Study: Math Test Items With Modified English Do Their Job
Modifying the English on math test items doesn't change the math knowledge assessed, concluded a study on math testing accommodations conducted by the Regional Educational Laboratory West and released this week by the Institute of Education Sciences.
The study compared the performance of three groups of 7th and 8th graders on two sets of 25 math test items, a set with modified English and a set that hadn't been altered linguistically. The three groups of students studied were English-language learners, students who had tested as proficient on English/language arts tests, and students who had not tested as proficient on English/language arts tests. The study says that the sets of test items were assigned to students within each class at random, with about half of students receiving one set and the other half receiving the other set.
The effect of the linguistic modification on students' math performance was the largest for English-language learners. The study's researchers found a smaller effect for students who weren't proficient on language-arts tests and virtually no effect for those who were proficient.
I turned to Jamal Abedi, an education professor at the University of California, Davis, and an advisor for the study, for insight into how the findings of the study might be applied to assessing English-language learners.
He said that the effect size found for English-language learners is "relatively small." The most important finding of the study, he said, is that modified English didn't interfere with the validity of the math test items.
Linguistic modification is thus "a very safe accommodation to use for ELLs and sometimes it shows effects but the effects are not huge," Abedi said.
He said his research shows that giving ELLs a customized dictionary or glossary is also a recommended testing accommodation for such students.