CSCL Long Paper
#652: An Item Response Theory Model for Multiple Resubmissions
Wed Jun 17, 10:00 AM–11:30 AM · ALP 2100
Part of Peer-feedback and Assessment
Assessment, Feedback & Formative Practices Learning Analytics & Educational Data Mining Computational Thinking & CS Education
Item response theory models (IRT) are commonly used to model students' responses to test items. Under IRT, the associations among item responses are modeled by a latent variable. However, most IRT models are not suitable for modeling multiple resubmissions which are common in a learning environment. Students may be allowed to revise and resubmit multiple times after each submission. In this paper, we develop an IRT model for multiple resubmissions. The utility of this model is demonstrated through analyzing a real data set from a recently developed novel authentic assessment of programming skills.
Speakers
- Xiang Liu — Educational Testing Service
Authors
Xiang Liu, Mo Zhang, Chen Li, Hongwen Guo, Amy Ko, Min Li