Graduate Record Examination in Physics

A review and analysis of the skyrocketing costs of US education parallelling a corresponding drop in education standing and quality suggests the US is doing something seriously wrong. We spend more as a percent of GDP for education than almost every other country in the world, and by some estimates spend twice as much as a percent of GDP as Japan.

US Education Scores Compared to Other Countries

According to the US Department of Education, the US ranks almost at the bottom of the scale in international competitions between seven "selected" countries in Science Education, NOT including countries like Japan and Germany. Per the "Technology Assessment and Forecast Report", Patent and Trademark Office, US Dept. of Commerce, of the "Top 20 Companies Receiving US Patents in 1994" (that is the largest number of US patents), 9 out of the 20 were Japanese companies, suggesting that Japan's scores could be off the following scale.

Amoungst the same seven countries, the US ranks almost at the bottom of the scale in Math Education:

SAT Math Scores Decline as the Gender Gap Widens

As affirmative action and other anti-productivity federal schemes have taken effect, both SAT and ACT scores have declined, and the gender gap in mathematics widened, from 17.6% in 1967 to 18.1% in 1989 on SAT Math, and from 12.2% to 13.7% on ACT Math.
SAT Math 1960 1965 1970 1975 1980 1985 1989 Change
Male 529 520 509 495 491 499 500 -29 points
Female 485 476 465 449 443 452 454 -31 points
Percent Difference 15.44% 15.94% 16.60% 18.47% 19.75% 18.65% 18.11%
Percent by which males scored higher than females averaged 17.65% over the last 3 decades. Source: College Entrance Examination Board, New York

ACT Math Scores Decline as the Gender Gap Widens

ACT Math 1960 1965 1970 1975 1980 1985 1989 Change
Male 24.5 22.3 19.3 18.9 18.9 18.6 18.3 -25.3%
Female 22.7 20.6 16.2 16.2 16.0 16.0 16.1 -29.1%
Difference 7.9% 12.2% 19.14% 16.67% 18.12% 16.25% 13.66%
Percent by which males scored higher than females averaged 14.84% over the last 3 decades. Source: The American College Testing Program.

Note: the narrowing in the gap from 19.14% in 1970 to 13.66% in 1989 is the result of declining male scores rather than increasing female scores. This suggests that narrowing this gap even further requires decreasing male scores even further, rather than increasing female scores.

A similar phenomena has occurred in the US economy. 

Graduate Record Exams

[email protected]
5th year graduate student
Physics Dept
University of Texas at Austin
I'm want to thank everyone that sent their Physics GRE scores and comments for the "informal" survey. The number of people that replied was fairly low so I can't say too much about the stats.
21 women
20 men
11 professors
19 grad students
6 post docs
5 misc
The average percentile score was:

44% (+/-20%) for the women who responded

75% (+/-14%) for the men

Even with such low numbers I think this is a pretty significant difference in scores. They barely overlap within their standard deviation. I called the GRE Board and they confirmed that there is a gender gap on the test and that white males generally score higher than women and minorities.

My own thoughts on why the gender gap exists are the following. Educational research on standardized tests have shown that women will spend more time `proving' that they have the right answer while men are more likely to `go with their gut feeling' and skip on to the next problem. Its a problem of confidence and guessing. If you don't guess on these tests and only answer the questions that you've had time to work out, there's very little chance that you will do as well as someone who is willing to guess and works through the test quickly. You don't have time to double check your work and women are more likely to do this. I know for a fact that I did not guess the first 3 times I took the Physics GRE and barely made it all the way to the end. The fourth time I studied by doing hundreds of multiple choice, learning to do them quickly and raised my score almost 200 points from my lowest score. I think the gender gap might disappear or at least narrow if the time allowed for the test was longer, or if it was fill in the blank.

