Thomas Lumley • 3/18/2017

Case-control efficiency

This technical article explores the statistical efficiency of weighted vs. unweighted logistic regression in case-control studies. It explains why the unweighted maximum likelihood estimator is often more efficient, how the efficiency loss varies with data distribution and model parameters, and provides an interpretation using a multiple imputation framework. The content is based on simulations and is related to an upcoming academic talk.

0 comments

#Logistic Regression #Statistical Efficiency #Case Control Study