Dimensionality of instructional quality in physical education. Obtaining students’ perceptions using bifactor exploratory structural equation modeling and multilevel confirmatory factor analysis

Abstract

BackgroundIn research on instructional quality, the generic model of the three basic dimensions is an established framework, which postulates that the three dimensions of classroom management, student support and cognitive activation represent quality characteristics of instruction that can be generalized across subjects. However, there are hardly any studies that examine if the three basic dimensions model could represent a suitable approach to measure instructional quality in physical education. Based on an extended model of the basic dimensions, a measurement model of instructional quality for physical education is presented, which integrates different theoretical approaches from the fields of educational and psychological research as well as different subfields of sports science in order to test the factorial structure of the corresponding measurement model.Methods1,047 students from 72 seventh to ninth grade classes from different German-speaking Swiss cantons participated in the study. The conceptualization of the instrument is based on a hybrid approach that integrates generic and subject-specific characteristics. The simultaneous analysis at the individual and class level using MCFA was supplemented by more complex methodological techniques within the relatively new B-ESEM framework at the individual level.ResultsThe postulated five-factor structure was initially tested using ICM-CFA and showed a good model fit (e.g., χ2/df = 2.32, RMSEA = 0.03, CFI = 0.97, TLI = 0.97, SRMR = 0.04). MCFA revealed a differential factorial structure at both levels of analysis with five factors at the individual level and four factors at the class level (e.g., χ2/df = 2.23, RMSEA = 0.03, CFI = 0.96, TLI = 0.96, SRMR within = 0.04, SRMR between = 0.10). ESEM and B-ESEM outperformed the ICM-CFA and showed an excellent model fit (B-ESEM: χ2/df = 1.19, RMSEA = 0.01, CFI = 1.00, TLI = 1.00, SRMR = 0.01). Inter-factor correlations and factor loadings are largely in line with expectations, indicating arguments for construct validity.DiscussionThe study represents a substantial contribution in linking physical education and the generic research on instructional quality. Overall, strong arguments for the factorial structure of the measurement model were demonstrated. The study can be interpreted as a first step in a multi-step procedure in terms of further validity arguments.