I’ve suggested Allen and Duncan Smith’s understanding of brain development and of data from large-scale studies isn’t as good as it might be. So what?
The role of evidence
It’s not clear why Allen and Duncan Smith feel compelled to use evidence from brain development to shore up their case for early intervention. We already know that maltreatment, neglect and poor attachment cause problems, often lasting ones; that’s why they’re called ‘maltreatment’, ‘neglect’ and ‘poor’ attachment. We’ve known for millennia that early intervention in children’s lives is more effective than late intervention. We can also demonstrate the effectiveness of an intervention without knowing how it works. The use of brain development as evidence is especially puzzling since Allen and Duncan Smith clearly don’t understand it well, and because people who understand it better suggest that the evidence linking maltreatment, neglect and attachment to abnormal brain development is still rather tenuous e.g. De Bellis (2005); Glaser (2000). Basing one’s case on unreliable evidence runs the risk of defeating one’s own aims.
Correlation and causality
Allen and Duncan Smith rely heavily on correlations to support their model. If I’ve understood it properly their model looks something like this:
But all the causal connections implied in the model rely on correlations. Assuming that correlated variables must be causally linked – ie that one causes another – is a basic error in data analysis. It’s not the first time the Centre for Social Justice has made this mistake. Their paper on marriage published in December 2009, assumes that correlations between marriage and length of relationship, mental and physical health, violence and abuse and outcomes for children, mean that marriage itself results in more stable relationships, improved mental and physical health and a reduction in violence and abuse. Not that non-violent couples in stable relationships, with adequate parenting skills, who enjoy good mental and physical health might be more likely to get married and to stay together. As with the brain development data it’s clear that the error has been made by the authors of the reports, rather than in the research on which the reports are based. In order to change behaviour it’s essential to identify its causes accurately.
The causes of social problems
A name that keeps cropping up in the child development literature is that of Urie Bronfenbrenner, a psychologist renowned for his Ecological Systems Theory of child development, often depicted as a set of concentric circles representing nested systems as shown below.
The point made by Bronfenbrenner’s ecological systems model is that many external factors at different levels of complexity interact to influence a child’s development. The same point is made by Mareschal et al’s neuroconstructivist model, except that they include factors internal to the child (genetic, epigenetic and behavioural). In the diagram below I’ve integrated the neuroconstructivist framework with Bronfenbrenner’s and Allen’s and Duncan Smith’s models, taken a cross-section, and highlighted Allen and Duncan Smith’s causal pathway in red. Even though the diagram is sketchy (I’ve omitted many factors and their possible links to outcomes), it’s clear that Allen and Duncan Smith’s model of the links between poor attachment and social problems is a narrow one. Although there is little doubt that low levels of maternal ‘attunement’ and ‘empathy’ could result in the social problems referred to, there are clearly many other factors that could also cause them, not taken into account by Allen and Duncan Smith.
Of course there’s no reason why government shouldn’t focus on one particular cause of social problems. After all, water purification, sewage treatment, education, free healthcare and democracy have each resulted in major improvements to quality of life. The difference is that the evidence demonstrating the adverse effects from drinking polluted water, lack of access to sanitation, education, healthcare and living under an authoritarian regime, is robust. The evidence demonstrating a causal pathway between poor attachment, brain abnormalities and an array of ‘social problems’, isn’t.
Blaming the parents
Decades of research have demonstrated that the causes of Allen and Duncan Smith’s ‘social problems’ are many, varied and often interact in complex ways. Identifying what interventions might be most successful in reducing social problems is a challenging task. And some potentially effective interventions, such as de-criminalising drug use or adopting alternatives to custodial sentences, are deeply unpopular politically. It’s much easier, and more intuitive, to allocate the blame for social problems to factors at Bronfenbrenner’s microsystems level than to tackle the complex, expensive and potentially embarrassing task of identifying possible causes at the economic, legislative or historical level. In locating the primary cause of social problems with parents – notably mothers – Allen and Duncan Smith are conveniently overlooking other possible causes at the level of the child (genetics, epigenetics, disease, diet, environmental toxins); the microsystems level (churches, schools, communities, quality of education, health and social care); the exosystems level (employment, economics, legislation, government policy); the macrosystems level (cultural assumptions, global trends) and the chronosystems level (constraints and affordances that have arisen historically). Again there is no doubt that bad mothering could cause abnormal brain development and could lead to social problems, but there is considerable doubt over whether it’s the main cause of either. In short, what Allen and Duncan Smith don’t say is as important as what they do say.
The real world is an uncompromising place
Concerned about the poor quality of evidence in the chapters on the brain in these two papers, I contacted Graham Allen. A member of his research team called me back. He assured me that the material on the brain in the Allen report had been approved prior to publication by a neuroscientist, Professor A. I contacted Professor A – who incidentally isn’t a neuroscientist. He said his role was to check the evidence on parenting programmes; the scientific advisor was Dr B. I contacted Dr B – also not a neuroscientist. No, he worked on the chapters on standards, what works and cost-benefit. Both Professor A and Dr B said they had questioned the use of the report’s cover image and had pointed out that it might be counterproductive.
This brings us back to the issues of expertise and use of evidence I highlighted in relation to Kanner and Bettelheim. The chapter on brain development in Allen’s Next Steps report doesn’t appear to have been written or checked by someone with expertise in the field – despite there being a number of high profile neuroscientists with excellent international reputations working in universities in the UK. It also raises questions about the role of ‘think tanks’. Despite inevitable shortcomings, universities have acquired a reputation for producing reasonably reliable, valid research findings; that is, research findings that reflect consistently and accurately the situation in the real world. Government policies based on a rigorous analysis of reliable, valid data are likely to be effective. Unfortunately politics tends to be not so much about the real world as about beliefs about the real world. Successful political campaigning involves persuading people that you are right and that those with opposing beliefs are wrong. It doesn’t matter, politically, if you haven’t understood the evidence or if you’ve misinterpreted the data, as long as you make a persuasive case for your policies. Of course for the people who benefit (or otherwise) from those policies, it’s quite important what evidence you are using, because if the evidence doesn’t reflect the situation in the real world, your policies won’t work. The adult social world might be open to manipulation and compromise, but as anyone who works with children, animals or materials will tell you, the real world is a pretty uncompromising place. The increasing reliance by politicians on material from think tanks founded by, or with strong links to politicians (as is the case with both the Centre for Social Justice and the Smith Institute who jointly published Early Intervention: Good Parents, Great Kids, Better Citizens) suggests we might see increasing amounts of policy-based evidence as distinct from evidence-based policy. It remains to be seen how the real world will respond.
De Bellis, M.D. (2005). The psychobiology of neglect, Child Maltreatment, 10, 150-172.
Glaser, D. (2000). Child abuse and neglect and the brain – a review, Journal of Child Psychology and Psychiatry, 41, 97-116.
Mareschal, D., Johnson, M., Sirois, S., Spratling, M.W., Thomas, M.S.C. & Westermann, G. (2007). Neuroconstructivism: How the brain constructs cognition, Oxford University Press.