An intensive longitudinal investigation was conducted on the social behavior of two three-year old boys in a nursery school setting over a four-month period to analyze observable stimuli in each subject's immediate social environment for the main determinants of his social interactive behavior. It was hypothesized that the daily rate of social interaction for each child would be highly variable, and that the fluctuations in a child's daily rate would be accounted for by the density with which key agents provided the social stimuli. A behavioral observation coding system which provided a sequential description of each subject's interactions in continuous form was used by two trained observers. With constant monitoring, observer reliability was maintained at 86 mean percent agreement. (Author)