TIGHT PRIORS CAN LEAD YOU DOWN RABBIT HOLES.

Two days ago, I wrote a long post suggesting that right-populist polemicists might have read too much into a climate scientist's response to Nature editors suggesting he and his colleagues narrow the focus of the paper.  That's the way the research enterprise works.

A simple confirmation of existing research with more recent data strikes me as a Minimal Publishable Unit for an archival journal, not necessarily the best use of space in Nature; whilst adding additional explanatory variables to capture those other effects might take more time off the tenure clock, and be hard to impound in ceteris paribus in any event.  Those considerations might be on the minds of the editors and the referees.

That post concluded,

I remind you, dear reader, that a paper challenging an existing paradigm has to be tightly argued, and it might have to be more convincing than the average Minimal Publishable Unit.

The simpler explanation might be that the accepted Nature article was more tightly framed than the more ambitious paper Nature turned down whilst in PLOS One accepted it.

Let me also reiterate an economics lesson. Journal space is a scarce resource, and, while journal editors might decry the lack of good submissions, the fact that rejection rates, particularly at the journals that get read, are high, suggests that shorter, tightly focussed papers are what get in.  And nothing concentrates the mind quite like that letter from The Review of Economics and Statistics that reads "If you can address these comments and shorten the manuscript, we will be happy to reconsider."

The polemicists, though, continue to embrace the "Corrupt Peer Review" hypothesis.

"Distinguished scientists who think that global warming is a problem but not a “crisis” get ostracised, cancelled or rejected by peer reviewers. Meanwhile, even the most trivial study that comes to an alarmist conclusion – such as a notorious one that found fish behaviour to be affected by carbon dioxide – gets rushed into print and celebrated in the media."

That was Matt Ridley, and he quoted a long-form x-twitter post (don't we have Blogger or Substack or Word Press?) from Patrick Brown, who is asking the polemicists to stand down.

Some outlets are reporting that I admitted to "manipulating data" or some such thing. This is totally false. Please read the actual text of my piece to understand the nuance of the argument I am making.

What follows is an explanation of the way scholarly publication works that ... squares with what I recall from years ago.

Overall, the point I am making is that the easiest pathway to a high-impact publication in this field is to focus on and highlight a climate change impact (even when it is just one driver in a complex, multi-causal system).

I used my paper as an example: My paper focused exclusively on a climate change impact, and it was published in a high-profile venue.

The editor-in-chief of Nature, Dr. Magdalena Skipper, has argued against my thesis by saying that reviewers asked us to account for other factors and I argued against it. That's not really true. A reviewer asked us about accounting for another climate variable (absolute humidity), and I voluntarily brought up other non-climatic factors in my response.

Besides, my claim is not that you don’t have to mention / caveat other non-climatic factors (which my paper did from the outset), my claim is that ALL you have to do is mention/caveat other non-climatic factors, and you are fine to move on.

Everything I am describing about my paper is totally conventional and banal. There is nothing about my paper that is unusual. That is the point.

His discipline's discourse practice does not include impounding things in ceteris paribus, although those mentions and caveats accomplish precisely that.

In addition, the simplest explanation might be that the editor prefers a pithier paper, with a well-defined focus, and perhaps the marginal effect of global warming on an otherwise declining rate of climate-caused deaths is such a focus.

As I say in the peer review rebuttal, It IS difficult to account for other non-climatic factors. But that COULD be a reason to reject the paper and disincentivize researchers from submitting similar papers. In an alternative universe where editors and reviewers were not so focused on climate change, we could have gotten reviews along the lines of: "when you make attribution statements, you have to account for the long-term build-up of fuel loads or long-term changes in ignition patterns or these numbers are meaningless. Reject this paper". Reviewers did not say anything like this. In our universe, with the focus so intently on climate change, not dealing with other factors is not remotely a barrier to publication.

Also I would argue that it is difficult to quantify these other non-climatic factors in part BECAUSE so much focus is on the climate component - e.g., gridded model results are easily at my fingertips. If more resources and effort had historically gone into quantifying changes in ignition patterns and fuels over time, then those gridded datasets would also be easily accessible and at my fingertips.

That final sentence is also important. If I had a nickel for each time I called out a graduate student lamenting the "lack of data" (meaning, no public-access government file, preferably downloadable) I would have had a lot more F.U. money at the time I called it a career.  Get your butt to the library, make use of interlibrary loan, maybe do a few trips to the National Archives or to the flagship university libraries in other states, and create your own grids.

Mr Brown concludes,

I hope that my raising these issues encourages more researchers to document, in a systematic way, other research/publication practices that can lead, in the aggregate, to an inaccurate public perception of the magnitude of climate impacts relative to everything else.

Yes, and work on getting it right.  One of my pet peeves is people using the expression "tease out" and my elaboration of that discontent goes right to whether Nature's editors have tight priors or not.

Here's the usual dynamics: investigator proposes a testable implication, investigator downloads from a public source some information that proxies reasonably well for the causes and the effects, investigator does the recommended laundering of the data (sorry, all this talk about filtering and screening is too much to resist), investigator runs the recommended regression package and comes up with a bunch of estimates lacking statistical significance, let alone the correct sign. Resulting seminar becomes a collection of suggestions for using additional scrubbing techniques or adding explanatory variables (often under the rubric of "controlling for") so as to get confirmation of the prior, rather than looking for possible reasons the prior is incorrect.

So might it be with rising global temperatures. Until someone obtains the signature of Midgaardsormen, the carbon content hypothesis may be provisional at best.

If you torture the data enough, Nature confesses?  Sorry, couldn't resist.