Longitudinal patterns in the development of physical activity (PA) and potential causal relationships between parent and child PA are examined.
Autoregressive models were used to examine bidirectional prospective paths between parent and child PA in a longitudinal sample of 351 Anglo and Mexican American families. PA was assessed independently in children and parents over a 13-y period.
There was little evidence for a causal path from mother PA to child PA.
Modeling does not appear to be the primary mechanism by which parents influence children’s PA behavior. Studies examining relations between parent and child behaviors should not rely on a single respondent for assessing both parent and child PA or on cross-sectional correlational data to make unidirectional causal inferences about determinants of child PA.