Accession Number : AD0639955

Title :   STEPWISE MULTIPLE REGRESSION STATISTICAL THEORY AND COMPUTER PROGRAM DESCRIPTION,

Corporate Author : BALLISTIC RESEARCH LABS ABERDEEN PROVING GROUND MD

Personal Author(s) : BREAUX,Harold J. ; Campbell,Lloyd W. ; Torrey,John C.

Report Date : JUL 1966

Pagination or Media Count : 49

Abstract : Multiple regression is a least squares statistical technique for quantitatively estimating an empirical linear relationship which may exist between variables. The classical approach often encounters acute computational difficulty, especially when many variables are involved and when little is known about the form of an appropriate functional linear model to represent the data. This can be overcome by using the technique of Stepwise Multiple Regression. The basic idea of the technique is that it allows one to overestimate the complexity of the 'candidate' linear model which is to be statistically analyzed. From this model a 'statistically significant' sub-model is determined by using certain criteria of fit and tests of significance to eliminate insignificant terms. (Author)

Descriptors :   (*COMPUTER PROGRAMMING, *LEAST SQUARES METHOD), CORRELATION TECHNIQUES, THEORY

Subject Categories : Statistics and Probability
      Computer Programming and Software

Distribution Statement : APPROVED FOR PUBLIC RELEASE