Smoking prevention: implications of study design, research setting, and goals.

Collaborations among various disciplines concerned with smoking prevention are gaining in currency. Such collaborations are predicated, in part, on recognizing both the implicit assumptions in different fields and the consequences of these assumptions. These issues, however, are often ignored in transdisciplinary research. In this article, we demonstrate how simple indices, such as the risk ratio, attributable fraction, or R(2), may have different implications for causal inference and interventions, depending on study setting and research goals.