Pedagogical agents significantly improve learners' self-efficacy expectations (g = 0.29), with effects moderated by grade level, facial expressions, and reliability reporting¶

Claim · All claims
Evidence · 1 study (3 entries) · quant-synthesis r2 · q3 · i1–i2
Currency · newest study 2025 · checked against its sources 2026-10-19 (automated, 3 of 3 entries) · not reviewed by a person

Subclaims¶

q3 i1 Across 57 comparisons from 29 studies, PAs had a small significant positive effect on self-efficacy expectations (g = 0.29, p < 0.001). → Gladstone 2025 q3 i2 PAs using facial expressions had a moderate significant effect on self-efficacy expectations (g = 0.46, p < 0.01), while those without had no significant impact. → Gladstone 2025 (2) q3 i2 The strongest self-efficacy effect was for students in grades 7-9 (g = 0.77, p < 0.001), though from only two comparisons in one study; adult learners outside postsecondary education showed g = 0.44 (p < 0.01). → Gladstone 2025 (3)

Evidence¶

Gladstone 2025¶

Gladstone, J. R., Schroeder, N. L., Heidig, S., Zhang, S., Palaguachi, C., & Pitera, M. (2025). Do Pedagogical Agents Enhance Student Motivation? Unraveling the Evidence Through Meta-Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-025-10050-2

q3 · i1 · quant-synthesis · r?

Three-level meta-analysis with CHERVE of PA-versus-non-PA comparisons measuring self-efficacy expectations. Pooled effect was "g = 0.29, p < 0.001" with significant within-study heterogeneity (Q(56) = 475.51).

"A random effects 3LMA with CHERVE found that the overall impact of PAs on learners' self-efficacy expectations was g = 0.29, p < 0.001 (kcomparisons = 57, kstudies = 29)."

Gladstone 2025 (2)¶

Gladstone, J. R., Schroeder, N. L., Heidig, S., Zhang, S., Palaguachi, C., & Pitera, M. (2025). Do Pedagogical Agents Enhance Student Motivation? Unraveling the Evidence Through Meta-Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-025-10050-2

q3 · i2 · quant-synthesis · r?

Moderator analysis within the self-efficacy 3LMA, based on 12 studies with 24 comparisons, found facial expressions significantly moderated the effect (F(2.00, 9.41) = 6.46, p = 0.02).

"PAs that used facial expressions had a moderate impact on learners' self-efficacy expectations (g = 0.46, p < 0.01), while those without facial expressions had no significant impact compared to non-PA systems."

Gladstone 2025 (3)¶

Gladstone, J. R., Schroeder, N. L., Heidig, S., Zhang, S., Palaguachi, C., & Pitera, M. (2025). Do Pedagogical Agents Enhance Student Motivation? Unraveling the Evidence Through Meta-Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-025-10050-2

q3 · i2 · quant-synthesis · r2

Grade level significantly moderated self-efficacy effects (F(4.00, 6.83) = 11.42, p < 0.01). Primary, upper secondary, and post-secondary learners showed no significant benefits or detriments.

"the strongest effect was found for students in grades 7–9 (g = 0.77, p < 0.001); however, there were only two comparisons extracted from one study (Plant et al., 2009). We also found a moderate, significant impact on adult learners outside of post-secondary education (g = 0.44, p < 0.01)."

Discussion¶

Learner Variables¶

• Self-Efficacy — outcome: instruction changes it

Related Claims¶

• Pedagogical agents have a small, statistically significant positive effect on learners' broadly defined 'motivation' (g = 0.20) — related
• Pedagogical agents significantly improve learners' interest (g = 0.30), with the effect depending on the type of control condition — related
• Pedagogical agents do not significantly affect learners' engagement (g = -0.02, non-significant) — related
• Pedagogical agents show a small-to-moderate but non-significant effect on learners' utility beliefs (g = 0.34, p = 0.05) — related
• Pedagogical agents do not significantly affect learners' value beliefs (g = 0.09, non-significant) — related
• Effects of agent nonverbal behavior depend on the task: facial expressions were valuable for attitudinal learning but detrimental for procedural learning — related
• Human-LLM agreement remains low across constructs (κ = 0.000 to 0.419), with self-efficacy showing the best alignment and Prior KSAs none — related
• Pedagogical agents show essentially no effect on learners' intrinsic motivation (g = 0.00) — related
• Most agent characteristics, including nonverbal communication, motion, and voice, did not moderate the instructional effectiveness of pedagogical agents — reports the opposite
• Pedagogical agents enhance learning compared with learning environments that do not feature agents — related
• Technology effects on TEFL outcomes trend upward across educational levels (primary g = 0.29, secondary g = 0.35, tertiary g = 0.44), though between-level differences are not statistically significant — related
• Learning with multimedia pedagogical agents is more effective than learning without these agents, with a small overall effect (g+ = 0.20) — related