Also this week: another social science journal bans p-values, figure aspect ratio vs. your data, JBS Haldane vs. Hannibal Lecter, and more.
Want to know what new Am Nat EiC Dan Bolnick thinks about “gatekeepers” and scientific publishing in the digital age? Here you go! I think it’s great for an EiC to spell out his thinking at such length.
The journal Political Analysis is banning p-values. Their reasons for doing so are:
- an isolated p-value is not evidence for a model or the associated scientific hypothesis
- lots of people have written lots of papers criticizing p-values
- relying on p-values leads to publication bias because non-significant results go unpublished
- banning p-values will stop p-hacking
- many social scientists misunderstand p-values
All of those reasons are pretty debatable to my mind, especially in the absence of any discussion of the alternatives. For instance, if you think p-values are widely criticized and misunderstood, wait ’til you see how widely criticized and misunderstood Bayesian posterior probabilities are! No statistical procedure works at scale. And if you think there’s no Bayesian equivalent of p-hacking, think again.
Sticking with Political Analysis, they’re switching from double-blind review to single-blind, because
banning p-values wasn’t controversial enough on its own it’s taking too many editorial staff resources to check for and correct author-identifying information in submissions. To my mind, that’s an argument for Am Nat’s approach of just leaving author names and institutions off the title page. That’s arguably better than nothing; reviewers won’t always recognize the authors. And hopefully leaving off the author names and institutions reminds reviewers who do recognize the authors to focus on evaluating the science. Thereby hopefully preventing perverse outcomes that could arise from blinding being seen through non-randomly with respect to author attributes.
How Pittsburgh illustrates the best playbook to revive old industrial cities with good universities: build a tech cluster around the university, welcome immigrants, fund the arts, use redevelopment to create a few nice neighborhoods, and go from there. What say you, Cleveland? (home of Case Western Reserve)
And finally, Stephen Heard imagines Valentine’s Day in the Ediacaran. 🙂