University Of Amsterdam
Faculty of Economics And Business
Master Thesis: MSc Business Administration
“ THE DIMENSIONS OF GENRE COMBINATIONS USED IN
CATEGORY SPANNING IN THE MOTION PICTURE INDUSTRY “
Written by:
Lizzy van Drom
11421002
Supervisor:
Dhr. J.F.E. de Groot
ABSTRACT
The aim of this thesis is to research the different dimensions of genre combinations used in category spanning in the motion picture industry, given the fact that the concept of category spanning is receiving increasingly more attention from researchers. However, all studies use the concept of category spanning in general and fail to look at the specific dimensions between certain genres. This study aims to get more insight into the interaction and dimensions between different genre combinations in category spanning. In other words, under which circumstances could some genre combinations used in category spanning be a better match than other combinations and does this results in a better performance and rating of the movie. The empirical setting of this research is the Dutch Movie industry from 2011 until 2015. Based on the conducted extensive literature review a total of 8 hypotheses were formed. As dependent variables rating and performance were used, and the independent variables were frequency, audience similarity, and genre consensus. The type of movie, either mainstream or arthouse was tested for any moderating effects. To test the hypotheses a movie database was created including all the released movies in the Netherlands from 2011 until 2015. Data collection is done via the Dutch organization FDN (Filmdistributeurs Nederland) and the IMDB (International Movie Database). In total 514 movies matched the criteria and were included in the database. The data has been analysed using correlation and regression techniques in SPSS. The results of the study showed that 3 of the 8 hypotheses could be accepted, namely, the following: (1) Audience similarity leads to a higher performance, (2) There is a positive relation between genre consensus and rating, and (3) The effect of genre consensus on rating is moderated by the type of movie. To conclude, this research has made the first steps in identifying the interactions between different genres used in category spanning, and showed that there are indeed differences in performance and rating resulting from the different combinations. In other words, not all genre combinations have the same positive effect, concepts such as, audience similarity and genre consensus play an important role.
Statement of originality
This document is written by student Lizzy van Drom, 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.
TABLE OF CONTENTS
1. Introduction………. 4
1.1 Background……… 4
1.2 Research Objectives………..………..….……… 6
1.3 Significance of the Study……….……… 7
2. Literature Review……….. 8
2.1 Theoretical Background…………..….……… 8
Categories……….……… 8
Category Spanning……..……… 10
Motion Picture Industry……… 12
2.2 Hypotheses…………..….……….………. 13 3. Methodology……….……….. 17 3.1 Operationalization Research……….….………….………. 17 Libby’s boxes………..……….……… 17 Empirical Setting………..……… 18 Research Goal………..……… 18
3.2 Dutch Motion Picture Industry……….…….….………..… 18
Movie Database………..……….……… 18 Sample………..……… 18 Genres……….………..……… 19 3.3 Dependent Variables ……….….……….……… 19 3.4 Independent Variables ……….….……….……….. 20 3.5 Control Variables ……….….……….……… 23 3.6 Moderating Variable……….….……….……… 24 4. Results……….. 25 4.1 Descriptive Statistics……….….…….……… 25 Pearson Correlations……….………..……… 26
4.2 Hierarchical Multiple Regression Analyses………..……..……… 26
Performance………..……….………..……… 27
Rating……….……….………..……… 28
4.3 Summary……….……….….………..……… 30
5. Discussion………..….………..……….. 31
5.1 Discussion and Implications……….……… 31
5.2 Limitations and Future Research……….……… 34
6. Conclusion……….……….. 36
7. References……….. 37
1. INTRODUCTION
1.1 Background
A famous principle in children’s fairytale movies is that eventually “right always wins out over wrong”. Some people say that this is not the case in real life, however, ironically enough, this is the case in the Motion Picture Industry. The past decade the Dutch Motion Picture industry faced enormous threats from budgetcuts due to the global economic crisis paired with the increasing amount of illegal downloading of movies. However, since 2010 the Dutch Motion Picture Industry has managed to overcome these problems, and saw an increase in the number of visitors and the total revenue from box office every coming year, suggesting that people deliberately choose the ‘right and honest way’ of watching a movie. In 2011, the amount of cinema visitors in the Netherlands was around 30,4 million with a total gross revenue of 240 million euros (Nederlandse Vereniging van Bioscoopexploitanten 2011, p. 64). In 2015, these amount have increased to an impressive 33 million visitors and a total gross revenue of 275,5 million euros (Nederlands Vereniging van Bioscoopexploitanten 2015, p. 64), and as these numbers indicate, the Dutch Motion Picture is an industry of importance.
Different from other industries is that in this industry, the type of good that is being sold to the consumer, a movie, is considered an experience good. Experience goods differ from normal goods. The quality of an experience good are difficult to observe in advance, and can only be ascertained upon consumption (Frey and Steiner, 2010). Moreover, experience goods have a considerable short life cycle, which is approximately only 1 to 16 weeks (Krider and Weinberg, 1998). After the opening weekend, a movie faces a rapid decrease in revenue and a high amount of competition, because of the constant release of new movies (Krider and Weinberg, 1998). These special characteristics and tough competition highlight the importance of understanding and targeting the right interests of your audience, which is based on factors such as story lines, actors, directors, and genres (Krider and Weinberg, 1998).
This research will focus on those movie categories, called genres, especially on category spanning, in other words membership in multiple categories/genres. Up to this point, studies in
the field of organizational and economic sociology have mostly been focusing on the influence and consequences of categories on a product’s performance and the evaluation of the audience. Many of the authors particularly focused on the specific effect of category spanning on performance and evaluation. These researches (e.g. Durand and Paolella, 2012; Dobrev, Kim and Hannan, 2001; Freeman and Hannan 1983; Hannan, 2010; Hsu and Hannan, 2005; Hsu, 2006; Hsu, Koçak and Hannan, 2009; Negro, Hannan and Rao, 2010; Negro and Leung, 2012; Pontikes, 2012; Ruef and Patterson, 2009; Vergne and Wry, 2014; Zuckerman and Kim, 2003) all (slightly) differ in terms of definitions, concepts, terminology, and thus, logically the effect category spanning can have. However, there are two important aspects they agree on. Firstly, for a category to work there needs to be some kind of agreement among the audience. In other words, the audience needs to reach an agreement about what the label means in order for a category to emerge. Secondly, the category can only persists as long as the level of the reached consensus about the meaning remains high (Hannan, Pólos and Carroll, 2007).
