\doc\web\99\17\cochiq.txt
Date sent:      	Wed, 15 Dec 1999 13:27:18 -0500
From:           	"Gregory M. Cochran" <74771.3230#compuserve.com>


- legacy may may 2x admit rate but only 30 point sat gap
At the upper reaches of the LSAT distribution, more than 1,100 white
students had scores of 170 or higher in 1992 (on a scale of 120 to
180) and three blacks. (0.3%)
On the MCAT, in three subtests whose scores go from 1 to 15), 1146,
1469, and 853 whites scored in thr 12 to 15 range; a maximum of 19
blacks scored that high. (2.2%)
On the GRE, blacks score at the 10 or 12th percentile of the white
distribution.  
Not imcome - little farm towns with a population of a few thousand
that can produce about as many students at the high end of the
distribution as blacks in Chicago. 


   Statistics is a subject worth understanding - take my word for it.   
The question is whether offspring of alums are preferentially admitted over
other students with similar qualifications, and the extent to which this
happens.  The answer, is yes, but not to a great degree;  on average,
childs of alums admitted  have only slighty lower test scores than others
admitted. - 30 SAT pts or so.    Contrast this with the differences
involved in 'affirmative' action.   They vary from school to schoo, but are
always much larger, and involve a much larger fraction of the students
admitted. At Harvard, AA admittees average only 100 pts  lower or so; this
smallish gap is made possible by funneling a high fraction of _all_ the

high-scoring blacks in the U.S. into Harvard.    But the aren't enough
high-scoring blacks for this to happens to the same extent in other
schools.  At Berkeley, for example, the gap was 288 ( in combined SAT
score)   Hardly any of the  blacks admitted at Berkeley under these
policies  would have been admitted in a race-blind process.   


      These gaps get bigger  as we look at more more exclusive processes,
ones that emphasize  the far edge of the  population distribution of
intelligence, 
They're bigger in grad school, for example - in general, except for
education majors, people going into grad school are smarter than the
average college grad.
At the upper reaches of the LSAT distribution, more than 1,100 white
students had scores of 170 or higher in 1992 (on a scale of 120 to 180) 
and three blacks.
On the MCAT,  in three subtests whose scores go from 1 to 15), 1146, 1469,
and 853 whites scored in thr 12 to 15 range; a maximum of 19 blacks scored
that high.
On the GRE, blacks score at the 10 or 12th percentile of the white
distribution.   This is not, for the most part, a consequence of  white
wealth and privilege.  I can find you
you  little farm towns with a population of a few thousand that can produce
about as many students at the high end of the distribution  as blacks in
Chicago. There are  single white families whose production of kids with
LSAT scores over 160 compares with the entire black population of the U.S. 
Just so, there are single black families that produce sprinters that
outclass all the white people on Earth put together.  Racial differences
are far clearer, far stronger at the far edge of distributions. 


      There are undoubtedly other small pieces to this puzzle -  the kids
of peope with vast clout  or bags of money probably manage to get admiited
perferentially also, but they are few in number. 


     One would expect the children of Yalies who graduated in the past two
or  three decades to have a higher-than-average chance of admission today
because their IQ and test scores will tend to be higher than average. It is
called heredity;  at least one parent was smart enough to get into an elite
school,  Because of assortative mating, these kids are fairly likely to be
smart.  Not as likely as their parent(s) who are alums ( regession towards
the mean)  -  but higher than average.  So, I'd expect that kids whose
father and/or mother went to Stanford have a better-than-average chance of
getting into Harvard, almost as high as that of the children of Harvard
alums.


     It is possible that one might maximize the income of _grandchildren_
by sending your kids to Harvard, since I'd guess that the average IQ of you
kid's mate would then be higher and lead to higher IQs in the next
generation.  It also would lead to fewer grandchildren, on average, since
high-IQ types, particular women, have realized that the next generation is
not worth the trouble.  


    Of course, there is almost no reason why anyone should care about
affirmative action in admissions, because what evidence we possess
indicates that long-term outcomes are essentially independent of how
exclusive the college is.  The market sees through the hype.    As far as
we can see,  kids do almost exactly as well (economically, anyhow) by going
to Penn State as to Yale, and it's cheaper.  In other words, the Ivy League
is a con.    The only legitimate beef I can see is the allocation of
financial aid.  If there are kids who would like to go to college, and who
would, on average, do significantly better than  AA admittees, but don't
because of financial reasons, we're  investing poorly. 


       Of course, I don't even know if college itself makes any economic
sense.   A quick look at the current distribution of undergraduate majors
suggests to me that at least 75% of undergraduates acquire only chugging
and rent-seeking skills in college - this pretty much agrees with my
college memories.   




       Gregory Cochran