Open AccessAN INITIAL INVESTIGATION OF A MODIFIED PROCEDURE FOR PARALLEL ANALYSIS
Author(s) -
Liu Ou Lydia,
Rijmen Frank,
Kong Nan
Publication year - 2007
Publication title -
ets research report series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.235
H-Index - 5
ISSN - 2330-8516
DOI - 10.1002/j.2333-8504.2007.tb02083.x
Subject(s) - factoring , computer science , ordinal data , missing data , software , variable (mathematics) , exploratory data analysis , popularity , statistics , algorithm , mathematics , data mining , machine learning , programming language , psychology , mathematical analysis , social psychology , finance , economics
Parallel analysis has been well documented to be an effective and accurate method for determining the number of factors to retain in exploratory factor analysis. Despite its theoretical and empirical advantages, the popularity of parallel analysis has been thwarted by its limited access in statistical software such as SPSS and SAS, especially in software that analyzes ordinal data. Among the few commonly used procedures, the Hayton, Allen, and Scarpello (2004) procedure requires manually computing the mean of eigenvalues from at least 50 replications. The O'Connor (2000) procedure overcomes that limitation, yet it has difficulties dealing with random missing data. To address these technical issues of parallel analysis for ordinal variables, we adapted and modified the O'Connor procedure to provide an alternative that best approximates the ordinal data by factoring in the frequency distributions of the variables (e.g., the number of response categories and the frequency of each response category per variable). Our procedure has a slightly different theoretical rationale from O'Connor's as well as a practical advantage in dealing with missing data.