Arthur Hu's Index

Home|Contact webmaster

SAT: Scholastic Aptitude Test

Discussion

America's biggest college entrance examination, and a controversial one because of differences between gender ethnic and racial groups. In truth, there's a bunch of interesting stuff in those numbers, but you won't make any sense of it unless you toss out your politically correct glasses. The SAT is your classic IQ-style norm-referenced test. Unless the wave of state tests, there is no pass point, no written standards of what every entering college student is expected to know and be able to do.

The two biggest no-nos in testing are 1. Tests should not be used as cutoff to determine admission and 2. you should not teach to a test. Guess what standards based testing is based on??

Under-minorities tend to get lower test scores. The standard explanation is that all the test scores show is you income, but scores are low for minorities even when you take parental income, grade point averages and parental education into account. Test scores are consistently lower for minorities for a given GPA

Asians are especially interesting because they don't all fall at the over or under minority line. Asians were 6 times over-represented in top scores in 1989, but also 5 times at the bottom end. By comparison, whites are under-represented at the lowest scores, but Blacks are best represented at the lowest scores, falling to only 3% of their population at the top end.

The new writing test puts Asians at only a 2 point disadvantage in 2008, but adds another 94 point gap to combine scores for blacks / african americans. Making the test harder only puts disadvantaged minorities farther behind.

SAT Spectrum

550V  George W Bush
480V  Bill Bradley

1510 California Institute of Technology
1300 UC Berkeley
1268 US Military Academy
1146 Evergreen State College
1132 Seattle University
1097 Simmons University
1080 Morehouse Collegee
 880 Tuskeegee Princeton Review 2004

Contents



@@achievement test / II

UC president in 2001 says he favors SAT-II achievement test to lessen
racial bias, but it might make things worse. Gap is narrower because
smaller, self-selected population takes the test, but if everybody
has to take it, the gap is probably even larger since average black
and white student is at the 85 and 88th percentile of their ranges.

Charles Murray says SAT-II gap is narrower: Feb 20, 2001 they produce somewhat smaller B/W differences, mostly in the region of .7 to .8 SDs, rather than the .9 SD difference in the SAT-V and SAT-M (which works out to a bit more than 1 SD in the combined score). You can see the differences for the individual achievement tests in an appendix of TBC (p. 663). But this fairly minor reduction in the difference is produced by a huge truncation of range. The typical black and white taking achievement tests as of the early 1990s (when I did these analyses) were at the 88th and 85th percentiles of their respective distributions. After correcting for restriction of range, the gap has to be real big.

Hausman says she suspects it's just that SATII has less score compression, usually only the most elite schools require them.


La Griffe du Lion says: 2/2001
Here, by race, are 1994 SAT data showing mean scores obtained on
several achievement tests and corresponding Math SAT I for the test
takers. All tests have a standard deviation ~100.

              white  black  difference

Chemistry      585    516     69
Math SAT I     650    565     85

Biology        562    490     72
Math SAT I     610    517     93

Physics        609    534     75
Math SAT I     674    598     76

Math Level I   557    487     70
Math SAT I     581    492     89

Math Level II  663    588     75
Math SAT I     659    574     85

Math Level IIc 672    622     50
Math SAT I     672    610     62

Except for physics, the racial gap is significantly narrower with
achievement tests, but hardly eliminated.

%%Spanish Test

SPANISH SAT-II EASY ACE FOR SPANISH SPEAKERS
\clip\2001\06\satspan.txt Wall Street Journal Bilingual Students Use
Language Tests To Get a Leg Up on College Admissions By DANIEL GOLDEN
Staff Reporter of THE WALL STREET JOURNAL Also helps Asian students,
but not blacks.



@@ACT COMPARISON

Meaning of hi/lo sat act rankings by
state Hi SAT rankings meaningless if few take the SAT. SAT mostly
given on coasts, ACT mainly in center of country Raymond Daye


ACT VS. SAT COMPARISON \doc\web\98\08\actsat.txt some minor
differences, but essentially equivalent competition.

ACT VS SAT details \doc\web\98\08\actsat2.txt Donna
Gardener The ACT is considered to be more curriculum-based. It has no
analogies on it and has approximately 1/4 math-related questions.  On
the ACT it is to a students' advantage to answer every question even
if he must guess.


@@bates

z84\clip\2004\08\batessat.txt
20-year Bates College Study of Optional SATs Finds No
    Difference in Student Performance
       LEWISTON, Maine, Oct. 1 2004(AScribe Newswire) -- In a milestone 
20-year study of its well-known policy for optional SATs for admission, 
Bates College has found no differences in academic performance or 
graduation rates between submitters and non-submitters.



@@bias

The most common charge is that the SAT discriminates against
minorities, and therefore should be dropped, or be replaced with
essay tests.

z74\clip\2003\10\satbias.txt The URL for this page is
http://www.theatlantic.com/issues/2003/11/mathews.htm.  The Atlantic
Monthly | November 2003 The The Bias Question 
In a surprising challenge to the SAT's reputation as an unbiased
measure of student learning, one researcher has argued that blacks do
better than matched-ability whites on the harder questions of the
SAT—something he believes their scores should reflect
by Jay Mathews 

z48\clipim\2001\04\sat.tif
Why Colleges Shouldn't Dump the SAT Robert Barro Hoover Institution
Business Week April 9, 2001 After taking in account parental income
and education, admission test scores for minorities were 17
percentiles lower than whites for same sex, 4 for hispanic, but
college grades for blacks or hispanics were lower than whites, even
with the same scores.  Women scored 4 points lower, but had higher
college grades, so maybe some basis for bias, but not minorities.

Patti Houseman April 2001
The SAT consistently underpredicts the performance of females and
overpredicts the performance of males, according to the center. 
Yes, but not by much. The last validity study I read (a few years 
ago) showed that after controlling for the easier grading in courses 
females are more likely to take, the magnitude of underprediction 
was .06 on a 4 point scale and the degree of overprediction for males 
about the same.


COLLEGE BOARD (SAT) POSTS COMPLAINT ABOUT OCR BAN ON "DISPARATE"
TESTS
\clip\99\13\nontest.htm
http://www.collegeboard.org/press/html9899/html/ocr2.html

the Resource Guide focuses exclusively on disparate impact resulting
from tests (or differences in outcomes) and ignores the level of
validity and utility offered by a test.

Second, the Resource Guide makes repeated references to “alternative
measures”, “alternative assessments”, “another way”, “practical
alternatives”, and “least discriminatory practical alternatives.” The
inconsistent terminology is confusing, but the Resource Guide offers
no guidance on what level of disparate impact would result in an
investigation. Must there be substantial statistical disparities or
would any disparate outcome result in an investigation? 

the Resource Guide establishes a false dichotomy between standardized
tests and all other measures used to make the same types of high
stakes individual decisions. The Resource Guide holds tests to a much
higher legal and professional standard than other measures (e.g.,
rank, ratings, samples of student work, grades) in terms of producing
uniform results across 
groups.

The tone and language used throughout is laced with legal terms,
often never defined in this document (or when defined there is no
citation or reference of where the definition is taken), or having no
common legal definition in case law. Examples, include terms such as
“educational necessity, due process challenges, de jure
discrimination, substantially serves educational purposes,
disproportionate adverse impact…” Clearly, this Resource Guide will
not be useful to most educators who routinely confront issues
involving high stakes educational testing. 

PRINCETON REVIEW CLAIMS SAT HARMS WOMEN, MINORITIES
\clip\98\12\princsat.txt Tuesday September 1, 12:01 pm Eastern Time
The Princeton Review: College Board's Report on SAT Test-Takers
Serves as an Indictment of the SAT Board's Own Data Contradict Their
Claims of the Test's Validity

\clip\98\08\hurtmin.txt 1/14/98 Education week
http://www.edweek.org/ew/vol-17/18sat.h17 "You're more likely to end
up with a diverse class\ because the SAT discriminates." " [but study
shows blacks would decline if it went to grade only, and grades were
raised accordingly]

\clip\98\08\sathist.txt http://www.edweek.org/ew/current/31weiss.h17
4/15/98 Education Week The SAT: Public-Spirited or Preserving
Privilege?  By Julian Weissglass Exploring elitist historical
assumptions that still underly the SAT.  [says that SAT was designed
to exclude minorities, they aren't valid and should be banned, now]

DIFFERENT GROUP AVERAGES MEAN THE TEST IS BIASED?
/images/972/112197/p03.gif
"Decade-old SAT scores better than you thought" Seattle Times Oct 26,
1997 The SAT is thought to be biased because minorities and women
score lower than whites and men, but defenders say it is because of
factors other than bias.

DO THE SATS ONLY PICK OUT THE RICH?
http://www.theatlantic.com/Atlantic/ issues/95sep/ets/fall.htm
\clip\97\04\fallow.txt Atlantic Magazine February 1980 The Tests and
The "Brightest": How Fair Are The College Boards?  by James Fallows
Comment - there are lots of Asians who get good test scores with low
incomes, conversely, there are lots of blacks with high incomes with
low scores. He does not investigate this question.

WOMEN SCORE LOWER ON SAT MATH, BUT GET BETTER FRESHMAN GRADES
\clip\97\19\satbias.txt BY laurie.snell@chance.dartmouth.edu
http://www.geom.umn.edu/docs/education/chance/teaching_aids/
student_projects/morgen/node1.html
Statistical Ideas Up: Sex Bias and the Scholastic Aptitude Test Previous:

@@black

BLACKS WITH HIGH SAT SCORES STILL DISADVANTAGED
\clip\98\15\blakscor.txt
http://www.washingtonpost.com/wp-srv/WPlate/1998-10/25/278l-102598-idx.html
Puzzling Percentiles By Karin Chenoweth Washington Post Sunday,
October 25, 1998; Page R07 African-American students who score 1200
or better on the SAT -- almost 5 percent of males and 4 percent of
females -- have fewer advantages than white students who score in
that range. They are more likely to come from families with lower
incomes and with fewer college degrees than white high scorers. They
attend schools where fewer of the parents have college degrees or
high incomes. And they are much more likely than white high scorers
to attend school in large central cities, where educational
opportunities are often more limited than in the suburbs.

\priv\95\01\hiblack.txt


@@Book

Lemann: The Big Test

Lehman advocates an even
bigger standards based test than the SAT

LEHMAN CLAIMS SAT UNFAIR TO POOR MINORITIES, ADVOCATES OBE STANDARDS
zip37\clip\99\18\satleh.txt Is the SAT a) Useful or b) Unfair?  By
Dan Seligman 09/29/1999 The Wall Street Journal Page A20 Doubtless
feeling a need to come up with something, he has an "Afterword"
imagining a totally different world wherein the federal government
guarantees decent schooling for everybody, the country's high schools
embrace an agreed-upon national curriculum, college admissions are
determined by mastery of this curriculum and just about everybody
gets a full-scale college education, apparently at public expense.

zip38\clip\99\19\lehmann.txt Atlantic interview with Lehmann, says 
way to get rid of affirmative action is to get rid of test score 
gap.

zip38\clip\99\19\bigtest.txt Bracey reviews Lemann's _The Big Test_
from the Sunday _WP_.
http://www.washingtonpost.com/wp-srv/WPlate/1999-10/17/105l-101799-idx.html
Sunday, October 17, 1999; Page X04 THE BIG TEST The Secret History of
the American Meritocracy By Nicholas Lemann Farrar Straus Giroux. 406
pp. $27 Reviewed by Gerald W. Bracey His treatment is the grandest in
scope, the most literate and most readable. In the end, though, it is
too glib and wrong -- dishonest even.


