THE EFFECTS OF VARIABILITY IN RATINGS ON CONSUMER APPEAL FOR A PRODUCT EXTENSION
- The Motion Picture Industry -
MIRUNA BRATULESCU Student Number: 10824480 Master Thesis
MSc Business Administration
Entrepreneurship and Management in the Creative Industries Supervised by Dr. J.J. Ebbers
1 STATEMENT OF ORIGINALITY
This document is written by Student Miruna Bratulescu, 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.
2 In an increasingly complex online medium, third party information has a powerful influence on demand and product performance in the market. When considering whether to purchase a brand’s extension, consumers usually read reviews by specialists or by other consumers regarding the previous product offered by the brand and many websites (such as Amazon.com, Ebay.com, Metacritic.com etc.) offer a multitude of reviews and ratings both by experts and regular users of the product. But what happens if the reviews or ratings for a product differ and lack consensus? Will consumers still purchase an extension of that brand?
Using branding literature as a background, I try to understand how a variability in ratings for a previous product impacts the success of the extension. The empirical setting is the motion picture industry and movie sequels are considered to be brand extensions. Through four between-subjects experiments, the results show that, in line with expectations, variability in ratings plays a major role in the appeal that consumer have for a product extension. An interaction was found between the variability in ratings and the level of the average rating, with consumers having more appeal for the extension when the previous product has low variability and a high average rating.
3 Table of Contents
1. Introduction Page 4
2. Literature Review Page 9
2.1. Brand Equity and Brand Image Page 9
2.2. Brand Extensions Page 10
2.3. Motion Picture Industry Page 12
2.4. Third Party Information Page 14
2.5. Variability Page 21
3. Method and Results – Study 1 Page 29
3.1. Description of Study 1 Page 29
3.2. Study 1 Results Page 35
3.2.1. Experiment 1 Page 36
3.2.2. Experiment 2 Page 41
3.2.3. Experiment 3 Page 45
4. Method and Results – Study 2 Page 49
4.1.Description of Study 2 Page 49
4.2. Study 2 Results Page 52
5. Discussion and Conclusion Page 61
5.1. Implications Page 61
5.2. Limitations and Future Research Page 66
5.3. Conclusion Page 68
6. References Page 69
4 Chapter 1. Introduction
Entertainment is not a basic need like food or shelter, nor even a secondary need like furniture or transportation. Seeing a movie, hearing a song, or watching a basketball game is completely optional.
(Anita Elberse)
Consumer demand in the creative industries is extremely uncertain. The success of an experiential product cannot be accurately predicted at an early stage, the ‘nobody knows’ principle is king and that is why producers are faced with the very difficult task of marketing and making their product known (Caves, 2003).
Customers are offered an experience and the quality of that experience cannot be objectively pinpointed, as the choices of customers reflect tastes and subjective judgments. That is why word of mouth and critics’ comments play a big role in the decision to purchase and can be especially useful when considering product extensions. In the words of Anita Elberse, the entertainment marketplace “is much like a minefield – it is hard to know where to step, and a misstep could trigger a catastrophe”.
The setting for this study is the creative industry and, particularly, the motion picture industry. The current prevalent business model in the movie industry seems to be that of the sequel. This model is essentially a tactic employed by studios to capitalize on the existing brand name and influence the purchasing decision by making the unpredictable a bit more predictable and trying to reconnect with the already established audience base. The production of sequels is encouraged
5 in the movie industry because it provides a built-in audience for subsequent products and, in this way, it is easier to relate to consumers.
From a list of highest grossing films of 2014 (Box Office Mojo), out of the first 20 movies, 10 are sequels (such as ‘The Hunger Games: Mockingjay - Part 1’ or ‘Captain America: The Winter Soldier’, which are in the Top 5). Out of 30 movies, 14 are sequels, each with a total gross between $102 million and $335 million (for The Hunger Games). Consequently, it is safe to say that almost 50% of the top 30 movies released in the US in the past year are sequels. For 2015, there are 31 sequels planned for release in the US. This data provides strong support to the claim that the sequel strategy is sought after and successful in this business.
In this study, I will build on insights from branding and signaling theory to explore consumer demand regarding extensions of a product (i.e. movies). As demand is highly uncertain in movie industry, the release of new products comes attached with a great risk; thus, sequels are an increasingly popular option used by movie studios to decrease this risk (Situmeang, Leenders, & Wijnberg, 2014). Movie sequels (as brand extensions) are employed as a means to capitalize on the already established brand name. This idea is supported in the literature by the ‘carry-over effect’ (Hennig-Thurau, Houston, & Heitjans, 2009; Keller, 1993) which shows how product quality signals ’carry-over’ to the next product of a brand, affecting its performance.
Signaling theory is fundamentally concerned with reducing information asymmetry between two parties, which occurs when “different people know different things” (Connelly et al., 2011; Spence, 2002). In the seminal paper on signaling theory, Spence (1973) focused on the job market to analyze the signaling function of education. The quality of job candidates is signaled by their education, the diplomas and results that they can show to potential employers to reduce information asymmetries. Since then, signaling theory has been applied in a wide variety of
6 disciplines ranging from anthropology to zoology (Bliege, Smith, & Bird, 2005) and, especially in the management literature, this theory has gained momentum. To give an example, CEOs can signal the value of their firm to potential investors through financial statements or by showing their concern regarding social and environmental issues (Connelly et al., 2011).
To put it briefly, in a broader business context, firms try to gain legitimacy by signaling the quality of their products to the audience, thus decreasing the risks associated with the purchase (Connelly et al., 2011). By using signals, the information asymmetry between the two parties – producer and consumer – is greatly reduced. This is especially valuable for experience goods, where the true quality of the good is known only after consumption has occurred.
Thus, an industry that is particularly ripe for applying signaling theory is the motion picture industry, characterized by high risks, but also high profits. Producers and studios try to signal the quality of their upcoming projects in such a way that it attracts a greater audience by reducing information asymmetries and the risk of purchasing a product without knowing its quality beforehand. In this industry, there are many types of signals used, such as famous actors and directors, exotic locations or special effects. Focusing on movie sequels in particular, reviews and ratings can act as quality signals (Basuroy, Chatterjee, & Ravid, 2003; Duan, Gu, & Whinston, 2008; Situmeang, Wijnberg, & Leenders, 2014).
