donation elicitation methods

This study investigates how moral framing—specifically care, fairness, and loyalty—impacts fundraising effectiveness on GoFundMe. Method: Leveraging a dataset of 14,088 real-world campaigns, we integrate moral dictionary–driven NLP text analysis with multidimensional regression modeling to quantify framing effects on donation outcomes. Contribution/Results: We provide the first empirical evidence that negative moral framing (emphasizing harm or injustice) significantly increases total donations and user engagement (e.g., comments, shares) but reduces average donation size. Loyalty framing consistently and robustly boosts donor count and private message volume across all campaign categories, exhibiting broad generalizability. Emergency-related campaigns demonstrate heightened sensitivity to moral framing overall. Collectively, findings reveal a “double-edged sword” effect of moral narratives: while certain framings enhance reach and participation, they may concurrently dilute donation depth. These results offer actionable, ethics-informed strategies for optimizing online charitable communication.

Standard data visualizations can induce “statistical numbing,” diminishing viewers’ empathy and prosocial responses to humanitarian crises. This study addresses this issue by conducting a data-video experiment that empirically tests the phenomenon within a visualization context and systematically compares the effects of three narrative strategies—data-driven, person-driven, and hybrid—on emotional responses and actual donation behavior. The findings reveal that person-driven narratives significantly enhance both empathy and donation amounts, whereas hybrid narratives yield the weakest effects, challenging the common assumption that combining data with personal stories is inherently superior. These results underscore the unique efficacy of individual narratives in motivating humanitarian action.

This study addresses the persistent gap between users’ willingness and actual behavior in data donation practices, focusing on how the presentation of personal data influences donation decisions—a dimension underexplored from a design-oriented perspective. Through a real-world experiment (N=24), the research evaluates three pre-donation data exploration frameworks: “self-focused,” “social comparison,” and “collective uniqueness.” Findings reveal that the “social comparison” frame significantly increases donation rates to 87.5%, outperforming the “self-focused” condition (62.5%), whereas the “collective uniqueness” frame reduces donations to 37.5% due to induced cognitive confusion and heightened privacy concerns. This work pioneers the integration of behavioral design into public-sector data donation, uncovering a pronounced framing effect in data selection and underscoring the critical role of interface design in fostering meaningful user participation.

Traditional social norm interventions exhibit limited efficacy in heterogeneous populations due to a lack of consensus regarding desirable behavior. This study introduces, for the first time, a multi-agent system to construct “virtual social norms,” engaging human participants in online discussions about donation behavior with agents representing distinct social identities—specifically, in-group versus out-group members. Integrating multi-agent interaction, behavioral experimentation, and norm perception measurement, the research demonstrates that this approach significantly enhances individuals’ perception of prosocial norms and their willingness to donate. Moreover, agents representing in-group identities prove more effective than out-group agents in fostering norm internalization and behavioral change. These findings establish a novel paradigm for social norm interventions that leverages artificial agents to navigate social identity dynamics in diverse populations.

Social media platforms are often criticized for amplifying antisocial behaviors and lacking effective mechanisms to foster prosocial tendencies such as curiosity. This study addresses this gap by constructing an independent experimental platform and conducting a randomized controlled trial with 2,282 U.S. adults in a highly controlled environment. Using AI-driven virtual users to simulate authentic social interactions, the research systematically manipulated platform social norms and interface design. Findings demonstrate that curiosity-inducing interventions significantly increased users’ question-asking frequency and textual markers of curiosity while reducing toxic language. Although these interventions decreased generalized engagement metrics—such as likes and comments—they did not adversely affect subjective user experience or time spent creating content. The study thus provides causal evidence and a practical design pathway for promoting prosocial behavior on digital platforms.

This study investigates how artificial intelligence can guide human cooperation in repeated collective risk dilemmas while highlighting its potential misuse to induce selfish behavior. By developing a personalized persuasion framework grounded in individuals’ social value orientations and conducting large-scale online behavioral experiments, the research reveals a pronounced asymmetry in AI interventions: prosocial framing temporarily increases contribution rates and group success, yet antisocial framing exerts stronger and more enduring negative effects. This asymmetry underscores the dual-use nature and inherent risks of AI-driven behavioral influence, demonstrating that AI’s capacity to steer social behavior functions as a double-edged sword. The findings provide empirical grounding for the design of responsible human-AI collaboration systems that mitigate unintended harms while promoting cooperative outcomes.

This study investigates whether online users are susceptible to prosocial behavioral nudges from both human peers and AI agents, and whether such interventions can trigger behavioral contagion. Conducting a field experiment on Reddit, the authors systematically compare the effects of sender type (human versus bot) and reward rationale—spanning four distinct justifications—on recipients’ subsequent activity and community-wide diffusion. The findings reveal, for the first time in a real-world social setting, that transparently disclosed automated agents play a critical role in platform governance. While symbolic rewards did not enhance user engagement or influence overall, rewards issued by bots under a “raffle” rationale even suppressed participation; however, they significantly increased direct interpersonal interactions among users.

This study investigates how museums can foster prosocial behavior toward vulnerable populations—such as refugees—through thematic curation. A randomized controlled field experiment was conducted at the Santa Maria della Scala Museum in Siena, where student participants were randomly assigned to either a care- and hospitality-themed guided tour or a standard art-focused tour, followed by an opportunity to donate to a refugee-supporting NGO. Results reveal that participants who received the thematic tour donated significantly more, with the effect particularly pronounced among female participants. This work provides the first empirical evidence that museums can function as effective instruments of behavioral public policy, demonstrating that targeted cultural experiences can meaningfully promote prosocial attitudes and actions, thereby offering cultural institutions a novel pathway to engage with pressing social issues.

This study addresses the ethical dilemmas in data visualization arising from contextual constraints that prevent full disclosure of raw data, reconceptualizing data disclosure ethics as a multi-stakeholder negotiation process rather than attributing issues to individual deception or misunderstanding. To explore ethical communication mechanisms, we designed and open-sourced Purrsuasion, an educational game in which students assume roles as constrained data providers and information seekers, engaging in iterative negotiation. Integrating mixed-methods analysis, gamified platform development, heuristic rubrics, and user interaction logs, our findings reveal that learners often settle on suboptimal visual designs and struggle to accurately infer authors’ intentions when envisioning ideal visualizations. Building on these insights, we propose a heuristic scoring framework to support socio-technical judgment, offering a novel pathway for ethics education and practice in data visualization.

This study investigates the impact of large language models (LLMs) acting as coordinators on consensus formation, fairness in resource allocation, and participant perceptions in real-world group decision-making. Through two incentivized experiments involving 879 participants, we compare real-time text-based discussions among three-person groups tasked with allocating charitable funds, under conditions with and without LLM coordination and across different coordination strategies. Results indicate that while LLM coordination does not significantly enhance group consensus or actual participation equity, it shifts funding allocations toward specific organizations by up to 5.5 percentage points and introduces two key governance risks: “algorithmic steering” and an “illusion of inclusivity.” Participants erroneously perceive discussions as more inclusive and express greater trust in the process, revealing the potential for AI-mediated deliberation to subtly manipulate outcomes and distort subjective experiences.