@@Bush, GW

http://www.johntaylorgatto.com/historytour/history8.htm
John Gatto:
Bush graduated from Yale, became governor of Texas, and president of
the United States—with a mediocre 550 [verbal SAT]
Bill Bradley had a horrifying 480 on the verbal part of his own SATs,
yet graduated from Princeton, won a Rhodes Scholarship, and became a
senator;

@@california SAT Scores

casat93.xls California 1993
SAT scores by race

@@california high schools

Taken from California Dept of Education Goldmine site:
1995 SAT scores and other statistics by county:
\priv\96\15\ALAMEDA  TXT
CONTCOST TXT
MARIN    TXT
ORANGE   TXT
LOSANG   TXT
SACRAMEN TXT
SANDIEG  TXT
SANFRAN  TXT
SANMATEO TXT
SANTBARB TXT
SANTCRUZ TXT
SHASTA   TXT
SONOMA   TXT
VENTURA  TXT
YUBA     TXT
PREFNOTE HTM
PRESRELS HTM
CALHIGH  HTM
CLASSAGE TXT

See rankings:Best CA High Schools


@@coaching

Testing helps, but by not more than 20 points out of 1600.

z48\clip\2001\03\satcoach.txt March 25, 2001 - NY Times Admissions
Test Courses Help, but Not So Much, Study Finds By GINA KOLATA The
study's data indicated that the average gain from coaching was no
more than about 20 points on the 1,600-point SAT test, similarly on
the ACT

The Test Under Stress New York Times Magazine Jan 10, 1999 SAT score
have never been so important, or so vulnerable to charges that
theyvdon't measur up.  Test coaching is part of the problem. It may
also be part of the solution.  Tony Schwartz. Tom Hayden proposes tax
funding of coaching for undrperforming students, some use SAT as
general skills tutorial.


@@College Ranking



@@decline

There isn't a real decline in SAT scores, it's just more people
taking the test.

VIETNAM WAR DRAFT DODGERS BROUGHT DOWN SAT SCORES, MILITARY IQ
http://www.intellectualcapital.com/issues/98/0611/icbusgraph.asp How
One War Dumbed Down Everything by Howard Wainer June 11, 1998
\clipim\99\04\onewar\onewar1.htm

path: asian.education.sat.decline file: \doc\95\06\satnodec.txt -
Corrected for socioeconomic changes, students score higher in 1994
than their counterparts in 1975 reference: "Private Schools Withing
the Public Schools" Seattle Times May 14, 1995 Dick Lilly E. Rhodes,
J. Houtz p.1


@@discrepant

ONE THIRD OF STUDENTS HAVE GPA'S THAT WILL NOT BE AS ACCURATE AS SAT SCORES
http://www.collegeboard.org/research/html/rs1.pdf
October 1997 The College Board's Office of Research and Development
[Note from AADAP: This report is in the PDF format, and requires the
downloadable Adobe Acrobat reader to be opened.]  "Relevant to recent
proposals to deemphasize standardized test scores in college admissions.
Reports that approximately two-thirds of all students have "nondiscrepant"
scores, meaning that their predicted freshman GPA in college will be
similar whether SAT scores or high school GPA are used.  Discusses
differences between this "nondiscrepant" group and students with
discrepant scores, and within the latter category, between HSD students
(those with standardized GPA at least one standard deviation above the
standardized SAT scores) and SATD students (those with standardized SAT at
least one standard deviation above the standardized high school GPA).
Reports that female and minority students are more highly represented in
the HSD group. Since students in the SATD group are more likely to take
courses in science and math and fewer remedial courses, the report
speculates that the use of freshman GPA rather than freshman course grades
may result in an artificially lower predictive validity for SAT scores."



@@Education Testing Service (ETS)

Home Page: < ahref="http://www.ets.org/">http://www.ets.org/
SAT
GRE
GMAT
TOEFL
LSAT

Corporate Headquarters
          Educational Testing Service
          Rosedale Road
          Princeton, NJ 08541 USA
          (609) 921-9000
          FAX: 609-734-5410
          e-mail: etsinfo@ets.org



@@gender gap

Gender Dif what is even more
interesting is the sex difference in quantitative scores on higher
level exams.  While the effect size favoring males on the SAT-math is
.31; on the GRE General it rises to .63, and on the GRE Math it's .87.

La Griffe du Lion  The gender difference in
math SAT scores cannot be explained away by either different standard
deviations or a larger female test-taking pool. Knowing that the best
and the brightest of both sexes take SATs, I solved this problem by
reconstructing, from the extreme high-end tail of the distribution,
the distribution that would be obtained if *every person* in the U.S.
of high school senior age took the exam.  The results of this
analysis yielded a gender gap of 0.303 SD from the >= 750 cutoff, and
0.313 SD from the >= 780 cutoff.  Both numbers agree well with the
gender difference in mean scores of 0.32 SD for just test takers.
 
For details, see:
http://members.xoom.com/griffedulion/women_and_minorities_in_science.htm


SAT GENDER GAP NARROWS \clip\97\12\gendgap.txt
/clip/97/12/gendgap/sat-gender-educ.html c:\clip\97\12\gendgap.txt
New York Times May 7, 1997 Gender Gap in Testing Narrower Than
Believed, Study Finds The most persistent gap isn't girls in math,
but boys in reading and writing, and its much closer than it was 30
years ago.

Rank of women better/worse

-- Girls Much Better
Verbal writing
Perceptual Speed
Short Term MemoryStudy skills
Verbal reading
--- Better
Math Computation
Abstract reasoning
Verbal Vocabulary reasoning
--- About Equal
Social science
--- Boys Better
Math Concepts
Spatial Skills
Natural Science
Geopolitical
--- Boys Much better
Mechanical/ Electronic



@@GPA Grade Point Average

ARE GRADES MORE VALID THAN TEST SCORES?


http://www.arthurhu.com/index/test.htm#satgpa
\doc\web\97\06\satgpa.wk1

A: BLACK GPA GAP ISN'T ANY CLOSER THAN TEST SCORE GAP
Many have argued that gpa is a much better and less biased indicator
than test scores, but blacks don't fare much better in GPA than test
score. Blacks are 3.7 times more likely to have GPA's as D, E, or F,
and only 1/3 as likely to have A+ grade averags.

B: BLACK GPA IS LESS RELIABLE AND OVERPREDICTS THAN TEST SCORE
Moreover, blacks are much more likely to attend segregated schools
with much lower grading standards. Thus test scores show that a Black
A+ is equivalent to a White A-, and a White B is better than a Black
A-. Studies cited in books like "Choosing Elites" by Klitgaard and
the LSAS validity study show that GPA combined with test scores
actually overpredict minority success, and that for minorities, GPA
is much less reliable by itself than test scores alone, which is to
be expected since SAT scores are uniform across all students, but
grading within schools is not.

Ranking of math SAT by GPA and Race
-----------------------------------
White A+    556
Asian A+    540
White A-    484
Black A+    480
MexicanA+   478
White B     425
Black A-    411

SAT vs. GPA 1995 by Race

        Total   White   Black   Mexican Asian   NativeAm
A+ 97-10   57219   42852    2029    1534    6592     340
A 93-96   134663   98497    6567    4104   14927     847
A- 90-92  145166  104153    8456    5045   14846    1068
B 80-89   488898  326618   54666   19207   34811    4678
C 70-79   148524   87082   28392    5470    7931    1716
DEF <70     4277    1996    1120     107     353      47
No resp    89246   13145    2624     856    2054     240
total    1067993  674343  103854   36323   81514    8936

SAT Math vs. GPA
        Total   White   Black   Mexican Asian   NativeAm
A+ 97-10     547     556     480     478     540     521
A 93-96      500     513     432     428     475     475
A- 90-92     470     484     411     410     440     448
B 80-89      408     425     356     363     382     391
C 70-79      354     372     317     327     337     354
DEF <70      337     362     312     318     320     345
No resp

Percent Total   White   Black   Mexican Asian   NativeAm
A+ 97-10   5.36%   6.35%   1.95%   4.22%   8.09%   3.80%
A 93-96   12.61%  14.61%   6.32%  11.30%  18.31%   9.48%
A- 90-92  13.59%  15.45%   8.14%  13.89%  18.21%  11.95%
B 80-89   45.78%  48.43%  52.64%  52.88%  42.71%  52.35%
C 70-79   13.91%  12.91%  27.34%  15.06%   9.73%  19.20%
DEF <70    0.40%   0.30%   1.08%   0.29%   0.43%   0.53%

Index   Total   White   Black   Mexican Asian   NativeAm
A+ 97-10   -1.19    1.00   -3.25   -1.50    1.27   -1.67
A 93-96    -1.16    1.00   -2.31   -1.29    1.25   -1.54
A- 90-92   -1.14    1.00   -1.90   -1.11    1.18   -1.29
B 80-89    -1.06    1.00    1.09    1.09   -1.13    1.08
C 70-79     1.08    1.00    2.12    1.17   -1.33    1.49
DEF <70     1.35    1.00    3.64   -1.00    1.46    1.78

Ratio   Total   White   Black   Mexican Asian   NativeAm
A+ 97-10    0.84    1.00    0.31    0.66    1.27    0.60
A 93-96     0.86    1.00    0.43    0.77    1.25    0.65
A- 90-92    0.88    1.00    0.53    0.90    1.18    0.77
B 80-89     0.95    1.00    1.09    1.09    0.88    1.08
C 70-79     1.08    1.00    2.12    1.17    0.75    1.49
DEF <70     1.35    1.00    3.64    1.00    1.46    1.78

@@gap (measured in standard deviation)
goodoc: notes sat gap http://docs.google.com/Doc?id=dfthx7q4_2159x3t72sfv

The SAT gap is nearly as large as the 1 Standard Devation difference
usually found in IQ tests, but the test includes more whites, and
fewer blacks, so under-estimates the difference in the comparable
population.

CA 1990 math black=392 white=504 SD=125, so CA difference =0.9 SD,
about the same margin as IQ tests, but with blacks slightly
underrepresented.

%%Journal of Blacks in Higher Education

Journal of Blacks in Higher Education
 
This is almost precisely what I was talking about back in the 1980s -
measured by SAT scores, black are short by a factor of 7 verbal and 10
in math, and whites with incomes under $10,000 outscored blacks with
incomes over $100,000. This is a journal of Black education, it only
lacks a similar analysis of how far Asians are ahead of whites.
 
http://www.jbhe.com/features/53_SAT.html
Whites from families with incomes below $10,000 had a mean SAT test score that was 17 points higher than blacks whose families had incomes of more than $100,000.
we find that 5.4 percent of all white SAT test takers scored 700 or above on the verbal portion of the test. But only 0.74 percent of all black SAT test takers scored at this level. Therefore, whites were nearly seven times as likely as blacks to score 700 or above on the verbal SAT. Overall, there are more than 39 times as many whites as blacks who scored at least 700 on the verbal SAT. 
On the math SAT, only 0.6 percent of all black test takers scored at least 700 compared to 6.4 percent of all white test takers. Thus, whites were more than 10 times as likely as blacks to score 700 or above on the math SAT. Overall, there were more than 53 times as many whites as blacks who scored 700 or above on the math SAT.
if these schools operated under a strict race-neutral admissions policy where SAT scores were the most important qualifying yardstick, these universities could fill their freshman classes almost exclusively with students who score at the very top of the SAT scoring scale. As shown previously, black students make up at best between 1 and 2 percent of these high-scoring groups. 
 that there will still be a large racial gap in standardized test scores 22 years from now, the deadline for race-sensitive admissions as outlined by Justice O’Connor. Thus, the huge and growing gap in SAT scores, and particularly the scores at the highest levels, becomes one of the nation’s most urgent problems
It doesn't cover Asians, but my 1980s findings were that that the primary reason some colleges are taking in 15-25 percent Asians or half in top California schools is because Asians are a huge percentage of top SAT scores.
 
