Genes, Brains, and Human Potential The Science and Ideology of Intelligence

PRETEND GENES 43

to be fully expressed. Th e average correlation between such pairs of twins
could be a direct estimate of heritability.
In an infl uential set of papers, Burt claimed to have done just that and
to have mea sured the IQs of twins reared apart. He arrived at an estimate
of the heritability of IQ of 0.83 (83  percent). Th is means that 83  percent of
the variation in IQ is associated with diff erences in genes; only 17  percent
results from diff erences in experience.
However, separated identical twins are relatively rare, and Burt seems
to have been suspiciously lucky in fi nding so many. An alternative ap-
proach has been to compare average resemblances of pairs of MZ twins
with those of DZ twins. Th e degree to which the resemblances correspond
with ge ne tic resemblances is also an index of heritability (or so it has
been claimed). Th is is the “classical twin method.”
Both approaches have been enthusiastically pursued since the 1950s,
with some refi nements, but using essentially the same basic logic and
procedures. For many reasons, though, they have been shrouded in doubts
and controversies. Many psychologists and other scientists now look
askance at Burt’s results, because suspicions were raised, aft er his death,
of fabrication of data. Th is came to light aft er Leon Kamin’s forensic scru-
tiny of Burt’s data.^4 And the classical method comparing MZ and DZ
twins is open to environmental confounds, as we shall see.
Before looking at methodological prob lems, it is impor tant to consider
the theoretical fl aws of the methods. Th ese mainly revolve around the
large number of simplifying assumptions that have to be adopted to pro-
duce the expectations with which the twin correlations can be compared.
In his 1956 paper, Burt used the words “assume” or “assumption” more t ha n
sixty times. Th e paper ends with the remarkable understatement, “Th e
assumptions adopted are perhaps not entirely free from criticism.”^5

ASSUMPTIONS

Th e formulas that behavioral ge ne ticists use to calculate ge ne tic eff ects on

variation are not straightforward. Because the mea sures are aspects of

populations, they are statistical, such as averages and variations, rather

than numbers of individuals in established categories. Because the

This content downloaded from 139.184.14.159 on Tue, 17 Oct 2017 13:51:26 UTC