The determinants of success for new editions in a product series are shown to originate in the reviews and sales of the previous products (Situmeang, Leenders, et al., 2014). Based on this reasoning, movie sequel success can be predicted by looking closer at the reviews, ratings and sales of the previous films in the series. If there is variability in the opinions regarding the movie, either in the consumer or in the expert opinions, this will have an effect on consumer purchase decisions (Situmeang, Wijnberg, et al., 2014; Sun, 2012). Situmeang et al. (2014) look at past
7 editions of video games and show that a high variability of evaluations (among reviewers of a certain type – expert or consumers) acts as a negative weighting factor on the purchasing of the sequel. Consequently, a lack of consensus among reviewers of a particular type weakens the strength of the signal (Situmeang, Leenders, et al., 2014); consumers will be confused as to the quality of the upcoming product, they will not know what to expect and this will increase the risk associated with the purchase.
With the spread of the Internet and the popularity of forums, consumers and experts alike are willing and even enthusiastic to share their opinions with fellow peers. This ease with which opinions and attitudes are shared nowadays greatly influences the ’carry-over effect’. Reviews act as signals of consumers’ and experts’ appreciations of the brand, influencing the performance of brand extensions. The Internet itself has become over the years a monumental database of user opinions and critical assessments, providing the perfect opportunity for analyzing how these opinions influence subsequent releases in a product line. The movie reviews website Imdb is the most influential source regarding movie information, having a global rank of 48, well above related sites – Rotten Tomatoes, 452 and Metacritic, 1234 (Alexa Ranks).
Despite the popularity of the sequel strategy, only one study has so far tackled the influence of both types of reviews (consumers’ and critics’) on the success of brand extensions (Situmeang, Wijnberg, et al., 2014). However, the interaction between word of mouth and critics’ reviews represents a gap in the literature. If both consumers and critics send signals regarding the quality of the product, the consumer is liable to receive contradictory signals and face ambiguity and confusion with regard to the purchase decision. It is critical to understand the role that each type of review has on consumer purchase decisions and how this variability would impact brand
8 extension success. Specifically, a high degree of variability could deter audiences from seeing the movie sequel.
The paucity of research to illuminate these issues resulted in this study, as I would like to shed light on the problem of variability between consumer and critics’ evaluations, using the movie industry and brand literature as a background. I believe that a lack of consensus in evaluations of previous editions by the two parties would lead to confusion regarding the quality of the extension. This confusion, depending on its expanse, would potentially deter consumers from making the purchase decision. Identifying movie sequels as brand extensions, I would like to answer the following research question: How does a lack of consensus between consumer and critic evaluations for the previous movie affect demand for the sequel?
The rest of the paper is organized as follows: in Chapter 2, the relevant literature concerning brand extensions and third party information is reviews. Then, in Chapters 3 and 4, the methodology and results are presented. Chapter 5 ends this study by discussing the implications, limitations and paths for future research.
9 Chapter 2. Literature Review
The literature review starts with an overview of the relevant branding literature, focusing on brand extensions and categorizing movie sequels as extensions. Then, the impact that third party information (i.e. critics and consumer reviews) has on consumer appeal and demand is explained with theories and examples. Moreover, the concept and operationalization of variability is presented. In the end, the hypotheses are introduced and developed.
2.1. Brand equity and brand image
Brand literature has its foundations in the associative network memory model, where information takes the form of nodes connected by links which vary in strength. Customers store brand knowledge as nodes in their memories, linking it to various cues. For instance, if a consumer thinks of caffeine, he might think of Pepsi or Coca-Cola. Producers struggle to capture “top-of-the mind” accessibility for their brand in memory and create strong, favorable and unique associations in consumer’s minds, so that when consumers are given a certain cue (such as caffeine) they think of Cola first and not Pepsi (Keller, 1993).
Brand equity is an extremely important concept in the business literature, because organizations can derive competitive advantage from developing and maintaining positive brand equity. The most widely adopted definition posits that brand equity is the incremental value of a product due to the brand name (Shocker, Srivastava, & Ruekert, 1994). This value is based on customer perceptions of that brand, the image of the brand conjured in their minds. Consequently, another concept emerges – brand image.
Keller defines brand image as the cumulative perceptions about the brand held in consumer memory (Keller, 1993) and it depends on the associations consumers have with the brand,
10 particularly how strong, favorable and unique these associations are. Brand image is developed over time through continuous effort from the producer’s part and is authenticated through the consumers’ direct and indirect experience with the products. Indirect experience is of particular interest as it refers to how consumers view the brand after being exposed to third party information, from experts (such as critics’ reviews) and/or peers (consumer recommendations) (Basuroy, Chatterjee, & Ravid, 2003; Chevalier & Mayzlin, 2006; Duan, Gu, & Whinston, 2008; Reinstein & Snyder, 2005; Sen & Lerman, 2007).
Summarizing, brand equity provides value to consumers and, through the brand image, it also provides confidence in the purchase decision (Keller & Aaker, 1992). This is especially helpful in the launch of a new product.
2.2. Brand extensions
In the branding literature, brand extensions form an area that is affected by the original brand’s equity. Brand extensions offer a risk-reducing strategy for managers: by taking advantage of brand name recognition, the risk associated with introducing a new product on the market is reduced as the strong brand name provides familiarity and knowledge. If the associations held in consumer memory regarding the parent brand are strong, favorable and unique enough, they will be helpful to the extension – consumers trust the brand and associate lower risks when purchasing its extension (Keller & Aaker, 1990; Park, Milberg, & Lawson, 1991).
Thus, the overall attitude and appeal consumers have towards the main brand will transfer to the extension. Attitude and appeal depend on the consumers’ perception of brand quality (defined as offering superiority and excellence). If there is a high quality perception towards the
11 original brand, consumers will have more favorable attitudes toward the extension (Keller & Aaker, 1990).
An important variable in the success of a brand extension is the perceived fit with the parent. Fit, in Keller’s view, refers to similarity between the original and extension products, focusing on the product-feature relationship. If there is high fit, the image of the main brand will transfer to the extension and the perceived quality will be enhanced. However, if the fit is low, undesirable beliefs and associations might be stimulated, rendering the extension derisory or even farcical. Keller and Aaker (1990) give the example of a McDonald’s photo processing service, which customers rate as inappropriate because there is no fit between the image of the restaurant and the extension; the restaurant is perceived as not having the competence to provide a photo service (Keller & Aaker, 1990).
A noteworthy addition to this point is offered by Park et al. (1991), who introduce brand concept consistency as another factor detrimental to brand extension success, along with product feature similarity. They argue that fit is not just one-dimensional (as proposed by Keller & Aaker), by bi-dimensional. Park et al. (1991) posit that evaluations of brand extensions depend on the fit between the new product and the existing brand, and this fit consists of two factors: product feature similarity and brand concept consistency. A brand concept is the abstract meaning of the brand as perceived by consumers and it originates both in the product’s features as well as the organization’s efforts to create meanings and images for the product (e.g. high price, beautiful design).
