Faculty of Economics and Business
Master's Thesis
MSc. in Business Administration – Marketing Track
Thesis Title:
Pass the popcorn:
The effects of online movie reviews on satisfaction in the
expectancy-disconfirmation-performance model
Author:
Marcela Torres-Muga De Olarte (5833140)
Thesis supervisor:
Frederik Situmeang
2
Statement of Originality
This document is written by Student Marcela Torres-Muga De Olarte who
declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original
and that no sources other than those mentioned in the text and its references
have been used in creating it.
The Faculty of Economics and Business is responsible solely for the
supervision of completion of the work, not for the contents.
3
Abstract
This study takes a signaling theory perspective and explores if peripheral signals (rating and tonality of earlier movie online reviews) influence user satisfaction. To capture this "signaling and carry-over- effect," the expectancy-disconfirmation and performance (EDP) model is adopted. The aim is to investigate how online reviews of earlier movies through disconfirmation, mediate the relationship between performance (the overall quality of the last movie) and user satisfaction for the last movie by a particular director. As predicted, disconfirmation based on tonality (emotions) had a greater (weak but significant) impact on satisfaction than disconfirmation based on numerical ratings (which showed no significant effects). This confirms that though rating and tonality both measure satisfaction, there are important differences as only review tonality seems to exhibit a significant signaling effect. This study proved that though tonality can largely be used as an extension of reviewer ratings, tonality seems to be particularly sensitive to emotionally laden signaling effects in terms of satisfaction at least when considering movie-goers. This is consistent with studies that show that emotions are a key determinant of satisfaction and that emotional responses constitute an important aspect of hedonic consumption experiences.
Moreover, this study also tackles variance, an understudied aspect of online reviews. Results for ratings and tonalities were mostly the same: users and critics (when combined into one measure) seem to disagree more in terms of tonality than in terms of ratings. However, regardless of if movie ratings or tonality is considered, there were no direct effects of variability on satisfaction or of variability on the relationship between expectation and satisfaction. Contrary to what was hypothesized, variability was not a significant moderator. Greater disagreement among previous movie online reviews by users and critics did not significantly weaken the signaling effect based on expectations of previous movies.
4
Table of Contents
1. Introduction
6
2. Literature Review
10
2.1. The Creative Industries & Movies as Experience Goods
10
2.2. Online Reviews: Definitions, Features & Affective Content
12
2.2.1. Definition & Characteristics
12
2.2.2. Role of Emotions
14
2.3. Online Reviews in the Context of Signaling Theory
15
2.3.1. Signaling Theory
15
2.3.2. Online Reviews- for Experience Goods- as Signals
16
2.4. Critic & User Online Reviews for Experience Goods
17
2.4.1. The Role of Critics
17
2.4.1.1. Source Credibility & Valence
18
2.4.1.2. Variance
19
2.4.2. User-generated Online Reviews
20
2.4.2.1. Volume & Valence
20
2.4.2.2. Variance
21
2.5. Literature Gap & Research Question
24
2.6 Online Movie Reviews & the "Carry-over Effect"
26
2.7. The Expectancy-Disconfirmation-Performance (EDP) Model
27
2.8. Hypotheses
30
3. Methodology & Research Design
34
3.1. Data Collection
34
3.2. Data Elaboration
35
3.3. Data Cleaning
36
3.4. Measuring the EDP Variables
37
3.5. Analysis Methods
38
3.6. Description of Variables
40
4. Data Analysis & Results
42
4.1. Hypotheses 1a & 1b: The
Relationship between Tonality & Rating42
4.2. Hypotheses 2a & 2b:
The Relationship between Disconfirmation & User Satisfaction43
4.3. Hypothesis 3:
The Different Effects of Rating & Tonality in the EDP Model45
4.4. Hypotheses 4a & 4b:
The Moderating Role of Variability48
5
5. Conclusions & Discussion
51
6. Managerial Implications
56
7. Limitations & Future Research
57
8. References
58
6
1. Introduction
“We need to see ourselves projected in other members of our species to, in turn, understand ourselves. Cinema, is that mirror. It is a bridge between the others and us.” ― Alejandro González Iñárritu
"Making a big Hollywood film that really affects people is as hard as making a small movie on a credit card." ― Cameron Crowe
"I'll tell you, I think that the Internet has provided an enormous boost to film criticism by giving people an opportunity to self-publish or to find sites that are friendly." ― Roger Ebert
The internet is unquestionably an integral part of everyday life for consumers, as a free community for the exchange of ideas and as the leading source of information. In particular, online reviews have become an indispensable tool for consumers to share their opinions and consider those of others before making a purchase. According to a survey by Bright Local (2014) cited by Forbes magazine, 88 percent of consumers trust online reviews as much as a personal recommendation. Similarly, Nielsen’s recent Global Trust in Advertising Report (2015) revealed online consumer reviews are the second most trusted source of brand and product information. In fact, research shows that 61 per cent of customers read online reviews before making a purchase decision (eConsultancy, 2015). In view of such inquiries, online reviews constitute an important facet of electronic-word-of-mouth (eWOM).
This holds especially true in the context of experience goods that fall under the creative industries. Already back in 2000, a report by Forrester Research in Cambridge revealed that approximately 50% of young internet surfers were using word-of-mouth recommendations to purchase experience goods such as CDs, movies, videos or DVDs, and games (Walsh, 2000). This statistic has grown significantly, especially considering the increase of user online reviews in the past decade. When purchasing experience products, more and more modern consumers are making their purchase decisions through eWOM on third-party websites (Zhu and Zhang, 2010).
This thesis looks specifically at online movie reviews, bearing in mind the importance of the film industry within the creative industries and the entertainment market on the whole. According to the Global Entertainment and Media Outlook 2016-2020 by PricewaterhouseCoopers, the entertainment market is expected to reach a worth of 2.14 trillion U.S. dollars in 2020. The world-wide entertainment and media market includes both digital and non-digital platforms and encompasses every broadcasting medium such as newspapers, magazines, TV and radio; and popular forms of entertainment such as:
7 movies, music, video games and books. Specifically, the global movie production and distribution industry is currently estimated at around $89 billion according to IBISWorld.
An advantage of the creative industries is the availability of large amounts of data, which includes online reviews on publicly available databases and review aggregate websites, such as
Metacritic.com, imdb.com, Rottentomatoes.com, etc. The web offers an incredible abundance of online
review sources, however, this study utilizes online movie reviews by both users and critics acquired from one specific platform: Metacritic.com.