An important element the authors do not agree on is the debatable effect category spanning can have on performance. According to Hsu (2006) having multiple genres targets a larger audience, and therefore, increases performance, however, it also makes it more difficult for audience to understand, in other words, the, so called, agreement is harder to reach, and this results in a lowered appeal. Thus, movies that use category spanning attract larger audiences, however, seem to be less appealing to those audience members. On the other hand, Zuckerman (1999), does not agree with these findings, and state that category spanning negatively influences the performance and evaluation, because, entities would experience reduced audience attention.
Noticeable, all the studies use the concept category spanning in general when researching it, they do not examine the specific interaction of certain combinations of genres. This results in a large research gap, since it is quite possible that a romantic-comedy movie could be a more successful category spanning example than a romantic-horror movie. Thus, the question remains what the effect is of different genre combinations in particular on performance and ratings. The purpose of this thesis is to take the first steps towards identifying these specific
1.2 Research Objective
In this study the focus will be shifted from the approach of researching category in general to more a focus on the dimension of the different genre combinations used in category spanning.
Specifically, this research aims to:
A. Making the first steps in identifying the specific genre ‘interactions’ and its effect on performance and rating.
B. Determining under which circumstances genre combinations positively influence a movie’s rating and performance.
In order to achieve this objectives, data from the Dutch Motion Picture Industry will be used to test the hypotheses by conducting hierarchical multiple regression analyses.
1.3 Significance of Study
In the first place this research will give new perspectives in the different possibilities and effects of certain genre combinations. It will contribute to the general discussion of the different theories about categorization and category spanning in the creative industries, in particular when using multiple genre combinations for a creative good, and possibly open new roads for future research. Secondly, the research has several (managerial) contributions to offer for film production and distribution companies. Knowing the effect of combinations of genres and their interaction is very useful information for producers and directors, because in all the steps of the process, such as the script, narrative style, marketing, of producing and distributing a movie, the type of genre plays an important role. It can positively influence the strategic process, overcome future mistakes, and optimize performance.
In the next section the literature on categorization, category spanning and the movie picture industry is reviewed. Based on this literature a total of eight hypotheses are formed. After the literature review, the research design and the variables are explained in the section Methodology. The data will be analysed using SPSS, the outcome of these analyses can be found in the results section. This research will end with the discussion, conclusion and references.
2. LITERATURE REVIEW
This chapter summarizes and analyzes the literature on categorization. The focus is on what categorization is, what problems are related to the concept and what the effect of categorization is on performance and rating. The concept of (strategically) category spanning will also introduced. Lastly, literature about the Motion Picture industry will be reviewed.
2.1 Theoretical Background
Categories and Classification
Increasingly more attention in research is given to the concept of categorization and categories, consequently several different definitions have emerged. In its broadest sense, the meaning of cateogirzation is the act of grouping ideas, objects, items or concepts into categories for a certain purpose. Hannan, Pólos and Carroll (2007) describe a category as a collective agreement made by the audience that is based on the extent to which the products share similarities. Similar to this, Negro, Hannan and Rao (2010) state that categories can be considered as social agreements. These agreements are about the labels assigned to sets of objects. Glynn and Navis (2013) state that categories are formed as social constructs to provide a conceptual system. This system can be constructed and reconstructed. Kuijken et al. (2013) argues that categories, or in other words, socially constructed cognitive orderings, form the core elements of the classification systems. The classification systems order and compromise the categories in certain domains and fields. Since those category systems provide clear boundaries, they are considered very helpful to identify market structures and consequently help to stabilize the market (White, 1981).
Another description of categories can be found in the article of Markus (1997), in here, categories are identified as schemas. These schemas form a structure of knowledge that helps to express objects with the same characteristics. Schemas can be used to identify products, producers and markets, whenever these schemas are expressed linguistically, the schemas are called labels. This distinction is important since labels make the understanding and meaning of
things more simple. Good examples of this important difference are simple objects like a house or a car. No one would call a car a big object with an engine, a steer, wheels and doors. They would all call it by its name, namely, a car.
One aspect the authors all agree on, is that in order for this schema, social construct, social agreement, or collective agreement to work, a certain level of agreement among the members is required. Hannan Pólos and Carroll (2007) describe this as follows: “A category can only emerge when an audience reaches agreement about what a label means, and a category persists so long as the level of such intentional consensus remains high”.
The producers of the goods, who are called the category members, influence the emergence and persistence of this aforementioned consensus. Diversity within the same category, in other words, small violations of the category code, made by the producers, threatens the durability of this consensus and thus the emergence and persistence of the category (Negro et al., 2010).
To further explain these violations of the category code, Zuckerman’s (1999) theory about this process consisting of two stages can be used. The author describes this process of categorization and the corresponding audience agreement in two stages. The first stage consists of the categories defined by the consumers, in order for them to easily compare the different products. They put different producers and products with similar characteristics into one category and the products and producers who do not match these characteristics are being excluded. As a result, the producers try to give their products the characteristics the consumer envisioned, so they can fit within a category, and will not be excluded. The second stage involves the differentiation of the products by the producers, they try to add value and become better than the competitors, without diverging too much from the category, because otherwise there would be no audience agreement and they could be seen as illegitimate. As this theory explains, small violations of the category code happen almost naturally.
Another aspect that influences the existence and persistence are the particular boundaries the category has. Those boundaries are lists of attributes that goods must posses to fall within that category. Some category boundaries can be more stable or clear than those of other
categories (Lamont and Molnár, 2002). For example, for most audience members the animation genre is clear, they know what falls within the boundaries of the category animation, however, when it comes to drama, the boundaries are less clear. Moreover, another possible cause for a product in which the consensus of to what category it belongs to is missing, is the situation in which a product spans multiple categories (Hannan, 2010). Theory about category spanning will be extensively reviewed on page 11.
Nonetheless, Vergne and Wry (2013) conclude that, even though categories result in issues concerning the boundaries, for both the consumer and the producer, the importance of categories is stronger and thus prevails. Simply stated, categories will keep playing a big role when it comes to creating coherence by splitting items into constructs.
