If you wanted to measure sex discrimination in school admissions, you measured strong discrimination in favor of men, when it was actually (weakly) in favor of women.
Understanding Bias: A Pre-requisite For Trustworthy Results
adam kelleher
3509

That’s not what the study concluded. They concluded that measurement is hard, not that bias exists. They explain (page 5 of the linked PDF) that their analysis of “bias” relied on the assumption of equal academic qualifications, which they believe to be an incorrect assumption.