science he thinks about most and forgetting about stuff like anti-vax. And I wonder about his fudging of the difference between pseudoscience/bad science that attracts many scientists (like ESP, back in the day), and pseudoscience/bad science that’s mostly only attractive to non-scientists (like anti-vax). But as we’ve discussed in old posts, there aren’t any clear bright lines to be drawn here (see here and here).
Speaking of Andrew Gelman and social psychology: the NYTimes magazine has a feature on how the rapidly-changing research landscape in social psychology (push for replicability, etc.) has affected one of its highest-profile researchers, Amy Cuddy. Andrew Gelman gets a look-in as one of the most prominent critics of Cuddy’s work. Amazing (and very fair) piece, weaves together a lot of stuff we’ve talked about in the past. I’ll also point you to this old post of mine on some of the same issues. Here’s Andrew Gelman’s response to the piece, which attracted a wide range of comments.
Semi-related: Developmental economist Chris Blattman asks what will the revolution be in economics, analogous to the “non-preregistered, small-N experiments are uninformative” revolution in social psychology. Related post from Brian on whether ecology is having (or will shortly have) a scientific crisis, and if so, what kind.
Sticking with Chris Blattman, he used to have a blog that was widely read in his field. Here’s why he stopped blogging (and why he might start again one day).
Machine learning program with no preprogrammed knowledge of Go except the rules of the game, and no access to any Go games except those it played against itself, taught itself to become easily the best Go player ever (human or computer) in three days. (ht a correspondent)
How Canadians (well, some of them) decided what college to attend 30 years ago. An interesting artifact from the Time Before The Intertubes. I predict that Stephen Heard will waste the rest of the day trying to track down a copy of the book discussed in the link. 🙂
I think my research falls in the lower left of this graph. Fortunately. 🙂
“I am the very model of a modern
major general digital humanist“. This is hilarious. A sample:
I know my methods irritate established new historicists
Provoking them to write Op-Eds dyspeptic and quite moralist
But though they write my coding off as empty and reductionist
I’d rather read more Python than some bloody deconstructionist
Ecological perspective on jogging:
Brian Martinson argues in this week’s Nature that every researcher should have a lifetime limit on the number of words they can publish. He acknowledges it would create different perverse incentives, but suggests they would be better than the perverse incentives of the current system. He broadly argues it would encourage replacing quantity with quality. He is most concerned about research quality (e.g. fewer errors), but I think it would apply to a lot of dimensions of quality. It will never happen, but I wish it would!
Hoisted from the comments:
My post on checking your causal inference methods by testing ridiculous causal hypotheses led to a great discussion, including a comment from the author of the paper that inspired the post.
The comment thread on our AUA about the research contributions of ecologists from developing countries is great too. Features numerous comments from researchers from Brazil. Touches on everything from the roles of local vs. international journals, to imposter syndrome, to detailed discussions of the particular challenges faced by researchers in particular countries.