Article

Why Do Online Product Reviews Have a J-Shaped Distribution? Overcoming Biases in Online Word-of-Mouth Communication

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Online word-of-mouth communication in the form of product reviews is a major information source for consumers and marketers about product quality. The literature has used the mean of online reviews to predict product sales, assuming that the mean reflects product quality. However, using a combination of econometric, experimental, and analytical results, we show that the mean is a biased estimator of product quality due to two self-selection biases (purchasing and under-reporting bias). First, econometric results with secondary data from Amazon.com show that almost all products have an asymmetric bimodal (J-shaped) distribution with more positive than negative reviews. Second, experimental results where all respondents wrote reviews show that their reviews have an approximately normal distribution with roughly equal number of positive and negative reviews. This implies two biases: (1) purchasing bias - only consumers with favorable disposition towards a product purchase the product and have the opportunity to write a product review, and (2) under-reporting bias - consumers with polarized (either positive or negative) reviews are more likely to report their reviews than consumers with moderate reviews. This results in a J-shaped distribution of online product reviews that renders the mean a biased estimator of product quality. Third, we develop an analytical model to derive the conditions for the mean to become an unbiased estimator of product quality. Based on these conditions, we build a new model that integrates three distributional parameters - mean, standard deviation, and the two modes of the online product reviews (to overcome under-reporting bias) and product price (to overcome purchasing bias). This model is shown to be a superior predictive model of future product sales compared to competing models.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... A customer only having one review is classified as not decided, because only one review is not enough to determine his/her (optimistic/ pessimistic) tendencies. However, because there is no ground truth for "true" product quality (i.e., the collective assessment of the product's perceived value across customers [15]), product sales rankings (i.e., the ordinal ranking of product sales within a product category (e.g., Amazon Best Sellers)) are used as an approximation of product quality instead [8]. Figure 1 indicates that optimistic and pessimistic reviewers are reviewers who have given high customer ratings to all products (the dotted line on the right-hand side) and low customer ratings to all products (the dotted line on the left-hand side), respectively. Realistic reviewers have consistently given high ratings to high-ranked products and low rankings to low-ranked products in the product sales rankings (the solid line from the lower left corner to the upper right corner). ...
... This is because many customers tend to write reviews when they are significantly satisfied or significantly dissatisfied with their purchased products. The results indicate that the current product ratings on a large number of electronic commerce websites do not represent true product quality, because the product ratings do not follow a normal distribution [15]. ...
... Fig. 7 The correlation coefficients of product sales rankings Fig. 8 The correlation coefficients of reviewers' sentiment scores Figure 9 shows the distributions of the original star ratings and the adjusted star ratings for 739 products that are listed in Amazon Best Sellers in Electronics. The original ratings in Fig. 9 follow a J-shaped distribution and correspond with the findings of Hu et al. (i.e., the J-shaped distribution in Fig. 3) [15]. The distributions of the adjusted ratings, where a ¼ 0.5 and a ¼ 1, are slightly shifted from a J-shaped distribution to a normal distribution. ...
Article
The authors of this work present a model that reduces product rating biases that are a result of varying degrees of customers' optimism/pessimism. Recently, large-scale customer reviews and numerical product ratings have served as substantial criteria for new customers who make their purchasing decisions through electronic word-of-mouth. However, due to differences among reviewers' rating criteria, customer ratings are often biased. For example, a three-star rating can be considered low for an optimistic reviewer. On the other hand, the same three-star rating can be considered high for a pessimistic reviewer. Many existing studies of online customer reviews overlook the significance of reviewers' rating histories and tendencies. Considering reviewers' rating histories and tendencies is significant for identifying unbiased customer ratings and true product quality, because each reviewer has different criteria for buying and rating products. The proposed customer rating analysis model adjusts product ratings in order to provide customers with more objective and accurate feedback. The authors propose an unsupervised model aimed at mitigating customer ratings based on rating histories and tendencies, instead of human-labeled training data. A case study involving real-world customer rating data from an electronic commerce company is used to validate the method.
... Not surprisingly, research on online reviews has increased tremendously recently, with topics including sales (Chevalier & Goolsbee, 2003;Chevalier & Mayzlin, 2006;Dellarocas, Zhang, & Awad, 2007;Duan, Gu, & Whinston, 2008), trust (Resnick & Zeckhauser, 2002), reviewer characteristics (Forman, Ghose, & Wiesenfeld, 2008;Otterbacher, 2012), motivations (Mackiewicz, 2008;Yew, 2009), rhetoric (Otterbacher, 2011), and biases (Dellarocas & Wood, 2008;Hu, Pavlou, & Zhang, 2007;Jøsang, Ismail, & Boyd, 2007;Mackiewicz, 2008;Otterbacher & Hemphill, 2012). Some of these studies have identified different outcomes based on the kind of things that are being reviewed. ...
... Barnesandnoble.com, Citysearch, and Yahoo Local (Chevalier & Goolsbee, 2003;Chevalier & Mayzlin, 2006;Dellarocas & Wood, 2008;Hu et al., 2007;Hu, Zhang, & Pavlou, 2009;Mackiewicz, 2008;Wang, 2010). This nonparametric distribution is referred to as the J-shaped distribution because it is marked by high levels of extremely positive ratings, some negative ratings, and very few moderate ratings (see Figure 1). ...
... Through a comprehensive analysis of product reviews on Amazon.com, Hu, Pavlou, and Zhang (2007;) investigated the J-shape distribution and identified that two sources of bias foster this J-shaped distribution: purchasing bias and under-reporting bias. The purchasing bias occurs because people who anticipate enjoying a product are the ones that tend to purchase it, while people who do not value a product are not likely going to buy it. ...
Thesis
Full-text available
This dissertation investigates how businesses are able to align the collective actions of a disconnected crowd with the strategic goals of the organization. I examined this questions within the context of the business review website Yelp through a quantitative analysis of nearly 60,000 business reviews, 17 in-depth qualitative interviews with reviewers, and a two-year ethnography. Interpreting the results of this data within the framework of the collective action space (Bimber, Flanagin, & Stohl, 2012) indicates that Yelp is able to manage the contributions of a relatively small subset of reviewers through the Yelp Elite Squad. Rather than simply motivating more reviews, the Elite Squad encouraged reviewers to interact more personally with other reviewers and accept increased institutional engagement with Yelp. In encouraging members of the crowd to produce online reviews within this context, Yelp was able to use organizational culture as a control strategy for encouraging Elite reviewers to adopt a pre-mediated reviewing approach to their reviews. This increased the frequency of moderate reviews and decreased the frequency of extreme reviews. This behavior ultimately furthers the organizational goals of Yelp, as moderate reviews are considered to be more helpful for reviews of businesses. Finally, implications for crowdsourcing, big data analysis, and theory are discussed.
... We believe this is the first such attempt to consider individual reviewers' unique reference points while evaluating actual review data, which may provide a foundation for more theoretically informed approaches (Z. Chen & Lurie, 2013;Hu et al., 2009;Nguyen et al., 2021;Wang et al., 2019). ...
... Contrary to previous studies that regarded 3 as the midpoint of 5-star ratings, such that ratings of 1 and 2 were seen to be negative and ratings of 4 and 5 were seen to be positive (Z. Chen & Lurie, 2013;Hu et al., 2009;Nguyen et al., 2021;Wang et al., 2019), we calculated the average of the past ratings given by a certain reviewer and used this as the reference point. By subtracting this reference point from the raw star rating (spanning 1-5 stars), we used continuous variable values as inputs for the analysis. ...
... In this context, this study-to the best of our knowledge-is the first to reflect each reviewer's unique reference points when dealing with actual review data on restaurants, thereby distinguishing itself from previous studies that have merely incorporated the midpoint 3-star rating as the uniformly equal reference point across different reviewers (Z. Chen & Lurie, 2013;Hu et al., 2009;Nguyen et al., 2021;Wang et al., 2019). In this sense, this study holds significance in that it lays the foundation for more theoretically precise measurements of the valence of ratings for future field studies. ...
Article
While some online reviews explicitly praise or criticize a product, others reveal a neutral evaluation. We predicted that extreme reviews would be considered more useful than moderate ones, and that negative reviews would be considered more useful than positive ones. To test these predictions, this study collected a dataset comprising 951,178 reviews of New York restaurants made by 142,286 reviewers on Yelp.com. By combining these two datasets, we incorporated each reviewer’s unique reference point into a model and showed that extremely positive or negative reviews were considered more useful than moderate ones and that negative reviews were considered more useful than positive ones. This dominance of negative over positive reviews was also more pronounced in the conditions of larger variance and lower average ratings for restaurants. Overall, these results support the presence and influence of extremity and negativity biases, particularly in the context of high preference heterogeneity.
... ) suggested a strong connection between product demand and the number of consumer reviews. One potential problem with utilizing movie ratings as a proxy for demand is that online review data is known to be biased because users tend to review items that they extremely like or dislike, thus causing the histogram to be J-shaped or U-shaped, see (Hu et al., 2007; Dellarocas and Narayan, 2007; Dellarocas and Wood, 2008), and citations therein. To check for the possibility of this bias in our data, we plot the histogram of the rating values on a scale from one to five in Figure 2 (right). ...
... Furthermore, previous literature (see Chen et al. 2004) suggested a strong connection between product demand and the number of consumer reviews. One potential problem with utilizing movie ratings as a proxy for demand is that online review data is known to be biased because users tend to review items that they extremely like or dislike, thus causing the histogram to be J-shaped or U-shaped, see (Hu et al., 2007; Dellarocas and Narayan, 2007; Dellarocas and Wood, 2008), and citations therein. To check for the possibility of this bias in our data, we plot the histogram of the rating values on a scale from one to five inFigure 2 (right). ...
... For example Clemons et al. (2006) demonstrate the relationship between variance of ratings and demand for products. Further, Hu et al. (2007) recommend controlling for the standard deviation of ratings as well as for two modes to overcome consumer under-reporting bias. Since in our case the distribution of ratings is symmetric, we do not control for the modes. ...
Article
Full-text available
We analyze a large data set from Netflix, the leading online movie rental company, to shed new light on the causes and consequences of the Long Tail effect, which suggests that on the Internet, over time, consumers will increasingly shift away from hit products and toward niche products. We examine the aggregate level demand as well as demand at the individual consumer level and we find that the consumption of both the hit and the niche movies decreased over time when the popularity of the movies is ranked in absolute terms (e.g., the top/bottom 10 titles). However, we also observe that the active product variety has increased dramatically over the study period. To separate out the demand diversification effect from the shift in consumer preferences, we propose to measure the popularity of movies in relative terms by dynamically adjusting for the current product variety (e.g., the top/bottom 1% of titles). Using this alternative definition of popularity, we find that the demand for the hits rises, while the demand for the niches still falls. We conclude that new movie titles appear much faster than consumers discover them. Finally, we find no evidence that niche titles satisfy consumer tastes better than hit titles and that a small number of heavy users are more likely to venture into niches than light users.
... Unfortunately, detecting discussions of defective products in online reviews and responding accordingly is a difficult process. The majority of online reviews describe positive experiences with products (Hu et al., 2009), and even most one-star reviews may not discuss product defects so much as generic negative sentiment relating to poor customer-product fit (Abrahams et al., 2015). Some researchers have suggested sentiment analyses of online reviews on the assumption that negative sentiment might follow a poor experience with a product of low quality (Hu et al. 2006(Hu et al. , 2014Yu et al., 2013). ...
... Table 1 captures a broad cross-section of analyses across product and service categories, and in general the research has found that online reviews are a valuable signal of quality. However, it is important to emphasize that some prior work has found that both extremely positive and extremely negative experiences are highly represented (Hu et al., 2009). Yet, for defect surveillance, representation of negative experiences is especially important (Abrahams et al., 2015). ...