Park et al. (1991) make the distinction between functional brands (having functional concepts, related to product performance, problem-solving, and practicality) and prestige brands (having symbolic concepts, related to self-expression). They demonstrate that both factors (feature similarity and concept consistency) are vital to brand extension success; however they emphasize
12 the importance of the latter factor for the prestige brand. This entails that, even if there is low feature similarity between the new product and the parent, a consistent brand concept will lead to extension success. This results in a greater extendibility of prestige brands rather than functional brands across different product classes. The reasoning is that prestige brand concepts are more accessible and more easily retrieved from memory (Park et al., 1991). A straightforward example would be that of a clothing company such as Hugo Boss, which successfully extends into dissimilar product classes – jewelry and fragrances – due to the fact that it maintains a consistent brand concept of luxury and value.
2.3. Motion Picture Industry
One of the most interesting industries in which branding literature can be successfully applied is the motion picture industry, because firstly, there is a lack of studies regarding hedonic, experiential brands – the bulk of the literature is focused on functional products, but the same concepts could very well be applied in an experiential context as well. Secondly, the extension strategy is very popular and profitable in this industry in the form of movie sequels. In 2014, out of the 20 highest grossing films, 10 are sequels, each having earnings between $102 million and $335 million. For 2015, 31 sequels are scheduled for release in the US, all with highly recognizable brand names, such as The Hunger Games or The Avengers (according to Box Office Mojo). Movies as brands
Generally, the branding literature is focused on utilitarian products, which accomplish a functional, practical task (Hirschman and Holbrok, 1982). However, I believe that branding theories could also be applied to hedonic products, whose consumption is linked to a sensory experience and enjoyment. Motion pictures can thus be seen as brands, striving to boost their equity, competing
13 for customer attention. The same reasoning is offered by Sood and Dreze (2006), comparing Hollywood with manufacturers of consumer packaged goods, who brand their products, with each film becoming a branded product.
Movie sequels as brand extensions
In the literature, there are three articles that have investigated movie sequels as brand extensions (Basuroy et al., 2003; Hennig-Thurau, Houston, & Heitjans, 2009; Sood & Drèze, 2006). Sood and Dreze (2006) analyze the psychological reactions of consumers to the title of the sequel; Basuroy and Chatterjee (2008) investigate the impact of various sequel characteristics on box office success and Hennig-Thurau et al. (2009) present an approach to measure the monetary value of brand extension rights.
In the movie industry, sequels are brand extensions that studios and producers use to capitalize on the success of the original product by launching another product that reprises mainly the same characters, but who are evolving in a new situation (Sood & Drèze, 2006). The sequel strategy is regarded as easier than introducing a completely new product on the market, because it already has an established audience base, so the effort put into marketing the new product could be significantly reduced.
Moreover, sequels are shown to generally perform better in the marketplace as the original product created a positive image, which will then be carried over to the extension. There is awareness, excitement and anticipation surrounding the release of a loved brand’s extension, and, thus, sales will increase (Karniouchina, 2011; Keller, 2003; Sood & Drèze, 2006). The phenomenon that the image of earlier products paves the way for product extensions, or later editions in a series, and influences how they are perceived is described as the “carry over mechanism” (Hennig-Thurau et al., 2009; Sood & Drèze, 2006). The carry-over mechanism
14 depends on how strong, favorable and unique are the associations created by the earlier products in consumer’s minds.
The experiences that consumers had with the editions that precede the sequel have been found to be a major contributor to the market success of these sequels (Völckner & Sattler, 2006). To explain this relationship, it has been argued that the ideas and impressions connected to the earlier editions are transferred to the next edition (Keller, 2003; Situmeang, Wijnberg, & Leenders, 2014; Sunde & Brodie, 1993) and that the popularity of the past editions builds up anticipation and excitement towards the new product in the series.
With the advent of the Internet and the increase of online blogging, ideas and impressions of the brand are shared between consumers on online discussions forums. Turning to the movie industry, consumers flock to websites such as IMDB or Rotten Tomatoes to decide what they should or should not watch next. This decision is heavily influenced by third party information.
2.4. Third Party Information: Word of mouth and Critics’ reviews
I make the distinction between two types of third party information: word of mouth (which refers to ratings and reviews given by consumers who have experienced the product) and critic reviews (offered by experts or specialists in the field, such as film critics). In the paragraphs below, I will explain how each type of third party information influences consumer appeal and product performance as suggested by the relevant literature.
Word of mouth (WOM)
Word of mouth involves informal communication among consumers regarding products, transferring knowledge, impressions and experiences (Liu, 2006). There are two types of WOM, online and offline. I choose to focus only on online WOM, because, in contrast to offline WOM,
15 opinions and ideas are kept on record and are easily accessible, and do not “disappear into thin air” (Sen & Lerman, 2007). Product review websites have quickly gained popularity and are now widespread on the Internet. Online reviews have become a typical component of many products from a wide range of categories (e.g., electronics, automobiles, travel, books, movies or music). Especially with the rise of online sales, consumers base their purchasing decisions on recommendations; they can easily rate and review any type of product, from functional to hedonic and websites such as Amazon.com offer a secure platform on which to do so. This study is focused on ex-post WOM, which is based on actual consumption – the views expressed only after experiencing the product.
The effects of WOM on customer purchasing decisions are well documented in the marketing literature (eg. Herr, Kardes, and Kim 1991). Many of these studies are focused on experiential goods and use the creative industries as a background, finding that consumer reviews play a key role in product sales (Dellarocas, Zhang, & Awad, 2007; Duan, Gu, & Whinston, 2008; Liu, 2006).
1) WOM Volume and Valence
In the WOM literature, the most important attributes of reviews that have been studied are volume and valence. Volume refers to the number of customer reviews, while valence refers to the nature of those reviews (positive or negative). Studies offer mixed results regarding the importance of the two attributes: some studies show that the volume (and not the valence) of product reviews has explanatory power for consumer behavior and market outcome, influencing customer purchase decisions.
For example, Liu (2006) looks at aggregate and weekly box office revenues to explain how online WOM influences motion picture success. He finds that WOM has an effect on consumers’
16 awareness and the volume of reviews, not the valence, has the most explanatory power of box office success. Providing an example from the music industry, (Dhar & Chang, 2009) show that the volume of blog posts about an album results in higher sales.
In addition to this, Duan et al. (2008) introduce an important distinction between awareness and persuasion effects. Awareness effects indicate that the reviews solely convey the existence of the product to potential customers; the reviews have an informative purpose. On the other hand, persuasive effects occur when the reviews shape the customers’ perceptions, influencing their purchase decisions. In their study, Duan et al. (2008) demonstrate that higher movie ratings do not lead to higher sales, so valence does not impact sales; however, volume – the number of reviews – is associated with sales, thus showing that reviews do not have a persuasive effect, but an awareness one.
