Purpose: This study aims to investigate the effectiveness of hybrid learning, blended learning and face-to-face learning in remediating misconceptions among primary school teacher education students. Design/methodology/approach: This research uses a quasi-experimental design, using a quantitative approach. Data were collected from 99 students using a pretest and posttest four-tier test. Findings: Hybrid learning proved to be the most effective, achieving a remarkable 90.32% success rate in remediating students' misconceptions, surpassing blended learning and face-to-face learning methods. Research limitations/implications: The data come from a single Indonesian university and focuses only on a science concept; therefore, the scope of findings may be limited. Practical implications: The research suggests that applying the conceptual change (CC) model in all learning types, particularly hybrid learning, effectively remediates misconceptions. Educators can use this insight to design impactful teaching strategies that combine online and traditional components, accommodating diverse learning styles and needs. Social implications: This research suggests that applying CC model in all learning types is actually able to remediate misconceptions, though hybrid learning is found to be the most effective one. Originality/value: This study addresses a research gap by investigating the effectiveness of different learning modes in rectifying misconceptions. Although prior studies have explored learning modes, few have directly compared hybrid, blended and face-to-face learning in correcting misconceptions. The findings offer insights for effective teaching strategies to address STEM-related misconceptions, benefiting educators in optimizing their approaches. Furthermore, the study's implications extend to the broader academic community, contributing to evidence-based teaching practices in science education and the development of effective strategies for addressing misconceptions in STEM courses.