The major limitation of disparate impact statistics as indicators of staffing discrimination is that they cannot tell anything about the intent or about the validity of the policy that has caused the disparate impact. Disparate impact statistics can only tell us what impact has been associated with a given policy. It cannot tell us any of the other things that we really need to know if we are to determine whether discrimination has occurred.
Disparate impact statistics only tell us what impact a particular policy has had on the demographics of a staff. If, for example, a firm institutes a test for potential employees, disparate impact statistics can tell us whether the implementation of the test has reduced the number of women or minorities who are getting hired. This is, of course, important, but it does not tell us everything we need to know.
In order to know if discrimination is actually occurring, we need to know why the test (or other policy) was implemented and how valid it is for the job in question. We need to know if the test truly measures some attribute that is legitimately important for the job. Unless we know things like this, we cannot know if discrimination has occurred. Disparate impact statistics cannot give us this information and they are, therefore, limited.