GRA 6020 Multivariate Statistics Factor Analysis Ulf H. Olsson Professor of Statistics
Ulf H. Olsson Factor Analysis Exploratory Factor Analysis (EFA) One wants to explore the empirical data to discover and detect characteristic features and interesting relationships without imposing any definite model on the data Confirmatory Factor Analysis (CFA) One builds a model assumed to describe, explain, or account for the empirical data in terms of relatively few parameters. The model is based on a priori information about the data structure in form of a specified theory or hypothesis
Ulf H. Olsson The EFA model
Ulf H. Olsson The CFA model In a confirmatory factor analysis, the investigator has such a knowledge about the factorial nature of the variables that he/she is able to specify that each xi depends only on a few of the factors. If xi does not depend on faktor j, the factor loading lambdaij is zero
Ulf H. Olsson EFA Eigenvalue of factor j The total contribution of factor j to the total variance of the entire set of variables Comunality of variable i The common variance of a variable. The portion of a variable’s total variance that is accounted for by the common factors
Ulf H. Olsson EFA-How many factors to retain Based on theory Eigenvalues 1 Checking the rows in the pattern matrix * See wordfile See SPSS file
Ulf H. Olsson Factor Solutions Principal Factor Method Extracts factors such that each factor accounts for the maximum possible amount of the variance contained in the set of variables being factored No distributional assumptions Maximum Likelihood Will be treated in detail later Multivariate normality
Ulf H. Olsson Rotation of Factors The objective is To achieve a simpler factor structure To achieve a meaningful structure Orthogonal rotation Oblique Rotation
Ulf H. Olsson Rotation Varimax Major objective is to have a factor structure in which each variable loads highly on one and only one factor. Quartimax All the variables have a fairly high loading on one factor Each variable should have a high loading on one other factor and near zero loadings on the remaining factors