Parents of acting-out children tend to use coercive discipline tactics more frequently than do parents of well-behaved children, whether the parenting is assessed concurrent with or antecedent to child adjustment (Campbell, 1990; Hetherington & Martin, 1986; Patterson, 1982). Coercive discipline is central to most models of child adjustment. However, empirically, parent–child interaction indexes in previous studies have accounted for only moderate portions of the variance in child outcomes, especially when used as longitudinal predictors. One implication of this finding is that models need to include additional predictor measures. Many researchers, influenced by general systems models, are now assessing various other possible predictors, including positive parenting in addition to coercion, marital quality, socioeconomic stress, child temperament, and child intelligence.

For the present chapter, however, the focus is more on a different implication of the limited variance in adjustment outcomes accounted for by parental coerciveness: in addition to inadequate definition of the predictor, there could also be inadequate definition of the outcome. Clinical accounts of the complexities of individuals' adjustment histories and modern systems theory suggest the need for more richly described criterion variables, including a multidimensional, multisource definition of adjustment, as well as a dynamic, cross-time definition. To build useful dynamic models, we must try to improve models of developmental stability versus change.