A Comparison of Several Methods for Determining the Number of Factors in Exploratory Factor Analysis

• Activity Number: 664
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Robert H. Pearson*+ and Dan Mundfrom and Adam Piccone
• Companies: University of Northern Colorado and New Mexico State University and DataLogix
• Keywords: Factor Analysis ; Multivariate ; Simulation ; Psychometrics

Abstract

Estimating the number of factors to retain is one of the most important steps in exploratory factor analysis. The minimum eigenvalue method remains the dominant approach in applied research. Two other approaches that have shown promise are parallel analysis and the minimum-average-partial technique. The purpose of this study was to compare the effectiveness of these three methods in correctly identifying the number of factors when applied to a wide range of factor structures. A Monte Carlo simulation was conducted in which the communality range, number of factors, and variable-to-factor-ratio were varied. The three techniques were then applied to both the full correlation matrix and a reduced correlation matrix found by replacing the diagonal elements with squared multiple correlations. Results indicate that the accuracy of the investigated methods interacts substantially with factor structure changes, but parallel analysis of the reduced correlation matrix consistently performs quite well.