Mathematics anxiety and self-efficacy among students in Hong Kong and Macao (China): socioeconomic disadvantage as a moderator of academic achievement.

PURPOSE: This study examines whether socioeconomic disadvantage alters the associations of mathematics anxiety and mathematics-specific self-efficacy with mathematics achievement among adolescents in Hong Kong (China) and Macao (China). The research problem is not whether anxiety, efficacy, and socioeconomic status are individually related to achievement; those relationships are already well established. The unresolved question is whether socioeconomic disadvantage functions as a boundary condition on the two psychological-achievement slopes when anxiety and self-efficacy are modeled jointly.

DATA AND

METHODS

This quantitative, observational, cross-sectional secondary analysis uses public PISA 2022 student data for Hong Kong (China) and Macao (China). The source data contained 10,291 relevant student records; after complete-case screening, 9,760 students from 209 schools were analyzed. Mathematics achievement was represented by PV1MATH-PV10MATH, mathematics anxiety by ANXMAT, mathematics self-efficacy by MATHEFF, and socioeconomic disadvantage by reverse weighted-standardized ESCS. Models were estimated separately across all 10 plausible values with final student weights, and sampling variance was estimated with the 80 official Fay-BRR replicate weights before plausible-value pooling.

FINDINGS: Anxiety was negatively associated with mathematics achievement (B = -9.03, SE = 1.53, p < 0.001), self-efficacy was positively associated with achievement (B = 34.82, SE = 1.62, p < 0.001), and greater socioeconomic disadvantage was associated with lower achievement (B = -14.88, SE = 1.69, p < 0.001). Neither the anxiety-by-disadvantage interaction (B = -1.77, SE = 1.52, p = 0.244) nor the self-efficacy-by-disadvantage interaction (B = -0.46, SE = 1.42, p = 0.744) was statistically significant. Equal-jurisdiction weighting, inverse-probability attrition weighting, bottom-SES-quartile models, and jurisdiction-specific models produced the same substantive moderation conclusion.

CONTRIBUTION AND PRACTICAL RELEVANCE: The findings show concurrent psychological and socioeconomic gradients while placing the moderation conclusion within the official PISA variance framework. Interventions should address emotional, motivational, and structural barriers in parallel rather than assuming that socioeconomic disadvantage necessarily magnifies the psychological gradients.