Fusing Specialized Surveys of Rare Populations to Larger Surveys for Generalized Inference: Cross-Sectional Survey Study.
Journal of medical Internet research April 27, 2026 Karilynn M Rockhill, Elizabeth A Bemis, Nicole Schow et al. 2 citations
Combining a large representative survey with a smaller survey focused on psychedelic drugs can produce generalizable estimates of rare behaviors like drug use without adding burdensome questions to the big survey. Researchers used calibration weighting to transport estimates from a psychedelic-enriched survey (two waves, total over 4,300 adults) to a representative anchor survey (two waves, total over 57,000 adults). The method showed good internal consistency, with transport biases under 0.4 percentage points for demographics, health, and substance use. External validity improved for health and substance use estimates after fusion. Using the fused data, recreational use of psilocybin (92.9%), LSD (93.2%), and MDMA (93.3%) was far more common than medical use (30.9%, 26.4%, and 21.1%, respectively). This approach expands surveillance epidemiology for rare behaviors.