Amazon Product Reviews – Do they create an information cascade?
http://snap.stanford.edu/class/cs224w-2012/projects/cs224w-033-final.v08.pdf
The article Investigating Information Cascades in Amazon Product Reviews discusses whether or not the reviews appended to each product page on Amazon.com follow the wisdom of the crowd and information cascade models. Researchers at Stanford University found that reviews normally did not follow cascade-like behavior, although 31% possibly did. This experiment iterated over 548,552 products and over 7,781,990 reviews. These researchers hypothesized that Amazon data would show a herding effect where product reviewers are influenced by previous reviews to herd around the same opinion and offer reviews that mirror those already existing. These experimenters also created a probability matrix that measured whether the reviewer would write their review based on public signals (other reviews about the product) or private signals (individual information about the product). Although it seems logical to suggest that Amazon product reviews would be influenced by one another (a bad review would influence another buyer to write another bad review), this article empirically proves that Amazon product reviews are independent and do not follow the wisdom of the crowd/information cascade model that we have learned in class.
This article is related to INFO 2040: Networks because it shows a real-life situation in which an information cascade could be present. In this case of Amazon, if one product review set off an information cascade, this cascade has the ability to influence future buyers who see either mostly positive or mostly negative reviews depending on the cascade. This would influence the spread of information and could economically effect the companies selling the products (for example, node switching from product A to product B in Q1 of Pset 8). Since in an information cascade, the individual’s action is based on previous actions rather than their personal information, they make decisions based on who made decisions before them. One would think that Amazon would be a great source of informational cascading, but this article proves not. Amazon also employs the long-tail business model, which is another topic we have discussed in class.