The Effects of Expert and Consumer-Generated Online Reviews on Consumer
Purchase Intention in the Performing Arts
On Stage and Online
Kirsten Elise Dedel - 10183248
August 2017- final version
Statement of originality
This document is written by Student Kirsten Elise Dedel 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
Table of contents
Statement of originality _______________________________________________________ 2 Abstract ___________________________________________________________________ 6 1 Introduction ____________________________________________________________ 7 2 Literature review _______________________________________________________ 11 2.1 eWOM in the performing arts _________________________________________ 11 2.2 (Electronic) word-of-mouth __________________________________________ 12 2.3 Purchase intention and sales _________________________________________ 14 2.4 Types of online reviews ______________________________________________ 16 2.4.1 Consumer-generated reviews _____________________________________ 16 2.4.2 Expert reviews _________________________________________________ 18 2.4.3 Consumer-generated review versus expert reviews ____________________ 19 2.4.4 Consumer-generated reviews combined with expert reviews _____________ 20 2.5 Performance type: popular versus highbrow performances _________________ 21 3 Methodology __________________________________________________________ 25 3.1 Design ___________________________________________________________ 25 3.2 Measurements _____________________________________________________ 27 3.2.1 Dependent variable _____________________________________________ 27 3.2.2 Experiment stimuli _____________________________________________ 27 3.2.3 Control variables ______________________________________________ 28 3.3 Procedure ________________________________________________________ 29 3.4 Sample ___________________________________________________________ 30 3.5 Analyses and predictions ____________________________________________ 31
3.5.1 Consumer-generated reviews _____________________________________ 32 3.5.2 Expert reviews _________________________________________________ 32 3.5.3 Consumer-generated reviews versus expert reviews ___________________ 33 3.5.4 Consumer-generated reviews combined with expert reviews _____________ 34 3.5.5 Type of performance ____________________________________________ 34 4 Results _______________________________________________________________ 37 4.1 Reliabilities _______________________________________________________ 37 4.2 Assumptions ______________________________________________________ 38 4.3 Correlations ______________________________________________________ 38 4.4 Control variables __________________________________________________ 40 4.5 Type of review and valance of the review ________________________________ 40 4.5.1 Consumer-generated reviews _____________________________________ 42 4.5.2 Expert reviews _________________________________________________ 43 4.5.3 Consumer-generated reviews versus expert reviews ___________________ 43 4.5.4 Consumer-generated reviews combined with expert reviews _____________ 44 4.6 Type of performance ________________________________________________ 45 4.6.1 Consumer-generated reviews _____________________________________ 47 4.6.2 Expert reviews _________________________________________________ 48 5 Discussion ____________________________________________________________ 50 5.1 Summary of study results ____________________________________________ 50 5.2 Discussion ________________________________________________________ 52 5.3 Managerial implications _____________________________________________ 54 5.4 Limitations and implications for future studies ___________________________ 55 6 Conclusion ___________________________________________________________ 56
List of references ___________________________________________________________ 57 Appendix 1 – Questionnaire English ___________________________________________ 62 Appendix 2 – Questionnaire Dutch _____________________________________________ 89
Abstract
Some first exploratory insights about eWOM in the performing arts sector have confirmed the relevance of eWOM for theatregoers, but the effects of eWOM on purchase intentions of consumers for performances in the performing arts are not studied yet and thus remain unclear. Therefore, this study aims to examine what the effects of eWOM, in the form of expert and consumer-generated online reviews, on purchase intentions of consumers in the performing arts are. Assumed is that there are differences in the effects on purchase intention between the different types of reviews, namely consumer-generated and expert reviews, the different valance of the review, namely positive and negative, and the different types of performances, namely popular and highbrow performances. These differences are tested through an experiment that manipulated the type of review, the valance of the review and the type of performance, whereby data was collected through online questionnaires among 305 consumers living in the Netherlands. The results show that there only is a statistically significant difference in purchase intention between negative reviews and no review, and between negative reviews and positive reviews. This effect is the strongest for consumer-generated reviews and for consumer-consumer-generated reviews and expert reviews combined; for expert reviews alone the negative effect of negative reviews is less strong. There is no statistically significant difference in purchase intention between positive reviews and no review, and there is also no statistically significant difference in purchase intention between popular and highbrow performances.
1 Introduction
When consumers want to buy a product or service, they will probably first search for input from their network to evaluate the product or service. 82% of the Americans search for recommendations from friends and family when considering a purchase of any kind and 67% is at least a little more likely to purchase a product after a friend or family member shared it via social media or email (Kapadia, 2016).
Word-of-mouth (WOM) - non-commercial communication about a product, service, brand or company between consumers - can thus be of great value for organizations. WOM is discussed in the literature frequently as a valuable instrument for creating awareness,
distributing information and winning new customers. In this social media era, consumers can communicate with each other in many different ways. While with traditional WOM
consumers could only communicate with each other directly in an oral form, consumers can now share their opinions about a product, service, brand or company also directly and indirectly with many other consumers on websites and social media. With the Internet and social media, third party recommendations and criticism can be given more quickly,
effectively and reaches more audience than before. Traditional WOM therefore has evolved into electronic word-of-mouth (eWOM) - WOM that is communicated through the Internet.
WOM and eWOM are especially interesting for organizations selling experience goods, which are goods where product or service characteristics, such as quality, are more difficult to evaluate. Because with experience goods and services the quality cannot be judged before consumers experience it themselves, consumers are likely to rely upon reviews and opinions of other consumers (Kim, Park & Park, 2013).
The creative industries - the totality of firms that produce experience goods with considerable creative elements (Peltoniemi, 2015) - encounter extreme uncertainty regarding the success potential of products and services, because consumers cannot have complete
information about the product or service prior to consumption. The experience goods in the creative industries includes for example movies, music, books, theatre, fine arts etc. Thus, especially in these settings WOM and eWOM are important concepts.
eWOM can have a big impact without many resources (Hausmann, 2012), so especially for those in the creative industries that have fewer resources, like a limited or no marketing budget, the impact of eWOM can be interesting. Yang, Kim, Amblee and Jeong (2012) already demonstrated that the effect of eWOM on sales of mass products - products targeting large market segments with greater marketing budgets and through corresponding diverse marketing channels, like mainstream movies - differs from the effect of eWOM on sales of niche products - products targeting relatively smaller or particular segment of market with smaller marketing budgets and consequently fewer marketing channels. Their findings suggest that the valence of a review affects whether consumers go to a non-mainstream movie, the valance of a review does not affect whether consumers go to a mainstream movie or not. For both mainstream and non-mainstream movies consumers are more likely to go to a movie if there is a higher amount of reviews about that movie, but this effect is stronger for mainstream movies.
