We appreciate the work by Myung et al1 in conducting a meta-analysis of randomized controlled trials evaluating computer- and Web-based smoking cessation programs. However, there is a fundamental problem with the heterogeneity of the studies included, an underlying issue that reflects the state of the science rather than the authors' technique.
Of the 13 studies described as “computer based,” 9 involved mailing tailored print materials, a technique that is already known to be effective from previous meta-analyses,2 and which could for the most part be accomplished with research assistants and a typewriter. The remaining 4 studies, in which the subjects interacted with a computer, showed no statistically significant effect. In the 9 studies described as “Web based,” in which all participants presumably directly interacted with a computer program via the Internet, the interventions varied to such an extent that no useful class effect can be inferred. One of the most effective studies involved provided subjects with up to $200 in incentives for visiting a Web site where they described their health habits, took an interactive quiz, and read a college life magazine,3 while another simply supplemented traditional counseling and physical materials with manually written e-mail messages from research assistants.4