308
PARTISAN REVIEW
are auxiliaries he has put silver, and iron and bronze in those who are
farmers and other workers. You will for the most part produce
children like yourselves but, as you are all related, a silver child will
occasionally be born from a golden parent, and vice versa, and all the
others from each other.
Socrates presents this story as a useful myth, whose acceptance would
promote social harmony and stability. But how is a person 's innate
worth to be ascertained at an early age?
Efforts to measure innate mental ability, initiated by Francis
Galton, a cousin of Charles Darwin, all reached the same impasse: tests
of (ostensibly) innate traits did not measure intelligence; intelligence
tests did not measure innate traits . Then the English psychologist
Charles E. Spearman (1863-1945) proposed a solution to this dilemma.
He conjectured that every intelligence test, broadly defined, assesses
two distinct and independent "factors" of intelligence: an innate
general factor g common
to
all tests and a specific factor that varies
from test to test. Tests differ not only through their specific factors but
also through the relative weights of the general and specific factors. A
test of pitch discrimination, for example, is not heavily "g-loaded"
while a test of ability to discern structure in a complex geometric figure
(Raven's "Progressive Matrices ," for example) is. Spearman conjec–
tured that "higher" mental abilities-mathematical ability, for
example-depend chiefly on g. He also
~uggested
that blacks, as a
group, are less richly endowed with g than whites. This "hypothesis "
is the cen tral thesis of Jensen's book.
To make the hypothesis credible one would need to rule out
nongenetic factors that might systematiccdly affect test scores. Jensen
briefly discusses one such factor-the "expectancy effect," described a
dozen years ago by Robert Rosenthal and Lenore Jacobson. Rosenthal
and Jacobson found that they could sometimes bring about substantial
improvements in the IQ scores of randomly selected students by telling
their teachers that a special test had shown that these students were
about to experience a spurt in mental development. Rosenthal has
recently summarized the outcomes of 345 studies of this kind . In 36
percent of the studies the measured effect was statistically significant at
the 5 percent level. The size of the effect, averaged over the entire
sample, was the equivalent of 10 IQ points. Among 43 studies with
special controls for cheating and observer errors the proportion with
statistically significant positive outcomes was 56 percent. These find–
ings suggest that the expectations of teachers can indeed strongly
influence the scholastic performance of their pupils. Thus it seems