... Although labeled data can be quite difficult to acquire, it can yield powerful results. Supervised text analytics have been used for a wide variety of applications, including financial projections (Eliashberg et al., 2014), fraud identification (Holton, 2009), Duan, Gu, and Whinston (2008) Volume of movie reviews is a significant indicator of box office sales Hu et al. (2008) Online Amazon review star ratings are predictive of product sales Hu et al. (2009) Consumers with extreme positive or negative experiences are most likely to post online reviews to "brag or moan" Cui, Lui, and Guo (2012) Both volume and star rating of electronics reviews are significant indicators of sales Phillips, Zigan, Silva, and Schegg (2015) Star ratings and assessments of room quality are predictive of hotel performance Abrahams et al. (2015) Products with high sales volumes and thus many reviews benefit from quality analytics, as reviews are more representative Qi et al. (2016) Topics discussed in online reviews and associated sentiment provide sufficient data to drive product improvement Chong et al. (2018) Existing reviews substantially impact consumer perceptions and purchase decisions Inf Syst Front and plagiarism detection (Oberreuter & VeláSquez, 2013). Within this category, deep learning-based word embedding analyses have become a popular state-of-the-art tool (Brahma et al., 2021). ...
Article
Full-text available
Online reviews contain many vital insights for quality management, but the volume of content makes identifying defect-related discussion difficult. This paper critically assesses multiple approaches for detecting defect-related discussion, ranging from out-of-the-box sentiment analyses to supervised and unsupervised machine-learned defect terms. We examine reviews from 25 product and service categories to assess each method’s performance. We examine each approach across the broad cross-section of categories as well as when tailored to a singular category of study. Surprisingly, we found that negative sentiment was often a poor predictor of defect-related discussion. Terms generated with unsupervised topic modeling tended to correspond to generic product discussions rather than defect-related discussion. Supervised learning techniques outperformed the other text analytic techniques in our cross-category analysis, and they were especially effective when confined to a single category of study. Our work suggests a need for category-specific text analyses to take full advantage of consumer-driven quality intelligence.
... Researchers have studied different types of online review biases, including (1) selfselection bias, which reflects various sampling bias due to users' self-selection at different stages of the review process (Hu et al., 2006(Hu et al., , 2009(Hu et al., , 2017Li and Hitt, 2008), (2) social influence bias, which results from the interactive effects among different user reviews (Askalidis et al., 2017) and leads to sequential biases when examined sequentially (e.g. Muchnik et al., 2013;Wu et al., 2017) and (3) biases related to data authenticity or credibility, which stem from deliberate manipulations of online reviews (e.g. ...
... The J-shaped distribution has a bimodal feature, with two modes or peaksan upper peak around the highest rating and a lower peak around the lowest ratingand a dip in between, reflecting a small number of moderate ratings. Researchers have found that the J-shaped distribution held irrespective of mean ratings, phases over time and volumes of reviews (Gao et al., 2015;Hu et al, 2009). ...
... Past studies have identified two mechanisms of self-selection bias in online reviews, including (1) acquisition bias (or purchasing bias)only consumers predisposed favorably toward a product would self-select to acquire a product and hence have a chance to write a review (Admati and Pfleiderer, 2004;Gao et al., 2015;Hu et al., 2006Hu et al., , 2009Li and Hitt, 2008) and (2) underreporting bias (also called nonresponse biases or polarity self-selection)consumers with extreme experiences are more likely to write reviews, i.e. "brag" or "moan" online (Hu et al., 2006), than modestly satisfied consumers (Bethlehem, 2010;Filieri, 2016;Gao et al., 2015;Schoenm€ uller et al., 2019). Acquisition bias means that a product's user population is often positively biased, as indicated by the common asymmetric right skewness of review distributions, and thus unrepresentative of the general population. ...
Article
Full-text available
Purpose Online review bias research has predominantly focused on self-selection biases on the user’s side. By collecting online reviews from multiple platforms and examining their biases in the unique digital environment of “Chinanet,” this paper aims to shed new light on the multiple sources of biases embedded in online reviews and potential interactions among users, technical platforms and the broader social–cultural norms. Design/methodology/approach In the first study, online restaurant reviews were collected from Dianping.com, one of China's largest review platforms. Their distribution and underlying biases were examined via comparisons with offline reviews collected from on-site surveys. In the second study, user and platform ratings were collected from three additional major online review platforms – Koubei, Meituan and Ele.me – and compared for possible indications of biases in platform's review aggregation. Findings The results revealed a distinct exponential-curved distribution of Chinese users’ online reviews, suggesting a deviation from previous findings based on Western user data. The lack of online “moaning” on Chinese review platforms points to the social–cultural complexity of Chinese consumer behavior and online environment that goes beyond self-selection at the individual user level. The results also documented a prevalent usage of customized aggregation methods by review service providers in China, implicating an additional layer of biases introduced by technical platforms. Originality/value Using an online–offline design and multi-platform data sets, this paper elucidates online review biases among Chinese users, the world's largest and understudied (in terms of review biases) online user group. The results provide insights into the unique social–cultural cyber norm in China's digital environment and bring to light the multilayered nature of online review biases at the intersection of users, platforms and culture.
... In our own data, we find a correlation of about 0.5 between the monthly number of ratings and the rental turns among the matched movies. Second, unlike other review data that are known to have selection bias because users tend to review items they extremely like or dislike (see Hu et al. 2007;Dellarocas and Narayan 2007;Dellarocas and Wood 2008 and citations therein), pure ratings may avoid this bias because giving a rating is much less costly to a user than writing a review. In our data, we plot the histogram of the rating values on a scale from one to five and find the rating of four to be the most frequent, followed by the ratings of three, five, two and one (more on, see Figure 2.3.3). ...
... For example Clemons et al. (2006) demonstrate the relationship between variance of ratings and demand for products. Further, Hu et al. (2007) recommend controlling for the standard deviation of ratings as well as for two modes to overcome consumer under-reporting bias. Since in our case the distribution of ratings is symmetric, we do not control for the modes. ...
Article
In the first part of the dissertation, we empirically examine the impact of expanded product variety due to the adoption of the Internet on demand concentration, taking endogeneity into consideration. We analyze two large data sets from the movie rental industry at both movie-level and consumer-level. We find that product variety diversifies the demand for niches more significantly than for hits in absolute terms (e.g., the top/bottom 1,000 titles). However, using relative terms (e.g., the top/bottom 10% of titles) to dynamically adjust for the changing product variety, we find that product variety increases the demand for the hits and reduces the demand for the niches. We further find that product variety increases monthly Gini Coefficients, a measure of demand concentration. We propose that new products appear much faster than consumers discover them. Finally, we find no evidence that niche titles satisfy consumer tastes any better than popular titles and that a small number of heavy users are more likely to venture into niches than light users. In the second part of the dissertation, we analyze a large, detailed operational data set from a restaurant chain to shed new light on how workload (defined as the hourly average number of diners assigned to a server) affects servers' performance (measured as hourly sales). We use an exogenous shock - implementation of a labor scheduling software - to disentangle the endogeneity between demand and supply in this setting. We find that when the overall workload is low, an increase in the workload leads to higher server performance. However, there is a saturation point after which any further increase in the workload leads to degradation of performance. In the focal restaurant chain we find that this saturation point is generally not reached and, counter-intuitively, the chain can reduce the staffing level and achieve both significantly higher sales (an estimated 35% increase) and lower labor costs (an estimated 20% decrease).
... Finally, a retailer's reputation is a key measure of vendor-specific quality. Economists (Hu et al. 2007; Holleschovsky 2015) have found that reputation and price have a positive relation, noting that customers have greater proclivity to pay more to retailers with a high-reputation H3: Vendor-specific quality in e-commerce has significant relationship with firm performance The aforementioned review highlights an introduction to e-commerce its importance. It also identified a list of indicators that used to measure the e-commerce performance which includes sales, profit, customers' satisfaction, and brand and inventory management, and cost reduction which refer to dependent variable (firm performance) as shown in the framework Figure (2). ...
... In this study, vendor quality was found to have a significant relation with both the quality of the product and service and the firm performance which is shown in Table (4: 19) summary of the statistic. Therefore, this research confirms the finding of study by (Hu et al. 2007;Holleschovsky 2015) which found that reputation and price have a positive relation, noting that customers have greater proclivity to pay more to retailers with a high-reputation ...
Article
Full-text available
In the last decade large organizations have realized the increasing impact of new and cutting-edge technology. The importance of technology is underlying on its role on competitive advantage. Information technology in the form of internet has allowed organizations and their management to benefit from better quality information and better decision making. The e-commerce through internet has been shown to be an extremely effective in attracting computer literate people, especially Internet has made a prolific influence on the business world. Thus, the internet has become an area where consumers are able to find what they need. However, at the same time the adoption of the technology is always associated with challenge which are related to technical and managerial issues. These issues form a big challenge to the users in many developing countries, especially in Iraq where the literacy about internet and technology use still humble. This leaves some negative repercussions on the development of small and medium enterprises. It adds heavy burden to both buyers and customers. On one hand, the buyers with the absence of effective communication with customers always lose market share due to market information scarcity and consequently lose their present in the market. On the other hand, the customers with the absence of e-commerce that has excellent quality in information, software and system have no potential to reach good deal with the sellers and consequently lose power of negotiation which affect their loyalty towards the company. Therefore, this research aims to bridge this gap by identifying the critical factors of e-commerce quality that influence SMEs performance. The quantitative method was used in this research driven by survey randomly distributed to 129 SMEs in ceramic industry in Iraq. The multiple research analysis was used to analysis the data. The finding of this research emphases that shows that the quality of website design has significant positive effect on SMEs performance at (P. value 0.000 and T test 3). The finding of this research emphases that shows that the vender specific design has significant positive effect on SMEs performance at (P. value 0.000 and T test 3).
... Given this bimodal distribution of total retweet count, we cannot help but think of the notorious J-shaped distribution often revealed in product reviews with many 5-star ratings, some 1-star ratings, and hardly any ratings in between. Hu et al. (2007) attribute the J-shaped distribution to two biases: "(1) purchasing bias -only consumers with a favorable disposition towards a product purchase the product and have the opportunity to write a product review, and (2) under-reporting bias -consumers with polarized (either positive or negative) reviews are more likely to report their reviews than consumers with moderate reviews." Our results show that similar biases seem to apply when people process and react to information of different truthfulness on the internet: (1) "following bias" -readers/followers of fact-checkers like PolitiFact tend to be people who are more passionate in debunking fake news and are more sensitive to the ruling of misinformation; and (2) "under-reporting bias" -people with more polarized views are more likely to spread information than those with moderate views. ...
... As compared to a Half True rating, a more conclusive rating, either at the True or the False end of the truthfulness spectrum, tend to get more retweets. This finding can be attributed to confirmation bias (Modgil et al., 2021) and under-reporting bias (Hu et al., 2007) in that people tend to seek and favor information that supports one's prior beliefs and, in response, to disseminate the information to others. Fact-checks with a conclusive rating, either true or false, make it a lot easier for people to spot information confirming their beliefs and hence tend to be shared more. ...
Article
Full-text available
Misinformation on social media has become a horrendous problem in our society. Fact-checks on information often fall behind the diffusion of misinformation, which can lead to negative impacts on society. This research studies how different factors may affect the spread of fact-checks over the internet. We collected a dataset of fact-checks in a six-month period and analyzed how they spread on Twitter. The spread of fact-checks is measured by the total retweet count. The factors/variables include the truthfulness rating, topic of information, source credibility, etc. The research identifies truthfulness rating as a significant factor: conclusive fact-checks (either true or false) tend to be shared more than others. In addition, the source credibility, political leaning, and the sharing count also affect the spread of fact-checks. The findings of this research provide practical insights into accelerating the spread of the truth in the battle against misinformation online.
... Second is the social influence bias, where new raters are influenced by existing ratings and, thereby, existing positive ratings dramatically affect future ratings [34]. Third is the under-reporting bias, where consumers who are greatly satisfied or dissatisfied are more likely to report a rating [15]. This last bias may be amplified when consumers view star rating systems as reflecting attitude extremity or deviation from the midpoint of an attitude scale [22]. ...
... This is because the survey is participation-based and customers may leave the chat without answering any questions. Figure 1 displays the histogram of satisfaction scores from the remaining 16.22% of the sessions, which follows the expected J-shaped distribution that is commonly seen in online ratings [15]. From the survey on satisfaction with chat sessions, the largest proportion of customers (45%) indicated they were Very Satisfied with the chat experience and a much smaller proportion (14%) expressed they were Very Dissatisfied with the service. ...
