Database Reference
In-Depth Information
60
males
females
total
50
40
30
20
10
females
discriminated
males
discriminated
0
0
20
40
60
acceptance rate to medicine (%)
Fig. 8.4 Illustration of redlining
the acceptance rate to medicine is reduced, the illegal discrimination is introduced,
females are discriminated (the grey area to the left). If the admission committee
chooses to increase the acceptance rates to medicine, then males actually become
discriminated, as more females are accepted than the explainable difference. The
light grey area illustrates the Simpson's paradox (Simpson, 1951), in which a re-
lation exists in different groups is reversed when the groups are combined. We see
that more males are accepted, but in fact males are being discriminated, because
females are applying to a more competitive program. Such situation was reported in
the Berkeley study (Bickel et al., 1975). The next section presents a more elaborate
example of a reverse discrimination.
8.3.4
Illustration of the Reverse Discrimination
We illustrate what happens if we do not take into account the explainable aspect of
discrimination. For simplicity let us modify our university example. Now there are
males and females applying for a job. The candidates are assessed based on their
experience. Male applicants happen to have on average longer job experience (in
years) than females. The recruitment company selects 6 candidates for an interview,
that have the highest experience level.
Figure 8.5 (a) shows the example situation. We see that males and females were
treated equally, thus all the difference in acceptance rates (the acceptance rate for
males is
4
2
25%) is explainable and happens due to the fact
that the experience attribute is correlated with the gender. On the other hand, if we
8 =
50%, for females
8 =
Fig. 8.5 Illustration of reverse discrimination
Search WWH ::




Custom Search