Premium
Two practical issues in using LISCOMP for analysing continuous and ordered categorical variables
Author(s) -
Poon WaiYin,
Lee SikYum
Publication year - 1999
Publication title -
british journal of mathematical and statistical psychology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.157
H-Index - 51
eISSN - 2044-8317
pISSN - 0007-1102
DOI - 10.1348/000711099159062
Subject(s) - categorical variable , continuous variable , covariance matrix , sample size determination , monte carlo method , covariance , confirmatory factor analysis , computer science , structural equation modeling , sample (material) , econometrics , correlation , mathematics , statistics , chemistry , geometry , chromatography
Practitioners use a variety of models and computer programs for structural equation modelling. In particular, the approach developed by Muthen and the associated computer program LISCOMP are often used when there are continuous and ordered categorical measurement variables. Two problems are frequently encountered in practical applications: determining the sample size required to produce reliable results; and whether continuous variables should be standardized. The objective of this paper is to address these problems using Monte Carlo studies based on the confirmatory factor analysis model. We show that when using LISCOMP continuous variables should be standardized and their correlation matrix analysed, even though in theory the covariance matrix can also be used. Furthermore, dependable results may be expected for correlation structure analysis only if the sample size is large enough. These findings have important implications for the use of LISCOMP.