Adjustment for biased sampling using NHANES derived propensity weights

The Consent-to-Contact (C2C) registry at the University of California, Irvine collects data from community participants to aid in recruitment clinical research studies. Self-selection into C2C likely leads bias due part enrollees having more years education relative US general population. Salazar et al. (Alzheimer’s Dementia Transl Res Clin Interv 6(1):e120023, 2020, https://doi.org/10.1002/trc2.12023 ) recently used examine associations race/ethnicity with participant willingness be contacted about To obtain representative estimates we use weighted estimation interest where weights are related probability self-selection convenience sample. selection probabilities estimated using National Health and Nutrition Examination Survey (NHANES). We create a combined dataset NHANES subjects evaluate trade-offs different approaches (logistic regression, covariate balancing propensity score, entropy balancing, random forest) for estimating membership NHANES. further propose methods estimate variance parameter that account uncertainty arises weights. Simulation studies explore impact weight on uncertainty. demonstrate approach by repeating analysis deduced contrast results two analyses. This method can implemented our estweight package R available GitHub.