Sticky Help, Bounded Effects: Session-by-Session Analytics of Teacher Interventions in K-12 Classrooms

TL;DR

This work examines how K-12 teachers decide when and whom to assist in ITS-supported classrooms and whether such help yields enduring benefits. It combines qualitative interviews with a large-scale log analysis to identify decision drivers (notably prior help history and engagement states) and to test their temporal effects using GLMM, CLPM, and AFM-based learning measures. The findings show that prior help increases the likelihood of future visits (stickiness) and that idle states can dampen subsequent teacher attention, while learning gains linked to help are largely contained within the same session. These results highlight the need for session-aware, equity-focused dashboards that surface attention gaps and support deliberate, need-based teacher orchestration across time.

Abstract

Teachers' in-the-moment support is a limited resource in technology-supported classrooms, and teachers must decide whom to help and when during ongoing student work. However, less is known about how students' prior help history (whether they were helped earlier) and their engagement states (e.g., idle, struggle) shape teachers' decisions, and whether observed learning benefits associated with teacher help extend beyond the current class session. To address these questions, we first conducted interviews with nine K-12 mathematics teachers to identify candidate decision factors for teacher help. We then analyzed 1.4 million student-system interactions from 339 students across 14 classes in the MATHia intelligent tutoring system by linking teacher-logged help events with fine-grained engagement states. Mixed-effects models show that students who received help earlier were more likely to receive additional help later, even after accounting for current engagement state. Cross-lagged panel analyses further show that teacher help recurred across sessions, whereas idle behavior did not receive sustained attention over time. Finally, help coincided with immediate learning within sessions, but did not predict skill acquisition in later sessions, as estimated by additive factor modeling. These findings suggest that teacher help is "sticky" in that it recurs for previously supported students, while its measurable learning benefits in our data are largely session-bound. We discuss implications for designing real-time analytics that track attention coverage and highlight under-visited students to support a more equitable and effective allocation of teacher attention.

Figures (4)

• Figure 1: LiveLab dashboard shows real-time student progress monitoring and remediation suggestions for teachers. Teachers could manually log instances of providing help to students using a button "Mark as Helped".
• Figure 2: Cross-lagged panel model with two time points illustrating the relationship between teacher help history ($h$) and student states such as idle or struggle ($s$). Stability effects ($\beta_h, \beta_s$) are shown in blue, cross-lagged effects ($\lambda_{hs}, \lambda_{sh}$) in red, and covariances are in gray.
• Figure 3: Proportion of student sessions with teacher help by number of prior visits (0--5). Error bars show 95% exact binomial confidence intervals.
• Figure 4: Lorenz-style concentration curve of teacher help across student sessions. A stronger departure from the $45^{\circ}$ line indicates greater inequality.