I am mystified by number of commenters saying that you can't draw conclusions because of the small sample size. This is not a small sample size. The results are robust in part because the data set is so large: 83000 observations.
One of the interesting points of TFA is that it pinpoints where in NIH's two-tier process of review black researchers are eliminated. They are more likely to be eliminated in the first round of peer review than in the second "scoring" round. This is not consistent with them submitting, on average, less worthy projects. If that were the case, they should have the same or a higher attrition rate at the second level of review. Black researchers are also less likely to resubmit grants, which could be a simple lack of mentoring or communication from the NIH.
We find it troubling that the typical measures of scientific achievement—NIH training, previous grants, publications, and citations—do not translate to the same level of application success across race and ethnic groups. Our models controlled for demographics, education and training, employer characteristics, NIH experience, and research productivity, yet they did not explain why blacks are 10 percentage points less likely to receive R01 funding compared with whites.
The middle-aged incompetent white managers brought into the startups were dumb enough to say racist things out loud.
And they preface their statements with "I'm not a racist but..." or "Don't take this the wrong way but..." or the ever-popular "Some of my best friends are black, but..."
If in any problem you find yourself doing an immense amount of work, the answer can be obtained by simple inspection.