This study is focusing on the Motion Picture industry, which is a creative industry, an important distinction, because in this industry, categories manifest themselves as genres. Categorization in the creative industries used four different dimensions, namely, differentiation, hierarchy, boundary strength and universality. To elaborate on this, genres are different from each other, genres are part of a hierarchy, since some genres are considered more exclusive than others, genres have different boundary strengths, as mentioned before, the animation genre is more clear than the drama genre, and lastly, genre systems are universally important across the entire field in which they are used (DiMaggio, 1997). Some studies have focused on the specific effects of genres as an independent value. According to Desai & Basuroy (2005) the genre of a specific movie can have an effect on the different signalling attributes of that movie. The research of Gazley, Clark, and Sinha (2011) suggests that the genre of a movie can have a direct effect on the purchase intent of consumers, and thus, that some genres are more in favour than others.
Category Spanning
An important aspect of categorization is category spanning, hybridity or straddling. Simply stated, category spanning is category membership in multiple categories. According to Vergne and Wry (2014) the definition is as follows: “An organization engages in category straddling when it has simultaneous membership in two or more categories located at the same level of
the classification hierarchy” (Vergne and Wry, 2014, pp. 71). To further elaborate on this definition, an American thriller is not an example of category spanning, whereas a action-thriller is, this is because the categories used need to be at the same level of the classification hierarchy. Country and genre are both salient categories within the motion picture industry.
As mentioned before, the effect of category spanning has received much attention in several studies. Up to now, there is no agreement among the researchers what the exact effect of category spanning is on performance. The nature of categories is very dynamic and therefore the effects as well, which makes it even harder to come to an overall agreement.
The important different perspectives on the effects of category spanning explained by the different studies will be discussed to get a clearer overview. Negro, Hannan and Rao (2010) state that spanning multiple categories leads to lowered appeal, they conducted a research among wine producers that used multiple sort of grapes, instead of those using grapes of a certain category. The producers using different sorts, the ones who category spanned, experienced a lower rating. This lower rating can be explained by the fact that the expectations the consumer had towards the particular category were not met. Hsu, Koçak and Hannan (2009) describe the consequences, like social and economic disadvantages, of category spanning by linking two different sides and their perspectives. The side of the producers when using category spanning is characterized by their reduced ability to effectively target the audiences of each category, and therefore results in decreasing appeal. Whereas the audience-side refers to the idea that the audience uses categories to make sense of products, therefore, when products use features from multiple different categories, the product becomes harder for the audience to make sense of and thus becomes less appealing. Kovács and Johnson (2014) state that category spanners face financial disadvantages and receive less audience attention. The negative effects, such as decreased product appeal, of category spanning can be summarized in three different points, according to the study of Kovács and Hannan (2010). The first problem is related to the learning and skills, because when improving and acquiring high expertise in one category, acquiring the same level of skills in another category becomes almost impossible. The second problem states that audience members prefer specialists over generalists. Even when the specialist and generalist are equal in terms of skills and expertise,
the generalist will still be the less favourable, since convincing the targeted audiences is more difficult for a generalist. Lastly, the audience finds it difficult to make sense of products that can be assigned to multiple categories. Leung and Sharkey (2013) also describe the perceived lower quality of the products of category spanners in the eyes of the audience, because category spanners do not narrowly focus their efforts. Zuckerman (1999) also explains the negative effect by the missing capability of the audience to make sense of the product, due to the fact that those category spanning products are difficult to classify, additionally, the author states that organizations which span multiple categories will suffer reduced audience attention. This reduced audience attention is explained by Zhao, Ishihara and Lounsbury (2013) as ‘illegitimacy discount’.
The opposite, namely positive effects of category spanning, are also mentioned in several studies. According to Pontikes (2012) category spanners enable themselves to operate in different categories and therefore attract larger audience. Jensen (2010) states this positive effect as well, calling it expanding market niches, and therefore gaining more and broader audience attention. Hsu (2006) describes the situation in which movies can successfully target different audiences by using category spanning, and because these movies target larger audiences, their performance increases as well. Zhao, Ishihara and Lounsbury (2013) supports this idea of Hsu (2006), and states that under certain circumstances category spanning can lead to positive outcomes in terms of audience attention.
Motion Picture Industry
The global total box office of the Motion Picture Industry in 2016 was around 38.3 billion dollars and it will most likely grow to 40 billion dollars within the next few years (Statista, 2017). An interesting aspect of this industry is that it is offering experience products, which means that consumers have to experience the creative good, in this case, watch the movie, before they know if they like the product (Gemser, Leenders and Wijnberg, 2008; Frey and Steiner, 2010). With experience products the quality is difficult to determine before consumption, that is why information about the movie’s quality, such as rating, awards and box office success are important information sources for the consumers (Anand and Watson, 2004). Genres convey
important information about the underlying features of the different offerings (DiMaggio, 1997). Sets of preferences held by the film audience’s members correspond to the genres, so films that are classified under the same genre share common elements that form the basis of the expectations of the audience they have for that particular genre. Examples of such elements are the structure of the events, the narrative style, the tone, the nature etc. (Dancyger and Rush, 2002). Some clear examples are the horror film that always has something that is haunted or someone who is possessed, or the science-fiction film, which is set in the future and most of the time deals with the invasions of ‘non-humans’, or the gangster film playing the contradictions resulting from wanting social and financial success (Hsu, 2006).
Genres play a enormous role in the production process. The envisioned genre, and the corresponding targeted audience, shape and provide the unifying elements of the production process, such as the way the movie and script is written and developed, the way it will be pitched to investors and what kind of actors and producers need to become a part of the movie. But also what kind of marketing is suitable for that specific genre and how much of the promotional budget should be devote to the movie (Hsu, 2006). Genres administrate clear frameworks for selecting movie projects, advising resource allocation decisions, and guiding and organizaing project’s development and facilitate coordination and communication among the project crew (Altman, 1999) (Schatz, 1981). According to Austin (1989), the most important reason for a audience member to go a specific movie is the genre. The audience’s perception and expectations are based on the genre. Furthermore, the audiences for different genres show variation in terms of demographics. For example, romance films are seen as a more female genre, while adventure or action films are considered male genres (Hsu, 2006). According to Fischoff, Antonio, and Lewis (1998) there are indeed significant differences in film audiences by gender, age and racial and ethnic groups.
2.2 Hypotheses
As mentioned previously, audience members use established categories to make sense of goods. In the case of movies, the audience sets their expectations based on the genres. For example, the expectations of the audience when visiting a horror movie, such as, scary and
fear, are different from the expectations when visiting a comedy, like laughter and joy. When movies use categories spanning, thus, incorporating features from multiple genres, setting and meeting the expectations becomes more difficult, and this results in poor fit and less appealing movies. However, not all genre combinations automatically result in poor fit and less appeal. Based on the literature on category spanning, I argue that when the audience is used to certain genre combinations, in other words when the frequency of that combinations used is higher, the expectations are easier drawn and met, and therefore, do not automatically result lower appeal and performance.
