MEASURING DEVELOPED ACADEMIC ABILITIES USING SPANISH‐LANGUAGE AND ENGLISH‐LANGUAGE TESTS: PAEG/GRE RELATIONSHIPS FOR PUERTO RICANS WHO ARE MORE PROFICIENT IN SPANISH THAN IN ENGLISH

Abstract

ABSTRACTThe objective of this study was to clarify the relationship between developed abilities for graduate‐level study as measured in two languages, Spanish and English, for bilingual Puerto Rican examinees more proficient in Spanish than in English. The sample consisted of 451 persons who had taken both the Graduate Record Examinations (GRE) General Test and the Prueba de Admisión Para Estudios Graduados (PAEG), a test in Spanish used for admission to graduate study and law schools in Puerto Rico. The PAEG contains sections that measure verbal, quantitative, and writing ability in Spanish. The verbal and quantitative sections contain item types that are similar to those of the GRE. The existence of the PAEG provided the rare opportunity to compare test performance in the students' native language with their test performance in a second language.In addition there were two measures of the students' English language proficiency. Besides the aforementioned verbal and quantitative tests in Spanish, the PAEG contains a subtest in English that measures reading comprehension and vocabulary. The normative sample for this test were non‐native speakers of English. For a subset of 102 individuals in this data set, the Test of English as a Foreign Language (TOEFL) was also available. The PAEG‐English and the TOEFL were used as measures of English proficiency to investigate how the relationship between the two aptitude tests was moderated by English proficiency.The analyses were patterned after the procedures followed by Alderman (1982) with undergraduate‐level English and Spanish measures. That is, several regression models were run in which a Spanish‐language measure (one PAEG ability test), an English proficiency measure (TOEFL or PAEG‐English), and a term for the interaction between the PAEG ability test score and the English‐proficiency test score were used to predict each criterion. The regression analyses had five different criteria: the GRE General Test verbal, quantitative, and analytic measures and two GRE Subject Tests, Psychology and Biology.The regression analyses revealed that each GRE General Test measure had approximately the same proportion of total variance explained by the predictors (63%‐67%; see Table 7), but the proportion of variance in test scores explained by the terms associated with measures of English proficiency varied substantially across tests. The proportion of variance explained by the English proficiency terms (both the main effect for score and the interaction) was highest for the GRE verbal test (34%), lowest for the quantitative test (8%), and intermediate for the analytical test (16%).Among the two groups with GRE Subject Test scores, the proportion of variance explained by the English proficiency terms for the Psychology and Biology tests was also intermediate between the values for the verbal and quantitative tests. The degree to which the variance in verbal scores was explained by the English‐proficiency terms was approximately double the corresponding value for the Subject Tests. Thus, students' test performance was affected by English proficiency on all tests, but these effects were largest by far on the GRE verbal test