Best Linear Unbiased and Invariant Estimators for Regression Parameters Based on Ordered Observations.
Abstract
The best linear unbiased estimators of the parameters in the multiple regression model are obtained when the samples are arbitrarily censored. The homogeneity of variance, polynomial regression and simple linear regression become special cases of the above model. The formulas take simpler forms when the underlying distribution is symmetric and the subsamples are of equal size and are symmetrically censored. Best constant-risk estimators of the parameters alpha, beta, and sigma in the simple linear regression model are obtained. These results are applied to symmetric uniform and negative exponential cases. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1974
- Accession Number
- AD0785448
Entities
People
- Z. Govindarajulu
Organizations
- University of Kentucky