Testing for invariance in a structural model of academic achievement across underachieving and non-underachieving students

Abstract

This work aimed to test the invariance of a causal structural model of the determinants of academic achievement in underachieving and non-underachieving students. A theoretical model of the relationships between personal, social, and familial variables and academic performance was derived empirically using data from a large sample obtained in a previous study, prior to testing for invariance across the two student groups. Underachieving students were identified using the Rasch model procedure. The sample comprised 259 underachieving and 258 non-underachieving students. The latter were selected randomly from a large non-underachieving sample of Spanish secondary education students. For model comparisons between groups, multiple-group causal structural analyses were performed, following a sequence of nested models with increasing constraints. The results showed a good fit of the model in both groups, although about half of the parameters were not invariant across groups. Underachieving students were characterized by their lack of learning strategies, an academic self-concept that exerted less influence on achievement, and a positive effect of the parent-school relationship on academic performance/achievement. Non-underachieving students were characterized by their use of metacognitive strategies, which led to higher academic achievement, a greater effect of self-concept on their achievement, the perception of parental support leading to higher performance, and the positive effects of peer acceptance on academic achievement.