From pixels to action: Virtual influencers promoting pro-environmental behavior through anthropomorphism, credibility, and strategic messaging

In the digital era, virtual influencers (VIs) have shown unique potential in public issue communication on social media, yet their comprehensive persuasive strategies require further investigation. Coupled with the unrealism of VIs, how to design effective strategies for them and enhance their credibility and attractiveness through analyzing the mechanisms of information evaluation have become focal goals of our study. Based on the Computers as Social Actors (CASA) Theory, Elaboration Likelihood Model (ELM), and Source Credibility Theory, we have employed an online mixed experiment to explore the role of anthropomorphism level, source type, and strategic messaging in the effectiveness of VIs’ pro-environmental persuasion. The findings reveal that human-like VIs are more persuasive, and VIs labeled as non-experts outperform those labeled as experts. Notably, non-expert VIs’ effectiveness is most pronounced under high anthropomorphism, reflecting a significant anthropomorphism × source-type interaction. Additionally, hope appeal can be more effective than fear appeal in our context. Both cognitive trust and emotional appeal play mediating roles in the persuasion process. However, cognitive trust mediates the relation between anthropomorphic images, source types, and persuasion efficiency. Strategic messages influenced the persuasive outcomes via emotional appeal. The results provide empirical evidence for optimizing VIs in environmental and health communication and updating the evaluation framework of VI green persuasion.