Mobile Device-Assisted Peer Review in High School Poetry Analysis: The Role of Revision Intensity
Mobile Device-Assisted Peer Review in High School Poetry Analysis: The Role of Revision Intensity

Summary/Abstract: Background/purpose. This study investigates how peer feedback translates into actual improvements in high school poetry analysis. It aims to connect feedback processes, feedback network structures, and final product quality by integrating process mining, social network analysis (SNA), and outcome modeling. Materials/methods. Thirty-four students (N = 34) from a high school in Semarang participated in two 90-minute sessions following the sequence: briefing – peer assessment – revision – presentation. Process mining (Optimal Matching and Partitioning Around Medoids) was used to identify workflow archetypes, while SNA examined how feedback was distributed across students. Feedback quantity and quality were linked to final task scores using ANCOVA and linear mixed models, and path analysis was used to test whether revision behavior mediated the effect of feedback on product quality. Results. Process mining revealed three workflow archetypes: Presentation-Leap, Linear-Fast, and Iterative-Revision. SNA indicated a sparse but moderately modular feedback network (density = 0.070; modularity = 0.448; Gini = 0.437; reciprocity = 7.7%). ANCOVA/LMM results showed that revision intensity was the strongest predictor of final quality, followed by weighted incoming feedback, whereas betweenness centrality contributed less. Path analysis confirmed that the effect of feedback on final quality was partially mediated by revision behavior. Conclusions. Meaningful revisions serve as a key mechanism linking peer feedback to improved performance in poetry analysis. Teachers are advised to incorporate mandatory revision checkpoints and to distribute feedback opportunities more equitably across students. The study is limited by its small sample size, single-school context, and nonrandom design; future research should employ multi-site or randomized designs and explore long-term learning outcomes to strengthen generalizability.

• Details
• Contents