On the Relative Efficiency of Crowdsourced Rating Mechanisms: Experimental Evidence

Abstract

This paper experimentally tests the effectiveness of three crowdsourced signaling mechanisms on their ability to resolve an asymmetric information problem over product quality for a set of consumers. Motivated by naturally occurring environments, the first mechanism allowed experimental subjects to submit a signal of high product quality only, the second allowed for a signal of high and a signal of low quality, and the third a five star rating scale of quality, which reported the average rating. In support of the primary hypothesis, the experimental results reveal positive returns to the degree of signal specificity in the rating system given to consumers. Namely, the five- star rating mechanism reduces the presence of asymmetric information more than the high quality signal only mechanism. The high-and-low quality signaling mechanism, which offers an intermediate level of signal specificity on quality, is not statistically different from the other two mechanisms. Furthermore, an analysis of individual characteristics finds evidence that the willingness of a subject to rate products, and thus aid in producing valuable information for other consumers, is critically linked to their pro-social tendencies.