How Does the Variance of Product Ratings Matter?

What do consumers look when they make a purchase decision? Consider that you want to buy a new laptop. First, you usually determine your needs and then you try to find a product that matches these needs. Trying to find a perfect match, you can address to multiple different sources and product descriptions offered both by companies and experts. This process, is greatly facilitated by the evolution of information technology. However, information technology also facilitates the spread of another type of product information which stems from a different source, but it is as influencing (or even more) than the information provided by conventional sources. This type of information are the user generated content.

User generated content is a very broad term that does not refer only to products. It can also exist for entertainment purposes such as videos or even for journalistic purposes. User product reviews are usually different methods that customers use to express their personal opinions and experiences about a certain product. They can take the form of a text or the form of a quantifiable scale usually from 1 to 5. Many retailers have adopted these review methods to help their customer in their decisions, while star ratings are one of the most prevalent review methods. Star ratings are characterized by 3 Vs, namely Volume, Valence and Variance. Volume refers to the number of reviews, valence to the degree of positive or negative sentiment (Mudambi, Schuff, Zhang, 2014) and variance to the distribution of the ratings within the scale. Sun, 2012 identifies a research gap regarding the variance characteristic and focuses her research on answering how consumers perceive different types of variance, how it affects subsequent demand and whether variance is interrelated with average rating (valence). The methods and the findings are discussed in the next paragraph, while the last paragraph focuses on practical implications.

The author recognizes that none of the three aforementioned Vs can deliver meaningful information about how customers interpret ratings when the product and consumer attributes are ignored. Consequently, an econometric model is developed that incorporates the notions of perceived quality and mismatch. The latter describes the level in which the attributes of a product allow customers to be satisfied, therefore, products with high mismatch are more likely to be niche products. The theoretical model suggests that the higher the product quality, the higher the average rating will also be, since every consumer will enjoy a high quality product more, irrespective of their preference match. Hence high average rating is perceived as result of high quality by consumers. Furthermore, the author suggests that when the average rating is low, consumers may not perceive the product as low in quality, if the variance, hence mismatch, is relatively high. The interesting explanation that the author provides is that high variance is perceived as the main factor of the low average rating, something that does not signal low quality, since the existence of bad reviews dramatically deteriorates the average score. The second finding of the model, which is the main proposition of the paper, is confirmed empirically. The effect was also reflected on sales which increased when both the conditions (high variance, low average) were met.

The research of Sun, 2012 is fruitful for different parties. Retailers that sell products with low ratings but with high variance, can exploit the opportunity of the well-matched consumers by increasing the price. For example, the Motorola in the image below has an average rating of 3 stars, however it also has a high rating deviation. Therefore, Amazon in this case could increase the price in order to exploit the group of people who still prefer an “old school” cell phone by expressing their opinion through high ratings. Finally, managers can use variance to make predictions about future demand, product life-cycle and make better portfolio decisions.

References:

Mudambi, S. M., Schuff, D., & Zhang, Z. (2014). Why Aren’t the Stars Aligned? An Analysis of Online Review Content and Star Ratings. In System Sciences (HICSS), 2014 47th Hawaii International Conference, 3139-3147

Sun, M. (2012). How does the variance of product ratings matter? Management Science, 58(4), 696-707.

http://www.amazon.com/Motorola-K1-Unlocked-Slot-International-Warranty/dp/B000JL4Y3Y/ref=pd_rhf_gw_s_cp_2_12Z5?ie=UTF8&refRID=0CFVB2S6H2494MEHE02R (Retrieved 24/4/2015)