ERIC Number: ED048354
Record Type: RIE
Publication Date: 1971-Feb
Reference Count: 0
A Modified Gauss-Jordan Procedure as an Alternative to Iterative Procedures in Multiple Regression.
Roscoe, John T.; Kittleson, Howard M.
Correlation matrices involving linear dependencies are common in educational research. In such matrices, there is no unique solution for the multiple regression coefficients. Although computer programs using iterative techniques are used to overcome this problem, these techniques possess certain disadvantages. Accordingly, a modified Gauss-Jordan technique, designed to function in the presence of linear dependencies is proposed. The modification is straightforward and incorporates a simple and efficient computer program. A comparative study, involving a series of experiments with both the modified Gauss-Jordan and iterative procedures is reported. Results indicate that the Gauss-Jordan modification performed best with simple problems and its primary advantages appear to be pedagogical. The complete computer program and examples of its output with various combinations of linear dependencies are included. (LR)
Publication Type: N/A
Education Level: N/A
Authoring Institution: N/A
Identifiers: FORTRAN Programing Language; Gauss Jordan Procedure
Note: Paper presented at the Annual Meeting of the American Educational Research Association, New York, New York, February 4-7, 1971