Hypothesis 1a: There is a positive relation between the frequency of a genre combination
and the performance of a movie.
Hypothesis 1b: There is a positive relation between the frequency of a genre combination
and the rating of a movie.
As mentioned before, category spanning in general can result in a movie that fits less within the genre boundaries set by the targeted audience members. Therefore, it becomes more difficult for the audience to categorical make sense of, and thus, more difficult to appreciate. This is because, based on the given category, expectations are set by the audience, and when there is no fit with the given category, the expectations will not be met. Hsu (2006) describes this as follows: “Paying attention to a broader, more diverse set of audiences therefore means less attention paid to establishing and communicating a clear fit to each” (Hsu, 2006; p. 424). However, in this case, the effect of category spanning in general is mentioned, whereas in some cases, when targeting the same audience a genre combination could actually work and result in a opposite effect, making it easier for the audience to appreciate. For example, a specific target group could have the genres thriller and war as favourite ones, combining those two genres when applying category spanning would in this case not mean having to pay attention to a broader more diverse set of audiences, because it is within the same audience group, and therefore not result in less appeal. A distinction between gender and age categories is made to split the audience in target groups.
Hypothesis 2a: There is a positive relationship between the audience similarity of a genre
combination and the performance of a movie.
Hypothesis 2b: There is a positive relationship between the audience similarity of a genre
combination and the rating of a movie.
Referring to the classification literature, a certain level of agreement about the category a good belongs to is required. If there is no agreement between the consumer and the producer on where the movie belongs to, it will influence the performance and rating. In other words, when the genre consensus is high, the level of agreement is high, and therefore the audience is more capable of understanding the category spanning construction, and consequently appreciate it more, because the chances of meeting the audience’s expectations are higher.
Hypothesis 3a: There is a positive relation between the genre consensus and the
performance of a movie.
Hypothesis 3b: There is a positive relation between the genre consensus and the rating
of a movie.
Another important factor, that can not be ignored, is the distinction between mainstream and art-house films and the possible effect on category spanning. Both type of films tends to be very different from each other. The mainstream films are produced in such way, e.g. a narrow range of familiar plots, expressions, and characters, that they are not challenging or difficult to understand for the audience, in other words, specifically designed to be easy to enjoy and appeal to a wide audience (Baumann, 2002). Those films are aiming for great commercial success (Holbrook, 1999).
On the contrary, arthouse films are considered as low-budget films that are made based on the principle of art for art’s sake. The idea of the director (artistic merit) is the key productive factor (Bagella and Becchetti, 1999; Baumann, 2002). These films are meant for a specific audience, instead of the mass audience (Bordwell and Thompson, 2001). Other differences include, content, narrative and structure. Authorial-expressivity and realism are in most cases the key drivers for the narrative (Bordwell, 1979). DiMaggio (1997) states the difference in preference
between high-art and mainstream art is attributed to the differences in social class. Consuming high art is a status symbol and can be used as a mean to construct social relationships.
In short, mainstream films are easier to understand and target a larger audience. The audience of mainstream movies are used to movies with multiple genres and the movie is especially designed for this larger target audience. Therefore, it is expected that the type of movie will moderate the effect of the independent variables, frequency, audience similarity and genre consensus, because, simply put, in the case of mainstream movies the audience cares less, thus, the effects will be weaker for mainstream movies.
Hypothesis 4a: The effects hypothesized in H1a, H2a and H3a will be weaker for mainstream
movies than for art house movies.
Hypothesis 4b: The effects hypothesized in H1b, H2b and H3b will be weaker for
3. METHODOLOGY
In this section the methodology will be explained in detail combined with a description of the data and variables. As mentioned before, the goal of this thesis is to research correlational relationships associated with genres and explain their dynamics in the Dutch movie industry, to do this, the formulated hypotheses will be tested quantitatively.
3.1 Operationalization Research
Libby’s Boxes
In figure 3.1, you can find the operationalization of the research in a explanatory model. In this section every variable and construct will be further explained in detail.
Genre Combinations Movie’s performance Movie’s rating Conceptual
Frequency Box-office revenue
IMDB rating Operational Control variables:
-
Budget-
Star power-
Director power-
Number of screens-
sequel or not Audience Similarity Arthouse vs. Mainstream Genre Consensus Number of visitors Figure 3.1Empirical Setting
The empirical setting of this research is the Dutch Motion Picture Industry. In section 3.2 this empirical setting will be explained further.
Research Goal
This research has two main goals. The first goals is to determine the effect of category spanning, more specifically, the use of specific genre combinations, on a movie’s performance and rating. Secondly, making the first steps in identifying under which circumstances these possible effects are moderated.
3.2 Dutch Motion Picture Industry
Movie database
The data forming the movie database that is used to analyse and test the hypothesis is coming from the Dutch organization, Film Distributeurs Nederland (FDN). This organization was willing to support my research and shared all their separate data, concerning the movie releases in the Netherlands, such as, data about a movie’s distributor, genre(s) (chosen by the producer), actors, directors, number of screens, box office revenue and number of visitors. From the International Movie Database (IMDB), additional information was collected, such as the rating, the genre(s), and the estimated budget. All this separate data was put together in one large
database
.
Sample
The sample used is coming from the Dutch Motion Picture industry. From 2011 until 2015 roughly 1825 were released in the Dutch cinemas. From this population of movies, a sample of movies was selected in a manner that included those movies that use category spanning, or in other words, have multiple genres. In total 518, of the 1825 movies, matched the criteria by having multiple genres. The reason for choosing this specific timespan is because of several reasons, firstly, previous research has proven that a time span of only a few years is adequate (Eliashberg and Shugan, 1997; Hsu, 2006; Zuckerman and Kim, 2003), secondly, it consists of the most recent data for which all the financial data is available, and lastly, this time span and
data is exhaustive, all the data is coming from FDN, the a organization that regulates all the movie releases in the Netherlands.
