Adaptive Assessment Facilitates Differentiated Learning
As we move to a model where learners have options in terms of how they learn, there is a new role for assessment in diagnosing how best to support an individual learner. This new role should not be confused with computerized adaptive testing, which has been used for years to give examinees different assessment items depending on their responses to previous items on the test in order to get more precise estimates of ability using fewer test items.
Adaptive assessment has a different goal. It is designed to identify the next kind of learning experience that will most benefit the particular learner. The School of One demonstration project used adaptive assessment to differentiate learning by combining information from inventories that students completed on how they like to learn with information on students’ actual learning gains after different types of experiences (working with a tutor, small-group instruction, learning online, learning through games). This information was used to generate individual “playlists” of customized learning activities for every student.
The online learning systems being developed through the Open Learning Initiative (OLI) at Carnegie Mellon University illustrate the advantages of integrating learning and assessment activities. The OLI R&D team set out to design learning systems incorporating the learning science principle of providing practice with feedback. In the OLI courses, feedback mechanisms are woven into a wide variety of activities. In a biology course, for example, there are
Interactive simulations of biological processes that students can manipulate; the student’s interaction with the simulation is interspersed with probes to get at his or her understanding of how it works;
“Did I Get This?” quizzes following presentation of new material so that students can check for themselves whether or not they understood, without any risk of hurting their course grade;
Short essay questions embedded throughout the course material that call on students to make connections across concepts; and
“Muddiest Point” requests that ask students what they thought was confusing.
Tutored problem solving gives students a chance to work through complex problems with the opportunity to get scaffolds and hints to help them. The students receive feedback on their solution success after doing each problem, and the system keeps track of how much assistance students needed for each problem as well as whether or not they successfully solved it.
When OLI courses are implemented in a blended instruction mode that combines online and classroom learning, the instructor can use the data that the learning system collects as students work online to identify the topics students most need help on so that they can focus upcoming classroom activities on those misconceptions and errors (Brown et al. 2006). There is a digital dashboard to give instructors an easy-to-read summary of the online learning data from students taking their course. OLI courses provide targeted feedback to students and capture data about student learning. This data, along with helpful tools for interpreting the data, are given to instructors. This information helps instructors to tailor their classroom lectures and activities to the topics with which students are struggling.
The OLI has developed learning systems for engineering statics, statistics, causal reasoning, economics, French, logic and proofs, biology, chemistry, physics, and calculus. A study contrasting the performance of students randomly assigned to the OLI statistics course with those in conventional classroom instruction found that the former led to better student learning outcomes in half the time (Lovett, Meyer, and Thille 2008).
Meshing learning and assessment in online and blended instruction, it is also possible to reduce the number of external assessments needed to audit the education system’s quality. Data streams captured by an online learning system can provide the information needed to make judgments about students’ competencies. These data-based judgments about individual students could then be aggregated to generate judgments about classes, schools, districts, and states.
U.S. Department of Education website: Assessment: Measure What Matters