The online UC Berkeley Film Studies library defines movie reviews as "assessments of the aesthetic, entertainment, social and cultural merits and significance of a current film." As online reviews for movies- and experience goods in general are hedonic in nature, these assessments rely greatly on subjective judgments and emotions (Gazley et al., 2011). Consumers take into consideration these subjective judgments and the perceived fit between the product and their tastes. Accordingly, the main function of an online movie review is to provide consumers with detailed and reliable information of past consumption experience; or so-called subjective "quality cues," (Troilo, 2015) regarding the focal movie. This study takes a signaling theory perspective where user and critic reviews (combined into one construct) act as "signals of a movie's quality" and therefore diminish the inherent information asymmetry between consumers (movie goers) and producers (film industry).
In the context of experience goods, studies mainly focus on how valence and volume- and to a lesser extent variability affect sales. This study on the other hand, examines the factors that determine user satisfaction. The focus herein involves the subtle antecedents that determine online reviews themselves, specifically in cases where carry-over effects are relevant.
Most studies treat determinants of product evaluations as being directly related to the underlying product. However, online reviews can also be regarded as "peripheral signals" not directly linked to the product being evaluated. For instance, Situmeang, Leenders and Wijnberg (2014) argue that peripheral signals that reflect performance of earlier editions also play a role in determining the evaluation for later editions in their study on sequels. Their study calls for further exploration of the so-called Pandora's "box of key carry-over mechanisms" where peripheral cues play a role.
8 further with a focus on movies by specific directors rather than sequels in a series. In the case of chronological movie installments by a creative agent like a particular director, previous creative works collectively reveal "patterns of performance" called "career trajectories" (Zickar and Slaughter, 1999: p. 212). Accordingly, these career trajectory patters signal a history of quality. Consequently, the overall reception of earlier works collectively, act as peripheral cues-, that in turn, create expectations regarding the latest work by a particular creative agent. This study thus takes online reviews of earlier movies by a particular director as peripheral quality signals of a director's past creative activities. The periphery signals used in this study are purely valence measurements and include both rating and tonality scores.
This study first focuses on the valence of earlier online movie reviews and their effect on user satisfaction. Most studies use only rating scores to measure valence direction. However, as emotions have considerable influence on the evaluation of hedonic goods (Ladhari, 2007; Bassi, 2010; Lee et al, 2016), emotional responses embedded in online movie reviews- are not properly captured by a rating score. This study is unique in that it also uses tonality scores to capture the valence of earlier reviews as peripheral cues that impact user satisfaction. By using tonality, this thesis therefore answers the call for more studies to employ textual data mining tools that generate review scores based on the content of the review (King, Racherla & Bush, 2014; Grabner-Kräuter & Waiguny, 2015; Situmeang et al, 2014).
Moreover, the expectancy-disconfirmation and performance1 (EDP) model is adopted to capture the "signaling effect" that in turn creates expectations for the latest movie installment. In marketing, the EDP paradigm has been used to study the antecedents of satisfaction (Oliver, 1980; 2010; Anderson and Sullivan, 1993). Seeing as the dependent variable in this study is user satisfaction (for the last movie) the EDP provides the conceptual framework to the test the hypotheses.
The second focus of this study deals with the degree of critic and user consensus of previous online movie reviews. Variability remains an understudied feature of online reviews. Given that earlier movie reviews are treated as peripheral signals, herein, variability raises the issue of signal consistency- (Gao, Darroch, Mather, & MacGregor, 2008). Expectations will be weak for the last movie, if reviews of earlier movies show inconsistency. Accordingly, a lack of consensus should decreases the effect of online movie reviews for previous movies- as "quality signals"- on user satisfaction and vice-versa.
1 It should be noted that performance herein is not related to the success of the movie in financial terms. In the EDP
9 On this basis, the main research question is as follows:
Through expectancy-disconfirmation-performance as the mechanism - how do online reviews of earlier movies affect user satisfaction - of the latest movie by the same director?
The two sub-questions are as follows:
1. As peripheral quality signals and measurement s of satisfaction, to what extent do rating and tonality exp ress themselves differently in the expectancy -disconfirmation -performance model?
2. Does the variability in online reviews of previous movies by a particular director influence (moderate) the relationship between expectations and user satisfaction for the l ast movie?
The aim is to investigate how online reviews of earlier movies through, disconfirmation, mediate the relationship between performance (the overall quality of the last movie) and user satisfaction (last movie). Given that affective cues are registered more rapidly than cognitive assessments; (Gorn, Pham, & Sin, 2001) the affective content embedded in a tonality score, should have a stronger "signaling effect" than a rating score in terms of disconfirmation. To what extent this holds true and how rating and tonality differ as periphery cues (of earlier movies) and measurements of satisfaction (latest movie) will be investigated. Moreover, this study also tackles variability, an understudied aspect of online reviews. The objective here is to examine if a lack of consensus in earlier movie reviews affect their strength as peripheral signals.
Thesis overview
This thesis is structured as follows: Chapter 2 provides the literature reviews where critic and user online reviews for experience goods will be discussed in the context of signaling theory. Chapter 2 culminates in the literature gap, the research questions are introduced along with the theoretical framework (EDP model) and hypotheses. Chapter 3 explains the methodology, research design and describes the variables used in testing the hypotheses. The results of the research will subsequently be presented in Chapter 4. The empirical findings are discussed in the Chapter 5 along with the conclusions. Afterward, the managerial implications will be discussed; and finally the limitations and suggestions for further research are provided.
10
2. Literature Review
This section first places movies as experience goods under the creative industries. Subsequently, online reviews are discussed in general terms along with the role of emotions. Moreover, online reviews for experience goods are placed in the context of signaling theory; and both critic and user generated reviews are discussed in relation to their characteristics. The literature study culminates in a literature gap and research questions are reiterated. The constructs in the expectancy-disconfirmation and performance (EDP) model will be introduced; and a brief academic discussion on satisfaction will be provided in the context of the model. Finally, the hypotheses will be introduced.
2.1. The creative industries & movies as experience goods
The literature emphasizes that consumer attitudes are inherently bi-dimensional since consumers purchase goods and services for both affective (hedonic-experiential) gratification- from sensory attributes; as well as instrumental, (utilitarian) reasons (Batra and Ahtola, 1990; Holbrook and Hirschman, 1982). Utilitarian needs are functional and include: practical, rational, objective, concrete, economic, and cognitive/rational decision-making. Conversely, hedonic needs are experiential as well as transformational and include: emotional, social, non-rational, subjective, abstract, symbolic, sensory, self-expressive, and aesthetic needs (Bassi, 2010).
