Affective contents in online reviews: The impact of reviews
and sales of earlier versions of a product on consumer and
expert reviews of new editions
Master’s Thesis
Author: Andre Akbar Kurnia (10286691) Supervisor: Frederik Situmeang
MSc Business Studies – Entrepreneurship and Management in Creative Industry Track University of Amsterdam
2 ABSTRACT
This study examines the effect of reviews of past editions to reviews of sequel in a series by measuring reviews’ affective contents from consumer and expert reviewers. In order to examine this, a set of hypotheses based on Situmeang, et al. (2014) is developed and takes video game industry as the research setting. A test on the final sample of 568 video game series with n = 1710 found that past editions reviews’ affective, from both consumer and expert, influence the sequel editions reviews’ affective. In addition, global product sales of past editions also influence the reviews affective content of the sequel.
Keywords: Sequels, Consumer reviews, Expert reviews, Affective Content, Global Product Sales
3 1. INTRODUCTION
It is known that consumers usually seek quality and product related information before they purchase new products. A way to do so is by looking at the product’s review that is available online. The usage of online review has grown rapidly that it is now has become an important element in marketing (Chen & Xie, 2008). Numerous researches have been conducted in regards with online review with focus on different aspects, such as the helpfulness and the content of the review (De Maeyer, 2012). Most of these studies found out that online reviews affect the sales level of certain products, like movies and books. For example, Lee, et al. (2008) find out that both positive and negative online review influence sales in different ways, while Dellarocas, et al. (2007) find that online movie ratings addition to their revenue-forecasting model resulted in improvement of the model’s predictive power. These researches further suggest that consumers rely on online reviews and information when they purchase something, whereas firms are aware that these reviews influence their sales of products (Dellarocas, 2006).
However, most of the research on online reviews only look at the numerical rating and only a few are about the textual content of online consumer review (Hu, et al., 2009; De Maeyer, 2012; Ho-Dac, et al., 2013). The textual or verbal content of the reviews is found to provide more information to consumers than mere numerical ratings (Ghose & Iperoitis, 2011). Moreover, verbal content and how it works become important to understand as most reviews ratings are on the extreme, either they are majorly positive or negative and giving less than useful information for consumers (Hu, et al, 2009; Archak et al, 2011). Therefore, this research aims to provide deeper understanding on the verbal aspect of online review as
4 represented by affective contents. In order to do so, this study will conduct text-mining method on affective contents in online consumer review as done by Ludwig, et al. (2013).
This research takes video games industry as its setting as it has grown tremendously and now holds a large part of the entertainment industry and there are relatively few researches have been conducted in this specific area (Zhu & Zhang, 2010). The continuing growth of video games industry is marked by the rather high number of sales of game software and contents that reached for approximately $14.8 billion in 2012 only (Entertainment Software Association, 2013). It is to note that video games, in this case, are different than movies or books although they are also creative industry’s products, as they are considered of more of a high-involvement product than books or movies. This means consumers spend more time to gather information and features about the product than low involvement products, which is why they are closely interrelated with online reviews. In addition, in this industry, new products often come in forms of sequels that rely on the same tricks and formulas (Hamlin, 2013). Successful video game titles such as Call of Duty, Assassin’s Creed, and Grand Theft Auto have sequels that give their producers huge amount of revenues. In relation with this, online reviews are then used as signal of the success rate and likeliness of past performance affecting newer sequel titles (Situmeang, et al., 2014). However, most research, including Situmeang, et al’s, (2014) only looked at the numerical ratings of online reviews and its impact to the sales performance of the product. This research then measures the effect earlier online consumer review as a signal of future success, with the focus on the affective content of the review. Moreover, the research also looks at the difference between expert and consumer reviews in order to provide better explanation on the topic in hand. Eventually, it leads into the research question of: What are the impact earlier online consumer review to newer online consumer review as measured by the affective content of the review?
5 By doing so, this thesis research attempts to shed a light on the topic of online consumer review textual content and its impact to product sales performance. Secondary data sources throughout a period of time are used in order to show the impact of past reviews and performances to future reviews and performances. Text analysis is used on this data source by using LIWC tool that is also used in previous research (Ludwig, et al., 2013). Moreover, the paper also compares the different review affective contents from both expert and consumer reviewers. The paper is then structured as follow: first, a literature review of relevant articles of the topic is presented. Then the conceptual framework will be provided, showing the relations of each variables and research hypotheses. The discussion and conclusion of the research will be provided in the end.
6 2. LITERATURE REVIEW
2.1 Sequels
Most researches regarding sequels often came from other sector of creative industry, especially in movies industry (Sood & Dreze, 2006; Basuroy & Chatterjee, 2008). Basuroy and Chatterjee (2008) explained that sequels are form of brand extension, which already has existing brand name attached on the product. Brand extensions, even it has existing product name on it, can take similar or different forms (Sood & Dreze, 2006). According to this definition, a certain product’s sequels, prequels, spin-offs, or reboots can be considered as brand extension. In this concept, the experiences that consumers had with previous editions of the product highly contribute to the market success of the products (Volckner & Sattler, 2006). This is because of that the ideas and impressions from earlier editions are also connected to the newer editions (Keller, 2003; Situmeang, et al., 2014) and that the popularity of earlier editions can also build up excitement and anticipation toward newer editions (Dhar, et al., 2012).The reason producers extend their brand are because they are beneficial in high budget media-related products, such as movies and video games, as the gross revenue from the products are available immediately after release. Also, releasing sequels can also reduce risks of a product to be failed in market, as sequels usually capitalize on previous product’s success (Sood & Dreze, 2006). However, the excitement and anticipation caused by previous edition can create high expectations which may lead into lower expectation level (Anderson & Sullivan, 1993) and lower sales than of the original edition of the product (Basuroy & Chatterjee, 2008).
7 Similar with every creative industry’s products, video games business also takes these considerations into account. Williams (2002) mentioned that video game developers tend to play it safe and reuse an already proven successful formula rather than taking more risks by developing a completely new video games. The motivations of doing so vary from high cost of video game development (Superannuation, 2014), already known brand name and market (Milewicz & Herbig, 1994), or companies’ stronger focus on business and financial interest (Tschang, 2007). This means that producers rarely consider to launch a sequel product if the previous edition of the product is a failure or not accepted well by consumers. On the other hand, consumers also need to consider signals from the available information before deciding to purchase the new product, especially when the product is an experience goods such as video game. As a result, both producers and consumers need to consider several signals before launching sequel product, such as preceding sales performance and consumers’ perception, which can often be found in online review of the product (Situmeang, et al., 2014). Therefore, this research then examine these reviews and how they affect past products and their sequels. The next section provides relevant literatures to further understand the concept of online reviews.
