The Effects of Nonlinearity in Regression Models. Part 1. A Procedure for Examination of Adequacy of the Linear Approximation.
Abstract
The difficulties inherent in assessing the amount of nonlinearity in a nonlinear estimation problem are first discussed. Then, a procedure is suggested for checking the nonlinearity compared with the null case when the model is linear. The nonlinearity will depend not only on the nonlinear model but also on the design points and data set available, as well as on the confidence level desired for parameter confidence statements. Both a visual test and an approximate numerical test are discussed and an example is given. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1974
- Accession Number
- AD0779373
Entities
People
- Douglas E. Shaw
- Norman Richard Draper
Organizations
- University of Wisconsin–Madison