Accordingly, hedonic products are likely to impact the affective component of consumer attitude in stronger ways than utilitarian products, which are essentially cognitive related (Adaval, 2001). Moreover, utilitarian goods are in most cases search goods whose qualities can be determined based on objective attributes by the consumer before purchase (Nelson,1970). Conversely, hedonic goods are in most cases experience goods whose qualities cannot be determined prior a purchase. Consumers can solely evaluate the quality of experience goods "through direct personal experience" (Troilo, 2015: p. 6). It should be noted that hedonic-experiential and utilitarian aspects of products and services are not mutually exclusive since consumers purchase goods that combine both motives (Dhar and Wertenbroch, 2000). For instance, a Tesla, as an incredibly powerful fancy electric sports car, serves both instrumental and hedonic purposes. A documentary movie is also characterized by having both instrumental and hedonic sensory attributes. Even though the consumption of many goods involves both
11 dimensions to varying degrees, (Batra and Ahtola, 1990) consumers evidently characterize some products as primarily utilitarian and others as primarily hedonic/experiential.
On this basis, this thesis deals with box-office-movies which fall under the umbrella of creative industries. The United Nations Educational, Scientific and Cultural Organization (UNESCO) 2 defines cultural and creative industries as “sectors of organized activity whose principal purpose is the production or reproduction, promotion, distribution and/or commercialization of goods, services and activities of a cultural, artistic or heritage-related nature.” Thus, creative industries can be defined as "experience goods" with significant "creative elements" which are "aimed at the consumer market via mass distribution" (Peltoniemi, 2015: p. 41). Moreover, these "creative elements" comprise stories and styles with the purpose of entertainment, identity building and social display. For example: music, movies, video games and novels serve the purpose of entertainment as activities in which people engage for amusement.
To reiterate, the dominant benefit of experience goods is hedonic consumption. According to Hirschman and Holbrook’s definition, hedonic consumption refers to those aspects of consumer behavior that are associated with “the multisensory, fantasy, and emotive aspects of one’s experience with products” (1982: p. 92). Hedonic goods thus have indulgent components and are purchased for the sake of enjoyment, pleasure and emotional arousal. Experiential consumption consequently "emphasizes the affective aspect of human beings" (Pine & Gilmore, 1998: p. 99).
Prior studies have asserted that emotions have considerable influence on the evaluation of experiences (Westbrook and Oliver, 1991; Bigné, Andreu, & Gnoth, 2005; Ladhari, 2007; Bassi, 2010; Lee et al, 2016). Pine and Gilmore (1998) generated an experience economy framework and suggested two criteria: level of customer participation and the environmental relationship- in order to identify four dimensions of experience: entertainment, aesthetic, educational and escapist experience. In the case of movies, an entertainment experience takes place when the consumer passively participates by taking in the experiential setting. This turns into an aesthetic experience when the consumer becomes engrossed in the experience while still passively participating. The consumer "merely enjoys being in the event environment without altering the nature of the environment" (Lee et al, 2016: p. 178). Moreover, the
2
12 aesthetic experience engages the senses, cognitions, emotions ad consists in "a gradual immersion and interaction" with the aesthetic product (Troilo, 2015: p. 71). An escapist experience occurs when the customer participates actively and is immersed in the event. A combination of the four dimensions is the ultimate objective of experience marketing (Pine & Gilmore, 1998).
Accordingly, these four "dimensions of experience" lay the basis for affective-based evaluations of movies. On this basis, the main aspects of experience goods- movies in this case- are their intrinsic personal and memorable characteristics that build upon an individual’s sensations and emotions (Bassi, 2010; Pine & Gilmore, 1998).
2.2. Online Reviews: Definitions, Features and Affective Content
This section introduces online reviews in general terms. This analysis deals with movies, experiential hedonic goods that are inherently affect-laden and for which emotions are important. Therefore, emotional content in online reviews will be discussed prior to placing online reviews for experience goods in the context of signaling theory.
2.2.1. Definitions and Characteristics
Online reviews fall under electronic-word-of-mouth (eWOM). Hennig-Thurau, Qwinner, Walsh and Gremler (2004) provide a comprehensive definition:
"eWOM is 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" (2004: p.39).
Accordingly, consumers interact with one another through eWOM, in order to exchange product-related information, and make informed purchase decisions (King et al., 2014). There exists an array of eWOM platforms, mainly: social networking sites (Facebook, Twitter, Instagram), online discussion forums, online review sites, blogs, and online shopping sites (Cheung & Thadani, 2012). However, marketing scholars regard online reviews as the most important type of eWOM (Cheung and Thadani, 2012; Jiménez and Mendoza, 2013) seeing as they provide consumers with detailed and reliable information of past consumption experiences. According to eConsultancy.com (2015), 61 per cent of customers read online reviews before making a purchase decision. Undoubtedly, most consumers take online reviews into consideration throughout the purchasing process.
13 In view of the above, eWOM has attracted considerable interest among researchers in the past decade. Thus, significant research has been conducted on online reviews in the following areas: consumers' motivations to engage in eWOM (Hennig-Thurau, Gwinner, Walsh, & Gremler, 2004), purchase intentions (Jiménez, Norma, & Mendoza, 2013), product sales (Zhu & Zhang, 2010), retail performance (Gu, Park, & Konana, 2012; Floyd et al., 2014), effectiveness across eWOM valence and product type (Park & Lee, 2009; Zhang, Craciun & Shin 2010), and what makes reviews “helpful” or “of value” to consumers (Mudambi & Schuff, 2010; Schindler & Bickart, 2012).
The result of a recent meta-analysis on how online product reviews affect retail sales concluded that online product reviews have a significantly greater influence on sales elasticities when they are "delivered by a critic, appear on a non-seller website, and include valence information" in the evaluation (Floyd et al., 2014: p. 227). Moreover, some eWOM studies show source expertise to amplify message impact, whereas others show consumers to rely more on non-expert sources (Senecal & Nantel, 2004).
In the context of experience goods, significant research has been directed at a at a better understanding of the effects of online reviews on product sales (Chevalier and Mayzlin, 2006; Liu 2006; Zhang and Dellarocas, 2006; Boatwright et al., 2007; Duan, Gu, & Whinston, 2008; Zhu and Zhang, 2010; Chintagunta et al., 2010; Sun, 2012; Basuroy and Ravid, 2014) and retail performance (Gu, Park, & Konana, 2012; Floyd et al., 2014). The effects of both user and critic online reviews on product sales for experience goods will be discussed in section 2.4. Another area of research that has received attention pertains to the characteristics of online reviews that shoppers find helpful and of value (Mudambi and Schuff, 2010; Schindler and Bickart, 2012). Findings include that moderate reviews are more helpful than extreme reviews (whether they are strongly positive or negative) for experience goods, but not for search goods. Pan and Zhang (2011) find that a review's valence and length have an effect on its helpfulness. Their study also notes that utilitarian and experiential products moderate this effect, with a greater benefit going to experiential goods than to utilitarian products.
