Abstract

Outliers and missing value are common problem in applied work. They can lead to inefficient of inferences if they are not properly handled. Bayesian technique had been applied to the two phenomena individually in literature. This work suggested the concept of Bayesian method to handle the problem of outliers and missing data simultaneously in regression model. The suggested Bayesian method was compared with some classical estimators through a simulation study when the regression is characterized by outlier and missing data. The criteria for assessing the performance of these estimators were mean squared error, root mean squared error, mean absolute error, and mean absolute percentage error. Also, in order to evaluate the performance of the model, Akaike and Bayesian information criteria were used. Results from the simulation revealed that Bayesian method of estimation can considerably improve estimation precision.