Is it a reasonable point to make a distinction between the predictive
validity of the GRE's in the imagined world where everybody gets to go to
graduate school and then we see how they do and the present convention of
using the GRE's as we do and then seeing what they tell us about
performance. . If 1100 does the job (i.e., accounts for variability in the
former world above) why distinguish between an 1100 and a 1400 (which
accounts for very little in the latter world)? Put differently ... maybe
the cutoff does all the work.
david