On the other hand, Godes & Mayzlin (2004) have looked at Usenet conversations about television shows and how they relate to Nielsen viewership ratings. They find evidence that volume does not have explanatory power; instead they show that the dispersion of conversations among different newsgroups has significant explanatory power. Also, Chevalier & Mayzlin (2006) emphasize the importance of valence for book reviews and show that the impact of one-star reviews is greater than the impact of five-one-star reviews.
2) WOM Negativity effect
In the WOM literature, there has been increased attention given to the “negativity effect”, whereby consumers place greater weights on negative rather than on positive messages or recommendations surrounding a product. An explanation would be that negative information tends to be more thought provoking (Rozin & Royzman, 2001). Drawing from psychology, Rozin & Rozyman (2001) give many examples in life where the negativity effect can be observed. A notable instance
17 would be the Hindu Indians, in which people of higher castes are contaminated (their social status decreased) from interaction with people from lower castes. However, lower castes do not benefit from this interaction. Consequently, contact with lower castes has a larger impact than contact with higher castes. This is called negativity bias, through which negative entities transfer properties by contact much more than positive entities do. They show that the principle holds in a wide range of domains.
In the context of consumer WOM, there is strong evidence that negative information is more valuable to the receiver than positive information – consumers are more likely to consider negative WOM as useful for decision making rather than positive WOM (Ahluwalia & Shiv, 1997; Skowronski & Carlston, 1987; Weinberger, Allen, & Dillon, 1981). For instance, Weinberger et al. (1981) found that unfavorable product ratings tended to have a greater impact on purchase intention than did favorable product ratings.
On the other hand, Sen & Lerman (2007) do not observe a negativity effect in the case of hedonic products. They analyze the negativity effect of online WOM for two types of products – functional and hedonic and find that this effect is moderated by product category differences. There is negativity observed in the case of utilitarian products (durables), but no effect in the case of hedonic products (such as movies, music etc.). The explanation for this would be that in the case of hedonic products, readers attribute the negative opinions to the reviewers’ internal (non-product related) motivations, whereas for utilitarian products, reviewers have external (product related) motivations. Moreover, consumers are usually in a good mood when reading reviews about hedonic products, as they are looking forward to choosing products that will entertain them (Sen & Lerman, 2007).
18 In addition to this, a study focusing on the travel industry found that both positive and negative reviews increase awareness in consumers, with the addition that positive reviews improve the attitude towards the hotels (Vermeulen & Seegers, 2009). The more ’buzz’ or conversation there is surrounding a product, the more consumers will want to purchase it, disregarding the actual nature of that buzz (Godes & Mayzlin, 2004). Thus, even negative reviews are better than no reviews at all.
With regard to new product development and brand extensions, WOM is shown to play a key role, because it creates ‘buzz’ around the product and consumers are more inclined to purchase if they are well acquainted with the product beforehand (Liu, 2006). Especially in industries where uncertainty is high and value cannot be pinpointed before the product is released, as is the case with motion pictures, video games or music for instance, customers base their decisions on recommendations by peers or experts. The motion picture industry specifically is known for high risks, where, out of ten movies produced, 6 to 7 do not recoup their costs (Vogel, 2015; Eliashberg and Shugan, 1997) and WOM is shown to be a critical factor in the success of the film (Elberse & Eliashberg, 2003; Hirschman & Holbrook, 1982).
Critics’ Reviews (CR)
Critics are seen as key actors, especially in the literature on cultural industries, because they form a link between products and audiences (Debenedetti, 2006). Consumers read book reviews before deciding what to purchase next, look at what restaurant critics have to say about the quality of the food or check the latest movie reviews from critics who themselves have become a brand (such as Peter Travers or Roger Ebert). This idea can hold true for many other industries as well, because critics offer an independent and well-written review of a product, service or experience, thus
19 informing and educating consumers. Reading a review is seen as a source of utility in itself, being a thought-provoking material and serving as a subject for everyday conversations (Cameron, 1995; Shrum, 1991). Also, consumers who are more knowledgeable about the product category tend to rely more on critical reviews than recommendations from peers (Debenedetti, 2006).
Critics are seen as separate from the actual producers, so they are free from bias and can offer an objective viewpoint, based purely on the merit of the work or product (Debenedetti, 2006).
However, an opposing view is that critics often collude with the market, molding their opinions on those of the target audience in order to increase their readability and, in effect, promote themselves by appealing to the mass market; quotes from their reviews can be incorporated into publicity material for the good (Larceneux, 2001). There can also be a ‘standardization of opinions’ as critics try to gain the approval of their peers and copy (to a certain degree) each other’s ideas and views (Debenedetti, 2006).
Many studies show that there is a positive correlation between critical evaluation and the commercial performance of a product (Basuroy et al., 2003; Chakravarty, Liu, & Mazumdar, 2010; Eliashberg & Shugan, 1997; Lampel & Shamsie, 2000; Reinstein & Snyder, 2005). Lampel and Shamsie (2000) make the distinction between products that have low versus high signaling properties, showing that critical reviews have a stronger impact on commercial performance for low-signaling products.
The motion picture industry has received the most attention regarding critical reviews and their impact. For instance, Reinstein & Snyder (2005) use the reviews of two famous critics (Gene Siskel and Roger Ebert) and find that positive reviews have a considerable effect only on the opening weekend box office sales. On the other hand, Basuroy et al. (2003) find that both positive and negative reviews are correlated with weekly box office revenue for a period of eight weeks.
20 Eliashberg & Shugan (1997) do not find evidence of correlation for early box office results, but instead show that critics’ reviews correlate with late and cumulative box office.
1) Influencers or predictors
In the literature, there is ambivalence with regard to the role of movie critics as either influencer or predictor. The predictor perspective suggests that the critics simply can statistically or intuitively predict the film’s success in the long run, but does not actually influence box office performance. This contrasts with the influencer perspective, whereby the critic has the power to influence the public’s opinion and thus affect subsequent success. Eliashberg & Shugan (1997) were the first to research whether the critic is either predictor or influencer of box office performance; they found that movie critics act only as predictors of long-term movie performance, but cannot influence early movie revenues.
On the other hand, Basuroy et al. (2003) find that critics have a dual perspective: they are both influencer and predictor, as positive and negative reviews correlate with box office revenue over the period of eight weeks, not only early on. Reinstein & Snyder (2005) use a different approach, categorizing movies by genres and by method of release (wide vs narrow). An influence effect is found only during the weekend box office for narrowly-released movies and for dramas. No such effect is observed for other genres or widely-released movies (blockbusters). Shrum (1991), in studying the Edinburgh Festival Fringe, came to the conclusion that critics ‘cannot make or break a show’, thus the influencer perspective is not substantial.
