Of course. My assumption can be stated as: the employer, for whatever reason, sees that empirically the correlation between e.g. GPA and work performance is dif

Of course. My assumption can be stated as: the employer, for whatever reason, sees that empirically the correlation between e.g. GPA and work performance is different for different populations, and therefore corrects it. If the "correction" is done beforehand, then the employer won't see this phenomenon.