Optimal Designs for Contaminated Linear Regression
One notion of robustness of linear regression designs refers to moderate deviations of the regression function from the ideal linear regression setup, which may be modelled by including additive contamination functions. These may be caused for example in polynomial regression by higher order terms, which have not been included in the ideal model. The paper shows how the concepts of optimal design theory for linear regression can be extended to contaminated linear regression and points out the main problems arising in the contaminated case.
KeywordsRegression Function Linear Estimator Borel Probability Measure Approximate Theory Optimal Robust Design
Unable to display preview. Download preview PDF.
- Huber, P. J. Robustness and designs. In: Srivastava, J. N. (ed.): A Survey of Statistical Design and Linear Models. North Holland, 1975 pp. 287–301.Google Scholar
- Marcus, M. B., Sacks, J. Robust designs for regression problems. Statistical Decision Theory and Related Topics II. Academic, New York, 1977, 245–268.Google Scholar