2.2 Online review and its roles
Online product review is a form of product information created by certain product consumer or user that is based on personal usage experience (Chen & Xie, 2008). It can also be viewed as a special form of electronic word of mouth as it has wider impact and reach due to the Internet (Dellarocas, 2003; Godes & Mayzlin, 2004; Hennig-Thurau, et al., 2004). Electronic word of mouth or eWOM can be defined as “any positive or negative statement made by potential, actual, or former customers about a product or company, which is made available to a multitude of people and institutions via the Internet” (Hennig-Thurau, et al., 2004). Moreover, earlier research have highlighted the impact of word-of-mouth in
8 marketing, such as its impact to product sales (Godes & Mayzlin, 2004; Chevalier & Mayzlin, 2006; Duan, et al., 2008). Nowadays, WOM are often expressed or written in online reviews that provide information about a product, which consumers can search and read from.
Then it brings us to the roles that review can play in marketing. Generally, review has two basic roles or effects: influence and prediction effects (Gemser, et al., 2007). Influence effect means that review actively affect or influence consumers’ selection process. While prediction effect means that product’s success or failure can be predicted by looking at the review. These roles mean that online reviews are known to influence certain organizational performances, including their sales, as supported by a growing number of research (Basuroy, et al., 2003; De Maeyer, 2012). Example related to its effect on sales is found on the research based on the data of Amazon.com and BN.com which found out that higher online consumer reviews lead to higher book sales (Chevalier & Mayzlin, 2006). Dellarocas et al. (2007) provided a model of box office sales that includes prerelease marketing, availability of theater, and critic review, while also included online movie ratings information and concluded that it can significantly improve the sales prediction accuracy. Clemons et al. (2006) found out that strongly positive ratings in product review positively influence product sales growth. In addition, online consumer review has significant implications for various management activities such as brand building, customer acquisition, quality assurance, and also product development (Dellarocas, 2003). Thus, in line with Chen and Xie’s (2008) paper, it has become an important part in modern marketing mix that needs to be managed well in order to create favorable organization’s performance. As a result, it becomes necessary for firms and organizations to understand more about online review and its role in marketing communication.
9 Another role of online review is that it can act as mean to signal specific product’s quality. Spence (1973) defined signal as manipulable attributes or activities that communicate information about characteristics of economic agents, such as firms or consumers. Whether it comes from consumer or expert, this type of information becomes more and more important because it can signal the product features and quality if the product is hard to evaluate before the actual purchase. These findings further support the argument that online reviews signal the quality of the product in order to minimize consumers’ uncertainties regarding the product. However, this is under the assumption that the acknowledged signals of the review are those that are linked directly to the product (Johnson & Levin, 1985; Kirmani & Rao, 2000; Situmeang, et al., 2014). In products that have sequels of editions, the effect of reviews on sales are not always under the assumption mentioned earlier (Situmeang, et al., 2014). This leads to the interesting question of whether the reviews are stable through time and whether the reviews from the past editions of products are related to the focal, newer editions. The research then takes this question into account and tries to further examine the relation through literature review and data analysis conducted later.
2.3 Relationship of past and future online review and performances
Numerous studies have examined the relation between past performances to future performances, suggesting that past products’ performance affect future or newer products (Keller, 2003; Dhar, et al., 2012; Situmeang, et al., 2014). Note that this usually apply to products that have sequels or newer editions. The past success can affect newer product in two different ways: either make them perform and sell better, or the opposite (Oliver: 2009; Basuroy & Chatterjee, 2008). The different outcomes can occur because new products do not meet the expectation built from the past, resulting in lower expectation and performance in the present time (Anderson & Sullivan, 1993; Oliver, 2009). In relation with online reviews, consumers build their perception for a certain product from the information that are provided
10 in other consumer’s reviews. Therefore, online reviews can be considered as important predictor for future successes of sequel products (Caves, 2000; Chen & Xie, 2008).
Based on these literatures, past and future performances, as well as reviews, are related and contribute to how product related information are perceived. Situmeang, et al., (2014) argued that past reviews are signals of consumers’ appreciation of the product series, therefore the reviews affect not only the sales performance of that particular product but also those products that will be available in the future. Therefore in order to measure the effect, it is suggested to look at two aspects from earlier versions of products: 1) the review and 2) the sales performance (Situmeang, et al., 2014). From here, there is a chance that focal product will be reviewed positively since it has been reviewed as such in the past. This is due to the nature that sequels are generally similar with the earlier product, in terms of characteristics and form. Similar with Situmeang, et al. (2014), this paper does not try to analyze the factors that influence sales performance, but rather the factors that lead to change in the reviews of sequel products, where past sales performance also act as a predictor.
However, lack of consensus between reviewers, either from expert or consumer, can affect the past reviews and sales in different way (Sun, 2012; Situmeang, et al., 2014). This differences can be attributed to the different characteristics that consumers and experts have (Wijnberg & Gemser, 2000). This study then attempts to address these problems mentioned in previous studies regarding the relation of past online reviews to the more recent reviews. This is achieved by looking at the verbal aspect of online review and sales performances of the products. The next part discuss relevant literatures of the verbal aspect of online review, which is the one of the main topics to study in this research.
11 Online consumer review has various dimensions, as supported by previous studies and research (De Maeyer, 2012; Dellarocas, et al., 2007). These dimensions consist of the verbal, valence, variance, volume, and helpfulness of the online consumer review. Verbal, as it says, is about the text content that an online consumer review has (De Maeyer, 2012). De Maeyer (2012) also provided summary of on the other dimensions: valence means the numeric value of a product rating mentioned in the review, variance is linked to the variety of product ratings, volume relates to the quantity of information that is available to consumers to access, and helpfulness mainly deals about how helpful the review is as seen from consumer’s perspective. These dimensions also have their own influence and benefits, such as high volume of online consumer reviews can lead to higher consumer’s awareness, confidence, and lower consumer’s uncertainty about a product (Chen, et al., 2004) or review’s helpfulness allows firms and organizations to have better understanding of the type and practice of product information that is best appreciated by consumers (De Maeyer, 2012).
Most research, however, only pay attention on the numeric ratings of reviews to support the importance of valence (Chevalier & Mayzlin, 2006; Dellarocas, et al., 2007; Chintagunta, et al., 2010), volume (Chen, et al., 2004), and variance (Sun, 2011) of online review. While on the other hand, research about the verbal dimension are still few even though it is found to contribute to organization’s performance, such as improving sales prediction and as price making determinant (Hu, et al., 2009; Archak, et al. 2011; Ghose & Ipeirotis, 2011; Ludwig, et al., 2013). Moreover, verbal content and how it works become important to understand deeper as most reviews are distributed in bimodal manner, meaning that most reviews ratings are on the extreme, either they are majorly positive or negative, that often give less than useful information for consumers (Hu, et al, 2009; Archak et al, 2011). Therefore, the verbal part of the review can be useful for consumer to deduce of this contrasting information. By doing so, consumers receive greater awareness of the product
12 features and benefits, while also gain better understanding which reviews are relevant and which reviewers have the same preferences as themselves (De Maeyer, 2012).
