Affiliation:
1. Eskişehir Osmangazi Üniversitesi Fen Fakültesi İstatistik Bölümü
2. ESOGÜ
Abstract
Factor analysis is a multivariate statistical analysis technique that has become very popular in recent years. In the factor analysis model, the error covariance matrix is assumed to be the multivariate normal distribution, and outliers are likely to be accounted for. Various estimation methods were compared with Monte Carlo simulation for the factor analysis model. The performances of the estimation methods were evaluated based on the ratio of the total variance explained and the criterion fit values. Considering the MLE, PCA, WLS, and GLS methods for classical factor analysis and the MCD, M, and S methods for robust factor analysis, the ratio of total variance explained, and fit values decreased as the sample size increased. When the number of variables increases, the ratio of total variance explained, and fit values increase at different sample sizes. It can be said that the WLS and GLS methods are better than others for classical factor analysis and the MCD and M methods are better than others for robust factor analysis.
Publisher
SDU Journal of Natural and Applied Sciences
Reference18 articles.
1. [1] Pison, G., Rousseeuw, P. J., Filzmoser, P., Croux, C. 2003. Robust Factor Analysis. Journal of Multivariate Analysis, 84(1), 145-172.
2. [2] Er, F., Sönmez, H. 2006. Öğrenci Başarı Notları İçin Robust Faktör Analizi Uygulaması. Anadolu Üniversitesi Bilim ve Teknoloji Dergisi, 7(1), 149-155.
3. [3] Browne, M. W., Shapiro, A. 1988. Robustness of normal theory methods in the analysis of linear latent variable models. British Journal of Mathematical and Statistical Psychology, 41, 193-208.
4. [4] Mooijaart, A., Bentler, P. M. 1991. Robustness of normal theory statistics in structural equation models. Statistica Nederlandica, 45, 159-171.
5. [5] Johnson, R. A., Wichern, D.W. 2007. Applied Multivariate Statistical Analysis. Fifth Edition, Pearson Education Int., New Jersey.