Synthetic cohorts place week-to-week change in the term's examination weeks, which real cohorts do not¶

Claim · All claims
Evidence · 1 study (2 entries) · design r2 · q2
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¶

q2 i? In every cohort the synthetic mean study time is higher in that year's own examination week and among the two highest of the eighteen, while in the real cohorts the mid-term week ranks ninth to fifteenth. → Inoue 2026 q2 i? The Spearman rank correlation between real and synthetic weekly β0 curves runs from -0.22 to 0.57. → Inoue 2026 (2)

Evidence¶

Inoue 2026¶

Inoue, H., & Yasutake, K. (2026). What Fidelity Metrics Miss: A Structural Check on Synthetic Educational Data. Proceedings of the 34th International Conference on Computers in Education. Asia-Pacific Society for Computers in Education. https://arxiv.org/abs/2609.27265 doi:10.48550/arxiv.2609.27265

q2 · i? · design · r2

Comparison of weekly mean study time placement across the four cohort pairs. The synthetic cohorts track a year-specific detail of the generator's documentation; "The real cohorts are not" so placed, and in two cohorts fall below the mean of the rest.

"The mid-term examination falls in week 22 in 2022 and in week 23 in the other three years, and in every cohort the synthetic mean study time is higher in that year's own examination week, and among the two highest of the eighteen. The real cohorts are not: there the mid-term week ranks ninth to fifteenth"

Inoue 2026 (2)¶

Inoue, H., & Yasutake, K. (2026). What Fidelity Metrics Miss: A Structural Check on Synthetic Educational Data. Proceedings of the 34th International Conference on Computers in Education. Asia-Pacific Society for Computers in Education. https://arxiv.org/abs/2609.27265. doi:10.48550/arxiv.2609.27265

q2 · i? · design · r2

Rank correlation between real and synthetic weekly β0 curves across the four cohorts, ranging from -0.22 to 0.57, indicating the synthetic curves do not reproduce where in the term the real variation falls.

"The synthetic cohorts therefore miss not only the size of the week-to-week movement but also its placement, and the Spearman rank correlation between the real and synthetic weekly β0 curves runs from -0.22 to 0.57."

Discussion¶

Related Claims¶

• Real cohorts separate from marginal-preserving surrogates of themselves in all four years, while three of four synthetic cohorts do not — related
• The variation gap traces to what the generator was disclosed: no released statistic constrains which band and which week are zero together — related
• Synthetic cohorts reproduce the level and shape of the weekly β0 partition but vary 2.6–4.9 times less across the term than real cohorts at a common working point — related
• In a control experiment on data of known construction, the check is calibrated, can be passed, and separability from marginals is not fidelity — related
• Partition descriptors and the component-membership–study-volume association are reproduced by the synthetic cohorts — related