New method for measurement error covariance estimation

A new robust indirect algorithm for measurement error covariance estimation is proposed in this paper. The residual covariance matrix is estimated using Hampel's three-part redescending M-estimators. Then measurement error covariance can be obtained from the residual covariance. Unlike the conventional indirect estimations of measurement error covariance, which are very sensitive to gross errors, credible results can be achieved either with or without the presence of external causes. Implementation results show the robustness of the proposed method.