Start Date
11-8-2016
Description
Effective use of online product reviews is hampered by their large variance in helpfulness. Therefore, there is a great need to better understand the determinants of review helpfulness. As one of the most studied factors, review length has been commonly assumed to have a linear relationship with review helpfulness. However, a recent study reveals a different finding, but it does not provide detailed explanations. This research fills the gap by investigating the factors that moderate the impact of review length on review helpfulness. Drawing on information theory aiming to reduce redundancy, we explore the interaction between review length and several other factors, including readability, the number of product features, and redundancy of product features. The results of regression models confirm all hypotheses in terms of interaction effects. Our findings not only extend the information theory to a new domain but also help improve the understanding of review helpfulness.
Recommended Citation
Kang, Yin and Zhou, Lina, "Longer is Better? A Case Study of Product Review Helpfulness Prediction" (2016). AMCIS 2016 Proceedings. 1.
https://aisel.aisnet.org/amcis2016/Intel/Presentations/1
Longer is Better? A Case Study of Product Review Helpfulness Prediction
Effective use of online product reviews is hampered by their large variance in helpfulness. Therefore, there is a great need to better understand the determinants of review helpfulness. As one of the most studied factors, review length has been commonly assumed to have a linear relationship with review helpfulness. However, a recent study reveals a different finding, but it does not provide detailed explanations. This research fills the gap by investigating the factors that moderate the impact of review length on review helpfulness. Drawing on information theory aiming to reduce redundancy, we explore the interaction between review length and several other factors, including readability, the number of product features, and redundancy of product features. The results of regression models confirm all hypotheses in terms of interaction effects. Our findings not only extend the information theory to a new domain but also help improve the understanding of review helpfulness.