What can interaction data tell us about learning?

Look beyond the finished text

Two completed texts may conceal very different writing journeys. One learner may progress with little support; another may revise repeatedly or use several levels of hints. Stage-level records make those differences available for investigation rather than leaving researchers with the final product alone.

Ask specific questions of the signals

PrepTalk records structured signals such as attempts, validation patterns, hint levels, timings and completion. These can help researchers locate recurring difficulties and examine how progression varies across tasks and stages.

• Which stages produce repeated attempts or recurring validation patterns?
• What patterns of progression follow different levels of support?
• How do those patterns vary across activity versions and proficiency levels?

Keep interpretation separate from observation

Progressing after a hint does not establish that the hint caused the improvement. Time spent on a stage is not a direct measure of cognitive effort, and completing an activity is not the same as demonstrating a lasting learning gain.

Research needs to connect those observations to explicit questions and an appropriate study design. Comparisons must also account for differences in the tasks and the learners. The value of the data is that it supports more precise investigation—not that it removes the need for it.

Study the process without retaining everything

A useful research record does not require an archive of every rejected draft or hint. PrepTalk’s pilot distinguishes approved structured metrics from transient interaction content. Consent and defined retention boundaries are part of the research design, not details to add after collecting data.