Whether and how does it influence the market performance of one movie artist’s new movie when the artist switched between movie genres
University of Amsterdam: Faculty of Economics and Business Mater Thesis Business Administration – Track Marketing Name: Manshu Zheng
Student Number: 11089385 Date: 24 June 2016 (final)
Statement of originality
This document is written by Student Manshu Zheng 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.
Abstract
This study mainly focuses on the relation between the time period one movie artist has been involved in one movie genre and the market performance of the artist’s new movie in this movie genre by analyzing 6421 movies produced from 1940 to 2015 in the United States. By conducting this study, it is proved by statistical analysis that the time period one movie artist has been involved in one movie genre has a significantly positive influence on the financial performance of the artist’s new movie in this genre and a significantly negative influence on the artistic performance of the artist’s new movie in this genre. Furthermore, the total genre similarity between the artist’s two movies can strengthen the positive relation between time period and the market performance of the artist’s new movie in this genre from the financial performance aspect.
Table of Content
1. Introduction ... 5
2. Conceptual Framework and Hypotheses ... 9
2.1 Time period one movie artist has been involved in one movie genre and the market performance of the artist's new movie in this genre .………….……….... 9
2.2 The moderating role of the time difference between the two movies ….…………... 13
2.3 The moderating role of total genre similarity between the two movies …..…….…. 14
2.4 Conceptual Framework ..……….. ………. 17
3. Data & Methods ... 17
3.1 Data & Sample ....………... 17
3.2 Dependent Variables ....……….. 18
3.3 Independent Variable ....………. 18
3.4 Moderators ....………. 18
3.5 Control Variables ....………..…. 19
3.6 Methods ...………. 20
4. Analyses & Results ... 21
4.1 Descriptive Statistics ...………. 22
4.2 Correlations ...………...… 23
4.3 Regression Analyses ...………. 25
4.4 Moderation Market Performance ...……….. 27
5. Discussions ... 34
5.1 Summary ..……….. 34
5.2 Theoretical implications ...………...………. 37
5.3 Managerial implications ...………...………. 37
5.4 Limitations and recommendations for future research ...………...……….…. 38
1. Introduction
As with all experience goods - people have to consume the product or experience to really know whether they like it or not (Caves, 2000), there is an inherent difficulty in assessing the quality of a movie prior to viewing (Reinstein and Snyder, 2005). This casts great risks on the movie industry. Much of the research propounds models that either predict or explain the box office success (Sawhney and Eliashberg, 1996; Jedidi, Krider and Weinberg., 1998; Swami Eliashberg and Weinberg, 1999; Eliashberg et al., 2000; Canterbery and Marvasti, 2001; Collins and Snell, 2002; Basuroy, Chatterjee and Ravid, 2003; Putler and Lele, 2003; De Vany and Walls, 2004; Holbrook, 2005; Reinstein and Snyder, 2005; Walls, 2005, 2009).
Though we are not able to predict the movie quality with any degree of certainty, there are nevertheless a number of hallmarks of quality that might influence audience behavior, including a movie’s tangible and intangible attributes (Neelamegham and Jain, 1999). Movies have very few tangible attributes (Eliashberg et al., 2000). One of them is the cast and the presence of big stars, which can indicate the movie quality on a number of levels (Ravid, 1999). Another attribute that may influence the market performance of a movie is the movie genre. Prior research has been conducted to examine the individual and interactive influence the two attributes have on the market performance of a movie.
Academic research on factors that may influence market performance of movies in previous years has investigated the individual effect of different factors such as movie genre, star power, and critics’ reviews. For example, Reddy, Swaminathan and Motley (1998) examined the critical role of critics and critical reviews, movie genres, and star talent in the success of Broadway shows. Similar issues have been examined in the recorded-music industry (Caves, 2000; Chung and Cox 1994), book-publishing industry (Caves, 2000; Greco, 2013), and movie industry (Austin and Gordon, 1987; De Silva, 1998; De Vany and Walls, 1999; Eliashberg and Shugan, 1997; Linton and Petrovich, 1988; Litman and Ahn, 1998; Ravid, 1999). There is also
some research focusing on the joint influences of these factors (Desai and Basuroy, 2005). However, little research has been done to investigate the influence of movie artists switching between movie genres, which can be quite dynamic and interactive.