The main features of online reviews that studies tend to focus on include: valence, volume, variability, dispersion and source credibility. Valence refers to the direction of the review-positive or negative. Volume refers to the total amount of posted reviews about a particular product or service. Variability captures the heterogeneity in consumer opinions since its shows the degree of agreement
14 among reviewers. Dispersion refers to the extent to which product-related conversations are taking place across a broad range of communities (Godes and Mayzlin, 2004). Source credibility refers to a user's perception of the credibility or reliability of the message source.
These key characteristics of online reviews establish how different effects are measured and quantified (King et al., 2014). A further explanation of these characteristics, particularly: valence, volume and variability will be given in section 2.3.2 in the context of signaling theory. A more detailed discussion of these measures, according to the literature, for both critic and user online reviews will be provided in section 2.4. First, the role of emotions in online reviews will be discussed.
2.2.2. Role of Emotions in Online Reviews
A few scholars have examined the role of emotions in online reviews and their effects on purchase intention (decision-making process)/conversion rates (Ludwig et al., 2013); emotional expressions on product evaluation- (Kim & Gupta, 2012); and how expressed emotions affect the helpfulness of a product review (Yin, Bond & Zhang, 2014; Ahmad & Laroche, 2015).
For starters, Ludwig et al. (2013) examine how positive and negative affect words impacts consumer conversion rates. They use text mining methods to ‘‘extract changes in affective content and linguistic style properties of customer book reviews on Amazon.com’’ (Ludwig et al, 2013: p. 87). They find a positive asymmetrical relationship between positive affective content and conversion rates. They also find that increases in positive emotional content in user reviews have a smaller effect on subsequent increases in purchase decision. Moreover, Ludwig et al. (2013) emphasize the need to further study content and style collectively since this can reinforce the impact of a review.
Studies conducted by Yin, Bond & Zhang (2014) and Ahmad & Laroche (2015) on how expressed emotions affect the helpfulness of a product review depart from a solely valence analysis by using Discrete Emotional Theory. Ahmed & Laroche (2015) build on cognitive appraisal theory to examine how discrete emotions (e.g., hope, happiness, anxiety, and disgust) embedded in the reviews affect the helpfulness votes of potential customers. They find that discrete emotions have differential effects on the helpfulness of the reviews. Similarly, Yin, Bond & Zhang (2014) explore the effects of emotions embedded in a seller review on its perceived helpfulness to readers. “Above and beyond a negativity bias,” they argue “the specific affective content in online reviews plays a major role in
15 determining their helpfulness” (Yin et al., 2014: p. 541). These studies demonstrate the importance of examining discrete emotions in online word-of-mouth, as they carry important practical implications for consumers and online retailers.
Finally, Kim and Gupta (2012) investigate how consumers interpret emotional expressions in online user reviews and how this impacts their product evaluations. In their investigation, the authors focus on a single product category: computers. Their results show that negative emotions in a single review decrease informative value and decrease negative impact on product evaluations. Moreover, the authors suggest future studies focus on hedonic products-since computers are single utilitarian products for which emotions may be unimportant. Kim and Gupta argue that exploring the effects of emotion provide "richer insights into how readers interpret affective content in online reviews" (2012: p. 990).
2.3. Online Reviews in the Context of Signaling Theory
The scope of this thesis will be discussed at length in sections 2.5, 2.6 and 2.7. However, since the analysis herein stems from a signaling theory approach, this section places the main characteristics of online reviews vis-à-vis signaling theory.
2.3.1. Signaling Theory
According to Conelly et al. (2011) signaling theory provides "a unique, practical, and empirically testable perspective on problems of social selection under conditions of imperfect information" (p. 63). In essence, signaling theory is primarily concerned with reducing information asymmetry between two parties (Spence, 2002). The notion of information asymmetry is of particularly relevance in the "study of marketplace exchanges" (Kirmani & Rao, 2000: 66). The most influential work that uses signaling theory to date was conducted by Spence (1073) who modeled "the signaling function of education" by using employer uncertainty regarding the abilities of job candidates.
The main three concepts in the signaling process are: the signaler- "an insider who obtains information about a product"- who in turn sends information of a product's quality- or a signal-to a
receiver- "an outsider who lacks information- who in turn observes and interprets the signal" (Conelly et
16 In the context of marketing and consumer behavior, information asymmetry can represent consumer uncertainty about the quality of the product. Several studies have applied signaling theory in areas of marketing such as: Kirmani & Rao (2000)- in conveying product quality information; Gammoh, Voss, & Chakraborty (2006)- in consumer evaluation of brand alliance signals; Basuroy, Desai, & Talukda (2006)- in sequels and advertising expenditures as signals in the motion picture industry on box office revenues; and Situmeang et al. (2014)- who examine online reviews from previous installments as signals and determinants for online reviews of subsequent editions in product series. Overall, these marketing studies that use signaling theory, position customers as receivers. In particular, Situmeang et al. (2014) identify both past sales performance as well as online reviews as "signals of quality."
2.3.2. Online Reviews for Experience Goods as Signals
In the context of signaling theory, user and critic reviews act as signals of a product's unknown quality and therefore diminish the inherent information asymmetry between consumers (movie goers) and producers (film industry). It should be noted that as signals, most studies treat online reviews as independent determinants directly linked to a focal product (Situmeang et al., 2014). However, in some particular cases, online reviews can also be regarded as peripheral signals not directly linked to the product being evaluated. This notion and usage of earlier online reviews as subtle determinants of product evaluation for sequential products, in particular, has to do with the carry-over-effect. This will be discussed in Section 2.6. in the context of the research question.
Consumers face uncertainty for experiential products such as movies because of their sensory nature and the need for direct experience (Troilo, 2015). Given that memorable experiences are highly personal, they rely greatly on subjective judgments and emotions (Gazley et al., 2011). In general, consumers take into consideration these subjective judgments or "quality cue indicators" (Troilo, 2015: p. 6) expressed in online reviews for experience goods. However, the various characteristics of online reviews measure, and therefore- "signal" different aspects. In particular, valence, volume and variability establish how different effects of online reviews are measured and quantified (King et al., 2014).
First, valence- the positive or negative communication direction- signals consumer attitude regarding the quality of the product. The theory behind measuring valence, or consumer attitude, is that positive opinions will encourage other consumers to adopt a product whereas negative opinions will
17 discourage them (Dellarocas & Zhang, 2007). Volume- the total amount of posted reviews-signals popularity and thus, creates awareness for products (Liu, 2006). As Liu (2006) asserts, “[the] greater the volume of eWOM, the more likely a consumer will be able to hear about a product. Not surprisingly, greater awareness tends to generate greater sales” (p. 77). Regarding variability, the issue of signal consistency-the agreement between multiple signals from one source (Gao, et al., 2008) is raised. "Conflicting signals confuse the receiver, making communication less effective" but signal consistency can help mitigate this problem (Conelly et al, 2011: 54). Accordingly, a lack of consensus should decreases the effect of online movie reviews as "quality signals."
