Machine Learning Analysis of Emotional Appeals in Charity Crowdfunding

Charity crowdfunding relies on empathy-driven support but often struggles to convert emotional appeals into actual donations. Drawing on the elaboration likelihood model, this study investigates how emotional appeals embedded in textual descriptions and facial images influence fundraising performance in charity-based crowdfunding. Using data from a leading Chinese crowdfunding platform, we apply machine learning techniques to analyze text and image content of charity crowdfunding campaigns. Our results show that emotions conveyed through textual descriptions, which tend to engage deeper cognitive processing, have a stronger persuasive impact than those expressed in facial images. For instance, anticipation in text is negatively associated with the amount raised, while sadness in facial expressions is positively associated with donation outcomes. To explain these findings, we introduce the concept of affective modality alignment, positing that emotional appeals are more effective when matched with the cognitive processing route typically activated by their delivery modality. Our research contributes to the theoretical understanding of dual-route persuasion in digital philanthropy and offers practical guidance for designing emotionally resonant charity crowdfunding campaigns.