All -

One might wonder how "aggregate" data might not show
patterns of discrimination while more detailed analysis could show
such patterns.

This is a well recognized but difficult problem in the identification
and analysis of discriminatory behavior.  Perhaps a small example can
help in understanding.  What follows is in general based on some
actual experience (admissions of women and men to graduate
programs at UC Berkeley a decade or two ago).

Consider two locations, which have the following "success" rates:

location 1:    Applied     Accepted       Rate

Male                100           80             80%
Female              16            14             87.5%

location 2:    Applied     Accepted       Rate

Male                 36            18             50%
Female            120            75             62.5%

Note that in both locations, the rate of "success" for women is higher
than for men.

Now consider the "aggregate" rates for both locations combined:

Combined:      Applied     Accepted      Rate

Male                 136            98            72.1%
Female              136            89            65.4%

The "aggregate" data indicate that overall, men are accepted at a
higher rate than women.

Now ask yourself:  Are there "patterns of discrimination" here?  If
so, what ought to be done to address the problems?

The central administrator, looking at the aggregate overview, clearly sees
that if anyone faces difficulties, it is the women.  On the other hand, the male "applicants" at each of the two locations would seem to have at least a potentially legitimate complaint with respect to local behavior.  Which way of looking at the situation is the right way?  Or, perhaps more to the point, does that last question even make any sense?

This is an example of what is often called Simpson's Paradox . . .

tom

p.s.  If you want to struggle with some of the meta-issues in the internal and shared negotiated spaces of power relations, you could look at books in the general field of Feminist Epistemologies, and authors like Sandra Harding, Donna Haraway, Nancy Hartsock, Elizabeth Potter, Lynn Nelson, or Judith Butler.  Some of their work is liable to leave you frustrated and angry, but then what is the life of the academic for, anyway?  :-)