Genres
In this research the 18 most common genres are used, namely: action, adventure, animation, children, comedy, crime, documentary, drama, family, fantasy, horror, musical, mystery, romance, science fiction, thriller, war, and western (FDN, n.d.). In theory, those 18 genres could make up to 153 different genre combinations (note: combinations of two genres), however, in practice, only 72 different combinations were used and produced from 2011-2015. This is not even 50% of all the options possible. Another remarkable finding is that more than 50% of the total amount of movies has one of the eight most occurrent genre combinations as a its genre. Some movies were classified under more than 2 genres, in table 3.2 you can find the frequencies.
3.3 Dependent variables
Two two main outcomes of interest in this research are performance, consisting of visitors, and appeal, consisting of the IMDB-rating.
Visitors: Performance is measured by using the visitors variable. This variable refers to the total amount of visitors in the Netherlands throughout the entire theatrical run of the movie, the data is provided by the FDN. It ranges from 81 visitors to 2.000.184 visitors.
IMDB rating: The overall appeal of a movie to its audiences is measured by the IMDB rating. The IMDB rating is based on the consumer who votes and rates movies on their website. These ratings are on a 1-10 scale.
Table 3.2: Genres
Amount of Genres Frequencies Percentage
1 genre (*not included in research) 1307 71,62
2 genres 432 23,67
3 genres 84 4,61
4 genres 1 0,05
5 genres 1 0,05
3.4 Independent variables
The following three independent variables are used in this study: frequency, audience similarity, and genre consensus.
Frequency: This variable refers to the frequency of use of a genre combination, consisting of two genres. It is formed by using the percentage of these frequencies. As mentioned before, the sample consists of movies with either 2, 3, 4, or 5 genres. The frequency variable only takes combinations of 2 into account. To compute these frequencies, all the movies with more than 2 genres are divided in separate combinations. Figure 3.3 shows an example of the movie, “OZ: The Great And Powerful”, to better explain this parting and calculation.
After the parting of the movies, the database consisted of a total of 704 movies, all having a genre combination with only two genres. Based on this total amount, the percentage of frequencies of the combinations were calculated. Once when the frequencies percentages were calculated the best proximities scores were added to the movies, and the other “double” ones were deleted. Resulting in the original database of 518 movies, including the calculated frequencies.
This approach has a disadvantage, namely the fact that the percentage value is calculated including the combinations that are later deleted. In other words, the weighting of frequency is
OZ: The Great And Powerful
ac, ad, fy ac, fy
OZ: The Great And Powerful
OZ: The Great And Powerful
7,8%
ad, fy
OZ: The Great And Powerful ac, ad
ac, ad, fy OZ: The Great And
Powerful 10 x 9 x 55 x 7,8% 1,4% 1,3% Figure 3.3
Genre combination Frequency of this genre combinations used in total Percentage of this frequency highest percentage frequency is used
based on combinations that are not included afterwards. However, not deleting those “double movies” would otherwise lead to performance results and ratings that are not really caused by that particular genre combination. Another thing taken into consideration is that when you would choose to not delete the “double movies”, you risk getting dependence in the data, while a regression analysis requires independent data.
Audience similarity: This variable refers to the level of similarity in audience. As explained in chapter 2, genres differ in terms of audience, however, some genres target broadly the same audience. To form combinations that target the same audience the research about the movie audience and the national market by Stichting Filmonderzoek is used. In this study the genre preferences per gender and age category are researched. In chart 3.4 the preferences are shown (Stichting filmonderzoek, 2013).
Combining two highly favourited genres of the same gender-category or age-category will lead to combinations that are similar in terms of audience, thus audience similarity is expected in such cases. The data of the chart is coming from a survey that has been conducted in collaboration with full service market research firm Ruigrok Netpanel and is carried out on behalf of EYE Film Institute Nederland, the Dutch Association of Cinema Operators (NVB) and
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% Male Female 16-22 23-30 31-40 41-54 55+ Chart 3.4
the Dutch Association of Film Distributors (NVF). The research focuses on the current customer, and it is based on a sample of 2101 respondents.
Based on the data and chart coming from this study, the value 1 is given to movies that (partly) consist of the following genre combinations:
These combinations score the highest on preference, therefore audience similarity is expected. The value 0 is given to all the other movies, that do not (partly) consist of one of these combinations, since their genre combinations do not accord with the preferences, and therefore, there is no audience similarity expected.
Genre consensus: This variable measures the agreement between the targeted genres chosen by the production side and the actual genres chosen by the consumers. The classification difference was assessed by comparing those of the FDN, given by the production side, and those of IMDB, chosen by the consumers. The majority of the movies experienced, in various degrees, dissensus in genre choice, in some cases, there was even no consensus at all. The genre consensus is a value between 0 and 1 and is calculated by dividing the amount of corresponding genres by the total amount of listed genres. Resulting in a score of 1 of a full genre consensus and a score of 0 in a full dissensus.
Table 3.5: Audience similarity
Best Match Male Female 16-22 23-30 31-40 41-54 55+
Option 1 co, war dr, ro co, ro co, ro co, fa co, fa co, th
Option 2 co, th co, ro co, fa co, dr co, th co, dr co, war
The calculation of the movie “OZ: The Great And Powerful” and “This Means War” can be found in the concrete example shown in figure 3.6.
3.5 Control variables
Number of Screens: This variable represents the number of screens a movie is shown on during the opening week. This data is coming from FDN and based on the amount of copies that have been distributed to cinemas in the Netherlands. This variable is used to control its effect on the dependent variable (Neelamegham and Chintagunta, 1999).
Budget: Budget is expected to have an effect on the dependent variables (Basuroy, Desai, & Talukdar, 2006). The data concerning the (estimated) budgets is coming from the website IMDB, in some cases, however, this information was not available. In these cases, the budget information is coming from other sources such as the production company, annual reports, wikipedia or numbers.com. There is chosen deliberately not to categories those budgets to make it comparable to the Dutch movie budgets, which are considerable lower, because categorizing those budgets would only lead to very rough and eventually inadequate amounts, and therefore does not outweigh the advantages.
Sequel: This variable is used to distinguish whether a movie is a sequel or not (0 or 1), and to control its effect on the dependent variables.
