The Trends in Mathematics and Science Study (TIMSS) and Progress in Reading and Literacy Study (PIRLS) provide important data on learners’ academic self-perceptions and motivational beliefs, which are constructs known to influence students’ educational outcomes. However, a critical methodological challenge emerges when comparing these measured constructs across socioeconomic groups. Observed score differences in the constructs could reflect genuine differences in underlying self-perceptions and motivation, or simply that groups interpret survey questions differently because items hold different meanings for students from different backgrounds. The study uses multigroup confirmatory factor analysis to investigate measurement invariance of self- concept, intrinsic motivation, extrinsic motivation, and affective engagement across socioeconomic groups using TIMSS 2019 (Grades 5 and 9 mathematics) and PIRLS 2021 (Grade 4 reading) South African data, comparing learners from the poorest 60% of schools (Q1-3) with those from Q4 and Q5 schools. Results reveal substantial measurement non- invariance, most pronounced between Q1-3 and Q5 groups. Failures at metric and scalar levels particularly affect mathematics self-concept and extrinsic motivation. Analysis of the PIRLS data reveals that language of instruction also moderates measurement equivalence. These findings highlight the need for caution when interpreting group differences in learner-reported self-perception and motivational constructs, especially in
highly unequal societies such as South Africa.