2) CR Volume and Valence
Another important point studied in the literature is the valence of reviews – positive or negative. According to Basuroy et al. (2003), negative reviews hurt revenue more than positive reviews help revenue. This effect can be seen only in the early weeks after the film is released.
21 Moreover, an interesting distinction is made between frequent and infrequent moviegoers Chakravarty et al. (2010), with the result that critical reviews have more influence on frequent moviegoers. Negative comments from critics tend to reduce demand in frequent moviegoers, but have no effect on infrequent moviegoers. Regular consumers of a product are more knowledgeable and develop “elite tastes” (Holbrook, 2005) and are thus more interested in the technical and artistic aspects of the movie – aspects covered by critics’ reviews. They are not interested in just a recommendation, they are looking for an educated opinion. Also, it is important to note that word of mouth has an inverse effect, whereby negative WOM influences infrequent moviegoers only.
Another valid point in the literature is that negative reviews are not detrimental to success. Shrum (1991) argues that reviews have more impact by providing visibility rather than through a positive evaluation; so even a negative review can help the product. Thus, the volume of reviews is more important than their valence. This is supported by a study concerning the book industry, focusing on reviews by New York Times book critics (Sorensen & Rasmussen, 2004). Persuasion and awareness effects on consumers are analyzed, the result being that reviews are primarily informative and even negative reviews can increase sales. The authors conclude that: “all publicity is good publicity”.
2.5. Variability
As noted above, the role of third party information is to reduce uncertainty regarding the purchase decision of consumers. Both consumer and critic reviews act as quality signals for the film, helping consumers make a purchasing decision. As such, opinions of experts and other users regarding previous products in a series guide potential consumers, reducing information asymmetries and assisting in forming an attitude towards the extension product. If this signal is ambiguous, meaning
22 there is variability of opinions, consumers will lack a clear understanding of the quality of the product.
In the motion picture industry, it is important to note that critics and regular consumers focus on different aspects of a film: experts usually analyze the stylistic value, acting and directorial merit, originality of the story and even technical aspects such as camera angles and lighting. They try to educate as well as entertain audiences with their writing. On the other hand, moviegoers just assess the overall movie, whether or not it was good, ending with a clearly formed recommendation to go or not to go. There are not many nuances and there is no emphasis on niche aspects of the film (Chakravarty, Liu, & Mazumdar, 2010; Holbrook, 2005). Consequently, the tastes expressed by critics do not necessarily mirror those expressed by consumers and variability will occur as critics and consumers express different views, rating the film according to those views. Austin (1986) suggests that film attendance increases if the public is in agreement with the views of the critics.
The degree of consensus between critics and consumers could thus play an influential part in the performance of a product’s extension, namely the movie sequel. Consensus here refers to how similar the opinions of experts and average consumers are, if they agree on their recommendations or not. If both consumers and critics send signals regarding the quality of the product, the consumer is liable of receiving contradictory signals and face ambiguity and confusion with regard to the purchase decision.
In the branding literature, several studies recognize variability as playing a major role. Dacin and Smith (1994) look at product portfolios of several brands and research how variability in the quality of products impacts the overall brand strength and a potential extension. They conduct an experiment exposing their respondents to product series with varying differences in
23 quality. The authors believe that if there is high variability of quality in this product portfolio, consumers’ judgements regarding the brand’s strength will be negatively altered because the quality signals sent by the brand are mixed and so they become less reliable. The results of their research support the above statement and the authors conclude that failure to manage variability will reduce brand strength and consumers’ favorability towards extension products.
In addition to this, Volckner and Sattler (2006) find that variability in quality decreases brand performance, as the strength of the brand becomes diluted by signals of non-stable quality which leads to consumer uncertainty. Consumers are also shown to regard variability of service quality as an indicator of inferior firm quality and reliability (Desai et al. 2008). Focusing on experience products, Situmeang et al. (2014) study the effect that evaluations of past editions have on sequel success in the video games industry. Looking at variability from two perspectives - within expert reviews and within consumer reviews - they find that high variability among evaluations of past editions negatively affects customers’ intentions to buy the sequel. In other words, variability in evaluations creates uncertainty regarding the decision to purchase the extension product as consumers develop less favorable attitudes towards extension products in the presence of high variability.
In the context of experience products and more specifically motion pictures, I would like to examine in detail the effect of variability on consumer appeal of an extension product by analyzing variability from three perspectives:
Firstly - by looking only at expert ratings, the effect that variability in expert ratings has on consumer appeal towards an extension product.
Secondly - by looking only at consumer ratings, the effect that variability in consumer ratings has on consumer appeal towards an extension product.
24 Thirdly - by looking at the interaction between expert ratings and consumer ratings, I would like to see how variability among experts and consumers shapes attitudes. More specifically - how a difference of opinion between the two groups (expert and consumer) regarding a previous product in a series impacts consumer purchasing decisions of the extension product.
To achieve this, I will build on insights from the paper by Situmeang et al. (2014), because they offer a comprehensive view of variability in both consumer and expert evaluation scores in the context of experiential products (video games).
However, it is important to note that my approach differs from their research along two dimensions. Firstly, in order to test their hypotheses and assess the impact of variability, Situmeang et al. (2014) used a database design by looking at sequel sales. However, I believe that an experimental design would be more suited to analyze how variability directly affects consumers’ judgements regarding an extension product and it would provide a higher internal validity. Secondly, in order to alleviate the important gap in the literature, I will look at the interaction between the ratings of experts and those of consumers, and how the purchasing decision is influenced when both reviews are simultaneously available, an approach which has not yet been covered by existing research. Consequently, I extend the definition of variability to also include the difference between the average ratings of consumers as compared to the average ratings of critics for a particular experiential product.
As a result, the aim of this research is to correctly identify the impact that variability has on the success of the brand extension, thus leading to the following hypotheses:
25 Variability within expert ratings - in the absence of WOM
Hypothesis 1a. In the context of experience products, audiences are more likely to have stronger appeal towards the extension product if there is low variability in expert ratings of the past edition.
Variability within consumer ratings - in the absence of CR
Hypothesis 1b. In the context of experience products, audiences are more likely to have stronger appeal towards the extension product if there is low variability in consumer ratings of the past edition.
Variability within consumer and within expert ratings
In order to better assess the impact that variability has, it is crucial to know how exactly consumers are influenced when both types of reviews are concurrently available. Few studies look at the interaction between expert and consumer recommendations, and which has more influence in the final purchasing decision.
