TV, Social Media, and College Students’ Binge Drinking Intentions: Moderated Mediation Models

Abstract

Many studies to date have examined how media influence health-related behavior through social norms. However, most studies focused on traditional media. In the era of traditional and social media integration, our study advances health and mass communication scholarship by examining the influence of both traditional and social media mediated through social norms. Also, we examined a boundary condition for the norms-mediated media influence process. Namely, in the context of college binge drinking, we predict that exposure to TV and social media prodrinking messages can influence college students’ binge drinking intentions through perceived peer descriptive and injunctive norms. We also predict that group identification will moderate this indirect effect. Our moderated mediation models were tested via structural equation modeling (N = 609). We found that college students’ exposure to social media prodrinking messages indirectly influenced their binge drinking intentions via perceived injunctive norms, and students’ identification with their peers moderated this indirect effect. However, neither descriptive nor injunctive norms mediated the influence of students’ exposure to TV prodrinking messages on their binge drinking intentions. Theoretical and practical implications of the findings are discussed.

Notes

¹ Media could influence people’s normative beliefs possibly because (1) media portrayal of an event helps people develop a knowledge structure about the event and (2) repeated media exposure increases the accessibility of that knowledge structure to influence people’s social judgment. In a study, Kean and Albada (2003) found that TV did not differ from people’s personal experiences in shaping people’s understanding of alcohol and heavy TV consumers might have more accessible alcohol use schema than light TV consumers.

² The communication classes where participants were recruited required all students to complete some research posted on the communication department research web portal. Our study was one of the studies on the web portal. After reading a brief description of our study, students made voluntary choice about their participation. Only those who were registered for our study were provided with the survey link. At the end of our survey, participants needed to enter their unique research identity code, which was assigned to them at the time of their registration on the research web portal. Hence, only those who completed our survey were given course credit. The identity code was an arbitrary number and did not contain any personal information regarding the participants. For the purpose of our study, we required that participants were U.S. college students aged at least 18. We did not require all participants to be binge drinkers because measuring binge drinking intentions of non-binge drinkers is a common practice.

Because of the nature of our topic (i.e., binge drinking), we provided participants a very detailed confidentiality statement in the consent process. Participants were told that the principal investigators (PIs) would keep all of their information confidential by means of storing data only in the PIs’ password-protected computer, deleting participants’ identifying information (e.g., IP address) immediately after data collection, and not asking participants for their names during the survey. We also informed participants that they had rights to withdraw anytime during our study. To conclude the survey, we provided participants with the contact information of our university health center in case participants needed drinking counseling.

³ All of our scales were developed from previous research. Cho and Choi (2011) assessed college students’ exposure to tanned images on TV. They asked participants how often they watched reality shows, soap operas, sitcoms, and talk shows in which tanned men/women appeared. We made the following adaptations to their scale: (1) changed the focus of the scale from tanned men/women to prodrinking messages and (2) added a 30-day time frame (i.e., exposure in the past 30 days). Also, compared with talk shows, alcoholic commercials are a more important source of prodrinking message on TV. Thus, instead of asking people about their exposure to prodrinking messages on talk shows, we asked how often people were exposed to alcoholic commercials in the past 30 days. Yoo et al. (2016, p. 177) assessed college students’ exposure to prosmoking messages on social media. They asked students how often they have “seen or heard comments, questions, pictures, videos, or other information about prosmoking on” a blog, on a social networking site such as Facebook, MySpace, or Google+, Flickr or Tumblr, Twitter, YouTube, and other social networking sites in the past 6 months. We adapted their scale by changing the focus of the question from “prosmoking” to “prodrinking.” Also, we changed past 6 months to past 30 days given 30-day time frame was used more often to assess media exposure in prior norms-based media effects research (e.g., Ho et al., 2014). To measure descriptive norms, Paek (2009, p. 442) asked people: “What percentage of your peers would you say smoke cigarettes at least once a week?” We asked people to estimate the percentage of their university peers who were engaged in binge drinking by adapting Paek’s scale. For injunctive norms, Paek (2009, p. 442) asked people: “How do you think your peers think and feel about smoking?” People were asked to provide answers on two semantic differential scales anchored by not at all OK/perfectly fine and completely disapprove/completely approve. We adapted her scale to make it comparable with the descriptive norms scale. Our binge drinking intention measure was similar to that of Real and Rimal (2007). Real and Rimal (2007, p. 172) asked people how often they would, would like to, and “intended to go out drinking with their friends” in the next year. However, our participants included people under legal drinking age (< 21 years old). In this case, it is recommended to include some short-term time window (e.g., 30 days) to assess people’s drinking (NIAAA, 2017). We adapted Real and Rimal’s scale by adding three timeframes used in Jang et al.’s (2013) study, which were the next 7 days, 30 days, and 12 months. Hence, our intention measure could better capture the drinking intentions of people with various binge drinking status.

⁴ Based on previous research (Cho & Choi, 2011), it is possible that binge drinking intentions influenced people’s estimation of social norms. To eliminate this concern, we reversed the direction of the arrow line from descriptive or injunctive norms to behavioral intentions. Because the revised models and the original models were non-nested models, we compared their Akaike’s Information Criteria (AIC) and Bayesian Information Criteria (BIC) to find a better model. It was found that the revised injunctive norms-mediated model showed worse fit than the original model: AIC_revised = 19800.94 vs. AIC_original = 19784.86 and BIC_revised = 19924.48 vs. BIC _original = 19908.39. So did the revised descriptive norms-mediated model: AIC_revised = 19779.94 vs. AIC_original = 19759.56 and BIC_revised = 19903.47 vs. BIC_original = 19883.09.