Facebook Recruitment Using Zip Codes to Improve Diversity in Health Research: Longitudinal Observational Study

Area of Science:

• Digital Health
• Public Health Interventions
• Social Media Research

Background:

• Facebook's advertising platform is widely accessible in the US and has been utilized for health research participant recruitment.
• The platform offers segmentation by age, gender, and location, but lacks direct targeting for race/ethnicity, posing challenges for diversity in study populations.
• Recruiting diverse participants is crucial for generalizability and equity in health research.

Purpose of the Study:

• To evaluate the effectiveness of zip code-based targeting within Facebook advertising for recruiting Black/African American and Hispanic/Latino daily smokers for a web-based, social media smoking cessation study.
• To compare the efficacy and cost-effectiveness of zip code targeting against general market campaigns.
• To assess the impact of the Cambridge Analytica scandal on recruitment outcomes and explore the utility of Facebook's Lookalike Audiences and Audience Network.

Main Methods:

• A general market Facebook advertising campaign was conducted across all continental US zip codes for 61 weeks.
• Two targeted campaigns focused on Black/African American and Hispanic/Latino adults, prioritizing zip codes with higher percentages of these groups and higher cigarette expenditures.
• A Spanish-language campaign ran for 13 weeks, using Facebook's language targeting. Outcomes and costs were compared pre- and post-Cambridge Analytica scandal.

Main Results:

• Zip code targeting significantly increased weekly queries and applicants from the targeted racial and ethnic groups compared to the general market campaign.
• The Black/African American-focused campaign yielded a mean of 9.48 queries (vs. 2.83) and 1.11 applicants (vs. 0.54) weekly.
• The Hispanic/Latino-focused campaign yielded a mean of 3.10 queries (vs. 0.71) and 0.36 applicants (vs. 0.10) weekly. Cost per participant did not differ significantly between campaigns.
• Costs for queries and applicants increased post-scandal for both targeted campaigns.

Conclusions:

• Facebook's zip code targeting is a viable strategy for recruiting diverse populations, particularly racial and ethnic minorities, for health interventions.
• The Audience Network placement feature should be avoided due to ineffectiveness or potential negative impact.
• Lookalike Audiences may not be essential for recruitment and raise concerns regarding data privacy and algorithmic transparency.