Some of the prior research on movies has treated movie stars as high-equity brands that enjoy both name recognition and positive image, and an association with particular movie types (Levin, Levin and Heath, 1997). And research on brand equity has shown that consumers favorably evaluate extensions of brands enjoying good reputation in categories that are similar to the parent categories (Aaker and Keller, 1990; Park, Milberg and Lawson, 1991). If the longer time period a movie artist has been involved in a certain movie genre, we can assume that he/she would be perceived by both audiences and critics as strongly associated with that specific movie genre and this would have influence on the market performance of the artist’s new movie in this movie genre.
However, prior research has suggested that audiences and critics evaluation can differ substantially, both regarding contents and impact (e.g. Gemser, Leenders and Wijnberg, 2008; Moon, Bergey and Iacobucci, 2010). Regarding the theory above, for the audiences, it could help them to easily associate the movie artist with a specific movie genre and happily pay for the artist’s new movie in this genre, which could positively leads to better financial performance of the new movie; however, for the critics, who would pay more attention to the artistic value of the movie and hold different standards from audiences, they may criticize if one movie artist has been involved in one specific movie genre for too long and did not make any breakthrough in his/her career or the artist’s performance in movies of the same movie genre has been always almost the same and shows no or very little progress, which would bring negative effect on critics’ reviews of the new movie and thus negatively influence the artistic performance of the artist’s new movie.
When a movie artist switched between different movie genres, it can be seen as entering different categories. Literature to date has suggested that producers and products spanning multiple categories have inferior market performance compared to those focusing on only one or few market categories (Hannan, 2010). There exist a number of studies demonstrating that social objects that combine features from multiple categories are discounted; in other words, they experience lower market performance and are deemed as less appealing by audiences (Hannan 2010, Negro and Leung, 2013). The time difference and the genre similarity between the artist’s two movies would impose on the market performance of the artist’s new movie as well.
Here is one example. Jim Carey, the world-renowned movie star, has always been perceived as a superstar in comedy movies and almost all of his comedy movies have been successful. When he firstly showed up in a romantic movie Eternal Sunshine of the Spotless Mind in 2004, many people had difficulty associating him with the soulful and loving character in the first place but then were fully surprised by his touching and excellent performance in the romantic movie and appreciated both the movie and Jim Carey very much. As a result, the romantic movie became a huge success both financially and artistically, with a box office of 72.3 million USD and a meta score of 89 out of 100. As shown in the Filmography of Jim Carrey, he has never played in romantic movies in his previous career before Eternal Sunshine of the Spotless Mind, that is to say, the time period he has been involved in romantic movie is very short. But how come his new romantic movie became such a success? His last movie before Eternal Sunshine of the Spotless Mind was Bruce Almighty, a comedy movie released in 2003. The time difference between the two movies is nearly one year and the two movies share quite large genre similarity as Eternal Sunshine of the Spotless Mind is in the genres of romance, comedy and science fiction. Theses could partially explain the outstanding market performance of Eternal Sunshine of the Spotless Mind.
If the movie artist’s switching between different movie genres happens in a short time, the level of acceptance of the new movie may be different for both audiences and critics, that is to say, the time difference between the artist’s two movies may moderate the effect of the time period one movie artist has been involved in one movie genre on the market performance of the artist’s new movie. Moreover, the impact of genre similarity between movies should also be taken into account when examining the relation. Variability is defined as the frequency and magnitude of changes in the environment over time (Glazer and Weiss, 1993). Generally, consumers do not like uncertainty, especially regarding quality. Variability often opens the possibility that large groups of consumers weigh past evaluations lower in their buying intentions regarding the next consumption (Situmeang et al., 2014). Volatility is one type of uncertainty that has been discussed a lot in the literature. Volatility generates uncertainty because it makes the qualities of sequels harder to predict (March and Olsen, 1975). Therefore, variability may be a negative weighting factor in the relationship between past performance and future performance.
