A Misunderstanding and Misuse of Parallel Analysis

• In Event: Roundtable Session 18
  In Roundtable Session: Exploratory Approaches to Latent Variable Model Building and Theory Development

Abstract

Horn’s Parallel analysis has emerged as one of the best tests of dimensionality for use with common-factor analysis. However, due to a fundamental misunderstanding of the underlying theoretical rationale, researchers performing common-factor analysis will often inappropriately and unnecessarily modify the parallel analysis procedures in an attempt to correct for a perceived theoretical inconsistency with the common-factor model. This paper seeks to rectify this basic misunderstanding by a careful examination of the original rationale behind parallel analysis and its predecessor, the Kaiser-Guttman rule. In addition to a theoretical argument, an extensive Monte Carlo simulation using the common-factor model will be conducted to provide empirical evidence.

Authors

• Brent Dale Hill, North Dakota State University

• Chris M. Ray, North Dakota State University