Powers’ research focuses on health disparities, with a specific focus on the Hispanic infant mortality paradox and race/ethnic comparisons of change in infant mortality over time. Most of his substantive work is intertwined with his methodological interests in survival modeling, regression decomposition, age-period-cohort models, and other topics. In addition to his focus on the Hispanic paradox, Powers investigates race (black-white) differences in the sources of change in infant mortality using an age-period-cohort perspective. Powers applies state of the art statistical computing approaches to substantive problems, and is affiliate of the Population Research Center and faculty associate in the Department of Statistics and Data Sciences.
Assesses women's preferences for contraception after delivery and compares use with preferences.
Builds on findings from recent research showing an erosion of infant survival advantage in the Mexican-origin population relative to non-Hispanic whites at older maternal ages, with patterns that differ by nativity. Quantifies the degree to which differences in the distribution and effects of risk factors contribute to the infant mortality gap at older maternal ages across the three populations of interest.
Shows that a Bayesian ridge regression model with a common prior for the shrinkage parameter yields estimated age, period, and cohort effects similar to those based on the intrinsic estimator. Allowing informative and non-informative priors on the shrinkage parameters associated with each temporal component produces results similar to mixed-effects APC models and the intrinsic estimator depending on the strength of the prior information on the shrinkage parameter for the cohort effects.
Reexamines the epidemiological paradox of lower overall infant mortality rates in the Mexican-origin population relative to U.S.-born non-Hispanic whites. A comparison of infant mortality rates among U.S.-born non-Hispanic white and Mexican-origin mothers by maternal age reveals an infant survival advantage at younger maternal ages when compared with non-Hispanic whites accompanied by higher infant mortality at older ages for Mexican-origin women.
Presents the methods that form the core of contemporary categorical data analysis; both the transformational and latent variable approaches and so synthesizes similar methods in statistical and economic literatures.