Yang et al. (2012) only studied this difference between mass and niche products within the movie industry, and not for other sectors. Other previous studies have examined the motivations of consumers to engage in WOM and eWOM, the impact of WOM and eWOM - whereby the effects on sales is the most discussed impact - and a range of variables that affect the impact of WOM and eWOM, such as volume and valance, but just like the research of Yang et al. (2012), within the creative industries this is often studied within the movie industry. Other activities and organizations in the creative industries have received little attention.
Yang et al. (2012) suggest that there is a difference between mass and niche products caused by the different marketing budgets for these products. So, the different effects of eWOM on sales in an industry where marketing budgets overall are much lower than in the movie industry might also differ from these effects in the movie industry, because the effect of valence that occurs here can impact the other effects. It would thus be interesting to test the effects of eWOM in other sectors of the creative industries where resources like marketing budget are more limited.
One sector that has very different dynamics than a sector like the movie industry is the performing arts sector (Peltoniemi, 2015). The performing arts are a specific type of service; a performing arts performance must provide a show experience for the audience on the one hand, but must on the other hand also meet cultural and artistic demands and contribute to education and welfare at the same time (Hume, Mort & Winzar, 2007).
Besides entertaining, artistic and cultural goals, performing arts organizations also have economic goals. A lot of performing arts organizations operate as non-profit
organizations and thus to maintain the balance between achieving artistic and cultural goals and economic goals is one of the major strategic challenges for performing arts organizations (Hume et al., 2007). So, resources like marketing budget in the performing arts are limited, because they have limited fund allocation. Also, actors and landlords must be paid regardless of how many people visit the show and therefore just one in five theatre shows make a profit (The Economist, 2016). Even popular shows can stop early if they had a few bad weeks and therefore investing in these shows is a gamble. It is thus for the performing arts sector in particular interesting to know if and how eWOM effects or predicts purchase intentions or sales of consumers.
Hausmann and Poellmann (2016) provided some first exploratory insights about eWOM in a performing arts marketing context and confirm the relevance of eWOM for
theatregoers. However, what the effects of eWOM on purchase intentions of consumers are for performances in the performing arts is not studied yet and thus remain unclear. Therefore, the research question of this study is:
‘What are the effects of eWOM, in the form of expert and consumer-generated online reviews, on purchase intentions of consumers in the performing arts?’
Moreover, most previous studies that examined the impact of (e)WOM on purchase intentions or sales by studying real sales results, and real (e)WOM. However, with this research design, changes to purchase intentions or sales might also be explained by possible effects of other external influences, and not only by effects of (e)WOM. With an experimental research design, those possible effects of other external influences are controlled for.
Therefore, in this study, quantitative research is done with an experimental design to answer the research question. A conceptual model based on the current literature is developed and this conceptual model is tested with empirical research. Hereby an experiment is used that manipulated three independent variables: the type of review, the valance of the review and the type of performance. The data needed to answer this research question is collected through online questionnaires among consumers living in the Netherlands (N=305).
First, the current literature about WOM and eWOM is reviewed in the literature review, whereby different hypotheses are formulated and thus the conceptual model is developed. Then in the methodology the method used for collecting and analysing data is described. Thereafter the results are given and discussed, and lastly the conclusion is drawn.
2 Literature review
To answer the research question, first the current literature about the main concepts of the research question is reviewed. This literature review gives an insight into previous research and thus will be the foundation for this study.
2.1 eWOM in the performing arts
As mentioned in the introduction, the performing arts are a specific type of service; a performing arts performance must provide a show experience for the audience on the one hand, but must on the other hand also meet cultural and artistic demands and contribute to education and welfare at the same time (Hume et al., 2007).
The quality of performing arts performances cannot be assessed before the purchase decision, but only during or after visiting the performance (Kotler and Scheff, 1997 as cited in Hausmann and Poellmann, 2016). The psychological risk of making a wrong choice is
therefore high, and leads to quality uncertainty for consumers. Therefore, consumers seek signs or evidence of quality, like reviews and recommendations of others (Hausmann and Poellmann, 2016).
Hausmann and Poellmann (2016) studied the general use of social media in theatre marketing and found that recommendations have a high relevance for theatre customers and are very relevant in a social media context. 71% of their respondents, who are German
consumers, stated that they use reviews by critics to seek information on a theatre, and 64% of their respondents stated that they use recommendation by people they know personally.
However, the respondents of their study are fans of theatre profiles on Facebook and thus have a certain affinity to the performing arts. 91% of the respondents stated that visiting theatre performances is a relatively or very important element in their recreational activities. The use of recommendations among visitors that are less familiar with theatre performances
are thus unknown. Furthermore, Hausmann and Poellmann (2016) only performed a descriptive data analysis and the effects of eWOM on for example sales of performing arts performances are not known yet.
2.2 (Electronic) word-of-mouth
Word-of-mouth (WOM) is ‘an oral form of interpersonal non-commercial communication among acquaintances’ (Arndt, 1967 as cited in Cheung & Lee, 2012). For organizations, this means that consumers communicate information about the organization’s products or services to each other beyond the organization’s control. Previous studies show the importance of WOM; it plays a major role for customers’ buying decisions (Richins & Root-Shaffer, 1988 as cited in Hennig-Thurau, Gwinner, Walsh & Gremler, 2004).
Nowadays, with the use of the Internet, not only can consumers communicate with each other directly in an oral form, but they can also share their opinions and reviews easily on websites and social media with many other consumers indirectly. Therefore, the traditional offline WOM has evolved into electronic word-of-mouth (eWOM). eWOM communication 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 institution via the Internet’ (Hennig-Thurau et al., 2004). Customers trust online reviews by unknown
consumers more than they trust traditional media, and eWOM thus plays an important role for the purchase decision process of customers (Cheung & Thadani, 2012), just like the
traditional WOM.
However, eWOM also differs from WOM in several dimensions. While the traditional WOM is limited in scalability and speed of diffusion because information is exchanged between small groups of individuals in private conversations or dialogs in synchronous mode, eWOM is not (Cheung & Thadani, 2012). With eWOM, information does not have to be
shared at the same time while all communicators are present, but can be exchanged in asynchronous mode to many different consumers (Cheung & Thadani, 2012). Another difference between WOM and eWOM is that eWOM communications are more accessible and persistent, because most of the information on the Internet is available for an indefinite period of time (Cheung & Thadani, 2012). Furthermore, eWOM is more observable through the presentation format, quantity and persistence and is therefore more measurable than WOM (Cheung & Thadani, 2012). The last difference is the credibility of WOM and eWOM.
Because with WOM the receiver knows the sender of the information, the receiver also knows the credibility of the sender and the message. With eWOM the credibility of the sender is not known and therefore the credibility of eWOM is more difficult to determine.
