As a commentary on the Special Issue, this paper discusses recent advances in the study of change across several time scales. It points out the importance of specifying time scales and putative patterns of change when characterizing problem behavior over developmental time scales. Methods for studying risk and protective mechanisms through observation of social interaction are also discussed as a way to study influences that operate in real time. Methods for studying episode-level interaction patterns that result from such real-time influences are also discussed, along with ways to integrate these with longitudinal assessments to study the effects of social interaction on problem behavior across developmental time scales.