Participants rated frustrated and bored instructors higher on negative than positive emotion scales, with the frustrated effect stronger when the agent was on screen¶

Claim · All claims
Evidence · 1 study (3 entries) · causal r2 · q3 · i1–i3
Currency · newest study 2022 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i3 The bored instructor was rated significantly higher on negative scales than positive scales (d = 0.91). → Lawson 2022 q3 i1 The frustrated instructor was rated significantly higher on negative scales than positive scales (d = 0.31), but this effect was mediated by agent presence. → Lawson 2022 (2) q3 i? For the frustrated instructor, the valence effect was significantly stronger in the agent present condition than the agent absent condition. → Lawson 2022 (3)

Evidence¶

Lawson 2022¶

Lawson, A. P., & Mayer, R. E. (2022). The Power of Voice to Convey Emotion in Multimedia Instructional Messages. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-021-00282-y

q3 · i3 · causal · r2

The 2×2×2 ANOVA on bored-instructor ratings in the Mturk within-subjects experiment showed a main effect for positive vs. negative emotion, which the authors read as differentiating positive from negative instructors regardless of agent presence.

"participants rated the bored instructor significantly higher on negative scales (M = 3.51, SD = 0.79) than on positive scales (M = 2.13, SD = 1.25), F(1, 99) = 83.55, p < 0.001, d = 0.91"

Lawson 2022 (2)¶

Lawson, A. P., & Mayer, R. E. (2022). The Power of Voice to Convey Emotion in Multimedia Instructional Messages. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-021-00282-y

q3 · i1 · causal · r2

The 2×2×2 ANOVA on frustrated-instructor ratings in the same experiment showed a main effect for positive vs. negative emotion, consistent with the positivity hypothesis but only weakly sized as printed.

"the frustrated instructor significantly higher on negative scales (M = 3.04, SD = 0.96) than on positive scales ( M = 2.58, SD = 1.17), F(1, 99) = 9.07, p < 0.003, d = 0.31"

Lawson 2022 (3)¶

Lawson, A. P., & Mayer, R. E. (2022). The Power of Voice to Convey Emotion in Multimedia Instructional Messages. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-021-00282-y

q3 · i? · causal · r2

Follow-up ANOVA on frustrated-instructor ratings showed the valence effect was moderated by agent presence, with the effect stronger for the agent present condition; the authors state the positivity result holds "but only for the agent present condition".

"however, this main effect was mediated by a significant interaction between agent type and positive–negative valence, F(1, 99) = 15.18, p < 0.001, in which the effect was stronger for the agent present condition that the agent absent condition."

Discussion¶

Related Claims¶

• Recognizing a frustrated instructor's emotional tone required the embodied onscreen agent, contradicting the voice hypothesis for that emotion — related
• Participants detected emotional valence but did not distinguish between discrete emotions of the same valence — related
• Participants rated happy and content instructors higher on positive than negative emotion scales, supporting the positivity hypothesis — related
• Voice alone conveys the emotional tone of happy, content, and bored instructors as accurately as voice plus an embodied onscreen agent — related
• Suggesting the SPP at detected boredom did not change its use: students were almost never bored, so SPP use was not reliably different between conditions — related