Feedback effects: Comparing the change resulting from peer and TA feedback to student solutions of Model-Eliciting Activities

Since 2002, students in a first-year engineering course at Purdue University have participated in several Model-Eliciting Activities (MEAs). MEAs are realistic, open-ended, client-driven engineering problems designed to foster students´ mathematical modeling abilities. The primary artifact produced by each team is a memo to the client describing a procedure for solving the engineering problem. Since 2007, three cycles of feedback and revision have been used in the course, with students receiving reviews from peers and teaching assistants (TAs) throughout the process. Prior to 2009, students received the first round of feedback from TAs, and the second from peers. After 2009, this order was reversed, and students received the first round of feedback from peers, and the second from TAs. This paper investigates whether the amount of change differed based on whether reviews were provided by TAs or by peers. Using the Levenshtein distance, the amount of change between consecutive drafts was calculated for four offerings of two different MEAs. Each MEA was offered with both the pre- and post-2009 sequencing, allowing a clear comparison of sequencing methods. Results from this study indicate that on the first revision, the amount of change resulting from TA reviews was indistinguishable from that resulting from peer reviews. However, on the second revision, the amount of change resulting from TA reviews was significantly different (greater) than the change resulting from peer reviews. Overall, the amount of change resulting from feedback was greater when peer reviews were provided first, and TA reviews were provided second. These results suggest that review expertise is more critical in the later stages of the process of solving an MEA.