Another interesting example can be the young, beautiful and talented actress, Emma Watson. She rose to prominence after landing her first professional acting role as Hermione Granger in the Harry Potter movie series at a very young age of 10, appearing in all eight Harry Potter movies from 2001 to 2011. This role earned Emma Watson worldwide fame, critical accolades, and over 10 million USD. However, as she grew up, it became harder for her to continue acting as that adorable little girl as appeared in the Harry Potter movie series. Then here we raise the following questions: what next step should she take? Should she continue her career in the fantasy genre movie? If so, how much influence would her showing up bring to the market performance of the next fantasy movie? Or, would it be better for her to switch to another movie genre, such as romance or drama? If so, which genre would be a better option? And how long should it take before she steps into another movie genre? Waiting long enough for people to forget that cute girl Hermione or straight after finishing the Harry Potter series when everyone
can still associate her with fantasy movies? All these questions are essential for both the artist himself/herself and for the movie company when they are choosing casts for a new movie. I personally find these questions really intriguing and would be worth investigating. Thus in this paper, the research question is： Whether and how does it influence the market performance of one movie artist’s new movie when the artist switched between movie genres?
This study could add value to academic studies about category spanning, especially for artists in the movie industry and implement what has not been studied yet about how movie artists’ switching between movie genres could affect both the financial and artistic performance of the artist’s new movie. Apart from its theoretical underpinning, this study makes an empirical contribution to the literature in several ways. This further could guide actors and actresses on how they should develop when making critical career decisions, for example, on which next movie to choose and also guide movie companies when they are making careful selection of movie casts for the new movie. In this way, the findings of this research could not only benefit movie artists themselves, but also reveal managerial relevance for the movie companies.
The structure of the paper is as follows – we will start with building the conceptual framework and formulating hypotheses, then describe the samples, the data, the measures and the model, after which the results of the study will be given and the paper concludes with providing our key findings, both theoretical and practical implications and discussing the limitations of the research as well as giving some suggestions for further research.
2. Conceptual Framework and Hypotheses
2.1 Time period one movie artist has been involved in one movie genre and the market performance of the artist’s new movie in this genre
take the global box office of the movie as the indicator of its financial performance and the meta score as the indicator of its artistic performance.
The impact of a movie star on a movie’s financial success has long been a hot subject to notice both in Hollywood and the academic literature. “As Hollywood has become a business driven by opening weekends, studios are more focused on event films that captivate a national audience. There’s no better way to do that,” says one top producer, “than casting Harrison Ford in a thriller or, say Tom Cruise as a pilot.”
Featuring popular stars in a movie is likely to have consumers expect a more entertaining (or high-quality) movie, as long as the movie is of a genre with which the movie star is typically associated. (Desai and Basuroy, 2005). Consistent with this argument, Levin et al. (1997) conducted an experiment and the results showed that an unknown movie was more attractive when associated with well-known stars than with less-known stars. They argue that familiar movie stars provide audiences with a heuristic device to make decisions on seeing a new movie (without requiring much additional information).
The most widely held belief within the movie industry concerns the role of actors and actresses as key design components and they carry the responsibility to attract a large base of audience of loyal fans (Donahue 1987, p. 34, p. 191). Based on this faith in "star power", Hollywood has traditionally relied on what Vogel (2014) refers to as "bankability" or "clout" and what Powdermaker (1950) describes as a "star system" that "provides a formula easy to understand and has made the production of movies seem more like just another business" (p. 228). That "formula" has resulted in payments to movie stars such as Tom Cruise, Arnold Schwarzenegger, Eddie Murphy, and Sylvester Stallone of $9, $12, $13, and $20 million, respectively, for their appearances in Days of Thunder, The Terminator II, Beverly Hills Cop, and Rocky V (Corliss, 1991; Fabrikant, 1990; Greenwald and Natale, 1990). There already exist a number of academic researches on examining the relationship between star power and a
movie’s financial performance. Papers by Faulkner and Anderson (1987), Litman and Kohl (1989), Wallace, Seigerman and Holbrook (1993), Prag and Casavant (1994), Sochay (1994), Sawhney and Eliashberg (1996), Neelamegham and Chintagunta (1999), Basuroy et al. (2003), and Ainslie, Dreze and Zufryden (2005) all report a positive relationship between the presence of a movie star and box office revenues.
