SAT and Affirmative Action


U.S. News & World Report November 11, 2002  has a very interesting 
article on the history  of college entrance/student selection exams 
in the USA.  The exams that the elite colleges use to help them 
decide which of the many applicants they should accept.   

In the late 19th and early 20th centuries it was the "College Boards". 
Exams given to high school seniors to test them on the information 
and knowledge that they learned in high school.  Those who had 
learned enough of the correct material were selected for the college 
because they were "qualified".   Anyone could study for the College
Boards by looking at the past years exams to see what they were 
expected to know, and high schools could know what to teach.

But the problem was that expensive elite "prep schools" better 
prepared their students for the elite colleges,  and the system was 
criticized for fostering an elite. The kids of the rich had a big 
edge over kids from poor families who might have as much or more
potential but who didn’t have as good a school system.

So the SAT was developed to measure not what a student had learned, 
but their aptitude for learning.  Hence Scholastic Aptitude Test.

This exam proved to correlate well with how a student did in college, 
and using it in the selection process increased the number of able 
kids from lesser quality high schools who were admitted into elite 
colleges (where they did well).

Sounds good?  Well as you may have heard, the SAT has fallen out 
of favor because it is said that "there is no way to study for it".  
It measures more "how you think" than "what you know".   Of course 
that was the idea, but each exam is seen by the SAT critics as just 
a "bag of tricks".  And besides the exam discriminates
against several "targeted minorities"  because they tend to score 
low on it. And (even worse) when members of those targeted minorities
are admitted even with lower SAT scores than other students, they 
drop out of school at a higher rate than the other students.  At 
the University of Wisconsin there is much concern over the higher 
drop out rate of students that were admitted under Affirmative
Action .

The story in the Wisconsin State Journal on Thursday February 6 is

"MINORITY GRADUATION RATES LOW".....About 42% of minority freshmen
(in 1996) graduated within 6 years compared to 63.7% of white
freshmen who enrolled then.  

(I think graduation rates were even higher for Asian freshmen).

Why is this no mystery to me? Because SAT was designed to predict
success in college.  And (statistically speaking) it does.

So there is a movement to abolish the test. And some elite 
schools don’t want any sort of testing used to select their 
students.


Of course no one dare suggest that the SAT is doing exactly what 
it was designed to do.   So now the SAT is being redesigned to 
measure what the student learned in high school, and is being 
called the Scholastic Achievement Test.   Sounds like the College 
Boards have returned.

We don’t want poor kids with the potential for college to be left 
out for lack of money.  And we should not sucker  targeted minority 
students into attending an elite college were they won’t do well, 
when they might have done just fine at another college.

So what to do?

Replies: (from the sci.econ post)

>>David Lloyd-Jones <dlj@rogers.com> wrote:
(of race based affirmative action)

>>>
>>>I think that's a perfectly sound way of looking at the proposition "Who 
>>>gets into law school?"
>>>

>Jim Blair wrote:
>>Hi,
>>
>>Law school? Or university?  
>>
>Jim,
> 
>I picked law school, rather than university, exactly because it's a club 
>leading to privilege, where the parallel with sports scores could be 
>expected to have both bearing and the right flavor.

Hi,

Is this a "bait and switch"?  The case before the courts is the University 
of Michigan and not just a law school. And are law (or medical) schools 
less competitive than "elite" universities?

Do "targeted minorities" that are admitted to law or medical schools with 
lower qualifications drop out at a higher rate?

> 
>On advanced universities I tend to agree with your point -- which I 
>think of more generally as Thomas Sowell's point -- that at least some 
>kids pushed beyond their SAT's may be sacrificial lambs. ....

>As a former Governor of Educational Testing Service (one meeting, 
>carrying my boss's proxy) I can rule ex cathedra on this one: ...
>what Jim wants has already been done.
 
>I.e. making the SAT a predictor of academic success has been its 
>intention from the start, back in the late 1940's; and every question 
>and inflection in the test is calibrated and recalibrated with just that 
>aim in mind. This is done against huge databases of actual subsequent 
>results, so that by now the thing is in fact pretty good at what it aims 
>for. The original social aim was quite explicitly elitist noblesse 
>oblige: Conant, at the time President of Harvard, and his sidekick Henry 
>Chauncey, were fed up with the number of dolts getting in, and wanted a 
>test that would qualify more working class kids for entry on "objective" 
>grounds, likelihood of success. They got together with the brilliant 
>educational psychologist Ben Wood and a few friends who could cough up 
>the necessary spondulics, and ETS was born.
>...
> 
>My general view of un-merited advanced placements is that they are a 
>necessary evil, and should be enforced rigorously, and very briefly. 
>Among other things we don't want to continue turning out white college 
>grads who have grown up in mono-racial environments. 