The following section expands on the characteristics of online reviews (regarded herein as signals) according to the literature. Both critic and user generated online reviews for experience goods will be discussed in relation to these key measures of online reviews.
2.4 Critic and User Online Reviews for Experience Goods
In the context of experience goods, both critic and user generated online reviews- regarded herein as signals of product quality- have attracted considerable interest from researchers. This section discusses the role of both critic and user generated online reviews in relation to the relevant characteristics. In particular, the literature focuses on the effect of volume, valence and to a lesser extent variability on sales.
2.4.1. The Role of Critics
The role of critics is a major focus in marketing research. This section discusses the role of critics in relation to source credibility, valence and variability (as volume is not relevant). The creative industries have historically been marked by a cultural hierarchy (Holbrook, 1999) with critics having the position of cultural gatekeepers (Caves, 2000). Seeing as movie critics are typically invited to early screenings, they generally write reviews before the film is available to the general public. For this reason, the influence of critics on consumer judgments is considerable as critics are generally one of "the first sources to diffuse information about new products" (West and Broniarczyk, 1998: p. 38), and their endorsements are used in advertising.
18
2.4.1.1. Source Credibility and Review Valence
Their professional status as experts and connoisseurs gives critics credibility. As such, the "source credibility" characteristics for online reviews is presupposed in regards to movie critics. However, Camara and Dupuis (2015) find that critics adopt reporting strategies that are consistent with the predictions of the literature on reputational cheap-talk (Ottaviani and Sørensen, 2006). Reputational cheap-talk, demonstrates that experts disregard "noisy signals" and conform to prevailing opinions in order to be recognized as good predictors. For instance, Camara and Dupuis (2015) find strong variations in movie critics’ abilities and biases with information manipulation occurring on average 10% of the time. Interestingly, this shows that critics are concerned with their reputations and they face incentives to manipulate demand.
There are numerous studies pertaining to the role of critics in the movie industry. As stated, most studies focus on the impact of critical reviews (valence) on movie revenues in the context of a film's box office sales (Holbrook, 1999; Basuroy, Chatterjee, and Ravid, 2003; Reinstein & Snyder, 2005; Boatwright, Basuroy & Kamakura, 2007).
For starters, Eliashberg and Shugan (1997) contrasted two standpoints: critics as influencers and critics as predictors. As influencers, the critics' expert opinion- valence of their review- drive attendance; and as predictors, they foretell how a movie would perform at the box office (Eliashberg and Shugan, 1997). Accordingly, critics with opinions that are correlated with early box office sales are characterized as influencers, while critics with opinions that are correlated with overall sales are identified as predictors. In their study, Eliashberg and Shugan (1997) state that the prediction role is more important. Basuroy et al. (2003) show that critics provide both roles; as revenues throughout the run of eight weeks and not just opening weekend are correlated with critical reviews. Other studies (Reinstein & Snyder, 2005: 30; Boatwright et al., 2007) on the other hand, find a more significant influencer effect.
According to Eliashberg and Shugan (1997), critics that are predictors have opinions that correlate with the tastes of their readers. Holbrook (1999) shows that in the case of films, consumers and professional critics emphasize different criteria when forming their tastes. However, shared taste between consumers and critics is statistically significant in their analysis, (albeit a weak correlation). Hence, the differences the author mentions are not significant enough to disprove shared taste (Holbrook, 1999). On
19 the other hand, Plucker, Holden, and Neustadter (2008) found evidence that professional critics’ ratings did not differ significantly from those of users.
Boatwright et al., (2007) argue that critics may still be influential even when having different opinions (expressed in valence ratings) from their audience. Critic reviews provide information about movies that permits consumers to form their "own quality expectations" separately from the general opinion of critics; so critics are still influential in a market whose "tastes do not correlate with their own" (Boatwright et al, 2007: p. 420). Reinstein and Snyder (2005) also find evidence of an influence effect (albeit weak) and that this effect differs across categories of movies. Also, they deem that expert reviews can be an important mechanism for transmitting information about goods of uncertain quality.
Basuroy, Chatterjee, and Ravid (2003) explored how critics affect the box office performance of films and how star power and big budgets moderate these affects. Three important findings of this study include: 1) The valence- both positive and negative critic reviews- are significantly correlated with box office revenue for the first eight weeks (this is consistent with their role as both influencers and predictors); 2) negative reviews have a progressively smaller influence over time, and there is an asymmetrical impact since negative reviews hurt revenue more than positive reviews help revenue in the early weeks of a film's release; 3) Star power and big budgets significantly decreases the impact of negative reviews but do not increase profits for positively reviewed films.
2.4.1.2. Variance in Critic Reviews
The direct effect of variability in critic reviews has received less attention in the literature (with the exception of West and Broniarczyk, 1998). As an offshoot of their study on the influence of individual critics Boatright et al., (2007) find that "critical consensus is positively correlated with market potential" (Boatwright, 2007: p. 414). West and Broniarczyk (1998), use prospect theory to examine consumer attitudes toward critic consensus for movies and restaurants. They find that consumers are sensitive to both the degree of consensus among the critics vis-à-vis the average critic rating in forming purchase intentions. In regards to movies, they find evidence that a higher variance increases the chance of purchase but only if the average rating is below an aspiration level (West and Broniarczyk, 1998). This is because variability with a low average critic rating offers consumers the opportunity to meet or exceed expectations, since a few critics evaluate the product favorably. Conversely, when the average
20 critic rating is high, consumers prefer consensus in critic opinions to avoid unfulfilled expectations (West and Broniarczyk, 1998).
2.4.2. User-generated Online Reviews
User reviews tend to reflect user experience and consumer satisfaction, which are mainly viewed as a source of product information (Li and Hitt, 2008). In recent years, posted-user-reviews have multiplied and become ubiquitous for both search and experience products. Quite a few studies have tried to evaluate the impact of user reviews on the success of movies (Liu 2006; Zhang and Dellarocas, 2006; Chintagunta et al., 2010; Duan et al., 2008; Basuroy and Ravid, 2014). Thus, similar to the literature examining the role of critics, user-generated online review studies focus on its impact on sales (box office success).
