All Times EDT
Keywords: propensity scores, inverse probability of treatment weights (IPTWs), generalized linear mixed model, confounding, causal inference
This study shows the application of inverse probability treatment weighting to estimate the causal effect of a multinomial exposure on a binary outcome in an observational data set with sparse repeated events. The application of these concepts is demonstrated using a data set of birth records to show the achieved isolated treatment effect of race-SES on infant survival.