Adaptive Learning Project Uses Video Games in the Classroom
talks about using computer games to measure students for adaptive instruction purpose.
The idea is actually very straightforward. Through interaction with video games, students are asked to make a series of choices related to sciences. The researcher then used the captured data to train the Neural Network. Once enough data have been collected, the students' behavior patterns are essentially captured. It is at this stage that instructors can test different instructional models to "students" (simulated here). Without even implementing the instructional model in classroom settings, instructors are now able to gain knowledge on how students will react to different instructional models. Thus, they can choose the model that has the highest probability of success.
The effectiveness of this will heavily depend on the quality of the test items (multiple choice questions embedded in video games). If psychometric property of the items is good enough, we can very accurately measure/model students.
We are facing one big problem in measuring learning outcomes in informal learning settings. We can't directly measure students! Currently most people are using various kinds of data to predict learning outcomes, because that data can be easily gathered. Throw in enough data, and we can build a very tempting statistical model; however, the validity of such a measure would be terrible. The research mentioned in this article, embedding testing items in informal settings, could be a possible solution to this problem. This is something we can think about for mSchool and Vialogues.
The researcher Richard Lamb
is an assistant professor in the educational measurement and quantitative methods program at Washington State University.