Systematically Testing Nonlinear Dynamic Changes in Parent-Child Interaction in the Context of Randomized Intervention Trials

In the past 15 years there increased interest in the application of nonlinear dynamic systems to understanding the dynamics of close relationships and children’s development of psychopathology. Although conceptually rich, it has been challenging to formally define concepts such as attractors and repellors using inferential statistics.  Recently, however, progress has been made in the application of time-sensitive statistical techniques in a multilevel context to test hypotheses regarding duration and transitions (Stoolmiller & Snyder, 2006).

In this study, we examine the hypothesis that in early childhood individual differences in parent-child positive dyadic engagement and coercive dyadic engagement function as attractors in early childhood.  We test the hypothesis that rapidity of the dyads return time to the positive engagement and coercive dyadic states is a reasonable index of an attractor state. Moreover, we examine whether randomization to the Family Check-up (FCU) can improve the dynamic exchange patterns with children and families at age 3 and 5 years of age.

A multiethnic risk sample of 731 families in 3 distinct geographical settings who were receiving assistance from the Women, Infants, and Children (WIC) program were randomly assigned to the yearly FCU. The families were videotaped in the home in a variety of socialization tasks, and later coded using the Relationship Affect Coding System, which captures continuous coding of the parent and child behavior for both content and affect. State space grids were used to conceptualize four dyadic and dynamic states, when the parent and child were both: 1) Positive engagement, involving continuous neutral and positive affective exchanges; 2) Coercive engagement, including both in negative affect or content, as well as three other combinations of the above states.

Multilevel hazard models were examined to evaluate the extent that random assignment to the FCU predicted dynamic patterns of parent-child interaction at age 3 and 5, controlling for both child characteristics as well as coder macro ratings of parenting skills. It was found that the intervention decreased the return time for families to dyadic positive engagement controlling for children’s inhibitory control and ratings of parents’ skills in positive behavior support. In addition, the FCU was associated with decreases in the staying time for dyadic coercive engagement, leading to shorter conflict episodes. The findings do suggest the promise of focusing interventions on increasing the propensity of families to get ‘stuck’ in positive engagement attractors as well as to avoid successfully avoid repellor states of mutual negativity in the prevention of problem behavior and psychopathology.