A More Accurate Growth Model: Using Multigrade Adaptive Assessments to Measure Student Growth
November 14, 2007
Steering Committee of the Delaware Statewide Academic Growth Assessment Pilot
Among the many proposed changes to NCLB, nearly everyone agrees that "adequate yearly progress" should henceforth be measured using some kind of growth model--i.e., student and school performance should abe gauged by academic improvement instead of how well they do on a test at a given point in time. But, as always, the devil's in the details. This short study out of Delaware shows that, in order to get accurate and useful data on student growth, states must implement tests that are more dynamic than the current crop of assessments. The problem, in short, is this: "no one would argue that a student who begins 4th grade with 1st grade skills and ends with 3rd grade skills hasn't made excellent progress, but current 4th grade tests cannot document this progress." Computerized "multigrade adaptive growth assessments," on the other hand, draw on a large database of questions across grade levels, and adapt each student's test to his or her level of aptitude. The study compared results from Delaware's current growth-model pilot program, which uses grade-level exams, to results from an adaptive growth assessment (the MAP test). The same 11,000 students took both tests. The adaptive assessment was much more sensitive to their progress. One caveat, which the authors acknowledge, is that the two tests are imperfectly aligned. Still, they argue, the study convincingly shows that the "progress of students who are well above or below grade level is effectively invisible under the current testing system." That's one more thing for Congress to chew on when feasting on NCLB reauthorization. See the study here.