ABSTRACT
In simple linear regression where the response variable has been subject to right-censoring, it is unwise to place the same statistical credence to all points in the scatterplot of response against covariate. Many methods that are based on traditional ideas of linear regression have been designed specifically to account for the effect of various censoring patterns. In the Buckley-James method for simple linear regression, to compensate for right-censoring, censored points in the scatterplot of observed data are moved vertically to create a renovated scatterplot where the bias of censoring is removed. The points are moved to estimated positions which are unbiased provided that there are large expected numbers of censored and uncensored observations across the support of the data. The method is simply based on an iterative solution to the usual least squares normal equations which have been modified to take account of the censoring.
