Methodological issues must be examined within the context of meta-theoretical perspectives. More specifically, at the level of drawing inferences from data, metatheory prescribes methodology. This argument is illustrated by examination of the validity of inferences in general, and external validity in particular, from two perspectives: a traditional perspective and a dialectical perspective. The traditional perspective, rooted in the mechanistic model, defines external validity as the generalizability of inferences across persons, settings, and time. External validity is also seen as the final validity concern, logically preceded by statistical conclusion, internal, and construct validity. The dialectical perspective, rooted in synthetic models, defines external validity as the adequate dimension-alization of organized complexities. External validity is also seen as the primary validity concern, logically followed by construct, internal, and statistical conclusion validity. Each of these contrasting views of the validity of inferences provides a useful perspective for the study of human development.