As mentioned earlier, online reviews are known to influence certain organizational performances, including their sales, as supported by a growing number of research (De Maeyer, 2012). Example related to its effect on sales is found on the research based on the data of Amazon.com and BN.com found out that higher online consumer reviews lead to higher book sales (Chevalier & Mayzlin, 2006). Dellarocas et al. (2007) provided a model of box office sales that includes prerelease marketing, availability of theater, and critic review, while also included online movie ratings information and concluded that it can significantly improve the sales prediction accuracy and Clemons et al. (2006) found out that strongly positive ratings in product review positively influence product sales growth. However, contrasting outcomes often found in research on numerical ratings and cues. For example, in contrast to Dellarocas et al. (2007) study, review’s volume instead of rating is the one that drives sales (Duan, et al., 2008; Liu, 2006). These mixed outcomes might come from methodological issue that cannot facilitate different factors and determinants (Ludwig et al, 2013) while can also resulted from the lack of numeric value’s ability to express more out of online consumer review (Ghose & Ipeirotis, 2011; Cao, et al., 2011; De Maeyer, 2012). Therefore, it can be suggested that the verbal dimension of online consumer review can give more detailed information on how firms and organizations should utilize online consumer review to yield favorable sales result.
Ghose and Iperoitis (2011) provided a quote that is considered a good start in digging deeper on the verbal contents of online consumer review by saying that “the text of reviews contains information that influences the behavior of the consumers, and that the numeric ratings alone cannot capture the information in the text”. Several studies have shown that there are several word categories that build the verbal dimension of online review, such as
13 sentiment words, affective words, and linguistic style. The sentiment and affective words are explained below.
Sentiment is a “view or opinion that is held or expressed” or “a feeling or emotion” (Oxford Dictionaries, 2014). Earlier attempts of research in this area measured the sentiment in reviews, whereby reviews are categorized into either positive and negative that depends on the use of specific words and phrases in the review content (Whitelaw, et al., 2005; Pang & Lee, 2008; Das & Chen, 2007; Hu & Liu, 2004). These content words are usually nouns, regular verbs, and many adjectives and adverbs, as they express the content of communication (Taucszik & Pennebaker, 2010). For example, in Archak, et al (2011) study of deriving pricing power based on text in consumer reviews revealed that words such as “amazing picture quality”, “great picture quality”, and “simple ease of use” are mentioned and used as an indicator of positive or negativity of the review.
While having the same characteristics of sentiment words, affective words are the more intimate and representative of certain consumers’ evaluation of a product (Ludwig, et al., 2013). Affective content words are words that represent emotion (such as happiness, anger, or sadness) which also reveal the intent of a text (Bird et al., 2002; Das, et al., 2005). It also refers to the “internal feeling state” of the writer (Cohen, et al., 2008) that is “consciously accessible as the simplest raw feelings evident in the moods and emotion” (Russell, 2003). Based on empirical research about treating feelings as information by Schwarz & Clore (1996), this paper has a clear rational of mining affective words in relation with online reviews.
Affective contents or words are more likely to influence consumers who have low motivation to engage in detailed cognitive processing of information, less resources to use with (such as time pressure), ambiguous other sources of information, and lack of expertise in certain area (Cohen, et al., 2008; Ludwig, et al., 2013). These conditions are often found in
14 online purchase process (Jones, et al., 2004). Moreover, the characteristics of text-based affective content words would mean that they can provide more accessible and understandable signals about targets or products in reviews (Cohen, et al., 2008). Those mentioned conditions represent the characteristics of video games industry. Which is why this may means that the use of word cues are most likely the effective way to measure affect within reviews (Ludwig, et al., 2013). These information are then used as underlying reasons of measuring affective words in online review’s content in this research.
Other than the sentiment and affective content words itself, style words are also important as they are the ones who communicate the contents (Taucszik & Pennebaker, 2010; Ludwig, et al., 2013). Evidentially, Ludwig et al. (2013) also found that the combination of affective words and linguistic style in online consumer reviews are more likely to influence consumers buying decision. Sentiment and linguistic style are not included in this study as it only takes affective words as variable to be measured in the analysis.
However, the usage of words or textual contents from online consumer review is not without its downside. Words and sentences used by reviewer in their online review sometimes are meant to be an irony or sarcasm, which are completely represent the opposite (Ludwig, et al., 2013; Pang & Lee, 2008). Other studies by Archak, et al. (2011) and Ghose, et al. (2007) revealed that even positive words such as “amazing”, “good”, or “great” do not always result positively to demand or sales, as consumers may consider those words are hyperbolic and therefore think the review as untrustworthy. On the other hand, similar findings are also found by Ghose, et al. (2007) whereas positive comments like “decent quality” or “well packaged” also resulted negatively, and reasoned that high number of reviews that use hyperbolic language mean that words such as “decent” are considered as neutral. In order to learn more about the role of textual content in online consumer review and its relation with measurable
15 outcome such as sales, further research in the same area are needed. This is due to the reason of the complexity and lack of consensus of findings with regards to the topic in hand. One way to do so is by more detailed study and measurement of the differences that positive or negative reviews made.
Based on the literature reviews above, this research then propose the main research question: What are the impacts of earlier online review to newer online consumer review as measured by the affective contents and linguistic style of online review? In order to get answer the question, these research use text-mining method by taking review’s affective contents into account as summarized by Pang and Lee (2008). The hypotheses tested in this paper are explained in the next section.
16 3. HYPOTHESES
Products that have same brand name are closely linked to each other and information from one part of the series is also applicable to other part of the series (Erdem, 1998). The performance of newer editions can also be determined by past editions’ image as it is carried over from the earlier to newer product (Keller, 2003). This study is based on Situmeang, et al., (2014) paper that examines the relation between past and newer reviews and sales performances. The paper found that online reviews from past editions influence consumers and experts alike (Situmeang, et al., 2014). These literatures then serve as underlying rational for further hypotheses to be tested below.
3.1 Affective Words Hypotheses
Affective words represents the feeling and emotion of the reviewer. These emotions and feelings can be categorized as positive or negative and good or bad (Pang & Lee, 2008). However, words used in different reviews are more likely to be distinct, as reviewers rarely use the same set of words to express their sentiment (Pang & Lee, 2008). But results from earlier study by Pang, et al. (2002) on movie review suggests that coming up with right set of
17 keywords solved the issue. Moreover, the affective contents within reviews are often come in two distinct polarities, which is about similar with how numerical review rating works (Pang & Lee, 2005; Das, et al., 2005).
It is a consumer’s tendency to link past performance with current performance as they learn about the franchise by consuming a series of product sequels or editions (Rothschild & Gaidis, 1981; Erdem, 1998). Therefore, the consumer reviews on the earlier editions of a product do not only signal the quality of the reviewed product, but also the series as a whole. Considering this concept of consumer reviews as indicators of quality, it is rational to expect some form of continuity in consumers’ evaluation about the subsequent editions of the series. Moreover, video game series are usually consumed by same consumers or fans of the series, which means their reviews would probably have similar affective contents from one edition to another. Therefore, it is safe to assume that if an earlier edition is reviewed positively, then so is the next editions of the product. Based on these and the nature of product sequels itself, hypothesis 1a is formed:
H1a: Consumer reviews’ affective content of past editions in the series have a positive relationship with consumer reviews’ affective content of the new edition of the product.
