In multiple linear regression, the ordinary least squares estimator is very sensitive to the presence of multicollinearity and outliers in the response variable. To handle these problems in the data, Winsorized shrinkage estimators are proposed and the performance of these estimators is evaluated through mean square error sense.
Jadhav, Nileshkumar H. and Kashid, D N.
"Robust Winsorized Shrinkage Estimators for Linear Regression Model,"
Journal of Modern Applied Statistical Methods: Vol. 13
, Article 7.
Available at: http://digitalcommons.wayne.edu/jmasm/vol13/iss2/7