Just because different groups perform differently doesn't mean that a test is flawed. Millions of people have taken tests and the data is solid. Minorities as a group consistently score about 2/3 to 1 standard deviation below the white mean and Asians score about 1/3 standard deviation higher. It doesn't matter what test it is, the SAT, GRE, Bar Exam, etc. Ashkenazi Jews average 2/3 of a std deviation above other whites.
I would urge Ms Allen to read this paper by Linda Gottfredson on this.
"The foregoing patterns of employability and trainability help to illustrate the implications of the one-standard-deviation (15-point) difference between American Whites and Blacks on valid, unbiased tests of (phenotypic) intelligence. The Hispanic-White difference is about half that, although it varies by subgroup (Cuban, Puerto Rican, and so forth). Asian Americans score at least as well as Whites on tests of intelligence, although they too differ by specific ethnic group (as do Whites). As shown in Figure 1, a one-standard-deviation difference in IQ represents a substantial average difference in trainability and employability.
There are successive increments in average IQ of roughly one standard deviation between assemblers, bank tellers, store managers, and attorneys, for example. Whereas the White average is about 100, which is typical of workers in jobs of moderate prestige and complexity (clerical, protective service, and crafts workers), the Black IQ bell curve is centered at about 85, a score that is typical of semiskilled workers.
Current minimum military enlistment standards are set above IQ 85, thus excluding at least half of Blacks. Indeed, a 1980 U.S. Department of Defense study showed that whereas 71 to 89% of White men 18 to 23 years old were eligible for enlistment on the basis of aptitude and education (the Air Force being the most stringent and the Army the least), the percentages were only 21 to 41% for Black men and 38 to 53% for Hispanic men (Eitelberg, 1988, p. 99). Figure 1 illustrates why the disparate impact of valid, race-neutral hiring becomes even more acute in higher levels of education and employment. The proportion of Blacks who are available at successively higher levels of IQ drops dramatically relative to that of Whites. The Black-White ratio changes from 5:1 for IQs below 75 to 1:30 for IQs above 125.
Noncognitive traits such as conscientiousness are important for good job performance (Hunter & Schmidt, 1996). Even when such noncognitive traits might substitute for deficits in cognitive capability, however, they would have to be higher on average among Blacks and Hispanics to compensate for these groups' lower average cognitive skills. Research reveals no such compensatory noncognitive advantage for performing cognitively demanding tasks."
<link>
Following publication of the Bell Curve the APA set up a taskforce to comment in psychometric testing including the question of bias.
Intelligence: Knowns and Unknowns Report of a Task Force established by the Board of Scientific Affairs of the American Psychological Association Released August 7, 1995:
Test Bias.
"It is often argued that the lower mean scores of African Americans reflect a bias in the intelligence tests themselves. This argument is right in one sense of "bias" but wrong in another. To see the first of these, consider how the term is used in probability theory. When a coin comes up heads consistently for any reason it is said to be 'biased," regardless of any consequences that the outcome may or may"not have. In this sense the Black/White score differential is ipso facto evidence of what may be called "outcome bias."
African Americans are subject to outcome bias not only with respect to tests but along many dimensions of American life. They have the short end of nearly every stick: average income, representation in high-level occupations, health and health care, death rate, confrontations with the legal system, and so on. With this situation in mind, some critics regard the test score differential as just another example of a pervasive outcome bias that characterizes our society as a whole (Jackson, 1975; Mercer, 1984). Although there is a sense in which they are right, this critique ignores the particular social purpose that tests are designed to serve.
From an educational point of view, the chief function of mental tests is as predictors (Section 2). Intelligence tests predict school performance fairly well, at least in American schools as they are now constituted. Similarly, achievement tests are fairly good predictors of performance in college and postgraduate settings. Considered in this light, the relevant question is whether the tests have a "predictive bias" against Blacks, Such a bias would exist if African-American performance on the criterion variables (school achievement, college GPA, etc.) were systematically higher than the same subjects' test scores would predict. This is not the case. The actual regression lines (which show the mean criterion performance for individuals who got various scores on the predictor) for Blacks do not lie above those for Whites; there is even a slight tendency in the other direction (Jensen, 1980; Reynolds &:Brown, 1984). Considered as predictors of future performance, the tests do not seem to be biased against African Americans."
