Completely unsurprising.
> Since McKinsey refused to turn over its numbers, Green and Hand had to reverse-engineer the firm’s 2015, 2018, and 2020 datasets.
So it's still not completely settled who is right. But the fact McKinsey wouldn't turn over their dataset is suspicious. When one researcher works in the open and another does not, conflicting results are likely to favor the open researcher.
> Green and Hand not only were unable to replicate the studies’ findings; they also found that each of the three studies had analyzed the data backward. Instead of looking at a firm’s diversity policies in the years leading up to a given year’s financial performance, McKinsey had reviewed each firm’s financial performance in the four or five years leading up to the year in which its researchers snapshotted their executive demographics. In other words, according to Green and Hand, the positive correlations that McKinsey researchers observed may have reflected “better firm financial performance causing companies to diversify the racial/ethnic composition of their executives, not the reverse.”