NCJ Number
52909
Date Published
1976
Length
36 pages
Annotation
METHODS OF MULTIVARIATE PREDICTION AND MULTIVARIATE COVARIATION ANALYSIS ARE CLASSIFIED AND EXPLAINED IN NONMATHEMATICAL TERMS. CLASSIFICATION SCHEMES ARE GIVEN TO HELP RESEARCHERS CHOOSE THE MOST APPROPRIATE METHOD.
Abstract
THE RATIONALE BEHIND MULTIVARIATE PROCEDURES AND THE USEFULNESS OF THESE TECHNIQUES IN ORGANIZATIONAL AND INDIVIDUAL PSYCHOLOGICAL RESEARCH ARE POINTED OUT. A NUMBER OF METHODS ARE DESCRIBED IN TERMS OF THEIR RATIONALE, FACTORS WHICH MAY INDICATE THEIR USE, AND CAUTIONS WHICH SHOULD BE OBSERVED IN INTERPRETATION OF FINDINGS. MULTIVARIATE PREDICTION METHODS COVERED INCLUDE LINEAR MULTIPLE REGRESSION, STEPWISE MULTIPLE REGRESSION, CANONICAL CORRELATIONS, DISCRIMINANT AND MULTIPLE DISCRIMINANT FUNCTIONS, CURVILINEAR MULTIPLE REGRESSION, RECIPROCAL AVERAGES PREDICTION, CONFIGURAL METHODS, MULTIPLE CUTOFF METHODS, WEIGHTED APPLICATION BLANK METHODS, AND DUAL PATTERN ANALYSIS. METHODS FOR ANALYZING MATRICES OF MULTIVARIATE CORRELATIONAL DATA ARE DIVIDED INTO THOSE THAT ANALYZE TOTAL VARIANCE (COMPONENTS ANALYSIS AND CLUSTER ANALYSIS) AND METHODS FOR ANALYZING COMMON VARIANCE (FACTOR ANALYSIS). THE DISCUSSION COVERS METHODS WHICH USE A COMBINATION OF THESE TECHNIQUES, ELEMENTS WHICH INDICATE THE USE OF ONE OR ANOTHER OF THESE APPROACHES, AND INTERPRETATION OF SCORES. NONHIERARCHIAL AND HIERARCHIAL METHODS OF CLUSTER ANALYSIS ARE DESCRIBED, AND INDICATIONS FOR USE ARE GIVEN. CHARTS AND GRAPHS ILLUSTRATE MAJOR POINTS. A BIBLIOGRAPHY IS INCLUDED. (GLR)