One of these studies, Senecal & Nantel (2004), looks at multiple online recommendation sources (other consumers, experts, recommender systems) and uses an experimental design to test how each source influences online product choices. The authors consider that the source entitled ‘other consumers’ is more trustworthy than human experts and online recommender systems (which make a recommendation based on user profile). Their results show that a human expert recommending a product, although possessing more expertise, is seen as less reliable than peers. This view is also supported in other studies, mentioned previously, which introduce the idea that critics collude with the market as they try to improve their readability by appealing to the mass market. Also, they are seen as working together with the producer, complementing the overall
26 advertising for the product and thus, they lose reliability (Bourdieu and Delsaut, 1975; Larceneux, 2001; Debenedetti, 2006). Summarizing, in an online context, consumers find peer reviews of products as being more trustworthy than expert reviews.
In addition to this, Smith et al. (2005) study the effect of recommendations on consumer purchasing decisions in an online environment. They base their study on the fact that consumers are overwhelmed by the amount of information available during online shopping experiences and show that consumers make decisions based on peer recommendations, irrespective of the recommender’s expertise, especially for experience products.
In an experimental setting, subjects had three options when faced with a purchasing decision regarding both experience and utilitarian products: they could access either a consumer or an expert recommendation, or they could decide not to consider any recommendation at all. Overall, more subjects (47%) chose to access the consumer recommendation and 31% chose to access the expert recommendation. Focusing on experience products only and assuming that consumer and expert reviews are concurrently available, audiences are significantly more inclined to find their peers’ recommendations as being more salient than expert ones (59% versus 20%). In summary, online consumers accept recommendations from other consumers rather than experts in order to effectively manage the high amount of product information available. Thus, it can be hypothesized that:
Hypothesis 1c. In the context of experience products, a high variability in consumer ratings has a stronger negative effect on consumer appeal than high variability in expert ratings. Audiences are more influenced by other users’ ratings than expert ratings.
27 Variability between expert ratings and consumer ratings
By looking at variability between expert ratings and consumer ratings, I try to understand how a difference of opinion between critics and consumers influences appeal. If there is a lack of consensus between the two types of raters, consumers might be deterred to purchase the product extension, as they receive mixed signals regarding the quality of the previous product in the series. Therefore, it is expected that:
Hypothesis 2a. In the context of experience products, consumers will have more appeal towards a product’s extension if the previous product in the series has low variability between expert and consumer ratings.
Moreover, I believe the average rating of the previous product might also play an important role in the attitude towards the extension and there might be an interaction between variability and average rating in the sense that variability has more impact on consumers when the average rating is high. A higher average rating for the previous product has been shown to increase sales for a subsequent product (Chevalier and Mayzlin, 2006). Also, a higher average rating indicates a better quality, which could transfer to the extension (Sun, 2012). A low average rating for a previous product means poor quality, so consumers might disregard the extension from the start, not even looking at the consensus or variability between experts and other users. Consequently, I believe that the impact that variability has on consumer appeal might be influenced by the level of the average rating. Therefore, it can be hypothesized that:
Hypothesis 2b. In the context of experience products, there is an interaction between variability and average rating for a previous product in a series. Consumers will have more appeal towards products with low (rather than high) variability and high average rating.
28 In the next two chapters, I will address in turn each research objective and the corresponding hypotheses. Thus, Chapter 3 includes the methodology and results related to Hypothesis 1a, 1b and 1c, and Chapter 4 includes the methodology and results related to Hypothesis 2a and 2b.
29 Chapter 3. Methodology and Results: Study 1
In this chapter, the first objective of the research is addressed: to examine audience appeal in the presence of variability within critics’ ratings and within peer ratings. This objective refers to Hypotheses 1a, 1b and 1c and is realized through Study 1. First, a detailed description of the study is offered, followed by results.
3.1. Description of Study 1
The first study is comprised of three experiments, testing Hypotheses 1a, 1b and 1c. All the experiments use a between-subjects approach, randomizing respondents. Overall, there are six treatments in this study and participants are randomly assigned to just one treatment. This is crucial for obtaining accurate results, as the experiments use similar data and images and, without using different treatments, there could be major carryover effects – meaning that the performance or behavior of respondents is altered by their participation in a previous treatment.
In the following paragraphs, I will offer a detailed explanation of each experiment.
Experiment 1 - Variability in expert ratings (in the absence of WOM)
The first experiment is based on Hypothesis 1a: In the context of experience products, audiences are more likely to have stronger appeal towards the extension product if there is low variability in expert ratings of the past edition. Thus, this experiment seeks to answer the following question: How does variability in expert ratings of the past edition influence consumer appeal towards the extension product?
30 Description of Experiment 1
The first experiment is focused on critic ratings for the previous product in a series. There are two movies - Movie X: Part 1 and Movie Y: Part 1 and each have been rated by three independent critics on a scale of 1 to 10. In order to accurately test the hypothesis, it is important for both films to have the same mean critic score, but different variabilities. The mean critic rating is 7 and Movie X: Part 1 has low variability of 0.66 and Movie Y: Part 1 has high variability of 6. This experiment has two treatments, using a between subjects design. In the first treatment, participants are shown a poster for Movie X: Part 1, followed by 3 critics’ ratings as could be seen on a movie review website such as Imdb or Rotten Tomatoes; there is low variability in the ratings. The same reasoning is used for the second treatment. Participants are shown a similar poster (as can be seen in Appendix 1) for Movie Y: Part 1, followed by other 3 critics’ ratings; there is high variability in the ratings.
It should be noted that, for all three experiments, respondents are shown a standard movie poster, featuring only the title of the film, and not pictures or names of famous actors. In this way, respondents are not distracted and will focus only on the ratings (The posters are included in Appendix 1). After seeing their respective treatments, participants are asked to rate the likelihood of seeing the product extension - either Movie X: Part 2 or Movie Y: Part 2.
Experiment 2 - Variability in consumer ratings (in the absence of CR)
The second experiment is based on Hypothesis 1b: In the context of experience products, audiences are more likely to have more appeal towards the extension product if there is low variability in peer ratings of the past edition. Thus, this experiment answers the following question:
31 How does variability in peer ratings of the past edition influence consumer appeal towards the extension product?
Description of Experiment 2
The second experiment is focused on consumers’ ratings for the previous product in a series and follows the same layout as the first. There are two movies - Movie X: Part 1 and Movie Y: Part 1 and each have been rated by three consumers who have seen them. Both films have a mean critic score of 7, but different variabilities. Movie X: Part 1 has low variability of 0.66 and Movie Y: Part 1 has high variability of 6. This experiment has two treatments, using a between-subjects design.
