| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 552620 | Decision Support Systems | 2014 | 12 Pages |
•We develop a multiple equation model to examine the inter-relationships between ratings, sentiments and sales.•Ratings do not have a significant direct impact on sales but have an indirect impact through sentiments.•Sentiments have a direct significant impact on sales.•Most helpful and most recent — play a significant role in determining sales.
It is generally assumed that ratings are a numeric representation of text sentiments and their valences are consistent. This however may not always be true. Using a panel of data on over 4000 books from Amazon.com, we develop a multiple equation model to examine the inter-relationships between ratings, sentiments, and sales. We find that ratings do not have a significant direct impact on sales but have an indirect impact through sentiments. Sentiments, however, have a direct significant impact on sales. Our findings also indicate that the two most accessible types of reviews – most helpful and most recent – play a significant role in determining sales. This suggests that information that is easily accessible and cognitive effort-reducing heuristics play a role in online purchase decisions. This study advances our understanding on the inter-relationship between ratings, sentiments, and sales and sheds insight on the relevance of ratings and sentiments over a sequential decision making process.
