Incorporating trend data to aid in the causal interpretation of individual-level correlations among variables: examples focusing on the recent decline in marijuana use.

Given the close correspondence of several trends beginning in 1979, it is tempting to conclude that increases in perceived risk and disapproval led to the decline in actual use of marijuana. In this chapter, two alternative interpretations are considered, reflecting different hypotheses about individual-level causal processes: (1) changes in use led to the changes in attitudes, or (2) changes in some other factor(s) (e.g., increased "conventionality") caused both changes in use and changes in attitudes. This chapter documents a series of analyses designed to untangle such issues by incorporating trend data along with individual-level, cross-sectional relationships. One analysis strategy shows that controlling attitudes could "account for" the time trend in marijuana use, whereas the reverse is not true. The second analysis strategy examines how time trends in marijuana use are affected by multivariate controls for attitudes, as well as other individual characteristics, and shows that only the attitude measures can "explain" the time trend in marijuana use. Although these analyses are viewed as helping to explain the recent secular trend downward in marijuana use, as well as the still more recent decline in cocaine use, their most important contribution to prevention intervention research may be that they support a very basic generalization about individual-level causal processes: individual attitudes about specific drugs affect individuals use of those drugs.