3.6.a. OZ: The Great and Powerful 3.6.b. This Means War
Production side Action Adventure Fantasy Production side Action Comedy IMDB Action Comedy Romance IMDB Action Family Fantasy
Genre consensus score: 2 / 3 = 0,6667 Genre consensus score: 2 / 3 = 0,6667
Star Power and Director Power: Based on several studies it is assumed that a specific movie actor/actress or director can draw a larger audience, this, so called, power, of the star and director seems to have a positive impact on performance (Hennig-Thurau, Marchand and Hiller, 2012; Sochay, 1994; Litman and Kohl, 1989; Desai and Basuroy, 2005). The information that is used to determine the star and director power is coming from the website The Numbers, Nash Information Services, LLC (www.thenumbers.com). This website is a premier provider of movie industry data and research service. The list top 100 worldwide grossing actors and the list top 50 grossing worldwide directors is used. If an actor or actress is listed in this top 100, the star power value will be 1, if a director is listed in this top 50, the director power value will be 1, in all other cases the value of star power, respectively director power, will be 0.
3.6 Moderating variable
Major: As discussed in chapter 2, the literature review, it is expected that the type of movie, either mainstream or arthouse, has a moderating effect. To construct this variable every movie is divided into either the value 1, meaning mainstream, or value 0, meaning arthouse. This division is based on two principles that are corresponding with the study of Gemser, Van Oostrum and Leenders (2006). First, by looking at the type of distributor, this can either be a major or an independent one. Secondly, combining the information of the previous step with the information about the amount of screens it was distributed on. When a movie was distributed by a major distribution company or distributed on more than 25 screens, it was considered a mainstream movie. The information about the distribution companies, including their market share, as well as the amount of screens was coming from FDN.
4. RESULTS
In the previous section the methodology and the research design were explained. In this section you can find the results obtained from SPSS by testing and measuring the data, however, only the important tables can be found in this section, all other outcomes can be found in the appendices. The descriptive statistics of the variables combined with a correlation matrix are shown and the hypotheses are tested using hierarchical multiple regression analyses. The variables visitors, budget and number of screens and transformed to the natural log (ln). The reason for this transformation is to improve interpretability.
4.1 Descriptive Statistics
Below, in table 1, are the most important descriptive statistics for the variables presented in a correlation matrix. Of the roughly total of 1825 movies displayed from 2011-2015 in the Netherlands, 518 movies matched the criteria by having multiple genres. Concerning those 518 movies, the best performing movie is Skyfall with a total number of visitors of 2.000.184. The movie with the highest rating of 8,6 is Interstellar. The lowest rated movie is De overgave, by scoring a 1,6.
Table 4.1 : Descriptive Statistics and Pearson Correlations
Variable Mean S.D Visitors IMDB Rating Frequency Audience Similarity consensus MajorGenre Number of Screens Budget Sequel Star Power Director Power
Visitors 10.628 1.776 1 IMDB Rating 6.414 .923 .130** 1 Frequency 5.410 3.497 -.014 .149** 1 Audience Similarity .344 .475 -.087* .013 .547** 1 Genre Consensus .644 .268 .015 .127** .227** .273** 1 Major .674 .469 .624** .001 -2.33** -.225** -.021 1 Number of Screens 3.620 1.110 .895** .016 -.093* -.154** -.032 .616** 1 Budget 16.677 1.717 .573** .121** -.041 -.280** -.065 .507** .581** 1 Sequel .151 .358 .250** -.018 -.118** -.180** -.074 .189** .254** .223** 1 Star Power .228 .420 .263** .223** -.012 -.160** .054 .221** .205** .385** .054 1 Director Power .095 .293 .314** .167** -.040 -.206** -.022 .225** .280** .319** .214** .218** 1
** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed)
Pearson Correlations
There are significant bivariate relationships between some of the research variables (frequency, audience similarity, genre consensus and major) and one or more dependent variables. To demonstrate, there is a significant positive correlation between frequency and the IMDB rating (r = .149, p = .001), a significant positive correlation between genre consensus and the IMDB rating (r =.127, p = .001), and a significant positive correlation between major and amount of visitors (r = .624, p = .001). Of course, conclusions can only be drawn regarding these effects when multivariate regression analyses have been conducted.
Furthermore, the control variables, all the dependent variables, and one or more research variables, show significant relationships. Based on this, it can be concluded that they may be responsible for (a part of) the effects of the research variables, and therefore included in the multiple multivariate regression analyses. Finally, it is noted that the independent variables, both the control and research variables, do not show correlations greater than .7, therefore there seems to be no risk of multicollinearity regarding the regression analyzes that will be conducted.
4.2 Hierarchical Multiple Regression Analyses
The hypotheses are tested using hierarchical multiple regression analyses with the number of visitors and the IMDB-rating as dependent variables. In the first step the control variables, number of screens, budget, sequel, star power and director power, were used. In the second step the research variables, frequency, audience similarity and genre consensus were added and in the third step the interaction variables, frequencyXmajor, audiencesimilarityXmajor and genreconsensusXmajor, were added. To avoid multicollinearity, the quantitative variables that were involved with the interaction variables were meancenterred.
The following assumptions are made regarding the regression model. Based on both the standardized residues and the normal pp plot can be concluded, that, for both regression analyses, the distribution of the residues not significantly differs from the normal distribution. Secondly, from the scatter plot of the standardized residues and the standardized predicted values can be concluded, again in both cases, that both the assumption of linearity and homoscedasticity are met. Subsequently, the Durbin Watson scores show that the residues are independent. Lastly, it is apparent from the VIF score that there is no multicollinearity. In the appendices you can find these analyses and plots.
Performance
The hypotheses regarding the performance of a movie, namely 1a, 2a, 3a and 4a, are tested using hierarchical multiple regression. Below in table 4.2 you can find the important outcomes of the regression.
Hypothesis 1a: There is a positive relation between frequency of a genre combination and
the performance of a movie
Table 4.2: Performance: “Visitors”
Variables Step I Step II Step III
B (std. error) B (std. error) B (std. error)
Controls Number of screens 1.338 ** 1.268 (.044) ** 1.277 (.044) ** Budget .044 (.027) .034 (.028) .022 (.028) Sequel .045 (.100) .105 (.098) .114 (.097) Star Power .260 (.089) ** .254 (.087) ** .277 (.087) ** Director Power .287 (.124) * .324 (.121) ** .303 (.120) * Independent Frequency .024 (.012) .041 (.023) Audience similarity .248 (.092) ** .410 (.189) * Genre consensus .081 (.135) .185 (.279) Major .399 (.100) ** .523 (.132) ** Interaction frequencyXmajor -.038 (.027) audiencesimilarityXmajor -.264 (.215) genreconsensusXmajor -.109 (.319) R2 .804 .818 .822 R2 change .804 .014 .004 F change 377.270** 8.954** 3.471*
** significant at the 0.01 level (2-tailed) * significant at the 0.05 level (2-tailed)
As regards to this hypothesis it can be concluded that there is no significant effect of frequency on performance, β = .041, p = .073. Therefore, Hypothesis 1a can not be accepted.
