Learned difficulty parameters show a moderate positive correlation with empirical KC difficulty on QATD2k¶
Claim · All claims
Evidence · 1 study · designr2·q2·i2medium
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: 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 i2 Predicted KC difficulty correlates moderately positively with ground-truth difficulty computed from empirical correctness statistics (r = 0.368, p < 0.001). → Shuyan Huang 2026
Evidence¶
Shuyan Huang 2026¶
Shuyan Huang, Alexander Scarlatos, Jaewook Lee, Andrew Lan. (2026). Interpretable Difficulty-Aware Knowledge Tracing in Tutor-Student Dialogues. arXiv preprint. https://arxiv.org/abs/2605.01097 doi:10.48550/arxiv.2605.01097
q2 · i2 · design · r2
Analysis on QATD2k test set, chosen because it is "collected from real-world educational environments"; ground-truth KC difficulty follows classical test theory as 1 − Ncorrect/Ntotal, predicted values min–max normalized, Pearson correlation computed across KCs (r = 0.368, p < 0.001).
"Results show a moderate positive correlation (r= 0.368 ,p<0.001 ), suggesting that the learned difficulty parameters align with empirical difficulty patterns across KCs to a certain extent"
Discussion¶
Related Claims¶
- Test difficulty value defined as ratio of observed to maximum score vector length times cosine of their angle equals test mean divided by number of items — related
- Point measure correlations were unchanged by the difficulty adjustment because the new person ability estimates are linear transformations of the old ones — related
- Intersection point k0 of item difficulty and discriminating curves provides a data-driven item-deletion criterion — related
- Scores on the debugging-oriented Python proficiency assessment moderately predict successful code repair — related
- Point-biserial item-total correlation is negatively related to test discriminating value and can be expressed via item and test difficulty and discrimination parameters — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
learned-difficulty-correlates-empirical-difficulty |
title |
Learned difficulty parameters show a moderate positive correlation with empirical KC difficulty on QATD2k |
description |
Learned difficulty parameters show a moderate positive correlation with empirical KC difficulty on QATD2k |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2026 |
recheck_by |
2029 |
generated |
by process:wiki-ingest on 2026-10-10 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — shuyan-huang-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.