A Large Black-White Scoring Gap Persists on the SAT 
http://www.jbhe.com/features/53_SAT.html
* Whites from families with incomes below $10,000 had a mean SAT test
score that was 17 points higher than blacks whose families had incomes
of more than $100,000.
* 5.4 percent of all white SAT test takers scored 700 or above on the
verbal portion of the test. But only 0.74 percent of all black SAT
test takers scored at this level. Therefore, whites were nearly seven
times as likely as blacks to score 700 or above on the verbal SAT.
Overall, there are more than 39 times as many whites as blacks who
scored at least 700 on the verbal SAT. 
* On the math SAT, only 0.6 percent of all black test takers scored at
least 700 compared to 6.4 percent of all white test takers. Thus,
whites were more than 10 times as likely as blacks to score 700 or
above on the math SAT. Overall, there were more than 53 times as many
whites as blacks who scored 700 or above on the math SAT.
* if these schools operated under a strict race-neutral admissions
policy where SAT scores were the most important qualifying yardstick,
these universities could fill their freshman classes almost
exclusively with students who score at the very top of the SAT scoring
scale. As shown previously, black students make up at best between 1
and 2 percent of these high-scoring groups.
* there will still be a large racial gap in standardized test scores
22 years from now, the deadline for race-sensitive admissions as
outlined by Justice O’Connor. Thus, the huge and growing gap in SAT
scores, and particularly the scores at the highest levels, becomes one
of the nation’s most urgent problems
* It doesn't cover Asians, but my(Hu) 1980s findings were that that the
primary reason some colleges are taking in 15-25 percent Asians or
half in top California schools is because Asians are a huge percentage
of top SAT scores.
* In 2006 the average black score on the combined math and verbal
portions of the SAT test was 863. The mean score for whites on the
combined math and verbal SAT was 1063, about 17 percent higher. 
* Not only are African-American scores on the SAT far below the scores
of whites and Asian Americans, but they also trail the scores of every
other major ethnic group in the United States including students of
Puerto Rican and Mexican backgrounds. In fact, few people realize that
American Indian and Alaska Native students on average score 118 points
higher than the average score of black students. On average, Asian
American students score 225 points, or 19 percent, higher than African
Americans

Note: JBHE has ordered every book by Richard Lynn that LRA have
published. They have the information about race differences and seem
unafraid to address it somewhat. LR. Andrews

@@Grade
@@Prediction

Blacks get grades equal to whites with much higher SAT scores, this
probably explains the "overprediction" found when comparing blacks
with equal grades and test scores

I'll need to put the table here sometime, but it's in the 1985 
College Bound Seniors book.

Underprediction
	The overprediction is no small amount either - according to Robert
	Kiltgard it amounts to about 240 points on the combined SAT.
	(Thernstrom's, p. 638, footnote 34)
	 In other words, a black with a combined SAT score of 1240 will
	generally perform at the same level as a white who scores only 1000.


ndeed, there is an article in pinc which deals with this
(http://www.cycad.com/cgi-bin/pinc/feb98/anon-discrim.html) 
and Ed Miller and Richard Epstein have both dealt
with aspects of this too.


The following PDF includes information on under/over prediction of
academic performance using the SAT for several ethnic groups, as well
as men and women.
Common Sense About the SAT (their title, not mine): 

http://www.collegeboard.org/index_this/research/html/rn01.pdf 
Once you get to the link, go to Table 7.  It shows that high 
school GPA overpredicts freshman GPA for blacks by .35 
on a 4 point scale.  The SAT overpredicts same by .23 on a 
4 point scale.

Table 8 prediction
SAT  / SAT + HSGPA
.18/.15     english not best lan
.09/.06     females
.08/.04     asian american
.01/.01     white
------ under predict ---
------ over predict ---------
-.10/-.06   males
-.13/-.13   hispanic
-.23/-.16   black
-.29/-.24   Am indian

Hausman

EITHER GPA OR SAT OVERPREDICT BLACK FROSH GPA
From: "Patricia Hausman" 
ETS data show that high school grades overpredict
freshman grade point average for black students
by .35 points on a 4 point scale.

The SAT overpredicts by .23 points (on a 4 point
scale).

These calculations are from 1994 and based on validity
data from 45 colleges.

Patti


@@Graduation rate

file: \clipim\99\03\scoregap\scoregap.htm
url: http://epn.org/prospect/40/40jencnf.html
Christopher Jencks and Meredith Phillips, "America's Next Achievement
Test: Closing the Black-White Test Score Gap," The American Prospect
no. 40 (September-October 1998): 44-53
when we compare blacks and whites with the same twelfth-grade test
scores, blacks are more likely than whites to complete college. Once
we equalize test scores, High School and Beyond blacks' 16.7-point
disadvantage in college graduation rates turns into a 5.9-point
advantage.


@@History


NEW SAT WILL MEASURE HOW GOOD SCHOOLS ARE, NOT STUDENTS
z56\clip\2002\07\murrsat.txt Wall Street Journal July 3, 2002
COMMENTARY SAT Reform Fails the Needy By CHARLES MURRAY Put more
bluntly, the SAT is backing off its historic mission of measuring how
smart students are.  Before the process goes any further, now is a
good time to ask who benefits.

z56\clip\2002\06\newsat.txt
http://seattletimes.nwsource.com/html/nationworld/134483919_sat28.html
Seattle Times June 28, 2002
Essay to be part of revamped SAT; critics say changes in test fail to
address issue of racial bias By Rebecca Trounson Los Angeles Times
The new test will cost students $10 to $12 more than the current one,
which is priced at $26 this fall... by March 2005, will include
questions based on advanced algebra, replace verbal analogies with
additional reading-comprehension questions. And, most significantly,
it will require each student to produce a handwritten essay


@@income

"The SAT is only a measure of parental income" WRONG! While this is
generally true within a race, it is NOT true between races. Rich
whites score better than poor whites . But poor Asians score as well
as middle class whites and upper class blacks.

\clip\98\09\racecolr.txt http:
//www.examiner.com/980607/0607race.shtml June 7, 1998, San Francisco
Examiner How race colors learning By Annie Nakao OF THE EXAMINER STAFF
Studies show that the most well-off African American youngsters -- children
of middle- and upper-income families that have come the farthest since the
civil rights movement -- generally perform worse than Asian and white
children from lower-income families.

California SAT scores for 1995 showed that whites accounted for 39 percent
of those scoring 750 or more in math (out of a perfect 800), and half of
those who scored 600 or more.

Asian American students comprised 46 percent of those scoring 750 or more
and 30 percent of those scoring 600 or more.

college-bound students with at least one parent who had a graduate
degree, the 1997 mean combined SAT verbal and math score
1165 Asians
1130 whites
1014 Mexican Americans
 951 blacks

A UC analysis last year of 1995 SAT scores in California found that
blacks from California families earning $100,000 or more per year had
a mean math score of 498, 1 point less than whites from families
earning less than $10,000 and only 7 points more than Asians whose
families made less than $10,000.

http://www.arthurhu.com/index/test.htm#satincome
\doc\96\06\satinc.wk1
Source: 1995 College Board SAT Profiles

Test scores
        white     |black    |mexican  |asian
income  verbamath |v    m   |v    m   |v    m
x$10,000          |         |         |
under 10  409  460| 320  315| 330  386| 343  482
10-20     418  459| 337  369| 349  403| 363  500
20-30     428  471| 352  382| 369  420| 497  518
30-40     433  478| 362  393| 384  431| 415  528
40-50     439  488| 375  405| 399  446| 432  537
50-60     446  498| 382  414| 409  456| 444  549
60-70     453  506| 385  415| 415  458| 453  558
over 70   475  533| 407  442| 430  478| 476  595
overall   448  498| 376  426| 356  388| 418  538

Black over 70 < white under $10,000
White over 70 < asian 40,000
Asian verbal scores comparable above $40,000 / yr

Ethnicity Ranked by 1995 Math SAT scores
--------------------------
595 Asian over   $70,000
538 Asian average
537 Asian $40-50,000
533 White over   $70,000
498 White average
488 White $40-50,000
482 Asian under  $10,000
478 Mexican over $70,000
460 White under  $10,000
431 Mexican $30-$40,000
442 Black over   $70,000
426 Black average
315 Black under  $10,000
Source: College Board Senior Profiles

See “Standardized Tests: The Interpretation of Racial and Ethnic Gaps”
at: 

http://members.xoom.com/griffedulion/testing.htm
This shows how to measure the "true" difference between races
measured in standard deviation, not just the arithmetic 
difference between scores. On low-g tests such as TAAS, minorities can 
narrow the difference slightly, but on very difficult tests
they may be limited by IQ.

Test scores difference
        white     black     mexican   asian
income  verbamath v    m    v    m    v    m
x$10,000
under 10    0    0  -89 -145  -79  -74  -66   22
10-20       0    0  -81  -90  -69  -56  -55   41
20-30       0    0  -76  -89  -59  -51  -31   47
30-40       0    0  -71  -85  -49  -47  -18   50
40-50       0    0  -64  -83  -40  -42   -7   49
50-60       0    0  -64  -84  -37  -42   -2   51
60-70       0    0  -68  -91  -38  -48    0   52
over 70     0    0  -68  -91  -45  -55    1   62
------------------------------------------------
overall     0    0  -72  -72  -92 -110  -30   40

The gap actually gets LARGER for math when you equalize income
Asians are better at math at EVERY income level, worse at verbal below $40,0

@@IQ equivalence

Lostrow says According to Arthur
Jensen, in 1970 a verbal score of 500 on the SAT was approximately
equivalent to an IQ of 117. (50th percentile) Thus for seniors in
high school, a 600 verbal SAT back then would represent an IQ of 130,
a 700 score would represent IQ143, 800 ~ 156 IQ. With renorming, a
current score of 500 SAT verbal probably is equivalent to about IQ
104.  In other words, the average score has dropped a standard
deviation (110 SAT points, or 13 IQ points).

RENORMED SAT NOW CUTS OUT AT ABOUT 134 IQ
Lostrow says the SAT is actually
better than some IQ tests at testing very high IQ, the top 800 score
was about equal to a 156-160 but about 134 now.  The pre-renormed SAT
was the best cost-effective instrument available to detect high IQs (
up to about 156-160, represented by the top verbal SAT score, 800).
The WISC and Stanford-Binet are frequently inadequate to test people
whose IQs are above 130-132 and who have uneven subtest scores, some
of which hit the test ceiling.  The renorming, has resulted in the
renormed SAT verbal test ceiling cutting off at roughly IQ 135-140.

From:           	"James"
The Bell Curve cites literature that estimates the SAT score for the 
total population.  It assumes that the score is a little less than 
900.  The standard deviation for scores of the test-taking population 
is about 230 points.  You can use the mean and standard deviation to 
estimate IQ equivalence.  By comparison, Mensa considers 1250 (post 
1974, pre-recentering) an acceptable score.
1300             SAT or CEEB scored prior to 9/30/74
1250             scored from 9/30/74 thru 1/31/94
NA               scored after 1/31/94

Igoe tries one equivalence
      Mean  SD  50Pct  84 97.5  99.5
IQ    100   16  100   116  132  148
SAT1  800  230  800  1030 1260 1490
SAT2  900  230  900  1130 1360 1590
1 - before recenter
2 - after recenter

estimates for the whole population are based on the PSAT and studies
comprising all students, not just SAT takers or prospective college
students.


@@Job

THE WALL STREET JOURNAL
http://online.wsj.com/article_print/0,,SB106729501444224900,00.html
Wall Street Journal
October 28, 2003
More Employers Are Asking
Job Seekers for SAT Scores
KEMBA J. DUNHAM
Staff Reporter of THE WALL STREET JOURNAL
Donna Chan is 23 years old and has been out of college since May 
2002, when she graduated from Wagner College on New York's Staten 
Island. So should anyone care how she did way back in high school on 
her SATs?



@@language and citizenship

Contrary to studies that show assimilation hurts academic peformance,
Asians who speak English as a first language score nearly as high in
math as those with another language (immigrant), and just as high as
whites in verbal. Immigrants score only 20 points higher, the same
spread as between whites and am-born Asians.

English as second language:

 \doc\95\12\sat95bk.wk1
 1995 Scholastic Aptitude Test Breakdown by Ethnicity
 Data from the College Board
 Data Entry and Analysis by Arthur Hu

                White          Black             Hisp             Asian
 Language p. 6  pct  verb math pct verb   math   pct verb   math  pct verb
 math
 Overall SAT         448  498        356    388        376   426        418
 538
 White gap             0    0        -92   -110        -72   -72        -30
  40

Primary
 English         95  450  499  90    358    389  42    406   449  29    449
 525
 English + Othe   4  415  467   7    341    378  32    365   412  29    443
 546
 Other            2  406  525   3    327    383  27    343   420  42    380
 541

 Citizen         98  449  498  93    357    387  88    383   430  60    452
 539
 Perm Res/ Refu   1  415  510   4    341    382   9    332   399  24    364
 514
 Citz of Other    1  424  541   3    374    433   3    334   420  15    384
 579
 Not Citizen      2             7                12               39

 Notes - Asian citizen V452 better in verbal than White V449
 Asian Perm Res V364 is comparable to Black V356 and Hisp V376, but
 Math M514 is still better than White Citizen M498 - equally smart, but
 penalized by low verbal scores
 Asian citizens of foreign countries are better in math and verbals than
 Perm Res better in math than Citizens (but citizens > white too)
 Nearly 40% of Asians not citizens, highest of any SAT group
 Asian with 1st English or other very close to White, as 2nd language close
to
 Black

@@major

Table 22-6: Percentage of students intending

 to major in selected fields and their average verbal and mathematics
SAT scores: 1995

Top 5 majors
Intended college major          Total Male  FemaleVerbalMath
Mathematics                          1     1     0   480   619
Physical sciences                    1     2     1   500   580
Engineering                          9    13     3   451   556
General/interdisciplinary            0     0     0   507   541
Biological sciences                  5     4     6   471   524
...
Computer/information sciences        3     4     1   416   495
...
Bottom 5 majors
Agriculture/natural resources        2     2     1   414   454
Education                            8     4    10   409   446
Home economics                       0     0     0   382   416
Public affairs and services          3     3     2   380   414
Technical and vocational             1     1     0   357   407

Note - Norman Matloff insists there is no differnce in ability to
program between english and computer science majors.


