To Infinity and Beyond
- Signals and Sequels in the Animation Industry -
Adrienn Juhász
(11420634)
MSc Business Administration – EMCI
University of Amsterdam
Supervisor: dr. Frederik Situmeang
Submission: 21.06.2017
Abstract
Film studios produce more and more sequels, so it is more important for them than ever to understand what their audiences really want to see. At the same time, we also need to start paying more attention to media’s identity shaping power on children. This study investigates what individual attributes of films contribute to good film evaluations and that how this affects sequel success, while considering different audiences and their interactions. Questionnaires were filled out by 96 respondents and analyzed using statistical methods. The findings contribute to scientific literature by proving the importance of characters used in films, by showing that children and their parents do have different preferences, but also that their taste spillovers influence decision making within households. It is concluded that the lower the taste spillover, the less children’s evaluation is able to influence their parents’ likelihood of watching a sequel. Whilst some guidelines for studios making animations are found, the study does encounter a number of potential limitations.
Statement of originality
This document is written by Adrienn Juhász 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.
Adrienn Juhász Amsterdam, 21 June 2017
Table of Contents
Introduction ... 4
Theoretical Framework and Hypotheses ... 7
Cultural Industries ... 7
Signaling Theory and Brand Extensions ... 8
Children, Parents, and Household Decision Making ... 11
Methodology ... 14
Research Setting ... 14
Data Collection ... 16
Sampling ... 16
Questionnaires ... 17
Potential biases, limitations ... 18
Variables ... 19 Control variables ... 20 Results ... 21 Descriptive Statistics ... 21 Test of Hypotheses ... 22 Model 1 ... 22 Model 2 ... 25 Discussion ... 28 Conclusion ... 31
Limitations and Future Research ... 32
References ... 33
Appendix A ... 38
Introduction
During the last decade there has been an unusual turn in the Hollywood film industry: instead of bringing forth novel ideas, one cannot help but notice the increasing number of prequels, sequels, and remakes produced that are trying to tap into our nostalgic feelings. In the years 2013 and 2014, seven out of the top ten grossing films were either sequels or prequels, and while Hollywood only created 14 sequels in 2008, in 2016 this number reached 35. (Stephen Follows, 2015) Though the number of sequels truly is increasing, their level of success varies: it must be a process of learning by doing if we consider the example of Toy Story, the first feature length animated film by Pixar using computer-generated imagery (CGI). It came out in 1995 and remains one of the most successful animated motion pictures of all time. While Toy Story 2 was a moderate success in 1999, Toy Story 3 in 2010 doubled its predecessor’s opening weekend box office results, became the first animated sequel to be nominated for an Academy Award for Best Picture, and it also grew into the second highest grossing animated film of all time. (IMDb)
Sequels can be seen as a type of brand extension – films keeping the same title and just changing the subtitle or adding the volume number to the end to exploit the established name of the first edition of their story. A number of researchers have studied what makes brand extensions successful and studied the effects of evaluations (e.g. Aaker & Keller, 1990; Bottomley & Holden, 2001; Situmeang, Leenders & Wijnberg, 2014), while some applied this to the film industry and studied movie sequels (Sood & Drèze, 2006; Basuroy and Chatterjee, 2008). Other scholars used sequels merely as a dummy variable in their studies and recognized that sequels on average generate higher revenue flows (e.g. Hennig-Thurau, Houston, & Walsh, 2006). However, none of these studies narrowed down their focus to individual attributes of the films. Other studies did look at what film attributes moviegoers specifically like (e.g. Gazley, Clark & Sinha, 2011), but not necessarily in terms of brand extensions. Since each genre targets a different audience, success factors can be highly varying - more specifically, in order to better understand what could make sequels truly successful, genres should be studied one by one, as some of the film attributes they are using can be different or can have a different level of importance. Interpreting this in terms if signaling theory: instead of asking how brand extensions
can be successful, the question could also be that what features need stronger signaling to make films and their sequels more outstanding and/or how to reduce uncertainty and information asymmetry audiences need to deal with (Spence, 2002 in Connelly, Certo, Ireland & Reutzel, 2011).
Furthermore, when looking at audiences we also need to ask ourselves whether this means one single group of consumers or multiple. More precisely, previous research discussing sequel success focuses on the decision of a single buyer (e.g. Sood & Drèze, 2006; Basuroy and Chatterjee, 2008), but if we consider that films as experience goods that can be consumed by, for instance, a family as a whole, that leads to the assumption of multiple audiences – since different age groups make up various consumer groups. Therefore, this thesis also wants to discuss a scenario in which a multitude of audience groups are present when exploring sequel success.
Consequently, all the above-mentioned aspects can come together in one single question:
How do film attributes positively perceived by different audiences affect their evaluation of the film and how this, then, influences their likelihood of watching (the
next edition of) the same film’s sequel?
Films are interesting to study in this case for their experience good properties as well. Unlike everyday household objects where utility is the most important determinant of purchase, films are watched solely for the experience and enjoyment they can give. Besides, whether individuals would truly like them or not cannot be predicted until they have the actual experience. Therefore, consumption of films aims mainly at pursuing fun, feelings, and fantasies. (Holbrook & Hirschmann, 1982). Finding out what exactly makes moviegoers reach the entertainment and magic they are seeking requires the closer study of signals in motion pictures.
Supporting these notions, the thesis will test which film attributes are relevant for different age groups (here children and their parents) and how these contribute to how much they enjoy the film. These findings will relate to the likelihood of watching (the next edition of) a sequel of the film. It will be proposed that the more an audience enjoys a movie, the more likely they are to watch the sequel. Children’s influential power is also explored. More precisely, it will be tested whether children’s evaluation
affects their parents’ likelihood of watching (the next edition of) a sequel, while also accounting for taste spillovers, thus considering a multitude of decision makers. The theoretical frameworks in which these questions will be investigated are mainly drawn upon signaling theory, brand extension theory, and household decision behavior theory.
The findings are expected to contribute to scientific literature and management practices: 1) seeing what types of movie attributes different audiences prefer can help filmmakers make more successful sequels and can increase the efficiency of their marketing campaigns, 2) precisely defining what children exactly pay attention to when watching films can also help studios in defining better these characteristics and focus on the effects of these in terms of how they shape children’s identity, 3) regarding the assumption on enjoyment and likelihood of watching a sequel, the findings can be generalized and applied to other genres too, and 4) the study draws attention to the importance of investigating multiple decision makers when examining purchase decision, which would apply not only to the motion picture industry but any other consumer good as well.
