Measuring teachers' beliefs about mathematics teaching: A Rasch analysis of a deep learning approach

Abstrak

Teachers' beliefs significantly influence the efficacy of mathematics education. The absence of an instrument to assess beliefs about deep learning poses a challenge for evaluating and enhancing mathematics teachers' competence in adopting deep learning. This research aims to develop and test the validity and reliability of an instrument to measure mathematics teachers' beliefs about implementing deep learning. This research involved 108 mathematics teachers from five provinces in Indonesia. The research data were collected using a questionnaire with a three-dimensional scale: mindful learning, meaningful learning, and joyful learning. The collected research data were then analyzed using the Rasch model, assisted by Winstep software version 3.73. The results of the first calibration indicated that the instrument's quality was very good, and the respondents' answers were quite consistent. The analysis was continued by examining each person and item more closely, resulting in the identification of three items as misfits. The second calibration was carried out by eliminating the misfit statement items. In the second calibration, a valid, reliable, and unidimensional instrument was obtained, which did not exhibit bias towards any of the research attributes. The instrument demonstrated very good reliability, and respondents' answers were quite consistent. After repeated calibration, a deep learning belief measurement instrument was obtained, consisting of 17 high-quality items with adequate psychometric properties.