Since California passed Prop 209 to ban race based affirmative action
in 1996, minority student population has increased in the state university
system, as have their graduation rates.  This result was predicted by me 
at the time.  

These increases were the result of more targeted minority students 
going to various state universities where they were a better match
(based on SAT scores) to their peers, rather than attending Cal 
Berkley or UCLA where they were in classes with students with 
higher SAT scores.


I wish that other proponents of race based affirmative action were as 
honest about this. The debate would be different if we all recognized that 
AA in college admissions exists to benefit white college students at the 
expense of some black sacrificial lambs.

But two objection. 

First I doubt that US universities are as "mono" as you think. At the UW 
we have students from all over the world. Of course as a big world class 
university that is expected. But even at little Miltion College I was 
suprised when of the 8 students in my first physics class in 1965 I had 
students from Kenya, Nigeria, Hong Kong and Saudi Arabia.  The other 4 
were the expected local WASPS.  I even had some black class-mates at 
private "semi-elite" Rockhurst College in the late 1950's.

Second if "diversity" is GOOD (and maybe it is) why is it applied only to 
race?  Why not to religion as well?  Both the UW Madison and Miltion 
College are/were "over-represented" in WASPS, Roman Catholics, athiests, 
and Jews, and "under-represented" in Mormons, Fundamentalist Christians,
Moslems, Hindus  and some others. Should applicants get bonus points for 
being members of an "under-represented" religion?

Jews are clearly "over-represented" in US higher education. Should they 
have a higher standard for admission than even Asians and WASPS?

>....Quick stern 
>temporary cures, however, are dreams. Policy by its nature does not work 
>that way.

Exactly. Affirmative Action is now a Big Business. Hey it provides my 
daugher in Boston with a job.  (She is a bureaucrat and my son-in law is a 
lawyer.  Where did I go wrong? :-(

> 
>In the absence of such possible dreams, I think that muddle is probably 
>the best that can be hoped for, and clarity very often a good thing to 
>avoid.
> 
>                                                 Best,
> 
>                                                     -dlj.

As long as those lambs don't realize why they are being sacrificed ;-)


                     ,,,,,,,
_______________ooo___(_O O_)___ooo_______________
                       (_)
jim blair (jeblair@facstaff.wisc.edu) Madison Wisconsin
USA. This message was brought to you using biodegradable
binary bits, and 100% recycled bandwidth. For a good time 
call:   http://www.geocities.com/capitolhill/4834


AND:


>Jim Blair <jeblair@facstaff.wisc.edu> wrote:
>[snip]
>> The point remains that the SAT is what it was designed to be: a predictor 
>> of success in college. Not a "perfect" predictor, but about as good as can 
>> be expected.

stephen.fromm@verizon.net (Stephen J. Fromm) wrote:
>
>No.  High school grades alone are a better predictor; the SAT plus
>high school grades is better yet.
>
>This is what I recall; go ahead and do your own google search.  I did
>find the following from http://www.fairtest.org/facts/satvalidity.html
> :
>
>"The SAT I is designed to predict first-year college grades - it is
>not validated to predict grades beyond the freshman year, graduation
>rates, pursuit of a graduate degree, or for placement or advising
>purposes. However, according to research done by the tests'
>manufacturers, class rank and/or high school grades are still both
>better predictors of college performance than the SAT I.
>"How well does the SAT I predict first-year college grades? The
>College Board and ETS conduct periodic studies of the SAT I. This
>usually involves examining the relationship between test scores and
>first-year college grades, generally expressed as the correlation
>coefficient (or r value). The College Board's Handbook for the SAT
>Program 2000-2001 claims the SAT-V and SAT-M have a correlation of .47
>and .48, respectively, with freshman GPA (FGPA). This number is
>deceptive, however. To determine how much of the difference in
>first-year grades between students the SAT I really predicts, the
>correlation coefficient must be multiplied by itself. The result,
>called r squared, describes the difference (or variation) among
>college freshman grades. Thus, the predictive ability (or r squared)
>of the SAT I is just .22, meaning the test explains only 22% of the
>variation in freshman grades. With a correlation of .54, high school
>grades alone do a better job, explaining almost 30% of the variance in
>first-year college performance."

Hi,

I think there is some mis-use of statistics here. The real issue is not 
the overall correlation between SAT and grades or graduation rate. That 
correlation is blurred by the fact that students tend to go to colleges 
with others of similar preparation/SAT. I bet the average Harvard freshman 
has a lot higher SAT than the average Milton College (or UW Whitewater) 
freshman. 

