On evaluating mass medias impact [letter]

The author of this letter to the editor raises concerns about Yoder Hornik and Chirwas use of ownership of a working radio and listening at least once per week to any program as a proxy for exposure to a radio drama about acquired immunodeficiency syndrome (AIDS) in Zambia. This methodology resulted in a misspecification of treatment groups; only 40% of the high-access group reported listening to the AIDS drama compared with 13% of the low-access group. Since the treatment group was not fully exposed and the control group was substantially contaminated accurate assessment of program effects was compromised. Moreover respondents with low access are likely to have fewer competing sources of information than those with high access increasing their likelihood of being influenced by the radio program. Finally salient knowledge and behavior levels are likely to be higher at baseline in the high-access group. Preferable would be measurement of exposure to the specific program under study and computation of this exposures association with the outcomes under review. Re-analysis of the original data revealed that knowledge about AIDS increased significantly on five knowledge scales as a result of program exposure and misinformation and incorrect knowledge decreased. Women exposed to the drama had higher levels of reported partner reduction than those not exposed. The concern is that researchers who adopt the methodology used in the Zambian study will neglect to collect program exposure data and thus be incapable of detecting program effects or measuring their magnitude.