| HBS Working Paper Series
Online Word of Mouth and Product Quality Disagreement
Studies of online word of mouth have frequently posited that the level of disagreement between existing product reviews can impact the propensity to review and the valence of future reviews. However, due to purchasing and reporting biases that result from unique facets of consumer behavior, the distribution of online reviews is frequently an amalgamation of two distributions: consumers who liked the product and consumers who did not. Consequently, statistical measures capturing only the dispersion of reviews, such as standard deviation, can be improved by a measure that specifically classifies reviews as belonging to these disjunct populations of consumers. We theoretically develop and empirically test a new measure of disagreement for online word of mouth using a new data set containing nearly 300,000 reviews for 425 movies over three years. We find this measure results in lower standard errors and has higher predictive power than standard deviation. Using this measure, we show that higher levels of disagreement among previously posted reviews lead to a higher propensity to post future product reviews. This effect is amplified by the average length of prior reviews but is decreased by the product's availability in the market. Further, we show that increased disagreement leads to future reviews of lower valence.
Keywords: online word of mouth;
online product reviews;
Marketing Reference Programs;
Social and Collaborative Networks;