@@math

 \doc\95\12\sat95bk.wk1
 1995 Scholastic Aptitude Test Breakdown by Ethnicity
 Data from the College Board
 Data Entry and Analysis by Arthur Hu

                White          Black             Hisp             Asian
 Math Coursework     Gap           Gap               Gap              Gap
 Overall               0  498       -110    388        -72   426         40
 538
 Algebra         97    0  498  96   -110    388  97    -73   425  94     37
 535
 Geometry        95    0  504  90   -109    395  94    -73   431  94     36
 540
 Triginometry    55    0  539  43   -106    433  42    -68   471  64     26
 565
 Precalculus     38    0  573  24   -104    469  32    -77   496  54     18
 591
 Calculus        23    0  613  12   -105    508  16    -76   537  40     13
 626
 Computer Math    9            10                10               13

 Black Calculus is equal to Whites who took Geometry, Asian Algebra, or
 Hispanic precalculus
 Asians advantage falls to 13 points with Calculus vs. 40 overall
 Nearly half of Asians took Calculus, 2X rate of Whites, 4X rate of Blacks


                White          Black             Hisp             Asian
 Years of Study pct  verb math pct verbal math   pct verbal math  pct verbal

 math
      4          70       521  62           406  62          451  77
 558
      3          27       450  33           361  35          390  20
 469
      2           3       402   4           332   4          356   3
 449
      1           0       395   0           319   0          329   0
 428
 1/2 or less      0       435   0           330   0          372   0
 434
 honors          29       597  18           487  25          518  42
 619

 42% (nearly half) of Asians took honors math courses compared to 29% White,


 18% Black
 Black 4 year = White 2 year, Hisp 3 Year, Asian 1/2 year or less
 Blacks who took honors courses (487) still less than Whites overall (498)
 Asian honors is highest, even higher than Asian $70,000+ income

@@Misuse


"the report is radical in that it acknowledges the abuse of the use
of SAT scores and that such abuse can be detrimental to women. The
abuse includes overweighing the SAT or using it as sole criterion in
some admission processes. Also, the test is employed as a proficiency
cut-off for some college math courses and scholarships which is a
potential misuse(330). "

@@news

\doc\web\97\07\eastsat.txt Seattle Times 8/28/97 Eastside students
score better on SAT this year

Combined Scores

Ranked by 1997 SAT Scores
-------------------------
1997 1996 City
-------------------------
1098 1090 Bellevue
1083 1066 Northshore
1057 1055 Issaquah
1067 1054 Lake Washington
1046      Washington State Average
1016      US Average

\clip\97\19\ocsat.txt Los Angeles Times Wednesday, August 27, 1997
O.C.'s SAT Scores Again Beat U.S., State Averages Education: Bigger
course loads boost achievement, official says. But some schools slip
from 1996 results.  By TINA NGUYEN, Times Staff Writer

\clip\97\19\calfsat.txt
http://www.sacbee.com/news/beetoday/newsroom/local/082797/ local03.html
Sacramento Bee Aug 26, 1997 State students' SAT scores up: Best
showing in 22 years By Jon Engellenner Bee Staff Writer

\clip\97\19\texsat.txt
http://www.dallasnews.com/texas-southwest-nf/tsw63.htm Dallas News
SAT math scores rise slightly 08/27/97


\clip\97\19\satmont.txt
http://www.WashingtonPost.com/wp-srv/WPlate/1997-08/27/112l- 082797-idx.html

Washington Post aug 27, 1997 p. A01 SAT Scores Rise In Montgomery,
Prince George's Averages Fall in D.C., Most of Northern Va.

(Seattle) Student test scores take a leap \clip\97\19\seatsat.txt
http://www.seattletimes.com/extra/browse/html97/altsatt_ 082797.html
Copyright © 1997 The Seattle Times Company Wednesday, Aug. 27, 1997
by Dick Lilly Seattle Times staff reporter

http://www.boston.com/dailyglobe/globehtml/239/Scores_on_SAT_ inch_up_statew

ide.htm
Scores on SAT inch up statewide \clip\97\19\masssat.txt Boston Globe
Mass. leads nation in rate of test-taking, lags in math By Jordana
Hart, Globe Staff, 08/27/97

Boston Schools
White V   535 (460)
Asian V   389 (310)
White M   524 (500)
Black M   392 (340)

\clip\97\19\connsat.txt
http://www.boston.com/dailynews/wirehtml/239/Connecticut_SAT_ scores_improve

__but.htm
Connecticut SAT scores improve, but poor kids still lag [Does not
mention that minorities lag even at high incomes] Associated Press,
08/27/97 01:46

\clip\97\19\moremin.txt [The Philadelphia Inquirer] Page One
Wednesday, August 27, 1997 More minorities taking SAT, but scores
still lagging Math results improved for all students, while verbal
results did not. The College Board also found indications of grade
inflation.

\clip\97\19\satcalf.txt SAT Scores Rise, but Trouble Spots Remain
Wednesday, Los Angeles Times August 27, 1997

\clip\97\19\satmath.txt New York Times August 27, 1997 S.A.T. Math
Scores Continue to Rise, Board Reports


@@Rate taking


 
@@recenter
@@norm - 1996 recentering
\clip\96\03\newscale.txt USA Today SAT scores use new scale

Official explanation:

     The College Board recentered the score scale of the SAT I:
Reasoning Test in 1995, re-establishing the original mean score of
500 on the 200-800 scale, in order to maintain the test's statistical
integrity and predictive validity. The scale had not been
recalibrated since 1941 when it reflected the norm of some 10,000
students from predominantly private secondary schools who applied to
the nation's most selective private colleges and universities. As
mean scores shifted below 500, the scores became stretched in the
upper half and compressed in the lower half, an intolerable
compromise to statistical standards.

     Now that scores are recentered on the renormed SAT I, they
reflect the more than two million students who take the test today, a
true cross section of a more diverse college-bound population.

     Although a student's score might change after recentering, the
rank order of scores, expressed as percentiles, stays the same. What
is more, a specific score on the verbal test now has the same
relative position and meaning as that score on the math test, so that
a 450 on verbal and math signifies comparable performance in both
areas. Just before recentering, when verbal and math means were out
of alignment, a score of 450 meant above-average performance on
verbal and below-average performance on math. Recentering permits
legitimate comparison of verbal and math scores and reduces earlier
confusion, but it has no effect on historical score trends, which
remain the same, or on the difficulty level of the test and the
relative standing of students to each other.

     Recentering has improved the quality of this sophisticated
instrument, maintaining its effectiveness as an objective standard in
American education.

asian.education.test.norm "Tally Change, not IQs boosting SAT scores"
USA Today May 31, 1995 Chart that gives equivalence from old to new
scores:

http://professionals.collegeboard.com/data-reports-research/sat/equivalence-tables

Conversion chart from new to old:
New->Old
Verbal        Math
730-800 -70   780-800 -20
700-760 -60   700-690 +10
600-670 -70   600-600 - 0
500-580 -80   500-520 -20


\doc\96\06\newsat.wk1 As cited in USA Today 08/23/96

Conversion of SAT scores from pre-1996 to recentered:

    New Dif New Dif
Old Verbal  Math
 800 800   0 800   0
 790 800  10 800  10
 780 800  20 800  20
 770 800  30 790  20
 760 800  40 770  10
 750 800  50 760  10
 740 800  60 740   0
 730 800  70 730   0
 720 790  70 720   0
 710 780  70 700 -10
 700 760  60 690 -10
 690 750  60 680 -10
 680 740  60 670 -10
 670 730  60 660 -10
 660 720  60 650 -10
 650 710  60 650   0
 640 700  60 640   0
 630 690  60 630   0
 620 680  60 620   0
 610 670  60 610   0
 600 670  70 600   0
 590 660  70 600  10
 580 650  70 590  10
 570 640  70 580  10
 560 630  70 570  10
 550 620  70 560  10
 540 610  70 560  20
 530 600  70 550  20
 520 600  80 540  20
 510 590  80 530  20
 500 580  80 520  20
 490 570  80 520  30
 480 560  80 510  30
 470 550  80 500  30
 460 540  80 490  30
 450 530  80 480  30
 440 520  80 480  40
 430 510  80 470  40
 420 500  80 460  40
 410 490  80 450  40
 400 480  80 440  40
 390 470  80 430  40
 380 460  80 430  50
 370 450  80 420  50
 360 440  80 410  50
 350 430  80 400  50
 340 420  80 390  50
 330 410  80 380  50
 320 400  80 370  50
 310 390  80 350  40
 300 380  80 340  40
 290 370  80 330  40
 280 360  80 310  30
 270 350  80 300  30
 260 340  80 280  20
 250 330  80 260  10
 240 310  70 240   0
 230 300  70 220 -10
 220 290  70 200 -20
 210 270  60 200 -10
 200 230  30 200   0
\

http://professionals.collegeboard.com/data-reports-research/sat/equivalence-tables/sat-score
Use this table to convert individual SAT I: Reasoning Test scores on
the original scale to equivalent scores on the recentered scale. 

Verbal Verbal 
Original Recentered Original Recentered 
800  800  500  580  
790  800  490  570  
780  800  480  560  
770  800  470  550  
760  800  460  540  
750  800  450  530  
740  800  440  520  
730  800  430  510  
720  790  420  500  
710  780  410  490  
700  760  400  480  
690  750  390  470  
680  740  380  460  
670  730  370  450  
660  720  360  440  
650  710  350  430  
640  700  340  420  
630  690  330  410  
620  680  320  400  
610  670  310  390  
600  670  300  380  
590  660  290  370  
580  650  280  360  
570  640  270  350  
560  630  260  340  
550  620  250  330  
540  610  240  310  
530  600  230  300  
520  600  220  290  
510  590  210  270  
      200  230

Math 
Original Recentered Original Recentered 
800 800 500 520 
790 800 490 520 
780 800 480 510 
770 790 470 500 
760 770 460 490 
750 760 450 480 
740 740 440 480 
730 730 430 470 
720 720 420 460 
710 700 410 450 
700 690 400 440 
690 680 390 430 
680 670 380 430 
670 660 370 420 
660 650 360 410 
650 650 350 400 
640 640 340 390 
630 630 330 380 
620 620 320 370 
610 610 310 350 
600 600 300 340 
590 600 290 330 
580 590 280 310 
570 580 270 300 
560 570 260 280 
550 560 250 260 
540 560 240 240 
530 550 230 220 
520 540 220 200 
510 530 210 200 
    200 200 


Source: The College Board.
http://167.8.29.8/life/dcovthu.htm
http://167.8.29.8/life/lds012.htm

RECENTERING INCREASED 800 SCORES BY 17 TIMES.
http://members.xoom.com/griffedulion/women_and_minorities_in_science.htm
La Griffe du Lion Volume 1, Number 1 September 1999 There were other
only rarely mentioned consequences.  The new scoring system blurs the
line between excellent and exceptional.  Truly special students get
lost in the crowd of very good students.  In the year just before the
SAT was recentered, 32 students nationwide scored a "perfect" 1600 on
the combined math and verbal exams. The following year, after
recentering, 545 students scored 1600, about 17 times more than under
the old system.  The SAT no longer distinguishes high-scoring
minorities from higher achieving Whites and Asians.


