On Randomizing Estimators in Linear Regression Models
S. Ermakov, Rainer Schwabe · Birkhäuser Boston eBooks · 2000
In this work we consider a special kind of randomization in the analysis of linear regression. This randomization is connected to the Δ 2 -distribution which was first introduced by Ermakov and Zolotukhin ( 1960 ) for decreasing the variance in the Monte Carlo calculation of integrals. The resulting resampling procedure allows for separating the systematic and the random components of the variance. Further, some problems are discussed in designing regression experiments.