The Effects of Text Complexity on Online Review Helpfulness

Abstract

ABSTRACT This research reported in this paper extends the literature on the helpfulness of online reviews. Previous research has assessed online reviews using standard unidimensional readability algorithms. This research extends previous work by investigating a multidimensional framework, and associated measures, of text complexity and its impact on the helpfulness of online reviews. Results show that as the amount of passive voice and negation in online reviews increase, the helpfulness of said review decreases. Other significant predictors of review helpfulness include word meaningfulness, lexical diversity, and the number of modifiers per noun phrase. Keywords: online reviews, computational linguistics, Coh-Metrix INTRODUCTION Advancements in internet-based technologies have resulted in reduced barriers to entry for user created content. Anyone with an internet connection can easily create content through websites, message boards, online forums, blogs and a variety of different social media outlets. One area where user generated content has garnered much interest is in online product reviews. Increasingly consumers are relying on peer based product reviews to inform their purchasing decisions. A recent article in Harvard Business Reviews states that 30% of US consumers go to Amazon to access their rich repository of product information and associated reviews to start their purchasing process (Simonson & Rosen, 2014). In addition, studies commissioned by Google have reported that consumers consult an average of 10 information sources before making a final purchase decision (Simonson & Rosen, 2014). These statistics are further supported by research that has found that customer reviews can positively impact customer sales (Clemons, Gao, & Hitt, 2006). The aforementioned results along with the fact that peer-based product reviews may have more credibility than seller based information (Bickart & Schindler, 2001) has established the importance of user generated content in the form of online product reviews in a consumer's purchasing intention. A central theme in previous online review research is the determination of what impacts the usefulness or helpfulness of a review. Research has explored impacts and relationships between many components of online review helpfulness including characteristics of the reviewers, different product types, attributes of the review, and the content of the review itself. The focus of this study is on the characteristics of the online review text that may affect its comprehension, and thus helpfulness, by potential consumers. While some work in this area has been completed, the research is limited by the simplicity of the constructs used to assess text difficulty. This exploratory research moves the online review literature forward by incorporating advancements in computational linguistics to explore how text coherence, structural complexity, and word complexity affect the helpfulness of online reviews. It takes a multidimensional approach to text difficulty through the use of a comprehensive suite of tools and indices called Coh-Metrix (McNamara, Graesser, McCarthy & Cai, 2014). This research is based on the premise that an online review that is more coherent, structurally simple, and uses less complex words that have fewer meanings will be more helpful to consumers. This paper is organized as follows. First, a brief review on online review helpfulness is presented. Next, computational linguistics and coherence is introduced which forms the basis of the research model to be explored. Hypothesized relationships between the model constructs are presented followed by a methodology section and results. Concluding remarks and areas for future exploration are also included. BACKGROUND Online Reviews Research on online reviews is quite varied and comprehensive. The focus of this paper research was to explore the helpfulness or usefulness of online product reviews. …