Unraveling the factors influencing deep learning in online education: the roles of metacognitive strategies, online learning self-efficacy, creativity-fostering teacher behaviors and self-directed learning behaviors

The rapid development of online learning offers unprecedented opportunities for flexibility and personalization in higher education. However, fostering deep learning in online education remains challenging due to limited interaction and support for students. This study examined how metacognitive strategies (MCS), online learning self-efficacy (OLSE), and creativity-fostering teacher behaviors (CTB) predict students’ deep learning (DL) in online higher education, with the mediating role of self-directed learning behaviors (SDLB). Participants were 423 undergraduates from a public university in southern China, enrolled in online courses. Validated questionnaires measured MCS, OLSE, CTB, SDLB, and DL, and data were analyzed using hierarchical regression and mediation modeling with bootstrapping. Results showed that OLSE (β = 0.420, p < .001), CTB (β = 0.263, p < .001), and MCS (β = 0.177, p < .001) significantly predicted DL, explaining 47.8% of its variance. SDLB partially mediated the effects of MCS (indirect β = 0.380), OLSE (indirect β = 0.281), and CTB (indirect β = 0.190) on DL. These findings highlight the interconnected roles of MCS, OLSE, and CTB in promoting students’ deep learning in online environments, and provide practical guidance for fostering SDLB to further enhance deep learning.