@@Requirement

%%Against

UC PREZ WANTS TO DUMP SAT
z48\clip\2001\01\nosat.txt STANDARDIZED TESTS AND ACCESS TO AMERICAN
UNIVERSITIES(1) Richard C. Atkinson President of University of
California http://www.ucop.edu/ucophome/pres/comments/satspch.html I
recommended that all campuses move away from admission processes that
use narrowly defined quantitative formulas and instead adopt
procedures that look at applicants in a comprehensive, holistic way.

BLACK MCWHORTER SAYS KEEP SAT
\clip\2001\03\mcwhsat.txt The Chronicle of Higher Education March 9,
2001 Eliminating the SAT Could Derail the Progress of Minority
Students By JOHN H. McWHORTER promising students whose grades have
suffered because of external factors are as likely to underperform on
those tests as on the SAT. Eliminating the SAT, which attempts to
measure aptitude and not exposure to specific course content, would
make it harder to identify such students

%%Black

GETTING RID OF SAT WOULD REDUCE BLACK ELIGIBILITY
z48\doc\web\2001\03\izumsat.txt
http://www.pacificresearch.org/action/action65.html
By LANCE IZUMI, Pacific Research Institute
Keep the SAT as a UC Admissions Requirement
the minimum GPA for admissions would have to be raised to 3.65.11
According to the study, "eligibility rates of white graduates
increased by 17 percent (from 12.7 to 14.8) and the Latino
eligibility rate increased by 5 percent (from 3.8 to 4.0)."12
However: "In contrast, proportionately fewer African-American
graduates within the fully and potentially eligible groups had
achieved a 3.65 or greater GPA. The African-American eligibility rate
falls from 2.8 percent to 2.3 percent, representing an 18 percent
decline."13

MOST BLACK COLLEGES REQUIRE SAT OR SIMILAR TESTS
Jim Woodhill 3/2001: Historically Black Colleges/Universities
(HBCUs).  The HBCUs do use the SAT, led by the top 3, Howard,
Morehouse and Spelman. .  Even HBCU's of lesser academic quality like
FAMU and Stillman require SAT or ACT test scores, I was told.


@@Overprediction

SAT SCORES OVERPREDICT MINORITY PERFORMANCE
\clip\97\19\caseend.txt http://www.usnews.com/usnews/issue/23comm.htm
The case for ending racial preferences by James Q Wilson Yet,
according to many studies, SAT scores [and grades] tend to
overpredict black performance in college. Blacks do not do as well as
their SAT scores predict.

@@parental education

In general, more education is better, but it does not hold across
races. Best educated blacks still equal to white hs dropouts, Asian
bachelors equal to white grad, Asian no diploma = white associate.

 \doc\95\12\sat95bk.wk1
 1995 Scholastic Aptitude Test Breakdown by Ethnicity
 Data from the College Board
 Data Entry and Analysis by Arthur Hu

Rank
592  Asian Graduate
550  Asian Bachelors
545  White Graduate
459  White HS Diploma
478  Asian No HS
-----Worst Asian
472  White Associate
-----Best Black
438  Black Graduate
418  White No HS
409  Black Bachelors
347  Black No HS

                White          Black             Hisp             Asian
 Parental Educa pct  verb math pct verbal math   pct verbal math  pct verbal

 math
 No High School   1  374  418   6    308    347  28    331   389  11    338
 478
 HS Diploma      34  414  459  50    340    371  40    376   422  26    382
 502
 Associate        8  425  472  10    354    384   7    390   435   5    398
 502
 Bachelors       30  460  513  20    379    409  14    420   468  30    426
 550
 Graduate        27  490  545  14    406    438  11    430   479  28    482
 592
 Beyond HS       65            44                32               63

 Comparative Distrib White         Black             Hispanic         Asian
 No High School Dipl 1.0             6.0              28.0             11.0
 HS Diploma          1.0             1.5               1.2             -1.3
 Associate           1.0             1.3              -1.1             -1.6
 Bachelors           1.0            -1.5              -2.1              1.0
 Graduate            1.0            -1.9              -2.5              1.0

Asians and Hispanics are most likely to have parents without a HS
diploma Asian are no more likely than whites to have college educated
parents, which fully explains their high college-going rate
Both groups are 2/3 beyond HS

 Hispanic have the highest rate of non-HS diplomas (1 in 4), but Asians
 are next
 Black with graduate education = white or Asian with no HS, Hisp Assoc
 White graduate education = Asian Bachelor
 Asian with no HS = White Associate

@@Population

http://www.collegeboard.org/press/senior99/html/990831a.html
zip36\clip\99\17\acad.htm
Minority students were one-third of SAT takers in the Class of 1999,
up from 25 percent 10 years ago.  SAT takers were 11 percent African
American, 9 percent Asian American, 4 percent Mexican American, 3
percent Hispanic/Latino, and 1 percent Puerto Rican. English was not
the native language for 39 percent of Asian American and
Hispanic/Latino, 28 percent of Mexican American, and 25 percent of
Puerto Rican students.

The SAT verbal average stayed at 505 for a fourth year, and did not
change this year for men (509), women (502), African American (434),
Asian American (498), Mexican American (453), and Other (511)
students.  American Indian/Alaskan Natives experienced the largest
one-year increase (4 points), followed by Puerto Ricans (3 points),
Hispanic/Latinos (2 points), and whites (1 point). Average scores for
women rose for American Indians (4 points) and Hispanic/Latinos (3
points), and fell 1 point to 495 for Asian Americans.  Average SAT
verbal scores for men rose 6 points for Puerto Ricans, 2 points for
American Indians and Asian Americans, and 1 point for
Hispanic/Latinos.

@@Private / Public

PRIVATE SCHOOLS HAVE HIGHER TEST SCORES \clip\97\19\satcalf.txt SAT
Scores Rise, but Trouble Spots Remain Wednesday, Los Angeles Times
August 27, 1997

1997 SAT Calf results:
Parochial School    526V 518M
Other Private       574V 582M

"as much as 84 points higher than public school averages"

http://www.capenet.org/facts.html#anchor1417287
CAPE Private School Facts
NCES reports 87.5 percent of private high school students apply to
college, compared to 57.4 percent of public high school students
(SASS). And reports from the College Board indicate private schools
account for 17 percent of all SAT test takers, even though they
enroll only 7.5 percent of the country's high school students.
Further, SAT scores for private school students are well above the
national average for every racial, ethnic, and income group.

 Class of '99 SAT Total Scores 
 Public Schools 
 1009 
 Private Schools 
 1068 

@@Race nonresponder

z75\clip\2004\04\satrace.txt
As Data Show Fewer Report Race, Minority Scores on SAT Questioned
Omissions Mean 'Achievement Gap' Is Difficult to Measure
By Jay Mathews
Washington Post Staff Writer
Sunday, April 4, 2004; Page A06
But buried in the [last august college board] ..  A record portion of
the test-takers, 25 percent, had declined to disclose their ethnicity.
Now, an independent analysis of years of SAT data suggests that
growing American disdain for racial categories may cast doubt on how
accurately the SAT can measure the achievement gap.
..the work of Howard Wainer, currently a professor of statistics at
the University of Pennsylvania's Wharton School, with SAT data from
1981 to 1985.  Wainer found that 12 percent to 14 percent of SAT
applicants failed to report their ethnicity. 
..The Shaker Heights non-responders were majority female, 82 percent
white and 18 percent African American, while the portion of total
students taking the SAT at Shaker Heights High was 59 percent
non-Hispanic white and 36 percent African American.
...the 2003 non-responder group had
more males than females, even though most SAT takers are female, and had an
average SAT verbal score of 510, slightly above the national average of 507.



@@Race distribution

Everyone knows that Blacks SAT scores are low, Whites are high, but
Asians are more likely to be found at the top AND bottom combined
scores.

%%California

ASIANS WOULD BE 45% OF TOP DECILE OF SATii WRITE/MATH/3RD TEST IN 2000
The study report is at
http://www.ucop.edu/sas/research/researchandplanning/pdf/sat_study.pdf (500K), with
some supporting documents at
http://www.ucop.edu/sas/research/researchandplanning/supporting.htm

Table 11
Racial/Ethnic Composition of Top Decile of
SAT I vs. SAT II Takers, Fall 2000 UC Applicant Pool

                        SAT I          SAT II         SAT II
                     Verbal+Math    Writing+Math  Writing+Math+3rd_Test
African_American         0.7%           0.7%            0.6%
American_Indian          0.3%           0.3%            0.2%
Asian_American          41.4%          42.3%           45.0%
Chicano/Latino           2.5%           2.5%            3.0%
White                   40.6%          40.2%           37.4%
Other/Unknown           14.5%          14.0%           13.7%



@@race percentile

http://www.collegeboard.com/prod_downloads/highered/ra/sat/CR_M_%20W_PercentileRanksGenderEthnicGroups.pdf
saved: \clip\2011\01\
gdc: SAT Percentile by Ethnic Group
SAT percentile ranks by gender and ethnic groups
based on 2006 cohort

Math 
   White  Asian Black
800   99    98     99 Top score isn't even top 99 for Asians!
750   98    92     99
700   94    81     99
650   85    68     98
600   71    53     94
550   54    40     88 Top 10 Black = White 54 Asian 40
500   35    26     76
450   19    15     59 Black 50 ~= White 13, as hard as IQ test
400    8     7     37
350    3     3     19
300    1     1      8
250   1-            3
200

Mean 536          429
SD   103          100
N    825k  138k  150k

As many Asians as blacks took the test, Asians about 5% ,blacks 10% of population
Asians get better scores even though twice as many took the test, so difference
 is even greater for the general population.





@@race scores

%%National Race

http://www.asianweek.com/2010/10/06/asian-americans-are-1-on-the-2010-sat/
Asian Americans are #1 on the SAT
By Arthur Hu– October 6, 2010(Edit Post)
Posted in: Education, Hu's on First, In the Classroom, National, News

The 2010 SAT college entrance test results are in. Historically, back
in the day when I went to MIT, the common wisdom was that Asians
scored better in math, but were disadvantaged because of the verbal
scores. The men outscored women in math, but women were better in
verbals. I’m probably the only guy who noticed that sometime in the
1990s, Asians started outscoring whites when the verbal gap shrank to
less than the math advantage, probably because of more American born
vs immigrants taking the test.  This year, the combined test scores
for Asians at 1636 is 56 points ahead of whites, but Mexicans were 211
and blacks were 303 points behind.

In math, there is no surprise in math that Asians overall at 591 beat
whites by 55 points, and Asian men scored highest at 605. But I’m
still the only guy who has noticed that Asian females at 577 scored
higher than white men at 555.  Now if colleges like MIT given women
admission preferences (their admit rate is about double the rate for
men) does it make sense to give Asian women a break if they get higher
test scores and apply in higher proportions than white men? And if we
break out any over-represented subgroup of women, what about other
groups such as Jews or any other religious/ethnic group?

In critical reading, men at 503 now outscore women by 5 points, while
Asian men are just about even at 1 point ahead of their women. Asians
overall lagged whites in reading, but not by much at just a 9 point
gap.  

 Overall, females did 12 points better than males in writing. However,
if writing was introduced to make the test more fair, it has proven if
anything to be an advantage for Asians who at 526 were 10 points ahead
of whites at 516. It has only made things worse for minorities such as
the African Americans, as their 420 was the worst of their 3 SAT
scores.

Here are some scores ranked from high to low:

math
Asian male   605
Asian        591 +55
Asian female 577
White male   555
White        536 =0
Male         534
white female 519
Female       500
black male   436
black        428
black female 422writing
asian female 532
asian        526 + 10
white female 523
asian male   520
white        516 = 0
white male   508
female       498
male         486reading
white male   530
white        528
white female 526
asian male   520
asian female 519
asian        519combined
asian   1636 +56 (mostly math)
white   1580 = 0
mexican 1369 -211, about 4x white-asian gap.
black   1277 -303, about 6x the white-asian gap.

