Social scientists routinely draw conclusions about cause and effect from their data. This book spells out the pre-statistical assumptions of multivariate research and explains in nonmathematical terms: the concepts of causal direction and system order; direct, indirect, and spurious statistical effects; signs and the sign rule; rules for introducing control variables, elaboration and explanation, "effects analysis," and path analysis. The book is not statistical in the sense of developing specific statistical tools. Rather, it explains the prestatistical assumptions required, whatever the technique. The importance of substantive knowledge about the "real world" is stressed, and the myth that causal problems can be solved by statistical calculations alone is repeatedly challenged.