Article
Customer ratings are valuable sources to understand their satisfaction and are critical for designing better customer experiences and recommendations. The majority of customers, however, do not respond to rating surveys, which makes the result less representative. To understand overall satisfaction, this paper aims to investigate how likely customers without responses had satisfactory experiences compared to those respondents. To infer customer satisfaction of such unlabeled sessions, we propose models using recurrent neural networks (RNNs) that learn continuous representations of unstructured text conversation. By analyzing online chat logs of over 170,000 sessions from Samsung's customer service department, we make a novel finding that while labeled sessions contributed by a small fraction of customers received overwhelmingly positive reviews, the majority of unlabeled sessions would have received lower ratings by customers. The data analytics presented in this paper not only have practical implications for helping detect dissatisfied customers on live chat services but also make theoretical contributions on discovering the level of biases in online rating platforms.
... In our case, the distribution of the Amazon reviews used is asymmetric, meaning that for a huge collection of reviews on a one-to five-star scale, the distribution shows a parabolic shape with a minimum at two (stars). For that reason, this distribution is called J-shaped distribution [11,12]. ...
... Consider complete phrasesInput: graph database дdb 1 set n to 2; 2 while there are n-gram nodes in дdb do3 for each n-gram node C in дdb do4 determine all (n − 1)-grams D in дdb representing subsets of C;5 for each (n − 1)-gram D do6 delete all occurrence edges of D incident to sentences also incident to C via occurrence edges; each (n − 1)-gram node D in дdb do10 recalculate frequency f (D) based on the remaining occurrence edges;11 if f (D) ≤f n−дr am then12 delete D from дdb together with all originating edges; SV D and σ SV D and update D; of Phrases with High Standard Deviation. As a last postprocessing step, we look at the standard deviations of the sentiment values σ SV i of each graph node. ...
Conference Paper
In this paper, we describe a new algorithm designed to generate lexical resources in the field of sentiment analysis. For this approach, based on corpora of customer reviews, we determine words and phrases as candidates for our sentiment lexicon solely by calculating a word co-occurrence measure and by considering word frequencies. The sentiment values of every single word or phrase are derived automatically from the review titles and the associated given ratings. We consciously renounce the use of natural language processing methods in order to ensure language independency of our algorithm. Furthermore, by using exclusively statistical methods, we are able to identify rather unusual word combinations, such as idiomatic expressions. This differentiates our work from most prior approaches which concentrate on single words or word-modifier combinations. An example lexicon is generated by the use of a corpus of 1.5 million German Amazon customer reviews.
... Meanwhile, the latter only presents the average rating score, which does not allow customers to understand the product in detail. It also exaggerates the negative effects of low ratings in a certain attribute [18,28]. In summary, both online review systems have two-sided effects that significantly impact the customers' perceptual process and utility functions. ...
... In the aggregated review system, a customer with extremely low satisfaction in a certain attribute is inclined to give an extreme rating score (e.g., zero or one star), which is consistent with Assumption 4 and some previous empirical results [18,20]. Hence, as shown in Fig. 5, we argue that a customer tends to rate an overall rating score of s = 0 when he or she is entirely dissatisfied with the delivery time. ...
Article
Full-text available
This paper examines the role of a subdivided or aggregated online review system to help online takeaway vendors select the most appropriate information strategy. First, we develop two models to depict the interaction between online vendors’ information strategies and consumers’ responses. Second, we take the multidimensional product attributes with their corresponding weights into consideration and illustrate that the sensitivity to product misfits, instead of the relative importance of product attributes, dominates profit maximization. Third, we make a comparison to find the most appropriate scenario to adopt a full or partial information strategy. When a large number of consumers satisfy the delivery time performance, an aggregated review system will be a better choice. Otherwise, vendors are advised to host a subdivided review system. Finally, we universally identify a variance boundary in the rating-star review system, which not only prevents consumers from expressing their real feelings but also makes observing consumer feedback and strategic adjustments inconvenient for online vendors. © 2018 Springer Science+Business Media, LLC, part of Springer Nature
... Schuckert, Liu, and Law (2016) stated that online review platforms do not offer any methods and procedures to help monitor the reliability of reviews. Previous research has shown that online generated consumers' reviews are subject to different types of biases (Hu et al., 2006;Hu, Pavlou, & Zhang, 2009;Lauw, Lim, & Wang, 2008;Li & Hitt, 2008;Munar & Jacobsen, 2014;Nguyen, Bin, & Campbell, 2012;Sikora & Chauhan, 2012). ...
... Different types of biases in online customer product reviews have been studied by many researchers and they may include self-selection bias, under-reporting bias, non-response bias, sequential bias, social influence bias, and cultural value bias (Askay, 2015;Groves, 2006;Hu et al., 2006Hu et al., , 2008Hu et al., , 2009Koh, Hu, & Clemons, 2010;Krishnan, Patel, Franklin, & Goldberg, 2014;Li & Hitt, 2008;Moe & Trusov, 2011;Purnawirawan et al., 2012;Sikora & Chauhan, 2012). Empirical studies have also found that, in some contexts, consumers behave and express their opinions differently online compared to offline (Levin, Levin, & Heath, 2003;Levin, Levin, & Weller, 2005;Nguyen et al., 2012;Ring, Tkaczynski, & Dolnicar, 2016). ...
Article
User-generated hotel ratings have been found to be an important element in customers’ decision making. Nevertheless, most hotel ratings online show average ratings that are extremely positive. Thus, the question is raised of whether online ratings reflect objective evaluation of the reviewers. This study examines the distribution of online ratings and compares it the distribution of offline ratings. Online hotel score ratings reported on Booking.com were extracted and compared with offline rating scores from a field survey. Online and offline hotel ratings of eight hotels located in Seoul were collected. The ratings were compared using Welch’s t-test. Overall, both online and offline ratings for all eight hotels show generally positive distribution curves. The distribution curves exhibit shapes analogous to those considered to contain self-selection bias. The randomized offline ratings did not produce more moderate ratings as predicted. Thus, no indication of non-response bias was found. Practically, this study supports the notion that hotels listed on online travel agencies such as Booking.com may already have passed certain quality criteria, thereby providing competitive marketplaces for customers searching for hotels online.
... The intention to purchase a product is not always the motivation to read a product review (Burton & Khammash, 2010). Also, not everyone who buys a product would write a review (Hu, Pavlou, & Zhang, 2007). People read reviews so that they can minimize the risk (if not possible to eradicate it) associated with a product they have never used (Burton & Khammash, 2010). ...
... Ganu et al. (2009) also stated that the user-input star rating (an extremeness rating component of argument quality) was possibly the most important metadata information in reviews. The focus is on extreme reviews, as online reviews are more extreme rather than moderate (Hu et al., 2007), and also because people assign more importance to extreme cues. In other words, they take more into consideration a piece of information that is either highly positive or highly negative (Fiske, 1980). ...
Article
Social media has boosted information sharing and user-generated content. Consequently, many restaurant goers rely on online reviews for dining recommendations. This study adds to the sparse literature on the influence of review extremeness, source credibility, website quality, and information usefulness on information adoption. Most notably, a modified information adoption model with the addition of website quality was tested in the context of restaurant review websites. Respondents answered survey questions based on what they saw in a simulated restaurant review website which depicted one of eight scenarios. Results showed that the more negative a review, the more useful it is perceived to be. Perceived source credibility of the review writer was positively related to the perceived information usefulness. The only component of website quality that played a significant role in determining information adoption tendency of the review readers was the quality of the information disseminated in the website. Lastly, information usefulness also was positively related to information adoption.
... Rating-based systems, however, suffer from self-selection bias. People who give ratings or write reviews generally do so to document negative experiences [23], thereby making the ratings 'Polarised' [24]. These ratings, thus, do not offer an exact picture of what most people think, and are mostly skewed [24,26]. ...
... People who give ratings or write reviews generally do so to document negative experiences [23], thereby making the ratings 'Polarised' [24]. These ratings, thus, do not offer an exact picture of what most people think, and are mostly skewed [24,26]. ...
Conference Paper
Evaluating trustworthiness of peers prior to engaging in any collaborative or transactional activity is crucial in sharing economy platforms. Current systems are heuristic in nature, limited in their approach to model trust. They provide non-personalized, unstructured data leading to increased time and effort while assessing trustworthiness of available options. These systems also fail to adapt to user's changing trust profile for different contexts and situations. In this paper, we propose a trust framework that helps structure factors essential in evaluating trustworthiness of peers and services they offer or seek. We apply this framework to build a recommender system that learns user-specific dynamic trust profile and provide personalized trustworthy recommendations. We implement this system in context of peer-to-peer ride-sharing where the system uses features from the framework, user feedback and platform usage data to learn a personalized dynamic trust profile for recommending the most trustworthy rides.
... This session generated the most unprompted positive comments on the survey documents. This high level of enthusiasm is consistent with the theory of word of mouth communication, in that consumers with either very positive or very negative views are much more likely to communicate their opinions than those with moderate views [42,44,45]. Customer 'delight' is considered to be the central driver of positive output WOM. ...
... Investment in promotion, sufficient that the target market is exposed multiple times to good quality materials and well-designed messages, is important if high levels of recruitment are to be achieved, and therefore good levels of WOM generated. While it is widely accepted that enjoyment is associated with exercise adherence, policy makers should consider that in order to activate output WOM, the most influential marketing tool, levels of enjoyment need to be very, not just moderately, high [42,44,45]. In addition, an intervention may require a certain longevity in order to fully capitalize on the sales impact of WOM. ...
Article
Full-text available
Background: The beneficial effect of physical activity for the prevention of a range of chronic diseases is widely acknowledged. These conditions are most prevalent in low-income groups where physical activity levels are consistently lower. Social marketing is the government's recommended approach to promoting physical activity but evidence of its effectiveness is limited. The purpose of this study was to examine the effect of a social marketing campaign on the monthly recruitment, attendance and retention levels at a community-based physical activity programme in a low income area. Methods: A six-month social marketing campaign was designed and delivered in a highly-deprived suburban neighbourhood. Analysis of variance was used to assess effects on recruitment and attendance. χ2 tests of independence were used to compare dropouts and adherers and effectiveness of recruitment mechanisms. Percentages were used to compare adherence rates at intervention, pre-existing sessions in the intervention area and control area sessions. Results: Attendance data were collected weekly and presented and analysed monthly to provide a view of changing participation over the six month intervention period, as compared to attendance at pre-existing sessions in the intervention area and in a control area. Recruitment into intervention sessions was significantly greater than into pre-existing and control area sessions in Month 1 (18.13v1.04 p = .007, 18.13v.30 p=.005), Month 5 (3.45v.84 p=.007, 3.45v.30 p<.001) and Month 6 (5.60v.65 p<.001, 5.60v.25 p<.001). Attendance at intervention sessions was significantly greater in all six months than at pre-existing and control area sessions; Month 1 (38.83v7.17 p<.001, 38.83v4.67, p<.001), Month 2 (21.45v6.20 p<.001, 21.45v4.00, p<.001), Month 3 (9.57v6.15 p<.001, 9.57v3.77, p<.001), Month 4 (17.35v7.31 p<.001, 17.35v4.75, p<.001), Month 5 (20.33v8.81 p=.007, 20.33v4.54 p<.001) and Month 6 (28.72v8.28 p<.001, 28.72v.4.00 p<.001). Drop-out rates in the intervention area were similar to the control area (66.2%v69.9%), and considerably lower than in pre-existing sessions (83%). In months one and two, traditional marketing techniques (posters/outdoor banners/flyers) had the greatest influence on recruitment compared to word of mouth communication (84.5%v15.5%). In months five and six word of mouth influenced 57.5% of new recruits. Conclusions: Direct comparisons with other programmes were difficult due to a lack of standard definitions of recruitment and adherence and limited reporting of findings. However when compared to pre-existing sessions and sessions delivered in a control area, monthly attendance patterns indicated that a reasonably well funded social marketing campaign increased recruitment into exercise sessions, maintained good levels of attendance and reasonable levels of adherence. Good attendance levels support on-going campaign success by offering evidence of peer and social support for the activity and increasing opportunities for social interaction. They also increase the capacity and reach of the word of mouth communication channels, the most effective form of promotion. Further study into methods of improving exercise adherence is required.
... Analyzing the impact of these positive and negative online reviews is an important topic for both networking and marketing communities. Product ratings on sites like Amazon typically have a large number of very high and very low scores, which create J -shaped histograms over the rating scale [6]. This is attributed to the "brag-and-moan" phenomenon among reviewers. ...
