Blame it on their peers?

Like father like son is not just a catch-phrase, says Michael Rutter.

The notes on the jacket of Judith Rich Harris's The Nature Assumption claim that "a pathbreaking psychologist explodes the myth of parental influence by revealing the power of peers who play a far more critical role than parents do in children's development". Psycho-linguist Steven Pinker, in his foreword to the volume, supports this notion, heralding the book as a "turning point in the history of psychology". High claims, indeed.

The book is ill served by such hyperbole because it creates unrealistic expectations and the reader is likely to be disappointed by the lack of critical discussion of either the weaknesses in the case being put forward or the strengths of alternative views. That is a pity because Harris sets some important challenges that need to be met, she assembles an impressive display of relevant research findings, and she puts forward some salient criticisms of widely held assumptions.

The Nature Assumption is an unapologetically evangelical tract, albeit a well-written, well-argued one, and Harris is quite forthright about her objective of bringing about conversion. From the outset she puts forward her twin purposes of dissuading readers that a child's personality is shaped or modified by parents and of convincing them that it is strongly influenced by experiences in the peer group. In other Words, Harris wishes to replace the "religion" of parental nurture with that of group socialisation on the grounds that scientific evidence supports the latter and disproves the former. But does it?

Harris is scathing about most studies of the effects of parenting. She points out that few consider possible genetic influences; that the statistical associations are as likely to reflect children's effects on parents as vice versa; that most confine attention to children's behaviour at home (ignoring the possibility that it may not be the same in other contexts); and that possible influences outside the family (such as school or the neighbourhood) are not considered.

These are fair criticisms and Harris is clearly right in claiming that the evidence in support of strong effects of parenting is largely lacking. But she goes on to argue that the associations found are so weak that, even if valid, they would not matter. Thus, she dismisses a correlation of .19 found in one study as "all but useless". That observation is flawed for two reasons. A correlation does not provide a good measure of the strength of an effect because it is so dependent on the number of individuals in the study to whom the risk factor applies. Thus, in the general population, Down's syndrome correlates only .036 with IQ (a far smaller correlation than .19) despite the fact that individuals with Down's syndrome have, on average, a 60-point IQ deficit. That is because not many people have Down's syndrome, so it plays a trivial role in accounting for individual differences in IQ despite its huge effect at an individual level.

Research has also shown that few risk factors have more than a slight effect. This is the case with respect to individual genes as much as with individual psycho-social risk factors. The main influence comes from multiple factors operating in combination and not from any one on its own.

The next plank in Harris's argument to dismiss the importance of parental influences notes that not only is there a lack of evidence to suggest they do matter, but that genetic findings suggest that it is unlikely that they could make much difference. Harris presents three main pieces of evidence: identical twins separated when young are as similar to one another as those brought up together. This sounds impressive, but is based on small, highly selective samples and covers a limited range of variables. Second, Harris argues that studies of twins show that genetic influences account for about 50 per cent of variation among individuals, while environmental effects tend to make children in the same family different rather than similar. But while the importance of genetic effects is clear and it is true that differences among children in the same family are striking, there are important exceptions to this tendency, such as delinquency, which shows a tendency to run in families.

Harris's third observation points to studies of the effects of parents' differential treatment of children, suggesting that it makes little difference to how children develop. In truth, there is little evidence one way or the other.

The book's weakness is in its approach to evidence from studies seeking to test hypotheses on specific environmental effects. It is all very well to lay into nurture researchers with an axe, hammer and sword - much of the criticism is justified - but all too often Harris leaps to a premature dismissal of awkward evidence. Thus, it is accepted that adoption of children from high-risk backgrounds is associated with a rise in IQ, but this she attributes solely to peer-group effects. Well, maybe. Most previous rejections of the effects of nurture (such as in Sandra Scarr's work) accept that extreme environments are likely to be influential even if environments within the normal range are not. Harris will have none of this. The possibility is cast aside as the "last slim hope". The evidence on extremes is stronger than she admits. Moreover, seriously adverse environments are more common than she would have us believe. Epidemiological studies have shown that the experiences of abuse, neglect, scapegoating and breakdowns of parenting are distressingly frequent. Because Harris is so partisan in her rejection of family influences, there is a danger that family researchers will fail to take on board that, if their findings are to be taken seriously, they will have to do a better job in testing for causal effects.

Equally, there is a danger that the general public will believe Harris's claim that she has shown that parents do not matter. She has not. What she has done is castigate family researchers for going beyond their evidence. Fair enough, but it is much too soon to dismiss family influences out of hand.

What about Harris's counter-proposal that the peer group constitutes a powerful environmental influence? Appropriately, she draws attention to the fact that humans are social beings and to the range of findings showing both the strong tendency for people to form social groups and the impact of group values on the behaviour of individuals. She notes that unimpaired children reared by mute-deaf parents gain spoken language and that children reared by foreign parents share the accent of their peer group. Both findings point to the importance of outside influences. Harris emphasises the importance of genetically influenced individual differences in behaviour, but she argues that the social groups are crucial in enhancing those differences.

So far so good: most researchers have accepted the need to consider environmental influences outside the home. Harris's book is useful in providing a persuasive set of arguments that all of us need to take this view on board in our research and in our policy recommendations. But there are three major concerns about her treatise. First, she claims too much. She asserts that if the peer group's culture differs from that of the parents, the peer group always wins. That is too strong. There have been few direct tests, but they have shown a far more complex picture.

Second, family researchers are taken to task for failing to consider children's effects on their parents, but Harris neglects the equally important need to assess selection factors when trying to evaluate the influence of peer groups, schools and neighbourhoods. Children choose which social groups to join and families select the schools their children attend and decide where they will live. There is a substantial research literature (not considered by Harris) dealing with the differentiation of selection effects and social influences. The evidence shows that both are operative, but does not demonstrate the power of group socialisation claimed by Harris.

The third concern is of a rather different nature - namely that Harris's focus on individual differences leaves a huge amount unexplained. Why, for example, have crime rates soared in the past 50 years in most industrialised countries but not in Japan? Why has the overall level of IQ risen? Why has the rate of marriage breakdown gone up? Why is the homicide rate in the United States many times that in Europe? The answers are not likely to be in the genes but it is not obvious that group socialisation provides any better explanation than family influences. The fact is that we really do not know. But, in any case, why do these social groups have the characteristics they have? Harris comments that cultures can be changed or formed within a generation, but how does that come about?

Most of us would be pleased to have Harris as the barrister pleading our case in court. She has mastered her brief, she is eloquent and entertaining, she makes people sit up and pay attention, and she opens our eyes to important considerations. What more could one want? But would you want Harris as the judge summarising the facts for the jury and seeking to provide an even-handed, balanced overview of the evidence? I think not.

Of course, as a researcher who has investigated school influences and the effects of living in an inner city area, and who has results indicating the importance of both, it might be expected that I would welcome Harris's powerful advocacy. To a degree, I do, because her book helpfully serves to put the issues on the policy agenda. But as a researcher who also studies genetic and family environmental effects I worry over her one-sided portrayal of the evidence.

There is still a great deal that we do not understand about influences on psychological development and it is much too soon to claim that all our eggs should be taken out of the "family" basket and put in the "peer group" one. As the old saying goes, "it ain't ignorance that does the harm; it's knowing so many things that ain't so".

Sir Michael Rutter is professor of developmental psychopathology, Institute of Psychiatry, King's College, London.