However, I have been arguing for this approach
for over 20 years, and as far as I can see, the test-
ing movement has been affected little by my
eloquence. Why? There are lots of reasons:
People keep insisting that the n Achievement score
is invalid because it will not predict grades in
school—which is ironic since it was designed pre-
cisely to predict life outcomes and not grades in
school. Or they argue it does not predict all types
of achievement (Klinger, 1966)—when, of course,
it is not supposed to, on theoretical grounds. But
the practical problem (outside the tedium of con-
tent coding) is the unreliability of operant thought
measures. Many of them are unreliable, though
not all. Costa's (1971b) ego development score
has a test-retest stability coefficient over a year of
.66, N = 223. Unreliability is a fatal defect if the
goal of testing is to select people, let us say, with
high n Achievement. For rejected applicants could
argue that they had been excluded improperly or
that they might have high scores the next time they
took the test, and the psychologist would have no
good defense. One could just imagine beleaguered
psychologists trying to defend themselves against
irate parents whose children had not gotten into a
preferred college because their n Achievement
scores were too low.