... IMDb and Rotten Tomatoes' user ratings 6 are often used as a predictors of a movie's quality and box-office potential. With the ready availability of OSN user opinion as poll data, researchers have proposed using pre- 6 For a fair comparison, we exclude scores from movie critics. ...
Article
Data from Online Social Networks (OSNs) are providing analysts with an unprecedented access to public opinion on elections, news, movies etc. However, caution must be taken to determine whether and how much of the opinion extracted from OSN user data is indeed reflective of the opinion of the larger online population. In this work we study this issue in the context of movie reviews on Twitter and compare the opinion of Twitter users with that of the online population of IMDb and Rotten Tomatoes. We introduce new metrics to show that the Twitter users can be characteristically different from general users, both in their rating and their relative preference for Oscar-nominated and non-nominated movies. Additionally, we investigate whether such data can truly predict a movie's box-office success.
... In addition and in contrast to the previous work, corrections, which are necessary due to the " Jshaped distribution " of online reviews, are applied . Reasons and implications of this " J-shaped distribution " are discussed in several publications (Hu et al., 2007; Hu et al., 2009). Online reviews are used for several other research projects, for an overview see (Tang et al., 2009). ...
... This information is then used by Netflix to refine the recommender systems that suggest movies to its customers. An important characteristic of customers' online CIV activity, however, is that these activities usually tend to be polarized at the extremes, with customers typically reporting only highly positive or highly negative experiences (Hu, Pavlou, and Zhang 2007). Moreover, strongly negative online evaluations have a disproportionate impact on others' purchase behavior (Chevalier and Mayzlin 2006). ...
Article
Full-text available
Customers can interact with and create value for firms in a variety of ways. This article proposes that assessing the value of customers based solely upon their transactions with a firm may not be sufficient, and valuing this engagement correctly is crucial in avoiding undervaluation and overvaluation of customers. We propose four components of a customer’s engagement value (CEV) with a firm. The first component is customer lifetime value (the customer’s purchase behavior), the second is customer referral value (as it relates to incentivized referral of new customers), the third is customer influencer value (which includes the customer’s behavior to influence other customers, that is increasing acquisition, retention, and share of wallet through word of mouth of existing customers as well as prospects), and the fourth is customer knowledge value (the value added to the firm by feedback from the customer). CEV provides a comprehensive framework that can ultimately lead to more efficient marketing strategies that enable higher long-term contribution from the customer. Metrics to measure CEV, future research propositions regarding relationships between the four components of CEV are proposed and marketing strategies that can leverage these relationships suggested.
... Fifth, distributions of reviews ratings are assumed to follow a Normal distribution. How the model would behave with non-Normal distributions, and particularly with realistic J-shaped ones (Hu et al., 2007), should be examined. To conclude, we have demonstrated that a model of opinion seeking that combined Bayesian update with a criterion of desired level of confidence can be applied to predict users' depth of opinion search. ...
Article
In this paper we describe a model of how people search online consumer reviews in service of purchasing decisions. The model is similar to other recent models of information seeking in that it updates estimates of products' utilities using Bayesian inference. It is different, in that it stops seeking further information when the confidence that one of the alternatives is the best exceeds a threshold. Findings from a controlled experiment support the model by suggesting that high variance in review ratings causes people to seek more information.
... While a positive bias is frequently found in online review systems, little research has empirically investigated the factors that contribute to this phenomenon. Scholars have suggested that social loafing (Resnick et al., 2000) or self-selection (Hu et al., 2007) contribute to this bias. Drawing from the SOS theory, this study investigated the extent to which fear of social isolation contributes to a positive bias. ...
Article
A positive bias of opinions has frequently been identified across online review websites, suggesting that the public is making decisions based on a limited range of opinions. While scholars often attribute this bias to social loafing or self-selection, this study investigates the positive bias from the perspective of opinion expression. Drawing from the spiral of silence theory, a qualitative analysis of discussion forums reveals that fear of isolation reduces the willingness of members to voice neutral and negative reviews. Additionally, communicative affordances of the interface were found to further suppress neutral and negative opinions. These results extend the spiral of silence theory into the context of non-anonymous multichannel online platforms and indicate the need to consider the role of communicative affordances in online opinion expression.
... Therefore, an increase in ratings can increase the dissimilarity of a shopper from the entire set of previous reviewers (Godes and Silva 2006), which leads to more purchase errors and lower ratings. Consumers with polarized opinions are more likely to write reviews (Hu, Pavlou, and Zhang 2007), and reviews can also be affected by the prices paid by the reviewers which may not be observable to the consumers who utilize these reviews to assist their purchase decisions (Li and Hitt 2010). However, none of these studies explicitly examines how the dynamics of reviews over time might depend on prior reviews, which is the focus of this study. ...
Article
Using book review data on Amazon.com, the authors extend current research into online consumer reviews by empirically investigating the context dependence effect in the review writing process. They find that when product quality remains constant, later reviews tend to differ from previously posted ones, and the difference is moderated by the popularity of the product, the variance of previous reviews, whether later reviews explicitly refer to previous reviews, and the age of the product and the reviews. This phenomenon can be explained by both consumer expectation and self-selection effects in review writing. The implications of this research can help practitioners understand the reviewing process and provide some guidelines for improving the objectivity of online product reviews.
... While a positive bias is frequently found in online review systems, little research has empirically investigated the factors that contribute to this phenomenon. Scholars have suggested that social loafing (Resnick et al., 2000) or self-selection (Hu et al., 2007) contribute to this bias. Drawing from the SOS theory, this study investigated the extent to which fear of social isolation contributes to a positive bias. ...
... In particular, online consumers tend to focus on negative reviews in order to reduce the risk of loss more than enhancing the gain (Kahneman & Tversky, 1979). This strongly supports the notion of negativity bias arguing that rational consumers recognize the purchasing bias, and they compensate for this bias by taking negative reviews more seriously and discounting the positive reviews (Hu, Pavlou, & Zhang, 2007). From the enjoyment aspect, the concept of hedonic consumption with regard to information search process suggests that consumers are likely to consider the excitement and pleasure that accompany purchase (Vogt & Fesenmaier, 1998), which supports the higher influence of positive reviews on inducing perceived enjoyment than negative reviews. ...
Article
Full-text available
Consumers tend to seek heuristic information cues to simplify the amount of information involved in tourist decisions. Accordingly, star ratings in online reviews are a critical heuristic element of the perceived evaluation of online consumer information. The objective of this article is to assess the effect of review ratings on usefulness and enjoyment. The empirical application is carried out on a sample of 5,090 reviews of 45 restaurants in London and New York. The results show that people perceive extreme ratings (positive or negative) as more useful and enjoyable than moderate ratings, giving rise to a U-shaped line, with asymmetric effects: the size of the effect of online reviews depends on whether they are positive or negative.
... Consequently, Li and Hitt (2008) suggest that consumer-generated product reviews may not be an unbiased indication of unobserved product quality. Furthermore, recent work has shown that the distribution of an overwhelming majority of reviews posted in online markets is bimodal (Hu et al. 2008). In such situations, the average numerical star rating assigned to a product may not convey a lot of information to a prospective buyer. ...
Article
Full-text available
Increasingly, user-generated product reviews serve as a valuable source of information for customers making product choices online. The existing literature typically incorporates the impact of product reviews on sales based on numeric variables representing the valence and volume of reviews. In this paper, we posit that the information embedded in product reviews cannot be captured by a single scalar value. Rather, we argue that product reviews are multifaceted, and hence the textual content of product reviews is an important determinant of consumers' choices, over and above the valence and volume of reviews. To demonstrate this, we use text mining to incorporate review text in a consumer choice model by decomposing textual reviews into segments describing different product features. We estimate our model based on a unique data set from Amazon containing sales data and consumer review data for two different groups of products (digital cameras and camcorders) over a 15-month period. We alleviate the problems of data sparsity and of omitted variables by providing two experimental techniques: clustering rare textual opinions based on pointwise mutual information and using externally imposed review semantics. This paper demonstrates how textual data can be used to learn consumers' relative preferences for different product features and also how text can be used for predictive modeling of future changes in sales. This paper was accepted by Ramayya Krishnan, information systems.
... To solve these problems, we proposed to detect anomaly items directly, which is equal to finding out items attacked by fake profiles directly. This is because the basic assumption of an item is that its intrinsic quality follows the uniform distribution [27]; the resulting rating distribution of this item remains stable without attack ratings. Once it changes greatly, the item is definitely considered under attack. ...
Article
Various types of web applications have gained both higher customer satisfaction and more benefits since being successfully armed with personalized recommendation. However, the increasingly rampant shilling attackers apply biased rating profiles to systems to manipulate item recommendations, which not just lower the recommending precision and user satisfaction but also damage the trustworthiness of intermediated transaction platforms and participants. Many studies have offered methods against shilling attacks, especially user profile based-detection. However, this detection suffers from the extraction of the universal feature of attackers, which directly results in poor performance when facing the improved shilling attack types. This paper presents a novel dynamic time interval segmentation technique based item anomaly detection approach to address these problems. In particular, this study is inspired by the common attack features from the standpoint of the item profile, and can detect attacks regardless of the specific attack types. The proposed segmentation technique could confirm the size of the time interval dynamically to group as many consecutive attack ratings together as possible. In addition, apart from effectiveness metrics, little attention has been paid to the robustness of detection methods, which includes measuring both the accuracy and the stability of results. Hence, we introduced a stability metric as a complement for estimating the robustness. Thorough experiments on the MovieLens dataset illustrate the performance of the proposed approach, and justify the value of the proposed approach for online applications.
... First, as in many online reviews contexts, the J-shape distribution of ratings can be partly ascribed to forms of rating behavioral bias. Among these, we believe that under-reporting bias is relevant to historic neighborhoods reviews, which indicates that consumers with moderate evaluations are less likely to report their reviews, compared to highly polarized consumers (Hu, Pavlou, & Zhang, 2014). The latter most commonly materializes as a form of positivity bias, since positive evaluations are more likely to be prevalent overall (Park, Cha, & Rhim, 2018). ...
Article
Full-text available
This paper implements a context-based text mining approach in online reviews of three European historic city districts, to assess the data-driven projection of experiences and major determinants of traveler satisfaction. Bivariate analyses, semantic networks of experience concepts, and word association statistics are used. Results are interpreted in contrast to experience deterioration issues caused by the loss of authenticity, as argued by recent literature. Overtourism signs are a powerful determinant of unfavorable evaluations, but images formed by reviewers are predominantly positive. Nonetheless, although references to authenticity and local identity are not common, destinations failing to project non-touristified aspects are less favored. Overall, evaluations are driven by gratification from offered products and services, as well as passive esthetic consumption of surroundings.
... Consequently, Li and Hitt (2008) suggest that consumer-generated product reviews may not be an unbiased indication of unobserved product quality. Further, recent work has shown that the distribution of an overwhelming majority of reviews posted in online markets is bimodal (Hu et al. 2008). In such situations, the average numerical star rating assigned to a product may not convey a lot of information to a prospective buyer. ...
Article
Full-text available
The increasing pervasiveness of the Internet has dramatically changed the way that consumers shop for goods. Consumer-generated product reviews have become a valuable source of information for customers, who read the reviews and decide whether to buy the product based on the information provided. In this paper, we use techniques that decompose the reviews into segments that evaluate the individual characteristics of a product (e.g., image quality and battery life for a digital camera). Then, as a major contribution of this paper, we adapt methods from the econometrics literature, specifically the hedonic regression concept, to estimate: (a) the weight that customers place on each individual product feature, (b) the implicit evaluation score that customers assign to each feature, and (c) how these evaluations affect the revenue for a given product. Towards this goal, we develop a novel hybrid technique combining text mining and econometrics that models consumer product reviews as elements in a tensor product of feature and evaluation spaces. We then impute the quantitative impact of consumer reviews on product demand as a linear functional from this tensor product space. We demonstrate how to use a low-dimension approximation of this functional to significantly reduce the number of model parameters, while still providing good experimental results. We evaluate our technique using a data set from Amazon.com consisting of sales data and the related consumer reviews posted over a 15-month period for 242 products. Our experimental evaluation shows that we can extract actionable business intelligence from the data and better understand the customer preferences and actions. We also show that the textual portion of the reviews can improve product sales prediction compared to a baseline technique that simply relies on numeric data.