Thank you,

Jennifer Siders

From [email protected] Jan 30 11:30:38 1996
Date: Mon, 29 Jan 1996 13:50:32 -0500 (CDT)
From: [email protected]
Subject: Re: GRE Scores

Score     China      USA
980->     14         2
960       3          2
940       2          2
920       1          3
900       2          3
880       4          2
860       5          2
840       1          1
820       0          3
800       1          7
780       2          6
760       1          2
740       1          5
720       0          8
700       0          7

There were no Chinese scores below 700, but about half of the american
scores lie below 700.  These data are from UT applicants 95-96

Jennifer Siders
January 29, 1996 -- I will be getting a detailed summary of Physics GRE results from ETS very soon and can send you a copy if you wish. What I have at the moment is the 1989-92 Volume and Score Data for the Physics GRE Test. I received this from someone who had attended an ETS meeting. It took me a while but I found someone at ETS who knew what I was talking about and is going to send me one directly. This way I can cite a source for the information. :-)
Gender & Country
Percent(of test takers)
Mean Physics GRE
I have data from a survey I did over the internet and also from 3 years worth of applicants to our Physics Dept that show a similar gap. There will be a special session at the APS/AAPT meeting May 4 on this topic where I will be presenting these results. 

Gender Gap in SAT Math Scores Widens

"there has been a difference of nearly a standard deviation between the mean mathematics scores of males and females"

Trends in SAT Scores and Other Characteristics of Examinees Planning To Major in Mathematics, Science, or Engineering. Research Report. Source: Scholastic Aptitude Test-ABSTRACT, ERIC_NO-ED376079.

This study analyzed data from the Scholastic Aptitude Test (SAT) taken between 1977 and 1988 to study trends in the numbers, test scores, and other characteristics of high school seniors planning to major in math, science, or engineering, and to compare these data with comparable data from examinees planning to major in other fields. Results indicated that: (1) the total test-taking population declined in number until 1983 and has since been increasing; (2) the percentage of examinees who planned to major in math, science, or engineering increased from 24% to 29% of the examinee population; (3) in 1988 examinees planning to major in math, science, or engineering obtained a mean verbal score 18 points higher and a mean mathematics score 31 points higher than the population average; (4) among students planning to major in math, science, or engineering, the mean mathematics score declined until 1981, increased until 1985, and declined thereafter; (5) among examinees who identify themselves as white and who plan to major in math, science, or engineering, there has been a differences of nearly a standard deviation between the mean mathematics scores of males and females; sex differences were not as great among black examinees; and (6) among examinees who identified themselves as black and who plan to major in math, science, or engineering, the mean mathematics score for males rose 18 points, and the mean for females rose 20 points. Appendices include Student Descriptive Questionnaires and definitions of major field categories in Science and Non-Science Tables. (MKR)

As education quality declined, US incomes and standards of living declined.

Some estimates show that, with the increased labor force size, Per Worker Income today is lower than it was in the 1950s.

Had incomes continued to increase at their pre-1973 rate, they would now be as much as SIX TIMES higher than they actually are. This of course would have required the quality of education to have increased, rather than decreased.

Why did education quality decrease as spending increased so dramatically? Would education quality have decreased if the budget for education had been decreased? Or would it have decreased even further?

Consider what happens to Koreans when they come to the US. Their IQ actually decreases by 6 points from 109 to 103. Yet Koreans in Belgium increase their IQ by 1 point.

Because we spend more per student than any other country, we expect education quality to be better. But the following chart is one more indication that exactly the opposite has occurred.

"What IS open to question is whether the tests measure innate, unchangeable abilities. As this chart shows, there are groups with average or better performance out there. However almost every conventional race based intervention in education does not result in equal performance. Only programs based on rigorous academics such as Barclay, and not race such as; conventional desegregation, busing students to expensive surburbs or affirmative action in hiring and multiculturalism in curriculum have been shown to result in superior performance."

