Two essays on the roles of rhetorical devices in online platforms

Date

2018-08

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

What factors affect the structure of funding of crowdfunded projects? Essay 1 examines the effects of rhetorical devices (i.e. emotional tone, linguistic style match, perceived clout, and narrative language style) found in project descriptions on the funding formation on a rewards-based crowdfunding platform (Kickstarter). Using functional data analysis and longitudinal data, the effects of these rhetorical devices are examined along with two dynamic variables (backers support and competitive intensity) across each project’s funding cycle. The results provide evidence of how and when rhetorical devices influence funding formation. For example, emotional tone (the ratio of positive to negative words in a project description) positively affects funding formation only at the tail-end of the funding cycle. In addition, the results suggest that the effectiveness of the rhetorical devices can be influenced by the two dynamic variables. For example, competitive intensity (the number of live, similar projects on the platform) reduces the effectiveness of narrative language style (the use of dynamic storytelling rather than formal and precise descriptions in a project description) across the entire funding cycle.

Essay 2 focuses on predicting the helpfulness of reviews using star ratings and hidden rhetorical devices. This essay investigates three main research questions: (1) Is the helpfulness of reviews affected by extremely positive (i.e. five-star) or negative (i.e. one-star) ratings? (2) Do rhetorical devices affect review helpfulness? (3) Do rhetorical devices moderate the effect of overall ratings on review helpfulness? Focusing on 207,720 reviews posted on Amazon.com over 18 years, Essay 2 examines the above questions using the negative binomial regression. The results show that extremely positive/negative reviews (i.e. one-star/five-star ratings) are perceived as more helpful by potential consumers than reviews with moderate ratings (i.e. three stars). The results provide evidence on how review content such as positive tone, negative tone, and cognitive language style affect review helpfulness. As regards emotional language, both a positive and a negative tone decrease the reviews’ helpfulness, as neither provide potential customers with adequate product-relevant information. However, cognitive reviews are more helpful, as they are more informative than non-cognitive or emotional reviews. Findings show that if a review with an extreme rating contains emotional language, its positive effect is diminished. In contrast, if a review with an extreme rating contains cognitive language, the positive effect of that review is strengthened. This essay expands the literature and examines the effects of textual information on review helpfulness. By providing new moderators, Essay 2 tries to offer a justification for inconsistent results in previous literature regarding the helpfulness of extreme reviews.

Description

Keywords

Crowdfunding, Rhetorical devices, Functional data analysis, Review helpfulness

Citation