of those under 18 drank, compared with 61 percent and 19 percent, respectively, for the U.S. population in these age groups.

Little research has addressed the potential effects of exposure to drinking on television on young people’s drinking beliefs and behaviors. Generally speaking, correlational studies have found small, but statistically significant, relations between television viewing and alcohol-related beliefs and behaviors. Thus, Tucker (1985) found that high school boys who were heavier television viewers drank more than lighter viewers. Similarly, Neuendorf (1985) reported that television viewing was related to beliefs about drinking among 10- to 14-year-old adolescents: Heavier viewers were more likely than lighter viewers to agree that people who drink are happy and you have to drink to have fun at a sporting event.

More recently, in a prospective study of 1,533 ninth-grade students, it was found that television viewing was related to initiation of drinking over an 18-month period (Robinson, Chen, and Killen, 1998). Specifically, each 1-hour increase in television viewing at baseline was associated with a 9 percent increased risk of initiating drinking during the following 18 months (OR = 1.09), after controlling for age, gender, and other media use. Unexpectedly, however, each hour of watching taped programming and movies on video was associated with an 11 percent average decrease in drinking initiation. Moreover, drinkers and nondrinkers did not differ in weekly hours of television viewing at baseline, and television viewing was not associated with increases in consumption among those young people who were already drinkers at baseline. A final study investigated reported television viewing and scores on a risky behavior scale that included drinking for a sample of 14-to 16-year-old adolescents (Klein et al., 1993). Although significant positive relations were found between viewing and involvement in risky behaviors for specific genres (e.g., cartoons), the results were inconsistent across genres and no effect was found for overall TV viewing. Moreover, data relating specifically to drinking were not presented.

These correlational studies suffer from potentially serious conceptual and methodological problems. Conceptually, none of the studies directly measured exposure to televised drinking portrayals. Rather, they relied only on measures of overall television viewing. The problem with such measures is that children watching equal amounts of television may be differentially exposed to alcohol portrayals depending on their program preferences and attention levels. More importantly, all of these studies used correlational analyses that cannot provide evidence for the direction of the relationship between television viewing and drinking beliefs and behaviors. Some unconsidered third variable may influence both viewing and drinking. This interpretation cannot be entirely discounted, even for the single longitudinal study.