Simultaneous Confidence Intervals for Linear Combinations of Two Inverse Linear Regression Parameters.
Abstract
Confidence intervals for the inverse linear regression parameter can be obtained from the least-squares estimators of the regression coefficients by applying the celebrated Fieller Theorem. In the present paper we develop a method based on the simultaneous application of Fieller's Theorem to obtain simultaneous confidence intervals to any linear combination of the inverse regressions of two independent regression lines, which are not parallel. Computer subroutine functions are provided to assist in actual computations. The exact confidence limits are compared numerically to asymptotic confidence limits in order to illustrate the deficiency of the method based on asymptotic formula. An application to comparative (non-dilution) bioassays is shown too. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 21, 1978
- Accession Number
- ADA058809
Entities
People
- Shelemyahu Zacks
Organizations
- Case Western Reserve University