The above discussion suggests that the longer one movie artist has been involved in one movie genre, the better audiences would associate him/her with movies in that genre and the better would the financial performance of his/her new movie in that genre will be.
Hypothesis 1a. The time period an artist has been involved in one movie genre is positively associated with the financial performance of artist’s new movie in this genre.
Signaling theory (Spence, 1973) explains how decisions are being made if limited information is available by focusing on the availability of signals of quality to the decision maker. The lower the ability of the individual consumer to evaluate the product on offer, the more important the presence of signals that do not directly derive from the product or the producer will be (Situmeang et al., 2014). Expert evaluations such as reviews in newspapers or other media can play an important role with respect to product performance, as has been found in many studies, especially with regard to the movie industry (Basuroy et al., 2003; Elberse and Eliashberg, 2003; Eliashberg and Shugan, 1997). Experts are neither the end user nor the producers relating to the movies, but are people who set standards based on their knowledge of the specific field to help consumers with their final decisions (Kwon and Easton, 2010; Wijnberg and Gemser, 2000; Priem, 2007). And according to the research of Reinstein and Snyder (2005), the distinctive characteristic of expert reviews is that they are issued by an independent party rather than the company itself. On the one hand, the independence of the expert may decrease the bias in the
information provided. On the other hand, the expert may not share the same incentive to deliver the information to consumers, reducing the influence on demand. Particularly in the creative industries, approval by experts can pose a determining impact on the market performance because of the difficulty for consumers to assess product quality before consumption (Wijnberg and Gemser, 2000; Priem, 2007; Caves, 2000).
Prior research has suggested that, even though there exist overlaps between experts’ evaluations and ordinary consumers’ evaluations (Holbrook, 1999), the critic standards of experts do seem to differ from ordinary consumers’ (Hirschman and Pieros, 1985; Holbrook, 1999; Plucker et al., 2009). In mainstream movies, evaluations of consumers tend to be based on the entertainment value of the movie whereas critics tend to analyze the artistic value of the movie according to a set of formal and artistic standards or norms (Hirschman and Pieros, 1985). Compared to critic’s reviews, the reviews of audiences are likely to be less scientific and focused more on the pleasure derived from the movie itself rather than evaluating the movie in a more technical and analytical method and expanding to a broader (social, historical, economic etc.) context (Hirschman and Pieros, 1985).
Moreover, consumers tend to be more appreciative of products or services that satisfy or conform to their current needs and wants; they usually find it difficult to see beyond the current state (e.g., Bell et al., 2004; Ordanini and Parasuraman, 2011). Experts, on the contrary, are more likely to value unconformity and have, to quote Priem (2007: 226), “reputation-based incentives” to identify innovation. Holbrook’s research on the movie industry (1999) found that audiences gravitate more towards movies that are accessible due to their greater realism in representing a familiar setting and catering for conventional values, and adherence to the blockbuster mentality. Contrarily, critics lean toward products that are challenging with respect to artistic values, cinematic styles and deviations from traditional values and setting (Holbrook, 1999).
Thus, as the critics view the movie with a more professional standard, the longer time period one movie artist has been involved in one movie genre may be reckoned as not making any progress and improvement and thus would decrease the artistic performance of his/her new movie in that genre.
Hypothesis 1b. The time period an artist has been involved in one movie genre is negatively associated with the artistic performance of the artist’s new movie in this genre.