The 2010 College Board Reports can be found here:
http://professionals.collegeboard.com/data-reports-research/sat/cb-seniors-2010


YEAR 2010
College Board Reports 
http://professionals.collegeboard.com/data-reports-research/sat/cb-seniors-2010
See Tables 8, 9, 10

Men higher than women BUT....
Asians men highest 605, Asian females at 577 higher than white men at 555
Asians 591 beat white by 
math 
Asian male   605
Asian        591 +55 
Asian female 577
White male   555
White        536 =0 
Male         534
white female 519
Female       500
black male   436
black        428
black female 422

Writing - may have been introduced to penalize Asians and help blacks,
but Asians are highest scoring race beating whites by 10 points with
526 vs white 516. Overall, females do better than males in writing.

writing 
asian female 532
asian        526 + 10
white female 523
asian male   520
white        516 = 0
white male   508
female       498
male         486

In critical reading, men at 503 now outscore women by 5 points, while
Asian men are just about even at 1 point ahead of their women. Asians
overall lagged whites by 9 points in reading. 

reading 
white male   530
white        528
white female 526
asian male   520
asian female 519
asian        519
men          503
women        498

combined 
asian   1636 +56 (mostly math) 
white   1580 = 0 
mexican 1369 -211, about 4x white-asian gap.
black   1277 -303, about 6x the white-asian gap.

Demographic Information
SAT: Mean Scores by Gender Within Ethnicity
Table 8: Total Mean Scores by Ethnicity
SAT Test-Takers Critical Reading Mathematics Writing
Test-Takers Who Described Themselves As: Number Pct Mean SD Mean SD Mean SD
American Indian or Alaska Native 8,550 1 485 107 492 104 467 102
Asian, Asian American, or Pacific Islander 166,064 11 519 122 591 125 526 126
Black or African American 196,961 13 429 98 428 97 420 93
black 420 writing is LOWEST  of 3 scores  
Mexican or Mexican American 85,761 6 454 98 467 97 448 92
Puerto Rican 24,365 2 454 103 452 103 443 100
Other Hispanic, Latino, or Latin American 112,254 7 454 104 462 104 447 100
White 838,235 54 528 102 536 102 516 102
Other 48,702 3 494 118 514 120 492 117
No Response 67,098 4 487 132 514 128 481 128
Total 1,547,990 100 501 112 516 116 492 111

Table 9: Male Mean Scores by Ethnicity
SAT Test-Takers Critical Reading Mathematics Writing
Test-Takers Who Described Themselves As:
                                 Number Pct Mean SD  Mean SD Mean SD
American Indian or Alaska Native  3,848   0  487 110 508  110 459 105
Asian, Asian American, or Pacific Islander 81,208 5 520 122 605 123 520 126
Black or African American 86,509 6 426 99 436 101 408 93
Mexican or Mexican American 38,157 2 459 100 486 100 444 93
Puerto Rican 10,820 1 456 105 468 107 437 100
Other Hispanic, Latino, or Latin American 47,751 3 460 107 484 108 444 101
White 397,002 26 530 104 555 104 508 103
Other 21,314 1 494 121 534 122 484 119
No Response 34,184 2 483 134 526 131 470 127
Total 720,793 47 503 114 534 118 486 112


Table 10: Female Mean Scores by Ethnicity
SAT Test-Takers Critical Reading Mathematics Writing
Test-Takers Who Described Themselves As: Number Pct Mean SD Mean SD Mean SD
American Indian or Alaska Native 4,702 0 484 104 479 98 474 99
Asian, Asian American, or Pacific Islander 84,856 5 519 122 577 125 532 125
Black or African American 110,452 7 432 97 422 93 428 93
Mexican or Mexican American 47,604 3 451 96 451 91 451 91
Puerto Rican 13,545 1 452 102 438 98 448 100
Other Hispanic, Latino, or Latin American 64,503 4 449 103 446 99 449 100
White 441,233 29 526 101 519 98 523 101
Other 27,388 2 494 116 498 115 498 115
No Response 32,914 2 491 131 501 124 493 127
Total 827,197 53 498 111 500 112 498 111


MINORITY UNDER AND ASIAN OVER GAP INCREASES 1993-2003
z70\clip\2003\08\sathi.txt
SAT Scores Are Highest Since 1974
Test Gap Between Whites, Minorities
Widens Among College Freshmen
By JUNE KRONHOLZ

Staff Reporter of THE WALL STREET JOURNAL
Wall Street Journal chart Aug 27, 2003 p. D2
"SAT Scores Are Highest Since 1974"
http://www.collegeboard.com/prod_downloads/about/news_info/cbsenior/yr2003/pdf/2003_TOTALGRP_PRD.pdf
z70\clipim\2003\08\26\2003_TOTALGRP_PRD.pdf
2003_WASHINGTON.pdf

z70\doc\web\2003\08\satrace.wk1
Average combined verbal and math scores for incoming college freshman
                  1993    2002    20031 yr    10 yr
Asian-Am          1042    1070    1083      13      41
White             1037    1060    1063       3      26
AmIndian           953     962     962       0       9
MexAmerican        910     903     905       2      -5
AfriAmerican       850     857     857       0       7
Males             1030    1041    1049       8      19
Females            981    1002    1006       4      25
All               1003    1020    1026       6      23

Gap vs White
Asian-Am             5      10      20      10      15
White                0       0       0       0       0
AmIndian           -84     -98    -101      -3     -17
MexAmerican       -127    -157    -158      -1     -31
AfriAmerican      -187    -203    -206      -3     -19
Males                0       0       0       0       0
Females            -49     -39     -43      -4       6
All                -34     -40     -37       3      -3
Wall Street Journal chart Aug 27, 2003 p. D2
"SAT Scores Are Highest Since 1974"

Std Dev = 102 math, 102 verbal		
Blacks are down by about 1SD in both scores.		

Asian Females score above white males in math		
SAT Math		
Asian M	592	
Asian F	560	
White M	552
White F	518

\doc\college\sat\sat89.wk1 
Figures from college board 1989
SAT Distribution 1989
Percent
Math     All      White   Black    Asian    Hisp     Mex      Latino   PR       Mex
Top                  68.7%     0.5%    20.8%     1.4%     0.3%     0.7%     0.1%     0.5%
500-540              76.5%     4.6%     2.6%     4.3%     1.4%     1.6%     0.7%     2.0%
Bot                  35.2%     3.1%     3.9%    10.1%     2.3%     3.6%     3.4%     3.0%

Verb     All      White   Black    Asian    Hisp     Mex      Latino   PR       Mex
Top                  75.3%     0.5%    12.6%     0.8%     0.2%     0.6%     0.1%     0.1%
500-540               9.4%     9.6%     5.1%     3.3%     0.9%     1.4%     0.5%     1.4%
Bot                  13.9%     9.7%    16.9%    10.7%     1.8%     3.9%     3.0%     3.7%

Both     All
Top                  69.2%     0.5%    20.1%     1.3%     0.3%     0.7%     0.1%     0.5%
500-540              45.0%     6.9%     3.8%     3.8%     1.1%     1.5%     0.6%     1.7%
Bot                  17.7%     8.5%    14.6%    10.6%     1.9%     3.9%     3.1%     3.6%
                  White   Black    Asian    Hisp     NA       Latino   PR       Mex
US Population        72.0%    14.0%     3.0%    12.0%     0.8%
Comb Relative        96.2%     3.6%   670.8%    10.9%    35.3%
                     62.5%    49.6%   126.1%    31.8%   143.1%
                     24.5%    61.0%   486.3%    88.0%   241.1%

White Index: W1.00 B-2.54 H-3.52 A-20.2 

-> Asians are 6.7 times at the top, but 4.9 times at the bottom, they are 
about equal to population at the middle.
-> Blacks are 0.036 times their population at the top, half at the 
middle, but 2.5 times compared to white share at the bottom.
-> Whites are only 1/4 their population at lowest SAT scores.

Registered           75.0%    10.0%     7.0%     6.0%     2.0%     2.0%     1.0%     3.0%
Population          104.2%    71.4%   233.3%    50.0%   250.0%
Parents college       0.55     0.55     0.33     0.58                                0.23
--->!! Asians twice as likely to have college parents
--->!! Asians going to college have same college parent rate as whites!


zip36\clip\99\17\ethgap.txt Los Angeles Times Wednesday, September 1,
1999 Ethnic Gap Widens in SAT College Exam Scores Education: Results
raise alarms that schools are failing to prepare nonwhites. The good
news is that more minorities are taking the entrance tests. By MARTHA
GROVES, RICHARD COOPER, Times Staff Writers

ASIAN SAT VERBAL GAP DOWN BY 30%
http://www.collegeboard.org/press/senior99/html/satt4.html
zip36\clip\99\17\satt4.htm,wk1
Asian verbal up 15 vs. 4 white, math up 15 vs 13 white
Verbal gap 1989=40 1999=28
Asian overall 1058 vs white 1055

\clip\98\11\satgap.txt
http://www.chronicle.com/daily/98/09/98090202n.htm
Chronicle of Higher Education, September 2, 1998
SAT Scoring Gap Widens Along Gender, Geographic, and Racial Lines
(ranked by math)
                                               Verbal    Math
                      Asian American             498     562
                      White                      526     528
                      American Indian            480     483
                      Other Hispanic and Latino  461     466
                      Mexican American           453     460
                      Puerto Rican               452     447
                      Black                      434     426

Average SAT scores of
college-bound seniors, by race/ethnicity: 1976-95

original: http://www.ed.gov/NCES/pubs/ce/c9622d02.html

--------------------------------------------

ASIANS AND MEN SCORE BEST ON SAT, BLACKS WORST.
\doc\web\98\03\sat96.wk1
National Average Scores on the SAT
by Sex, Racial and Ethnic Group, 1996

Source: The Chronicle of Higher Education, 1997

Ranked by Combined score

        Verbal  Math    Combined
Asian        496     558    1054
White        526     523    1049
Male         507     527    1034
All          505     508    1013
Female       503     492     995
Mexican      455     459     914
Black        434     422     856

\doc\web\97\06\satrk.wk1
1995 SAT Scores by Ethncity, Ranked and Compared
Analysis by Arthur Hu

Verbal    Scores   Index Math        Score  Index
White        448    1.00 Asian        538    1.08
Asian        418   -1.07 White        498    1.00
NativeAm     403   -1.11 NativeAm     447   -1.11
Other Hi     389   -1.15 Other Hi     438   -1.14
Mexican      376   -1.19 Mexican      426   -1.17
Puerto Rican 372   -1.20 Puerto Rican 411   -1.21
Black        356   -1.26 Black        388   -1.28

Asians and Whites trade ranks for verbal and math, otherwise, then
come Native Americans just under Whites, other Hispanics, many Cuban,
Mexicans, Puerto Ricans just slightly under Mexicans, and Blacks get
the worst test averages.

Source: College Board

--------------------------------------------------
 \doc\95\12\sat95bk.wk1

 1995 Scholastic Aptitude Test Breakdown by Ethnicity
 Data from the College Board
 Data Entry and Analysis by Arthur Hu

                White          Black             Hisp             Asian
 Language p. 6  pct  verb math pct verb   math   pct verb   math  pct verb
 math
 Overall SAT         448  498        356    388        376   426        418
 538
 Combined            946             744               802              956
 White gap             0    0        -92   -110        -72   -72        -30
  40

As a ratio  W1.00 B-1.27 H-1.18 A1.01

@@Rate

\doc\94\18\priv\calfsat.txt  CALIFORNIA'S SAT SCORES DIP SLIGHTLY
8/25/94 SJM

A higher percentage of Asian-Americans and blacks took the SAT in 1994
than did whites. Nearly 59 percent of Asian-American students took the
test, followed by 40 percent of blacks, 38 percent of whites, and 25
percent of Hispanics.


@@percentile

\doc\96\04\satpct.wk1
1995 SAT Percentile Chart
(before recentering for 1996)

National - All Races

        Percentile
        Verb    Math
        Pct     Pct     Comb    Comb %  Comb pc
750-800   100.0%  100.0%  17,261    0.8%  100.0%    1500
700 -740   99.9%   98.5%  55,846    2.6%   99.2%    1400
650 -690   98.8%   94.4%  98,744    4.6%   96.6%    1300
600 -640   96.3%   87.6% 146,022    6.8%   92.0%    1200
550 -590   91.7%   78.5% 190,070    8.9%   85.1%    1100
500 -540   84.0%   68.4% 249,136   11.7%   76.2%    1000
450-490    73.5%   55.6% 305,443   14.3%   64.6%     900
400-440    58.2%   42.3% 341,427   16.0%   50.3%     800
350 -390   40.2%   28.4% 300,146   14.1%   34.3%     700
300 -340   24.4%   16.0% 244,438   11.4%   20.2%     600
250 -290   12.0%    5.5% 125,335    5.9%    8.8%     500
200-240     4.8%    1.1%  62,118    2.9%    2.9%     400

@@perfect 1600

http://www.collegeboard.com/prod_downloads/about/news_info/cbsenior/y
r2003/pdf/2003CBSVM.pdf
944 students got a 1600 in 2003 out of 1406234 total test takers. 
1 out of 1489 test takers got a 1600. 

