Educability-and-Group-Differences-1973-by-Arthur-Robert-Jensen

46 Educability and Group Differences
when a2. is zero or close to zero that the trait itself is not genetic,
when what it really means is that all individuals in the population
are genetically the same or very nearly the same on the trait and
that the phenotypic differences among individuals are mostly
attributable to non-genetic or environmental factors. Similarly,
when cr| is zero it does not mean that the environment is unimportant
in the development of the trait; it only means that environmental
factors do not contribute to individual differences in
phenotypes.
We can sum up by emphasizing that h2 is not to be thought of as
a constant, like n or the speed of light, but as a population statistic
like the birth rate or the mortality rate, which can differ from one
time to another depending on a multitude of conditions.
But does this mean that h2 is irrelevant to the individual member
of a population in which h2 has been estimated Not at all, if we
understand our model correctly. Formulas 1 and 3, above, tell us
that individual differences in P are composed of individual differences
in G plus individual differences in E. Since h2 tells us the
proportion of variance in P accounted for by variance in G, the
correlation rGp, between genetic value2 and phenotype, will be the
square root of h2, i.e., h. This means that by knowing h2 we can
estimate an individual’s genetic value, given his phenotype, and
the standard error of the genetic value, SEG, is
SEC = (2.5)
where ot — SD of obtained scores.
This means that approximately 68 percent of individuals will
have true genotypic values (G) that lie within one SEG of their
estimated genotype; approximately 95 percent will have G within
2 SE qs of their estimated value, and approximately 99-7 percent
estimated Gs will lie within 3 SEQs of their true G. In short, we
can estimate an individual’s genetic value from a knowledge of his
phenotype and of h2 of the trait in question in the population of
which the individual is a member. If we take the estimate of h2
for IQ of 0-80, and if IQ has a standard deviation of 15 points in
the population, then the total variance of P, or oj,, is 152 = 225.
The standard error of G for an individual, therefore, will be
■J22S x yj 1 —0-80 = 6-75 IQ points. This means that 68 percent
of our estimates of individuals’ genetic values will be less than
about 7 points off, 95 percent will be less than 14 points (i.e.,