Ethnic Group IQ Source SAT Verbal SAT Math
Borderline Retarded 70-75 IQ - Seligman p.151 ? 267
Black Delinquents 74 Shuey p. 498 ? ?
Africans (Africa) 75 Lynn as cited by Murray p. 288 ? 302
Southern Black children 80.5 Shuey ? ?
White Delinquents 81 Shuey p. 498 ? 347
Black Felons 81 Shuey p. 498 ? 347
Black Females 82 Interpolated from SAT estimate ? 355
Black Average 85 IQ Seligman p.149 351 377
Colored From Africa 85 Lynn as cited by Murray p. 288 ? ?
Dull-Normal 80-90 IQ Textbook by Wechsler ? ?
Border Black Children 87 Shuey ? 393
Northern Black Children  87.6 Shuey ? ?
Black G12 NAEP Math 89 (7) ? ?
Black US Northern Average 90 IQ Jensen(4) ? 415
Koko the Gorilla 85-95 IQ Seattle Pacific magazine(1) ? ?
Urban North Black 91.1 Shuey ? ?
Hisp G12 NAEP Math 92 (7) 379 430 
White Felons 92 Shey p. 498 ? ?
White Chronic Welfare  92 Murray ? ?
Overall Criminals  92 Murray ? ?
Black Calif Math CLAS 92 A.Hu estimate (6) ? ?
White Female Average 96 Extrapolated from SAT ? 457
Malays 96 Lynn 1977 (Seligman) 460
Black Reading CLAS 96 A.Hu estimate (6)  ? ?
White Unwed - No Welfare 98 Murray ? ?
Mixed Child of Black 100 Murray p. 310 (5) ? ?
Black Children Boston, NYC 100 See note ? ?
Adopted Asian Children 100 Sandra Scarr ? ?
American Students 100 Cattell's culture Fair Test ? ?
British 100 Lynn 1977 (Seligman) ? ?
Average Intelligence 90-109 IQ Textbook by Wechsler ? ?
Mainland China 101 Lynn (Seligman) ? ?
White CLAS Reading & Math 100 A.Hu set by definition (6)  447 489
Flynn's Asian Studies 100  ?? ? ?
White Male Average  102  Extrapolated from SAT ? 502
Asian Calif CLAS Math 102 A.Hu (7) ? 504
White G12 NAEP math 102 (7) ? ?
British Children 102 IQ Lynn & Song 1994 ? ?
Malnourished Korean Adult  102 Winick, Meyer and Harris 1975 (2) ? ?
Asian Average 103-4 Murray 405 521
Korean (est) 105 IQ Lynn & Song 1994 ? 528
Moderately-nourished Adult Korean 106 (2)Winick, Meyer and Harris 1975 536
Chinese Hong Kong 106 IQ Chan & Lynn 1989  ? ?
Chinese Hong Kong 106.7 Lynn 1982 (Seligman) ? ?
Japanese 9 Year Olds 107 IQ Shigehisa & Lynn 1991 ?
Korean 108.5 IQ Lynn & song 1994 ? ?
Asian G12 NAEP math 109 (7) 559
Korean Children in Korea 109 IQ Lynn & Song 1994 ? ?
Chinese Hong Kong 109.8 Lynn 1982 (Seligman) vs. British 100 Raven's matrices ? 566 
Bright-Normal 110-119 IQ Textbook Wechseler ? ?
Korean Adults in Belgium  110 Frydman & Lynn 1989 ? ?
Adopted Black Children 110 Sandra Scarr ? ?
Chinese 110 Lynn 1977 (Seligman) ? ?
Well Nourished Korean Adult  112 Winick, Meyer and Harris 1975 (2) ? 581
Jewish Boys 112.8 verbal Ann Arbor Institute (Seligman) ? ?
Orthodox Jewish Children (applying to Hebrew schools in 1950's)  114.9 Miles B. Storfer ? ?
Other Storfer Jews 110-115 Miles B. Storfer ? ?
Jews 115 Review of Bell Curve ? 603
Blacks at UC Berkeley 115 A.Hu top 15% UC definition ? ?
Blacks in Math Program 116 Barclay school Baltimore MD ? 611
Adopted White children 120 Sandra Scarr  ? 641
Blacks at MIT 120 A.Hu based on SAT 1985 ? ?
Whites at UC Berkeley 124 A.Hu top 5% top 1/3 of UC ? 671
Superior 120-129 IQ Text Wechsleler ? ?
Above Superior 130 & above By Definition 715
White/Asian MIT 138 A.Hu (top 0.5% of population) 771

(1) In many cases, conversions were made from SAT scores to an estimate of the US population based on percentile to IQ conversion.