... Specifically, online consumers tend to focus on negative reviews in order to increase the utility of their decisions by reducing the risk of loss (Kahneman & Tversky, 1979). This strongly supports the notion of negativity bias, arguing that rational consumers recognise the purchasing bias, and they compensate for this bias by considering negative reviews more seriously than positive reviews (Hu, Pavlou, & Zhang, 2007). From the enjoyment aspect, the characteristics of tourism products, which refer to experiential (or hedonic) products, suggest that consumers tend to take into account the elements of excitement and pleasure when searching for travel information (Vogt & Fesenmaier, 1998). ...
Chapter
Full-text available
This study estimates the effect of online consumers’ star ratings on perceived evaluations of consumer reviews such as usefulness and enjoyment. The data includes 5090 online reviews of about 45 restaurants located in London and New York respectively. The results reveal curvilinear (U-shaped) relationships between star ratings and usefulness and enjoyment. That is, online consumers perceive extreme ratings (positive or negative) as more useful and enjoyable than moderate ratings. Additionally, the findings of this research indicate the usefulness of the negative binomial model, which allows researchers to manage the features of count data as well as address the heteroscedasticity in linear regression and the overdispersion problem in the Poisson regression model.
... However, the profit of preannounced pricing policy does not increase monotonically with ; when is too high, the profit begins to decrease. Park and Nicolau [37], Kahneman and Tversky [41], and Hu, Pavlou, and Zhang [42] suggest that rational consumers tend to focus on negative reviews more seriously. While positive reviews satisfy people's enjoyment, they make little impact on usefulness. ...
Article
Full-text available
Crowdfunding marks a popular and sustainable means by which small and microentrepreneurs obtain financial resources for their innovative project. Consumers increasingly rely on online reviews to make purchase decisions. However, the crowdfunding nowadays lacks a form type of review system. This paper is designed to extend research on the optimal pricing decision with review system for the reward-based crowdfunding. Firstly, a Bayesian analysis is established to construct consumers’ belief update process in presence of review system. Secondly, we take the strategies without the review system as a benchmark to explore the impacts of review system under preannounced pricing and responsive pricing. Finally, through the equilibrium analysis, we find that the review system has a positive impact on the creator under responsive pricing policy. The fraction of favorable review has a large effect on the profit of preannounced pricing. When the fraction is about 80%, the profit is the maximum. Generally speaking, the review system will make more profit for the creator.
... Following this, a variety of technological initiatives by retailers have been geared towards providing a facilitative shopping experience for customers. These interactive technologies played a key role in a retail context and enable retailers with a full range of information capabilities and better customer orientation and online product review with product sales (Hu et al., 2009;Zhu & Kraemer, 2005). These technologies may have a direct or indirect impact on the strategies of the retailers, that is, facial tracking technology provides a subtle competitive advantage (Varadarajan et al., 2010). ...
Article
Full-text available
Purpose: The major intent of this study is to empirically explore the technology–organization– environment (TOE) factors that influence retail outlets’ adoption of technology applications in India. Methodology: In line with the objectives of the study, a conceptual model based on the TOE framework was proposed. A quantitative survey was conducted with organized retail outlets in Madhya Pradesh, India to validate the proposed model. The survey generated 134 valid responses. Partial least squares-structural equation modeling (PLS-SEM) was used to assess the research model. Findings: The findings elucidated that factors of the TOE framework significantly influence the technology adoption intention in the organized retail sector. It could be recommended that this framework is a robust instrument to envisage the adoption of technology applications in the retail sector. Research implications: The conceptual framework can be used by the retail sector in framing strategies for marketing that can enhance their potential customer base. It can also help to formulate policies which can act as a catalyst in adopting the technology applications. Practical implications: Retail vendors can use this model to increase their prospective customer base. Originality: The research contributes to the diversifying academic literature on technology application adoption by using factors within the TOE framework to explain organized retail outlets’ intention to adopt such technologies. Keywords - Organized retailers, TOE framework, Technology applications, Technology adoption intention
... We interpret this finding to question the typical positivity bias for online reviews (Hu et al., 2009). While a four-star rating may intuitively seem 'positive', or a three-star rating 'neutral', the voices of the platform members themselves tell a different story: They overuse negative sentiment for anything other than the full five-star rating. ...
Article
Full-text available
Prominent among the social developments that the web 2.0 has facilitated is digital social reading (DSR): on many platforms there are functionalities for creating book reviews, ‘inline’ commenting on book texts, online story writing (often in the form of fanfiction), informal book discussions, book vlogs, and more. In this article, we argue that DSR offers unique possibilities for research into literature, reading, the impact of reading and literary communication. We also claim that in this context computational tools are especially relevant, making DSR a field particularly suitable for the application of Digital Humanities methods. We draw up an initial categorization of research aspects of DSR and briefly examine literature for each category. We distinguish between studies on DSR that use it as a lens to study wider processes of literary exchange as opposed to studies for which the DSR culture is a phenomenon interesting in its own right. Via seven examples of DSR research, we discuss the chosen approaches and their connection to research questions in literary studies.
... The findings explain that review information has a significant effect on consumers' purchase decisions (e.g., Chevalier & Mayzlin, 2006;Mayzlin, 2006). Hu, Pavlou, and Zhang (2009) explained that most product reviews showed a J-shaped distribution, and most were five-star ratings with some one-star ratings. They asserted that people tend to rate products on the extreme if they are expressing biased views more than those who expressed a moderate view. ...
... One concern in our estimation is that the individual properties' online popularity may be underestimated by the volume of online reviews, as online reviews are primarily written by consumers who have polarized experiences (either terrible or excellent); customers with neutral experiences tend not to write reviews. This polarized distribution of online review valence is the so-called "J-shaped" reviews effect (Hu, Pavlou, & Zhang, 2007). As the missing neutral reviews are not accounted for in measurements of the online popularity of a property based on online review volume, our estimation may be biased. ...
Article
Full-text available
Commercial hosts are becoming increasingly common in peer-to-peer accommodation sharing, but the interplay between commercial and individual hosts is less identified. This study investigates the effect of properties managed by commercial hosts on the individual hosts in the neighborhood. Specifically, we hypothesize that an increase in commercial properties, which have competitive advantages, would penetrate neighborhood markets and cannibalize the online popularity of individual properties. We test these hypotheses using a large-scale and longitudinal dataset collected from a leading peer-to-peer accommodation sharing platform in Beijing. The findings show that an increase of commercial properties is associated with a decrease in the popularity of individual properties in the neighborhood. However, the negative effect of commercial properties is weakened where there is a higher price difference between two ownership types and a higher density of tourist attractions. The implications on service operations, pricing, and locational strategies for peer-to-peer accommodation sharing businesses are discussed. KEYWORDS: peer-to-peer accommodation sharing; commercial property; individual property; online popularity of properties; China
... Park and Nicolau (2015) show that extreme ratings (positive and negative) are more useful to customers than moderate ratings, and that the effect of reviews in the tourism sector is asymmetric: negative reviews are perceived as more useful as the aim of reducing losses is more salient than increasing gains, in line with Kahneman and Tversky (1979). This applies when rational consumers operate in contexts of uncertainty and risk (Hu et al., 2007), whereas positive reviews are crucial for enjoyment aspects and purchasing decisions. The impact of eWOM on sales and consumer decisions is also moderated by the life-cycle stage and by specific characteristics of the product (You et al., 2015). ...
Purpose This study aims to compare the rating dynamics of the same hotels in two online review platforms (Booking.com and Trip Advisor), which mainly differ in requiring or not requiring proof of prior reservation before posting a review (respectively, a verified vs a non-verified platform). Design/methodology/approach A verified system, by definition, cannot host fake reviews. Should also the non-verified system be free from “ambiguous” reviews, the structure of ratings (valence, variability, dynamics) for the same items should also be similar. Any detected structural difference, on the contrary, might be linked to a possible review bias. Findings Travelers’ scores in the non-verified platform are higher and much more volatile than ratings in the verified platform. Additionally, the verified review system presents a faster convergence of ratings towards the long-term scores of individual hotels, whereas the non-verified system shows much more discordance in the early phases of the review window. Research limitations/implications The paper offers insights into how to detect suspicious reviews. Non-verified platforms should add indices of scores’ dispersion to existing information available in websites and mobile apps. Moreover, they can use time windows to delete older (and more likely biased) reviews. Findings also ring a warning bell to tourists about the reliability of ratings, particularly when only a few reviews are posted online. Originality/value The across-platform comparison of single items (in terms of ratings’ dynamics and speed of convergence) is a novel contribution that calls for extending the analysis to different destinations and types of platform.
... The findings explain that review information has a significant effect on consumers' purchase decisions (e.g., Chevalier & Mayzlin, 2006;Mayzlin, 2006). Hu, Pavlou, and Zhang (2009) explained that most product reviews showed a J-shaped distribution, and most were five-star ratings with some one-star ratings. They asserted that people tend to rate products on the extreme if they are expressing biased views more than those who expressed a moderate view. ...
Article
This study examined generational discrepancies among Indian Gen X and Millennial consumers' attitude toward online star ratings. The results show online star rating involvement has an impact on perceived ease of use and perceived usefulness of star ratings as well as the intention to use online ratings for both generations. However, Millennial consumers find star ratings to be user-friendly, but there is no effect on their attitude toward star ratings. The results also indicate that high star ratings have an effect on Gen Xers' attitudes, while Millennials find that low star ratings have an impact on their attitudes. In addition, Gen Xers rely on low star ratings while Millennials prefer to use high star ratings for intention to use. For managerial suggestions, e-commerce retailers may focus on a product quality indicator such as star ratings for Millennials. However, for Gen Xers, more descriptive information can be more effective.
... In general, people react more strongly to negative information than to positive (Baumeister et al. 2001). However, a majority of online reviews are favorable (Hu et al. 2009). Thus, negative reviews are considered Brarer^and salient (Feldman 1966;Kanouse and Hanson 1972). ...
Article
Full-text available
This research explores the effects of profanity on perceptions of online reviews’ usefulness. Based on the prior research that described asymmetric attribution effects, we predict that when consumers see profanity in an online review, their reactions differ depending on the valence of the review. Specifically, when profanity is used in a negative review, it should reduce review usefulness because of decreased perceived reviewer objectivity. Among positive reviews, profanity increases review usefulness through greater perceived reviewer credibility. Through analysis of Yelp data, we show that the effect of profanity on usefulness depends on review valence. Experimentally, we demonstrate the opposing mediating effects of perceived objectivity and credibility on the usefulness of the review.
Conference Paper
As the new platform in e-commerce, online reviews change the purchase means of Internet consumers. Therefore, it is important to analyze the influence of online reviews on consumer purchase decisions in order to enhance the penetration ratio of Internet consumers in internet users of China. To further verify whether the product's average scores can reveal the true quality of products, using empirical analysis and mathematical modeling methods, the paper makes a statistical description for the data about business war stories, home video and the bar phone on Amazon during from January to February of 2010. Finally, we found average score of online reviews may mislead the consumer purchasing decisions and give the rationalization proposals to Internet retailers and consumers, while noting the contributions and deficiencies of this article, and discusses the future research directions.
Conference Paper
As an emerging platform for online shopping, online review changed the network consumer's purchase mode. In this paper, we study the effect of potential online purchaser influenced by the online reviews, by means of the questionnaire. The final study shows that: the Internet consumers when making purchase decisions are mostly affected by the average scores of products; and not significantly affected by the following three factors: the proportion of comment buyers accounted for all buyers, the sooner or later of online reviews and whether the contents reflecting the latest product information or not.
Article
A growing reliance on customer reviews prompts firms to develop strategies to encourage customers to post online reviews of their products. However, little research investigates the behavioral consequences of writing a review. The act of sharing personal opinions through reviews is a rewarding experience and makes customers feel socially connected. With an application of reverse alliesthesia theory, the current study predicts that such rewarding experiences drive online reviewers to seek other rewards, such as impulsive buying. Three lab-based and two field studies demonstrate such an emotional review–reward effect: sharing emotional information in the public realm of customer reviews, rather than forming similar opinions privately, drives participants to make more impulsive buying decisions.
