Abstract

This study investigates how multimodal emotional expressions influence user behavior on digital charity platforms. Drawing on framing theory and cognitive load theory, we propose the concept of multimodal emotional consistency to capture the alignment between textual and visual emotional expressions. Using a dataset of 23,859 online charity crowdfunding projects, we employ deep learning-based emotion classification to extract textual emotions and facial emotion recognition to identify visual emotions, resulting in 4,474 valid samples. We construct key variables including emotional consistency, textual emotional intensity, and visual emotional intensity, and examine their effects on fundraising performance using ordinary least squares (OLS) regression. The results show that emotional consistency has a significant positive effect on donation outcomes, suggesting that aligned emotional signals across modalities enhance information persuasiveness. Textual emotional intensity exhibits a significant negative effect, indicating that overly emotional language may reduce user trust. In contrast, visual emotional intensity has a positive effect, highlighting the advantage of visual cues in eliciting user empathy and engagement. Furthermore, we find a nonlinear relationship between emotional consistency and fundraising performance, revealing diminishing marginal effects. This study contributes to the literature on digital platform user behavior by uncovering how multimodal emotional design shapes donation decisions, and provides practical implications for optimizing content strategies in online charity crowdfunding.

Download

Share

COinS