On Robust Estimation in Case{control Studies with Errors in Covariates

We discuss robust estimates based on leverage downweighting of the coe cients in logistic regression case-control studies when the true predictor X is an unobserved continuous random variable, and instead a surrogateW is observed with nondi erential measurement error. The asymptotic theory of the estimates is derived, and consistent standard error estimates obtained. An illustrative example shows that the adverse e ects of outliers upon standard methods can be somewhat ameliorated by our techniques.