Two mathematicians had written to observe that NSF applications include an "impact" portion, in which "diversity" (etc.) was likely to be mentioned, which might explain the results of the study purporting to find an increase in such terms in funded projects. But now other readers have written in to point out that the original study looked only at abstracts, not the full application. As a classics professor elsewhere wrote to me:
I'm not a scientist or mathematician, and I've never applied to the NSF and am never likely to, but your correspondents' critique of this study seems to be entirely off the mark.
It took me less than 5 minutes...to find and download the same data set as the study uses (https://www.nsf.gov/awardsearch/download.jsp). From which I discovered that the data set comprises ONLY the general abstracts, describing the topic of research. It does NOT comprise the entire grant application which include the "broader impact" statements. Hence it is immune from criticism on those grounds.
A fairer criticism might be that at least some of the terms searched for in the study have broader meanings than the "social justice" one. For example, if one does a simple search for "diversity" in the entire data set, the first two results listed are studies of "phylogenetic diversity" (respectively in phytoplankton and prairie grass). If we search for "inclusion", the first result concerns "melt inclusions in crystals of plagioclase". Clearly the funding of such projects tells us nothing about the so-called "politicization" of science. (The study makes a cursory acknowledgement of the problem when it comes to "diversity", but not with "inclusion", and in any case makes no attempt to quantify it.) But the study does break down its results by individual terms, and some of those terms are less susceptible to double meanings of that sort. To that extent the results are likely to have some validity.
I'm opening comments for further thoughts from readers about the study and its methodology.