Basuroy and Ravid (2014), who examined both expert opinions and user reviews, find that while "user reviews matter, expert opinions carry more weight." However, comparing the effect of professional critical reviews in the presence of ever-present user generated content is not relevant. As the literature shows that box office sales are influenced by volume, suggesting the importance of the awareness effect among users; as well as valence to different extents. However, as will be discussed below, there is discrepancy in the literature as to which measure of online review is more significant.
2.4.2.1. Volume and Valence in User-generated Online Reviews
For user generated online reviews (in experiential goods) there is disagreement in the literature as to whether volume or valence is more important in driving sales. Scholars agree that both of these characteristics matter but there is no consensus as to what degree volume matters over valence, and vice versa. Some studies have found that sales in creative industries are influenced to a greater degree by volume more so than by valence, some find the opposite effect especially when controlling for a movie's quality as well as other market fixed effects.
In particular Liu (2006) examined the effects of online reviews posted on Yahoo Movies on box office sales during the first week of the premier of 40 movies. Liu's findings showed that sales are influenced to a greater degree by the awareness of the film among consumers, generated by the number of posted user comments (volume) than by the opinions expressed by consumers (valence). Moreover,
21 Chen, Wu & Yoon (2004) found that consumer ratings- valence- are not correlated with sales for books using amazon.com. Similarly to Liu- they find that volume- number of consumer reviews is positively associated with sales.
Chintagunta et al. (2010), test the three measures of online reviews: volume, valence and variability, to distinguish their effect on revenues. After controlling for advertising, distribution, competition, age, and movie quality, however, they find that only the valence of user ratings has a positive and statistically significant impact on box office revenues. Similarly, in their study whimsically titled The Lord of the Ratings, Zhang and Dellarocas (2006), examine both the impact of critic and amateur user reviews on the box office performance of movies. Their results reveal that valence measures- (star ratings) of online reviews have significant impact on sales. In fact, when controlled for quality, they found that propensity to write reviews-volume- is not significant.
In an attempt to reconcile this debate, Duan et al. (2008) show that volume and valence occupy different roles in influencing product sales. Taking into account the dual nature of online reviews as a
precursor to and an outcome of product sales; they find that "valence indirectly increases revenue by
generating higher volume" (Duan et al, 2008: p. 241). Essentially, they find that even if the rating of user online reviews have little impact on movies' box office revenues, they find that volume does. Like Liu's study, this also suggests that volume generates knowledge of a movie among consumers implying the importance of awareness effect (Liu, 2006; Duan et al, 2008).
Zhu and Zhang (2010) find that for video games, online reviews are more influential for less popular games. As such, their results show that for niche experiential products, the impact of online consumer reviews on product sales depends on product and consumer characteristics. As such, valence impacts the success of niche products more so than any other measure, as even a few negative reviews tend to be decrease sales in their study.
2.4.2.2. Variability in User-generated Online Reviews
Volume and valence are the measures of user online reviews that dominate in the literature. It is implied from the literature that consumers regard high variance experiential products as niche products, that some enjoy and other dislike. There are a few studies that directly examine variability, degree of
22 consensus and the distribution of ratings in user online reviews for experience goods (Sun, 2012; Zhang and Dellarocas, 2006; Situmeang et al., 2014).
In her study of books on Amazon.com, Sun (2012) explores the interaction of the variance of ratings and the average rating to analyze the extent to which consumers differ from each other in their enjoyment of particular books. Sun ascribes this variation to the underlying product characteristics. Similar to the study conducted by West and Broniarczyk (1998) on critic variability, Sun finds that books with a low average user ratings can still be profitable given that the variance of ratings is high. In the case of Amazon.com, Sun finds that "a high standard deviation of ratings increases a book’s relative sales when average rating is lower than approximately 4.1 stars" (Sun, 2012: p. 697).
On the other hand, Zhang and Dellarocas (2006) examine the impact of amateur reviews on the box office performance of movies, but finds that variability of movie reviews does not influence box-office performance in the early weeks of the movie's premier. The level of disagreement associated with a movie is not significant for most specifications, which suggests that review variance is not an influential variable in the early stage of a movie’s life cycle (Zhang and Dellarocas, 2006).
Situmeang et al., explores the effects of variability of both expert and user reviews not on sales but as determinants of the review of subsequent editions in a sequel.3 A lack of agreement in both expert and user reviews tends to diminish "the transferability of the communities’ appreciations" in a series of editions from one to the next for both experts and users (Situmeang et al, 2014:81). However, as online reviews are but one signal of quality, the issue of signal consistency (consensus among user and critic reviews) works the same way. For instance, in their study of video game sequels, Situmeang et al. (2014) showed that high variability in expert and user reviews diminishes "the transferability of the communities’ appreciations" in a series of editions from one to the next (Situmeang et al., 2014: p. 81).
3
23
Table 1: Literature review on the effect of online reviews on sales for movies and experience goods
Articles Critic Reviews User Online Reviews
Characteristics Effect on sales
OR Not OR Volume Valence Variability
West & Broniarczyk (1998)
✔ ✔ ✔A higher variance increases the chance of purchase but only if the average rating is below an aspiration level.
Elberse and Eliashberg (2003)
✔ ✔ ✔Less positive reviews correspond to a higher number of opening screens, but more positive reviews mean more opening revenue.
Basuroy et al. (2003)
✔ *entertainment
magazines*
✔
✔ Valence has comparable effects: Negative reviews are more harmful to a film's box office performance than positive ones are helpful. ✔Negative reviews have lesser influence over time. Chen, Wu & Yoon
(2004)
✔ ✔ ✔ ✔Consumer ratings are not correlated with sales for books using
amazon.com.
Snyder (2005) ✔ *Siskel & Ebert*
✔ ✔Critics’ influence on opening weekend box office revenue is smaller than previous studies would suggest but is still significant. Chevalier and
Mayzlin (2006) ✔ ✔
✔Improvement in reviews for a book at one site leads to a relative increase in the sales. ✔Marginal (negative) impact of 1-star reviews is greater than the(positive) impact of 5-star reviews on sales. Liu (2006) ✔ ✔ ✔ ✔Volume offers significant explanatory power for both aggregate
and weekly box office revenue in the early movie opening week. Zhang and
Dellarocas (2006) ✔ ✔ ✔ ✔ ✔
✔Significant influence of valence measure (star ratings) but volume is not significant once controlled for quality. ✔Variance does not play a significant role in the early weeks.
Boatwright et al. (2007) ✔ *Variety Magazine* ✔ ✔
✔ Critics marginally impact sales when estimated individually than aggregately. ✔Some critics are more influential. ✔Critical consensus is positively correlated w/ market potential.
Duan, Gu, &
Whinston (2008) ✔ ✔ ✔
✔Box office revenue and valence greatly influences eWOM volume. ✔The rating has no significant impact on movies’ box office revenues; they affect it indirectly through volume.
