Date of Award
8-1-2022
Degree Type
Thesis
Degree Name
Doctor of Philosophy (PhD)
Discipline
Business
Department
Marketing and International Business
First Advisor
Prof. CUI Geng
Second Advisor
Prof. PENG Ling
Abstract
This dissertation contains two essays. For Essay One, previous studies on review helpfulness focus on what makes a review helpful and how to predict review helpfulness. In so doing, researchers hope to identify the most helpful reviews for consumers and improve the recommendation system. However, little is known about the effect of these helpful reviews on product performance. Thus, this paper investigates how helpful reviews or the helpfulness votes influence product sales. Since product sales are only available at the group (product) level, estimating the effect of helpfulness votes presents a challenging multilevel problem. This research considers both the disaggregating (individual) and the aggregating (group) approaches and compares four competing models in their parameter estimates and model fitness. The results suggest that the average number of votes performs the worst while the mean-adjusted model slightly improves predictive power. Among them, the total number of helpfulness votes renders the best predictive performance.
For Essay Two, crowdfunding has become a trendy way to raise funding nowadays. Budding entrepreneurs try to make a convincing pitch to attract potential backers' interest. Existing studies have found that linguistic styles such as the narrative tone, the use of emotional or informational arguments, concreteness, precision, and interactivity are signals of underlying project quality. Nevertheless, this body of research lacks proof of the effect of micro-level linguistic elements on the success of crowdfunding. In this essay, we conduct two studies to investigate the effect of word-level and topic-level linguistic characteristics on crowdfunding outcomes.
In Study One, we adopt a multimethod approach which includes N-gram natural language processing model, penalized logistic regression (PLR), and linguistic analysis to analyze the narratives of projects on Kickstarter. We find that "speaking the same language" and careful choice of words is critical to the success of crowdfunding. Further, the psychological meanings of the words and phrases associated with the success and failure of crowdfunding Our findings will provide a valuable insight for entrepreneurs to prepare their pitches. In Study Two, we switch our focus from the choice of word to the choice of topic. We use topic entropy to measure the theme complexity for each project pitch and examine how it would affect the probability of crowdfunding success. We find a significant prediction power of the topic entropy, with the lower (higher) the value, the more probability the success (failure) of the project. Among successful projects, certain words and topics that have more positive or negative impacts vary depending on the movie genre.
This essay is one of the first in marketing research to use advanced text analysis to evaluate the effect of micro-level linguistic features on message persuasiveness. In addition, this work has further proved the power of language in effective marketing communications.
Keywords
Online Reviews, Consumer Behavior, Helpfulness votes, Multilevel Model, Crowdfunding, natural language processing (NLP), Linguistic Inquiry and Word Count (LIWC), topic entropy, promotion pitch
Language
English
Copyright
The copyright of this thesis is owned by its author. Any reproduction, adaptation, distribution or dissemination of this thesis without express authorization is strictly prohibited.
Recommended Citation
Bao, Z. (2022). The effects of online reviews and promotional messages on product performance: Review helpfulness and the power of language (Doctoral thesis, Lingnan University, Hong Kong). Retrieved from https://commons.ln.edu.hk/otd/153/