<p>There’s little point in looking to NSF’s data for research expenditures. For one, it focuses only on science and engineering. More importantly, it selectively includes affiliated institutions: JHU operates APL for the DOD, so why not include JPL for Caltech, which operates it for NASA? Or LBNL for Berkeley, which operates it for the DOE? Or SLAC, which is operated by Stanford for the DOE (and until recently it owned SLAC as an institution, and still owns the plant)?</p>
<p>Then there are the affiliated institutions that aren’t technically operated by the university but that have an influence on the university and its activities. For example, LLNL at Berkeley; the Stanford Research Institute (SRI International) at Stanford; or the Scripps Institute by UCSD.</p>
<p>Another glaring problem is the arbitrary separation of state university schools: for some, they count all campuses’ research expenditures, while for others they list them separately.</p>
<p>And having certain kinds of schools definitely makes the comparison mucky. Some have medical schools, some don’t; but even for those that do have medical schools, there are other kinds of schools that make the comparison difficult. For example, Berkeley doesn’t have a medical school but simply adding UCSF’s total to it also gives Berkeley an unfair advantage over universities that don’t have nursing, pharmacy, and dentistry schools also (which UCSF does).</p>
<p>Basically, there’s too much uncertainty to really pay attention to the NSF research expenditures study. Though it is interesting to see aggregate figures of research and what the sources of funding are (federal, industry, etc.).</p>