A collective generative AI relevance intervention significantly improved high school students' relevance as identification with math immediately after the intervention compared with a control group¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r1 · q3 · i2 medium
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q3 i2 The collective generative AI condition (E2) showed significantly higher post-intervention relevance as identification with math than the control group, with a medium effect size (η2 = 0.073). → Guo 2026 q3 i? The omnibus MANCOVA effect of the intervention on the two post-test outcomes was significant. → Guo 2026 (2)

Evidence¶

Guo 2026¶

Guo, Z., Fryer, L. K., & Shum, A. (2026). Utilizing generative AI to promote high school students' personal relevance to math and interest in the math class: An intervention. doi:10.31234/osf.io/srxga_v1

q3 · i2 · causal · r1

Quasi-experimental MANCOVA with 218 Chinese high school students in three intact classes, controlling prior relevance, interest, self-efficacy, math grade and gender. The planned comparison showed E2 "reported higher relevance as identification with math than" C, η2 = 0.073, a medium effect.

"the students in E2 (adj. Mean = 2.08, SE = 0.10) reported higher relevance as identification with math than that reported by the students in C (adj. Mean = 1.57, SE = 0.10) (F = 11.11, p = 0.001, η2 = 0.073)"

Guo 2026 (2)¶

Guo, Z., Fryer, L. K., & Shum, A. (2026). Utilizing generative AI to promote high school students' personal relevance to math and interest in the math class: An intervention. doi:10.31234/osf.io/srxga_v1

q3 · i? · causal · r1

Omnibus MANCOVA on post relevance as identification and post math class interest, controlling five covariates; the article reports the overall intervention effect as "significant (Wilk's λ =0 .94, F = 3.13, p = 0.015 < 0.05)". No effect size was printed for the omnibus test. The omnibus test shows only that the conditions differed overall; the collective-AI-versus-control contrast itself is the planned comparison in the entry above.

"the overall effect of utilizing generative AI on the two dependent variables was significant (Wilk's λ =0 .94, F = 3.13, p = 0.015 < 0.05)"

Discussion¶

Learner Variables¶

• Domain Context — outcome: instruction changes it

Related Claims¶

• Learners invest more effort in tasks they perceive as interesting, personally important, or useful to their future goals. — the general claim this finding bears on, with the studies behind it

• An IES practice-guide presentation recommends personalized problem contexts, asserting without cited studies that students engage more actively and perform better in them — the general claim this finding bears on, with the studies behind it

• The generative AI relevance intervention did not significantly increase students' interest in the math class — related
• The individualized generative AI condition showed no significant effect on relevance as identification with math compared with the control group or the collective condition — related
• The collective generative AI intervention's effect on relevance as identification with math persisted one week later through its immediate effect — related
• Interviews indicate individualized AI-generated identity-based relevance was perceived as overly abstract, while collective AI dialogue connected math to current concerns and cross-domain applications — related