LIKELIHOOD ANALYSIS FOR ERRORS-IN-VARIABLES REGRESSION WITH REPLICATEMEASUREMENTS

This paper advocates likelihood analysis for regression models with me asurement errors in explanatory variables, for data problems in which the relevant distributions can be adequately modelled. Although comput ationally difficult, maximum likelihood estimates are more efficient t han those based on first and second moment assumptions, and likelihood ratio inferences can be substantially better than those based on asym ptotic normality of estimates. The EM algorithm is presented as a stra ightforward approach for likelihood analysis of normal linear regressi on with normal explanatory variables, and normal replicate measurement s.