we conclude the section by presenting an agenda for such diversity-related research, rather than a summary of what is known.
Good evidence exists that the periods of operation of some national campaigns have been associated with periods of improving overall levels of problematic health behaviors and health outcomes. The operation of the National High Blood Pressure Education Program (Roccella, 2002), National Cholesterol Education Program (2001), Back to Sleep Campaign (Willinger et al., 2000), the CDC AIDS campaign, several European AIDS campaigns (Wellings, 2002; Dubois-Arber et al., 1997), the urban vaccination campaign in the Philippines (Zimicki et al., 1994), and the California (Pierce, Emery, and Gilpin, 2002), Florida (Sly et al., 2001), and Massachusetts (Siegel and Biener, 2000) antitobacco campaigns all are associated with periods of sharp change in their target behaviors. Although simple association of trends over time with campaign initiation is not sufficient grounds for claiming a causal effect, many of the evaluations have additional reasons for attributing the observed change to their efforts. These include reports of high exposure to their messages, evidence for change in intermediate process variables that were the direct targets of the campaigns, and evidence that those reporting more exposure to the campaigns were particularly likely to change (Hornik, 2002).
However, evidence that some campaigns are associated with good outcomes is not evidence that their particular strategies for dealing with diversity were effective, particularly when compared with alternative approaches. What sort of evidence would be relevant to the diversity issue? We outline a variety of levels of evidence that might bear on this issue.
This evidence would require examination of the trajectories of the subgroups over the period of campaign operation. Presum-