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

Abstract

Background Facebook’s advertising platform reaches most US households and has been used for health-related research recruitment. The platform allows for advertising segmentation by age, gender, and location; however, it does not explicitly allow for targeting by race or ethnicity to facilitate a diverse participant pool. Objective This study looked at the efficacy of zip code targeting in Facebook advertising to reach blacks/African Americans and Hispanics/Latinos who smoke daily for a quit-smoking web-based social media study. Methods We ran a general market campaign for 61 weeks using all continental US zip codes as a baseline. Concurrently, we ran 2 campaigns to reach black/African American and Hispanic-/Latino-identified adults, targeting zip codes ranked first by the percentage of households of the racial or ethnic group of interest and then by cigarette expenditure per household. We also ran a Spanish language campaign for 13 weeks, targeting all continental US zip codes but utilizing Facebook’s Spanish language targeting. The advertising images and language were common across campaigns. Costs were compared for advertisement clicks, queries, applications, and participants, and yields were compared for the final three outcomes. We examined outcomes before and after the Cambridge Analytica scandal that broke in March 2018. Finally, we examined 2 promoted Facebook features: lookalike audiences and audience network placement. Results Zip code targeting campaigns were effective for yielding the racial or ethnic groups of interest. The black-/African American–focused versus general market campaign increased black/African American weekly queries (mean 9.48, SD 5.69 vs general market mean 2.83, SD 2.05; P<.001) and applicants (mean 1.11, SD 1.21 vs general market mean 0.54, SD 0.58; P<.001). The Hispanic-/Latino-focused versus general market campaign increased Hispanic/Latino weekly queries (mean 3.10, SD 2.16 vs general market mean 0.71, SD 0.48; P<.001) and applicants (mean 0.36, SD 0.55 vs general market mean 0.10, SD 0.14; P=.001). Cost metrics did not differ between campaigns at generating participants (overall P=.54). Costs increased post- versus prescandal for the black-/African American–focused campaign for queries (mean US $8.51, SD 3.08 vs US $5.87, SD 1.89; P=.001) and applicants (mean US $59.64, SD 35.63 vs US $38.96, SD 28.31; P=.004) and for the Hispanic-/Latino-focused campaign for queries (mean US $9.24, SD 4.74 vs US $7.04, SD 3.39; P=.005) and applicants (mean US $61.19, SD 40.08 vs US $38.19, SD 21.20; P=.001). Conclusions Zip code targeting in Facebook advertising is an effective way to recruit diverse populations for health-based interventions. Audience network placement should be avoided. The Facebook lookalike audience may not be necessary for recruitment, with drawbacks including an unknown algorithm and unclear use of Facebook user data, and so public concerns around data privacy should be considered. Trial Registration ClinicalTrial.gov NCT02823028; https://clinicaltrials.gov/ct2/show/NCT02823028