A THB Follow Up: Climate Research Fails a Science Integrity Test

Something is rotten in academic publishing

This is fine.

The ridiculous ChatGPT image above appeared in a recent peer-reviewed paper, and was quickly sussed out and widely mocked. Within days, the paper was retracted. How could it not be retracted? Fake information should not appear in a peer-reviewed journal.

Well, unless that journal is the Proceedings of the National Academy of Sciences (PNAS).

Last February I documented how PNAS had published a paper — Grinsted et al. 2018 — that was based on a fake ‘dataset” of hurricane losses assembled by a successive set of interns for a now defunct insurance company, for purposes of marketing insurance products. The “dataset” was never intended for research.

I won’t rehash the details here — you can take a deep dive here. The “dataset” simply does not exist and thus the paper’s analysis and conclusions are based on false information. This case is not complicated or difficult to understand.

At the time, I requested of the editor of the paper, Kerry Emanuel of MIT, that the paper be retracted based on its reliance on a fake dataset. Emanuel agreed to escalate my request to PNAS.

Today, I heard back from PNAS that the dataset and paper are without problems and the matter is now closed. I reproduce the full PNAS email to me below. Some highlights:

• Remarkably, PNAS did not actually examine the dataset in question;

• PNAS apparently relied instead on an unrelated letter I submitted on the paper’s methods from 2019 — the information I shared with them earlier this year was based on new information, which was apparently ignored.

• Remarkably, one of the “expert” reviewers wrote: “the conclusions of the Grinsted et al. article do not depend on the ICAT dataset due to the robustness analyses performed with the other datasets.” This is easily shown to be false — The “other” datasets are subsets of the fake dataset, as I explained to PNAS but which they apparently did not read or ignored. Remarkable.

• The “expert” reviewer wrote that my request: “does not change my impression of the validity of the data, and again ignores the robustness analyses performed with the other datasets.” There are no other datasets. Rather than rely on an “impression,” PNAS could have actually looked at the data and done some simple analyses to confirm my assertions — it might have taken all of 10 minutes.

Today, I emailed Kerry Emanuel, who edited the original paper, and asked if he agreed with the PNAS verdict. He said that he did agree that the paper was fine.

Writing last February, I observed:

This is one of those rare cases where a fatal error in research is obvious, consequential and simple for anyone to document for themselves. I have requested of PNAS that G18 be retracted, and corrections from IPCC and the UCNCA must occur as well. This is a simple but important test of scientific integrity for climate science. Let’s see what happens.

Climate science — and PNAS specifically — has failed the test.

Don’t take my word for it — look at the evidence and the data and come to your own conclusions. For me, this episode is yet another example of how some (not all) climate science has a tendency to go off the rails. This is a particularly egregious example.

Here is the PNAS response, from the PNAS Ethics Editor, received earlier today:

Comments welcomed! If you think PNAS made an error here, please share on your favorite social media platform and click the little heart. Science is self-correcting, but sometimes that takes a while.