Since enjoyment and focusing on a specific genre are of key importance here, feature films of this study will be animated feature-length films. Besides, if we consider films to be experience goods consumed for their high fun-factor, animations seem to be clearest choice – they certainly carry more magic and fantasy than any other genre on the market. The reason for choosing the animation industry besides this is twofold. First of all, previous studies on animation mainly focus on industry, production, and labor characteristics in different geographic locations (e.g. Tschang & Goldtsein, 2004, 2010; Yoon & Malecki, 2009; Cole, 2008; Fuerst, 2010) while less on specific film attributes (e.g. Davis, 2007). Secondly, the main target audience of animated films is children – thus, choosing this genre gives opportunity for seeing differences between the preferences of different age groups. The importance of studying how the minds of children develop is undeniable and in today’s world, especially in developed and developing countries, the amount of media influencing their development is immense, greater than ever. “It is (…) of extraordinary
importance to bring the two topics of, on one hand, media globalization, on the other, children, youth and media closer to each other” (von Felitizten & Carlsson, 2002, p.
14). As animated films are part of this worldwide entertainment targeted towards children, their study is crucial. An excellent example of showing how important
animations can be is described by Ron Suskind in his 2014 bestseller Life, Animated:
A Story of Sidekicks, Heroes, and Autism. The author tells the story about how
fictional Disney characters helped his autistic son understand the world – but this is just one of many examples of cartoon’s identity shaping power.
The thesis firstly lays out the theoretical background and hypothesis, followed by describing methodology. The results section describes statistical findings, while a discussion outlines the implications of these. Concluding remarks, limitations, and possibilities for future research close the paper.
Theoretical Framework and Hypotheses
Cultural Industries
In order to ultimately uncover what factors could make sequels in the animation industry successful, we first have to take a step back, look at the broader picture and ask: why are cultural industries characterized by such high levels of uncertainty?
In the following, cultural industries will be referred to as defined by Peltoniemi (2015, p. 1): “Cultural industries are those that produce experience goods
with considerable creative elements and aim these at the consumer market via mass distribution”. She states that the creative elements are to serve the goal, among
others, of identity building. This definition suits the purposes of this study for numerous reasons. First of all, it recognizes films as cultural experience goods. Secondly, as the research considers feature-length animated films with high box office results, pointing at mass distribution as means of reaching consumers is also fitting. Lastly, the identity building aspect can have an enhanced role, e.g. greater than general, in case the target audience is children.
Core characteristics of the cultural industries are persistent oversupply of output and constant extreme uncertainty. The first explains that there are always more creative goods on the market than can be consumed, which makes finding success in this industry a challenging endeavor (Hesmondalgh, 2002; Thorsby, 2001). The phenomenon of “nobody knows” describes the uncertainty lingering around all cultural goods (Caves, 2003). The aforementioned experience goods are especially
haunted by this type of uncertainty, thus predicting whether relevant gatekeepers in either end of the value chain will like them is not definable until they are out on the market. For example, a writer can ask readers about his story proposition, but he cannot be certain whether his book will actually be successful until people read that book. Thus, predicting the reaction of the audience of a cultural good is truly difficult – yet not completely impossible.
There are certain indicators that can help estimating future success of a cultural good; these are taste of consumers and popularity of goods. (Peltoniemi, 2015). Some researchers went against the basic assumption of extreme uncertainty, saying that taste is not pre-existent but is cumulative, thus builds up over time in an incremental manner (Blaug, 2001; Cowen, 1989; McCain & Towse, 2003; Thorsby, 2001). Relating to this thesis, their statement would imply that by understanding preferences of children and parents, thus their taste, better, predicting future success of films is less difficult than one would generally anticipate. On the other hand, popularity of goods can also be a reflective indicator. For instance, a fine balance between the novelty of a good and consumers’ familiarity with it can be critical: in the film industry, creating a story that is both unique in style but also is appealing for many audiences is generally sought by filmmakers (Alvarez, Mazza, Pedersen, Svejenova, 2005). Moviegoers may, for instance, look for characters they already know but in a setting that they have not encountered before – by the virtue of this, sometimes “sequels (…) outperform other films” (Peltoniemi, 2015, p. 11). The key takeaway is the following: there is constant and extreme uncertainty when it comes to cultural goods, but by understanding taste and popularity, e.g. preferences, the obstacle could be better handled.
This applied to films means that audiences can have different taste in what individual film attributes they are looking for, but understanding these could decrease the uncertainty of not knowing what consumers want exactly.
Signaling Theory and Brand Extensions
Information asymmetry fundamentally means that different people have different information at hand – in other words, what they know is dissimilar. One tool to bridge this asymmetry is through sharing signals between two parties that are to give information about intent and quality (Stiglitz, 2002). Applied to the creative
industries as discussed above, this would mean using signals to decrease the uncertainty that characterizes this sector.
This study will greatly draw upon the signaling timeline worked out by Connelly et. al. (2011) (see Figure 1). According to this timeline, a signaler (person, product, or firm) has some underlying quality. This signaler sends the signal to the
receiver, who makes meaning out of the received signaler. The timeline ends with the
receiver sending feedback to the signaler.
Applied to this study, signalers are products, more precisely animated feature films. The signals they send are the different film attributes that were discussed above. Audiences of films, so here children and parents, are the receivers of the signals. They evaluate the film after seeing it and the feedback they “send” is whether they watch the sequel or not – not wanting to see the next edition is like a negative feedback, while the opposite is a positive feedback.
It is proposed that studios are able to use different signals to decrease the information asymmetry between themselves and their audience. Since films have experience good properties, the level of how much audiences enjoy them will greatly depend on the perceived signals from these films. Furthermore, receivers do have the power to select among signalers (Connelly et. al., 2011), which here can be seen as the competitive arena in which studios and their films fight for the attention of audiences by sending different signals. Whilst the features preferred by various audiences may differ, as it is expected to happen between children and parents, their importance to reach the desired level of enjoyment is consistent across genres and is not explicit for animations. In this manner it generally could be proposed that the more preferred signals receivers get from the signaler, the more positive their
feedback will be. Therefore, it will further on be assumed that the more positively one evaluates the individual attributes of a film, the more positive the film’s overall evaluation will be.
According to Hem, Chernatony & Iversen (2003), brand extension strategy is becoming an ever more widespread approach for reducing risk, since drawing upon an already-recognized brand and consumers’ good association with it can lead to exploitation of existing resources. In the film industry this would mean using a successful movie title and producing one or more editions of the story, in other words make a sequel out of it. As mentioned in the introduction, the tendency to do so has been increasing. A type of positive association, especially in case of films with experience good properties, is how much one enjoyed the first edition of a sequel. Völckner & Sattler (2006) also found that experiences consumers have with previous editions of a sequel could have strong implications for the level of success of the newer edition is about to yield. Relating to the previous hypothesis this implies a chain-like effect: the more features of a film an audience likes, the more positive their evaluation of the film will be. Situmeang et. al. (2014) also found that past evaluations are good indicators of sequels future success. If we assume that a film’s evaluation depends on the signals in this film and that future success is dependent on how many moviegoers the new edition of the sequel produces, then the following relationships should be tested:
H1A: There is a positive relationship between the evaluation of individual attributes of a film and the overall evaluation of that film.