Zhu, and Zhang (2010)
✔ ✔ ✔ ✔For video games, online reviews are more influential for less popular and online games.
Chintagunta et al.
(2010) ✔ ✔ ✔
✔Main driver of box office performance is the valence not the volume of reviews.
Sun (2012) ✔ ✔ ✔ ✔ A higher variance in ratings improves a book’s relative sales rank when the average rating but only when it is lower than 4.1 stars. Basuroy and Ravid
(2014)
✔ ✔ ✔ ✔ ✔Critics remain more influential than consumer ewom in predicting box office revenue.
24
2.5. Literature Gaps and Research Questions
As seen in Table 1, studies on online reviews for experience goods primarily focus on their effect of sales using 2 main measurements: valence (in terms of numerical ratings) and/or volume in their analysis (Liu, 2006; Duan et al, 2008; Basuroy et al, 2003; Chintagunta., et al., 2010). However, studies that focus on movie ratings may not be sufficient to capture the impact of a wide variety of emotions described in consumer reviews for experience goods (Ahmad and Laroche, 2015). Scholars agree that a further understanding of content (with the use of text-mining techniques) remains an understudied aspect of online reviews (King et al., 2014; Grabner-Kräuter & Waiguny, 2015; Situmeang et al, 2014). As King et al. (2014) put it:
"Numerical ratings allow consumers to communicate product performance related to various attributes, it is the actual text that provides them with the opportunity to articulate the nuances of their overall experience and convey useful information" (p. 175).
Moreover, there are only a few studies that analyze the role of affective content in online reviews as discussed in Section 2.2.2. (Ludwig et al., 2013; Kim and Gupta, 2012; Yin et al., 2014; Ahmad and Laroche, 2015). In particular, Kim and Gupta (2012) examine how consumers interpret emotional expressions in online user reviews for a single utilitarian product category: a computer. However, the generalizability of their work is limited since computers are utilitarian products for which emotions may be unimportant. On this basis, Kim and Gupta suggest future work should be done on the impact of "eWOM emotions on the evaluation for hedonistic emotion-laden products" (2012: p. 990).
This thesis aims to fill this research gap by examining emotional content in online reviews- and ultimately comparing the impact of both rating and tonality scores embedded in online reviews- on user evaluation/satisfaction.
Tonality refers to the degree of intensity embedded in the textual content of the review-where a
high number is associated with a more positive, upbeat style; and low numbers suggest greater negative. A lack of emotionality or different levels of ambivalence can also occur with neutral scores. Accordingly, this thesis answers the call to employ textual data and mining tools that generate review scores based on the content of the reviews by using a tonality4 score.
4 A tonality score was generated using Linguistic Inquiry and Word Count (LIWC2015). This will be further
25 The second research gap this study aims to fill deals with the degree of online review consensus. As previously discussed, most studies examine the effect of volume and valence on sales- while fewer studies examine variability in online reviews (Sun, 2012; Zhang and Dellarocas, 2006; Situmeang et al., 2014). However, instead of focusing on community consensus, this analysis examines if the degree of consensus in reviews impacts the signaling strength of earlier online movie reviews. As peripheral signals of quality for future installments, online reviews for earlier products should become less relevant where variability in reviews is high.
As such, the scope of this study covers two key characteristics of online reviews: valence (as expressed by not only rating but also tonality) and degree of consensus (variance).
To reiterate from the introduction, the main research question is:
(Through expectancy-disconfirmation-performance as the mechanism) how do critic and user online reviews of earlier movies affect user satisfaction- of the latest movie by the same director?
The two sub-questions analyzed herein are:
1. As peripheral quality signals and measurements of satisfaction, to what extent do rating and tonality express themselves differently in the expectancy-disconfirmation and performance model?
2. Does the variability in both user and critic reviews of previous movies by a particular director influence/moderate the relationship between expectations and user evaluation/satisfaction of the last movir?
Rather than focusing on the effect of online reviews on sales like most studies, the analysis herein examines how previous online reviews of earlier movies affect user satisfaction of the latest movie by a particular director. Accordingly, this thesis takes a similar approach to Situmeang et al's (2014) study on video games sequels; since previous online reviews are treated in part as determinants of future product reviews.
To capture the signaling effect, this study will integrate two expectancy-disconfirmation and performance models (see section 2.7). One using rating and the other using tonality as signals. The aim is to provide insights regarding how emotional content in online reviews (an understudied aspect of EWOM)- captured by tonality- leads to expectations; and how disconfirmation based on previous movie ratings and another based on tonality scores- mediates the relationship between performance/overall
26 quality (of last movie) and user satisfaction/evaluation (for last movie by a particular director). Moreover, (as discussed in the methods) it should be noted that user and critic reviews are combined into one construct to signal the quality of the product being evaluated: movies-sorted by particular directors.
2.6. Online Reviews as "Peripheral Signals" and the "Carry-Over Effect"
Exploring the effect of online reviews of previous movies by a particular director on the latest movie by that director leads to carry-over effects. Most studies on online reviews identify quality standards as being directly linked to the product being evaluated- but that is not always the case. In some cases, like in this analysis, online reviews can also be regarded as peripheral signals not directly linked to the product being evaluated. For instance, Situmeang, et al. (2014) argue that in the case of sequels-each new edition is strongly associated with previous editions in a series so that other determinants apart from quality of the underlying product play a part in the evaluation . They emphasize that for sequels: "signals that pertain not to the product itself but to previous editions (and the performance) can affect the review of the focal product" (Situmeang et al., 2014: p. 74).
Considering consumer and critic reviews, as "peripheral signals" of the quality as perceived by consumers, one can expect some form of continuity in the average response of consumers regarding the latest movie by same director. Performance/overall quality of the last movie by a particular director builds up anticipation and expectations regarding later movies. However, unlike sequels where ideas and expressions connected to earlier editions are transferred to the next edition, movies are stand alone products. Also, even movie sequels usually change directors throughout the entire-run of the franchise- like in the Harry Potter Movies (4 different directors for the 8 installments). Generally, it is rarer for one particular director to be attached to the same franchise than not, but it can also be the case; for instance: Lord of the Rings and The Hobbit trilogy (Peter Jackson); the Star Wars franchise (George Lucas) a Sam Raimi's Spider-Man film trilogy (2002, 2004, and 2007).
