Several features of science are crucial. Scientific claims are intended to be true and rationally supportable. Scientific knowledge is based on empirical research and rigorous reasoning. Science provides a basis for explaining the phenomena it considers. And science depends upon the idea of critical research communities: peers and collaborators who challenge, test, evaluate, and extend a set of results. In addition to these core features, positivist philosophers have added a few features drawn from the experience of the natural sciences. These observers hold that science involves the discovery of strong generalizations and laws that describe and govern empirical phenomena; that explanation means subsumption under some of these laws; and that natural phenomena constitute a law-governed, interconnected system.
Positivism is a poor theory of the social sciences, because the phenomena of social life do not conform to the law-governed ontology stipulated for natural phenomena. But let us stick with the minimal set of features and consider what kind of science is possible for social life.
First, sociology involves description. Social phenomena are observable, and it is straightforward to design rigorous research efforts aimed at establishing the facts about a particular domain. This aspect of sociology involves rigorous empirical study of social phenomena. Examples of descriptive research include ethnographic research and micro-sociology along the lines of the Chicago School. But large-scale description is feasible as well, including empirical description of large social patterns and institutions.
Descriptive findings often take the form of statistical estimates of the frequency of a feature within a group--for example, rates of suicide among Protestants (Durkheim). Properties may be correlated with one another within a given population; variation in one variable may be associated with variation in another variable. Descriptive research can thus sometimes reveal patterns of behavior or social outcomes--for example, patterns of habitation and health status. And patterns such as these invite efforts to find causal relationships among the characteristics enumerated.
Second, sociology involves discovery of social causation and mechanisms. Is there such a thing as social causation? What does social causation derive from? What is the ontology of "social necessity" (analogous to natural necessity)--the way in which one set of circumstances "brings about" another set of outcomes? In general we can begin with an ontology grounded in purposive social action by agents within institutional settings and environments. Social causation derives from the patterns of behavior that are produced in this setting. (For example, we can explain the degradation of environmental quality of a common resource as the consequence of free-riding behavior.)
Third, sociology can provide explanations of some social outcomes as a causal consequence of proposed social mechanisms. Once we have a generic idea of what social causal mechanisms look like, we can turn to specifics and try to discover the processes through which behavior
is created and constrained. So we can try to discover or hypothesize the mechanisms through which tropical agriculture tends to under-serve farmers (Bates). Social theories are hypotheses about social causal mechanisms; so theories provide a basis for explanation of social phenomena.
Research and explanation along these lines creates major and visible limitations on the degree of systematicity, interconnection, and determination that sould be expected of social phenomena. The social world is highly contingent, the product of many independent actors. So we should only expect a weak degree of systematic variation among social phenomena.
Finally, the epistemic setting provided by the disciplinary institutions offers a basis for estimating the rational credibility of social science knowledge: journals, peer review, tenure evaluation. Social research and explanation remains fairly close to the level of the facts. Researchers in the disciplines and sub-disciplines are charged to test and explore the empirical and theoretical claims of their peers.
The results of a science including these components will be empirically disciplined, theoretically eclectic, and systemically modest. The goal of providing an over-arching theory that demonstrates the systematic integration of the social is abandoned.
(It is worth noting that there is a period in the history of sociology when the epistemic values of the discipline were most consistent with this view. That was the period of the Chicago School. See Andrew Abbott's Department and Discipline: Chicago Sociology at One Hundred).