Understanding civic cognitive assessment tasks: Associations between linguistic features and students’ task performance

This study investigates how text features and item characteristics are associated with early adolescents’ performance in large-scale assessments in civic education (CIVED). Examining responses to multiple-choice items from the IEA CIVED assessment, the study identified linguistic features related to item difficulty. These features account for more than one-third of the variance in difficulty level of 38 test items for a national sample of 3000 US ninth-grade students. On this test in CIVED short words, words with concrete meanings, intentional verbs and certain noun phrases were likely to facilitate students’ comprehension and ability to correctly answer an item. Contrariwise, lengthy words, words with abstract meanings and complex syntactic structure hindered student performance. Some linguistic features significantly predict the performance of students from homes with high literacy resources, but do not reach significance for less advantaged students. Kintsch’s situation model helps to interpret these results.