Identifying patterns of resilience using classification trees

This paper introduces and illustrates the value of a classification trees approach in the study of resilience. The inherently interactive nature of the resilience construct makes this approach useful. Classification trees are a person-centred approach to data analysis, which successively split the sample into pairs of increasingly homogeneous groups of individuals. We outline the approach and then illustrate using adult educational outcomes for children in the British Cohort Study of 1970 who had experienced foster care. The insights gained from the classification tree approach are contrasted with those obtained from standard regression approaches.