Measurement Invariance and Validity of Mathematics Self-Concept Scale in TIMSS 2023: Evidence from Hong Kong

Abstract

The present study examined the factor structure, measurement invariance across sex and language, concurrent and discriminant validity and achievement-related criterion validity of the eight-item mathematics self-concept scale from TIMSS 2023 in a sample of 4463 eighth-grade students in Hong Kong. Cronbach’s alpha was high (α = 0.86) and a confirmatory factor model accounting for wording-related method variance fit the data substantially better than a strictly unidimensional model. Equivalence testing revealed that the mathematics self-concept scale achieved strict measurement invariance across both sex and language groups. Correlations provided strong concurrent/convergent evidence with liking mathematics (r = 0.59) and moderate evidence with valuing mathematics (r = 0.36), discriminant evidence through a small association with science self-concept (r = 0.20), and achievement-related criterion evidence through a positive association with mathematics achievement (B = 10.65) using a multilevel modeling approach. Findings support the mathematics self-concept scale’s practical utility for describing students’ perceived mathematical competence and for making cross-group comparisons while cautioning against overreliance on a strict unidimensional structure that ignores negative wording effects. Implications, limitations and future directions of the present study are discussed.