论文信息 - A comparison between maximum likelihood and generalized least squares in a heteroscedastic linear model

A comparison between maximum likelihood and generalized least squares in a heteroscedastic linear model

Abstract We consider a linear model with normally distributed but heteroscedastic errors. When the error variances are functionally related to the regression parameter, one can use either maximum likelihood or generalized least squares to estimate the regression parameter. We show that likelihood is more sensitive to small misspecifications in the functional relationship between the error variances and the regression parameter.

D. Ruppert | R. Carroll

[1] R. Welsch,et al. Efficient Bounded-Influence Regression Estimation , 1982 .

[2] David Ruppert,et al. Robust Estimation in Heteroscedastic Linear Models. , 1982 .

[3] T. Hammerstrom. Asymptotically Optimal Tests for Heteroscedasticity in the General Linear Model , 1981 .

[4] David Ruppert,et al. On Robust Tests for Heteroscedasticity , 1981 .

[5] Wayne A. Fuller,et al. Least Squares Estimation When the Covariance Matrix and Parameter Vector are Functionally Related , 1980 .

[6] Raymond J. Carroll,et al. On Estimating Variances of Robust Estimators When the Errors are Asymmetric , 1979 .

[7] M-Estimates for the Heteroscedastic Linear Model. , 1979 .

[8] W. Fuller,et al. Estimation for a Linear Regression Model with Unknown Diagonal Covariance Matrix , 1978 .

[9] D. V. Gokhale,et al. A Survey of Statistical Design and Linear Models. , 1976 .

[10] P. Bickel,et al. Using Residuals Robustly I: Tests for Heteroscedasticity, Nonlinearity , 1975 .

[11] George E. P. Box,et al. Correcting Inhomogeneity of Variance with Power Transformation Weighting , 1974 .

[12] P. Sen,et al. Theory of rank tests , 1969 .