Cheung and Thadani (2012) have made an integrative framework of the different previously studied effects of eWOM communication (see figure 1). This framework is based
on the four elements of social communication. These four elements of social communication are the communicator, the stimulus, the receiver and the response (Hovland, 1948 as cited in Cheung & Thadani, 2012). The communicator (source) is the person who transmits the communication; the stimulus (content) is the message transmitted by the communicator; the receiver (audience) is the individual who responds to the communication; and the response (main effect) is made by the receiver to the communicator. eWOM is a form of the stimulus element of social communication and involves both receivers, which are the information-seeking customers, and communicators, which are the information-sharing customers.
Most of the research on eWOM in the literature is about the response of eWOM; whereby the most investigated responses of eWOM are attitude, purchase intention and purchase (Cheung & Thadani, 2012).
Besides the impact of eWOM communication, the motivation of consumers to engage in eWOM communication is also studied in the literature. According to Hennig-Thurau et al. (2004) the primary factors for consumers to engage in eWOM communication are their desire for social interaction, the potential to enhance their own self-worth, their concern for other consumers and their desire for economic incentives. Cheung and Lee (2012) found similar motives of consumers to engage in eWOM communication: sense of belonging, reputation and enjoyment of helping other consumers.
2.3 Purchase intention and sales
As mentioned before, attitude, purchase intention and purchase are the most investigated responses of eWOM and thus many studies established the relationship between eWOM and purchase intention, or between eWOM and actual sales.
Kim et al. (2013) found that the volume of eWOM, in the form of consumer-generated reviews, about movies affect box office outcomes of those movies. Besides the volume of
eWOM, Kim et al. (2013) also examined the valence rating, i.e. whether the review is positive or negative, of eWOM. They found no effect for the valance rating of eWOM on box office outcomes. So according to these results only the volume and not the valance of eWOM in the form of consumer-generated reviews affect box office outcomes.
Similarly, Liu (2006) found that the volume of eWOM on the Yahoo Movies Website offers significant explanatory power for box office revenue, in contrast to the valance. Duan, Gu and Whinston (2008) also found that box office sales are significantly influenced by the volume of eWOM in the form of online consumer-generated reviews, but not by the valence. In contrast, Dellarocas, Zhang and Awad (2007) found that both the volume and the valence of eWOM in the form of online reviews have a positive effect on box office revenues.
Babić Rosario, Sotgiu, De Valck and Bijmolt (2016) conducted a meta-analysis across 26 product categories that were studied in primary previous studies: movies, music albums, books, video games, restaurant services, audio players, apparel, cars, cellular phone devices, cellular phone services, computer memory, digital cameras, electronics, financial services, furniture, garden products, green tea, hotel stays, houseware, Internet services, mobile applications, perfume, software, video cassettes and DVDs, and video players. Hereby the product categories books (39%), movies (20%) and digital cameras (8%) are studied the most often in previous studies. Then results show that the volume of eWOM has a stronger impact on sales of those products and services than the valance of eWOM. This supports the findings of Kim et al. (2013), Liu (2006) and Duan et al. (2008), who all found that the volume of consumer-generated reviews affects box office outcomes. However, while Kim et al. (2013), Liu (2006) and Duan et al. (2008) conclude that the valence rating of consumer-generated reviews has no significant effect, Babić Rosario et al. (2016) conclude that valance only has a less strong effect on sales than volume and not that valance has no significant effect on sales at all.
Marchand, Hennig-Thurau and Wiertz (2017) studied the effect of
consumer-generated reviews and microblogs on the sales of video games. They argue that the volume of consumer-generated reviews and microblogs influences sales before and after launch of the video game, while the valance of microblogs never influences sales and the valance of consumer-generated reviews only affect sales a couple of weeks after release of the video game. Their findings are thus similar to the findings about eWOM in the movie industry.
So, overall can be concluded that for eWOM in general the volume of eWOM affects sales, meaning that if there is a greater amount of eWOM about a certain product or service, sales will be higher than sales of a product or service with a lower amount of eWOM.
2.4 Types of online reviews
Just as with traditional WOM, eWOM can take up different forms or types. A specific type of eWOM is online reviews. Online reviews are, like other eWOM, an important source of information on product quality to consumers (Chevalier & Mayzlin, 2006).
In general, there are two types of online reviews: consumer-generated reviews and reviews written by professional editors or critics; the expert reviews. Expert reviews are perceived to be of higher quality than consumer-generated reviews, while consumer-generated reviews are perceived as more reliable (Bickart & Schindler, 2001; Huang & Chen, 2006; Smith, Menon & Sivakumar, 2005 as cited in Parikh, Behnke, Almanza, Nelson &
Vorvoreanu, 2016).
2.4.1 Consumer-generated reviews
Consumer-generated reviews are reviews written by consumers, based on personal
experiences of the consumers and posted on for example websites or social media. Zhang, Ye, Law and Li (2010) showed in their study about reviews of restaurants that
consumer-generated reviews about the food quality, the environment and the service of restaurants are all positively associated with the online popularity of restaurants (i.e. the visits to a
restaurant’s webpage). Zhang et al. (2010) also showed that the volume of
consumer-generated reviews significantly increases the online popularity of restaurants, i.e. the volume of online reviews relates positively to product performance. This is in line with the results of most other studies about the relationship between eWOM and sales that are mentioned before.
However, not all generated reviews are the same, because consumer-generated reviews are posted on different media channels and the characteristics of each conversation channel shape how, when, and what type of WOM is used (Berger and Iyengar, 2013 as cited in Hennig-Thurau, Wiertz & Feldhaus, 2015). For example, the rise of real-time interactive social media channels has introduced microblogging WOM (MWOM) as a new type of WOM, which is defined as ‘any brief statement made by a consumer about a commercial entity or offering that is broadcast in real time to some or all members of the sender’s social network through a specific web-based service (e.g., Twitter)’ (Hennig-Thurau et al., 2015). This can also be seen as a form of a consumer-generated review. In this respect, Hennig-Thurau et al. (2015) found that for MWOM only negative reviews negatively affect sales of movies and that there is no significant effect of positive reviews, in contrast to the findings of other studies that valance has no effect.
But overall, previous studies indicate that the volume of online reviews positively relates to performance regardless of the content of the reviews. Kim et al. (2013) ascribe the importance of the volume of online reviews for movies to the characteristic of experience goods; the more conversation there is about a product, the greater the probability that
consumers are informed about it and this leads to consumption decision (Godes and Mayzlin, 2004 as cited in Kim et al., 2013) because people develop a favourable attitude toward things simply because they are familiar with them (Zajonc, 1968 as cited in Kim et al., 2013). I.e.
any review, positive or negative will be better than no review at all, because it creates awareness. Therefore, the first hypothesis is:
Hypothesis 1a (H1a): Positive consumer-generated reviews have a positive effect on purchase intention of consumers in the performing arts.
Hypothesis 1b (H1b): Negative consumer-generated reviews have a positive effect on purchase intention of consumers in the performing arts.
