Predicting teachers’ sense of efficacy: A multimodal analysis integrating SEM, deep learning, and ANN

Abstract

This study aims to investigate the predictive role of cultural intelligence, motivation to teach, and “culturally responsive classroom management self‐efficacy” (CRCMSE) in teachers’ sense of efficacy. The study utilized a combination of “structural equation modeling” (SEM), deep learning, and “artificial neural network” (ANN) to analyze data collected from 1061 preservice teachers. The SEM analysis indicated that cultural intelligence, motivation to teach, and CRCMSE significantly predicted the sense of efficacy of the teacher candidates, accounting for 59% of the variance. Additionally, the ANN model accurately predicted the teachers’ sense of efficacy with 75.71% and 75.17% accuracy for training and testing, respectively. The sensitivity analysis revealed that CRCMSE played the most crucial role in predicting the preservice teachers’ sense of efficacy. The deep learning model also predicted the sense of efficacy with an overall accuracy of 74.18%. The utilization of a multimodal analysis approach facilitated the identification of both linear and nonlinear relationships between the constructs.