La Griffe du Lion: In the year just before the SAT was recentered, 32
students nationwide scored 1600 on the combined math and verbal exams.
The following year, after recentering, 545 students scored 1600. In
those days a bit more than 1 million took SAT I.  Compression made it
possible for very selective schools to admit high scoring minorities
who had suddenly become indistinguishable from higher scoring
non-minorities.



http://www.arthurhu.com/index/test.htm# satprediction
@@Prediction

Myth: Grades alone are a better predictor than SAT scores.
Fact: College board study shows SAT alone to be the better
predictor.

Source: college board
http://www.collegeboard.org/sat/html/admissions/stat003.html
"For Admission Staff: Understanding SAT Program Scores"

Recentered SAT I Scores As a Predictor of College Grades
Mean Correlations with Course Grades:

                        Total               Male              Female
                 OriginalRecentered Original Recentered OriginalRecentered
                  Scale     Scale     Scale    Scale     Scale     Scale
 SAT Verbal        .42       .43       .40      .40       .45       .46
 SAT Math          .46       .46       .44      .44       .48       .49
 SAT Total         .50       .51       .49      .49       .53       .54
 High School GPA   .48       .48       .47      .47       .49       .49
 SAT plus High
 School GPA        .59       .59       .57      .58       .61       .62
 SAT Increment     .10       .11       .11      .11       .12       .13

The new scale has a .51 correlation to course grades, higher than
.48 for grades alone, .49 vs. .47 for men, and an even bigger spread,
.54 vs .49 for women. Studies show that grades alone have less
predictive ability for minority students. The new scales are slightly
better than the old ones with respect to predictive ability.

The college board says to use these validity coefficients to compare
the ability of SAT scores on the original and recentered scales to
predict students' performance in college. The higher the coefficient
(i.e. closer to +1.00) the better the test as a predictor.

Data are from studies conducted on the traditional SAT at 45 colleges
and universities.

@@profile

Students who take the SAT I represent more than 90 percent of new
students who will enter four-year colleges this fall.

overall
http://www.collegeboard.org/press/html/960822a.html
details

http://www.collegeboard.org/press/html/960822b.html

http://www.collegeboard.com/sss/rsc/s/studentprofile2002_2003_partial_28660.xls
z78\clip\2004\05\studentprofile2002_2003_partial_28660

College Bound Seniors by State 2003 / National
http://www.collegeboard.com/about/news_info/cbsenior/yr2003/html/2003reports.html

washington f:\clip\2004\05\2003_WASHINGTON.pdf
W75 A12 B4 Mex3 Lat1 Other 4

CALIFORNIA.pdf
W40 B7 Mex16 Lat5 A23 Other 5 None 
Out of 131,133 46,867 were no response
vs. 52,573 white, 29,567 Asian

\clip\96\03\sat96.htm Student profiles
1996 W69% B11% H8% A7.6% N1% O3%
1987 W78% B 9% H5% A6.6% N1% O1%

@@religion

see religion, college
c:\doc\95\07\relig90.wk1

SAT-Comb  95/90* W1.0 B-1.27 H-1.18 A1.01 J*1.10 I*1.10 *

Note that Asians overall scored only slightly better than whites
overall when both math and verbals are counted. Asians were outscored
by Episcopals, Presbyterians, Black Unitarians, Jews and Hindus. Only
White Unitarians and Quakers outscored Jews.

@@Sample

Sample test info
"10 Real SATs" comes from the College Entrance Examination Board (The
College Board to most).  Most bookstores can order it or you can call
direct to College Board Publications, Two College Way, Forrester Center, WV
 25438-4100, phone 800-323-7155.  ISBN 0-87477-567-8.  $17.95.

@@SCORES

------------------------------------------------------------

\doc\96\03\sat95.xls
Big multi-sheet of SAT by race

\priv\96\14\sat95.met - Care scan of main pages and state

------------------------------------------------------------

%%Year 2008

http://professionals.collegeboard.com/data-reports-research/sat/cb-seniors-2008
College-Bound Seniors 2008
SAT national and state reports provide broad context of student performance
Total Group Report: College-Bound Seniors 2008 

Asians generall lead in math, lag in verbal, but combined scores
matched whites by 1990s.

Now 3 scores, 1610, 27 more than 1583 for whites Blacks 1280, down by
303 - asians overall 1/10 as far ahead as blacks behind

Asians 11% vs blacks 12% of test takers, Asians about 2X pop, blacks
at parity

Asians 581 math, whites 537 (+44), black 426 (-111) Asians 1/3 ahead
as blacks behind

Male 533 vs female 500 (+33)

Asian female 567M vs white male 555M (+12)
asians behind in critical reding 15 pts
Writing test is a big win for Asians -
only 2 down overall. white female 526, but asian female 523, higher than
white or white men
favors females 501 to 488 (+13)

How big is the math gap in income?
Asians go up $70-$100,000 from $60-$80,000 to $160-$200,000 or $0 to $70,000
Blacks go down $200,000 from over $200,000 down to bottom $0-$20,000

Asian math same as A average grade
White math same as below A-
Black math same as C

Asian / white math same as between pre-calculus + calculus taken
black algebra 1

Social science course work highest for European history (543) and
ancient history (524), mostly dead white guys

Scores go up with languages, 574 for over 4 years vs. 443 for 1/2
year or less

Highest math scores are for Chinese 600, Korean 587, Japanese and Latin 556
and Hebrew 552

Studnets who took extra subject tests in language, Latin 671, Chinese
668, Korean 659, Japanese 652, Spanish 555.

Asian parents put their kids through music - over 4 years 537 vs 512 for
1/2 year or less. +25

Computers none-510, 527 (+17) spreadsheet or database, 521 computer
programming

By intended college major, Math (609), Science (586). Low agriculture
(481), construction 460, public administration social services 452,
security protective services 458. 515 is average. Area, Ethnic
Cultural and Gender studies at 521 is just above average

Sorted by SAT math score			
		total	notes
asian	male	596	
asian	total	581	
asian	female	567	asian female > white male
white	male	555	
white	total	537	
Other 	male	533	other race male also high
total	male	533	
white	female	521     34 behind white men	
total	total	515	
Other 	total	512	
amind	male	509	amind male
no resp	male	503	
total	female	500	
Other 	female	496	
no resp	total	492	
amind	total	491	
mexican	male	482	Mexican, other hisp
otherHi	male	482	
no resp	female	482	
amind	female	475	
pr	male	470	
mexican	total	463	
otherHi	total	461	
pr	total	453	
mexican	female	448	
otherHi	female	445	
pr	female	440	
black	male	434	lower than all other minorities
black	total	426	
black	female	420	

Asians have highest math SD = 124, vs 116 total, 103 white 98 black

Press release
http://www.collegeboard.com/press/releases/197846.html
Females have narrowed the performance gap with males in critical
reading, closing the gap to 4 points, compared with 7 points a decade
ago, and females continue to outperform males on the writing section —
by 13 points this year.

Growth among Asian Americans (population) and African Americans (real
improvement, mostly GW Bush years) reached 61 percent and 52 percent,
respectively. 

Of all three sections of the SAT, the writing section is the most
predictive of students’ freshman year college performance for all
students, demonstrating

Female students made up 57 percent of Hispanic and 57 percent of
African American SAT takers in 2008. More than half of the Asian
American (51 percent) and white test-takers (53 percent) were women. 


http://www.nytimes.com/2005/08/31/education/31sat.html SAT Math Scores
at Record High, but Those on the Verbal Exam Are Stagnant By KAREN W.
ARENSON Published: August 31, 2005 The score on the math test was 520,
of a possible 800, up two points from the previous year. The verbal
score was 508, the same as in the previous year. This score is the
highest on the verbal test since 1986 but far below the record of 543
in 1967,

       1995                                     2005
Math   506 508 511 512 511  514 519 516 519 518 520 +14
Verbal 504 505 505 505 505  505 506 504 507 508 508  +4

increase in the number of students who take challenging courses like
pre-calculus, calculus and physics. They said 48 percent of the
students who took the test had studied pre-calculus, up from 37
percent a decade earlier. The proportion who had studied calculus rose
to 27 percent from 22 percent over the same period, while those who
had studied physics jumped to 51 percent from 46.


flat verbal scores reflected a lack of agreement about what should be
taught. 

critics fairtest: show that college readiness skills, as measured by
these indicators, have not improved significantly."

62 percent of the Asian-American students had taken pre-calculus,
compared with only 32 percent of African-Americans and 34 percent of
Puerto Ricans.

Students from families with incomes of $10,000 to $20,000, for
example, earned an average verbal score of 443 and an average math
score of 463. Students from families with incomes above $100,000
scored more than 100 points higher: their average verbal score was
554, and their average math score 565.

Students were more likely to have taken demanding courses like
pre-calculus if their parents were college graduates or earned high
incomes


Asian-Americans the stars in SAT math
Gale Holland 
Tuesday, September 02, 2008
http://www.thestandard.com.hk/news_detail.asp?we_cat=16&art_id=71045&sid=20421134&con_type=1&d_str=&fc=4
Hong Kong The Standard 
Asian-American students topped the math scores, with an average of
564, compared with 549 for white students, 453 for Hispanic students
and 429 for black students, from among 1.5 million students who took
the Scholasic Aptitude Test admissions exam for 2008 in the United
States. ...Minority participation was 40 percent, up from 33 percent
10 years ago. But achievement gaps between students of different
ethnic backgrounds persisted.


\clip\2008\09\sat08.txt
Class of '08 Fails To Lift SAT Scores
Flat Results Fuel Debate Over Costly Test-Preparation Courses, Provide
Ammunition
For Critics of No-Child-Left-Behind Policy
By JOHN HECHINGER
August 27, 2008, Wall Street Journal
oAfrican-American and Hispanic students lagged behind the general 
population. black scores declined, 72 pts below national avg.

oAsian-American students posted strong scores, especially in math.

"Asian-Americans continue to post stellar results. On the math
section, the group achieved an average score of 581, 66 points better
than the average.  Asian-Americans also outperformed in
critical-reading and writing, though by less.  "

"Some students have said that college-admissions offices have placed
unofficial quotas on the number of spots allowed for Asian-American
students. Using the group's superior SAT scores as evidence, these
critics argue that Asian-Americans are held to a higher admissions
standard at top colleges. The U.S. Department of Education is
investigating such an allegation at Princeton University. The school
denies any discrimination and says test scores are only part of its
admissions criteria.  "

Among Asian-American families "there's a strong understanding that
test preparation is an important part of success," he says. 

Asian-American students tend to take harder courses, such as 
higher-level math and those given to prepare for Advanced Placement exams. 
"The country really needs to focus on this simple and honest truth -- by 
doing hard work in schools, that's how you get a better education," he 
says. "That's how you get better SAT scores."

2008: SAT Scores Flat as Test-Taking Edges Upward
http://www.edweek.org/ew/articles/2008/09/03/02sat.h28.html?tmp=219949777
** Asians post gains in math and writing **
ASIANS GAIN MATH AND WRITING, MEN STILL AHEAD IN MATH
White students and those identified as “Asians, Asian-Americans, or
Pacific Islanders” posted the only gains on the math section; all
other groups showed point losses in the low single digits. And on the
writing section, only students in the Asian/Pacific Islander category
improved on last year’s crop of scores.

the mean math score on the SAT for men among this year’s graduating
class was 533 out of a possible 800 points, 33 points higher than the
mean score for women. That’s the smallest difference in years, but
still substantial.

\DOC\94\14\SAT95.WK1 - SAT scores and improvement

------------------------------------------------------------
\priv\95\01\confwork.doc

The national norm for math on the Scholastic Aptitude Test is 467 out
of a possible 800.  In 1981, Asian Americans averaged 513.  In
California a remarkable 68% of Japanese-born students scored over 600.
as did 66% of students born in Korea. (comment vs. 35.6% of all US
Asians, 28021 out of 78,693, so foreign born Korean or Japanese are
twice as good as average, probably comparable to Chinese)

"Asian-born students head for the head of the class"
By Ellie McGrath Time March 28, 1993 p 52

doc\college\sat\sat.wk1 1983
distribution

doc\college\sat\sat88.wk1 distributions by race 1983, 1988

*doc93a\sat93us.xlw - SAT distributions 1993 NEEDS WORK!