H1B: There is a positive relationship between the evaluation of a film and the likelihood of watching (the next edition of) the sequel of this film.
H1C: The relationship between the evaluation of individual attributes of a film and the likelihood of watching (the next edition of) the sequel of this film is mediated by the overall evaluation of the film.
Accordingly, the aim of studios should be to find out what characteristics of their films audiences prefer and emphasize these when coming out with a sequel. Sood & Drèze (2006) also show that evaluation of sequels gets better as the overlap
between editions of the same sequel is higher. Other scholars also reflect on the importance of fit between parent brand and extension (Dacin & Smith, 1994; Hem et. al., 2003; Henning-Thurau et. al., 2009). To find out what exactly to bring back in the newer editions, however, one needs to study various audiences – the next section will focus exactly on this.
Children, Parents, and Household Decision Making
The mysterious minds of children are something that we are all eager to understand – the curiosity and imagination in the little heads have no limits, kids may believe as many as six impossible things before breakfast. There are many ways to distinguish groups of different ages and generations but one of the most common ways to do so is looking at individuals’ cognitive development. According to psychological literature (Piaget 1929, 1954 in Zhang & Sood, 2002), children go through four main stages: 1) sensorimotor (0-2 year old), 2) preoperational (2-7 year old) where surface stimuli are perceived, 3) concrete operational (7-11 year old) where deeper stimuli are also somewhat perceived, and 4) formal operational (12 year old and older) where complex, both surface and deeper stimuli are processed, thus the cognitive development reaches adult-like thinking. Seeing the different stages is crucial, because it also implies that children as consumers react to completely different signals at these different stages. John (1999) redefines the latter three stages of cognitive development in terms of consumer behavior, resulting in 1) perceptual, 2) analytical, and 3) reflective. Fundamentals of consumer behavior are learnt before reaching the final stage around the age of twelve, by which time children start perceiving marketing stimuli like adults (Valkenburg & Cantor, 2002 in Calert, 2008). Based on this and their own research, Zhang & Sood (2002) found that children and adults vary in their perception of cues – the younger group relies more on “surface” cues when it comes to brand extensions, such as brand name, while the older group puts more emphasis on “deeper” cues as well. By the virtue of this phenomenon it is stated that extensions targeted at these two groups really are different, since they need to be.
This research chooses to focus on children between the age of eight and eleven, thus the ones at their concrete operational stage of cognitive development and analytical stage of consumer development but ones who can also already be surveyed
(Borgers & Hox, 2000). At this age children do perceive signals differently from adults, thus differently from their parents as well. Along these lines, applied to films, it can be expected that children focus on different attributes from their parents, having a different base for evaluating how much they enjoyed a film. Regarding the features relevant for this study this could imply that, for instance, children pay attention to title and characters, but parents may also be interested in who directed the animated feature film and what the country of origin (COO) is. Overall, in other words:
H2: Children’s film attribute preferences are different from parents’ film attribute preferences.
Henceforth, if we believe this to be true, it can similarly be argued that the enjoyment of the film will depend on different factors for these two audience groups, which is confirmed if Hypothesis 1 turns out to be true.
Furthermore it can be theorized that interactions between evaluators can have an effect on the likelihood of watching a sequel. The roots of this assumption are in household decision behavior (HDB) theory, which focuses on three main topics: 1) who the decision-maker in the family is, 2) what the outcomes of these decision in the household are, and 3) what factors determine who becomes the decision maker (Qualls, 1987). This can be interesting to study in the case of animated feature films, since even though the main target audience and users are children, the final decision makers are still the parents. Darley & Lim (1986) found that children have a different degree of influence on purchase decision depending on the type of good – for instance, regarding movie attendance, children could influence when to go to see the film. The study does not, however, describe whether children influence the decision on purchase, so whether to go at all. Influence from here on is defined as “both the
effort and the ability to affect or to sway a decision” (Lee & Beatty, 2002, p. 26)
Other studies, however, do clearly describe children’s and adolescents’ power to influence purchase decisions within the family (e.g. Foxman, Tansuhaj & Ekstrom, 1989; Lee & Beatty, 2002) and their effect on household economics and demand patterns (Browning, 1992). Consequently, regarding the decision whether to watch a sequel could depend not only on the level of enjoyment from the parents’ side. More precisely, there could be a scenario in which parents did not enjoy a film they watched with they children at all – nonetheless, the children may have liked the same
film so much that they are able to convince their parents about watching the sequel. This is also due to the fact that children’s power to influence the purchase choices of their parents has been increasing over time (Calvert, 2008). The scenario could also happen vice versa, thus the parents wanting to see the sequel but get discouraged by their offspring’s negative evaluation. Put differently:
H3: The relationship between parents’ evaluation and their likelihood of watching the sequel is moderated by their children’s evaluation of the film. The relationship is stronger for more positive evaluations by children.
If true, this could especially be important in terms of marketing. If children were able to convince their parents to watch a sequel that originally they did not want to would mean that marketing efforts and signaling should rather focus on children – after all, then, they have the final say. Besides, since the group of children studied in this thesis concerns the age group of 8-11, the confirmation of Hypothesis 3 would also contradict comparative resource theory (CRT). CRT states that one’s power to influence a decision of the household depends on the level of resources provided by the same person (e.g. Sanchez & Thomson, 1997). Clearly children in their first years of school do not provide resources that allow for the purchase itself, thus, according to theory, should not have the power to influence the final decision. However, here it is proposed that they do when it comes to watching (the next edition of) a sequel.
Despite their potential influencing power, however, one also needs to account for the level of taste spillovers between the decision-making groups. More precisely, it can be assumed that similar preferences for films could make the “negations” about whether to watch a sequel easier. However, the more discrepancies there are in preferences of children and their parents, the harder it could get to change evaluations of either party. Consequently, the following hypothesis should also be tested:
H4: The difference in children’s and parents’ attribute preferences moderates the moderating effect of children’s evaluation of the film on the relationship between parents’ evaluation and their likelihood of watching the sequel. The relationship is weaker for greater values of difference.
The proposed hypotheses are summarized in Figure 2. The following section will discuss the research methodology of this study, including research setting, variables, and details regarding data collection.
Methodology
Research Setting
“…if we seek ‘the magic dust of creativity’ then we need to look no further than the cultural industries” (Jeffcut & Pratt, 2002, p. 226) – and if we want to find
the most magical of them all, we arrive at the animation industry.
Animation is “The process or technique of filming successive drawings or
positions of puppets or models to create an illusion of movement when the film is shown as a sequence; a film produced in this way (…) typically involving a computer moving images produced in this way” (Oxford English Dictionary, 2010). Cultural
industries, accordingly involving the animation industry too, have been described as a chart business, meaning that their success depend on being able to be the best in terms of value for a limited period of time with a certain type of output (Jeffcut & Pratt, 2002). This description is especially true for blockbuster movies of Hollywood whose lifecycle is theaters is rather limited, as a “typical run is three or four month,
[which] leads to (…) the trade-off between the pressure to open early” (Krider &
Winberg, 1998, p. 2) and the pressure to wait and avoid competition.
