When Virtual Influencers Look More Human: Consumer Reactions and Conversations Around Product Endorsements
Date of Award
Spring 5-2026
Degree Name
Bachelor of Science (BS)
Advisor(s)
Marina Carnevale
Abstract
This research examines how the humanlikeness of virtual influencers in product endorsement posts shapes consumer discourse on social media. Real-world behavioral data from 1,200 Instagram comments were collected from four influencers representing different levels of humanlikeness: APOKI (@imapoki), Imma (@imma.gram), Aitana (@fit_aitana), and Mikan (@mikan.mandarin). Using LIWC-22, a text analysis program that sorts words into psychologically meaningful categories, and SPSS, the extent to which humanlikeness influenced comment length, sentiment, and emoji use was analyzed. Significant differences in the average length of consumer responses across influencer types, as well as in positive tone and emoji use, challenge the assumption of a straightforward cause-and-effect relationship between human likeness and consumer responses. Influencer types did not significantly affect the likelihood of product mentions. Together, these findings indicate that the design characteristics of virtual influencers—especially their level of humanlikeness—play an important role in shaping the volume and emotional tone of consumer discourse on social media about endorsed products. At the same time, the absence of differences in product mentions suggests that humanlikeness primarily affects the intensity and emotional tone of consumer discourse rather than whether consumers mention the product itself.
Recommended Citation
Guo, Betty, "When Virtual Influencers Look More Human: Consumer Reactions and Conversations Around Product Endorsements" (2026). Gabelli School of Business Honors Thesis Collection. 167.
https://research.library.fordham.edu/gabelli_thesis/167
