Information Cascades in Ratings for Various Movies
Do I Follow My Friends or the Crowd? Information Cascades in Online Movie Ratings [1] – https://pubsonline.informs.org/doi/abs/10.1287/mnsc.2014.2082
This paper, written by Young-Jin Lee, Kartik Hosanagar, and Yong Tan, describes how ratings for online products are generated and how these ratings are often influenced by previous ratings. The authors studied how oftentimes, previous ratings can have some influence on us socially and impact our ratings on products, in particular, online movie ratings. It begins by providing an introduction. It is common knowledge that “online user-generated reviews are an important source of product information on the Internet. They help prospective buyers in gaining from the experience of other users who have tried the product” [1]. However, some recent literature has revealed a lot of interesting information about how we are influenced by user ratings. In one such study mentioned in the paper, they found that viewers often give negative reviews if they see that the prior reviews are also negative. The study attributes this to viewers wanting to follow the norm and wanting to show that they think the same as everyone else. This can be attributed to human behavior and wanting to present themselves as equally as competent and smart as all the other people reviewing. Additionally, if viewers observe that the previous reviews are centered or balanced, they typically shift to support this by not providing extreme opinions. The study cites this as something called the “multiple audience effect” [1]. This effect “indicates that consumers try to give more balanced opinions if they observe heterogeneous ratings in the community” [1]. In the lecture, we talked about herding and how people often go toward what the group thinks. Differentiation is the opposite, meaning deviating from what the norm is. Then the authors hypothesize how there is a “herding behavior of subsequent posters for movies with a large volume of user ratings and a differentiation behavior of subsequent posters for movies with a smaller volume of ratings” [1]. They also provide additional hypotheses that include how the rating of your friends also affects your rating. They performed data collection and different experiments to test their hypotheses. They concluded that the social influence of the ratings also depends on whether the movie that everyone is ranking is popular. They noticed that consistent with hypothesis 1, differentiation behavior occurred more frequently with smaller and more niche movies, that inevitably had fewer people watching them. Additionally, they found that with larger and more popular movies, there was more of a chance for herding behavior to happen. Additionally, they concluded that herding behavior always occurs if the prior ratings include some of the reviewee’s friends. This is typical human nature.
This relates to the topics that we have been learning in the class, specifically about the Herding experiments we talked about in class as well as information cascades. In class, we talked about a Herding experiment involving an urn with two possibilities. The possibilities are that it had either a majority of blue elements or a majority of red elements and these were equally likely combinations. Then students appear sequentially and draw a random ball, look at it, and then guess what the majority color is in the urn. This is an example of a situation where herding could occur, or where students’ decisions are influenced by the previous things they have seen, even if they are not necessarily accurate. I think it was very interesting to see the example of herding and how common it is. The urn example makes sense but it is probably quite rare to see that example in reality, where students line up and draw from an urn and make a guess about if it’s mostly blue or red. However, rating movies is something that we all do or at least deciding what to watch based on reviews. The implications of the study could suggest that many of us would watch a movie we know isn’t as interesting to us, because our friends said it was good. Or we would have harsher opinions about a movie if the consensus of it was balanced and had few opinions but we’d follow what the crowd said about popular movies that had tons of reviews. This also relates to information cascades and how people make a decision in sequential order, something we also discussed in class. As the study says, with fewer reviews, future reviews are more scattered and have more of a range. However, as there are more and more reviews for a movie, future reviews tend to congregate more. We can think of reviews as an information cascade, which each next review being influenced in some way from the previous set of reviews.
This study demonstrates how subconscious some of these behaviors are. Herding and differentiation behavior is prevalent throughout our lives; it’s just the way the human brain operates.