This segment of the film industry using different techniques to bring fictional and non-fictional stories to life is unique in terms of 1) evolution of technology, 2) styles/visual techniques, 3) type of characters, and 4) feasibility of product globalization. Advancements reinforce faster output generation and contribute to the growth of the industry (Leyshon, 2001; Power & Jansson, 2004). Consequently, mass production occurs where studios need to focus not only on creating blockbusters but also on potentially expanding their parent brand to different media platforms (Yoon & Maleski, 2009).
Technology also played a role in the evolution of different worlds of production in the industry, nevertheless described by two styles of animated feature films used in both eras: specialized (artisan-style) and standardized (globalized, mainly CGI) (Yoon & Maleski, 2009). Sequels occur mainly in the latter style, since the artistic drive and the wish to be original goes against creating sequels that re-use pre-existing resources – however, this does not mean that there potentially would be no demand for new editions in the more artistic style as well.
Other distinctive characteristics of animated films can be seen comparing them to live action films. Firstly, unlike live action, animated films oftentimes use non-human characters either solely or in combination with human characters in their storytelling. Besides being more fun in some ways, this also helps the genre
“overcome racial and ethnic differences among exporting and importing cultures”
(Havens, 2007) and make globalization of the product smoother. Another aspect of easier foreign distribution of these films is the fact that their dubbing is more uncomplicated than that of live actions.
Based on the above-described characteristics that are specific for animated feature films and including ones that could be reflective as deeper signals in terms of what audiences look for, the following film features will be tested as to whether preferred by different audiences: style (artisan and CGI), characters (human and non-human), cast & crew, studio producing the film, COO, and the title (parent brand) of the film. These characteristics will be considered as signals films can use.
Summed up, the setting of this study to test the above-described hypotheses is the animated motion picture industry. While the broader motion picture industry has been tested for audiences’ preferences of films and scholars have focused on brand
extensions and sequels in this industry, there is a lack of studies specifically focusing on animations. Since this thesis also includes aspects of HDB and investigates differences between age groups, choosing the animation industry gives opportunity for studying all distinctive fields of research – signaling theory, brand extensions, HDB – at once.
Output will be generated and interpreted via the software IBM SPSS Statistics Version 24.
Data Collection
The study is a quantitative research using primary data that is to be collected by questionnaires (see Appendix A). These will be sent to a primary school in Budapest, Hungary, namely Szemere Bertalan Primary and Secondary School. Data will be collected by sending out questionnaires both to children and their parents in paper-and-pencil (PP) form. The school itself has been chosen on a convenience basis. It is the primary school where I spent the first eight years of my own education and where I still have contact with some of my old teachers. Anonymity of respondents will be kept.
Sampling
The sample for this study consists of two main samples: children and parents. The sample of children expectedly consists of approximately n1=190 students
between the age of 8 and 11. Since children are expected to provide the survey to one of their parents, the sample of parents (n2) is of the same size.
Darley (1986) published a study named ‘Family decision making in leisure-time activities: an exploratory investigation of the impact of locus of control, child age influence and parental type on perceived child influence’. The sample for this study included 106 parents. Since somewhat similar in nature, this thesis will also aim for this number of respondents per group, allowing for a highly safe › 50% non-response rate.
In this study non-probability sampling is used. The sample, on one hand, is a convenience sample, since the exact research site and the children and parents to whom the questionnaires will be sent out to were selected on a convenience basis. This availability sampling raises the likelihood of biases and does not guarantee
generalizability of the results – however, due to the timeframe restraining the study, this was still found to be the best solution.
Questionnaires
The Questionnaire for Parents consists of 30 questions and can be found in Appendix A. It has been developed based on multiple previous studies – the variables and the related questions are described in further detail below. The questions have been adapted to this study, some with greater extent of change than others. When sent out, the questionnaire also includes a cover sheet explaining the study, the anonymity protecting respondents, as well as a raffle respondents can participate in (see below). Parents also have the option to ask for the results of the study by providing their email addresses.
The Questionnaire for Children consists of 30 questions and can also be found in Appendix A. The questionnaire firstly has been developed in the exact same manner as the one for parents, however, due to the young age of the targeted group of respondents, it had to be further adapted. As Borgers, Leeuw & Hox (2000, p. 63-64.) found, when it comes to children between ages of 8-11, “questionnaires have to be
specifically developed (…), one cannot use the standard questionnaires used for adults”. One needs to use a very detailed introduction while also paying special
attention to wording every question – young children interpret sentences very literally. Due to the possibility of lacking motivation and concentration, the questionnaire includes an extra assignment at the end, including drawing their favorite fictional character – only in order to make the study more fun for them. By the same virtue, the design of the questionnaire is more playful with colors as to draw children’s full attention.
Both questionnaires are PP questionnaires. The reason for using this method instead of a computer-based system is the following: since children will be asked to inform their parents about the questionnaires, giving them the opportunity to hand over physical questionnaires is expected to increase the response rate over having to go to a viral link.
Respondents will also be informed that their participation will give them the chance to win in a raffle. The raffle here is used as a lottery incentive, defined as “a
every recipient who responds is entered in a drawing (similar to a lottery) for one or more prizes” (Porter & Whitcomb, 2003, p. 390). Since this study is set in the motion
picture industry, the prize to win is tickets to the movies. More specifically, the following is stated: if a child participates, he or she will have the chance to win a single cinema gift ticket. However, if the child’s parent is also willing to participate, they have the chance to win three tickets. Consequently, after collecting responses, two raffles will be held: one for children whose parents decided not to participate, and one for the group where both parent and child answered the questionnaires. Porter & Whitcomb (2003) found that lottery incentives are a common method with generally effective results of increasing response rate, thus it is hoped to help this study as well.
The original questionnaires found in Appendix A will be directly translated to Hungarian. Back translation or parallel translation will not be applied.
Potential biases, limitations
The sample of this study is expected to encounter numerous biases. First of all, since the study involves convenience sampling, results are less generalizable (Marshall, 1996).
Secondly, generalizability is also affected by conducting the research in Budapest, Hungary. In order to get the clearest picture, the study would need to involve respondents from different parts of the world, especially since the focus is on globally distributed feature length animations. However, the scope of the research does not allow for this.
Thirdly, the time frame constraining the research does not allow for a more credible research. For instance, using a focus group that would watch a film and then would directly be able to answer questions about this film could prove to be more efficient.
Lastly, using children as a group of respondents also has its risks. Borgers & Hox (2000) write that their responses are less reliable. The age group investigated in this study (8-11) mostly just recently acquired the skill of reading while their language skills are also in a developing phase. Henceforth, these characteristics can also affect the results of the study.
