Accession Number : AD0766911

Title :   A Technique for Predictive Modeling of Multivariable and Multivariate Data.

Descriptive Note : Technical rept.,

Corporate Author : WASHINGTON UNIV SEATTLE DEPT OF PSYCHOLOGY

Personal Author(s) : Gillo,Martin W. ; Shelly,Maynard W.

Report Date : APR 1973

Pagination or Media Count : 29

Abstract : MAID-M, a survey technique operative in FORTRAN 4, is presented. It performs predictive modeling for a multivariate criterion from a basically unlimited set of predictor variables. Based on additive multivariate measures of association, it identifies the smallest combination of predictor variables which accounts for a maximal proportion of the variation space of a given set of criterion variables. MAID-M allows the user to interact with the program for the purpose of exploring different alternatives to solutions recommended by the program. Estimates of stability as well as estimates of validity for specific empirical solutions are provided optionally. An example with real data illustrates the new technique. (Author)

Descriptors :   (*MULTIVARIATE ANALYSIS, ALGORITHMS), (*PSYCHOLOGICAL TESTS, MULTIVARIATE ANALYSIS), COMPUTER PROGRAMMING, SAMPLING, MATHEMATICAL PREDICTION

Subject Categories : Psychology
      Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE