A1451
Title: Multilevel regression and poststratification
Authors: Yajuan Si - University of Michigan (United States) [presenting]
Abstract: Multilevel regression and poststratification (MRP) is a popular method for addressing selection bias in subgroup estimation, with broad applications across fields from social sciences to public health. The inferential validity of MRP is examined in finite populations, exploring the impact of poststratification and model specification. To enhance the fitting performance of the outcome model, modeling the inclusion probabilities conditionally on auxiliary variables and incorporating flexible functions of estimated inclusion probabilities as predictors in the mean structure is recommended. A statistical data integration framework is presented that offers robust inferences for probability and nonprobability surveys, addressing various challenges in practical applications. Simulation studies indicate the statistical validity of MRP, which involves a tradeoff between bias and variance, with greater benefits for subgroup estimates with small sample sizes, compared to alternative methods. The methods are applied to the Adolescent Brain Cognitive Development (ABCD) study, which collected information on children across 21 geographic locations in the U.S. to provide national representation, but is subject to selection bias as a nonprobability sample.