Variables
Evaluation of Film. Sood & Drèze (2006) studied movie sequels as experiential goods
and their evaluation. In their study, participants were asked to evaluate a movie coming out in the near future based a plot. Thus, a similar approach is applied in this study for one of the independent variables, the Evaluation of Films. As mentioned above, the questionnaire firstly provides a plot of an animation. After reading it, the questionnaire asks respondents to evaluate this film on six scales, each having a 7-point range where higher numbers stand for more favorable evaluations. The six scales are the following: bad movie/good movie, forget it/must see, uninteresting/interesting, wait for rental/see opening night, will be a flop/will be a hit, sounds worse than most films/sounds better than most films. These six scales are combined into an overall evaluation index (Cronbach’s α = 0.95).
This variable will result in two independent variables, one being Evaluation of Film overall in Model 1 and the other Evaluation of Film by Children only in Model 2.
Likelihood of Watching (Next Edition of) Sequel. The dependent variable of this study
is in Model 1 is The Likelihood of Watching (Next Edition of) Sequel. In Model 2 for Hypotheses 3 and 4 it changes solely to Parents’ Likelihood. Since the likelihood of watching the film is correlated with the decision whether actually to do so, the measure of Purchase Decision developed by Berger & Fitzsimons (2008) is used. Their two questions regarding purchase decision have been adapted to the current study, both are answerable on a 7-point Likert-scale. The authors found these two questions to be highly correlated (r = 0.81), thus both questions are kept for the purposes of this study as well and they are averaged to create a purchase likelihood index (Cronbach’s α = 0.7). After describing the plot of the sequel of the firstly described film (thus this being the second plot), respondents are asked the following questions: “How interested are you in watching this sequel?” and “How likely would you be to watch this sequel?”
Preferred Film Attributes. To see whether children and parents have different
preferences when it comes to movie attributes, a number of characteristics will be tested. Respondents will be asked to rate eight attributes on a scale from 1 (not at all)
to 7 (very much) regarding how much they care about each when choosing or watching an animated film. Based on the study by Yoon & Malecki (2009), who discuss specific characteristics of animated films, the first four attributes tested are: 1) artesian animation, 2) CGI animation, 3) human characters, and 4) non-human characters. Numerous scholars studied film characteristics and their effects on box office results as well as evaluation by movie goers (Ravid, 1999; Chang & Ki, 2005; Henning-Thurau et.al., 2009; Gazley et.al., 2010; Asad, Ahmed & Rahman, 2012). Based on these, the following attributes have also been chosen to be tested: 5) cast & crew, 6) studio, 7) COO. Lastly, a feature that is important from both movie evaluation studies perspective as well as in the field of brand extensions, 8) title is also included.
The plot of the film that is to be evaluated includes only some of these attributes. It will be checked whether the evaluation of only the attributes present leads to a better overall evaluation – this way Hypothesis 1A can be tested.
A simple t-test will help to test Hypothesis 2 regarding possibly differing preferences of children and parents.
Difference in Preferred Film Attributes. To test Hypothesis 4, a variable standing for
the Differences in Preferred Film Attributes is created. By summing the absolute difference (∆) between the importance of each attribute in the minds of parents and their children this variable can be created.
Control variables
Perceived Fit. Ahluwalia (2008) studied brand extensions and measured perceived fit
of products by asking respondents to indicate on a 7-point Likert-scale how similar-dissimilar and how consistent-inconsistent they found the extension. Perceived fit is important to control for in this study, since the effects for sequel watching likelihood to be tested are evaluations based on signals rather than perceived fit – and previous researchers indicated perceived fit as the most important prerequisite for extension success (e.g. Aaker & Keller, 1990; Park, Milberg & Lawson, 1991), thus its bias needs to be accounted for.
Demographics. The study will also control for gender (male = 0; female = 1), age
group (parent = 1; child =0), and age.
Results
The following section presents the results of the statistical analysis of the study. It starts with briefly introducing descriptive statistics, which is followed by the results of testing the hypotheses. The relevant output generated by SPSS can be seen in the Tables 1-8.
Descriptive Statistics
The expected sample of the study was overall n=380 (n1=190 + n2=190), however, the
final sample consists of no more than n=98 (n1=49 parents + n2=49 parents)
respondents overall. As the school said, many of the children did not bring back the questionnaires for the given deadline, therefore many filled out surveys could not be used due to this reason.
Table 1. Descriptive Statistics
Parent Age Female
N Valid 98 98 98 Missing 0 0 0 Mean 0,5 25,68 0,74 Median 0,5 21 1 Std. Deviation 0,50 16,83 0,79 Variance 0,25 283,31 0,62 Skewness 0 0,22 4,96 Std. Error of Skewness 0,24 0,24 0,24 Minimum 0 8 0 Maximum 1 60 7
This total sample consists of 67.3% females and 31.6% males. The youngest respondents of the questionnaire are 8 years old, while the oldest is 60. The age of the children who filled out the surveys is between 8 and 11, while parents’ age varies between 31 and 60. The female dominance is interesting to see both in the parent and in the child samples and is in accordance with the finding that women are more likely to participate in surveys than men (Moore & Tarnai, 2002 in Smith, 2008).
Table 2. Correlations 1 2 3 4 5 6 7 8 9 1. Age 1 .201* -.240* -.230* -.220* 0,093 -.307** -.228* -0,089 2. Gender .201* 1 0,135 0,132 -0,038 0,099 -0,020 -0,043 -0,075 3. Likelihood -.240* 0,135 1 .759** .224* .222* 0,178 .296** 0,091 4. Evaluation -.230* 0,132 .759** 1 0,133 .257* 0,199 .270** 0,163 5. CGI -.220* -0,038 .224* 0,133 1 0,049 .241* .321** 0,058 6. Human Characters 0,093 0,099 .222* .257* 0,049 1 0,139 0,077 -0,012 7. Non-Human Characters -.307** -0,020 0,178 0,199 .241* 0,139 1 0,190 0,002 8. Studio -.228* -0,043 .296** .270** .321** 0,077 0,190 1 .518** 9. COO -0,089 -0,075 0,091 0,163 0,058 -0,012 0,002 .518** 1
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).
Test of Hypotheses Model 1
Hypothesis 1. The dependent variable of this study is Likelihood of Watching the
Sequel, which is based on two questions. To combine Q26 and Q27, firstly a reliability test is run, which gives highly reliable Cronbach’s α = 0.87. Thus, the two questions can be combined into the likelihood index previously described by Berger & Fitzsimons (2008). Furthermore, in order to test Hypothesis 1B, an evaluation index needs to be created the same way. The reliability test of Q12-16 results in a Cronbach’s α = 0.898. Even though the original study of Sood & Drèze (2006) used a Cronbach’s α = 0.95, this result is still above the generally accepted threshold of 0.7, thus is accepted. Controlling for the perceived fit between the original edition of the
B Std. Error Beta t Sig.
