Assessing Integrated Writing Tasks for Academic Purposes: Promises and Perils

Abstract

The five studies presented in this special issue offer unique evidence, analyses, and theoretical rationales for assessment tasks that involve writing in reference to information from source material with substantial content. I review the five studies in respect to five “promises” and five “perils,” concluding that, collectively, the promises were mostly fulfilled, but so were most of the perils. The promises are that these task types (a) provide realistic, challenging literacy activities; (b) engage test takers in writing that is responsible to specific content; (c) countertest method or practice effects associated with conventional item types; (d) evaluate language abilities consistent with construction-integration or multiliteracies models of literacy; and (e) offer diagnostic value for instruction or self-assessment. The perils of these task types, however, are that they (a) confound the measurement of writing abilities with abilities to comprehend source materials; (b) muddle assessment and diagnostic information together; (c) involve genres that are ill-defined and so difficult to score; (d) require threshold levels of abilities for competent performance, producing test results that may not compare neatly across different ability levels; and (e) elicit texts in which the language from source materials is hard to distinguish from examinees' own language production.