Evaluating the Impact of Test Accommodations on Test Scores of LEP Students & Non-LEP Students.

Using a quasi-experimental analysis of variance (ANOVA) design, this project examined the effects of the use of accommodations with students of limited English proficiency (LEP) and non-LEP students and whether the use of accommodations affected the validity of test score interpretations. Major accommodations examined were extra time, and extra time with extended oral presentation. Samples of fourth (n=292) and seventh grade students (n=159) were tested using the Terranova multiple assessment mathematics test, as well as a math skills test and the LAS reading comprehension test. Descriptive findings showed that LEP students scored lower than non-LEP students on mathematics tests and teacher-reported skill levels. Major predictors of mathematics achievement were LAS reading proficiency level (a proxy of LEP status), whether students received an accommodation, and teacher rating of reading skill. ANOVA analyses were conducted to compare the mean scores of students in accommodated tests versus those with no accommodation. These showed the accommodation effect was significant, with those students in the extra time condition showing the highest scores. A discriminant analysis showed that the best predictors of membership in different English fluency groups were Spanish fluency, time in the United States, reading grade, and mathematics test score. LEP students were more likely than non-LEP students to be misclassified into a fluency group. Examination of students' writing samples showed clear differences in mathematics and language achievement, depending on the student's language proficiency level. The project's results provide information on whether accommodations provide valid inferences for special needs children as well as non-LEP children. (Contains 1 figure, 16 tables, and 25 references.) (SLD)