In sequels, the performance of new editions is somewhat determined by the extent that there is a carry-over of the consumers’ image of the past editions to the next editions (Keller, 2003). Although they tend to be stand alone products, a carry-over effect is still present when evaluating the last movie by a particular director. A lot of movies tend to be strongly associated with its director, and a director's star power carries a lot of weight (Basuroy et al.; 2003). Similar to sequels, therefore, past performance track
27 records can signal predicted future behavior (Perks and Halliday, 2003) for a particular movie director. As such, past reviews are also signals of consumers and experts reception of a director's creative activity. Accordingly, the carry-over effect is closely related to expectations in this analysis. By examining earlier films by a particular director over the span of his or her career; one can recognize a director’s creative activity more clearly, than would be possible by merely examining individual films. As a result, this creates a "history of quality" of a particular director's overall creative activities. It is argued herein that this overall quality cue, generates expectations that affects user satisfaction for a director's latest movie.
2.7. Expectancy-Disconfirmation and Performance model
T
he expectancy-disconfirmation and performance (EDP) model is adopted from marketing and consumer behavior studies (Oliver, 1980, 2010; Anderson and Sullivan, 1993) to capture the "signaling effect" that in turn creates expectations for the latest movie installment.As seen in Figure 1 below, the four main constructs in the model are: expectations, performance (overall quality), disconfirmation, and satisfaction. The manner by which each construct is measured in this analysis will be explained in the methods.
Figure 1: Expectancy-disconfirmation and performance model
Source: Oliver, R. L. (2010). Satisfaction: A behavioral perspective on the consumer. Second edition.
Accordingly, the EDP model suggests a dynamic relationship between prior expectations of consumers, post-experience perceptions of the performance (or overall quality) of the good or service, the disconfirmation (either positive of negative) of these prior expectations based on performance and; finally, resulting satisfaction (Morgeson 2012:).
28 The construct expectations denotes the pre-purchase initial anticipation about a product's overall performance/quality. The model theorizes expectations as a determinant of satisfaction because expectations provide a standard or reference level (Bhattacherjee, 2001). Consumers subsequently form evaluative judgments about the focal product or experience based on expectations (Oliver, 2010). In this study, online reviews of previous movies act as peripheral signals that consequently, endow moviegoer-consumers with expectations regarding the latest film by a particular director.
Performance denotes the overall quality and inherent attributes of the focal product or
experience good. Several studies in the marketing literature have empirically tested and verified the direct impact of performance on satisfaction (Ladhari, 2007; Hui and Ho, 2007). It is established in the literature that there is a positive relationship between performance (overall quality) and satisfaction since consumers are satisfied when the product performance is favorable, and vice versa. Moreover, as seen in Figure 1, performance in the EDP model affects consumer satisfaction directly and indirectly as the relationship is also mediated by the degree of disconfirmation. Thus, according to the model- performance is expected to have a direct effect on CS- in addition to the indirect effect through disconfirmation.
Disconfirmation of expectations is the central component of the model. Disconfirmation is
calculated by subtracting performance from expectations (P-E); accordingly both expectations and performance determine the disconfirmation concept. Moreover, disconfirmation is a dichotomous variable. Positive disconfirmation occurs when performance is higher than consumer's expectation- implying better-than-expected performance. Similarly, negative disconfirmation occurs when performance is lower than consumer's expectation- implying worse-than-expected performance (Oliver, 2010). Thus, consumer satisfaction is hypothesized primarily as a function of disconfirmation, which affects the relationship between performance (overall movie quality of last movie) and satisfaction (user evaluation of the last movie). However, seeing as performance also has a direct effect on satisfaction, it is interesting to compare, the significance of both its direct and indirect (through disconfirmation) effects.
Oliver (2010) defines satisfaction as: "a judgment that a product/service feature, or the product or service itself, provided (or is providing) a pleasurable level of consumption-related-fulfillment, including levels of under-or over-fulfillment" (p. 8). Accordingly, consumer satisfaction is
29 conceptualized with cognitive and emotional components. In examining the antecedents and determinants of satisfaction, Anderson (1973) and Oliver (1980, 2010) conclude that higher expectations, generally lead to lower satisfaction. In the case of movies, accordingly- higher expectations should lead to negative disconfirmation-meaning consumers will be slightly disappointed regardless of movie quality.
Direct effect of performance or indirect effect through disconfirmation on satisfaction?
Prior to introducing the hypotheses, it should be noted that the impact of performance on satisfaction directly or indirectly through disconfirmation has generated different results in the literature. For starters Oliver, (1980, 2010) and Bigné, Andreu, & Gnoth, (2005) show that user satisfaction increases when performance is higher than expectations, and decreases when expectation is higher than performance- because of disconfirmation. In their study of recreational theme parks, Bigné et al. (2005) show that consumers’ willingness to pay more for a service is due to disconfirmation and not to performance only; and that disconfirmation influences satisfaction. On the other hand, other studies have shown that performance outperforms disconfirmation as a determinant of satisfaction (Hui and Ho, 2007; Spreng and Chiou, 2002). For instance, in their study of tourist satisfaction, Hui and Ho (2007) found that performance had a strong significant direct effect on satisfaction than no indirect effect through disconfirmation. Similarly, Spreng and Chiou (2002) concluded that performance had a stronger effect on satisfaction directly than through disconfirmation.
30
2.8. Hypotheses
This study stems from insights by Oliver (1980,2010) and Bigné et al. (2005) in order to test the extent to which peripheral signals (online reviews for earlier movies) influence satisfaction through disconfirmation-which in turn should mediate the indirect effect of performance- (overall quality of the last movie) on user satisfaction. Moreover, the signaling affect should be stronger for tonality due to the role of emotions since emotions are known to mediate between cognitive evaluation and satisfaction (Mano and Oliver, 1993; Bigne et. al, 2005; Ladhari, 2007). Moreover, a lack of consensus among user and critic online reviews of earlier movies should diminish the signaling effect.
1. Correlation between tonality and rating
Tonality and rating are both measurements of satisfaction and product evaluation. Due to the hedonic experiential nature of movies, both rating and tonality in online reviews capture the interplay of cognition and affect of users. In particular, a tonality score provides an indication of the affective mental thought process that consumers have before generating a rating score.
Accordingly, these measurements should positively correlate. Hence, a starting point would be discerning the positive correlation between the emotional content (degree of positive or negative intensity) of an online review expressed by tonality- and the rating score-given by the same reviewer. This relationship may seem evident but defining a positive relation between tonality and rating may also reveal differences regarding these measures.
It is therefore hypothesized that:
H1a: User review tonality is positively correlated to user rating.
H1b: Critic review tonality is positively correlated to critic rating.
Moreover, since tonality scores express the affective process that lead to ratings, tonality should be more susceptible to differences in expectations. Tonality could be a stronger indicator of disconfirmation as it is more representative of a consumer's emotional evaluation of a movie than a rating. On this basis, the remaining hypotheses will test for differences between tonality and rating in the EDP model.
