Developing Effective Messages about Excessive Alcohol Consumption: Data-Driven Audience Segmentation for Resource Maximization

Theoretical Background and research questions/hypothesis: Excessive drinking is responsible for 1 in 10 total deaths among working-age adults each year and cost the U.S. $249 billion in 2010. The fourth leading preventable cause of death in the United States, assessment of knowledge and attitudes about excessive drinking was identified as critical to developing communication strategies—and with a small budget, efficiency in information collection was essential. This effort sought to understand:

RQ₁: What are the alcohol consumption profiles of the American general population?

RQ₂: How can those profiles inform audience research (both methodological and logistical decisions) and subsequent message development?

Methods: An audience segmentation analysis—focused on frequency of drinking, intensity of drinking, and prevalence of binge drinking—was conducted using publicly available survey data. More specifically, latent class cluster analysis was used to segment the Behavior Risk Factor Surveillance System (BRFSS) survey samples from 2013 and 2014. Both survey samples were weighted by the final analytical weights to allow for estimates of non-institutionalized adults (i.e., 18+) residing in the U.S.  

Results: The analysis resulted in six mutually-exclusive consumption profiles, which were named and narratively described to increase their practical value:

1. Binge Drinkers (16% of population): Drink frequently and in greatest volume. Young-to-middle-aged, with almost half under 35 years old. Socio-economically diverse; skew mid-to-high income, but about 25% make less than $25k per year and nearly half have a HS diploma or less.
2. Drinking from Disadvantage (14% of population): Mostly young-to-middle-aged and less socioeconomically affluent in addition to being the most racio-ethnically diverse. Drink infrequently but in higher intensity.
3. Casual Drinking Young Adults (12% of population): Primarily young adults entering college or the workforce. Relatively racio-ethnically diverse. Tend to drink infrequently but in somewhat higher volume.
4. Nightcap Generation (15% of population): Mostly older, White, married, and mid-to-high income. Despite drinking frequently, do not drink in great volume.
5. Mid-life Moderation (28% of population): Mostly middle-aged, White, married, and affluent. Drink somewhat frequently and of moderate intensity.
6. Abstaining but Troubled (15% of population): Older, socio-economically struggling people in poorer mental and physical health who tend to be female, separated or divorced, and not in the labor force.

Conclusions: A data-driven approach to audience segmentation informed priority audiences and logistical considerations (i.e., screening criteria, grouping variables, analysis plan, and location selection) for upcoming qualitative research in three states. The focus groups will assess knowledge and attitudes related to excessive drinking, and the findings will serve as the foundation for message development. Importantly, the replicable segmentation process, the creation of an interpretable narrative around the results, and subsequent application allowed the team to maximize limited resources and stretch the learning potential of a small number of focus groups. The impact of this process would also scale to larger efforts. 

Implications for research and/or practice: When public health researchers and practitioners may not have the resources available to conduct original research, leveraging publicly-available data can be an effective, efficient, and budget-friendly means to segment a population toward end goals of prioritizing and locating audiences as well as developing and disseminating tailored messages.