In the first treatment, participants are shown a poster for Movie X: Part 1, followed by 3 consumers’ ratings as could be seen on a movie review website; there is low variability in the ratings. In the second treatment, participants are shown a similar poster for Movie Y: Part 1, followed by other 3 consumers’ ratings; there is high variability in the ratings. Then, participants are asked to rate the likelihood of consuming either Movie X: Part 2 or Movie Y: Part 2.
Experiment 3 - Variability in consumer and in expert ratings
The third experiment is based on Hypothesis 1c: In the context of experience products, a high variability in peer ratings has a stronger negative effect on consumer appeal than high variability in expert ratings. Thus, this experiment seeks to answer the following question: Is consumer appeal influenced more by variability in peer ratings or variability in expert ratings?
32 Description of Experiment 3
The third experiment encompasses both consumer ratings and critic ratings for the previous product in a series. As in the first two experiments, there are two movies - Movie X: Part 1 and Movie Y: Part 1 and each have been rated by three independent critics and three consumers. The same mean rating of 7 is kept for both movies. This experiment has two treatments, using a between-subjects design. The treatments have a mean rating of 7 (for consumer ratings as well as critic rating), but mixed variabilities.
In the first treatment, participants are shown a poster for Movie X: Part 1, followed by 3 critic ratings and 3 consumer ratings as could be seen on their favorite movie review website. Treatment 1 has high variability for the consumer ratings (6), but low variability for the expert ratings (0.66). In the second treatment, participants are shown a similar poster for Movie Y: Part 1, followed by other 3 critic ratings and 3 consumer ratings. Treatment 2 has low variability for the consumer ratings (0.66), but high variability for the expert ratings (6). After seeing their respective movie poster and ratings, consumers are asked to assess the likelihood of seeing the extension product, Movie X: Part 2 or Movie Y: Part 2.
Variables and Measures
Independent Variables
Variability in expert ratings and variability in consumer ratings are the main independent variables, as they are used to examine effects on audience appeal towards a product extension. These variables were calculated based on the Situmeang, Wijnberg, & Leenders ( 2014) research, by taking the variance of the evaluation scores of past editions in the same series. Variance is the most
33 commonly used measure of variability in the branding and marketing literature (Dacin & Smith, 1994; Situmeang et al., 2014; Völckner & Sattler, 2006).
There is scarcity of research regarding how this variance is defined and delimitated. Two studies stand out. Sun (2012) studies the influence of variance in experiential product ratings on the seller‘s performance and market strategy. Using data from two industries, motion picture and book industry, he focuses on consumer heterogeneity – consumers have different tastes, and, thus, rate products differently. Sun (2012) seeks to understand whether a high or a low variance in product ratings is better for the seller, concluding that high variance is associated with niche products and sellers need to attract well-matched consumers for these niche products. However, there is no clear definition of what constitutes low or high variance.
Turning to the branding literature, Luo, Raithel, & Wiles (2013) study the impact of brand rating dispersion on firm value. They define dispersion as variance in consumer ratings of brands. And again they offer no clear delimitation between a low variance and a high variance. Their research shows that if there is high dispersion, the brand’s rating will have a lower impact on returns, noting the fact that high dispersion reflects inconsistencies of the brand. They conclude by emphasizing the importance of dispersion as a brand management tool that should be widely implemented and used. However, as in Sun’s research (2012), there is no clear delimitation between low dispersion (variance) and high dispersion (variance).
In consequence, due to the lack of substantial research regarding what constitutes low or high variance of an experiential product, I will operationalize Low Variability as variance that is below 1 and High Variability as variance that is above 5.
34 Dependent Variable
The main dependent variable is audience appeal and it is used to assess how audiences respond to an experiential product’s extension after being exposed to critic and peer ratings of the previous product in the series. Audience appeal is measured by the likelihood of seeing the movie sequel. Responses were recorded on a 6-point Likert Scale (anchored by Very unlikely to Very likely).
Control Variables
All the experiments were controlled for age (birth year), gender (male=1 and female=2) and education (1=High School and below, 2=Bachelor Degree and 3=Master or PhD Degree). In addition to this, four possible covariates were added:
Frequency of movie attendance and Frequency of using movie review websites
Studies suggest that frequent users of a product will differ in their responsiveness towards online information (Austin, 1986; Chakravarty, Liu, & Mazumdar, 2010). A distinction is made between frequent and infrequent moviegoers, with the finding that the former category is less susceptible to persuasive messages or reviews from consumers regarding the product; they are more susceptible to critic reviews. Thus, it is relevant for this study to check how familiar respondents are with motion pictures as well as review websites as this might influence how they perceive the variability in the ratings. Participants were asked how frequently they go to see a film and responses were recorded on a 5-point Likert Scale (anchored by Never to Very Frequently). Then, participants were asked how often they use movie review websites such as Imdb or Rotten Tomatoes to make a choice regarding a film; responses were recorded on a 5-point Likert Scale (anchored by Never to Very Frequently).
35 Importance of critical reviews and importance of consumer reviews
Consumers may be more inclined towards a specific type of reviewer as suggested by Chakravarty et al. (2010). Frequent moviegoers are thought to mostly look at critics’ ratings and reviews, whereas infrequent moviegoers trust other consumers more than critics. Consequently, it might prove useful to check this inclination. Participants were asked how important are critical reviews and how important are consumer reviews when considering which movie to watch; responses were recorded on a 4-point Likert scale (anchored by Not Important to Very Important).
Measurement - SPSS 22 was used for all statistical analyses.
3.2 Study 1 Results
This subchapter provides an overview of the results for Study 1. For each experiment, firstly, general data overview is provided, including demographic characteristics of the sample. Descriptive statistics of the study and correlations follow. Lastly, derived hypotheses are tested using various statistical techniques.
As noted before, Study 1 is comprised of three between-subjects experiments, each with two treatment conditions. For all the experiments, respondents were randomly assigned to one of the two treatment conditions so that there was an equal number of respondents in each group. Overall, there were 60 respondents for each experiment, with 30 participants in each treatment group. The detailed analysis of each experiment is offered below.
36 3.2.1. Experiment 1
Data Overview
With respect to the demographic variables, it can be seen from the overall frequencies of gender that, from all participants, 45% were male and 55% female. In more detail, the results of a cross-tabulation show that there were 13 males and 17 female respondents in the first treatment and 14 males and 16 females in the second.
With regard to educational levels, the majority of respondents (73.3%) have a Masters or PhD Degree. By doing a cross-tabulation to see the educational levels per treatment, 19 of respondents in the first treatment have a Masters or PhD versus 25 respondents in the second treatment. The age of participants in this experiment is between 20 and 25, with the average age being 23 years old overall. The mean age of respondents for the first treatment is 22.6 years, and for the second treatment is 23.3 years old. Overall, there are no significant differences between the two treatments regarding the control variables.