The arguments provided earlier in terms of consumer reviews, are also likely applicable to the relationship of expert reviews of past and newer editions. This is due to the reason that new sequel’s characteristics are usually similar to the original products’ characteristics. This similarities then can cause similar responses toward the original product and the sequel (Aaker & Keller, 1990). Moreover, expert reviewers are more likely to provide similar reviews for sequels to avoid damaging their credibility and reputation. A recent study by Vermulen and Seegers (2009) found out that expert reviews’ sentiment do not differ much from consumer reviews’. They do, however, may provide different reviews if the new edition diverges significantly from the original one.
18 Previous arguments suggest that expert reviews will remain about the same than that of consumers. However, there are also arguments that oppose this idea. Professional or expert reviewers are more likely to detect lack of creativity and innovation in a product, which suggests that products with similar attributes may lead to lower rate of expert reviews. This condition is more likely to happen in high art creative industry, where innovativeness plays an important role in judging quality (Wijnberg & Gemser, 2000). Situmeang, et al. (2014) then found out that expert reviews of new editions are highly affected by those of past editions. Therefore it can be hypothesized that:
H1b: Expert reviews’ affective content of past editions in the series have a positive relationship with expert reviews’ affective content of the new edition of the product.
3.2 Sales Hypotheses
Sales performance of earlier editions of a product signals the quality and attractiveness of the version as it tells the number of products sold (Hennig-Thurau, et al., 2009). Earlier studies have shown that market performance of the previous editions in a product series has significant and positive effect on newer edition’s market performance (Hennig-Thurau, et al., 2009; Situmeang, et al., 2014). A signal theory can be used to explain this kind of relationship. Along with reviews, sales performance then can be used as a signal to see how
19 well the whole series have been famous and evaluated by consumers in general across past editions of the series.
In addition, sales performance also indicate how many consumers know and aware about the product series. The higher the number of people know, the higher the chance of word of mouth communication among them, which eventually affect sales (Duan, et al., 2008). Moreover, consumers talk more about movies that are financially successful over unsuccessful ones, which again would affect the revenues (Moon, et al., 2010). Therefore, the more people review the product series, the more likely it will be for them to talk with other consumers about the product as they encourage people to try the new product and create sentiments towards the product. These arguments bring us to the next hypothesis:
H2a: Sales performance of past editions in the series has a positive relationship with consumer reviews affective content of the new edition of product
As for expert reviews, however, the impact of sales performance may be less than the impact on consumer reviews. Consumer reviews and sales performance, in this case, are both measurements of how well consumers evaluate the series (Situmeang, et al., 2014). Expert reviews, however, are clearly different from consumer reviews. First, they are professionals, as expert reviewers are sometimes called as independent professional reviewers, who express their opinions before the launch of the product (Debenedetti, 2006). Even so, it does not mean that they are immune to bribes (Eliashberg & Shugan, 1997; Mol & Wijnberg, 2007), but sales performance itself is less likely to influence their evaluation regarding a product. Expert reviewers are assumed to be unbiased, willing to inform the actual to the public, and to evaluate based on how they perceive the product itself, paying less attention to consumers perception of earlier editions.
20 However, conditions in real world said otherwise, especially in market driven industries such as video games industry (Phillips, 2011). In such industry, the dominant selectors are the consumers (Wijnberg & Gemser, 2000). A number of studies highlighted the issue that because of such nature in the industry, reviewers are often pressurized by profit-seeking purpose and maintaining readership level to ensure their business prevail (Chen, et al, 2012; Hu, et al., 2012). Because of this, these reviewers and media are assumed to acknowledge what the readers or consumers said. This way, consumers’ perceived preferences and evaluations are likely to influence the evaluation of expert reviewers. Eventually it means that success of earlier editions, as represented by sales performance, may directly or indirectly affect expert reviews. Situmeang, et al., (2014) argued that when there is clear and strong signal that earlier product edition is successful, reviewing the same product negatively means that expert writes against what market wants. By doing so, experts face the potential problem of severing themselves and the media they write in, particularly in industry where market selection is dominant. Moreover, if an expert’s opinion is not in line with the consumers, the review readers may start to doubt the credibility of that particular expert. Therefore, it can be hypothesized that:
H2b: Sales performance of past editions in the series has a positive relationship with expert reviews affective content of the new edition of product
21 4. RESEARCH DESIGN AND METHODOLOGY
4.1 Sample
The research data sample is taken from the online websites that provide widespread and global performance records of creative products’ (video games) sales and reviews. The online websites used are vgchartz.com for the sales database and metacritic.com for online reviews. These online websites have also been used by previous studies (Hennig-Thurau, et al., 2009; Zhu & Zhang, 2010; Situmeang, et al., 2014). Moreover, video games that have a minimum number of 2 editions in one series are collected as sample since this study aims to understand the effect of past sales and reviews to future sales and reviews. The data only focus on video games from major consoles of Sony Playstation PS1 and PS3, Microsoft Xbox and Xbox 360, and Nintendo Wii from year 2000 – 2013. The final data then consist of 568 video game series with n = 1710.
The data for online reviews, including expert and consumer reviews are taken from the Metacritic (www.metacritic.com) website database. This research adapt the approach taken by
22 Situmeang, et al., (2014) in obtaining the reviews data. For each review type, affective content analysis of the textual content of the review is used, similar to what has been done in Ludwig, et al. (2013). The reviews’ text will be analyzed by using text mining/sentiment LIWC analysis tool. The usage validity of LIWC have been supported by numerous studies (Ludwig, et al., 2013; Tausczik & Pennebaker, 2010). The LIWC program can analyze hundreds of standard ASCII text files or Microsoft Word documents in seconds (LIWC, 2007). The validity of LIWC has been confirmed in more than 100 studies that applied this methodology to various texts (Ludwig et al., 2013) and research suggests that LIWC accurately identifies emotion in language use (Tausczik and Pennebaker, 2010). This tool use online dictionary that will be related with the affective words, which will reveal the level of intensity of affective content in the reviews.
Moreover, this research use the Internal Pennebaker LIWC dictionary as words source. There are a total of 406 words that represent positive emotions and 499 words of negative emotions. Words such as “happy” or “nice” are used to determine the positive affective content measurement in the reviews, the same case is also applied to negative words like “bad” or “ugly”. For example, if the word “happy” appeared in the positive emotion dictionary and is counted as 1 in the total amount of positive affective content within review. Whereas “bad” came from negative emotion dictionary and is counted as 1 in the total amount of negative affective words in the review’s text.