2.2 The moderating role of time difference between the two movies
Within the time-based resource-sharing (TBRS) model framework, a large body of evidence demonstrates that, in complex span tasks, memory recall performance is a function of the time during which intervening activities occupy attention, which suggests that memory traces suffer from a temporal decay while attention is diverted away (Barrouillet, Bernardin, and Camos, 2004; Barrouillet, Bernardin, Portrat, Vergauwe, and Camos, 2007; Barrouillet, Portrat, and Camos, 2011; Vergauwe, Barrouillet, and Camos, 2009, 2010). Previous studies have shown that memory recall performance in this kind of span task is dependable on the ratio between the time contributed to perform the distractive task (i.e., the processing time) and the free time available to reactivate memory traces (i.e., the refreshing time): The higher this ratio, the poorer the recall performance (Barrouillet et al., 2004, 2007; Barrouillet, Portrat and Camos, 2011). One possible reason for this relation is to assume that memory traces decay with time. Research of Barrouillet, Paepe and Langerock (2012) has added proof to the growing body of evidence that working memory depends on time-related factors strongly and constituted a further evidence for temporal decay in the short term, suggesting that the causal effect of time on forgetting cannot be ignored easily.
Thus we assume, according to the discussions above, that both audiences and critics could not easily associate the movie artist with one movie genre if there is longer time difference between the artist’s two movies, which will result a moderating effect on the relation between time period one movie artist has been involved in one movie genre and the market performance of the artist’s new movie in this genre. The critics would devalue the new movie even more if there is longer time difference between the artist’s two movies as they would deem that the movie artist lacks diligence and is not focusing on his/her career. Moreover, longer time difference between the artist’s two movies in one genre also indicates a perceived lack of innovativeness and breakthrough. Thus, the following hypotheses are formed:
Hypothesis 2a. The time difference between the artist’s two movies moderates the effect of the time period an artist has been involved in one movie genre on the new movie’s financial performance. The effect of the time period an artist has been involved in one movie genre on the new movie’s financial performance is less positive when there is a longer time difference between the two movies.
Hypothesis 2b. The time difference between the artist’s two movies moderates the effect of the time period an artist has been involved in one movie genre on the new movie’s artistic performance. The effect of the time period an artist has been involved in one movie genre on the new movie’s artistic performance is more negative when there is a longer time difference before the switch.
2.3 The moderating role of total genre similarity between the two movies
Genre, a French word, which means types or classes of sub products within a given literary product (Abrams, 2011) and has become an essential element in the production and
distribution strategies of movies. Beyond expectation, very little investigation on movie genres (e.g., horror, romance, or drama) has been done, and very limited research that existed has taken genre only as a control variable (e.g., Neelamegham and Chintagunta, 1999). On the contrary, research studies in communication and media (Austin and Gordon, 1987) have not only classified movies in different genres but have also shown that consumers regard genre labels as a very convenient and easy way to categorize movies and distinguish among movie types (Desai and Basuroy, 2005). Moreover, genre labeling can also help to clarify or elaborate on the movie story (Austin and Gordon, 1987). When people go to movies, they all bear some extent of expectations, including some conceptions associated with the specific movie genre. These expectations might in turn build genre preferences among consumers and affect their choices on movies (Austin and Gordon, 1987).
Research up to date has shown that many consumers take a movie’s genre as the most vital, and probably the number one factor they consider when it comes to deciding which specific movies to see (Austin and Gordon, 1987; De Silva, 1998). The arguments above hold the assumption that audiences should have stored some information/knowledge about movies specific to various genres in their memory. The literatures on product familiarity in general, and in particular on product satisfaction, implies that special characteristics of the movie product market (vs. other product markets) make it highly possible that consumers store movie information (e.g., expectations and attitude) at the aggregate level of genres, as opposed to at the specific level of individual movies (Cadotte, Woodruff and Jenkins, 1987; Woodruff, Cadotte and Jenkins, 1983). This phenomenon occurs mostly when product types (e.g., dramas) within a product class (e.g., movies) have many versus few brands (e.g., individual movies), and the audiences have already had experience with lots of them, but no one specific brand stands out as the desired reference brand. The reason behind it probably is that audiences do not have extensive experience with any one of them, because there is little possibility in such product
markets. As a result, it becomes crucial to examine the influence of this “aggregate” variable on the market performance of individual movies that it includes (Desai and Basuroy, 2005).