But are the graduation rates that different?

Rather it is how well a lower SAT student does when in a collge (and in 
classes) with high SAT students.  

And the women vs men thing should consider that women tend to score better 
on verbal but lower on math. But then they tend to take courses where 
verbal skills count for more than math. 

There is even a suggestion that the gender gap should be reduced by giving 
the verbal score more weight relative to the math part.

The thing here is how well students match their classmates.

Me from amother thread:


..... But the
whole idea of "elite" colleges is that they deal with bright and well
prepared students.

Then there are a range of other (non-elite) colleges that accept students
who are less well prepared.  One key to increasing the success rate of
students in college is to better match each student to the college.  As
the California system is now doing since they match by preparation and SAT
rather than to achieve racial balance. (called "diversity")

So (by neglecting the fact that "targeted minority students" are admitted 
to elite colleges when they have SAT scores of 200 points lower), how do 
you explain the fact that they graduate at about a 20% lower rate from 
those elite schools?


AND:

>Jim Blair <jeblair@facstaff.wisc.edu> wrote :

>> the women vs men thing should consider that women tend to
>> score better on verbal but lower on math.

callen@efn.org (Chris Allen) wrote:
>
>Fairtest.org takes the position that women and men have equivalent
>math ability:

Hi,

Yes, Fairtest seems to be committed to the proposition that every 
identifiable group is equal in every respect, no matter what any test or 
evaluation indicates.

>
><beginquote>
>71
>
>SAT-Math Scores and Course Selection in High School and College
>
>Looking only at the one-quarter of SAT test takers who took calculus
>in high school, men still scored 36 points higher on the math portion
>of the SAT (which tests math knowledge only through algebra and
>trigonometry, not calculus). For the 29% of students who took honors
>science courses in high school, the gender gap on SAT-Math is still 26
>points. Overall, 68% of women SAT test-takers complete four years
>of math in high school; for males the figure is barely higher at 70%.
>Similarly, there is little difference between the average math grades
>men receive in high school (3.02) and those received by women (3.01).
>Yet, the gender gap for these students was 37 points.
>
>In a 1992 study, ETS researchers Howard Weiner and Linda Steinberg
>analyzed the first-year college math grades and previous SAT scores
>of nearly 47,000 college students and found that women who earned
>the same grades in the same courses as men had averaged 33 points
>lower on the math section of the SAT.6 ETS research shows that
>college course selection explains some of the gender gap but never
>more than half.

Sounds like the gender gaps being discussed here are much smaller than the 
several hundred point gap between the "targeted minorities" admitted to 
elite colleges have as compared to the other students at the same school.

Since grades are "rounded off" to the nearest letter (quintile?) in most 
schools, how many SAT points difference would be expected to correspond to 
a letter grade difference?

>
>Causes of the Gap
>
>Research has suggested several possible explanations for the SAT
>gender gap: 1) the multiple-choice format (women do better on essay
>and other types of assessments); 2) the SAT rewards males' greater
>willingness to take educated guesses on the test; 3) males do better
>on
>"speeded" tests where most students cannot finish the exam;7

For each of these, the next question is WHY? Like why are men better at 
guessing?  Or more willing to take a chance on a good guess?

They can't consider the possibility that the SAT score actually reflects 
the ability to solve problems better than grades do. And that grades might 
reflect a woman's ability to get guys to help her study?  Or that (mostly 
male?) teachers might give a (pretty?) co-ed a break on grades ;-)

> 4)
>students who belong to groups expected to perform better on highstakes
>tests score higher than students belonging to groups expected
>to achieve lower scores.8
><endquote>

This sounds like a reversal of cause and effect. Who "expects" certain 
"groups" to do better?  And is that "expectation" based on past 
performance on similar exams?

But it was a good reply Chris.
>
>
>From the section "Appendix C: Gender Bias on the SAT" of
>fairtest.org's
>special report _Test Scores Do Not Equal Merit_.
>ftp://ftp.fairtest.org/optrept.pdf
>
>
>BTW, the last "possible causes" proposition is known as
>"stereotype threat". (Jensen, 1980).
>http://www.amazon.com/exec/obidos/tg/detail/-/0029164303/qid=1047384562/sr=1-1/ref=sr_1_1/102-6772501-7240141?v=glance&s=books
>
>
>-Chris

                     ,,,,,,,
_______________ooo___(_O O_)___ooo_______________
                       (_)
jim blair (jeblair@facstaff.wisc.edu) Madison Wisconsin
USA. This message was brought to you using biodegradable
binary bits, and 100% recycled bandwidth. For a good time 
call:   http://www.geocities.com/capitolhill/4834