Similar to the equation formula used in Ludwig, et al.’s (2013) paper, this research also use the formula of:
𝑨𝑨𝑨𝑨ᵢ = � 𝑷𝑷𝑨𝑨ᵢƁ ∩ 𝒊𝒊=𝟏𝟏 − � 𝑵𝑵𝑨𝑨ᵢ ∩ 𝒊𝒊=𝟏𝟏 Ɓ
23 ACi represents the overall intensity of affective content in the reviews for video game i. The subscript (B) denotes the body of the review text, and the calculations for affective content of the review. PAiB and NAiB stand for the positive affective- and negative affective content in the reviews. Therefore the score of the equation will be the total amount of positive affective contents in the review minus the total amount of negative affective contents in the reviews. Additionally, the data for sales performance are gathered from www.vgchartz.com. This website contains the number of sales of legal copies of video games in global scale. The number of sold copies is considered a good indicator of sales performance since re-sales and illegal copying are assumed to be proportional to legal sales (Situmeang, et al., 2014). The summary of the data information are shown in table 1. These information consists of the number of expert and consumer reviews of each consoles and their global sales.
4.2 Variables and operationalization
The dependent variables in this research are the consumer review’s affective content (ConAff) and expert review’s affective content (ExpAff). As mentioned earlier, each of expert and consumer’s affective content in reviews have been analyzed and quantified with LIWC. The LIWC result then provide sum of both positive and negative affective emotions in each video game reviews. Then, the sum of positive affective contents minus the sum of negative affective contents resulted in the final sum of emotion in video game review, which can be positive or negative.
The first independent variables for this research is the global sales of video game (Sales). The global sales number of video game is chosen to measure the performance of video games. Moreover, the number of sold copies is considered a good indicator of sales performance since re-sales and illegal copying are assumed to be proportional to legal sales
24 (Situmeang, et al., 2014). Then, for examining the effect of reviews of previous editions to the evaluations of their sequels all non-sequel games were deleted from the data, while the expert- and consumer reviews of previous editions of sequels were established. This resulted in the independent variables of positive affective content of consumer reviews of previous editions (ConsAffPE) and positive affective content of expert reviews of previous editions (ExpAffPE). Similar approach is also done for the sales of previous edition of sequels of videogames (SalesPE).
This research then use change in age rating from previous edition to sequel (Teen, Mature and Everyone), change in genre from previous edition to sequel (First Person Shooter, Racing and Action Adventure), expert scores (Metascore) and consumer scores (Userscore) as control variables. They control for possible attribute change that may influence consumer or expert reviews. The data for control variables are taken from metacritic.com. The overview list of the variables measured in this research is shown on Table 1 below.
Table.1 Overview of the variables
Variable Name Description
Sales Worldwide unit sales (million copies) of video games.
ConsAff Positive affective content in consumer reviews of video game. ExpAffi Positive affective content in expert reviews of video game.
ConsAffPE Positive affective content in consumer reviews of previous edition of video game.
ExpAffPE Positive Affective Content in expert reviews of previous edition of video game.
SalesPE Worldwide unit sales (million copies) of previous edition of video game.
FPS Genre classification of video game is FPS (First Person Shooter). Dummy variable.
25 Racing Genre classification of video game is Racing. Dummy variable. ActionAdventure Genre classification of video game is Action Adventure. Dummy
variable.
Mature ESRB rating of video game is Mature. Dummy variable. Teen ESRB rating of video game is Teen. Dummy variable. Everyone ESRB rating of video game is Everyone. Dummy variable. Metascore Expert scores of video game.
Userscore Consumer scores of video game.
4.3 The model of hypotheses testing
The hypotheses are tested by using the statistical model mentioned below. The model incorporates expert and consumer review’s affective content and of previous and the sequel, as well as the global sales from both previous editions and the sequel. The control variables (age classification, genre classification, user score, and meta score) are also included in the model.
ConsAff = α + β1 ConsAffPE + β2 ExpAffPE + β3 SalesPE+ β4 Teen + β5 Mature + β6 Everyone + β7 FPS + β8 Racing + β9 ActionAdventure + β10 UserScore + β11 Metascore+ β12 ExpAff+ ϵ
26 ExpAff = α + β1 ExpAffPE+ β2 ConsAffPE + β3 SalesPE + β4 Teen + β5 Mature + β6
Everyone + β7 FPS + β8 Racing + β9 ActionAdventure + β10 UserScore + β11 Metascore+ β12 ConsAff+ ϵ
5. RESULTS 5.1 Descriptive statistics and correlations
Table 2 shows the correlation matrix and descriptive statistics of the variables used in this research. The results provided by correlation analyses can be used as preliminary evidence to support the hypotheses. The correlation analyses show preliminarily that affective contents of sequel review is highly correlated with the past review’s affective contents from their respective communities (r = .597, p < .01 for consumers, and r = .572, p < .01 for experts). In addition, there is also a significant and positive relationship between past sales and review’s affective contents (r = .258, p < .01 for consumers and r = .205, p < .01). Moreover, it is also found that past consumer review’s affective content is significantly
27 correlated with the expert’s review’s affective content of the sequel (r = .370, p < .01). Similar relation is also found between past expert review’s affective content and consumer’s review affective content of the sequel (r = .410, p < .01).
As for sales performance, it is found out that the affective content in consumer reviews of previous editions has a significant positive correlation with sales performance of the sequel (r = .453, p < .01). The same also applied to the relationship between affective content in expert reviews or previous editions with sales performance of the sequel (r = .251, p < .01). In addition, previous editions’ sales also have a strong and significant correlation with the sales of the sequels (r = .673, p < .01).
28
Table 2 Descriptive Statistics and Correlations
Variable name Mean SD 1 2 3 4 5 6 7 8 9 10 11 12 13
1 Sales 1.194 2.410 2 Sales PE 1.348 3.089 .673** 3 Consumer Affective Content 107.353 152.360 .453 ** .258** 4 Consumer Affective Content PE 118.757 159.064 .387 ** .407** .597**
5 Expert Affective Content 113.898 108.636 .289** .205** .521** .370**
6 Expert Affective Content
PE 122.206 112.470 .251 ** .251** .410** .540** .572** 7 UserScore 6.923 2.121 0.001 -0.035 .136** .065* .182** .089** 8 MetaScore 74.593 12.581 .302** .152** .400** .302** .524** .343** .464** 9 FPS 0.050 0.218 .083** .070* .099** .135** -0.016 -0.019 0.028 0.047 10 Racing 0.050 0.218 0.018 -0.008 -0.045 -.059* 0.007 -0.006 0.062 -0.024 -0.053 11 Action Adventure 0.074 0.261 -0.012 0.003 0.029 0.053 0.005 .055* .096** -.067* -.065* -.065* 12 Mature 0.228 0.420 .164** .125** .310** .339** 0.029 .078** .080* .123** .324** -.125** .186** 13 Teen 0.273 0.446 -.078** -.079** -.058* -.061* 0.009 -0.020 .104** -.146** -0.049 -0.041 -0.013 -.333** 14 Everyone 0.312 0.464 -0.026 -0.011 -.138** -.162** -0.031 -.076** 0.016 .060* -.154** .178** -.166** -.366** -.413**
29 5.2 Hypotheses test results
The hypotheses are tested by using linear regression method. The summary of regression results for both consumer and expert review’s affective content are shown in table 3. The results show that the squared multiple correlations (R2) of the dependent variables are .535 for consumer affective content in review and .510 for expert affective content in review. The numbers means that approximately 53,5% and 51% variance of both consumer and expert community are explained by these models.
