Task type strongly determines trust in educational robots, with highest trust for procedural instruction, moderate trust for teaching, and lowest trust when robots request personal information¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · q3 · i3 large
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i3 In a repeated-measures ANOVA, TASK had a significant large main effect on trust ratings (η² = .439). → Anna-Maria Velentza 2026 q3 i? Post hoc comparisons showed instruction tasks produced significantly higher trust than teaching and personal-information tasks, and teaching higher than personal information. → Anna-Maria Velentza 2026 (2)

Evidence¶

Anna-Maria Velentza 2026¶

Anna-Maria Velentza, Konstantina Nikou, Anne-Gwenn Bosser, Nikolaos Fachantidis. (2026). What Robots Do Matters More Than What They Look Like: Task Context Shapes Trust in Educational HRI. https://www.google.com/search?q=%22What+Robots+Do+Matters+More+Than+What+They+Look+Like%22 doi:10.1109/ro-man72286.2026.11714558

q3 · i3 · causal · r2

Two-way repeated-measures ANOVA with ROBOT and TASK as within-subject factors in the N = 81 video-based experiment. The article reports a significant TASK main effect, "F (1.88, 148.18) = 61.88, p < .001, η 2 = .439".

"Results revealed a significant main effect of TASK, F (1.88, 148.18) = 61.88, p < .001, η 2 = .439, indicating that trust levels differed significantly across activities."

Anna-Maria Velentza 2026 (2)¶

Anna-Maria Velentza, Konstantina Nikou, Anne-Gwenn Bosser, Nikolaos Fachantidis. (2026). What Robots Do Matters More Than What They Look Like: Task Context Shapes Trust in Educational HRI. https://www.google.com/search?q=%22What+Robots+Do+Matters+More+Than+What+They+Look+Like%22

q3 · i? · causal · r2

Bonferroni-adjusted pairwise comparisons in the same repeated-measures ANOVA. Estimated marginal means were highest for Instructions (T2, M = 3.43), followed by Teaching (T1, M = 3.14), and lowest for Personal Information (T3, M = 2.88); only test statistics with p < .001 are printed, no effect size.

"Post hoc comparisons for TASK (see Fig. 6) showed that T2 (Instructions) produced significantly higher trust scores than both T1 (Teaching) and T3 (Personal Information), all p < .001."

Discussion¶

Related Claims¶

• Robot appearance showed no meaningful effect on trust: pairwise robot differences did not reach significance after Bonferroni adjustment — related
• The interaction between robot appearance and task on trust was only marginal, suggesting appearance may slightly modulate expectations without determining trust — related
• Hypotheses that specific robot appearances would be trusted more in matching tasks (humanoids for teaching, task-oriented robots for instructions, soft robots for personal disclosure) were not confirmed — related
• Trust evaluations were strongly and positively correlated across all robot–task combinations, suggesting trust in social robots may be a stable, generalized attitude — reports the opposite