EVALUATION OF FACTOR ANALYTIC RESEARCH PROCEDURES BY MEANS OF SIMULATED CORRELATION MATRICES.
Abstract
Since the effectiveness of a method of data analysis in revealing underlying influences in observations depends on the validity of the structural model on which the data analysis is based, evaluative studies of data analysis methods are needed in which the methods are applied to bodies of data for which desired results are known. With knowledge of desired results, the validity of a data analysis method may be ascertained by a comparison of the actual results with the desired results. In the area of factor analysis, desired results from bodies of actual data are seldom known. To solve this problem, a procedure was developed for computing correlation matrices which are more similar to real data correlation matrices than are correlation matrices computed from the factor analysis structural model. The correlation matrices computed by the new procedure are termed simulated correlation matrices. They have the desirable property of being developed from known input, against which results from analyses of the matrices may be evaluated. In the present investigation, three methods of factor extraction were studied as applied to 54 simulated correlation matrices.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1967
- Accession Number
- AD0659060
Entities
People
- Ledyard R. Tucker
- Raymond F. Koopman
- Robert L. Linn
Organizations
- University of Illinois Urbana–Champaign