The results of the analysis of the models suggest that consumer’s affective contents in reviews of the new edition are negatively related to expert affective contents in reviews of past editions (β = -.099, p < .01) and sales of past editions (β = -.231, p < .001), but positively related to consumer’s affective contents in reviews of past editions (β = .438, p < .001). As for expert’s affective contents in reviews of the new edition, they also have negative relationship with consumers’ affective content in reviews of past editions (β = -.115, p < .01), although they have positive relationship with previous editions’ expert review affective contents (β = .424, p < .001) and sales (β = .081, p < .05). These results then support H1a, H1b, and H2b.
Apart from the hypotheses variables, the research also found evidence that expert’s affective contents in reviews are positively related with consumer’s affective contents in reviews (β = .294, p < .001) and vice versa (β = .310, p < .001). However, the past editions of both reviews affect each other negatively, as consumer’s affective content of past editions have negative relation with expert’s affective content of sequels (β = -.115, p < .01) and conversely past editions’ expert affective contents have negative relation with sequels’ consumer affective contents (β = -.099, p < .01).
30 From control variables, mature age rating is known to be positively related to consumer’s affective content in review (β = .148, p < .001 but related negatively to expert’s affective content in review. (β = -.110, p < .01). Metascore or expert score is also positively related to both consumer and expert’s affective content in reviews (β = .084, p < .05 for consumer and β = .272, p < .001 for expert). The rest of the control variables have more than .05 significance level, suggesting that their impact on consumer and expert affective contents inessential.
Table 3
Regression Result Consumer Affective Content Regression Result Expert Affective Content
Variable Name B SE β Sig Variable Name B SE β Sig (Constant) -40.613 33.239 .222 (Constant) -156.853 23.139 .000 ExpAffPE -.140 .050 -.099 .006 ConsAffPE -.075 .026 -.115 .004 ConsAffPE .408 .032 .438 .000 ExpAffPE .419 .033 .424 .000 SalesPE 9.930 1.496 .231 .000 SalesPE 2.435 1.105 .081 .028 FPS -3.876 20.314 -.005 .849 FPS -4.817 14.593 -.010 .741 Racing 5.536 21.394 .007 .796 Racing 6.307 15.368 .011 .682 ActionAdv -16.324 17.596 -.025 .354 ActionAdv 5.685 12.647 .013 .653 Mature 60.682 15.976 .148 .000 Mature -31.692 11.534 -.110 .006 Teen 18.522 13.982 .048 .186 Teen -10.611 10.049 -.039 .291 Everyone -4.568 13.630 -.012 .738 Everyone -7.514 9.789 -.029 .443 Userscore -4.798 3.740 -.038 .200 Userscore 4.622 2.685 .053 .086 Metascore 1.147 .467 .084 .014 Metascore 2.582 .322 .272 .000 ExpAff .420 .051 .294 .000 ConsAff .217 .026 .310 .000 R2 .535 .510
31 6. DISCUSSION
6.1 Theoretical implications
This research provides new insights on the concept of affective content in product review that carries over time. Earlier studies did not explore if the affective content in evaluations of earlier editions has effect on the performance of sequels. Previous studies, however, have already discussed the concept of carry over effect of product quality perception from across a series of product editions (Hennig-Thurau, et al., 2009; Keller, 1993). These previous editions then emit signals that are available through consumer and expert’s evaluations of the products (Situmeang, et al., 2014). As mentioned earlier, this study follows Situmeang, et al., (2014) research that explored the connection between consumer and expert reviews in a series of editions. This study then provides deeper understanding on the mechanism of such relationship, by using affective contents within review as a measurement.
Earlier research by Anderson and Sullivan (1993) and Oliver (2009) suggest that original product’s success results in high expectations of the sequel, which leads into lower consumer satisfaction. The findings of this research, however, proved otherwise and show that both consumer and expert review’s affective contents are influenced by affective contents from past editions’ reviews by their respective communities. Moreover, they both affect the newer editions’ reviews in a rather similar strength. This result is consistent with Situmeang, et al.,’s (2014) research that suggests past reviews from both consumer and expert influence the reviews of the sequel. This means that positive reviews of past editions are more likely result in positive reviews of new editions.
32 Hennig-Thurau et al. (2009) and Situmeang, et al. (2014) have shown that positive evaluations of past editions are associated with sequel performance success. In addition, Ludwig, et al. (2013) also found that positive affective content in reviews lead into better conversion rate in sales. Consistent with those studies, the findings of this research provide evidence that past editions’ sales performance positively affect the review of the new edition. It means that higher sales of original products will result in better reviews of the new products. However, it is also found that sales of past editions have higher impact on consumer’s review affective than to expert’s review affective. This can be attributed to the argument that professional critics or experts are more likely to punish a lack of creativity and innovation of making new products (Situmeang, et al., 2014) and are therefore often less positive than consumer reviews.
In addition, the research also found several crossover relationships of consumer’s review affective content on those of expert’s. First, consumers’ affective contents affect experts’ affective contents stronger than when it is the other way around. This means that experts are more likely to alter their reviews to meet the evaluations of the majority, which is also supported by Situmeang, et al., (2014). Second, it is found that past editions’ consumer reviews negatively affect the new editions’ expert reviews. Similar relationship is also found between expert reviews of past editions on consumer reviews of new editions. The issue of trust may have caused this relations as community members usually trust reviews that come from their own rather than the ones from outside the group (Das & Chen, 2007). Therefore, rather than increasing the positive reviews of the sequel, both past editions’ consumers and experts’ reviews tend to negatively affect each other.
33 6.2 Managerial implications
The results of this research suggest several implications for marketing and managerial purposes.First, this study is the first to attempt to examine the influence of affective content in reviews of earlier editions on the reviews of sequels. As, it is a new way of studying reviews, therefore it provides new insights on the concept of reviews’ influence, especially in marketing sector. It is then suggested that firms and managers look on the textual aspect of online reviews more extensively than before, as it conveys more information than the numerical rating of reviews. Firms and managers are also suggested to consider the importance of past reviews as signals of quality of a series of product and as consideration in launching new product or sequel.
Second, this research provide further support on studies that examine the differences between consumers and experts, which has been the objective of several studies (Godes & Mayzlin, 2004; Gemser, et al., 2007; Zhu & Zhang, 2010). In this study, differences between consumer and expert reviews affective contents are found and that they are able to cross-influence reviews. Therefore it is important to take past reviews from both communities as consideration of launching new product or sequel as they are the ones who provide evaluations on the product. However, it is also suggested that firms that produce goods should not take past expert reviews too much into consideration if past consumer reviews are already good, since consumer reviews affect expert reviews stronger. This means that expert reviews are more likely to be revised in respect of later editions to meet the general opinion of consumers.
