I am going to use the example of Michael Rutter. There are many examples from which to choose, Garmezy in the United States to name just one, but Rutter was the one that had most influence on the way I thought about evidence.

Rutter starts with the question of definition. What is autism? A subject to which he has devoted a lifetime of study. I will use examples of ordinary childhood mental ill-health, although I could have chosen other scientists putting some form on say obesity or Alzheimer’s. There is a consistency to the way good scientists work.

The starting point is epidemiology, the study of disease, its origins and sequelae. Rutter and his colleagues spent a lot of time in places like Camberwell in South London and the Isle of Wight, visiting families with children, ordinary families, what would be called in the trade a normative sample.

In the home they would note down the features of common disorders, when did anxiety become an impediment, how was it different from depression or, to use a word typical of the day, delinquency. Eric Taylor, a long time colleague of Michael Rutter used the same approach with regards to hyperactivity.

The result of this early work was a series of heuristics, a rough sense of what these different types of mental ill health might comprise, and how they differ from each other.

The next step was to codify the heuristics into a questionnaire or checklist, what became known as the Rutter questionnaires, which could be applied in numbers to whole populations of children, again in Camberwell and the Isle of Wight.

Later on Robert Goodman improved the questionnaires to produce the Strengths and Difficulties Questionnaire, or SDQ, which has now been applied in over 100 countries.

The SDQ is a measure, or a screener. It comprises 25 items. It doesn’t ask a child or a parent or a teacher who knows the child ‘does Michael have a conduct disorder?’. But the 25 questions predict with a high degree of accuracy whether Michael or any other child has a disorder of depression, conduct or peer relationships. A child scoring positive on the SDQ for one of these disorders would be recognised as such by a psychiatrist whatever her training.