Unfortunately, there is a lot of internal corruption in academia, which makes this feedback ineffective. If you apply for a grant, you already have to outline results of our research. Heather Sarsons, the author, is a doctoral fellow in the "Inequality and Social Policy program". She is paid to find inequality. If she won't find any, she would be unsuccessful. The same holds for colleagues--who actually do peer review. They are on the same ship.
It's not a critique of this work, but the overall climate in academia. About 80% of papers in health science was found to be unreproducible. Guess why.
The comparable figure for Physics is something like the top 100-odd schools produce the half of the research. That seems like a much fairer distribution.
Take the inherent fuzziness of social science research, compound it with prestige gaming, compound with the impossibility of reporting a negative result, and what you get is shoddy research.
I think it's both. Contentious though this may be, I think that the reproductibility problems in academia invite us to consider the idea that we may be hiring political activists to do research.
>"The majority of schools require faculty to apply for tenure after 7 years. I consider one year before and after the 7th year to account for people who go up for tenure early or late because of a leave of absence, for example. I put universities into bins of 3 based on their ranking and assume that an individual is denied tenure if that person moves to a lower-ranked university group after 6-8 years. For example, a person who moves from Harvard to MIT after 6 years is not assumed to have been denied tenure since he moves within the same bin of schools. Someone who moves from Harvard to UCLA after 6 years is assumed to have been denied tenure since he moves to a lower group of schools. As another example, a person who moves 5 or fewer years after his initial appointment is not assumed to have been denied tenure since he moved before the tenure window (years 6 through 8 at an institution) starts."
Arbitrary binning of universities, using a change in employer as a proxy for getting tenure, using number of papers published as a measure of productivity, fitting lines to a relationship that does not appear linear (and in fact cannot be linear since fraction getting tenure must be in the range 0-1).
This issue may exist, but I don't think this analysis is capable of producing anything convincing.