The Quest for Comparability: Studying the Invariance of the Teachers' Sense of Self-Efficacy (TSES) Measure across Countries.

Ronny Scherer,Shaljan Areepattamannil,Malte Jansen,Herbert W Marsh,Trude Nilsen,Fabio Lucidi

doi:10.1371/journal.pone.0150829

Abstract

Teachers’ self-efficacy is an important motivational construct that is positively related to a variety of outcomes for both the teachers and their students. This study addresses challenges associated with the commonly used ‘Teachers’ Sense of Self-Efficacy (TSES)’ measure across countries and provides a synergism between substantive research on teachers’ self-efficacy and the novel methodological approach of exploratory structural equation modeling (ESEM). These challenges include adequately representing the conceptual overlap between the facets of self-efficacy in a measurement model (cross-loadings) and comparing means and factor structures across countries (measurement invariance). On the basis of the OECD Teaching and Learning International Survey (TALIS) 2013 data set comprising 32 countries (N = 164,687), we investigate the effects of cross-loadings in the TSES measurement model on the results of measurement invariance testing and the estimation of relations to external constructs (i.e., working experience, job satisfaction). To further test the robustness of our results, we replicate the 32-countries analyses for three selected sub-groups of countries (i.e., Nordic, East and South-East Asian, and Anglo-Saxon country clusters). For each of the TALIS 2013 participating countries, we found that the factor structure of the self-efficacy measure is better represented by ESEM than by confirmatory factor analysis (CFA) models that do not allow for cross-loadings. For both ESEM and CFA, only metric invariance could be achieved. Nevertheless, invariance levels beyond metric invariance are better achieved with ESEM within selected country clusters. Moreover, the existence of cross-loadings did not affect the relations between the dimensions of teachers’ self-efficacy and external constructs. Overall, this study shows that a conceptual overlap between the facets of self-efficacy exists and can be well-represented by ESEM. We further argue for the cross-cultural generalizability of the corresponding measurement model.

Highlights

Teachers’ self-efficacy has been in the focus of educational psychologists for many years
The present study aims to test the hypothesis of potential construct overlaps as manifested by significant cross-loadings in the Teachers’ Sense of Self-Efficacy (TSES) measurement models with respect to the factor structure and its measurement invariance across 32 countries and selected country clusters
To check whether or not the structure of the TSES measure with a perfect item-factor link was supported by the data (Research Question 1), we fitted a confirmatory factor analysis (CFA) model to the total sample in a first step and assumed three correlated traits, each representing one factor of self-efficacy

Summary

Introduction

Teachers’ self-efficacy has been in the focus of educational psychologists for many years. In order to assess these three facets, Tschannen-Moran and Woolfolk Hoy established the ‘Teachers’ Sense of Self-Efficacy (TSES)’ scale [6], which formed the basis for a number of studies that were concerned with comparing the measurement of teachers’ self-efficacy across countries and cultures [9,10,11,12]. Such crossnational comparisons, even though very interesting, rely on one key assumption: that the construct can be measured invariantly across countries, meaning that the same measurement model applies [13]. If the invariance assumption is violated, inferences on differences in teachers’ self-efficacy across countries are compromised [14]

Methods

Results

Discussion

Conclusion