2.4.2 Expert reviews
Expert reviews are reviews written by professional editors or critics. Zhang et al. (2010) found that expert reviews are negatively associated with the online popularity of restaurants, in contrast to consumer-generated reviews that were all positively associated with the online popularity of restaurants. This could be due to that for restaurants expert reviews are
perceived to be less credible than consumer-generated reviews.
In contrast, Kim et al. (2013) found that the valance rating of expert reviews affects box office outcomes positively. However, the low credibility of the expert reviews in the study of Zhang et al. (2010) is probably causing the negative effect, because credibility is one of the antecedents affecting purchase intention (Cheng & Thadani, 2012). In the performing arts, the credibility of expert reviews is higher than the credibility of consumer-generated reviews (Chiou, Hsiao & Su, 2014), so in the performing arts the credibility of expert reviews is probably higher than the credibility of expert reviews for restaurants. The findings of Kim et al. (2013), which indicate that positive reviews will lead to higher sales and negative reviews will lead to lower sales, are therefore more likely to correspond with the effect of
expert reviews in the performing arts. Therefore, the second hypothesis can be stated as follows:
Hypothesis 2a (H2a): Positive expert reviews have a positive effect on purchase intention of consumers in the performing arts.
Hypothesis 2b (H2b): Negative expert reviews have a negative effect on purchase intention of consumers in the performing arts.
2.4.3 Consumer-generated review versus expert reviews
Kim et al. (2013) studied both the impact of online consumer-generated reviews and expert reviews. The valance rating of expert reviews affects box office outcomes positively, in contrast to the valance rating of consumer-generated reviews. This means that consumers might have different expectations from other consumers than from experts when searching for information about movies.
Continuing with the credibility of reviews, Cheung & Thadani (2012) propose that eWOM credibility is positively associated with eWOM adoption, and that eWOM adoption is positively associated with purchase intention. So, it can be assumed that a higher credibility would eventuality lead to a higher purchase intention.
In the performing arts, expert reviews are perceived to be more credible than consumer-generated reviews (Chiou et al., 2014). Expert reviews will thus probably have a greater effect on purchase intention than consumer-generated reviews. Also, Hausmann and Poellmann (2016) found that consumers in the performing arts slightly prefer
recommendations by critics, i.e. expert reviews, to recommendations by people they know personally because experts are more experienced and are more credible. Thus, consumers in
the performing arts probably also prefer expert reviews to consumer-generated reviews. Therefore, the third hypothesis is:
Hypothesis 3a (H3a): Positive expert reviews have a stronger effect on purchase intention of consumers in the performing arts than positive consumer-generated reviews.
Hypothesis 3b (H3b): Negative expert reviews have a stronger effect on purchase intention of consumers in the performing arts than negative consumer-generated reviews.
2.4.4 Consumer-generated reviews combined with expert reviews
Consumers can also use multiple sources of information when evaluating a product or service, they can thus use consumer-generated reviews and expert reviews together when deciding if they will visit a performance. Because it is expected that positive consumer-generated reviews as well as positive expert reviews have a positive effect on purchase intention of consumers in the performing arts, the combination of those two will probably also have a positive effect on purchase intention.
Furthermore, it is expected that negative consumer-generated reviews have a positive effect on purchase intention of consumers in the performing arts, but negative expert reviews have a negative effect on purchase intention of consumers in the performing arts. Then the combination of negative consumer-generated reviews and expert reviews will probably have a negative effect on purchase intention, because it is expected that the effect of expert reviews is stronger than the effect of consumer-generated reviews. Therefore, the fourth hypothesis is:
Hypothesis 4a (H4a): Positive consumer-generated reviews and expert reviews combined have a positive effect on purchase intention of consumers in the performing arts.
Hypothesis 4b (H4b): Negative consumer-generated reviews and expert reviews combined have a negative effect on purchase intention of consumers in the performing arts.
2.5 Performance type: popular versus highbrow performances
Because in the literature the effect of eWOM valence on product sales is mixed, Yang et al. (2012) distinguished between mainstream and non-mainstream movies when studying the effect of eWOM valance. They found that the valence of eWOM has a significant effect on box office revenue only for non-mainstream movies, but not for mainstream movies.
Therefore Yang et al. (2012) argue that the effect of eWOM valance on sales can be diluted if marketing channels become more diverse with larger marketing budgets, which is the case with mainstream movies. In contrast, the volume of eWOM has a greater effect on box office revenue for mainstream movies than for non-mainstream movies. So only for niche products positive reviews will increase sales, while negative reviews will decrease sales. For mass products, it does not matter whether reviews are positive or negative. Here it only matters whether there are a lot of reviews, because a higher number of reviews will increase sales.
Shrum (1991) classifies performing arts genres in two categories: popular types and highbrow types. Genres that he classified as popular are for example comedy, musical and cabaret and genres that he classified as highbrow are for example theatre and dance. Hereby popular performances can be compared to mass, or mainstream products and highbrow performances can be compared to niche, or non-mainstream products.
When following the results of Yang et al. (2012), it can be therefore expected that the valance of consumer-generated reviews will only affect purchase intention for highbrow performances and not for popular performances. However, looking at the effect of consumer-generated reviews in general also volume of consumer-consumer-generated reviews must be taken into
account. Because the volume of eWOM has a greater effect on sales for mainstream movies than for non-mainstream movies, it can be expected that the effect of eWOM on purchase intention is stronger for popular performances than for highbrow performances. So at least there will be some effect of consumer-generated reviews on purchase intention for popular performances. Therefore, the fifth hypothesis is:
Hypothesis 5a (H5a): The positive effect of positive consumer-generated reviews on purchase intention of consumers in the performing arts is stronger for highbrow performances than for popular performances.
Hypothesis 5b (H5b): The effect of negative consumer-generated reviews on purchase intention of consumers in the performing arts is positive for popular performances, but negative for highbrow performances.
Then for expert reviews, Shrum (1991) argues that critics are only important for highbrow performances. He found that positive reviews of critics are only associated with higher sales when the performance is a highbrow performance. For popular performance genres, the visibility provided by the reviews (i.e. volume) is more important than the evaluative function of those reviews (i.e. valance). Gemser, Van Oostrum and Leenders (2007) support these results. They argue that reviews influence the behaviour of consumers only for art house movies, and that for mainstream movies consumers rely mainly on other sources of information. However, valance is not taken into account in this study.
In contrast Berger, Sorensen and Rasmussen (2010) found that negative reviews in the New York Times negatively affect the sales of books by well-known authors, i.e. established products, but positively affect the sales of books that had lower prior awareness, i.e. unknown
products. This means that the valance of expert reviews only influences the sales of established products and, the sales of unknown products are thus only influenced by the volume of expert reviews. Like popular performances, established products probably also have greater marketing budgets, and correspond trough more diverse marketing channels than unknown, or new products. Therefore, established products can be compared with popular performances, and unknown products with highbrow performances.
