A Latent Dirichlet Allocation approach to understanding students' perceptions of Automated Writing Evaluation
Wilson, J., Zhang, S., Palermo, C., Cruz Cordero, T., Zhang, F., Myers, M. C., Potter, A., Eacker, H., & Coles, J. (2024). A latent Dirichlet allocation approach to understanding students' perceptions of automated writing evaluation. Computers and Education Open, 6, Article 100194. https://doi.org/10.1016/j.caeo.2024.100194
This article examines U.S. middle school students' perceptions of the MI Write automated writing evaluation system using survey responses and open-ended comments. The authors apply latent Dirichlet allocation and follow-up qualitative analysis to identify common themes in students' feedback and compare topic prevalence by perceived usefulness of the tool.
Key takeaways
- Four latent topics emerged from students' comments: desire for more in-depth feedback, improved user experience, greater personalization, and increased fairness in automated scoring.
- Students who viewed MI Write as less useful were more likely to emphasize the need for deeper feedback.
- Students who found MI Write useful were more likely to highlight its value as a learning tool while still wanting greater personalization.
- The study suggests latent Dirichlet allocation can help uncover patterns in student perceptions of automated writing evaluation systems from open-ended responses.
Why it matters
Provides an empirical analysis of middle school students' perceptions of automated writing evaluation and demonstrates the use of latent Dirichlet allocation alongside qualitative interpretation to inform AWE design and implementation.
Coauthors
Joshua Wilson, Saimou Zhang, Corey Palermo, Tania Cruz Cordero, Fan Zhang, Matthew C. Myers, Andrew Potter, Halley Eacker, Jessica Coles
Topic tags
External links