Sociological theory, if it is to advance significantly, must proceed on these interconnected planes: (1) by developing special theories from which to derive hypotheses that can be empirically investigated and (2) by evolving a progressively more general conceptual scheme that is adequate to consolidate groups of special theories.
The term "middle-range theory" does not refer to a specific theory, but is rather an approach to theory construction. Raymond Boudon defines middle-range theory as a commitment to two ideas. The first is positive, and describes what such theories should do: sociological theories, like all scientific theories, should aim to consolidate otherwise segregated hypotheses and empirical regularities; "if a 'theory' is valid, it 'explains' and in other words 'consolidates' and federates empirical regularities which on their side would appear otherwise segregated." The other is negative, and it relates to what theory cannot do: "it is hopeless and quixotic to try to determine the overarching independent variable that would operate in all social processes, or to determine the essential feature of social structure, or to find out the two, three, or four couples of concepts ... that would be sufficient to analyze all social phenomena".