In unidimensional models, scores are assumed to be indicators of a single dimension, but it is not always possible to provide unidimensionality in the data. For this reason, the factor structure of the measurement tools can be examined with different alternatives such related factor models, bifactor and higher order factor models. The purpose of the current study is to investigatethe factor structure of the job satisfaction scale developed by Hackman and Oldham (1980) and adapted into Turkish by Gödelek (1988) and Güler (1990). The adapted version of the scale was not based on factor analysis results. And it was seen that this scale is using usually based on old results. The original and adapted form of the scale require responding to incomplete items. In this study incomplate items were turned into questions sentences. To examine and obtain additional information for revision of the factor structure of the scale it was administered to 318 teachers in the city of Burdur in Turkey. The scale is a 5-point Likert type consisting of 14 items. In the current study the factor structure of the scale wastested with one-factor, two-factor and three-factor CFA (Confirmatory Factor Analysis) and two-factor and three-factor Bifactor CFA (B-CFA), two-factor, and three-factor ESEM (Exploratory Structural Equation Modelling) and Bifactor ESEM (B-ESEM) models. The results of the analysis showed that among all the models, the fit indices obtained with the three-factor B-ESEM remained at an acceptable level. Model data fit could not be achieved in one and two factor models. In addition, some items in the scale were found to exhibit cross-loading. In order to obtain more valid and reliable scores, it has been suggested to use the scale unidimensional.

Bifactor model, cross loading, exploratory structural equation modelling, job satisfaction, confirmatory factor analysis.