(Constant) 2,777 0,893 3,108 0,003 CGI 0,015 0,087 0,017 0,168 0,867 Human Characters 0,225 0,105 0,211 2,136 0,035 Non-Human Characters 0,074 0,086 0,09 0,871 0,386 Studio 0,083 0,082 0,124 1,017 0,312 COO 0,059 0,073 0,091 0,8 0,426 Gender 0,626 0,296 0,209 2,118 0,037 Age 0,017 0,031 0,2 0,542 0,589 Age Group -1,258 1,023 -0,45 -1,23 0,222
Notes: R squared = 0,217 (adjusted R squared = 0,146)
sequel and next edition of the sequel is necessary too, thus the two questions (Q28A-B) need to be combined as well for perceived fit. The reliability test gives a Cronbach’s α = 0.77, thus they can also easily be combined.
Model one takes the overall final sample of n=98 respondents (n1 = 49
children + n2 = 49 parents) to test the relationships proposed by Hypothesis 1A-C. To
do so, first the relationships between the evaluation of individual attributes of the film and the evaluation of film as a whole is tested. To test Hypothesis 1A, attributes are first chosen that are present in the film and attributes whose preference was asked for are matched. The remaining characteristics this way out of the original eight are 1) CGI, 2) human characters, 3) non-human characters, 4) COO, and 5) studio. In order to test whether there is a positive relationship between the evaluation of these attributes and the evaluation of the film, a regression analysis is performed with the five attributes regressed on film evaluation. Age, gender, and whether one is a parent or child are controlled for. The analysis shows a low R2=0.217. However, looking at the coefficients of the individual attributes we can see that evaluation of human characters in a film does have a significantly positive effect on the evaluation of the film (β = 0.225, p ‹ 0.05). Thus, Hypothesis 1A cannot be rejected.
Hypothesis 1B describes the second indirect effect of the mediating relationship. To test if there is a positive relationship between the evaluation of the film and the likelihood of watching its sequel, a regression analysis is used. Control variables are age, parent/child, gender, and perceived fit. With β = 0.971, p ‹ 0.000 and R2=0.584 the relationship is strong and significant and is in accordance with
previous studies (e.g. Völckner & Sattler, 2006; Situmeang et. al., 2014). Summed up, evaluation of a film has a positively influences the likelihood of watching (the next edition of) its sequel.
Table 4. H1B - Regression Analysis Results
B Std. Error Beta t Sig.
(Constant) -0,16 0,633 -0,253 0,801 Evaluation 0,971 0,097 0,737 10,058 0 Age Group -0,015 0,903 -0,004 -0,016 0,987 Age -0,008 0,027 -0,079 -0,304 0,762 Gender 0,236 0,266 0,064 0,887 0,377 Perceived Fit -0,021 0,075 -0,02 -0,275 0,784
Since H1A is somewhat significant and H1B is significant, we could assume that the mediating relationship described by H1C is also significant. However, while R2=0.215 for the direct effect between preferred attributes and likelihood of watching
the sequel, adding the mediating variable this value increases to R2=0.608. Interestingly, looking at the individual film attributes’ coefficients it can be seen that the mediating relationship holds for the studio attribute (β = 0.176, p = 0.079 in the direct relationship and β = 0.076, p = 0.301 in the total effect). With an almost significant result this means that there is a mediated relationship between evaluation of a studio and the likelihood of watching a film, where evaluation of the film is the mediator. Thus, Hypothesis 1C cannot be rejected, since it holds for one out of five attributes.
Table 5. H1C Direct Effect - Regression Analysis Results
B Std. Error Beta t Sig.
(Constant) 1,866 1,109 1,682 0,096 Age Group -1,089 1,26 -0,317 -0,864 0,39 Age 0,009 0,038 0,092 0,249 0,804 Gender 0,663 0,37 0,179 1,795 0,076 Perceived Fit 0,047 0,112 0,046 0,425 0,672 CGI 0,093 0,108 0,09 0,863 0,39 Human Characters 0,23 0,139 0,176 1,659 0,101 Non-Human Characters 0,027 0,102 0,027 0,261 0,795 Studio 0,176 0,099 0,214 1,775 0,079 COO -0,017 0,09 -0,021 -0,183 0,855
Notes: R squared = 0,215 (Adjusted R squared = 0,134)
Table 6. H1C Total Effect - Regression Analysis Results
B Std. Error Beta t Sig.
(Constant) -0,73 0,842 -0,868 0,388 Evaluation 0,945 0,101 0,717 9,333 0 Age Group 0,302 0,929 0,087 0,325 0,746 Age -0,016 0,028 -0,157 -0,58 0,564 Gender 0,226 0,267 0,061 0,844 0,401 Perceived Fit -0,06 0,08 -0,058 -0,747 0,457 CGI 0,111 0,077 0,107 1,429 0,157 Human Characters 0,084 0,102 0,064 0,825 0,412 Non-Human Characters -0,044 0,076 -0,044 -0,585 0,56 Studio 0,076 0,073 0,091 1,041 0,301 COO -0,062 0,065 -0,078 -0,95 0,345
Model 2
Hypothesis 2. Hypothesis 2 investigates whether different age groups, in this specific
case parents and children, have different preferences when it comes to film attributes. In order to analyze this assumption, an independent samples t-test is performed that controls for age and gender. It is possible to use these test because the dataset meets all required assumptions: the test variables are measured on a scale (Likert-scale for measuring preferences), while the grouping variable is categorical (parent=1 or child=0); independence of observations is met; there are no significant outliers; there is normal distribution; and variances are homogeneous.
The analysis shows mixed results. The preference for certain attributes is clearly significantly distinctive, these are: CGI (p = 0.028), non-human characters (p = 0.004), studio (p = 0.027), and title (p = 0.001). However, for the other half of the attributes the hypothesis does not hold for. These are: artesian technique (p = 0.953),
F Sig. t df Sig. (2-tailed) Difference Mean Difference Std. Error Lower Upper Artesian Technique 9,019 0,003 -0,059 96 0,953 -0,02041 0,34418 -0,70359 0,66278 -0,059 86,471 0,953 -0,02041 0,34418 -0,70456 0,66374 CGI 0,542 0,464 2,226 - 96 0,028 -0,73469 0,33012 -1,38999 -0,0794 -2,226 95,604 0,028 -0,73469 0,33012 -1,39002 -0,07937 Human Characters 0,631 0,429 0,688 96 0,493 0,18367 0,267 -0,34632 0,71367 0,688 95,75 0,493 0,18367 0,267 -0,34634 0,71368 Non-Human Characters 2,024 0,158 -2,932 96 0,004 -1 0,34104 -1,67695 -0,32305 -2,932 92,139 0,004 -1 0,34104 -1,67731 -0,32269 Cast & Crew
0,281 0,597 -0,804 96 0,423 -0,32653 0,40605 -1,13254 0,47948 -0,804 95,538 0,423 -0,32653 0,40605 -1,13259 0,47953 Studio 0,862 0,355 -2,247 96 0,027 -0,93878 0,41785 -1,7682 -0,10935 -2,247 95,296 0,027 -0,93878 0,41785 -1,76828 -0,10927 COO 0,099 0,754 -0,555 96 0,58 -0,2449 0,44152 -1,1213 0,6315 -0,555 95,86 0,58 -0,2449 0,44152 -1,12132 0,63152 Title 1,196 0,277 -3,463 96 0,001 -1,16327 0,33588 -1,82999 -0,49654 -3,463 95,239 0,001 -1,16327 0,33588 -1,83006 -0,49647 Table 7. H2 – Independent Samples T-Test Results
human characters (p = 0.493), cast & crew (p = 0.423), and COO (p = 0.58). Overall, Hypothesis 2 cannot be rejected, but its argumentations do not hold significantly.
