Crowdfunding has emerged as an essential alternative financing channel in recent years. However, it is recognized that most investors in crowdfunding are laypeople with limited experience in investment and therefore have a relatively weak ability to understand complex project information. Consequently, crowdfunding project success depends heavily on adequately managing language complexity in various stages of project information revelation. Unfortunately, previous studies on the language complexity effect have no consistent conclusions, thus leaving a critical research gap to be tackled. Moreover, earning is common when humans improve their decisions based on experience in repeated tasks.

The first essay, “Time-Varying Effect of Language Complexity on Crowdfunding Success,” examines the time-varying effect of language complexity in various project stages. I propose a dual-process framework of project information comprehension to reveal the underlying mechanism. I operationalize the comprehension process in crowdfunding as the interaction effect of language complexity measured by an innovative machine learning approach and investor experience proxied by investment timing. I lay a firm foundation for this research effort by examining the impacts of multiple relevant theories, such as the elaboration likelihood model in the context of crowdfunding. I empirically examine the time-varying effect of language complexity and comprehension effect by employing a panel dataset collected from Kickstarter. My findings confirm that the impact of language complexity on project performance changes as the project progresses, leading to a new strategic approach for crowdfunding management concerning project information revelation. I also find that the impacts of textual features (the central route signals) on investors will dominate at the high level of comprehension, and non-textual heuristic factors (peripheral route signals) will dominate at the low level of comprehension; this is an expansion of the elaboration likelihood model while applying it in the context of crowdfunding. I examine why my findings are of significant theoretical and practical importance.

The second essay, “Transitional Bias in Project Owners’ Learning in Crowdfunding Description,” shows that people may not “improve” their project description writing style in the correct direction in crowdfunding. My investigation provides evidence that this phenomenon is due to human bias in perceiving the causal relations between project description writing and project success. Specifically, humans could be biased by the temporal change of the cause-effect correlation. Complementing the recent finding in psychology (Soo and Rottman 2018) that human causal learning is affected by data presentation order, the transitional bias identified in this study is toward temporal change. I show that the transitional bias effect can help explain the “incorrect” learning in crowdfunding after controlling the herding effect (both rational and irrational herding) and associative learning. I show that the transitional bias effect differs from the well-studied recency or primacy effect due to the attention to earlier or later observations. I also show that user experience can mitigate transitional bias. According to my findings, other venues, such as e-commerce platforms, should be scrutinized for transitional bias.

This thesis makes significant theoretical contributions to the literature on crowdfunding. It has several practical implications for relevant stakeholders, such as the platform, the project owners and backers, the policy maker, and the government.