day life. To assess the likelihood that the positive association between exposure to television advertising and adiposity reflected a causal influence from advertising to adiposity, the committee examined three issues: (1) whether the association could be explained by adiposity affecting exposure to television advertising; (2) whether the association could be explained by other variables that the advertising exposure measure (television viewing) could also represent; and (3) whether the association could be explained by a third variable that influenced both the advertising exposure measure and adiposity.
If the significant association found between advertising exposure and adiposity represents a causal relationship, then the question becomes which influences the other. Because advertising exposure was measured by television viewing, the obtained association could logically be explained as “adiposity influences exposure to television advertising (as measured by television viewing)” or “exposure to television advertising (as measured by television viewing) influences adiposity.” Of the 73 results that examined the relationship between children’s and teens’ exposure to television advertising and their adiposity, 1 was experimental and the remaining 72 were correlational; 17 of these 72 correlational studies employed a longitudinal design (all panel studies). The experimental and longitudinal studies provide the best evidence as to the direction of any causal relationship that might exist between advertising exposure and adiposity.
The one experimental study (Robinson, 1999) was a well-conceived and well-conducted randomized controlled trial intervention. Nearly 200 third and fourth graders and their teachers and parents participated. Over a period of 6 months, children in the experimental intervention engaged in an 18-lesson classroom curriculum to reduce television, videotape, and video game use. The intervention effort was concentrated into the first 2 months, and part of the intervention sought to bring parents into the process. The analytic framework compared the intervention and control groups on a variety of outcome measures, while covarying on age, gender, and pretest score for the specific outcome measure. As compared to the control group, the intervention group had significantly less increase in BMI and in three of four other measures of adiposity. Also, for both child and parent measures, for the intervention group compared to the control group, there was a highly significant reduction (4–6 hours a week) in television viewing and a significant reduction (about one a week) in meals eaten in front of the television set. In addition, children in the intervention group compared to the control group reported relative reductions in servings of high-fat foods, although the differences were not significant (p = 0.12). There were no