Evaluation of interventions for gay men: two desiderata

This paper was presented in a session at the AIDS Impact conference devoted to a debate on the methods that should be used to evaluate educational interventions. The paper highlights two desiderata for evaluation of interventions directed at gay men. First, the view is presented that there is no acceptable substitute for assessing the effect of an intervention on gay men's sexual behaviour (rather than, for example, their AIDS-related attitudes or beliefs). This view is justified in terms of (a) the differences that exist between AIDS-related thinking in the cold light of day and during actual sexual encounters; and (b) the often faulty nature of intuitions about the factors that contribute to sexual risk-taking and the ways in which it might be reduced. Second, it is argued that the randomized control study design represents the best means for ensuring that interventions will be as effective as possible. Criticisms which have been made of this design are discussed and the conclusion drawn that they do not amount to a strong case against it.