Descriptive statistics
As preliminary analysis, it is necessary to test the assumptions of normality and homogeneity of variance (Field, 2009). Overall, the data follows a normal distribution as can be observed in the histogram and Q_Q and P_P plots. Skewness and kurtosis are also close to 0. Applying the rule of thumb and dividing the score for skewness (-0.26) by its standard error (0.30), results in -0.87, which is between the accepted ±1.96 limits. For kurtosis (-0.86), dividing by the standard error (0.60), gives -1.43, which is also within the accepted ±1.96 limits. The Levene’s test (Field, 2009) shows that the homogeneity of variance is not violated (p=0.109). The above data is found in more detail in Appendix 2.
37 Focusing on the main dependent variable, that measures consumer appeal for a product extension, it is important to note that the mean for treatment 1 (M=5.1, SD=0.84) is higher than the mean for treatment 2 (M=2.57, SD=1.07). The minimum point for this variable is 1 and the maximum is 6, so a mean of 5.1 reflects the fact that consumers have high appeal towards extensions coming from products with low variability in ratings; low variability in critic ratings is perceived more favorably than high variability.
Table 1.1. Dependent Variable Descriptives
Dependent Variable: How likely are you to see Movie X: Part 2 based on the critic ratings for the previous film?
N Mean Std. Deviation
Low Variability Group 30 5.10 .845
High Variability Group 30 2.57 1.073
Total 60 3.83 1.596
Covariates
Looking at the four possible covariates, several points are important to note. With regard to the first variable, movie-going frequency, the majority of respondents report on seeing movies Often (46.7%) and All the time (28.3%), resulting in an overall high mean of 4 (minimum value 1, maximum value 5). Thus, it is safe to say that respondents are familiar with this industry and they have informed opinions and attitudes.
The next variable analyzed is review website frequency of use. Respondents reported that they use movie review websites such as Imdb, Rotten Tomatoes etc. Often (40%) and All the time (30%), resulting in a high mean of 4 (minimum value 1, maximum value 5).
38 The third variable is importance of critic ratings. Participants consider ratings by critics as mostly Moderately Important (33.3%) when making a product choice (M= 2.30; minimum value=1; maximum value=4).
On the other hand, the next variable – importance of consumer ratings – has a higher mean (M= 3.12) and the majority of consumers consider ratings by other consumers as Very Important (40%) and Important (36.7%) when deciding on a product extension.
Table 1.2. Descriptive Statistics
Minimum Maximum Mean
How often do you watch movies? 1 5 4.00
How often do you use movie review
websites? 1 5 4.00
How important are ratings by film critics for
you when choosing which movie to watch? 1 4 2.30
How important are ratings by other consumers for you when choosing which movie to watch?
1 4 3.12
Correlations
In this part, correlations between all variables are explored. Control variables that were measured in this study were tested for correlation to the dependent variable, in order to know which of them should be included in the hypothesis testing (Field, 2009).
As can be seen in Appendix 2, age, gender and education are not significantly correlated with consumer appeal.
Before conducting a one-way between-groups analysis of covariance, the assumptions of covariance must be met. Firstly, all the four possible covariates were tested for independence – the covariates must not differ across the independent variable groups (Field, 2009). T-tests were
39 conducted and the results show that there are no significant differences for any of the covariates along the two levels of the independent variable
Another assumption to be tested is that the covariates are correlated with the dependent variable. However, this assumption is not met by any of the covariates. Furthermore, there are moderately strong correlations (p>0.05) between three of these variables, which would hinder the reliability of the model (results are shown in more detail in Appendix 2).
Concluding, the four variables cannot be used as covariates in this experiment and the model will be tested using a one way analysis of variance instead. Moreover, the above results are replicated for the other two experiments part of Study 1 and the four variables cannot be accurately used as covariates in neither experiment.
Hypothesis Testing
A main effect of variability in critic ratings on consumer appeal for a product extension was hypothesized (H1a). It was expected that low variability in critic ratings for the previous product leads to more favorable consumer attitudes towards the extension. This hypothesis was best tested using a one way between groups analysis of variance.
First, the homogeneity of variance is tested and the assumption that the variance of scores is the same for both treatments is met, as the significance value is greater than 0.05 (p=0.109) (Appendix 2).
Looking at the main results of the ANOVA, there was a statistically significant effect found, with F(1,58)=103.27, significant at the p<0.001 level.
40 Table 1.3. One-way ANOVA
Dependent Variable: How likely are you to see Movie X: Part 2 based on the critic ratings for the previous film?
Sum of
Squares df Mean Square F Sig. Between Groups 96.267 1 96.267 103.270 .000
Within Groups 54.067 58 .932
Total 150.333 59
With regard to the effect size, it is useful to compute Eta Squared (between groups sum of squares over total sum of squares).
This results in a value of 0.64 (96.26 / 150.33), meaning that 64% of the variance in the dependent variable (audience appeal) is explained by the independent variable (variability in expert ratings). Also, another variable that measures effect size is Cohen’s d, which in this case is 2.16, with a correlation coefficient ( r ) of - 0.8 (showing a strong negative correlation). According to Cohen (1988), all these values indicate a very large effect.
Table 1.4. Correlations
Variability in expert ratings
How likely are you to see Movie X: Part 2 based on the critic ratings for the
previous film? Variability in expert ratings Pearson Correlation 1 Sig. (2-tailed) N 60
How likely are you to see Movie X: Part 2 based on the critic ratings for the previous film? Pearson Correlation -.800 ** 1 Sig. (2-tailed) .000 N 60 60
41 Thus, it can be concluded that variability has a very significant effect on consumer appeal and the first hypothesis is accepted. When there is low variability in expert ratings for a previous product in a series, consumers will have more favorable attitudes towards the product extension, in this case, the movie sequel.
3.2.2. Experiment 2
Data Overview
By looking at the demographic variables, it can be seen that, overall, there were 43.3% male and 56.7% female participants. In more detail, the results of a cross-tabulation show that there were 14 male and 16 female respondents in the first treatment and 12 males and 18 females in the second.
With regard to educational levels, it can be seen that the majority of respondents (63.3%) have a Masters or PhD Degree and the rest, a Bachelor Degree. By doing a cross-tabulation to see the educational levels per treatment, 16 of respondents in the first treatment have a Masters or PhD versus 22 in the second treatment. The age of participants in this experiment is between 20 and 26,
Table 1.5. Cohen - Thresholds for interpreting effect size
Test Relevant effect size
Effect size threshold
Small Medium Large Very Large
Standardized mean difference d, ∆, Hedges’ g .20 .50 .80 1.30 Correlation r .10 .30 .50 .70