Conference Paper
In this paper we present an approach to generate lists of opinion bearing phrases with their opinion values in a continuous range between -- 1 and 1. Opinion phrases that are considered include single adjectives as well as adjective-based phrases with an arbitrary number of words. The opinion values are derived from user review titles and star ratings, as both can be regarded as summaries of the user's opinion about the product under review. Phrases are organized in trees with the opinion bearing adjective as tree root. For trees with missing branches, opinion values then can be calculated using trees with similar branches but different roots. An example list is produced and compared to existing opinion lists.
Article
There has been a growing use of social media by patients to share their healthcare experiences and produce information that can be helpful to other patients seeking healthcare services. These stories can reveal issues in healthcare quality. However, faced with the inherent risks of social media, healthcare providers have been skeptical about the value of these stories, and many healthcare systems have adopted restrictive and protective policies to control the use of social media by healthcare providers. This study explores healthcare providers' and administrators' perspectives on patient stories on social media and whether they can use the stories to evaluate healthcare experiences. Semi-structured interviews (n = 21) were conducted with healthcare providers and administrators, including physicians, nurses, and quality managers in Ontario, Canada, between April 2018 and May 2019. Inductive and data-driven thematic analysis was used to analyze the data. Several barriers prevent healthcare providers from realizing the benefits of social media, including concerns about the quality of patients' feedback, the professional codes of conduct, and the time and effort required to process these stories. The study findings suggest that cultural changes in the healthcare system might be required to foster the use of social media for healthcare quality improvement and enable the development of a safe patient-provider communication environment that facilitates the exchange of constructive feedback between the two parties without the fear of legal consequences, breaches of patient privacy, or violation of professional codes of conduct.
Article
This study investigates the successful creation of specialized books in management literature, by focusing on consumer demand for expert knowledge in business topics. Applying a sales-ranks conversion model based on a sample of 60,000 books from Amazon.de, I test hypotheses on the effects of (1) author characteristics (professional background, reputation, experience, demographics), (2) book characteristics (publisher, price, page count, reading excerpts, cover), and (3) promotional activities (recommendations, reviews, discussions) on book sales for a sample of 1,435 management books. Data were collected from the Amazon website using Perl-based scripts. The regression results demonstrate that demand for management knowledge depends on a mix of these factors; yet, success is largely determined by author characteristics. In fact, the market’s perception of the author alters consumers’ price sensitivity and renders observable book characteristics insignificant; promotional activities amplify this effect. The paper shows how to target demand for management knowledge more effectively.
Article
This paper uses online users' reviews and sales information from May 2003 to March 2007 to study a new experiential product: online video game (or virtual world). The results suggest that, first of all, for a hedonic product such as an online video game, the availability of online review system could not fully transform experience attributes into search attributes. The empirical study confirms that negative word-of-mouth has more significant impact than positive reviews. The study empirically verifies that the existence of direct network externalities (i.e. a product's value increases when more consumers join the network) is crucial for a hedonic product. The size of the user base signals the quality of the game, and works like a search attribute for potential users. A large user base will reduce the negative impact from unfavorable reviews.
Conference Paper
Online product reviews sometimes display a consistent, positive bias. Prior research has largely focused on reviewer honesty but other possible explanations are lack of product knowledge and self-selection bias. A semantic Web method is proposed to deliver more accurate reviews to online customers. An agent-based model is used to show that such a method can significantly improve information delivery even with noisy knowledge data.
Article
This paper investigates whether firms' responsiveness to customer reviews affects firms' online reputation. Responsiveness is measured by the intensity (fraction) of reviews that are responded to whereas online reputation is measured using TripAdvisor's average customer review rating (1–5 scale). Our analysis is applied to the hotel market in Manhattan (New York). To deal with the endogeneity of a hotel's responsiveness, we instrument it with the responsiveness level displayed by nearby competitors. This identification strategy is motivated by the fact that hotels have a greater tendency to respond to reviews not only because a particular review demands attention, but also because they seek to comply (catch up) with their competitors' level of responsiveness. The results show that one standard deviation increase in a hotel's responsiveness level would result in an improvement of 0.055 stars in TripAdvisor's average online rating (equivalently, an increase of 0.09 SD). Importantly, not accounting for endogeneity would lead to an erroneous conclusion that such effect is nonexistent. In addition, our results show that the effect is heterogeneous. Specifically, the effect is particularly strong for: (a) hotels responding more intensely to negative reviews, (b) independent hotels (vis‐à‐vis chain hotels), (c) hotels with a less established online reputation, (d) hotels with more volatile online ratings, and (e) hotels with more experience responding to negative reviews. We discuss the possible mechanism between firms' responsiveness and online reputation.
Chapter
Der Begriff Customer Engagement gewinnt in Wissenschaft und Praxis gleichermaßen zunehmend an Bedeutung. Er beschreibt den integrativen Ansatz zum Aufbau und Ausbau einer wechselseitigen Beziehung zum Kunden, welche maßgeblich in der Einbindung des Kunden in Unternehmensaktivtäten besteht. Der Customer Engagement Value stellt dabei einen wichtigen Bestandteil zur Quantifizierung dieses Ansatzes dar. Dabei liefert er ein Vorgehen zur monetären Bewertung transaktionaler (Kauf von Gütern oder Dienstleistungen) sowie nichttransaktionaler (Empfehlungen, Einfluss auf andere Kunden und Feedback) Wertbeiträge von Kunden. Er kann somit als Maß zur Einschätzung der Attraktivität von Kunden sowie als Entscheidungsgrundlage für die Ausgestaltung integrativer Marketingmaßnahmen dienen. Dieses Kapitel vermittelt eine ganzheitliche Perspektive des Customer Engagements Values, welche den Weg der Konzeptualisierung über die formale Operationalisierung bis hin zu Beispielen für die Einbindung des Customer Engagements in Marketinginitiativen beschreibt. Dabei stellt der Customer Engagement Value ein wertorientiertes Konzept dar, welches als monetäre Kenngröße mit vier eigenständigen Komponenten formuliert wird: dem Customer Lifetime Value, Customer Referral Value, Customer Influencer Value und dem Customer Knowledge Value.
Article
Full-text available
The extremely short life cycle and the rapid decay in revenues after opening coupled with the rapid and frequent introduction of new competitive products makes the timing of new product introductions in the motion picture industry critical, particularly during the high-revenue Christmas and summer seasons. Each studio wants to capture as much of the season as possible by opening early in the season. At the same time, each wants to avoid head-to-head competition. The authors model competition between two motion pictures in a share attraction framework and conduct an equilibrium analysis of the product introduction timing game in a finite season. The following three different equilibrium configurations emerge: (1) a single equilibrium with both movies opening simultaneously at the beginning of the season, (2) a single equilibrium with one movie opening at the beginning of the season and one delaying, and (3) dual equilibria, with either movie delaying opening. A key factor is the product life cycle, which can be captured well with a two-parameter exponential decline. The authors relate the life-cycle parameters to these possibilities with the general result that the weaker movie may be forced to delay opening. These results are related to case studies of the opening of recently released movies. A statistical analysis of the 1990 summer season in North America provides support for the conclusions and suggests that current release timing decisions can be improved. The authors discuss the rationale of "avoiding the competition" in the general context of product introduction timing.
Article
Full-text available
Critics and their reviews pervade many industries and are particularly important in the entertainment industry. Few marketing scholars, however, have considered the relationship between the market performance of entertainment services and the role of critics. The authors do so here. They show empirically that critical reviews correlate with late and cumulative box office receipts but do not have a significant correlation with early box office receipts. Although still far from any definitive conclusion, this finding suggests that critics, at least from an aggregate-level perspective, appear to act more as leading indicators than as opinion leaders.
Article
Full-text available
Marketing practitioners and theorists routinely cite the power of the personal referral on customer behaviour. However, relatively few companies have tried to harness the power of word of mouth (WOM). Scholars have been pondering WOM over 2400 years, although modern marketing research into WOM started only relatively recently, in the post-war 1940s. WOM can be characterized by valence, focus, timing, solicitation and degree of management intervention. Most recent WOM research has been conducted from a customer-to-customer perspective, even though WOM is found in other contexts such as influence, employee and recruitment markets. Marketing research into WOM has attempted to answer two questions. What are the antecedents of WOM? What are the consequences of WOM? This paper integrates that research into a contingency model and attempts to identify researchable gaps in our knowledge.
Article
Full-text available
W e present a framework and empirical estimates that quantify the economic impact of increased product variety made available through electronic markets. While efficiency gains from increased competition significantly enhance consumer surplus, for instance, by leading to lower average selling prices, our present research shows that increased product variety made available through electronic markets can be a significantly larger source of consumer surplus gains.. One reason for increased product variety on the Internet is the ability of online retailers to catalog, recommend, and provide a large number of products for sale. For example, the number of book titles available at Amazon.com is more than 23 times larger than the number of books on the shelves of a typical Barnes & Noble superstore, and 57 times greater than the number of books stocked in a typical large independent bookstore. Our analysis indicates that the increased product variety of online bookstores enhanced consumer welfare by $731 million to $1.03 billion in the year 2000, which is between 7 and 10 times as large as the consumer welfare gain from increased competition and lower prices in this market. There may also be large welfare gains in other SKU-intensive consumer goods such as music, movies, consumer electronics, and computer software and hardware.
Article
Full-text available
The authors investigate how critics affect the box office performance of films and how the effects may be moder- ated by stars and budgets. The authors examine the process through which critics affect box office revenue, that is, whether they influence the decision of the film going public (their role as influencers), merely predict the deci- sion (their role as predictors), or do both. They find that both positive and negative reviews are correlated with weekly box office revenue over an eight-week period, suggesting that critics play a dual role: They can influence and predict box office revenue. However, the authors find the impact of negative reviews (but not positive reviews) to diminish over time, a pattern that is more consistent with critics' role as influencers. The authors then compare the positive impact of good reviews with the negative impact of bad reviews to find that film reviews evidence a negativity bias; that is, negative reviews hurt performance more than positive reviews help performance, but only during the first week of a film's run. Finally, the authors examine two key moderators of critical reviews, stars and budgets, and find that popular stars and big budgets enhance box office revenue for films that receive more neg- ative critical reviews than positive critical reviews but do little for films that receive more positive reviews than neg- ative reviews. Taken together, the findings not only replicate and extend prior research on critical reviews and box office performance but also offer insight into how film studios can strategically manage the review process to enhance box office revenue.
Article
Full-text available
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
Full-text available
While marketing scholars have emphasized the importance of customer satisfaction, few studies have examined in detail consumers' responses to dissatisfaction. This study examines correlates of one possible response-telling others about the dissatisfaction-and identifies variables that distinguish this response from others. Variables investigated include the nature of the dissatisfaction, perceptions of blame for the dissatisfaction, and perceptions of retailer responsiveness. Marketing management and consumer behavior research implications are discussed.
Article
Full-text available
The use of the WWW as a venue for voicing opinions, complaints and recommendations on products and firms has been widely reported in the popular media. However little is known how consumers use these reviews and if they subsequently have any influence on evaluations and purchase intentions of products and retailers. This study examines the effect of negative reviews on retailer evaluation and patronage intention given that the consumer has already made a product/brand decision. Our results indicate that the extent of WOM search depends on the consumer's reasons for choosing an online retailer. Further the influence of negative WOM information on perceived reliability and purchase intentions is determined largely by familiarity with the retailer and differs based on whether the retailer is a pure-Internet or clicks-and-mortar firm. Managerial implications for positioning strategies to minimize the effect of negative word-of-mouth have been discussed.
Book
Full-text available
Offering a unifying theoretical perspective not readily available in any other text, this innovative guide to econometrics uses simple geometrical arguments to develop students' intuitive understanding of basic and advanced topics, emphasizing throughout the practical applications of modern theory and nonlinear techniques of estimation. One theme of the text is the use of artificial regressions for estimation, reference, and specification testing of nonlinear models, including diagnostic tests for parameter constancy, serial correlation, heteroscedasticity, and other types of mis-specification. Explaining how estimates can be obtained and tests can be carried out, the authors go beyond a mere algebraic description to one that can be easily translated into the commands of a standard econometric software package. Covering an unprecedented range of problems with a consistent emphasis on those that arise in applied work, this accessible and coherent guide to the most vital topics in econometrics today is indispensable for advanced students of econometrics and students of statistics interested in regression and related topics. It will also suit practising econometricians who want to update their skills. Flexibly designed to accommodate a variety of course levels, it offers both complete coverage of the basic material and separate chapters on areas of specialized interest.
