Monday, January 30, 2006

Pseudo-Science in the Service of Political Correctness

UnderstandingPrejudice.org, which is funded by the National Science Foundation (your tax dollars) and a branch of McGraw-Hill, is "a web site for students, teachers, and others interested in the causes and consequences of prejudice." In its pages one can "find more than 2,000 links to prejudice-related resources, as well as searchable databases with hundreds of prejudice researchers and social justice organizations."

I came across the site while I was searching for information about The Authoritarian Personality, about which I was instructed when I took my one and only college-level psychology course from Prof. Milton Rokeach, author of The Open and Closed Mind (related links). UnderstandingPrejudice.org reminds me very much of The Authoritarian Personality and Prof. Rokeach's teachings, in that it perverts science and logic in an effort to "prove" that conservative views are based on blind prejudice and will lead humanity into the abyss of authoritarianism.

Here is an example of what UnderstandingPrejudice.org tries to pawn off as rigorous analysis:
[C]onsider the following hypothetical problem:
Suppose your school or organization is accused of sex discrimination because the overall percentage of female job candidates offered a position in the last five years is less than the overall percentage for male candidates. To get to the bottom of this problem, you launch an investigation to see which departments are discriminating against women. Surprisingly, however, the investigation finds that within each department, the percentage of female job applicants who are offered a position is identical to the percentage of male applicants who are offered a position. Is this possible? Can each department practice nondiscrimination, while the organization as a whole hires more men than women?
This problem is a variant of Simpson's Paradox [link added] (a well-known paradox in statistics), and the answer to it is yes -- nondiscriminatory conditions at the departmental level can result in hiring differences at the organizational level. To see how this might happen, imagine a simplified organization with two equally important departments, Department A and Department B, each of which receive the same number of job applications. As shown in Table 1, if Department A were to offer a position to 10% of its job applicants (female as well as male), and Department B were to offer a position to 5% of its job applicants (female as well as male), neither department would be discriminating on the basis of sex. At the level of the organization, however, more positions would be going to men than to women, because of the higher number of jobs offered by Department A than Department B. Unless there is a good reason for this difference in hiring, the pattern may represent a form of institutionalized sex discrimination.

Table 1. A Hypothetical Example of Sex Discrimination


Number
of Applicants
Number
of Job Offers
Percentage
Offered Jobs
Department A


Women 500 50 10%
Men 1000 100 10%
Department B


Women 1000 50 5%
Men 500 25 5%
Combined Total


Women 1500 100 6.67%
Men 1500 125 8.33%

First of all there's the presumption that the school is acting discriminatorily if it does not offer jobs to female applicants at the same rate as it offers jobs to male applicants. This is stated without mentioning the possibility that qualified candidates are more likely to be male in some cases (e.g., math teachers) and female in other cases (e.g., art teachers).

Moreover, the writer of the quoted passage blithely promotes the illogical proposition that the school as a whole can discriminate even if individual departments do not discriminate. But the whole cannot be greater than the sum of the parts. If each department does not discriminate with respect to applicants for its positions, that's that: Department A cannot discriminate against Department B's applicants, and vice versa. The aggregation of departmental statistics is therefore nonsensical.

Nevertheless, if the thought police find it necessary to aggregate departmental statistics in order to point the finger of suspicion at an institution, you can be sure that the thought police will aggregate those statistics. Of course, if Department B were to bend over backward toward women by giving them 75 job offers instead of 50, the aggregate statistics would come out "right": a total of 125 offers for women and 125 offers for men. That result -- discrimination in favor of a "protected" group -- is precisely the objective of the thought police, which is why they stoop to statistical tricks.

I know. I've been there.

Related posts:

The Cost of Affirmative Action
The Face of America
Is There Such a Thing as Legal Discrimination?
More on the Legality of Discrimination
Affirmative Action: A Modest Proposal
Race, Intelligence, and Affirmative Action
Affirmative Action: Two Views from the Academy
Affirmative Action, One More Time
A Law Professor to Admire
Guilty Until Proven Innocent