Hypothesis 3. Next, the first moderating hypothesis is tested. To analyze whether or
not children’s evaluation can affect the relationship between parents’ evaluation and their likelihood of watching the film’s sequel, a regression analysis needs to be performed. Since in this and the coming test the dependent variable is not the overall likelihood of the whole sample, only likelihood of parents is used. Similarly, evaluation of children and evaluation of parents are used as two separate independent variables. After mean centering ad creating the interaction effect between parent evaluation and child evaluation, the regression analysis is performed. Mean centering is necessary before creating the interaction effect in order to decrease multicollinearity. The analysis shows non-significant results with β = 0.154, p = 0.132. Therefore, it can be stated that children’s evaluation does not positively moderate the relationship between parents’ evaluation and their likelihood of watching the sequel as it was expected – the two-way interaction does not work.
Table 8. H3 - Regression Analysis Results
B Error Std. Beta t Sig. Tolerance VIF
(Constant) -2,781 1,166 -2,384 0,022
Evaluation 1,155 0,13 0,898 8,863 0 0,615 1,627
Evaluation Child 0,229 0,156 0,149 1,466 0,15 0,609 1,643 Evaluation Inter. 0,154 0,1 0,181 1,535 0,132 0,451 2,216
Notes: R squared = 0,735 (Adjusted R squared = 0,716)
Hypothesis 4. Hypothesis 4 tests whether the effect described by Hypothesis 3 is
lower in cases of greater differences in preferences of parents and children. An initial factor analysis is conducted next, which firstly shows that the difference item related to the studio making the film is associated with two components, thus needs to be disregarded. The next factor analysis conducted without this variable shows similar results for the variable associated with title. The final, third, factor analysis shows clear results. Interestingly it shows three main components present: the first evolving around artesian technique and CGI, the second around human and non-human characters, and the third around cast & crew and COO.
Another regression analysis is performed to test the three-way interaction described by the moderated moderation of Hypothesis 4. Results of the regression analysis show curious findings: while the positive relationship between parents’ evaluation and likelihood is significant in model 1 (Sig. F change = 0.000), this changes to Sig. F change = 0.238 in model 2 when adding the moderating effect of children’s evaluation. However, when adding the difference variable in model 3, it becomes Sig. F change = 0.035. The three-way interaction has an almost significant result of β = -0.017, p = 0.081. Although significance is not lower than 0.05, due to the small sample size used the results are accepted as significant. This implies that the relationship between parents’ evaluation and their likelihood of watching the sequel while influenced by their children’s evaluation truly is less significant when the difference in their preferences is greater. Coming back to Hypothesis 3 this interestingly means that while children’s evaluation by itself does not decrease on increase parents’ likelihood of watching a sequel, when difference in preferences comes in the picture, this changes: in the three-way interaction, the moderating effect of children’s evaluation can become positive for lesser levels of difference in preferences. However, when there are mode differences, the moderating effect becomes negative. Consequently we can say that Hypothesis 4 is strongly supported. These results prove to be especially significant since the sample sizes for conducting these tests is only n1=45 parents and n2=45 children. While this is a low number, it
implies that with a greater number of respondents the study could reach even clearer results.
Table 9. H4 – Regression Analysis Results
Model B Std.
Error Beta t Sig. Tolerance VIF
1 (Constant) -0,995 0,536 -1,858 0,07 Evaluation 1,09 0,103 0,846 10,546 0 1 1 2 (Constant) -2,781 1,166 -2,384 0,022 Evaluation 1,155 0,13 0,898 8,863 0 0,615 1,627 Evaluation Child 0,229 0,156 0,149 1,466 0,15 0,609 1,643 Evaluation Inter. 0,154 0,1 0,181 1,535 0,132 0,451 2,216 3 (Constant) -1,739 1,25 -1,391 0,172 Evaluation 1,104 0,132 0,858 8,384 0 0,535 1,871 Evaluation Child 0,196 0,148 0,128 1,326 0,192 0,604 1,655 Evaluation Inter. 0,091 0,098 0,107 0,928 0,359 0,422 2,371 Difference Sum -0,036 0,021 -0,139 -1,711 0,095 0,846 1,183 Evaluation Inter. Diff. -0,017 0,009 -0,148 -1,79 0,081 0,824 1,214
The next section will discuss the findings and potential implications of both the rejected and supported hypotheses.
Discussion
The aim of this study was to find out whether individual film attributes can affect overall film evaluation to an extent that it results in wanting to see the film’s sequel. Answering the initial research question of the study, it can be concluded that audiences evaluate films differently based on whether they belong to the ‘child audience’ or ‘grown-up audience’, since these two have differing preferences. Their likelihood to watch a film’s sequel, however, depends on the studio making the film and on how great the taste-spillover is between these audiences. These results are significant since previous research (e.g. Aaker & Keller, 1990; Bottomley & Holden, 2001; Situmeang et. al., 2014) generalized findings on relationships between evaluations and sequels or applied their findings to films in general, in a non-genre specific manner (Sood & Drèze, 2006; Basuroy and Chatterjee, 2008). The statistical findings of the study point to a number of important implications regarding film evaluations, children’s preferences, and decision making related to these – in this section we will take a closer look at all these.
First of all, it was found that there is a positive relationship between positively evaluating human characters of a film and the overall evaluation of the film. The reason why this specific attribute turned out to be the one and only significant, unlike for instance title or COO, can be explained by people’s tendency to identify themselves with characters. This is a mechanism by which audience members experience events happening to a character as if it was happening to them (Cohen, 2011). Speaking in practical terms, this would mean that when I see the character of
Simba in the Lion King losing his father, I might tear up – and seeing that scene who
would not? Although in this study the identification and importance of human characters were specific for animated films, Igartua (2010) found that the identification process has an effect on the level of enjoyment of feature-length films regardless of genre. “Although identification plays a major role in media research,
the attempts to conceptualize the nature of identification and the theoretical treatment of this concept have been less satisfactory” (Cohen, 2011, p. 246). Thus, the support
of Hypothesis 1C of this study takes a small step in expanding the currently lacking research on identification.
