Why understanding the evidence is key to policy

Think tanks play a vital role in informing government and policy makers on the results of research as it applies to social policy so it is crucial, that they can trust the information they receive. However, as Professor Elizabeth Meins argues, complex academic findings need careful handling, and policy recommendations should be based on a detailed understanding of all the evidence.

photo credit: Ulka

The Sutton Trust’s report Baby Bonds: Parenting, attachment and a secure base for children is the latest in a line of well-intentioned, but misinformed calls for policy to promote secure attachment in children.

We already have the Parent Infant Partnership UK that aims to help ensure that all parents have securely attached children who will grow up to be “fulfilled, socially responsible adults”. Andrea Leadsom MP claimed in 2012 that secure attachment was “essential” for healthy development. I wouldn’t advise you to read her article, but if you do, please take everything she says with a huge dose of salt.

When attachment is used outside of its academic context, I worry that it does more harm than good…. I am against the complex academic findings on attachment being misused by think-tanks in their attempts to influence government policy.

Making recommendations

Personally, I would avoid making national policy recommendations unless I was really sure about my facts, but it seems that most people are considerably more gung-ho than I am. The executive summary of the Sutton Trust report states that:

  • secure attachment “is fundamental to [children’s] flourishing”,
  • “children without secure parental bonds are more likely to have behaviour and literacy problems”,
  •  [children] without strong bonds may be more likely to be NEET [not in education, employment or training], and less likely to be socially mobile and get good jobs in later life”.

The report concludes that “promoting secure attachment should be a focus for policy and services for families with children under age three”, and ends with the statement that targeted intervention to promote secure attachment “would represent a sound preventative investment”.

Unfortunately, none of the above claims about secure attachment are true. More unfortunately, it is likely that busy policy makers will, at best, only read the executive summary.

Gold standard

The ‘gold standard’ measure of early attachment is the strange situation procedure, which classifies toddlers into one of four attachment categories. The best evidence shows that in regular, middle-class families, 15% of children are insecure-avoidant, 9% are insecure-resistant and 15% are insecure-disorganised. This means that over a third of us are insecurely attached, so insecure attachment certainly isn’t abnormal.

Let us start with the things the report got right. There is some truth in its claim that early insecure attachment relates to higher levels of behavioural difficulties, but the effects are much more complex than the report suggests.

Several review articles have pooled and analysed data from multiple studies (so-called meta-analyses) to provide the most rigorous test of whether attachment predicts children’s later development. A meta-analysis showed that insecure attachment, and specifically insecure-disorganised attachment, are both associated with an increased risk of later externalising behaviours (e.g., conduct problems, hyperactivity).

However, the effects are not strong and the pattern of findings is different for boys and girls. While the report stated that the effect “was larger for boys than girls”, it failed to point out that disorganised attachment was associated with fewer externalising behaviours in girls. Does this mean that we should be trying to facilitate disorganised attachment in girls?

Dangerous claims

Another meta-analysis reported that the effect of insecure attachment on later internalising behaviours (e.g., anxiety, social withdrawal) was small, but statistically significant. For these types of behavioural difficulty, the link was specifically with insecure-avoidant attachment.

So making claims about the negative impact of “insecure attachment” is dangerous. In the cases where there is some effect, it is a specific sub-type of insecure attachment that predicts specific outcomes; the problem is that it is not always the same sub-type that relates to less optimal development.

What about the claim that early secure attachment is “fundamental to [children’s] flourishing” whereas insecure attachment is related to literacy problems? A meta-analysis showed that the effect of attachment on children’s later cognitive development is negligible. Although this meta-analysis (conducted almost 20 years ago) also reported that attachment predicted children’s language development, the authors stated that, due to the small number of studies included in the analysis, their findings “can only be considered provisional and exploratory”.

Only one study, conducted on a sample of around 300 children growing up in poverty in the US, has investigated relations between the child’s early social environment and their long-term educational attainment. In the publications from this study, attachment is often combined with other measures assessing the quality of the parent–child relationship or early home environment. This means that it is impossible to understand whether early attachment predicts children’s later educational attainment.

Unwarranted conclusions

The Sutton Trust report claimed that early insecure attachment predicted delinquency in adolescence and poorer job status in adulthood. This simply is not true. These claims are based on findings that measured attachment in adolescence or adulthood. Consequently, they tell us nothing about whether attachment security in early childhood predicts these long-term outcomes. So the report’s claim that secure attachment should be promoted because it provides “a surer basis for education and social mobility” hardly seems warranted on the basis of the evidence.

The report makes the classic error of assuming that attachment is stable over time. In fact, fewer than half of children show stability in attachment security over periods as short as 6 months. Five remarkable studies have investigated how strange situation classification in infancy relates to attachment in early adulthood. Three of them, including the largest observational, longitudinal study ever conducted in developmental psychology, found poor stability.

Interestingly, all studies implicated factors in the intervening years (e.g., parental divorce, family functioning in the teenage years) in young adults’ attachment security.

Interpreting the information

I am not pointing out the problems with this report because I am “anti-attachment”. I have devoted over 25 years of my life to studying the effects of early mother–baby interaction and attachment on children’s later development, and think that John Bowlby and Mary Ainsworth, the founders of attachment theory and research, are peerless in their contributions to developmental psychology. But I am against the complex academic findings on attachment being misused by think-tanks in their attempts to influence government policy.

It is a neat idea to think that attachment security acts like a crystal ball for predicting children’s futures. Unfortunately, the reality is not that simple. As well as the complexities in understanding whether attachment really does predict later development, there is also the problem of knowing the toddler’s attachment security in the first place.

You can not causally observe the caregiver and child together and then conclude that the child has a particular type of attachment; you have to do a formal observational assessment, which then must be coded by a trained and reliable researcher in order to establish the child’s attachment security.

Attachment labels tend to be used much too causally, and all too often, “insecurely attached” is interpreted as “unattached and unloved”. When attachment is used outside of its academic context, I worry that it does more harm than good. It can make parents feel concerned or guilty unnecessarily, and it muddies the waters for professionals who are trying to work out how best to intervene.

The belief that a person’s future is determined by the time they’re two years old is much too fatalistic for me. Surely none of us really believe this, otherwise, why are we bothering to try to help?

Elizabeth Meins is Professor of Psychology at York University.


Leave a Reply