Comment by DangitBobby

They did indeed reduce standards. From the article:

> Throughout the ‘90s and ‘00s, the FAA faced pressure to diversify its field of air traffic controllers, historically a profession that has been primarily white men, notably from the NBCFAE.4 In the early 2000s, this pressure focused on the newly developed air traffic control qualification test, the AT-SAT, which the NBCFAE hired Dr. Outtz to critique from an adverse impact standpoint. As originally scored, the test was intended to pass 60% of applicants, but predictions suggested only 3% of black applicants would pass.5 In response, the FAA reweighted the scoring to make the test easier to pass, reducing its correlation with job performance as they did so.6 In its final form, some 95% of applicants passed the test.7

> This was a bit of a shell game. In practice, they divided it into a “well qualified” band (with scores between 85 and 100 on the test, met by around 60% of applicants) and a “qualified” band (with scores between 70 and 84), and drew some 87% of selections from that “well qualified” band.8 Large racial disparities remained in the “well qualified” band. As a result, facing continued pressure, the FAA began to investigate ways to deprioritize the test.

> Why not ditch it altogether? Simple: the test worked. It had “strong predictive validity,” outperforming “most other strategies in predicting mean performance,” and it was low cost and low time commitment. On average, people who performed better on the test actually did perform better as air traffic controllers, and this was never really in dispute. When they tested alternative measures like biographical data, they found that the test scores predicted 27% of variance in performance, while the “biodata” predicted only 2%. It just didn’t do much.9

> The NBCFAE continued to pressure the FAA to diversify, with its members meeting with the DOT, FAA, Congressional Black Caucus, and others to push for increased diversity among ATCs. After years of fiddling with the research and years of pressure from the NBCFAE, the FAA landed on a strategy: by using a multistage process starting with non-cognitive factors, they could strike “an acceptable balance between minority hiring and expected performance”—a process they said would carry a “relatively small” performance loss.10 They openly discussed this tension in meetings, pointing to “a trade-off between diversity (adverse impact) and predicted job performance/outcomes,” asking, “How much of a change in job performance is acceptable to achieve what diversity goals?” There is a trade-off between diversity (adverse impact) and predicted job performance/outcomes. How much of a change in job performance is acceptable to achieve what diversity goals?