This identification process is even more crucial to look at when considering how media characters can influence children’s identity building. While the younger respondents of this study found human characters of films similarly important as their parents, they had a significantly greater affection for non-human characters of animations. This is not surprising, since fun characters in these films are often used to come more easily across to children, sometimes in the lead role, other times as sidekicks. They are also often used to overcome cultural or ethic differences, and can be more easily globalized (Havens, 2007). The animation industry, which is highly unique in terms of what characters it uses, needs to pay special attention to how they
build these characters. Animation studios know that characters drive actions and pay
attention to building lifelike characters. They also know that stories need to aim at developing characters “that elicit a response and connect with the audience” – as Porter & Susman (2000) described, who both belong to the creators of Toy Story and
Toy Story 2. Children tend to like when their role models are nice, understanding, and
helpful (Anderson & Cavallaro, 2002), thus animation studios such as Pixar, Disney, or DreamWorks, should aim at creating both human and non-human characters that carry these ever-lasting traits of kindness. This would have two positive effects: on one hand, children would identify with positive characters when looking for guidance and, on the other hand, their evaluation of films would stay also more positive. According to previous studies, this evaluation could mean that next edition of the film will also be a success (Situmeang et. al., 2014). Consequently, this thesis specifically suggests that sequels could be successful when keeping positively perceived characters from the first (or previous) edition of the sequel.
It was also found that the mediating effect of film evaluation between evaluation of a studio making a film and likelihood of watching a sequel holds. This is good news for studios, since their “sequels represent an increasingly important
new product introduction strategy” (Sood & Drèze, 2006, p. 352). The reason why
sequel success may depend on, among other factors, the studio itself could be related to lower uncertainty. When choosing whether to watch the (next edition of the) sequel whose parent film had a positive evaluation, audiences may have an already existing appraised picture of the studio that made the film. They can see it as legitimate, trustworthy, and be more likely to believe in the next work created by this studio. This
would be a typical behavior of consumers of cultural goods (Peltoniemi, 2015). Thus, using the studio as a marketing tool for emphasizing quality and lowering information asymmetry and associated uncertainty between the studio and its audiences is highly recommended.
The notion that children have different film attribute preferences from their parents was supported. It adds the study of Zhang & Sood (2002), who described that children tend to focus more on surface cues such as brand name. In this study, children had found the title of an animation more important than their parents, supporting the authors’ argument specifically in cases of experience goods. This finding can shed light on the challenge faced by studios that try to create films fitting both adults and children – should they keep film attributes that only the young ones care for or should they appeal to grown-ups as well? We can find the answer by seeing the implications of Hypothesis 3 and 4.
It was found that when children and their parents have less similar taste, children are less able to influence the relationship between parents’ film evaluation and their likelihood of watching the sequel. Vice versa, when they have more similar taste, children can affect more positively the parents’ likelihood of watching the sequel. Darley & Lim (1986) found that children could influence their parents’ decision making to different extents depending on the type of good. Regarding films they found that children can influence when to watch a film, but they have not found if they can influence whether to watch it at all. Therefore this thesis extends this field of research by adding that their household decision making power is greater when 1) the experience good is a film and 2) when their taste is more similar to their parents’. This finding contradicts CRT (e.g. Sanchez & Thompson, 1997) as well and says that household decision-making power does not necessarily depend on how much of the resources one contributes to the household, but on the level of taste spillover between the decision makers. However, this is not a general argument, but in case of this study holds specifically for animated feature films. Consequently, studios making animations should try to impress both children and grown-ups and not only keep attributes that appeal to children. If the family members have similar tastes, studios may be able to kill two birds with one stone.
Speaking in terms of signaling theory, (Connelly et. al., 2011) the findings mean that studios as signalers need to take advantage of all available film attributes used as signals to appeal to different receivers, i.e. different audiences. This way the
feedback can be positive, meaning audiences’ higher likelihood of watching the sequel. They need to realize this uncertainty lowering effect and see their audiences as consumers who look for an optimal balance between novelty and familiarity (Peltoniemi, 2015). Brand extensions of studios, so their sequels, can leverage characters and the studio name itself to find a solid fit between parent product and extension.
Conclusion
This study explored new aspects of film sequel success and added insights on the currently infant literature on the animation industry, while also observing some angles of the relationship between household decision-making and experience goods. It was found that human characters of a film significantly affect the evaluation of a film. The study re-confirmed previous findings on the strong positive relationship between film evaluation and likelihood of watching the film’s sequel (e.g. Sood & Drèze, 2006) and explored taste differences between children and their parents. The thesis also analyzed children’s level of influence on their parents’ decision making and concluded that it is dependent on the level of taste spillovers, and thus proved for the importance of accounting for multiple decision-making groups.
The most important finding of the study is the clarification that when watching a film, children focus on different things than grown-ups. By the virtue of the identification process and media’s power to influence children’s identity building as well as the statistically significant importance of both human and non-human characters, it is crucial for studios to build these characters in a fashion which has positive and constructive effects on children.
Overall, the success recipe for making successful sequels “to infinity and
beyond”1 can be summed up the following way: take a pinch of great characters, emphasize your studio’s presence and legitimacy when coming out with a sequel, and try to appeal both to children and parents. There might still of course be cases when children do like the animation and parents do not, but no one can blame them after all; as the Little Prince said, “Grown-ups are certainly very strange.” (de Saint-Exupéry, 1943)
Limitations and Future Research
The greatest limitation of the study is its low sample size. While in some cases the overall sample of n=98 could be used, for testing other hypotheses this was only n1=49 and n2=49. While in quantitative research finding the right sample size is
essential (Bartlett, Kotrlik & Higgins, 2001) and this study aimed for a sample of at least 106 respondents per group following Darley’s (1986) study, the non-response rate hit a very high 74%. Though the findings are still relatively significant considering the final sample size, the study should be re-conducted in the future with a sufficiently big sample of respondents.
Secondly, responses given by children between the ages of 8-11 are not completely reliable (Borgers & Hox, 2000). Due to lacking experience, even though I rephrased the survey question to a simpler language for children, a better, more child-friendly questionnaire could benefit the study as well. Children could also better answer such a questionnaire about a film if they were able to see the film before, therefore a possible experiment built around the same topic could also enrich the discussion.
Lastly, while this thesis focused solely on animations to prove differing tastes of various audiences, it could be interesting to see whether this applies to other genres too. For example, do teenagers and their grandparents like the same things about a contemporary documentary? This study implies that film evaluations may be genre-dependent, thus studies across multiple film genres could expand the scope of the current field of research on film evaluations.
