Tuesday, July 9, 2019

Measuring the Impact of Test Disengagement

 
Academic assessments are used to measure educational attainment, assess proficiency, evaluate schools and programs, for certification/licensure, and to inform other important decisions. For test scores to validly indicate what students know and can do, students must give good effort on the assessment. Measurement practitioners have long known that test takers are not always engaged, however, and that disengagement can threaten score validity and negatively bias test scoresi,ii. Disengagement has been seen both in high-stakes timed tests, where some test takers rapidly-guess as time runs out, and in untimed, low-stakes tests, where low test-taker motivation is a more likely cause.

Before computer-based tests (CBTs) were introduced, inferences about test taker engagement had to be made at the test event level, most frequently by asking a test taker immediately after testing to report their own level of engagement on the test. With CBTs, new insights are available: item response time permits an item-by-item assessment of engagement through the identification of rapid-guessing behavior. Using data from MAP® Growthassessments, an adaptive assessment system for K-12 students, this research illustrates the nature of rapid-guessing behavior, explores how it differs from solution behavior, provides a model of what happens when test takers disengage, and addresses how disengagement should be managed during scoring.

Research has shown that test takers rarely exhibit rapid-guessing behavior throughout a test, but rather may move multiple times between solution behavior and rapid guessing. Rapid guessing is affected by characteristics of the item, the test taker, and the context in which the item is administered.

No comments: