Hedonic Analysis for High-involvement Consumer Electronics Using Online Product Reviews

Abstract

In recent years online product reviews have been considered as a valuable source of information to assist people in making buying decisions Most of prior studies on the effect of online product reviews have utilized the factors which manufactures cannot control by themselves such as the number of reviews the average review rating as independent variables in their regression models However those factors cannot provide direct implications for manufacturers For example managers cannot easily increase the number of reviews to rise the product price or demand In contrast they have to trace the causes of why the amount of reviews grows Thus in order to offer more straightforward suggestions we adopt the concept of hedonic analysis which decomposes the demand of a commodity into several product features to identify which of them impact its demand mostly Further previous surveys find that consumers spend time and effort conducting pre-purchase searches for high-involvement products and visit several external review websites during the search process In this article we take smartphone market as our research target and propose a framework to demonstrate these assumptions Our framework utilizes opinion mining techniques to extract sentiment words and features from online product reviews and combines those extracted items with basic characteristics obtained from a specification of each product to form hedonic regressions In order to examine influences of reviews in more detail we separate the sources of product reviews into two groups—a retailer-hosted website and third-party hosted websites and take this differential into our models Thus we construct three different regression models—(1) considering basic characteristics only; (2) considering the information extracted from a retailer-hosted review source and basic characteristics; (3) considering the information extracted from third-party review sources and basic characteristics to test which one has the highest value of the coefficient of determination Finally we provide managerial implications for firms and help them make proper strategies based on the experiment results

Date of Award	2015 Jun 30
Original language	English
Supervisor	Sheng-Tun Li (Supervisor)