Theses and Dissertations

Date of Award

8-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Information Systems

First Advisor

Francis Kofi Andoh-Baidoo

Second Advisor

Jun Sun

Third Advisor

Geng Sun

Abstract

Crowdfunding has become a transformative mechanism for financing innovation, connecting creators directly with backers while bypassing traditional funding barriers. Despite its rapid global growth, only about 33.7% of campaigns succeed, largely due to ineffective communication of value propositions and lack of backer engagement. This dissertation investigates the complex interplay between project founders, backers, and emerging AI technologies to understand the factors influencing donation intention and to enhance crowdfunding success prediction.

Adopting the behavioral decision-making framework, I draw on multiple theoretical perspectives- signaling theory, the dual-process model of persuasion, and the cognitive-affective trust framework to examine founder- backer interactions. Using a multi-method approach, I integrate decision-tree modeling, experimental surveys, and deep learning to explore these dynamics. Using a global dataset of 2,108 campaigns, the research uncovers how founder experience, persuasive comments, and rich media increase backer support. It further examines how generative AI particularly domain-specific tools can enhance trust by producing project content perceived as reliable, transparent, and informative. Experimental results show that cognitive trust in AI-generated content leads to emotional engagement and increased donation likelihood, though overly positive tech reputation can paradoxically weaken this trust link. To better capture backer engagement during campaigns, a novel deep learning framework (DTV-AMI) is introduced, which leverages attention mechanisms across timeliness, semantic diversity, voting patterns, and multimodal interactions in comments. Using an online dataset of 1,317 campaigns with 285,094 comments, the DTV-AMI model demonstrates that recent, popular comments attract backers’ attention and improve predictions of project success. The model significantly improves success prediction accuracy and offers practical implications for creators and platforms aiming to optimize engagement and build trust in the evolving crowdfunding landscape.

Comments

Copyright 2025 Maliha Alam. All Rights Reserved.

https://proquest.com/docview/3246803822

Download

Share

COinS