responsible positions (Yerkes, 1921). However, it
is not really clear whether the Army made much
use of the masses of data supplied by Yerkes and
his eager assistants. A careful reading of his
memoirs reveals that Yerkes did little more than
produce favorable testimonials from high-ranking
officers. In the main, his memoirs say that the
Army could have saved millions of dollars and increased
its efficiency, if the testing data had been
used.
To some extent, the mountains of test data had
little practical impact on the efficiency of the Army
because of the resistance of the military mind to
scientific innovation. However, it is also true that
the Army brass had good reason to doubt the validity
of the test results. For example, an internal
memorandum described the use of pantomime in
the instructions to the nonverbal Beta examination:
For the sake of making results from the various
camps comparable, the examiners were ordered to
follow a certain detailed and specific series of ballet
antics, which had not only the merit of being
perfectly incomprehensible and unrelated to mental
testing, but also lent a highly confusing and distracting
mystical atmosphere to the whole performance,
effectually preventing all approach to the
attitude in which a subject should be while having
his soul tested. (cited in Samelson, 1977)
In addition, the testing conditions left much to be
desired, with wave upon wave of recruits ushered
in one door, tested, and virtually shoved out the
other side. Tens of thousands of recruits received a
literal zero for many subtests, not because they
were retarded but because they couldn’t fathom the
instructions to these enigmatic new instruments.
Many recruits fell asleep while the testers gave esoteric
and mysterious pantomime instructions.
On the positive side, the Army testing provided
psychologists with a tremendous amount of experience
in the psychometrics of test construction.
Thousands of correlation coefficients were computed,
including the prominent use of multiple
correlations in the analysis of test data. Test construction
graduated from an art to a science in a few
short years.