Research Pilot -- Spring 2026 -- Appleton Area School District
349 Appleton students and 13 teachers were surveyed before and after using Byline (students) and Revision History (teachers) for 8 weeks
Key takeaways
Teachers
Increased trust in student writing authenticity
Students
Decreased concern about being falsely accused of using AI
Students
Increased engagement with writing
Alongside an amazing team in Appleton Area School District, a public school district in Wisconsin, and Dr. Joshua Wilson, Associate Professor in the School of Education at the University of Delaware, we ran a research pilot with 349 students and 13 teachers, gauging teacher and student perception changes before and after implementing Revision History with teachers and Byline with students.
The results were very encouraging when comparing pre-pilot to post-pilot perception changes. Teachers reported a large and statistically significant increase in trust that students were authentically doing their work. Students reported a smaller but statistically significant decrease in concern about being falsely accused of using AI, along with small but statistically significant gains across various measures of engagement with writing.
Why we did this research
It has never been more difficult for teachers to assess the authenticity of student writing -- and, in turn, to gauge a student's level of effort and learning. To help keep students honest and better understand their students' work journey, 200,000+ teachers now use Revision History to make their students' writing process visible.
We built Byline for two reasons:
Provide teachers with even more insight into students' writing journey, in addition to what happens inside a Google Doc or Slide (student website visits, Chrome extension activity, AI use)
Provide teachers with the ability to put teacher-enabled/disabled, guardrailed, transparent AI in front of students to incentivize and teach responsible AI use
But before rolling out Byline, we wanted to conduct research to ensure we were helping boost trust in classrooms -- on both the teacher and student sides.
Finding 1
Teachers: increased trust in student writing authenticity
The teacher sample is small (13 matched responses), so read these as exploratory, but all three shifts below survived statistical correction. Scores are means on a 1-5 scale.
Finding 2
Students: decreased concern about being falsely accused of using AI -- and the more they used Byline, the more they gained
Across 349 matched responses, student worry about being falsely accused of using AI showed a small but statistically significant decline, concentrated among students who actually used Byline. Bars show baseline-adjusted gains in scale points by degree of Byline usage.
Students were not randomly assigned to usage levels, so these are associations with usage intensity, not causal effects.
Finding 3
Students: increased engagement with writing
Across the same 349 matched responses, all changes below are statistically significant.
How the study worked
Students and teachers completed surveys before and after the Spring 2026 pilot, answering on a 1-5 agreement scale. Responses were matched pre-to-post, and each outcome was tested with paired-samples t-tests, with appropriate controls applied to make sure any overall changes were statistically meaningful rather than just random chance.1
The analysis was conducted by Dr. Joshua Wilson, Associate Professor in the School of Education at the University of Delaware, whose research focuses on how automated writing evaluation can improve the teaching and learning of writing.
This research was funded by Tabwrite. Dr. Wilson designed and conducted the analysis.
1 To address Type I error across multiple tests, we applied the Benjamini-Hochberg procedure to control the false discovery rate.
What this study can and can't tell you
We think research pages should be honest about limitations, so here are ours:
- This was a single-district, single-semester pilot without random assignment -- it shows real pre/post change, not proof of causation.
- The teacher sample (13 matched responses) is small; teacher results should be read as exploratory.
- All outcomes are self-reported survey responses, not writing-quality measures.
- Student effect sizes are modest in absolute terms (about 0.2-0.3 points on a 5-point scale), though consistent across a large matched sample.
Full technical reports, including item-level statistics and methodology, are available to districts on request.
Try Byline free for a year
District-wide Revision History subscriptions include a free one-year trial of Byline. Reach out and we'll get your teachers and students set up.
Contact us at info@revisionhistory.com