*doc\94\8\sat93us.xlw - look at  sheet

doc\95\01\sat93us.xlw

1993 SAT math 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

Patti Hausman: the average difference between Asian males and Asian
females (47 points) is essentially identical to the average
difference (46 points) between white males and white females.
[despite cultural differences and social conditioning, this is a very
precise difference]

\doc\95\01\SATIQ.WK1
SAT as white percentile, and IQ
distribution Analysis By Arthur Hu.

The SAT test appears to be about as difficult as IQ tests with a
similar 1 standard deviation spread

Elementary and high school NAEP test are much closer in performance,
at a 35-40th percentile level

Differences in races is dependent on how difficult the test is.
Simple tests where everybody passes will be much closer than those
that flunk most who take the test.

 1993 SAT Math Scores by Percentage, Percentile and IQ
 vs. White = 50 percentile, 100 IQ
 SCORES     NATIONA WHITE   BLACK   MEXICAN ASIAN   NATIVE  PUERTO
 MEAN           478     494     388     428     535     447     409
 Percentile              50      15      28      65
 IQ Scale               100      85      92     106
 Note Chinese / Japanese / Korean IQ would be 112
 due to overrepresentation at MIT and Berkeley

 At each range, the race composition of students in the range
 SCORES     NATIONA WHITE   BLACK   MEXICAN ASIAN   NATIVE  PUERTO
 TOTAL       100.0%   64.2%    9.9%    3.1%    7.5%    0.7%    1.2%
  750-800    100.0%   60.0%    0.6%    0.6%   25.3%    0.3%    0.2%
  700-740    100.0%   65.9%    1.0%    0.8%   18.7%    0.3%    0.3%
  650-690    100.0%   70.3%    1.8%    1.2%   13.7%    0.4%    0.3%
  600-640    100.0%   72.9%    2.6%    1.5%   10.6%    0.5%    0.5%
  550-590    100.0%   73.5%    3.6%    2.0%    8.5%    0.6%    0.6%
  500-540    100.0%   72.2%    5.3%    2.6%    7.1%    0.7%    0.9%
  450-490    100.0%   68.8%    8.0%    3.2%    6.1%    0.8%    1.1%
  400-440    100.0%   64.3%   11.2%    3.8%    5.5%    0.8%    1.4%
  350-390    100.0%   58.0%   15.6%    4.5%    4.7%    0.9%    1.7%
  300-340    100.0%   49.0%   21.3%    4.9%    4.5%    0.9%    2.2%
  250-290    100.0%   40.1%   27.0%    4.6%    4.3%    0.9%    2.7%
  200-240    100.0%   31.6%   30.5%    5.1%    4.7%    1.0%    3.2%

 Asians are 25% of top math scores overall
 Blacks are 31% of bottom math scores, 0.6% of top


SAT Scores Vs. US Population

\doc\web\98\02\satrate.wk1
Analysis By Arthur Hu 1/19/98
Index: W1.00 B-21.77 H-7.84 A8.54 N-2.90
1993 High Math SAT Scores Compared to 18 yr old Population
SCORES  WHITE   BLACK   ASIAN   NATIVE  Mexican Puerto
TOTAL     64.20%   9.90%   7.50%   0.70%   3.10%   1.20%
750-800   60.00%   0.60%  25.30%   0.30%   0.60%   0.20%
Pop       68.90%  15.00%   3.40%   1.00%   5.40%   1.10%
Rate       0.871   0.040   7.441   0.300   0.111   0.182
Index       1.00  -21.77    8.54   -2.90   -7.84   -4.79

doc\94\5\priv\college.txt US News survey SAT ranges
doc938\satrac93.xls US 93 test averages

@@ses

Patricia Hausman" 
from Jensen's Straight Talk about Mental Tests (page 148).
"Studies of the SAT's predictive validity for college students
from different socioeconomic (SES) backgrounds show no
indication of bias disfavoring (i.e. underestimating) applicants
of lower SES.  The SAT does not consistently under- or over-
estimate college grades of students from a wide range of SES . . . 
The tests can 'read through' the veneer of social class background 
to identify academic talent more objectively and accurately than 
teachers' marks of interviews by college admissions officers."


@@Standard Deviation

ASIAN AMERICANS HAVE THE LARGEST SAT STANDARD DEVIATION
\doc\96\03\sat95.xls
1995:
Group   Verbal       Math
        Male Female  Male Female
Asian   127  127     134  133
White   107  104     122  115
Mexican 102   97     115  100
Black    97   97     108   96

Asians have higher average math and combined SAT scores, but they
also have a larger standard deviation on both math and verbal, which
reflects poor verbal performance and low education of low-income
Asian immigrants combined with high performing educated immigrants.

Mexican and Black std devs are simply lower since they don't have a
high performing population to balance out the white curve.
Analysis by Arthur Hu


@@state scores

It is misleading to compare states by SAT since states with low SAT
taking rates will be dominated by high scoring students.

\doc\96\06\satstate.txt 1996 test scores by state
\doc\96\03\satstate.wk1
Comparing of States based on SAT scores alone is discouraged
the college board (but if you insist, these are the numbers)
States with highest averages have very low % of test takers
indicating only the best students are taking test, not that
the average student is better.

Nationally, only 41% of graduates take the SAT.
Source: College Board 1995

Top 10 states with low rates of testing
                1995         Grads
State           VerbMathComb %TakeRank
North Dakota    515 592 1107    5%       1
lowa            516 583 1099    5%       2
Minnesota       506 579 1085    9%       3
Utah            513 563 1076    4%       4
Wisconsin       501 572 1073    9%       5
South Dakota    505 563 1068    5%       6
Kansas          503 557 1060    9%       7
Nebraska        494 556 1050    9%       8
Illinois        488 560 1048   13%       9
Missouri        495 550 1045    9%      10

Ranking of states over 40% taking SAT
(This actually isn't a bad comparison)
                1995         Grads
State           VerbMathComb %TakeRank
Oregon          448 499  947   51%       1
Washington      443 494  937   48%       2
New Hampshire   444 491  935   70%       3
Alaska          445 489  934   47%       4
Delaware        429 488  917   88%       5
National        428 482  910   41%       6
Maryland        430 479  909   64%       7
Connecticut     431 477  908   81%       8
Massachusetts   430 477  907   80%       9
California      417 485  902   45%      10
Vermont         429 472  901   68%      11
=============== median ========================
New Jersey      420 478  898   70%      12
Maine           427 469  896   68%      13
Virginia        428 468  896   65%      14
Texas           419 474  893   47%      15
New York        419 473  892   74%      16
Hawaii          407 482  889   57%      17
Florida         420 469  889   48%      18
Rhode Island    425 463  888   70%      19
Indiana         415 467  882   58%      20
Pennsylvania    419 461  880   70%      21
North Carolina  411 454  865   60%      22
D.C.            412 445  857   53%      23
Georgia         406 448  854   65%      24
South Carolina  401 443  844   58%      25

\clip\97\19\satstate.txt CNN 08/27/97- Updated 03:46 PM ET 1997 SAT
scores by state

%%WASHINGTON STATE

2010: WASHINGTON STATE LEADS SAT FOR PAST 8 YEARS 
http://seattletimes.nwsource.com/html/editorials/2012888651_edit14sat.html
Monday, September 13, 2010 - Page updated at 06:16 PM Kudos to
Washington students who put the state No. 1 for SAT results SAT scores
of college-bound seniors nationwide were flat this year, but credit
Washington state for bucking the trend, registering average scores
that are the highest in the nation.  ... our students have led the
country in SAT scores for eight years running.

Another high mark for the growing number of students taking the
voluntary test. About 54 percent of Washington's high-school students
took the test, making it the highest among Western states, which were
at 43 percent, and the nation, at 47 percent.



"SAT math scores climb, verbals level off" Seattle Times Aug 26, 1997
Ranked by math score
------------------------
Asian 496V 560M  1056C
All   523V 523M  1046C
Men   507V 530M  1037C
White 526V 526M  1052C
Women 503V 494M   997C
AfrAm 434V 423M   857C

Asians best in math, combined, 2nd best in verbals. Men lag women in
verbals by 4 points.



@@Striver

zip37\clip\99\17\strive2.txt 9 Sep 1999 New Republic THE END OF
MERITOCRACY Should the SAT Account for Race? [Strivers] Opposing
opinions by Nathan Glazer and Abigail Thernstrom 

WEIGHTING SAT SCORE BY SOCIAL DISADVANTAGE
http://interactive.wsj.com/archive/retrieve.cgi?id=SB936061265207782969.djm
zip36\clip\99\17\striver.txt New Weights Can Alter SAT Scores As
Family Factors Determine 'Strivers' By AMY DOCKSER MARCUS Staff
Reporter of THE WALL STREET JOURNAL ""A combined score of 1000 on the
SATs is not always a 1000," Mr. Carnevale says. "When you look at a
Striver who gets a 1000, you're looking at someone who really
performs at a 1200. This is a way of measuring not just where
students are, but how far they've come."  "

CHAVEZ KNOCKS "STRIVERS" SAT HANDICAP POINTS
http://jewishworldreview.com/cols/chavez1.asp
zip37\clip\99\17\chstrive.txt Jewish World Review Sept. 9, 1999 /28
Elul, 5759 Linda Chavez Scoring the SATs --- the givers, not takers 


@@Time

Some students get extended time

z41\clip\2000\05\sattimet.txt From the Washinton Post Metro Section,
B1, Monday , May 8, 2000 Parents’ Push Gives Students More Sat Time
By Jay Mathews, Washington Post Staff Writer URL:
http://washingtonpost.com/cgi-bin/gx.cgi/AppLogic+FTContentServer?pagename
=w pni/print&articleid=A23620-2000May7

@@Writing

EVEN THE SAT HAS SWITCHED TO HOLISTIC
ESSAY QUESTIONS My understanding of the TSWE (Test of Standard
Written English) is that it has been replaced by the SAT II Writing
Test This SAT II subject test (formerly called Achievement Test) is a
one hour test that requires students to write a simple essay in 20
minutes on a topic that is given, and then work for 40 min. on a
multiple choice grammar test: correct usage, errors, sentance
structure, etc.

ASIANS ABOUT EQUAL IN MEAN, BUT LARGER SD MEANS MORE ASIANS AT TOP SCORES
http://www.collegeboard.com/prod_downloads/highered/ra/sat/2006-total-group-writing-subscore-report.pdf
2006 Total Group Writing Subscore Report
\doc\web\2009\11\satwriting.xls
SAT Writing Subgroups 2006

Performance by subgroups
Performance across the essay, multiple-choice questions, and the
overall writing section demonstrated similar trends to those seen on
the other sections of the SAT Reasoning Test. Table 1 summarizes the
performance on the writing section by gender and ethnicity.

# of students Essay Multiple-choice Total Writing
Section               Essay          Multiple Choice  Total
              N       Mean    SD      Mean    SD      Mean    SD
1 White          778,885     7.5     1.6      52      10     519     100
2 Asian          127,826     7.5     1.8      51      12     512     119

3 AmericanIndian   8,754     7.0     1.6      47      10     474     101
4 Hispanic       145,724     6.8     1.6      45      10     451      98
5 Black          143,305     6.5     1.6      43       9     428      93

Overall       1,376,74     7.2     1.7      50      11     497     109
Female         740,103     7.4     1.6      50      11     502     108
Male           636,642     7.1     1.7      49      11     491     110
Other           49,898     7.3     1.7      49      11     493     110
NoResponse     122,348     7.0     1.9      48      12     482     127


Whites have slightly higher mean 519 vs 512, but larger SD 119 vs 100, so
there will be more Asians at the top of the curve
Blacks lag Hispanics and Native Americans


	Pct N
Overall		1,376,745	100.0%
Female		740,103		53.8%
Male		636,642		46.2%
AmericanIndian	8,754		0.6%
Asian		127,826		9.3%
Black		143,305		10.4%
Hispanic	145,724		10.6%
White		778,885		56.6%
Other		49,898		3.6%
NoResponse	122,348		8.9%