In the research of Kovács and Hannan (2010), they argue that the effect of category spanning on how audiences accept it depends on the fuzziness of the categories. When a set of categories lacks contrast, or in other words, have very fuzzy boundaries, spanning categories does not cause much additional confusion for audiences, thus the penalties associated with spanning should be slight. However, when the contrasts of the categories spanned are high, audiences will find it difficult to interpret the producer, so spanning categories will be devalued more.
Hypothesis 3a. The genre similarity between the artist’s two movies moderates the effect of the time period an artist has been involved in one movie genre on the new movie’s financial performance. The effect of the time period an artist has been involved in one movie genre on the new movie’s financial performance is more positive when the artist’s two movies are more similar.
However, for movie critics, as discussed earlier, they hold higher standards for evaluating the movie. When the artist’s two movies are similar, the critics would devalue the new movie from an artistic perspective as the movie artist does not challenge himself/herself and is repeating his/her role in the previous movies, especially when that artist has been involved in that movie genre for a long time. Thus, the time period’s effect on the new movie’s artistic performance will be moderated by the genre similarity between the two movies.
Hypothesis 3b. The genre similarity between the two movies of the same artist moderates the effect of the time period one artist has been involved in one movie genre on the new movie’s
artistic performance. The effect of the time period an artist has been involved in one movie genre on the new movie’s artistic performance is more negative when the artist’s two movies are more similar.
2.4 Conceptual Framework
3. Data & Methods
3.1 Data and sample
This study uses a cross-sectional research design to study the relationship between the time period one artist has been involved in one movie genre, the time difference between the artist’s two movies, the total genre similarity between the two movies and the market performance of the artist’s new movie in this genre.
The raw sample in this study consists of 6421 movies produced from 1940 to 2015 in the USA, and the data was downloaded from Metacritic.com. Metacritic.com is a website that aggregates reviews of music albums, games, movies, TV shows, DVDs, and formerly, books. For each product, a numerical score from each review is obtained and the total is averaged.
The raw data includes the movie name, the release date, the movie artists played in the movie, movie genres, gross box office, the meta score, the production budget and the user score. Because some data required for this research in some movies are missing, only 3704 movies are used for this research.
3.2 Dependent Variables
The dependent variable in this research is the market performance of movies. This variable is measure by examining two indicators: (1) the financial performance (indicated by gross box office of the movie) and (2) the artistic performance (indicated by meta score of the movie). The box office is in USD and depreciation is not taken into account in this study. The meta score is the score of the movie on the Metacritic website. Scores are given by movie critics and the weighted score (based on the fame of the critic) is finally assigned and thus would be a good indicator for the movie’s artistic performance. Usually, "very good"(critical favorites) movies have scores above 70.
3.3 Independent Variable
The independent variable in this research is the time period one movie artist has been involved in one movie genre. Whether or not one movie is in one movie genre is coded as a binary: 1 = in this movie genre, 0 = not in this movie genre. Then VBA in Excel was adopted to calculate the time period one artist has been involved in one movie genre using the data of the movie’s release date.
3.4 Moderators
There are two moderators for this study: (1) the time difference between the artist’s two movies in the same genre and (2) the total genre similarity between the artist’s two movies in the
same genre. Time difference was calculated using the same method as calculating the time period (the independent variable).
