Nature and Nurture 275Precautions required when interpreting behavioural
genetic studies
A number of problems can arise when behavioural genetic studies are
interpreted without recognising the limitations of the methods used:
1 For many years, behavioural genetics studies were carried out by re-
searchers who were more interested in genes than environments. Using
twin and adoption studies, they were happy to estimate environmental
effects indirectly without actually measuring the environment. In effect,
the environment was assumed to be responsible for any variance that
could not be explained by genetic effects. This assumption is a simplifica-
tion since there are other potentially important sources of unexplained
variance including measurement error, which is often greatly under-
estimated, and the role of chance in brain development. Fortunately,
the best behavioural genetics studies nowadays include sophisticated
environmental measures in genetically sensitive designs (for example,
twin studies), generating direct measures of relevant shared and non-
shared environments.
2 The average variation in environment across a given population may
not be great, leading to an underestimate of its potential effects. For ex-
ample, height in Western Europe is estimated to have had a heritability
of over 90% in the seventeenth century, and has a similar heritability
today. It would, however, be wrong to conclude that environmental fac-
tors such as nutrition have only a slight role to play in the determination
of height – on the contrary, mean adult height has increased by about
15 cm over that time, and this cannot plausibly be attributed to a change
in the gene pool. At first sight, it is puzzling to find that height can be so
sensitive to nutrition despite a heritability of over 90%. The explanation
lies in the restricted environmental variation at each time point. In the
past, most Western Europeans resembled one another in being poorly
nourished; nowadays they resemble one another in being well (or
excessively) nourished. With relatively little variation in nutrition, most
of the variance in height is genetic at any one moment. Nevertheless,
the major change in average nutrition over three centuries had a major
effect. What applies to nutrition applies to parenting too – for most of
the population, there are not great differences in child-rearing style at
any one time. This will make the effect of ‘shared environment’ seem
small, but it does not mean that parenting cannot make a big difference,
especially when it is outside the normal range. Sometimes the effects
of extreme environments are evident from ‘experiments of nature’,
where severe man-made or natural disasters affect lots of children or
adolescents, almost at random. For example, severely deprived orphans,
such as those studied from Romania in the late 1980s, all had very
delayed development. For children in these circumstances, the shared
environment does indeed matter. Randomised controlled trials that alter
the environment can also be very informative since there should not be