More preschool behavioural assessments improve predictions of aggression

This is the finding of a study by Professor Marcel Zentner (University of Innsbruck), Milana Smolkina (Institute of Psychiatry, King´s College London) and Professor Peter Venables (University of York) published in the British Journal of Developmental Psychology.

Professor Zentner said: "Identifying if a preschool child is at risk for the (later) development of behaviour disorders is a daunting diagnostic challenge. In psychology, as much as in other branches of science, forecasting is a process prone to error.

"In most long-term studies, life-outcomes such as delinquency or introversion have been predicted from behaviours assessed only at one single point in time (typically sometime in the first three or four years of life). Yet, this is quite like trying to forecast a cricket player´s batting ability from a performance in a single game. Few people would bet their savings on it. Just as any champion can have a bad day, so can a usually untroubled child."

It is a well-known statistical principle that insufficient behavioural sampling weakens any existing connection between a predictor and an outcome. However, the researchers are the first to examine the extent of this problem in a real life developmental study. The data used came from the Mauritius Child Health Project, a large-scale birth cohort study co-initiated by Professor Venables.

In this study aggressive-impulsive behaviours of 4-year-old children had been assessed over a period of 15 weeks, once every week. At ages 8 and 10 years, the same children were reassessed for aggressive symptoms, such as fighting and destroying other children´s belongings. Thus, it was possible to find out how much prediction accuracy would be affected by the number of initial observations.

The results showed a gradual strengthening of the association between early and late childhood aggression the more weekly observations informed the preschool rating:

"We´re talking about large increases, suggesting that single occasion measures of preschool behaviours may underestimate their predictive significance by as much as 50 per cent," Professor Zentner said.

"This is a significant finding in light of the widespread reliance on single occasion, behavioural measures of early childhood characteristics in long term studies of psychological development. As about six per cent of children and young people aged five to 16 show such disorders, the costs to the public sector are substantial. Our study indicates that strengthening preventive efforts by more reliable assessment techniques could pay off."