Multiple Linear Regression.
Abstract
Multiple linear regression theory provides an estimated covariance structure for the estimates of the parameters of the linear function based on given data. However, when the deviation form is used to calculate these parameter estimates, the portions of this matrix which involve the constant term are generally missing. This report presents equations which can be used to calculate these missing covariances from quantities which are generally available. Most standard regression references discuss calculation of confidence limits for point estimates of the dependent variable when these point estimates are calculated from the regression equation. This report presents equations for similar limits for the independent variables, again from quantities generally available when a deviation-form routine is used. A different interpretation is suggested for these limits than is seen in the references. A numerical example is provided. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1978
- Accession Number
- ADA057958
Entities
People
- George R. Humfeld