Policy makers, like practitioners and members of the general public, are constantly faced with the need to decide when to take action on the basis of research findings supposedly showing that a particular individual characteristic or environmental circumstance is associated with a markedly increased risk for some negative outcome. Thus, over the years, campaigners have argued for the apparent need to prevent mothers from taking jobs outside the home, or to stop unmarried mothers from having children, or to restrict immigration, or to avoid immunization on the grounds that each of these carried serious risks for the children. But do they? How can we decide which research findings should lead to action and which should not?

In part, that issue involves asking which findings we should believe (is the claimed association real?); in part, it requires consideration of whether the causal inferences are justified; in part, it means questioning whether the proposed risk mechanisms are truly the ones that carry the risk; and, finally, it means considering whether the risks operate generally or only in certain circumstances. These questions constitute the major challenge for the whole field of social and behavioral sciences, and my purpose in this chapter is to discuss how they may be tackled. My messages are to caution against uncritical acceptance of claims regarding causal influences but to recognize that good research strategies are available to test causal inferences and to appreciate that these have led to some reasonably solid conclusions.