A version of the research act, termed naturalistic behaviorism, is presented and compared to existing formulations of the method. A sensitizing framework for organizing naturalistic studies is presented. Special attention is given the problems of sampling, measurement, and causal analysis. Examples from an ongoing study of early childhood socialization are presented as tentative solutions to the sampling and analysis issues. Central weight is given to introspective-investigator accounts of social process. Existing formulations of naturalism as a distinct approach to empirical inquiry in the social sciences suffer from several overriding flaws. On the one hand, naturalistic theorists and practitioners have seldom been in agreement on what they mean by the method. For some (Catton, 1966), it is seen as rigorous positivism. For others (Matza, 1969), it is viewed as humanism in disguise. For still others (Barker, 1968; Hutt and Hutt, 1970; Willems and Rausch, 1969; Wright, 1967), it is compared to ecological psychology and/or ethology-a bare kind of behaviorism that studies people in their natural habitats. Here the naturalist, like the ethologist, makes little effort to record, probe, and study such socialpsychological processes as attitudes and definitions of the situation. There are those (Lofland, 1971) who view naturalism as a deep commitment to collect rich, often atheoretical ethnographic specimens of human behavior. These statements also suffer from a failure to specify the empirical phenomena to which the method is directed (e.g., if one observes behavior, what kinds of behavior?). Nor has there been any systematic attention given such traditional and perduring methodological problems as measurement, sampling, validity, reliability, and causal analysis. The basic unit of naturalistic analysis has never been clarified and the role of the naturalistic observer in his studies remains clouded. This conceptual diversity has led many to take a sceptical, if not irreverent, view of the naturalistic approach, viewing it as soft science or journalism. Perhaps the basic deficiency of prior naturalistic formulations has been the absence of a more general theoretical perspective that would integrate all phases of the sociological act. With few exceptions the dominant scientific paradigm has been imported from physics, chemistry, or biology.' In this article I offer a view of naturalism that takes as its point of departure the social behaviorism of Mead (1934; 1938) and the symbolic interactionism of Blumer (1969). I call this version of the research act naturalistic behaviorism and mean by the term the studied commitment to actively enter the worlds of native people and to render those worlds understandable from the standpoint of a theory that is grounded in the behaviors, languages, definitions, attitudes, and feelings of those studied. Naturalistic behaviorism attempts a wedding of the covert, private features of the social act with its public, behaviorally observable counterparts. It thus works back and forth between word and deed, definition and act. Naturalistic behaviorism aims for viable social theory, it takes rich ethnographic descriptions only as a point of departure. This version of behaviorism recognizes that humans have social selves and as such act in ways that reflect their unfolding definitions of the situation. The naturalist is thus obliged to enter people's minds, if only through retrospective accounts I am grateful to Philip Bechtel, Herbert Blumer, H. M. Blalock, Donald Dixon, John Lofland, Peter Manning, Anselm Strauss, and Clark McPhail for their comments and criticisms on earlier versions of this article. l The major exceptions here are Becker (1970a), Lofland (1971), and Schatzman and Strauss (in press).