Investigating the effects of strategy use and second language test performance with high- and low-ability test takers: a structural equation modelling approach

Abstract

Implicit in the research on strategy use is the notion that high- and low-ability language learners utilize strategies differently and that these differences are related to differential performance; however, to date, no study has tested these assumptions statistically. This study uses structural equation modelling (SEM) to examine the relationships between strategy use and second language test performance (SLTP) with high- and low-ability test takers. 1,382 test takers were given strategy questionnaires and a language test, and separate baseline models of strategy use and SLTP were established for each group. These models were then estimated simultaneously with cross-group equality constraints imposed. The similarities and differences in this multi-group model provided the means for evaluating the viability of the parameter estimates across ability levels. The results showed that the metacognitive strategy use (MSU) and SLTP models produced almost identical factorial structures for each group, while the cognitive strategy use (CSU) models were somewhat different. Then, while some evidence of cross-group equivalence was observed, the cross-group tests of invariance could not be uniformly supported in the data; the effect of MSU on CSU was found to be variant across the groups, as was the effect of the retrieval processes on lexico-grammatical ability, suggesting that the variation in lexico-grammatical ability and reading ability stems from the test takers’ use of the retrieval processes in concert with the metacognitive strategies.