Minor implications for firms are that text analysis method of reviews can be used to communicate better with consumers and understand them more. Text analytics help firms to predict consumers’ behavior and gain better insights of them. Moreover, firms can also place reviews that have high affective content to further promote their products and attract
34 customers. They can also ask for consumers’ cooperation to provide affective reviews of the product, as it is already proved that better affective reviews lead to better future performance.
6.3 Limitations and future research suggestions
This research has several limitations. First, the dataset is only based on video games industry. Other type of product from different industries should be included in the research to provide higher validation of the findings, preferably the products with different characteristics from video games. Products with high consumer involvement, for instance, may be beneficial to be researched under similar approach. Different affective words may occur and have different meaning than those examined in this research. Therefore, future research should consider investigating the role of affective contents for other type of products, particularly those with different characteristics.
Second, this research used global sales data as source for sales variable, which is taken from different website from the reviews. As a result, it may not represent the actual purchasing behavior of the consumers. Moreover, due to the high anonymity of the online consumer reviews in this study, the actual purchase behavior could not be determined. Future research then should address these issues and use data that reflect purchasing behavior of consumers.
Third, it is also important to note that affective words in reviews are not always literal. Ironic connotations are often used to communicate the opposite of the actual meaning of the word. This study is limited because of so even though irony plays less of a role in heuristic processing. Future research may investigate linguistic properties that represents ironic statements to help identify the sentiment orientation of the reviews and avoid errors in understanding the meaning. Moreover, future research may research other kinds of
text-35 mining research, such as review’s linguistic style and examine their relations with product performance over time.
Fourth, this research did not take the consensus of reviewers’ communities into account. Situmeang, et al. (2014) highlighted the influence of consensus of past editions reviews on sequel reviews from both consumer and expert point of views, where lack of consensus result in lower review score of the product. It is then suggested to take the consensus of consumer and expert affective contents as research topic for the future as it will provide deeper understanding and insights on how the relationship of past and sequel evaluations work.
Fifth, this research only take video game new editions without including the possible spinoffs of the franchise. Firms often capitalize their products’ success by extending their reach to media other than the original products’. For example, after the success of the comic book series, the Harry Potter or Walking Dead franchise created more profits from movie productions and video games. Future studies may investigate how expert and consumer reviews in a product media are influenced by the performance and evaluations of those from other media.
6.4 Conclusion
The aim of this research is to examine the impact of past editions online review to new editions online review, as measured by review’s affective contents. Furthermore, this research also compares the consumer and expert reviews. This research is done by using regression method on the data that consist of video game reviews and global sales. The research results have provided support the hypotheses. It is shown that affective contents of the past product reviews influence the affective contents of new product reviews in a product series. Moreover,
36 both consumers and experts past reviews have similar strength in influencing new reviews. Another result then revealed that past editions’ consumer and expert reviews also influence the sales performance of newer editions.
37 REFERENCES
Aaker, D. A., & Keller, K. L. (1990). Consumer evaluations of brand extensions.The Journal
of Marketing, 27-41.
Anderson, E. W., & Sullivan, M. W. (1993). The antecedents and consequences of customer satisfaction for firms. Marketing science, 12(2), 125-143.
Archak, N., Ghose, A., & Ipeirotis, P. G. (2011). Deriving the pricing power of product features by mining consumer reviews. Management Science, 57(8), 1485-1509.
Basuroy, S., & Chatterjee, S. (2008). Fast and frequent: Investigating box office revenues of motion picture sequels. Journal of Business Research, 61(7), 798-803.r, K. L. (1990). Consumer evaluations of brand extensions.Journal of marketing, 54(1).
Basuroy, S., Chatterjee, S., & Ravid, S. A. (2003). How critical are critical reviews? The box office effects of film critics, star power, and budgets. Journal of Marketing, 67(4), 103-117.
Bird, H., Franklin, S., & Howard, D. (2002). ‘Little words’—not really: function and content words in normal and aphasic speech. Journal of Neurolinguistics,15(3), 209-237.
Cao, Q., Duan, W., & Gan, Q. (2011). Exploring determinants of voting for the “helpfulness” of online user reviews: A text mining approach. Decision Support Systems, 50(2), 511-521
Caves, R. E. (2000). Creative industries: Contracts between art and commerce (No. 20). Harvard University Press.
Chen, P. Y., Wu, S. Y., & Yoon, J. (2004). The impact of online recommendations and consumer feedback on sales.
38
Chen, Y., & Xie, J. (2008). Online consumer review: Word-of-mouth as a new element of marketing communication mix. Management Science, 54(3), 477-491.
Chen, Y., Liu, Y., & Zhang, J. (2012). When do third-party product reviews affect firm value and what can firms do? The case of media critics and professional movie reviews. Journal of Marketing, 76(2), 116-134.
Chevalier, J. A., & Mayzlin, D. (2006). The effect of word of mouth on sales: Online book reviews. Journal of marketing research, 43(3), 345-354.
Clemons, E. K., Gao, G. G., & Hitt, L. M. (2006). When online reviews meet hyperdifferentiation: A study of the craft beer industry. Journal of Management
Information Systems, 23(2), 149-171.
Chintagunta, P. K., Gopinath, S., & Venkataraman, S. (2010). The effects of online user reviews on movie box office performance: Accounting for sequential rollout and aggregation across local markets. Marketing Science, 29(5), 944-957.
Cohen, J. B., Pham, M. T., & Andrade, E. B. (2008). The nature and role of affect in consumer behavior. Handbook of consumer psychology, 297-348.
Das, S. R., & Chen, M. Y. (2007). Yahoo! for Amazon: Sentiment extraction from small talk on the web. Management Science, 53(9), 1375-1388.
Das, S., Martínez‐Jerez, A., & Tufano, P. (2005). eInformation: A clinical study of investor discussion and sentiment. Financial Management, 34(3), 103-137.
De Maeyer, P. (2012). Impact of online consumer reviews on sales and price strategies: a review and directions for future research. Journal of Product & Brand
39
Debenedetti, S. (2006). The role of media critics in the cultural industries. International
Journal of Arts Management, 30-42.
Dellarocas, C. (2003). The digitization of word of mouth: Promise and challenges of online feedback mechanisms. Management science, 49(10), 1407-1424.
Dellarocas, C., & Narayan, R. (2006). A statistical measure of a population’s propensity to engage in post-purchase online word-of-mouth. Statistical Science, 21(2), 277-285.
Dellarocas, C., Zhang, X. M., & Awad, N. F. (2007). Exploring the value of online product reviews in forecasting sales: The case of motion pictures. Journal of Interactive
marketing, 21(4), 23-45.
Dhar, T., Sun, G., & Weinberg, C. B. (2012). The long-term box office performance of sequel movies. Marketing Letters, 23(1), 13-29.
Dhar, T., Sun, G., & Weinberg, C. B. (2012). The long-term box office performance of sequel movies.Marketing Letters, 23(1), 13–29
Duan, W., Gu, B., & Whinston, A. B. (2008). Do online reviews matter?—An empirical investigation of panel data. Decision Support Systems, 45(4), 1007-1016.