Thus, based on the results of Berger et al. (2010), it can be expected that the valance of expert reviews will only affect purchase intention for popular performances, and will not affect purchase intention for highbrow performances. While the results of Shrum (1991) are found in the same sector as the sector used for this study, i.e. the performing arts, these results are not based on the online environment, while the results of Berger et al. (2010) are. The results of Berger et al. (2010) thus seem more in accordance with the current setting. Therefore, the sixth hypothesis is:
Hypothesis 6a (H6a): The positive effect of positive expert reviews on purchase intention of consumers in the performing arts is stronger for popular performances than for highbrow performances.
Hypothesis 6b (H6b): The effect of negative expert reviews on purchase intention of consumers in the performing arts is negative for popular performances, but positive for highbrow performances.
The overall conceptual model based on these hypotheses can be found in Figure 2. Based on the different attributes of each of these constructs, different effects on consumer purchase intention are expected.
Consumer purchase intention Type of review Type of performance Valance of the review
3 Methodology
In this section, the methods used for collecting and analysing data are described. First the chosen research design is described, then the measurements used to measure the different variables are given, the procedure of collecting data is explained, the characteristics of the used sample are given, and the analyses used to test the hypotheses are described.
3.1 Design
The overall research design chosen to answer the research question is quantitative research. The conceptual model (see figure 2) is tested with empirical research by using an experiment. An experiment allows for studying the probability of a change in independent variables, in this study type of review, valance of the review, and type of performance, causing a change in another, dependent variable, in this study consumer purchase intention (Saunders, Lewis &Thornhill, 2012). Hereby the possible effects of an alternative explanation to the manipulation of the independent variables are limited and threats to internal validity are eliminated, because all groups (control and experimental groups) are subject to exactly the same external influences other than the manipulation, and the manipulation is consequently the only explanation for any changes to the dependent variable consumer purchase intention.
Internal validity of this research method is thus higher than most previous studies, whereby real-life data is used to examine the impact of (e)WOM sales. A disadvantage of this experimental design is that external validity is more difficult to establish and the
generalisability of the results is thus lower than the findings of previous studies.
The experiment manipulated the type of review, the valance of the review and the type of performance. This is thus resulting in a 4 (consumer-generated reviews versus expert reviews versus consumer-generated and expert reviews versus no reviews) x 2 (positive reviews versus negative reviews) x 2 (popular versus highbrow performance)
between-subjects factorial design. Only in the control scenario of no reviews there is no valance, so in total there are 14 scenarios.
Respondents are shown two scenarios; one of the scenarios of the popular performance and one of the scenarios of the highbrow performance. This is done to increase the response rate for the different scenarios. Because the two different performances are not related to each other, it is not expected that the two scenarios shown affect each other. Also, the sequence in which the two scenarios are shown to the respondents is determined randomly, so if one scenario would affect the results of the other scenario the potential effects on the results are limited. The possible scenarios for both the popular performance and the highbrow
performance are shown in figure 3. Respondents were assigned randomly to the experiment’s treatments, so that the composition of the respondents for the different scenarios is about similar and the composition is thus not affecting the results.
The data is collected through self-completed online questionnaires among consumers living in the Netherlands. With a questionnaire, it is possible to measure the variables among consumers and to examine and explain the relationships between the different variables (Saunders, Lewis, & Thornhill, 2012). Also, an online questionnaire is a relative cheap and simple method and an online questionnaire makes it easier to get a larger sample size
(Saunders et al., 2012). This is important considering the limited time and available resources. To get a good representation of the population, a probability sample from the population is drawn. A sample size of 384 is needed to have a sample that is representative of the
population at a 95% confidence level and with a 5% margin of error (Saunders et al., 2012). A sample size of 30 for each scenario is needed to be able to test the differences between the groups of the different scenarios, because with 30 cases in each group the group size is considered large, and then the assumption that the data for each group is normally distributed is not particularly important for analyses (Saunders et al., 2012).
3.2 Measurements 3.2.1 Dependent variable
The dependent variable consumer purchase intention is measured at consumer level with a three items scale adapted from Xia and Bechwati (2008). An example item is ‘It is very likely that I will go to this theater performance’. The scales are ranged on a seven-point Likert scale from (1) strongly disagree to (7) strongly agree, so a high score means that the consumer has a high intention to visit the performance after seeing information about that performance and possible consumer-generated and/or expert reviews. There is one item counterbalanced, which will be recoded for analysis.
3.2.2 Experiment stimuli
The independent variables type of review, valance of the review and type of performance are manipulated with different scenarios in the experiment. For the type of performance, the popular performance and highbrow performance used are real performances that are not performed yet; prior knowledge of the consumers about the performance is thus limited or absent. This is done to improve the internal validity of the experiment, because prior product
knowledge can have potential confounding effects (Lee, Park & Han, 2008). Furthermore, the names of the cast are not given in the information about the performance, so that any star-power does not affect the results.
For the type of review and the valance of the review the consumer-generated reviews for the experiment were created based on real online consumer-generated reviews on different social media sites as Twitter, Facebook and Instagram. Consumers generally read about six to eight reviews of about three to four lines each when wanting to buy something (Lee et al., 2008). Therefore, seven short consumer-generated reviews are shown in the scenarios with consumer-generated reviews. Also, the expert reviews were based on real expert reviews on the websites of some Dutch newspapers. Because expert reviews in general are much longer than consumer-generated reviews, only one expert review is shown in the scenarios with expert reviews. The different scenarios used can be found in the questionnaire in Appendix 1.
3.2.3 Control variables
Because personal attitude towards online reviews can also have a potential confounding effect on the results (Lee et al., 2008), this is used as control variable. Personal attitude toward online reviews is measured at consumer level with a three items scale adapted from Lee et al. (2008). An example item is ‘When I buy a product or service, I always read reviews on the Internet’. The scales are ranged on a seven-point Likert scale from (1) strongly disagree to (7) strongly agree, so a high score means that the consumer usually uses online reviews when they want to buy something. There are no items counterbalanced.
The trustworthiness and expertise of the reviews are also used as control variables. Trustworthiness is measured at consumer level with a five items scale adapted from Ohanian (1990). The scales are ranged on a seven-point bipolar scale, for example from (1)
reviewers are trustworthy. There are no items counterbalanced. Expertise is measured at consumer level with a five items scale adapted from Ohanian (1990). The scales are ranged on a seven-point bipolar scale, for example from (1) unknowledgeable to (7) knowledgeable, so a high score means that the consumer thinks the reviewers are expertized. There is one item counterbalanced, which will be recoded for analysis.
