Online Program

Understanding and Using Ordinal Factor Analysis (303696)

Keywords: Ordinal data, Factor analysis, Polychoric correlations

Factor analysis is widely used across different industries. The prevalent method for factor analysis assumes that the variables under consideration are continuous. The common practice is to use this method even in case of ordinal variables. However, it is known that the Pearson correlations are not suited for ordinal variables and tend to yield underestimated coefficients. This method of factor analysis for ordinal variables will result in biased factor loadings and unreliable factor structure which can lead to a poor model fit. It is strongly recommended to use polychoric correlation in case of ordinal variables. In this presentation, the method of ordinal factor analysis will be introduced by making use of polychoric correlations. Ordinal factor loadings and model fit will be compared with those obtained from Pearson correlations. Guidelines on the estimators, model fit indices and sample size will be provided. The data analysis will be performed in R software. A resource guide will also be provided to the audience which will contain in-depth information about the method.