Kelly, Y., Heffernan, N., Heffernan, C., Goldman, S., Pellegrino, G. & Soffer, D. (submitted). Estimating the Effect of Web-Based Homework. The Artificial Intelligence in Education Conference. Memphis. 2013. (Watch the videos and related archived material here http://web.cs.wpi.edu/~nth/PublicScienceArchive/Kelly.htm and permanently archived at http://www.webcitation.org/6E6lv54G8 )
A version of the paper has been accepted for publication, but the final is not yet submitted.
In the above referenced study we submitted for publication we wanted to see how a teacher goes over homework differently (used with permission). The control condition video below shows the teacher putting up the answers, giving students time to check their answers and then ask questions. The video is boring as it takes a long time for someone to ask a question and even then it is not clear that a good discussion was generated. You may also notice that several students didn't have their work with them, preventing them from checking their answers or asking questions. The second video shows the teacher having her item report on the screen that shows what questions were the toughest. She knows before class begins what questions she wants to go over. Listen to the video and notice how the teacher shares the common wrong answer, and starts a discussion about common misconceptions. Additional studies have shown similar findings. We tested students just before going over the homework, and then just after. Students learned more in the condition where the teacher was using the ASSISTments data.
Control: must correct answers first