English-language proficiency as a moderator variable in the prediction of academic achievement of Hispanic students

The purpose of this study was to assess the validity of the Standard Progressive Matrices (SPM) and a revised version of the Mill Hill Vocabulary Scale (MHV-R) in English or Spanish in predicting the academic achievement of Hispanic middle high school students, and to determine the extent to which English language proficiency (ELP) moderated the nature of the relation. A second purpose was to study the differences between the language of the vocabulary tests in the prediction of achievement across different ELP groups. The sample consisted of 792 students for whom English was their second language. Students were placed in one of three groups, according to their ELP level. Achievement was measured by the Stanford Achievement Test (Total score, and subtest scores of Reading Comprehension, Mathematics Applications and Mathematics Computations), students\u27 grade point average, and a math course grade. Results indicated that the addition of a vocabulary measure to the SPM improved the prediction of all achievement measures, across groups, except when predicting the math grade in the most English proficient group. In groups with less ELP, the MHV-R English was a better predictor than the Spanish MHV-R. However, in the group with the highest ELP the Spanish measure was as good of a predictor as the English scale. The predictive validity of the English MHV-R did not vary as a function of the ELP level of the student, but the predictive ability of the Spanish MHV-R was somehow influenced by the degree of students\u27 ELP. Results of a factor analysis of the SPM and the MHV-R provided support for Cummins\u27 theory of language proficiency by suggesting that, the nature of the constructs being measured by these predictors changed across ELP groups. For students at lower levels of ELP, the English MHV-R was primarily measuring proficiency in English language, while still being able to predict academic achievement. For students in the highest ELP group the predictors measured the same general ability