Hypothesis 2a: There is a positive relationship between the audience similarity of a genre
combination and the performance of a movie.
Regarding this hypothesis it can be concluded that there is a significant positive effect of audience similarity, in other words, a genre combination which consists of (partly) the most favourite genres of one or both genders and/or one of more age categories, on performance, β = .410, p = .030. Therefore, hypothesis 2a is accepted.
Hypothesis 3a: There is a positive relation between the genre consensus and the
performance of a movie.
Regarding this hypothesis it can be concluded that there is no significant positive relation between the genre consensus score and the performance of a movie, β = .185, p = .508. Therefore, hypothesis 3a is not accepted.
Hypothesis 4a: The effects hypothesized in H1a, H2a and H3a will be weaker for mainstream
movies than for art house movies.
It turns out that all interaction variables are not a significant addition to the model (β = -.038, p = .159, β = -.264, p = .215, β = -.109, p = .319). Therefore, hypothesis 4a is not accepted.
Rating
The hypotheses regarding the rating of a movie, namely 1b, 2b, 3b and 4b, are tested using hierarchical multiple regression. On the next page, in table 4.2, you can find the important outcomes of the regression.
Hypothesis 1b: There is a positive relation between the frequency of a genre combination
and the rating of a movie.
As regards to this hypothesis it can be concluded that there is no significant effect of frequency on rating, β = .000, p = .989. Therefore, hypothesis 1b is not accepted.
Hypothesis 2b: There is a positive relationship between the audience similarity of a genre
combination and the rating of a movie.
Regarding this hypothesis it can be concluded that there is no significant positive effect of audience similarity, β = .274, p = .231. Therefore, hypothesis 2b is not accepted.
Hypothesis 3b: There is a positive relation between the genre consensus and the rating
of a movie.
Table 4.3: Rating
Variables Step I Step II Step III
B (std. error) B. (std. error) B (std. error)
Controls Number of screens -.089 (.461) -.0664 (.053) .001 (.001) Budget .037 (.032) .038 (.034) 8.953E-010 (.00) Sequel -.114 (.119) -.073 (.119( -.117 (.121) Star Power .431 (.106) ** .400 (.106) ** .411 (.104) ** Director Power .424 (.147) ** .411 (.146) ** .310 (.150) * Independent Frequency .038 (.015) * .000 (.028) Audience similarity -.098 (.112) .274 (.229) Genre consensus .371 (.164) * 1.206 (.338) ** Major -.060 (.421) .145 (.160) Interaction frequencyXmajor Xmajor .041 (.033) audiencesimilarityXmajor -.465 (.261) genreconsensusXmajor -1.050 (.386) ** R2 .078 .109 .131 R2 change .078 .031 .022 F change 7.781 ** 3.913 ** 3.850 *
** significant at the 0.01 level (2-tailed) * significant at the 0.05 level (2-tailed)
Regarding this hypothesis it can be concluded that there is a significant positive relation between the genre consensus score and the rating of a movie, β = 1.206, p = .000. Therefore, hypothesis 3a is accepted.
Hypothesis 4b: The effects hypothesized in H1b, H2b and H3b will be weaker for
mainstream movies than for art house movies.
The positive effect of the genre consensus is moderated by the type of movie, β = -1.050, p = .007, this means that the effect of genre consensus on the rating for mainstream movies significant weaker is than for arthouse movies. This is not the case for the other research variable, therefore, hypothesis 4b is only accepted partially.
4.3 Summary
The summarized results can be found in table 4.4.
Table 4.4 : Summarized Results
Hypothesis Relationship Result Comments
1 a Positive relation between frequency and performance not accepted
1 b Positive relation between frequency and rating not accepted
2 a Audience preference similarity leads to a higher performance accepted
2 b Audience preference similariry leads to higher rating not accepted 3 a Positive relation between genre consensus and performance not accepted
3 b Positive relation between genre consensus and rating accepted
4 a
The effects of the independent variables on performance mentioned in H1a, H2a, and H3a, are moderated by the type of movie, either arthouse or mainstream.
not accepted
4 b
The effects of the independent variables on
rating mentioned in H1b, H2b, and
H3b, are moderated by the type of movie, either arthouse or mainstream.
partially accepted
Only significant moderating effect on genre consensus
5. DISCUSSION
5.1 Discussion and Implications
Several hypotheses were tested in an attempt to making the first steps in identifying the dimensions between genre combinations used in category spanning and its effect on performance and rating. Performance was measured by using the visitors variable. The variable rating was operationalized by using the IMDB-rating. Below every hypotheses and its concluded results will be discussed.
Hypothesis 1A and 1B (rejected)
-
There is a positive relation between the frequency of a genre combination and the performance of a movie-
There is a positive relation between the frequency of a genre combination and the ratingof a movie.
Both hypothesis 1A and 1B can not be accepted since the regression analyses did not show any significance. Relating these hypothesis to the literature it could be explained by the fact that even though combinations are used more often, expectations could still be difficult to met, and thus poor fit and less appeal, which means that frequency was not the best independent variable to use. As Negro, Hannan and Rao (2010) state the lower lower rating can be explained by the fact that the expectations the consumer had towards the particular category were not met. Hsu, Koçak and Hannan (2009) describe the consequences of category spanning, by referring to the idea that the audience uses categories to make sense of products, therefore, when products use features from multiple different categories, the product becomes harder for the audience to make sense of and thus becomes less appealing.
Another explanation for the rejecting of hypotheses 1a and 1b could be that this thesis only looked at genre combinations of two. Which means that even though consumers are used to certain combinations, combining this with another genre (e.g. 3 genres in total) could lead to a
complete other combination in the eyes of the consumer, and therefore the audience can still not easily make sense of the category spanning example.
Hypothesis 2A (accepted) and 2B (rejected)
-
There is a positive relationship between the audience similarity of a genre combinationand the performance of a movie.
-
There is a positive relationship between the audience similarity of a genre combinationand the rating of a movie.
Hypothesis 2A is accepted, which means that whenever there is audience similarity within a genre combination, or in other words, a genre combinations consists of (partly) the most favourite genres of one or both genders and/or one or more age categories, leads to higher performance of the movie. As discussed in the theory this can be explained by the fact that, as for example, Pontikes (2012) states, category spanners enable themselves to operate in different categories and therefore attract larger audience. Surprisingly, when hypothesis 2B was tested no significant relations were found. The operationalization of audience similarity can be improved in further research to test the relationship again.
