Errors in Variables or Bad Leverage at Some Observations ?
MetadataShow full metadata
Errors-in-variables is a long-standing, difficult issue in linear regression; and progress depends in part on new identifying assumptions. I characterize measurement error as bad-leverage points and assume that fewer than half the sample observations are heavily contaminated, in which case a high-breakdown robust estimator may be able to isolate and downweight or discard the problematic data. In simulations of simple and multiple regression where eiv affected 25% of the data and R 2 was mediocre, one high-breakdown estimator had small bias, very good coverage, and precision that improved when the sample size increased.