Behind the Review Curtain: Decomposition of Online Consumer Ratings in Peer-to-Peer Markets

Daniel Kaimann

doi:10.3390/su12156185

Abstract

Peer-to-peer markets are especially suitable for the analysis of online ratings as they represent two-sided markets that match buyers to sellers and thus lead to reduced scope for opportunistic behavior. We decompose the online ratings by focusing on the customer’s decision-making process in a leading peer-to-peer ridesharing platform. Using data from the leading peer-to-peer ridesharing platform BlaBlaCar, we analyze 17,584 users registered between 2004 and 2014 and their online ratings focusing on the decomposition of the explicit determinants reflecting the variance of online ratings. We find clear evidence to suggest that a driver’s attitude towards music, pets, smoking, and conversation has a significantly positive influence on his received online ratings. However, we also show that the interaction of female drivers and their attitude towards pets has a significantly negative effect on average ratings.

Highlights

Due to the high volume of user-generated ratings per product in peer-to-peer markets, customers must make a selection to reviews when they search for reliable products or services
We are interested in the following: How can we decompose the variance of online consumer ratings? In which context are the specific determinants of online ratings? How do these determinants interact with each other to indicate an observable product quality?
It is the possible decomposition of online consumer ratings in peer-to-peer markets that this paper addresses, using data from a setting that is especially suited for an econometric study of the disaggregation of reputation mechanisms

Summary

Introduction

Due to the high volume of user-generated ratings per product in peer-to-peer markets, customers must make a selection to reviews when they search for reliable products or services. Few studies have attempted to decompose online ratings either on traditional two-sided markets or on emerging peer-to-peer markets, i.e., ridesharing, accommodation, or home services. Using data from the leading peer-to-peer ridesharing platform BlaBlaCar, we analyze 17,584 users registered between 2004 and 2014 and their online ratings, focusing on analyzing the explicit determinants reflecting the variance of online ratings. After we determine the consumer’s decision elements for the rating assessment, we apply the Shapley value decomposition to identify the expected marginal contribution of the important positive significant independent variables to the given average rating score. The Shapley value decomposition especially calculates the contributors to the online rating by decomposing the R-squared (share of explained variance) of an OLS model into contributions or groups of regressor variables. We present our empirical results, followed by a discussion of the findings and managerial implications

Literature Review

Data and Model

Empirical Analysis

Findings

Concluding Remarks