This study reviews and extends the considerable literature demonstrating th
at the various College Board examinations (most importantly the Scholastic
Aptitude Tests) make a small underprediction of women's college grades rela
tive to those of men in all fields except engineering. This finding persist
s even when corrections are made for differences in the fields that women a
nd men study and for sample selection bias. Because of this underprediction
, women most probably are underrepresented relative to their merit in fresh
man classes and scholarship competitions at selective public universities.
The differences in predicted grades are small, but account for an underrepr
esentation of women by at least 5% of the freshman classes of the Universit
y of California at Berkeley (200 to 300 a year) in the late 1980s. Various
solutions to this underprediction by the SATs and the dilemmas they pose fo
r public universities Such as Berkeley are explored.