Relating to customers of performing arts organizations, Johnson and Garbarino (2001) found significant differences between customers who are subscribers, customers who are occasional subscribers and customers who are no subscribers in terms of the information sources these customers use for decision-making; for non-subscribers, the importance of all sources of information is the highest in their decision-making process and for subscribers the importance of these sources of information is the lowest. This is supported by the theory of Hausmann and Poellmann (2016), who argue that especially visitors that are unfamiliar with theatre experiences seek signs or evidence of quality to eliminate the quality uncertainty caused by the high psychological risk of making a wrong choice that comes with the
experience good attribute of the performing arts. Consumer subscription to a theatre and the amount of theatre visit per year are therefore also used as control variables. These variables are measured at consumer level by asking consumers if they have a subscription to a theatre (ranged (1) yes and (2) no), and how often they visit a theatre performance (ranged (1) never or less than 1 time a year, (2) 1 to 5 times a year, (3) 5 to 10 times a year, and (4) more than 10 times a year). Furthermore, the demographics gender, age and education are also used as control variables.
3.3 Procedure
The questionnaire was made with Qualtrics. The questionnaire was made in Dutch and in English, so both Dutch and foreign consumers living in the Netherlands were able to fill in the
questionnaire. The English version of the questionnaire can be found in Appendix 1, the Dutch version of the questionnaire can be found in Appendix 2. In the questionnaire in Appendix 1, the dummy variables used for analysis are displayed between brackets after each possible answer.
The link to the questionnaire was shared actively on social media and via the personal network of the researcher. Participants could fill in the questionnaire online when and where they wanted. The questionnaire was online for a period of two weeks.
Participants could win a Podium Cadeaukaart or Bol.com gift card worth € 30, to stimulate them to fill in the entire questionnaire. This two types of gift cards are chosen to attract consumers who like the performing arts (with the Podium Cadeaukaart), as well as consumers who do not (with the Bol.com gift card).
3.4 Sample
The sample consists of 348 respondents. However, 116 respondents did not complete the entire the questionnaire. 34 of these respondents did not answer any of the questions,
therefore these 34 cases are excluded from further analyses. The other 82 respondents who did not complete the entire questionnaire, saw at least one of the scenarios and answered at least a part of the questions. Respondents who answered all the questions for only 1 scenario can still be used for the between-subjects factorial analyses and therefore these 82 cases are not
excluded from further analyses. Then 2 respondents who completed the questionnaire did also not answer any of the questions, so these 2 cases are also excluded from further analyses. Furthermore, there are outliers detected in 7 cases and these cases are also excluded from further analyses.
So, the final sample consists of 305 respondents. Cases with missing values are deleted pairwise during further analyses. This means that only the cases without missing data
in the analysed variables are used in the analysis. For each scenario, the sample size is bigger than 30.
The average age of the respondents is 41.02 years (SD = 15.930), with a median of 45 years, a minimum age of 13 years, and a maximum age of 84 years. 55.7% of the respondents is male and 44.3% is female. 0.7% of the respondents has primary school as current or highest obtained level of education, 4.9% lbo/vmbo, 9.2% havo, 4.3% vwo, 22.3% mbo, 37.0% hbo, and 21.6% wo. 95.7% of the respondents is Dutch, 1,3% Belgium, 0.7% German, 0.7%
Indian, 0.3% Iranian, 0,3% Luxembourgish, 0.3% Polish, 0,3% Spanish, and the nationality of 0,3% of the respondents is unknown. 97.0% of the respondents is currently living in the Netherlands and 3.0% is not.
8.9% of the respondents has a theatre subscription or is for example ‘Friend of’ a theatre, and 91.1% has no theatre subscription and is no ‘Friend of’ a theatre. 43.3% of the respondents visits theatrical performances less than 1 time a year or never visits theatrical performances, 49.8% of the respondents visits theatrical performances 1 to 5 times a year, 6.2% 5 to 10 times a year, and 0.7% more than 10 times a year.
3.5 Analyses and predictions
The statistical program SPSS is used to analyse the outcomes of the questionnaire. First frequencies are checked, counterbalanced items are recoded, and the reliability of the used measurement scales is tested with a reliability analysis. Then scale means are computed and the assumptions of normal distribution and homogeneity of variance are checked.
Furthermore, a correlation analysis is done to study the relationships between the continuous variables consumer purchase intention, personal attitude towards reviews, trustworthiness of the review, and expertise of the review. Finally, the hypotheses are tested.
3.5.1 Consumer-generated reviews
To test whether positive consumer-generated reviews have a positive effect on purchase intention of consumers in the performing arts (H1a), and whether negative consumer-generated reviews also have a positive effect on purchase intention of consumers in the performing arts (H1b), a between-subjects factorial ANOVA is used. A factorial ANOVA tests for differences between groups when there are two independent variables (Field, 2009), in this case thus the type of review and valance of the review. The main effect of each independent variable on the dependent variable purchase intention is tested and also the interaction effect between the independent variables is tested, so if the effect of one independent variable on the dependent variable is the same across all levels of the other independent variable.
Predicted is a statistically significant difference in purchase intention between the group with positive consumer-generated reviews and the group with no review, and a significant difference in purchase intention between the group with negative consumer-generated reviews and the group with no review, whereby purchase intention for the group with positive consumer-generated reviews is higher than purchase intention for the group with no review, and purchase intention for the group with negative consumer-generated reviews is also higher than purchase intention for the group with no review.
3.5.2 Expert reviews
This between-subjects factorial ANOVA is also used to test whether positive expert reviews have a positive effect on purchase intention of consumers in the performing arts (H2a), and whether negative expert reviews have a negative effect on purchase intention of consumers in the performing arts (H2b). Hereby the differences between the groups of the two independent variables type of review and valence of review are tested. Again, the main effect of each
independent variable on the dependent variable purchase intention is tested and also the interaction effect between the independent variables is tested.
Predicted is a statistically significant difference in purchase intention between the group with positive expert reviews, the group with no review, and the group with negative expert reviews, whereby purchase intention for the group with positive expert reviews is higher than purchase intention for the groups with no review and negative expert reviews and purchase intention for the group with negative expert reviews is lower than purchase intention for the groups with no review and positive expert reviews.
3.5.3 Consumer-generated reviews versus expert reviews
Whether positive expert reviews have a stronger effect on purchase intention of consumers in the performing arts than positive consumer-generated reviews (H3a), and whether have a stronger effect on purchase intention of consumers in the performing arts than negative
consumer-generated reviews (H3b), is also tested with the between-subjects factorial ANOVA of the two independent variables type of review and valence of review and the dependent variable purchase intention.
Predicted is a statistically significant difference in purchase intention between the group with positive expert reviews and the group with positive consumer-generated reviews, and a significant difference in purchase intention between the group with negative expert reviews and the group with negative consumer-generated reviews, whereby purchase intention for the group with positive expert reviews is higher than for the group with positive
consumer-generated reviews, and purchase intention for the group with negative expert reviews is lower than for the group with negative consumer-generated reviews.
