Automated gaming-the-system detectors built with XGBoost reliably recognize gaming in both problem-solving and self-explanation steps¶
Claim · All claims
Evidence · 1 study · associationalr2·q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person · recheck by 2028: tool and model accuracy changes quickly
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? Tenfold student-level cross-validation showed the detectors reliably predicted gaming in unseen students in the problem-solving (AUC = 0.89, k = 0.50) and self-explanation (AUC = 0.99, k = 0.95) steps. → Baker 2025
Evidence¶
Baker 2025¶
Baker, R. S., Richey, J. E., Zhang, J., Karumbaiah, S., Andres-Bray, J. M., Nguyen, H. A., Andres, J. M. A. L., & McLaren, B. M. (2025). Gaming the system mediates the relationship between gender and learning outcomes in a digital learning game. Instructional Science, 53, 203–238. https://doi.org/10.1007/s11251-024-09679-3
q2 · i? · associational · r2
Detectors were built by hand-labeling log data via text replay coding (800 problem-solving and 1500 self-explanation clips; inter-rater k = 0.74 and 0.88) and training an Extreme Gradient Boosting (XGBoost) classifier for each step type.
"Based on the cross-validation results, we determined that the models could reliably predict gaming in unseen students in both the problem-solving (AUC = 0.89, k = 0.50) and self-explanation (AUC = 0.99, k = 0.95) steps."
Discussion¶
Related Claims¶
- Machine-learned detectors of science inquiry skill made accurate inferences after a student ran a simulation three times — related
- Gaming the system is significantly negatively correlated with test performance in the digital learning game — related
- Automated detectors of disengaged behaviors inferred from interaction-log data alone reached AUC over 0.8 for gaming and correlated 0.55 with observed off-task behavior — a broader claim this one bears on
- No significant gender difference in gaming the system appears in the non-game control condition or in the problem-solving step — related
- Generalizable detectors of students' affective states in DLPs remain an unsolved problem — a broader claim this one bears on
- Male students game the system significantly more than female students in the self-explanation step of a digital learning game, across three studies — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
xgboost-gaming-detectors-decimal-point |
title |
Automated gaming-the-system detectors built with XGBoost reliably recognize gaming in both problem-solving and self-explanation steps |
description |
Automated gaming-the-system detectors built with XGBoost reliably recognize gaming in both problem-solving and self-explanation steps |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2025 |
recheck_by |
2028 |
generated |
by process:wiki-ingest on 2026-10-18 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — baker-2025 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.