March 7, 2001
Published in: ''More Dropouts, Higher Standards,'' letter to the
editor, <cite> Academic Questions,  </cite>  12:5 ( Fall 1999)



June 3, 1999
Editor@aq.nas.org

Dear Academic Questions, 

  Lenore Ostrowsky's Spring 1999 article, "College Dropouts and
Standardized Tests,"   begins with a discussion of dropout  rates and
the strong implicit assumption that a high dropout rate is a bad
thing.

 Many people assume this, but why?  There are two reasons why high
dropout rates are  a sign of health, not mistakes.

First, high dropout rates show high standards. As  Ostrowsky points
out, Cal Tech has a high dropout rate. This is not special; when I
last looked, the University of Illinois had higher dropout rates than
lesser colleges in that state, and this is common for good colleges.
For that matter, Harvard University has high rates of faculty dropout,
if you consider denial of tenure to be  a form of dropout.  In each
case, the existence of dropouts is a sign of high standards.

  Second, retaining poor students has what we economists call an
"opportunity cost".  Suppose the University of X can admit 1000
students per class.  This year, the 1000th-highest test score is
Smith, and the 1001st is Jones, so we admit Smith. Smith, however, is
lazy, and his grades are marginal. One choice is to work hard with
tutors and counselling to retain Smith. That choice has an opportunity
cost, though, as well as a resource cost: Smith is filling up one of
our 1000 slots. A second choice is to flunk out Smith and admit Jones
as a transfer student for sophomore year.  Since Jones is almost as
good as Smith in his observable characteristics, and Smith is clearly
below average in what is hard to observe, how can one say that the
right thing is to keep Smith?

  I suspect  college administrators are usually  unaware of this
argument, and that college dropout rates are not too high, but  too
low.   That said, nothing in this reasoning says that test scores
should not be used.  Rather, all available indicators should be used
to decide who gets a degree from  the University of X, including test
score, but    freshman year performance is an  excellent  indicator
too.

 Eric Rasmusen  
 Indiana University
Erasmuse@indiana.edu