(2) Koreans adopted by Americans, Winick, Meyer and Harris 1975

(3)American men = Asian women at visiospatial tasks (math) (Lynn / Seligman)

(4) Blacks in New York and Boston are equal to grade level in some grades, equal to 100 IQ by definition. Some schools that are mostly black in Chicago also perform at the 50th percentile.

(5) Study of children of mixed parents of European and Black soldiers found no difference

(6) A.Hu Converted 1995 CLAS distribution to IQ with white=100

(7) National Assessment of Educational Progress given to national cross-section of students, 1990

(8) "A better idea" The Economist Dec 2, 1995 p. 23 Barclay public school in Baltimore Md. emulates a tough private school to achieve 85 percentile math scores with 90% black students, 82% free lunch who previously were scoring 20-30 percentile. Superintendent was fired who had opposed "rich black man's curriculum" inappropriate for poor students.

(9) Flynn looked at studies that put Asian IQ higher and concluded that after you accounted for increases in IQ since the tests were "normalized", Asian IQs were actually no higher than whites might expected to be at the same time. However, he does not dispute that Asians still have superior academic performance in school, which is just as useful as having a high IQ.

This is my latest version -- Arthur Hu's "IQ Spectrum" -- 12422 107th Pl NE Kirkland WA 98034

Note that blacks are, as a group, generally below whites, but it is not difficult to find groups who score 100 average or far above under ideal conditions. Those who say that tests are invalid are implying that performance is equal when every test devised by man shows differences in skills such as vocabulary, abstract reasoning, and mathematics.

Date: Fri, 2 Feb 1996

Here's something else you might find helpful - note Asian females have a gender gap, but its nearly erased by the racial gap for Asians.

1993 SAT Scores Ranked by Race and Gender (unpublished special tabulations, College Board)
559 Asian male
518 White male
512 Asian female (only 6 points below WM)
472 White female
451 Hispanic male
408 Hispanic female
400 Black male
[email protected]

The following data is published by the College Entrance Examination Board, and/or the Educational Testing Service.

The Gender Gap... Is the SAT fair to women?

Studies have shown that women slightly outperform men in their first year of college. So why does the SAT, a supposedly "fair" test, yield 45-point lower scores to women?
                1994 SAT Test Scores for:

                       Women       Men

Average for ALL races   881        926

The Ethnic Gap... Is the SAT fair to minorities?

Now let's take a look at similar data for ethnic groups. The differences are even larger!
                1994 SAT Test Scores for:

Asian American            951

Caucasian                 938

Mexican American          799

Black                     740

Average for ALL races     902

The Income Gap... Is the SAT fair to middle class and poor students?

Perhaps the most interesting data is the correlation between test scores and family income. On the average, well-off students score 234 points higher than students whose familes earn under $10,000 per year.
                1994 SAT Test Scores for:

under $10,000/yr           766

$10,000 - $20,000/yr       812

$20,000 - $30,000/yr       856


$50,000 - $60,000/yr       929

$60,000 - $70,000/yr       948

over $70,000/yr           1000

Our opinion...

Your SAT score is a better prediction of your income, race, or gender than it is of your ability to perform in college. So why does it not predict your grades in college accurately, and why do college grades not predict income, when SAT scores do?


Affirmative Action, which allowed the number of female educators to increase by 350,000 in the last 3 decades, while the number of male educators declined by more than 100,000, leading to lack of discipline in schools, which destroyed education quality, and the ability of the US to compete in what is now misappropriately labeled the "global economy". This crippled US manufacturers, so "Manufacturing jobs" are leaving our shores as fast as they can, leaving behind "service jobs" (which very soon will have little to service).

Why Fatherhood?