Assessing the construct validity of the conception of teaching and learning questionnaire (CTLQ) for Chinese university students in Hong Kong: Going beyond the use of goodness of fit indices

This paper argued that the assessment of construct validity of the Conception of Teaching and Learning Questionnaire (CTLQ) does not need to stick to the use of goodness of fit indices only. Additional statistical measurements can help to determine construct validity. By systematically investigating the convergent and discriminant validity of the CTLQ, it was found that the initial 5-factor model does not have satisfactory discriminant validity, although its fitness indices are very good. Further examination showed that discriminant validity could be improved by combining the two factors related to constructivist teaching and learning conception, with a slight reduction in the goodness of fit of the revised measurement model. As the fit indices of the revised model are still good, the revised model is the best possible model with both construct validity and good model fit. This optimum model would not have been found if construct validity had not been explicitly examined. Results indicated that the factorial structure of the traditional conception is more complex than the unidimensional constructivist conception. This imbalance in structural complexity may reflect the different degree of exposure to the two approaches of teaching and learning. It may also highlight the need to incorporate more Chinese-specific elements of student-centred teaching in the CTLQ.