Ubiquitous in natural science, and remarkable in their simplicity, physical laws rely on a few independent variables in relationships that explain behaviors of complex systems. Formulation of laws involves making simplifying assumptions, finding limiting conditions under which they hold, and recognizing cases where they do not hold. Somewhat less rigorous than pure law, a "lawlike regularity" must regularly provide a "reasonably good" explication of the evidence. While laws reflect accurately a specific system's behavior, the utility of a lawlike regularity lies in the variety of cases to which it applies, rather than precision of fit to any particular case.
Awareness of how a lawlike relationship fits into other available information outside the regularity is as important as discovery of the regularity itself. Additional information may have great theoretical and empirical importance, or lead to refinements of the lawlike relationship. A regular relationship cannot be said to describe the lawlike aspects of a process unless there exists a viable theoretical explanation of why the observed relationship is what it is. Without a theoretical model of the process underlying a regularity, its worth as a lawlike relationship cannot be determined. Moreover, the operation of such a model must be verifiable independently of the regularity itself. As in the physical sciences, explanations of social scientific systems need not be complicated, complex, or even profound.
Neither laws nor lawlike relationships are necessarily causal; indeed they may originate in the coincidence of chance with purpose. Specification of causal ordering suffers from each of two contradictory weaknesses:
Because models can at best approximate reality, some variables are always excluded from any equation; it is therefore not possible to establish dear causality among several interrelated variables.
Reality does not consist of simultaneous causal dependencies, but rather of recursive, hierarchical relationships; it is therefore, possible to determine dear causal relations.