Open AccessCausal Effects of Alcohol-Related Facebook Posts on Drinking Behavior: Longitudinal Experimental Study
Author(s) -
Hanneke Hendriks,
Wouter de Nooy,
Winifred A. Gebhardt,
Bas van den Putte
Publication year - 2021
Publication title -
jmir. journal of medical internet research/journal of medical internet research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.446
H-Index - 142
eISSN - 1439-4456
pISSN - 1438-8871
DOI - 10.2196/28237
Subject(s) - longitudinal study , social media , psychology , odds , alcohol consumption , social psychology , natural experiment , odds ratio , demography , medicine , alcohol , logistic regression , computer science , world wide web , biochemistry , chemistry , pathology , sociology
Background Adolescents and young adults frequently post alcohol-related content (ie, alcoholposts) on social media. This is problematic because both social norms theory and social learning theory suggest that viewing alcoholposts of peers could increase drinking behavior. It is therefore paramount to understand the effects of exposure to alcoholposts on viewers. Objective This study aimed to investigate the causal effects of exposure to alcoholposts on alcohol consumption by using a rigorous design. Methods We conducted a 6-week longitudinal study during which alcoholposts were measured by a newly developed app that copied Facebook posts shared by participants (n=281) to a new social media environment. In addition, daily questionnaires assessed alcohol use. Effects of natural alcoholposts (ie, posted by the participants) were assessed in phase 1, and effects of experimental posts (ie, posted by fake participants) were explored in phase 2. Results Results showed that natural alcoholposts increased the occurrence and quantity of drinking the following day. That is, exposure to a single additional alcoholpost increased the log odds of drinking the next day by 0.27 (b=.27, credible interval [CI] .18 to .35). Furthermore, the number of natural alcoholposts had a positive (predictive) effect on the number of glasses drunk the next day (b=.21, CI .14 to .29). In phase 2 when experimental posts were also present, these effects decreased. Experimental posts themselves had hardly any effects. Conclusions This study illustrates clear and direct effects of exposure to alcoholposts on next-day alcohol consumption and suggests that alcoholposts represent an important societal problem that interventions need to address.