Genre similarity was measured using a commonly used method in the category spanning literature (e.g., Kovács and Hannan, 2010), one can measure category similarity by observing how common it is in the industry for two genres to co-occur in products produced by the same artist. This is an adaptation of the simple and widely used measure of similarities developed by Jaccard (1901), which is calculated as follows: let i and j denote two different genres, then the similarity between i and j can be defined as the ratio of the number of co-occurrences (i.e., producers that are active in both i and j) to the number of all producers active in i and/or j. Formally this can be expressed as:
Similarity(i,j)=(|i∩j|)/(|i∪j|)
The range of the output of the above function is between 0 and 1. The higher the number, the more similar the two movie genres. 0 denotes perfect dissimilarity and 1 denotes perfect similarity.
After that, a total genre similarity between two movies is calculated by using simple addition in Excel.
3.5 Control Variables
This study controls for several factors. The first control variable is the production budget. This control variable is vital as it would influence both the movie quality and the movie artists showing in the movie, according to the study of Ravid (1999) that big-budget movies may signal high revenues, regardless of the source of spending. The assumption is that the higher the production budget, the better the financial performance of the movie will be. Production budget, the same as box office, is in USD.
Furthermore, the analysis controls for the user score of the movie. A higher user score in general indicates a better artistic performance of the movie. Thus the assumption is that the higher the user score of the movie, the better the meta score of the movie will be.
3.6 Methods
As this study contains two dependent variables, the analysis will be done twofold. The proposed hypotheses will be tested via hierarchical linear regression analysis. Regression analysis is used when one or several independent variables are hypothesized to affect one dependent variable. Thus, regression analysis finds the best fitting straight line through a set of points. The relationship can be described in the following equation:
Y= β0+ β1X1+ β1* β2X2+ β1* β3X3+ 𝜀
In the formula, Y represents the dependent variable, market performance measured by either the financial performance of the movie, indicated by box office or the artistic performance of the movie, indicated by meta score. The regression coefficients are represented by β0, the intercept, and β1, which represent the time period one artist has been involved in one movie genre as the independent variable, and β1* β2 which represents the interaction effect between time period and time difference (the moderator 1), and β1* β3 which represents the interaction effect between time period and total genre similarity (the moderator 2). Moreover, 𝜀1 stands for the difference between the estimated Xi and the actual Xi (Field, 2009).
Table 1 shows a summary of the combinations of variables used in the various regression analyses. To test for moderation, a modeling tool called PROCESS developed by Hayes was used. This model integrates many of the functions of existing statistical tools for testing moderation into one model. The model that is appropriate to test hypothesis X is the simple moderation model in which a predictor influences an outcome directly and moderator variables
affect the strength and/or direction of the relation between a predictor and an outcome: enhancing, reducing, or changing the influence of the predictor (Fairchild & Mackinnon, 2009).
To measure the effect of time period on market performance of the artist’s new movie, a hierarchical regression is used while controlling for the variables that were elaborated on above. As shown in Table 4, Step 1 includes the control variables, as discussed above. In the second step, the control variables are combined with the independent variable (Time period) to test the effect on the dependent variable measured by box office without moderating effect. In moderating analyses section, the moderation effect of time difference and total genre similarity are tested respectively while still controlling for production budget and user score. The same method was used for box office and meta score since the market performance is measured by both of them.
Table 1: Summary of Regression Analyses
Control Variable Dependent Variable Independent Variable Moderator
Production Budget User Score Box Office Meta Score Time Period difference Time Total Genre Similarity
Model 1 X X X Model 2 X X X X Model 3 X X X X X Model 4 X X X X X Model 5 X X X X X X Model 6 X X X Model 7 X X X X Model 8 X X X X X Model 9 X X X X X Model 10 X X X X X X
4. Analyses & Results
In this chapter, the statistical analyses of the data will be presented. This section will start with an overview of the descriptive statistics, correlations between the variables and a test for
multicollinearity. Following will be the regression analyses described in the method section. The results will be used to draw conclusions regarding the proposed hypotheses of this study.
4.1 Descriptive Statistics
The descriptive statistics of the dependent variables, the independent variable, the moderators and the control variables can be found below in Table 2.
