Factors underlying religious orientation scale - a methodological approach -
This study translated the 21-item Religious Orientation Scale (ROS) into Persian and explored its factorial validity in Iran by administering it to 329 undergraduate university students and employing three methods of factor extraction, i.e., Maximum Likelihood (ML), Principal Axis Factoring (PAF) and Principal Component Analysis (PCA). Among the three methods, ML seems to be favored in the literature recently because it forms the basis of the structural equation modeling (SEM) upon which studies such as Brewczynski and MacDonald's (2006) are developed. The ML, PAF and PCA all extracted four latent variables (LVs) when they were applied to the participants' responses and the LVs were rotated via Varimax with Kaiser Normalization. When the highest loading of a cross loading item was kept and its loadings on other LVs were removed, it was found that the three methods had the same items loading on factors three and four. The one-way ANOVA analysis of the mean of loadings and post hoc tests, however, showed that the PCA differed significantly from the ML and PAF. It was also found that the first factor extracted by the PAF is the same as the second factor of the ML and vice versa. Based on the items loading on the first two factors it is suggested that the PAF be adopted as the best method of factor extraction in both exploratory and confirmatory studies.