[Silent A]

I should further clarify that the authors say that the primary effect of any bias appears to effect mostly white players but that their analysis is incapable of determining if white refs are being favourable to white players or if black refs are discriminating against whites.

Another interesting result is that teams with more white players than the NBA average have a better points for-against differential with white ref crews and teams with more black players than average do better with black ref crews.

Also, 3 white/0 black crews show similar results to 2 white/1 black crews. The big changes start to occur when you go to 1 white/2 black and then 0 white/3 black.

Again, the study is incapable of identifying which group is being biased (white or black refs) because they have no way of identifying a "no-discrimination" baseline. But they've done a good job of identifying that a bias does exist.