 
Howard Wainer, Educational
Testing Service, Harvard Educational Review, Fall 1992
Study of 48,000 college students in
Eastern universities majoring in math.
 The overall median SAT math scores in the US in 1993 were 502 for men
and 457 for women. 
 The "base score" is 200 and the difference between men and
women is (502 minus 457) divided by (457 minus 200) or 17.5%. 
 SAT scores are more accurate in predicting future incomes and job performance
than in predicting college grades. 
 46,920 math students in Howard Wainer's article in the Fall 1992 Harvard
Educational Review illustrates the DISTRIBUTION of math skills. 
SAT range 
Men's Letter Grade 
Women's Letter Grade 
Number of Men 
Number of Women 
Percent Men 
Percent Women 
A (592 & up) 5,693 0 22.0% 0% B (575591) 7,357 0 28.4% 0%
C (549574) 6,931 5,144 26.8% 24.5% D (532548) 2,881 6,383 11.1% 30.4%
F (524531) 3,030 0 11.7% 0% (493523) 0 5,394 0% 25.7% (476492) 0 2,235
0% 10.6% (475 & Below) 0 1,872 0% 8.9%
Physical Characteristics 
Average Height 
Average Weight 
Average Strength 
Emotional Stability 
Suicides per Year 
Life Expectancy in Years 
(592 & up) 
A 
 
5,693 
0 
22.0% 
0% 
Female 
5' 4" 
120# 
? 
? 
5,700 
79 
Difference 
10.94% 
41.66% 
3X 
3X? 
4.69X 
8.48% 
 50.4% of the men's groups scored higher than the highest women's group
 68% of the individual men scored higher than the highest women.
 64.5% of the women scored lower than "D" men.
 45.2% of the women scored lower than all of the men.
 Percent of women in the men's "A" range = 0%.
 Percent of women in the men's "B" range = 0%.
 Percent of women in the men's "C" range = 24.5%.
 Percent of women in the men's "D" range = 30.4%.
 Percent of women in the men's "F" range = 0%.
 Women's group's mean score was 30.7% higher than women nationally,
and men's group's was 25.8% higher than men nationally, a 4.9% difference
which indicates that the women's group was more selective.
 There were 4,864 or 18.8% fewer women in the study than men, possibly
because there were not enough highscoring women available to have even
sized groups?.
 Sample sizes for women in Advanced Math were one half (1/2) that for
men and for Calculus were 40% lower than that for men while the sample
sizes of women in Remedial Math and Regular Math were larger.
 Women's Remedial Math scores were 48.3% lower than their Advanced Math
scores. Iincreasing their sample size by 23.1% with students whose average
scores were 48.3% lower would have reduced women's "Grade Means"
from 536 to 517, which is 19 points or 5.7%.
IF 500,000 women took the SAT math test, and IF 1 out of 50,000 scored
at the men's A level, THEN there might be 10 women in this sample who scored
at this level, which would be 1 woman out of 312 men.
Education's failure to produce competent employees particularly in math
makes knowledge of SAT scores even more important than ever. To claim that
women are discriminated against in education, or in the workforce, simply
because they earn 40% less than men, when so much credible evidence shows
that at best 1 woman for every 312 men has basic competence in math, is
dangerous. It is not "Why aren't women paid as much as men?"
but "How much more valuable to a company is a worker who has good
math skills?" How much more would you pay a worker to reduce math
errors in your business by this much?
 