Hypothesis 3A (rejected) and 3B (accepted)
-
There is a positive relation between the genre consensus and the performance of a movie.-
There is a positive relation between the genre consensus and the rating of a movie. The analysed results of hypothesis 3B shows a significant relation and is therefore accepted. Hypothesis 3A shows a significance of p = .508, it is debatable whether this hypothesis is marginally significant and therefore could possibly accepted as well. In sum, this means that from the producer-side the category must carefully be selected, and correctly ‘transferred’ among the audience members. Because, as proven, when the producer side fails to do so, it suffers rating and possibly performance as well.Hypothesis 4A and 4B (partially accepted)
-
The effects hypothesized in H1a, H2a and H3a will be weaker for mainstream movies thanfor art house movies.
-
The effects hypothesized in H1b, H2b and H3b will be weaker for mainstream movies thanfor art house movies.
It was expected that, because of the differences in audience between mainstream and arthouse movies (Baumann, 2002; Bagella and Becchetti, 1999; Bordwell and Thompson, 2001), the type of movie the effect on the independent variables, frequency, audience similarity and genre consensus, moderates. Simply put, because it was expected that the mainstream audience would ‘care’ less. However, the testing of the hypothesis only showed that the effect of genre consensus on the rating for mainstream movies significant weaker is than for arthouse movies. In short, this means that genre consensus does matter less for mainstream movies.
Not all hypotheses could be accepted, possible reasons for biases and (wrongfully) rejected hypotheses will be explained in the limitations section.
Implications
There is no complete proof to accept all the hypotheses and therefore state the different dimensions between genres and its effects on performance and rating. However, some accepted hypothesis show that there is indeed a difference, and that those dimensions can not be ignored. Therefore, this research gives new perspectives in the different possibilities and effects of certain genre combinations. It will contribute to the general discussion of the different theories about categorization and category spanning in the creative industries, in particular when using multiple genre combinations for a creative good, and possibly open new roads for future research. Secondly, the research has several (managerial) contributions to offer for film production and distribution companies. Knowing the effect of combinations of genres and their interaction is very useful information for producers and directors, because in all the steps of the process, such as the script, narrative style, marketing, of producing and distributing a movie,
the type of genre plays an important role. It can positively influence the strategic process, overcome future mistakes, and optimize performance.
5.2 Limitations and Future Research
Similar to other studies, this research has some limitations, that could possibly have caused biases, however, it goes without saying, that those limitations were as much reduced as possible.
The first limitation is the dependent variable rating, only the IMDB rating was used to measure this variable. This was because the amount of websites that rate all international movies is scarce, however, combining multiple ratings and calculating an average would result in a more adequate number. Future research can try to expand the measures of this variable by, for example, including domestic rating websites. Also the other dependent variable, visitors, can be expanded for future research. For example, distinctions could be made between different periods of times, such as the amount of visitors during opening weekend, cumulative, and individual weeks, and comparing those differences.
Secondly, the control variable, niche volume of individual genre, is missing in this research. Not all genres attract an evenly large audience, and therefore the effect of a specific genre has on rating and performance should be controlled.
Moreover, the setting of this research is the Dutch Motion Picture Industry. Because of this, it is not certain that the same results and mechanics apply in other countries, for example, in one of the largest markets concerning the Motion Picture Industry, the United States. The audience of different countries, and their related preferences could be completely different. Future research could replicate this thesis either in different countries, or across multiple countries.
Additionally, this study has included a wide range of different foreign movies, the only criteria was for the movie to be released between 2011 and 2015. It is quite possible that this wide range of movies results in a bias, because Hollywood blockbusters are completely different
still have the same genres. Future research can look at a more specific empirical setting, for instance, only movies from the Netherlands released in the Netherlands, important to note, that in order to generate a large enough sample size, a larger period is needed.
As an independent variable in this thesis I have used frequency, as mentioned before, the frequency is calculated including the combinations that are later deleted. In other words, the weighting of frequency is based on combinations that are not included afterwards. However, not deleting those “double movies” would otherwise lead to performance results and ratings that are not really caused by that particular genre combination. Another thing taken into consideration is that when you would choose to not delete the “double movies”, you risk getting dependence in the data, while a regression analysis requires independent data. Future research can improve this variable’s operationalization.
Another limitation regarding the operationalization of the independent variable is the one of audience similarity. This variable is constructed by using one research with a sample size of approximately 2100 respondents, to improve the generalizability of this research, it would be better to use multiple researches that focuses on audience preferences, or conduct your own research with a special focus on this aspect. Future research could for example use the insights of ticket sale demographics, or conduct surveys, to realise an reliable audience similarity variable.
6. CONCLUSION
Since previous studies failed to look at the specific dimensions between genres, and just looked at the concept in general, this thesis was an first attempt to get more insight in these dimensions and interactions. The aim of this research was to find out under which circumstances some genre combinations used in category spanning could be a better match than other combinations and if this results in a better performance and rating of the movie. The empirical setting of this research is the Dutch Movie industry from 2011 until 2015. All the relevant data concerning the movie releases in the Netherlands was coming from FDN (Filmdistributeurs Nederland) and the IMDB (International Movie Database). In total 514 movies matched the criteria and were included in this research. Of the total 8 hypotheses, 3 were accepted based on correlation and hierarchical multiple regression analyses conducted in SPSS, namely (1) Audience similarity leads to a higher performance, (2) There is a positive relation between genre consensus and rating, and (3) The effect of genre consensus on rating is negatively moderated by the type of movie.
Although, only half of the hypotheses could be accepted, the results still indicate that there are indeed certain dimensions and interactions between certain genre combinations. Which means that not all genre combinations result in the same effects on performance and rating. Therefore, this research did proof that even though a lot of research is already done on this specific subject, the end is not near, there are still many effects, circumstances and interactions that require more research.
In sum, this thesis expanded the growing body of existing literature focusing on the relationship between membership in multiple categories and its effect on performance and rating. This thesis has shown that the dimensions between categories, in this case genres, can not be ignored, and this strengthens the believe that the dimensions and relationships are more nuances than one would believe. To be more specific, not all genre combinations have the same positive effect, concepts such as, audience similarity and genre consensus play an important role. However, only half of the hypotheses were accepted, therefore I would strongly
