NotesFAQContact Us
Search Tips
ERIC Number: ED315448
Record Type: Non-Journal
Publication Date: 1990-Jan
Pages: 16
Abstractor: N/A
Reference Count: N/A
The Use of Structure Coefficients in Regression Research.
Perry, Lucille N.
It is recognized that parametric methods (e.g., t-tests, discriminant analysis, and methods based on analysis of variance) are special cases of canonical correlation analysis. In canonical correlation it has been argued that structure coefficients must be computed to correctly interpret results. It follows that structure coefficients may be useful in interpreting regression results. A small data set for a random sample of 12 of 367 participants in a study of cardiovascular risk factors is used to demonstrate structure coefficients and their use. This example and one drawn from educational research on testwiseness illustrate that failure to utilize structure coefficients can lead to serious distortions. The relationship between beta weights and structure coefficients is examined. The comparison of beta weights and structure coefficients provides important information about the predictive model. Both are important to the researcher interpreting regression results. Five data tables illustrate the examples. (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A