Abstract
This study tests whether the impact of alcohol advertising exposure on intentions to drink and actual consumption is mediated by cognitive responses to advertising messages and positive expectancies about alcohol use. The model was tested using survey data of two important age cohorts, 15 to 20 years (n = 608) and 21 to 29 years (n = 612). The findings show that alcohol advertising was influential in shaping young people's attitudes and perceptions about alcohol advertising messages. The attitudes and perceptions predicted both positive expectancies and intentions to drink of those under the legal drinking age, but did not affect the young adults' expectancies and consumption. Positive expectancies were powerful predictors of intentions to drink and consumption for both groups. The effects of alcohol advertising on intentions to drink of those aged 15 to 20 years were mediated by cognitive responses to advertising messages and positive expectancies. The mediation effect was not evident among those between 21 and 29 years.
Notes
1Because of current alcohol laws restricting the legal drinking age at 21, we defined those aged 15 to 20 as underaged youth, and 21 to 29 as young adults in both the data collection and discussion of the findings.
2 CitationSmart (1988) adequately addressed the question of “does alcohol advertising affect overall consumption?” by reviewing research on alcohol advertising published up to 1987. Unfortunately, empirical studies using cognitive measures such as positive and negative beliefs about alcohol (CitationGrube & Wallack, 1994) and expectancies about alcohol drinking (CitationBrown, Creamer, & Stetson, 1987; CitationChristiansen et al., 1989) were not reviewed. Also not included in Smart's work were longitudinal studies (e.g., CitationChristiansen et al., 1989) to examine long-term exposure to alcohol advertising as acknowledged by the author (CitationSmart, 1988, p. 315).
3 CitationGrube & Wallack (1994) used a cross-sectional data of 468 fifth and sixth graders personally interviewed in northern California to examine the effects of television beer advertising on schoolchildren's knowledge, beliefs, and intentions to drink alcohol. The data were obtained during the first phase of a 3-year panel survey (CitationGrube, 1995). The use of measures of positive and negative beliefs about alcohol drinking in the authors' information-processing model was a necessary step going beyond using simple exposure (e.g., CitationAtkin et al., 1984), thus enhancing the overall quality of research on alcohol and youth. Prior to the CitationGrube & Wallack (1994) study, CitationWallack et al. (1990) content analyzed 195 episodes of prime-time television programming (173 hours) in the fall season of 1986. The purposes of their study were to (1) identify the frequency of alcohol appearances and alcohol acts, (2) examine the characteristics of drinking and non-drinking characters in prime-time television programming, and (3) see whether alcohol portrayals on television had changed from 1976 to 1986. The findings of the CitationAtkin et al. (1984) study were generated from a correlational data of 665 teenagers from the seventh through twelth grades in Michigan, California, New York, and Georgia.
Note: amale = 1, female = 0. byes = 1, no = 0. cIntentions to drink were administered to the 15–20-year-olds only, yes = 1, no = 0. dAlcohol consumption was administered to the 21–29-year-olds only.
Note: amale = 1, female = 0. byes = 1, no = 0. Entries are standardized beta coefficients. c p < .10. d p < .05. e p < .01. f p < .001.
Note: amale = 1, female = 0. byes = 1, no = 0. Entries are standardized beta coefficients. c p < .10. d p < .05. e p < .01. f p < .001.
Note. amale = 1, female = 0. byes = 1, no = 0. Entries are standardized beta coefficients. c p < .10. d p < .05. e p < .01. f p < .001.
Note: amale = 1, female = 0. byes = 1, no = 0. Entries are standardized beta coefficients.c p < .10. d p < .05. e p < .01. f p < .001.
Note: amale = 1, female = 0. byes = 1, no = 0. Entries are standardized beta coefficients. c p < .10. d p < .05. e p < .01. f p < .001.
Note: amale = 1, female = 0. byes = 1, no = 0. Entries are standardized beta coefficients.c p < .10. d p < .05. e p < .01. f p < .001.
4The CitationChristiansen et al. (1989) study was based on a two-year longitudinal sample consisted of more than 800 seventh and eighth graders in two suburban Detroit public school systems.