Not everybody is a fan of Michael Rutter, or of scientists who follow a similar approach. But more or less everybody agrees that his work has transformed how we think about child health and development (just as similar work has transformed how we think of adult health and development).

The observation and quantitative work has led to a definition on which most agree, and a tool to operationalise the definition. The tool is not 100 per cent accurate. It needs the validation of a trained clinician. But it isn’t dependent on the state need to justify intervention, and it isn’t subject to variation in how clinicians see the world.

When we agree on the definition we can begin to understand cause. If we are confident that Michael has a conduct disorder but his brother does not then we can investigate what might have happened that led one family member to get sick while the other remained well. We can work out the risks for conduct disorder.

And if we get to agree on those risks -coercive parenting perhaps, a delinquent father, maternal deprivation- then we can begin to formulate interventions to prevent those risks.

Good measures like the SDQ also tell us about incidence rates. Before the epidemiology the amount of mental ill-health was little understood. Most of what was seen was exhibited by boys in institutions, misleading the lay person into thinking that conduct disorder was rare and reserved for the working class. We now know from say Meltzer’s work with the ONS that around a fifth of children and young people experience mental ill-health, and that it crosses all social strata.

When we have the same measures applied at successive periods of time we can chart historical patterns, viz Maughan and Collishaw’s groundbreaking work showing how mental health of adolescents improved in the 1990s.