This paper presents arguments for greater use of sophisticated research strategies and statistical analyses of data in dealing with the real-life complexity of social interaction. Three possible procedures are proposed and illustrated in a study of interactions between mothers and their 8- to 18-month-old children. The first strategy is to operationalize complex concepts and define loose constructs behaviorally and quantitatively (e.g. maternal responsiveness.) The second general research strategy is to analyze static complexity. It is suggested that a wide range of variables be measured in the same investigation and that rather than computing individual correlations between variables, multivariate statistical analyses should be performed. This method of analysis would serve to uncover broad patterns of behavior which would more likely reflect the complexity of human nature. The third general research strategy suggested involves examining behavior patterns and mother-child relations as they change over time. This strategy calls for a longitudinal design and the use of cross-lagged panel correlation. (BRT)