3.5.4 Consumer-generated reviews combined with expert reviews
To test whether positive consumer-generated reviews and expert reviews combined have a positive effect on purchase intention of consumers in the performing arts, and negative
consumer-generated reviews and expert reviews combined have a negative effect on purchase intention of consumers in the performing arts, again a between-subjects factorial ANOVA is used, whereby the differences between the groups of the two independent variables type of review and valence of review are tested. Again, the main effect of each independent variable on the dependent variable purchase intention is tested and also the interaction effect between the independent variables is tested.
Predicted is a statistically significant difference in purchase intention between the group with positive consumer-generated reviews and expert reviews combined, the group with no review, and the group with negative consumer-generated reviews and expert reviews combined, whereby purchase intention for the group with positive consumer-generated reviews and expert reviews combined is higher than purchase intention for the groups with no review and negative consumer-generated reviews and expert reviews combined and purchase intention for the group with negative consumer-generated reviews and expert reviews
combined is lower than purchase intention for the groups with no review and positive consumer-generated reviews and expert reviews combined.
3.5.5 Type of performance
A between-subjects factorial ANOVA is also used to test whether the positive effect of positive consumer-generated reviews on purchase intention of consumers in the performing arts is stronger for highbrow performances than for popular performances, whether the effect of negative consumer-generated reviews on purchase intention of consumers in the
whether the positive effect of positive expert reviews on purchase intention of consumers in the performing arts is stronger for popular performances than for highbrow performances, and whether the effect of negative expert reviews on purchase intention of consumers in the
performing arts is negative for popular performances, but positive for highbrow performances. Predicted is a statistically significant difference in purchase intention between the group with positive consumer-generated reviews about a popular performance, the group with positive consumer-generated reviews about a highbrow performance, and the group with no review, whereby purchase intention for the group with positive consumer-generated reviews about a highbrow performance is higher than purchase intention for the group with no review about a highbrow performance, purchase intention for the group with positive consumer-generated reviews about a popular performance is higher than for the group with no review about a popular performance, and the difference between the group with positive consumer-generated reviews about a highbrow performance and the group with no review about a highbrow performance is bigger than the difference between purchase intention for the group with positive consumer-generated reviews about a popular performance and the group with no review about a popular performance.
Also, predicted is a statistically significant difference in purchase intention between the group with negative consumer-generated reviews about a popular performance, the group with negative consumer-generated reviews about a highbrow performance, and the group with no review, whereby purchase intention for the group with negative consumer-generated reviews about a popular performance is higher than purchase intention for the group with no review about a popular performance, and purchase intention for the group with negative consumer-generated reviews about a highbrow performance is lower than for the group with no review about a highbrow performance.
Furthermore, predicted is a statistically significant difference in purchase intention between the group with positive expert reviews about a popular performance, the group with positive expert reviews about a highbrow performance, and the group with no review, whereby purchase intention for the group with positive expert reviews about a popular performance is higher than purchase intention for the group with no review about a popular performance, purchase intention for the group with positive expert reviews about a highbrow performance is higher than for the group with no review about a highbrow performance, and the difference between the group with positive expert reviews about a popular performance and the group with no review about a popular performance is bigger than the difference between purchase intention for the group with positive expert reviews about a highbrow performance and the group with no review about a highbrow performance.
Predicted is also a statistically significant difference in purchase intention between the group with negative expert reviews about a popular performance, the group with negative expert reviews about a highbrow performance, and the group with no review, whereby purchase intention for the group with negative expert reviews about a popular performance is lower than for the group with no review about a popular performance, and purchase intention for the group with negative expert reviews about a highbrow performance is higher than purchase intention for the group with no review about a highbrow performance.
4 Results
In this section, the results of the analyses as described in 3.5 are given. First the results of the reliability analysis to test the measurement scales are given, then the results of the checked assumptions are given, the correlations between the main variables are described, and then results of the hypotheses tested are given.
4.1 Reliabilities
A reliability analysis is done to test whether the used measurement scales are reliable enough to be used for further analyses. The scales are reliable if the Cronbach’s alpha of a scale is 0.7 or higher. This indicates that the data collection techniques yield consistent findings; so, all the items in the scale are measuring the same thing (Saunders et al., 2012).
The consumer purchase intention scale has a high reliability, with a Cronbach’s Alpha of .87. The corrected item-total correlations indicate that all the items have a good correlation with the total score of the scale (all above .30). Also, none of the items would substantially affect the reliability if they were deleted.
The personal attitude towards online reviews scale also has a high reliability, with a Cronbach’s Alpha of .77. Here the corrected item-total correlations also indicate that all the items have a good correlation with the total score of the scale (all above .30) and none of the items would affect the reliability substantially if they were deleted.
Then the trustworthiness of the reviews scale is also highly reliable, with a Cronbach’s Alpha of .91. Again, the corrected item-total correlations indicate that all the items have a good correlation with the total score of the scale (all above .30) and also, none of the items would substantially affect the reliability if they were deleted.
The reliability of the expertise of the reviews scale is high too, with a Cronbach’s Alpha of .79. For 1 item, the corrected item-total correlation does not indicate that this item
has a good correlation with the total score of the scale (under .30). However, the reliability of the scale is already high, and deleting this item will not lead to an increase in Cronbach’s alpha above 0.10. This item will therefore not be deleted from this scale.
All the scales used are thus highly reliable and therefore the scale means are used for the variables in further analyses.
4.2 Assumptions
All continuous variables, i.e. consumer purchase intention, personal attitude towards reviews, trustworthiness of the review, and expertise of the review, have a normal distribution (statistic values of skewness and kurtosis are between -1 and 1).
The variances for consumer purchase intentions are equal for the different types of reviews, F(3, 497) = 0.64, ns, but the variances are significantly different for the different types of valance of the reviews, F(2, 498) = 17.45, p < .001 and for the different types of performance, F(1, 499) = 10.36, p < .01 Therefore, a 1 / square root transformation is done for consumer purchase intention to correct for the unequal variances. After this transformation, the variances are equal for the different types of reviews, F(3, 497) = 0,61, ns, the two types of valance of the reviews, F(2, 498) = 1.19, ns, and the different types of performances , F(1, 499) = 2.18, ns. This transformation causes that a higher number of the variable purchase intention equals a lower level of consumer purchase intention, instead of a higher level of consumer purchase intention.
4.3 Correlations
To study the relationships between the variables gender, age, education, theatre subscription, amount of theatre visits per year, personal attitude towards reviews, trustworthiness of the
review, expertise of the review, and consumer purchase intention, a correlation analysis is done. The correlations between the variables are shown in Table 1.
There is a significant negative correlation between age and consumer purchase intention, r = -.10, p < .05. There is thus a small negative effect between age and consumer purchase intention, whereby purchase intention of older consumers is higher than purchase intention of younger consumers.