Article
Full-text available
Product reviews by third parties are growing in popularity. This paper examines when and how a manufacturing firm should adapt its marketing strategies to such reviews. For example, should a firm receiving an unfavorable review reduce its price or adjust its advertising? Should a winning product of a product review (e.g., “editor’s choice”) boost its advertising expenditure to spread the good news? How should firms’ strategic responses to product reviews differ across different types of product reviews (description vs. recommendation) and different advertising media (the reviewer’s publication vs. other media)? We develop a theory to address these issues and derive firms’ optimal responses to product reviews under different product/market/review/media conditions. We show that firms should choose rather than as a strategic variable in response to product reviews when enough consumers value horizontal product attributes. Surprisingly, we find that using a review-endorsed (i.e., advertisements containing third-party award logos) to broadcast its victory can hurt the winning product of a product review. Also, it is not necessarily wise for the winning products to boost to spread the good news. Data from two industries—printers and running shoes—are used to illustrate some of our findings.
Article
Full-text available
Optimal Reward Programs Sellers who plan to capitalize on the lifetime value of customers need to manage the sales potential from customer referrals proactively. To encourage existing customers to generate referrals, a seller can offer exceptional value to current customers through either excellent quality or a very attractive price. Rewards to customers for referring other customers can also encourage referrals. We investigate when referral rewards should be offered to motivate referrals, and derive the optimal combination of reward and price that will lead to the most profitable referrals. We define a delighted customer as one who obtains a positive level of surplus above a threshold level, and consequently recommends the product to another customer. We show that the use of referral rewards depends on how demanding consumers are before they are willing to recommend (i.e., on the delight threshold level). The optimal mix of price and referral reward falls into three regions: 1. When customers are easy to delight, the optimal strategy is to lower the price below that of a seller who ignores the
Article
Full-text available
In spite of the high financial stakes involved in marketing new motion pictures, marketing science models have not been applied to the market evaluation of motion pictures. The motion picture industry poses some unique challenges. For example, the consumer adoption process for movies is very sensitive to word-of-mouth interactions, which are difficult to measure and predict the movie has been released. In this article, we undertake the challenge to develop and implement MOVIEMOD—a prerelease market evaluation model for the motion picture industry. MOVIEMOD is designed to generate box-office forecasts and to support marketing decisions for a new movie after the movie has been produced (or when it is available in a rough cut) but before it has been released. Unlike other forecasting models for motion pictures, the calibration of MOVIEMOD does not require any actual sales data. Also, the data collection time for a product with a limited lifetime such as a movie should not take too long. For MOVIEMOD it takes only three hours in a “consumer clinic” to collect the data needed for the prediction of box-office sales and the evaluation of alternative marketing plans. The model is based on a behavioral representation of the consumer adoption process for movies as a macroflow process. The heart of MOVIEMOD is an interactive Markov chain model describing the macro-flow process. According to this model, at any point in time with respect to the movie under study, a consumer can be found in one of the following behavioral states: undecided, considerer, rejecter, positive spreader, negative spreader, and inactive. The progression of consumers through the behavioral states depends on a set of factors that are related to the marketing mix, as well as on a set of more general factors that characterize the movie-going behavior in the population of interest. This interactive Markov chain model allows us to account for word-of-mouth interactions among potential adopters and several types of word-of-mouth spreaders in the population. Marketing variables that influence the transitions among the states are movie theme acceptability, promotion strategy, distribution strategy, and the movie experience. The model is calibrated in a consumer clinic experiment. Respondents fill out a questionnaire with general items related to their movie-going and movie communication behavior, they are exposed to different sets of information stimuli, they are actually shown the movie, and finally, they fill outpostmovie evaluations, including word-of-mouth intentions.These measures are used to estimate the word-of-mouth parameters and other behavioral factors, as well as the movie-specific parameters of the model. MOVIEMOD produces forecasts of the awareness, adoption intention, and cumulative penetration for a new movie within the population of interest for a given base marketing plan. It also provides diagnostic information on the likely impact of alternative marketing plans on the commercial performance of a new movie. We describe two applications of MOVIEMOD: One is a pilot study conducted without studio cooperation in the United States, and the other is a full-fledged implementation conducted with cooperation of the movie's distributor and exhibitor in the Netherlands. The implementations suggest that MOVIEMOD produces reasonably accurate forecasts of box-office performance. More importantly, the model offers the opportunity to simulate the effects of alternative marketing plans. In the Dutch application, the effects of extra advertising, extra magazine articles, extra TV commercials, and higher trailer intensity (compared to the base marketing plan of the distributor) were analyzed. We demonstrate the value of these decision-support capabilities of MOVIEMOD in assisting managers to identify a final plan that resulted in an almost 50% increase in the test movie's revenue performance, compared to the marketing plan initially contemplated. Management implemented this recommended plan, which resulted in box-office sales that were within 5% of the MOVIEMOD prediction. MOVIEMOD was also tested against several benchmark models, and its prediction was better in all cases. An evaluation of MOVIEMOD jointly by the Dutch exhibitor and the distributor showed that both parties were positive about and appreciated its performance as a decision-support tool. In particular, the distributor, who has more stakes in the domestic performance of its movies, showed a great interest in using MOVIEMOD for subsequent evaluations of new movies prior to their release. Based on such evaluations and the initial validation results, MOVIEMOD can fruitfully (and inexpensively) be used to provide researchers and managers with a deeper understanding of the factors that drive audience response to new motion pictures, and it can be instrumental in developing other decision-support systems that can improve the odds of commercial success of new experiential products.
Article
Full-text available
This study investigates consumers’ usage of online recommendation sources and their influence on online product choices. A 3 (websites) × 4 (recommendation sources) × 2 (products) online experiment was conducted with 487 subjects. Results indicate that subjects who consulted product recommendations selected recommended products twice as often as subjects who did not consult recommendations. The online recommendation source labeled “recommender system,” typical of the personalization possibilities offered by online retailing, was more influential than more traditional recommendation sources such as “human experts” and “other consumers”. The type of product also had a significant influence on the propensity to follow product recommendations. Theoretical and managerial implications of these findings are provided.
Article
Full-text available
An attribution model of information processing is proposed and experimentally tested to explain the alleged disproportionate weighting of unfavorable product information. The findings of the experiment generally support hypotheses proposing that unfavorable ratings, as compared to favorable product ratings on the same attributes, prompt significantly stronger attributions to product performance, belief strength, and affect toward products.
Article
Full-text available
The effects of word-of-mouth (WOM) communications and specific attribute information on product evaluations were investigated. A face-to-face WOM communication was more persuasive than a printed format (experiment 1). Although a strong WOM effect was found, this effect was reduced or eliminated when a prior impression of the target brand was available from memory or when extremely negative attribute information was presented (experiment 2). The results suggest that diverse, seemingly unrelated judgmental phenomena--such as the vividness effect, the perseverance effect, and the negativity effect--can be explained through the accessibility-diagnosticity model. Copyright 1991 by the University of Chicago.
Article
Plots, probability plots and regression tests tests using moments other tests for univariate normality goodness of fit tests testing for outliers in univariate samples power comparisons for univariate tests for normality testing for normalitywith censored data assessing multivariate normality testing for multivariate outliers testing for normal mixtures robust methods computational methods and issues. Appendices: data sets used in examples critical values for tests.
Article
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Article
When a population is assumed to be composed of a finite number of subpopulations, a natural model to choose is the finite mixture model. It will often be the case, however, that the number of component distributions is unknown and must be estimated. This problem can be difficult; for instance, the density of two mixed normals is not bimodal unless the means are separated by at least 2 standard deviations. Hence modality of the data per se can be an insensitive approach to component estimation. We demonstrate that a mixture of two normals divided by a normal density having the same mean and variance as the mixed density is always bimodal. This analytic result and other related results form the basis for a diagnostic and a test for the number of components in a mixture of normals. The density is estimated using a kernel density estimator. Under the null hypothesis, the proposed diagnostic can be approximated by a stationary Gaussian process. Under the alternative hypothesis, components in the mixture will express themselves as major modes in the diagnostic plot. A test for mixing is based on the amount of smoothing necessary to suppress these large deviations from a Gaussian process.
Article
Abstract ¶ Despite the widespread popularity of online opinion forums among consumers, the business value that such systems bring to organizations has, so far, remained an unanswered question. This paper addresses this question by studying the value of online movie ratings in forecasting
Article
This paper uses detailed weekly data on sales of hardcover fiction books to evaluate the impact of New York Times book reviews on sales. In order to weigh the relative propensity of reviews to inform and to persuade, the analysis utilizes a measure of review opinion obtained through a systematic reading of each review. The estimates indicate that in the case of book reviews, any publicity is good publicity: even negative reviews lead to increases in sales. We interpret this finding as evidence that book reviews serve largely to inform consumers about books' content and characteristics (including the books' existence). However, positive reviews have a larger impact on sales than negative reviews, suggesting that reviews also have a persuasive effect.
Article
Finite mixture models are a useful class of models for application to data. When sample sizes are not large and the number of underlying densities is in question, likelihood ratio tests based on joint maximum likelihood estimation of the mixing parameter, θ, and the parameter of the underlying densities, λ, are problematical. Our approach places a prior distribution on λ and estimates θ by maximizing the likelihood of the data given θ with λ integrated out. Advantages of this approach, computational issues using the EM algorithm and directions for further work are discussed. The technique is applied to two examples.
Article
This article offers a practical guide to goodness-of-fit tests using statistics based on the empirical distribution function (EDF). Five of the leading statistics are examined—those often labelled D, W , V, U , A —and three important situations: where the hypothesized distribution F(x) is completely specified and where F(x) represents the normal or exponential distribution with one or more parameters to be estimated from the data. EDF statistics are easily calculated, and the tests require only one line of significance points for each situation. They are also shown to be competitive in terms of power.
Article
This article reports an experiment designed to investigate the short-term sales effects of product-related conversations. The results show that exposure to favorable comments aids acceptance of a new product, while unfavorable comments hinder it.
Article
Online reputation mechanisms are emerging as a promising alternative to more traditional trust building mechanisms, such as branding and formal contracting, in settings where the latter may be ineffective or prohibitively expensive; a lot of electronic trading communities fall under these categories. Although a number of commercial websites already employ various forms of reputation mechanisms, rigorous research into their properties is still in its infancy. This fledgling field can benefit from past results in economics and game theory. Moreover, in order to translate the stylized results of game theory into concrete managerial guidance for implementing and participating in effective reputation mechanisms further advances are needed in a number of important areas: First, the design space of such mechanisms needs to be scoped and the effects of different design choices on performance need to be better understood. Second, the economic efficiency of various classes of reputation mechanisms needs to be quantified and compared to that of alternative mechanisms for building trust. Third, the robustness of those mechanisms against boundedly rational players, noisy ratings and strategic manipulation needs to be studied and improved. This paper surveys past results that have been derived in a variety of contexts, but which are relevant as a basis for building online reputation systems, presents two analytical models that illustrate the role of such systems in electronic markets and identifies opportunities for further MS/OR research in this fascinating area.
Article
We examine three tests, widely employed in the statistics literature, that help determine the number of modes in an empirical nonparametric kernel density. The tests are described and applied to the Penn World Table (PWT) data set on world-wide per capita income to study convergence. Our results suggest that the distribution of world-wide per capita income, contrary to conventional wisdom based on informal inspection, may have been multimodal throughout the period covered by the PWT data set.
Article
This paper uses actual WOM information to examine the dynamic patterns of WOM and how it helps explain box office revenue. The WOM data was collected from the website of Yahoo! Movies. The results show that WOM activities are the most active during pre-release and the opening week, and that movie audiences tend to hold relatively high expectation before release, but become more critical in the opening week. More importantly, WOM information offers significant explanatory power for both aggregate and weekly box office revenue, especially in the early weeks after opening. However, most of this explanatory power comes from the volume of WOM, not its valence as measured by the percentages of positive and negative messages.
