One of the particularities of applied anthropology is working on demand, and performing research on demand requires changing fields constantly. This diversity of fields has led to an awareness in applied anthropology that the focal point of observation varies from study to study, and that depending on the particular scope or decoupage, researchers do not see the same thing. This scales-of-observation method has four empirical principles: (a) What one observes at one scale vanishes at another scale. (b) The causes explaining actors’ behavior vary based on the scale of observation; they can stem from situational effects or meaning effects, or suggest statistical correlation. (c) Knowledge acquired at one scale is complementary and cumulative with that of other scales of observation. However, they cannot be fused into a single, global description. Indeed, although reality is continuous, observation between the “macro” and the “micro” is discontinuous. Discontinuity stems from the importance of the situational effects in anthropology and organizational sociology. These two approaches are most often centered on the interactions among actors operating under situational constraints. All generalizations are thus limited to scales pertaining to the same type of causality. (d) Part of the conflict among schools, disciplines, or professions regarding explanations for human behavior and changes within a community, an organization, a society, or an individual can most often be explained by different choices in the scale of observation. The scales-of-observation method is a mobile tool of knowledge founded on the anthropological practice of the cultural detour, in this case scientific cultures. It is an inductive epistemological theory on the variability of the explanatory causes of human behavior and falls under methodological relativism. Consequently, the scales-of-observation method is also a tool of negotiation among actors who are involved collectively in a project of social change, but with contradictory interests or objectives.