Evaluating the validity of accommodations for English learners through evidence based on response processes.

摘要

English learners (ELs) represent one of the fastest growing student populations in the United States. Given that language can serve as a barrier in EL performance, test accommodations are provided to help level the playing field and allow ELs to better demonstrate their true performance level. Test accommodations on the computer offer the ability to collect new types of data difficult to obtain via paper-and-pencil tests. Specifically, these data can be used as additional sources of validity evidence when examining test accommodations. To date, limited research has examined computer-based accommodations, thus limiting these additional sources of validity evidence. The purpose of this study was to evaluate the validity of computer-based test accommodations on high school History and Math assessments using evidence based on response processes, specifically accommodation use and response time. Two direct linguistic accommodations, non-ELs, two EL groups, and five research questions were investigated in this study.;Accommodation use results indicated significant differences in use across the three student groups, with ELs using accommodations more frequently than non-ELs. However, there were still high percentages of all three groups not accessing any accommodations on individual items. Accommodation use was more common on History than on Math, and decreased as the assessment progressed. Results suggest future research focus on students actually using the accommodations when conducting research on the effectiveness of accommodations.;Response time results showed ELs taking longer to process test items as compared to non-ELs regardless of receiving test accommodations. Receiving accommodations significantly impacted processing time for some of the items on History, but not on Math. Similarly, History showed a relationship between the number of accommodations on test items and response time, but Math did not. These results suggested that the Math content knowledge may have played a larger role in response time than the accommodations. Positive relationships between test performance and response time were found in both subject areas. The most common predictors of both accommodation use and response time across both subject areas were sex, Hispanic status, and socioeconomic status. Implications of the results and suggestions for future research are discussed.