Poster #33 - A Monte Carlo Study of Common and Not-So-Common Criteria for Number of Dimensions in Exploratory Factor Analysis

• In Event: 77.059 - MTCC Tuesday Poster Session 18
  In Poster Session: 77.059-4 - Educational Statisticians SIG Poster Session: Statistical Issues and Investigations

Abstract

Parallel analysis (PA) is useful for determining the number of dimensions in many conditions. This present study compared two versions of PA to other common and some relatively unknown criteria. Performance of these criteria was compared for these studied conditions: sample size, number of variables, number of dimensions, and correlations among factors. The results of Monte Carlo analyses in R provided guidelines for employing PA and the other criteria to yield the best results under studied conditions. In general, PA performed most consistently and correctly across the most conditions, followed by minimum average partial, modified average root, indicator function, and then Kaiser's rule, imbedded error, and broken stick. Results showed strong potential for combinations of criteria to be useful.

Authors

• Pornchanok - Ruengvirayudh, University of La Verne

• Gordon P. Brooks, Ohio University