Graduate Record Exam

Bell Curves & Gaussian Distributions

As if though the race and gender gap wasn�t already huge enough, there was another huge but completely unreported [and as yet, unexplained] jump in the gap in the 2000-2001 tests

1. Race/sex gap leaps from 234 to 267 points.
2. Ain't no black geniuses.
3. The top 97% of applicants to the University of Texas are WHITE men who are DENIED ADMISSION.
4. University of Texas applicants are representative of nationwide scores.
5. The distribution of GRE scores are equivalent to that for MCAT scores.
6. This pattern is identical with SAT, TIMSS, and all other standardized test scores.
7. Race and sex differences are 20 times greater than year to year variations in GRE test scores.
8. Converting GRE Verbal Scores to IQ.

Scoring of the GRE Quantitative test is similar to that for the SAT Math test which consists of one essay question which can be scored only by subjective means, opening it up to controversy about whether or not it's valid or accurate. The most objective part of the test are the 225 five part multiple choice questions which attempt to measure the probability of a student succeeding in college or graduate school. About 1/4th of these questions can be answered strictly from rote memorization. For simplicity we include the three quarters of the test which require some form of reasoning or calculation which can be measured objectively.

Both tests have a base score of 200 which is assigned when a student signs their name. Of the 600 points between 200 and 800, another 150 points can be gained by knowing the subjects but not exhibiting any reasoning or critical thinking skills, for a total score thus far of 350. If a person were to just guess on the balance of the five part multiple choice questions without evaluating the answers in any way whatsoever, he would on average get 20% of them correct just by chance, for another 90 points and a total score of 440.

This score of 440 is conservative. In TIMSS physics, which is scored similarly and which has questions of equal validity and depth, a score of 446 for American girls represents a score which is 30 points lower than if they'd just guessed on this part of the test.  A score of 446 is not indicative of any reasoning or critical thinking skills at all. These skills can be measured accurately only between 446 and 800.  And because many Asian students (more than 10% of them) score higher than 800, the SAT is not a very accurate IQ test, particularly when American blacks consistently score lower than 431 in both SAT Verbal and SAT Math, lower than 391 in GRE Verbal, and lower than 416 in GRE Quantitative.

While the College Board does provide the standard deviation for black students in both SAT and GRE, scores this low most likely do not follow a normal Gaussian Distribution. For the average to be 431, there can't be too many students who score higher than this, because it's almost impossible to score lower than this, so their estimate that the standard deviation, say for black women who took GRE Verbal, is 88, is the most optimistic estimate.

GRE vs. IQ

The correlation of Graduate Record Exam quantitative scores with IQ's enables us to express the huge race and sex differences in terms that more people might understand or appreciate.  Where many people might not grasp what a 234 GRE point spread between Asian men and black women means, expressing it as a 61 point spread in IQ might make more sense.  This is a test that more than one quarter of a million students spend a lot of money to take very year, because they know that both graduate schools and employers depend heavily on the outcome of these scores in admissions and hiring decisions.  Whatever quantity is measured by this test is a desirable feature in both students and employees.  People who don't do well on this test aren't, or should not be, hired or admitted, even under affirmative action programs.

Year after year, from test to test, gaps between races and the sexes have been fairly consistent [other than the as-yet unexplained jump in the gap between 1997 and 2001].  Within each race, when expressed as IQ, the gap between black men and black women is the lowest, with black men scoring 10 points higher than black women.  The biggest gap is 19 points for "Other Hispanic", "White", and "Other".  The gap between Asian men and women, and Mexican men and women, is 17 points.  Puerto Ricans are at 16, and American Indians are at 13.

As seen in the graph above, within each sex, the IQ of Asian men is 116, which is 13 points higher than White men at 103, and 51 points higher than black men at 65; and the IQ of Asian women is 99, which is 15 points higher than White women at 84, and 44 points higher than black women at 55.

Across races and sexes, the most notable points are that the average IQ of White women is equal to Mexican men, 1 point higher than Indian and "Other" men, 2 points higher than Puerto Rican men, and 19 points higher than black men, whereas the IQ of black women is 28 points lower than Indian men and 61 points lower than Asian men.

Is it reasonable to believe that American black women have an average IQ of 55 rather than the 85 touted in the news media?  Could reality be 30 points lower than perception?  What other evidence is there to support it?

Without worrying too much about why females within each race score the equivalent of 13 to 19 IQ points lower than males within each race, and accepting point blank that whatever is being measured is of value to both industry and education, this 10 point gender gap is predicted, explainable, and completely reasonable.  Many different ways of measuring IQs in Africa have confirmed that two African nations have an average IQ of 59, and two dozen nations have an average IQ lower than 70.  It's reasonable to believe that those who had been taken as slaves were not selected from the intelligentsia, as the purpose of slavery was brawn, not brains.  It's not unreasonable to believe that the average IQ for slaves was from the lowest intellectual strata of Africa, or 55 for women and 65 for men.  And there's no reason to believe that this average increased during slavery, by miscegenation with Caucasoids, nor by the huge education expenditures which left them with the world's highest murder and violent crime rates.

To accurately reflect the unannounced, unexpected, unpublicized, unknown, unexplainable, unpredictable INCREASE in the "gender gap" in the midst of all the media hoopla about the "narrowing of the gender gap", we must add that to the above charts, in red, as follows:

 GRE vs. brain size per Professor Rushton