Article
Nonparametric tests of modality are a distribution-free way of assessing evidence about inhomogeneity in a population, provided that the potential sub populations are sufficiently well separated. They include the excess mass and dip tests, which are equivalent in univariate settings and are alternatives to the bandwidth test. Only very conservative forms of the excess mass and dip tests are available at presently, however, and for that reason they are generally not competitive with the bandwidth test. In the present paper we develop a practical approach to calibrating the excess mass and dip tests to improve their level accuracy and power substantially. Our method exploits the fact that the limiting distribution of the excess mass statistic under the null hypothesis depends on unknowns only through a constant, which may be estimated. Our calibrated test exploits this fact and is shown to have greater power and level accuracy than the bandwidth test has. The latter tends to be quite conservative, even in an asymptotic sense. Moreover, the calibrated test avoids difficulties that the bandwidth test has with spurious modes in the tails, which often must be discounted through subjective intervention of the experimenter.
Article
(This article originally appeared in Management Science, January 1969, Volume 15, Number 5, pp. 215--227, published by The Institute of Management Sciences.) A growth model for the timing of initial purchase of new products is developed and tested empirically against data for eleven consumer durables. The basic assumption of the model is that the timing of a consumer's initial purchase is related to the number of previous buyers. A behavioral rationale for the model is offered in terms of innovative and imitative behavior. The model yields good predictions of the sales peak and the timing of the peak when applied to historical data. A long-range forecast is developed for the sales of color television sets.
Article
Existing innovation diffusion models assume that individual experience with the product is always communicated positively through word-of-mouth. For certain innovations, however, this assumption is tenuous since communicators of the product experience may transfer favorable, unfavorable, or indifferent messages through word-of-mouth. This paper examines a diffusion model for products in which negative information plays a dominant role, discusses its implications for optimal advertising timing policy and presents an application to forecast attendance for the movie Gandhi in the Dallas area.
Chapter
This book provides an elementary-level introduction to R, targeting both non-statistician scientists in various fields and students of statistics. The main mode of presentation is via code examples with liberal commenting of the code and the output, from the computational as well as the statistical viewpoint. A supplementary R package can be downloaded and contains the data sets. The statistical methodology includes statistical standard distributions, one- and two-sample tests with continuous data, regression analysis, one- and two-way analysis of variance, regression analysis, analysis of tabular data, and sample size calculations. In addition, the last six chapters contain introductions to multiple linear regression analysis, linear models in general, logistic regression, survival analysis, Poisson regression, and nonlinear regression.
Article
ABSTRACT For online marketplaces to succeed and prevent a market of lemons, their feedback mechanism (reputation system) must differentiate among sellers and create price premiumsfor trustworthy ones as returns to their reputation. However, the literature has solely focused on numerical (positive and negative) feedback ratings, alas ignoring the role offeedback text comments. These text comments are proposed to convey useful reputation information about a seller’s prior transactions that cannot be fully captured with crude numerical ratings. Building upon the economics and trust literatures, this study examines the rich content of feedback text comments and their role in building buyer’s trust in a seller’s benevolence and credibility. In turn, benevolence and credibility are proposed to differentiate among sellers by influencing the price premiums that a seller receives from buyers. This paper utilizes content analysis to quantify over 10,000 publicly-available feedback text comments of 420 sellers in eBay’s online auction marketplace, and match them withprimary data from 420 buyers that recently transacted with these 420 sellers. These dyadic data show that evidence of extraordinarypast seller behavior contained in the sellers’ feedback text comments,createsprice premiums for reputable sellers by engendering buyer’s trust in the sellers’ benevolence and credibility (controlling for the impact of numerical ratings). The addition of text comments and benevolence helps explain a greater variance in price premiums (R,=20-30%). By showing the economic value of feedback text comments through trust in a seller’s benevolence and credibility, this study helps explain the success of online marketplaces thatprimarily rely on the text comments (versus crude numerical ratings) to differentiateamong,sellers and prevent a market of ‘lemon’ sellers.
Article
Information systems and the Internet have facilitated the creation of used product markets that feature a dramatically wider selection, lower search costs, and lower prices than their brick-and-mortar counterparts do. The increased viability of these used product markets has caused concern among content creators and distributors, notably the Association of American Publishers and Author's Guild, who believe that used product markets will significantly cannibalize new product sales. This proposition, while theoretically possible, is based on speculation as opposed to empirical evidence. In this paper, we empirically analyze the degree to which used products cannibalize new product sales for books - one of the most prominent used product categories sold online. To do this, we use a unique dataset collected from Amazon.com's new and used book market-places to measure the degree to which used products cannibalize new product sales. We then use these estimates to measure the resulting first-order changes in publisher welfare and consumer surplus. Our analysis suggests that used books are poor substitutes for new books for most of Amazon's customers. The cross-price elasticity of new book demand with respect to used book prices is only 0.088. As a result only 16% of used book sales at Amazon cannibalize new book purchases. The remaining 84% of used book sales apparently would not have occurred at Amazon's new book prices. Further, our estimates suggest that this increase in book readership from Amazon's used book marketplace increases consumer surplus by approximately $67.21 million annually. This increase in consumer surplus, together with an estimated $45.05 million loss in publisher welfare and a $65.76 million increase in Amazon's profits, leads to an increase in total welfare to society of approximately $87.92 million annually from the introduction of used book markets at Amazon.com.
Article
Do dissatisfied customers engage in more or less word of mouth than satisfied customers? There is theoretical and empirical support for both possibilities. To better understand this issue, the authors developed a utility-based model of the relationship between customer satisfaction and word of mouth. The hypothesized functional form-an asymmetric U-shape-cannot be rejected based on data from the United States and Sweden. In addition, the estimation results based on the two samples are similar, suggesting that the proposed relationship is generalizable. The findings also indicate that although dissatisfied customers do engage in greater word of mouth than satisfied ones, common suppositions concerning the size of this difference appear to be exaggerated. Peer Reviewed http://deepblue.lib.umich.edu/bitstream/2027.42/68654/2/10.1177_109467059800100102.pdf
Article
The dip test measures multimodality in a sample by the maximum difference, over all sample points, between the empirical distribution function, and the unimodal distribution function that minimizes that maximum difference. The uniform distribution is the asymptotically least favorable unimodal distribution, and the distribution of the test statistic is determined asymptotically and empirically when sampling from the uniform.
Article
The emergence of online communities has enabled firms to monitor consumer-generated online word-of-mouth (WOM) in real-time by mining publicly available information from the Internet. A prerequisite for harnessing this new ability is the development of appropriate WOM metrics and the identification of relationships between such metrics and consumer behavior. Along these lines this paper introduces a metric of a purchasing population’s propensity to rate a product online. Using data from a popular movie website we find that our metric exhibits several relationships that have been previously found to exist between aspects of a product and consumers’ propensity to engage in offline WOM about it. Our study, thus, provides positive evidence for the validity of our metric as a proxy of a population’s propensity to engage in post-purchase online WOM. Our results also suggest that the antecedents of offline and online WOM exhibit important similarities.
Article
This article presents a network analysis of word-of-mouth referral behavior in a natural environment. The relational properties of tie strength and homophily were employed to examine referral behavior at micro and macro levels of inquiry. The study demonstrates different roles played by weak and strong social ties. At the macro level, weak ties displayed an important bridging function, allowing information to travel from one distinct subgroup of referral actors to another subgroup in the broader social system. At the micro level, strong and homophilous ties were more likely to be activated for the flow of referral information. Strong ties were also perceived as more influential than weak ties, and they were more likely to be utilized as sources of information for related goods.
Article
The creation of online consumer communities to provide product reviews and advice has been touted as an important, albeit somewhat expensive component of Internet retail strategies. In this paper, we characterize reviewer behavior at two popular Internet sites and examine the effect of consumer reviews on firms' sales. We use publicly available data from the two leading online booksellers, Amazon.com and BarnesandNoble.com, to construct measures of each firm's sales of individual books. We also gather extensive consumer review data at the two sites. First, we characterize the reviewer behavior on the two sites such as the distribution of the number of ratings and the valence and length of ratings, as well as ratings across different subject categories. Second, we measure the effect of individual reviews on the relative shares of books across the two sites. We argue that our methodology of comparing the sales and reviews of a given book across Internet retailers allows us to improve on the existing literature by better capturing a causal relationship between word of mouth (reviews) and sales since we are able to difference out factors that affect the sales and word of mouth of both retailers, such as the book's quality. We examine the incremental sales effects of having reviews for a particular book versus not having reviews and also the differential sales effects of positive and negative reviews. Our large database of books also allows us to control for other important confounding factors such as differences across the sites in prices and shipping times.
Article
Managers are very interested in word-of-mouth communication because it can have a tremendous impact on a product's sales. However, there are at least three significant challenges associated with measuring word of mouth. It is our primary objective in this paper to address these challenges. First and foremost, how does one even gather the data? Since the information is exchanged in private conversations, direct observation is (or at least has traditionally been) quite difficult. Second, even if one could observe the conversations, what aspect of them should one measure? The third challenge comes from the fact that word of mouth is not exogenous. While the mapping from word of mouth to future sales is of great interest to the firm, we must also recognize that word of mouth is at the same time an outcome of past sales. Our core result is that on-line conversations may offer an easy and cost-effective opportunity to measure word-of-mouth. However, simply counting on-line conversations may not be informative. On the other hand, measuring the "dispersion" of these conversations across communities is. Specifically, we show that a measure of dispersion has explanatory power in a dynamic model of sales, while pure counts do not. As a context for our study, we have chosen new TV shows during the 1999/2000 seasons. Our source of word-of-mouth conversations is Usenet, a collection of thousands of newsgroups with very diverse topics.
Article
Companies spend lots of time and money on complex tools to assess customer satisfaction. But they're measuring the wrong thing. The best predictor of top-line growth can usually be captured in a single survey question: Would you recommend this company to a friend? This finding is based on two years of research in which a variety of survey questions were tested by linking the responses with actual customer behavior--purchasing patterns and referrals--and ultimately with company growth. Surprisingly, the most effective question wasn't about customer satisfaction or even loyalty per se. In most of the industries studied, the percentage of customers enthusiastic enough about a company to refer it to a friend or colleague directly correlated with growth rates among competitors. Willingness to talk up a company or product to friends, family, and colleagues is one of the best indicators of loyalty because of the customer's sacrifice in making the recommendation. When customers act as references, they do more than indicate they've received good economic value from a company; they put their own reputations on the line. And they will risk their reputations only if they feel intense loyalty. The findings point to a new, simpler approach to customer research, one directly linked to a company's results. By substituting a single question--blunt tool though it may appear to be--for the complex black box of the customer satisfaction survey, companies can actually put consumer survey results to use and focus employees on the task of stimulating growth.
Article
Despite the interest in measuring price sensitivity of online consumers, most academic work on Internet commerce is hindered by a lack of data on quantity. In this paper we use publicly available data on the sales ranks of about 20,000 books to derive quantity proxies at the two leading online booksellers. Matching this information to prices, we can directly estimate the elasticities of demand facing both merchants as well as create a consumer price index for online books. The results show significant price sensitivity at both merchants but demand at Barnes and Noble is much more price-elastic than is demand at Amazon. The data also allow us to estimate the magnitude of retail outlet substitution bias in the CPI due to the rise of Internet sales. The estimates suggest that prices online are much more variable than the CPI, which understates inflation by more than double in one period and gets the sign wrong in another.
Article
We analyze a model where an altruistic, but possibly overconfident sender broadcasts one of a finite set of messages to rational receivers. If broadcasting is costless and the sender is rational, there is an informationally efficient equilibrium, but multiple equilibria may arise, and asymmetric equilibria might be more informative than the symmetric equilibrium even if the prior is symmetric. Although overconfidence on the part of the sender reduces informativeness in some cases, it may also eliminate less informative equilibria and lead to better information transmission. Overconfidence can also improve the informativeness of the message when broadcasting is costly. Copyright 2004 by the Economics Department Of The University Of Pennsylvania And Osaka University Institute Of Social And Economic Research Association.
Article
The primary aim of the paper is to place current methodological discussions in macroeconometric modeling contrasting the ‘theory first’ versus the ‘data first’ perspectives in the context of a broader methodological framework with a view to constructively appraise them. In particular, the