The field of developmental psychopathology (Achenbach, 1974; Sroufe & Rutter, 1984) identifies risk factors and models the processes by which risks relate to behavior problems. One type of model posits that risks add in a linear way to increase the chances that a child will develop problems. This type of model has been well supported. Research to date has established that adverse rearing conditions, such as excessively hostile parenting or poverty (e.g., Rothbaum & Weisz, 1994), and adverse personal qualities of children, such as neurological dysfunctions or difficult temperament (Moffitt, 1993; Rothbart & Bates, 2006), are correlated with behavior problem outcomes. The more such adversities are present, the stronger the prediction of later behavior problems (Appleyard, Egeland, van Dulmen, & Sroufe, 2005; Deater-Deckard, Dodge, Bates, & Pettit, 1998). Some additive risk models postulate mediating factors and thus provide a more satisfying account of developmental process; for example, the experience of physical abuse leads to deviant social cognition, which, in turn, leads to aggressive behavior (Dodge, Bates, & Pettit, 1990; Dodge, Pettit, Bates, & Valente, 1995).

However, linear combinations of even relatively large numbers of predictors tend to account for less than half of the variance in behavioral adjustment (e.g., Deater-Deckard et al., 1998). Part of this shortcoming may be due to failure to specify enough of the many possible risk factors.