Proceedings of the SIGCHI Conference on Human Factors in Computing Systems
CLARISSE: A Machine Learning Tool to Initialize Student Models
ITS '02 Proceedings of the 6th International Conference on Intelligent Tutoring Systems
A scalable solution for adaptive problem sequencing and its evaluation
AH'06 Proceedings of the 4th international conference on Adaptive Hypermedia and Adaptive Web-Based Systems
Error-Flagging support for testing and its effect on adaptation
ITS'10 Proceedings of the 10th international conference on Intelligent Tutoring Systems - Volume Part I
Limiting the number of revisions while providing error-flagging support during tests
ITS'12 Proceedings of the 11th international conference on Intelligent Tutoring Systems
Hi-index | 0.00 |
Previously, providing error-flagging support during tests was reported to lead to higher scores. A follow-up controlled study was conducted to examine why, using partial crossover design. Two adaptive tutors were used in fall 2009 and spring 2010, and the data collected during their pre-test stage was analyzed. The findings are: (1) When a student solves a problem correctly on the first attempt, error-flagging support helps the student move on to the next problem more quickly without pausing to reconsider the answer. But, it may also encourage students to use error-flagging as an expedient substitute for their own judgment; (2) Given error-flagging support, many more students will arrive at the correct answer by revising their answer, which explains why students score higher with error-flagging; (3) Students will use error-flagging to reach the correct answer through trial and error even though the problems are not of multiple-choice nature. However, at least some students may engage in informed (as opposed to brute-force) trial and error. (4) Error-flagging support provided during tests could cost students time. (5) Given how often students move on after solving a problem incorrectly, without ever reconsidering their answer, providing error-flagging support during testing is still desirable.