A SYSTEM OF IBM TYPE 709 PROGRAMS FOR OBTAINING PRINCIPAL AXES FACTORS OF LARGE ORDER CORRELATION MATRICES
Abstract
IN A FACTOR ANALYSIS SYSTEM WHERE THE PRINCIPAL AXES PROGRAM OPERATES DIRECTLY ON THE CORRELATION MATRIX, THE NUMBER OF VARIABLES WHICH CAN BE FACTORED IS LIMITED IN PRACTICE BY THE SIZE OF SYMMETRIC MATRIX WHICH CAN BE CONTAINED IN CORE. With a 32K core, this amounts to a maximum of around 240 variables. The number of observations is not restricted. The system reported here restricts the number of observations, but allows for a relatively large number of variables. Instead of computing a correlation matrix, which is the minor product moment of the normalized data matrix, this system computes the major product moment of the normalized data matrix. The principal axes program operates on the latter matrix, which is of the order of the number of observations, and produces factor scores rather than factor loadings. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1961
- Accession Number
- AD0269318
Entities
People
- George Burket
Organizations
- University of Washington