Combining Social Media Recruitment with Aggregated Relational Data to Study People Who Use Drugs

• In Event: Novel Methods In Drugs and Health Research

Abstract

Studying the social networks of stigmatized populations presents two persistent methodological challenges: recruiting participants who avoid institutional contact, and collecting network data without enumerating individual connections. This study addresses both challenges by combining social media advertisement recruitment with aggregated relational data (ARD) collection among people who use drugs (PWUD) in King County, Washington. We recruited participants through Facebook and Instagram conversion campaigns using neutral imagery, monetary incentives, and quota sampling across age and gender categories. Rather than requiring participants to list specific social contacts, we collected aggregated relational data (ARD) by asking respondents to estimate the number of people they know who have a particular trait (e.g., opioid use), distinguishing between trusted close contacts and broader acquaintances. A three-day pilot campaign with a $40 budget recruited 143 eligible participants at $0.28 per completed survey—substantially below the $0.52–$17.00 range reported in existing social media recruitment studies. The sample demonstrated notable demographic diversity: 37% Black or African American, 19% Hispanic, 17% Asian, and 7% American Indian or Alaska Native. Substance use patterns closely aligned with national estimates, with 85% reporting lifetime use of both opioids and stimulants. Network data revealed a structural divergence between relationship types: acquaintance networks followed a right-skewed distribution extending beyond 20 contacts, while trust networks concentrated heavily at 0–2 people. These pilot findings demonstrate that social media advertisements can rapidly and cost-effectively recruit diverse samples of PWUD for network research, and that ARD methods successfully capture meaningful variation in network structure among a population typically excluded from quantitative network studies. Full-scale data collection targeting 2,000 participants is currently underway. We discuss implications for researchers seeking to apply these methods to other hard-to-reach populations.

Authors

• Ihsan Kahveci, University of Oxford

• McKenna Parnes, University of Washington

• Brittany Blanchard, University of Washington

• Tyler McCormick, University of Washington