Dynamic reciprocal association of adolescents' math self-efficacy with math achievement and parental homework support: A cross-lagged panel model analysis

OBJECTIVES: This study aimed to examine the bidirectional relationship of self-efficacy with social support and academic achievement in a mathematics learning context. METHODS: Three-wave longitudinal data were collected from first-year junior high school students during a period of home-based online learning. At each data collection point, students reported their math self-efficacy, parents reported their support for students' math homework, and the school provided recent math exam scores. RESULTS: Analysis of a cross-lagged panel model with family socioeconomic status and students' age at T1 as covariates found that students' prior math self-efficacy predicted parental subsequent homework support, but not vice versa. Moreover, math self-efficacy exhibited alterable reciprocal association with math achievement. Specifically, new-domain knowledge learning led to a stronger prediction of math achievement on math self-efficacy, while old-domain knowledge learning led to the dominant prediction of math self-efficacy on math achievement. Mediation analyses revealed that prior math achievement indirectly predicted future parental homework support through current math self-efficacy, whereas no significant mediation emerged from prior homework support to future math achievement via self-efficacy. CONCLUSIONS: These findings reveal a unidirectional relationship between self-efficacy and social support in family math learning environments, and demonstrate bidirectional association between math self-efficacy and academic performance with a dominance transition tied to learning content, suggesting the need for deeper investigation into these dynamic mechanisms.