There is a significant positive correlation between theatre subscription and consumer purchase intention, r = .14, p < .01. So, there is a small positive effect between theatre
subscription and consumer purchase intention, whereby purchase intention of non-subscribers is lower than purchase intention of theatre subscribers.
There is a significant negative correlation between the amount of theatre visits per year and consumer purchase intention, r = -.26, p < .001. There is thus a medium negative effect between the amount of theatre visits per year and consumer purchase intention, whereby purchase intention of frequent theatre visitors is higher than less frequent visitors.
Variables M SD 1 2 3 4 5 6 7 8 9 1 Gender 1.44 0.50 - 2 Age 41.02 15.92 -.29*** - 3 Education 5.40 1.43 .00 -.26*** - 4 Theatre subscription 1.91 0.28 -.07 -.18*** .06 - 5 Theatre visits / year 1.64 0.63 .10* .20*** .00 -.40*** - 6 Personal attitude towards reviews 5.10 1.03 .15** -.18*** .07 -.03 -.02 (.77) 7 Trustworthiness of the review 4.72 1.02 .09 -.16** .08 .03 -.01 .14** (.91) 8 Expertise of the review 4.44 0.85 .05 -.02 .03 -.01 -.00 .10 .52*** (.79) 9 Consumer purchase intention 0.68 0.20 .03 -.10* -.06 .14** -.26*** -.01 .00 -.10* (.87) Note. * p< .05. ** p < .01. *** p < .001.
There is a significant negative correlation between expertise of the review and consumer purchase intention, r = -.10, p < .05. There is thus a small negative effect between expertise of the review and consumer purchase intention, whereby purchase intention is higher when expertise is higher.
There are no significant correlations between consumer purchase intention and the control variables gender, education, personal attitude towards reviews and trustworthiness of the review. There are thus no effects between these control variables and consumer purchase intention, and therefore these control variables will not be included in further analyses.
4.4 Control variables
A between-subjects factorial ANOVA showed that the effects of the control variables age (F(1, 362) = 0.31, p = .58), theatre subscription (F(1, 362) = 0.16, p = .69), and expertise of the review (F(1, 362) = 1.87, p = .17) on consumer purchase intention were not significant. Therefore, these control variables will not be included in further analyses. The effect of the control variable amount of theatre visit per year on purchase intention is significant, F(1, 362) = 19.72, p < .001. This control variable will therefore be used as only covariate in further analyses.
4.5 Type of review and valance of the review
Hypotheses 1a, 1b, 2a, 2b, 3a, 3b, 4a and 4c are tested with a between-subjects factorial ANOVA. The mean and standard deviations of consumer purchase intention are shown in Table 2 and a summary of the results of this factorial ANOVA are shown in Table 3. There was a non-significant main effect of type of review on consumer purchase intention, F(2, 493) = 0.94, p = .39, η2 = .00. There was a significant moderate main effect of valance of the review on consumer purchase intention, F(1, 493) = 54.99, p < .001, η2 = .10. The Tukey
post-hoc tests revealed that purchase intention was significantly lower with negative reviews compared to no reviews (p = .00) and compared to positive reviews (p = .00). There was no statistically significant difference between positive reviews and no reviews (p = .79). There was a significant low interaction effect between type of review and valance of the review on purchase intention, F(2, 493) = 3.75, p < .05, η2 = .02. This indicates that the effect of valance of the review on consumer purchase intention is not the same across the different type of reviews. A factorial ANOVA whereby the effect of valance on consumer purchase intention is tested for the groups of type of review separated, showed that the significant effect of valance on purchase intention is strong for consumer-generated reviews F(1, 139) = 39.72, p < .001, η2 = .22, and for consumer-generated and expert reviews combined, F(1, 141) = 22.29, p < .001, η2 = .14, but low for expert reviews, F(1, 140) = 5.10, p < .05, η2 = .04.
Positive Negative No review 0.64 (0.19), n = 72 Consumer-generated review 0.62 (0.18), n = 68 0.79 (0.18) , n = 74 Expert review 0.66 (0.20), n = 70 0.71 (0.19), n = 73
Consumer-generated review and expert review combined
0.60 (0.18),
n = 72
0.75 (0.20),
n = 72
Table 2. Descriptive statistics of purchase intention per group (mean, standard deviation, and group size)
SS DF MS F η2 Sig.
Amount of theatre visits per year 1.31 1 1.31 40.28 .08 .00
Type of review 0.06 2 0.03 0.94 .00 .39
Valance of the review 1.79 1 1.79 54.99 .10 .00
Type of review x valance of the review 0.24 2 0.12 3.75 .02 .02
Error 16.08 493 0.03
Total 253.51 501
Note. p< .05.
Figure 4. Type of review and valance of the review
4.5.1 Consumer-generated reviews
Hypothesis 1a stated that positive consumer-generated reviews have a positive effect on purchase intention of consumers in the performing arts and hypothesis 1b stated that negative consumer-generated reviews have a positive effect on purchase intention of consumers in the performing arts.
Figure 4 shows that as expected purchase intention for the group with positive consumer-generated reviews is higher than purchase intention for the group with no review. However, the results of the factorial ANOVA and Tukey post-hoc tests showed that while there was a significant moderate main effect of valance of the review on consumer purchase intention, the difference in purchase intention between the group with positive consumer-generated reviews and the group with no review was not statistically significant. Hypothesis 1a is therefore not supported.
On the other hand, the difference in purchase intention between the group with
However, as Figure 4 shows, purchase intention for the group with negative consumer-generated reviews is not higher than purchase intention for the group with no review, but lower. Hypothesis 1b is therefore also not supported.
4.5.2 Expert reviews
Hypothesis 2a stated that positive expert reviews have a positive effect on purchase intention of consumers in the performing arts and hypothesis 2b stated that negative expert reviews have a negative effect on purchase intention of consumers in the performing arts.
Surprisingly, Figure 4 shows that purchase intention for the group with positive expert reviews is not higher than purchase intention for the groups with no review, but lower. Also, the difference in purchase intention between these two groups was not statistically significant. Hypotheses 2a is therefore not supported.
Figure 4 shows furthermore that, as expected, negative expert reviews and purchase intention for the group with negative expert reviews is lower than purchase intention for the groups with no review and positive expert reviews. These differences in purchase intention between the group with negative consumer-generated reviews and the group with no review, and between the group with negative consumer-generated reviews and the group with positive consumer-generated reviews, are statistically significant. Hypotheses 2b is therefore
supported.
4.5.3 Consumer-generated reviews versus expert reviews
Hypothesis 3a stated that positive expert reviews have a stronger effect on purchase intention of consumers in the performing arts than positive consumer-generated reviews and hypothesis 3b stated that negative expert reviews have a stronger effect on purchase intention of
