Development of website-based deep learning media for backward roll gymnastics instruction

Abstract

This study aimed to develop and evaluate website-based gymnastics learning media integrated with a deep learning approach to improve backward rolling skills among Physical Education students at Universitas Negeri Padang. This research employed a Research and Development (R&D) design using the ADDIE model, consisting of Analysis, Design, Development, Implementation, and Evaluation stages. The product was validated by material and media experts, followed by small-scale and large-scale trials. Effectiveness testing involved an experimental group (n = 30) and a control group (n = 27) using a pretest–posttest control group design. Material and media expert validation scores reached 93.34% and 95.00%, respectively. Practicality scores were 86.81% in the small-scale trial and 85.90% in the large-scale trial. Significant differences were found between the experimental and control groups for backward roll angle (t = 4.721, p < 0.001) and backward roll round performance (t = 3.586, p = 0.001). The developed website-based learning media was valid, practical, and effective in improving students’ backward rolling skills and can serve as an innovative instructional tool in gymnastics learning.