By Pierre Moessinger
Things are made out of components, and often have properties their components, once disconnected, do not possess ; such properties are « emergent ». A level is characterized by some unity among such properties. Sociology is dealing with several levels ; new levels will surely be discovered as sociological knowledge progresses. Multiplicity of levels seems to be the major specificity of sociology (and not the fact that it deals with intentions, feelings, or thoughts : so does also psychology). Explanation cannot ignore levels, because components of systems act as components (i.e., they would act differently if disconnected form other components in the system) and because emergent properties of a system cannot be explained out of the blue, but only from the interactions between the components of the system within its environment. Thus, there cannot be a good explanation at one level only.
Sociology should tend to be explanatory ; it cannot be purely hermeneutical, only descriptive, or simply « critical », for example. Although there is no satisfactory theory of explanation in sociology, there are many seminal attempts at explaining. I believe that explanation in sociology works through the construction of hypothetico-deductive systems, their integration into current knowledge, and the testing of them. True, sociological problems always refer to social phenomena, i.e., changes in social systems. The mechanisms of such changes are, most of the time, coordinations or interactions, be they between social systems, between individuals, or between social systems and individuals. Thus, good explanations (ultimately) refer to concrete systems. But since social systems have emergent or structural properties, explanations should also show how such properties emerge, and analyse the components of such systems. Many examples of attempts at explaining social systems can be found in Voir la société: liens micro-macro.