Figure 4.

Revised developmental taxonomy for the emergence, desistence and persistence of conduct disorder across childhood and adolescence. The model depicts five clusters of children or adolescents within a matrix of individual and environmental vulnerabilities for conduct disorder. It is the combined effects of these vulnerability dimensions, and their nature and timing over the first two decades of life, that result in differential probabilistic risks for conduct disorder and the likelihood of persistence or desistence over time. The model illustrates our argument that the differences between childhood-onset and adolescence-onset conduct disorder are quantitative, rather than qualitative, in nature (although childhood-onset forms tend to be more severe and persistent, and may require a higher loading of environmental or individual-level risk factors). As the childhood-onset persistent and childhood-limited conduct disorder subtypes appear to share many risk factors in common, we propose that individual-level risk factors are high in both subtypes. Note that those in the ‘normative experimentation’ cluster would not merit a diagnosis of conduct disorder. Although we have been informed by the empirical literature, the dashed lines around the clusters indicate that the available evidence is too limited to confirm their exact position in the two-dimensional matrix. The boxes also show where the highest concentration of individuals in each group is expected to occur, but they are not necessarily restricted to these regions of the matrix