Eliashberg, J., & Shugan, S. M. (1997). Film critics: Influencers or predictors?.The Journal of
Marketing, 68-78.
Entertainment Software Association. (2013). Essential Facts About the Computer and Video Game Industry.
Erdem, T. (1998). An empirical analysis of umbrella branding. Journal of Marketing
40
Gemser, G., Van Oostrum, M., & Leenders, M. A. (2007). The impact of film reviews on the box office performance of art house versus mainstream motion pictures. Journal of
Cultural Economics, 31(1), 43-63.
Ghose, A., & Ipeirotis, P. G. (2011). Estimating the helpfulness and economic impact of product reviews: Mining text and reviewer characteristics. Knowledge and Data
Engineering, IEEE Transactions on, 23(10), 1498-1512.
Godes, D. B., & Mayzlin, D. (2004). Firm-Created Word-of-Mouth Communication: A
Field-Based Quasi-Experiment. Division of Research, Harvard Business School.
Hennig‐Thurau, T., Gwinner, K. P., Walsh, G., & Gremler, D. D. (2004). Electronic word‐of‐
mouth via consumer‐opinion platforms: What motivates consumers to articulate
themselves on the Internet?. Journal of interactive marketing, 18(1), 38-52.
Ho-Dac, N. N., Carson, S. J., & Moore, W. L. (2013). The Effects of Positive and Negative Online Customer Reviews: Do Brand Strength and Category Maturity Matter?. Journal
of Marketing, 77(6), 37-53.
Hamlin, J. (June 2013). The Sequel Cycle: Its Time for a New Way of Making Video Games. Retrieved from http://dustycartridge.com/features/the-sequel-cycle-its-time-for-a-new-way-of-making-video-game-sequels/ (accessed May 10, 2014)
Hu, M., & Liu, B. (2004). Mining and summarizing customer reviews. InProceedings of the
tenth ACM SIGKDD international conference on Knowledge discovery and data mining (pp. 168-177). ACM.
Hu, N., Bose, I., Koh, N. S., & Liu, L. (2012). Manipulation of online reviews: An analysis of ratings, readability, and sentiments. Decision Support Systems,52(3), 674-684.
41
Hu, N., Zhang, J., & Pavlou, P. A. (2009). Overcoming the J-shaped distribution of product reviews. Communications of the ACM, 52(10), 144-147.
Johnson, R. D., & Levin, I. P. (1985). More than meets the eye: The effect of missing information on purchase evaluations. Journal of Consumer Research, 169-177.
Jones, Q., Ravid, G., & Rafaeli, S. (2004). Information overload and the message dynamics of online interaction spaces: A theoretical model and empirical exploration. Information
systems research, 15(2), 194-210.
Keller, K. L. (2003).Strategic brand management: Building, measuring, and managing brand equity(2nd ed.). Upper Saddle River, NJ: Prentice-Hall.
Kirmani, A., & Rao, A. R. (2000). No pain, no gain: A critical review of the literature on signaling unobservable product quality. Journal of marketing, 64(2), 66-79.
Lee, J., Park, D. H., & Han, I. (2008). The effect of negative online consumer reviews on product attitude: An information processing view. Electronic Commerce Research and
Applications, 7(3), 341-352.
Liu, Y. (2006). Word of mouth for movies: Its dynamics and impact on box office revenue. Journal of marketing, 74-89.
Ludwig, S., de Ruyter, K., Friedman, M., Brüggen, E. C., Wetzels, M., & Pfann, G. (2013). More Than Words: The Influence of Affective Content and Linguistic Style Matches in Online Reviews on Conversion Rates. Journal of Marketing, 77(1), 87-103.
Milewicz, J., & Herbig, P. (1994). Evaluating the brand extension decision using a model of reputation building. Journal of Product & Brand Management, 3(1), 39-47.
Mol, J. M., & Wijnberg, N. M. (2007). Competition, selection and rock and roll: The economics of payola and authenticity. Journal of Economic Issues, 701-714.
42
Moon, S., Bergey, P. K., & Iacobucci, D. (2010). Dynamic effects among movie ratings, movie revenues, and viewer satisfaction. Journal of Marketing, 74(1), 108-121.
Oliver, R. L. (2009).Satisfaction: A behavioral perspective on the consumer. New York: McGraw-Hill.
Pang, B., & Lee, L. (2008). Opinion mining and sentiment analysis. Foundations and trends
in information retrieval, 2(1-2), 1-135.
Phillips, C. (2011). Sequels dominate fall video games releases. Available at http:// www.timesfreepress.com/news/2011/sep/02/0902-e1-game-developersunleashing-a-wave-of/. (accessed May 10, 2014)
Rothschild, M. L., & Gaidis, W. C. (1981). Behavioral learning theory: its relevance to marketing and promotions. The Journal of Marketing, 70-78.
Russell, J. A. (2003). Core affect and the psychological construction of emotion.
Psychological review, 110(1), 145.
Schwarz, N., & Clore, G. L. (1996). Feelings and phenomenal experiences. Social
psychology: Handbook of basic principles, 2, 385-407.
Situmeang, F. B., Leenders, M. A., & Wijnberg, N. M. (2014). History matters: The impact of reviews and sales of earlier versions of a product on consumer and expert reviews of new editions. European Management Journal, 32(1), 73-83.
Sood, S., & Drèze, X. (2006). Brand extensions of experiential goods: Movie sequel evaluations. Journal of Consumer Research, 33(3), 352-360.
43
Sun, M. (2011). Disclosing multiple product attributes. Journal of Economics & Management
Strategy, 20(1), 195-224.
Sun, M. (2012). How Does the Variance of Product Ratings Matter?. Management
Science, 58(4), 696-707.
Tausczik, Y.R. and Pennebaker, J.W. (2010). The Psychological Meaning of Words: LIWC and Computerized Text Analysis Methods. Journal of Language and Social Psychology. 29 (1), 24-54.
Tschang, F. T. (2007). Balancing the tensions between rationalization and creativity in the video games industry. Organization Science, 18(6), 989-1005.
Vermeulen, I. E., & Seegers, D. (2009). Tried and tested: The impact of online hotel reviews on consumer consideration. Tourism Management, 30(1), 123-127.
Völckner, F., & Sattler, H. (2006). Drivers of brand extension success. Journal of
Marketing, 70(2), 18-34.
Whitelaw, C., Garg, N., & Argamon, S. (2005). Using appraisal groups for sentiment analysis. In Proceedings of the 14th ACM international conference on Information and
knowledge management (pp. 625-631).
Wijnberg, N. M., & Gemser, G. (2000). Adding value to innovation: Impressionism and the transformation of the selection system in visual arts.Organization Science, 11(3), 323-329.
Williams, D. (2002). Structure and competition in the US home video game industry. International Journal on Media Management, 4(1), 41-54.
Zhu, F., & Zhang, X. (2006). The influence of online consumer reviews on the demand for experience goods: The case of video games.