Since one movie includes several movie artists, there are in total 81644 data listed on a one artist - one movie base for the independent variable (time period one artist has been involved in one movie genre) and the two dependent variables (box office and meta score of the movie). Moreover, since one movie occupies several movie genres and some artists switched between different movie genres, the number of data for variables concerning time difference and total genre similarity are 27109 and 24876 respectively, also on a one artist - one movie base. There are some data missing for the control variables: production budget and user score, thus the number of data concerning these two variables are 44492 and 71551 respectively, on a one artist - one movie base.
As shown, the box office varies from 0 to 760,507,625 USD and the overall mean is 30,936,104.14 USD with a standard deviation of 56,288,464.93. The meta score varies from 1 to 96 out of 100 and the overall mean is 53.86 with a standard deviation of 17.533.
The production budget varies from 1,100 to 425,000,000 USD and the overall mean is 39,017,907.13 USD with a standard deviation of 42,411,226.24. The average production budget is higher than the average box office and the reason may be that the data of some movies’ production budget is missing. Another control variable, user score varies from 1.4 to 9.2 out of 10 and the overall mean is 6.793 with a standard deviation of 1.337.
The independent variable, time period one movie artist has been involved in one movie genre varies from 0 to 6,420 and the overall mean is 1066.05 with a standard deviation of
1847.533. This means that the average time period of one artist being involved in one movie genre is 1066.05 days.
The moderator time difference varies from 1 to 21,644 and the overall mean reports 1,630.64 with a standard deviation of 2496.904. So on average, it takes 1,630.64 days before one artist’s new movie being released. Another moderator, total genre similarity varies from -0.667 to 10.79 and the overall mean reports 2.3334 with a standard deviation of 1.7213. This indicates that the average total genre similarity between one artist’s two movies is 2.3334 on a scale with 1 represents the same movie genre.
Table 2: Descriptive Analysis
N Min Max Mean SD
Box Office 81644 0 760507625 30936104.14 56288464.93 Meta Score 81644 1 96 53.86 17.533 Production Budget 44492 1100 425000000 39017907.13 42411226.24 User Score 71551 1.4 9.2 6.793 1.337 Time Period 81644 0 6420 1066.05 1847.533 Time Difference 27109 1 21644 1630.64 2496.904 Genre Similarity 24876 -0.6669 10.7929 2.3334 1.7213 Valid N (listwise) 16254 4.2 Correlations
To test for multicollinearity between variables, two methods were adopted. First, a correlation analysis has been done to assess the correlation between the dependent variable, independent variables, moderators and control variables of this study. As displayed in Table 3, none of the correlations are extremely high, above 0.80 (Field, 2009). This is the first indicator that there is no multicollinearity between the variables. The highest correlation (0.661**) is the correlation between box office (dependent variable) and production budget (control variable), which is in line with our assumption that the higher the production budget, the better the box
office. The significantly positive correlation (0.575**) between meta score (dependent variable) and user score (control variable) is also in line with our expectation from the literature that a higher user score indicates a higher meta score.
The correlations between the independent variable (time period one artist has been involved in one movie genre) and the two dependent variables (the box office and the meta score of the artist’s new movie in this genre) are both significant and are in line with the expectations: the time period should be positively correlated with box office (0.108**) and negatively correlated with meta score (-0.06**).
An interesting observation is that total genre similarity has a positive correlation with meta score (0.010) (though not significant) while a negative correlation should be expected following the literature. This positive correlation could imply that when an artist shows up in a movie that is more similar to his/her previous movies, the artistic performance of the new movie would be better, as well as the financial performance of the new movie (0.027**). In contradiction to expectations from the literature is that time difference has a positive correlation with both the box office (0.019**) and the meta score (0.064**), however, a negative correlation should be expected. This could imply that the longer it takes before one artist shows up in a new movie, the better both the financial and the artistic performance will be.
Second, the Variance Indicator Factor (VIF) is used as a second indicator to test whether or not there consists multicollinearity between the variables. The VIF indicates whether or not a predicator has a strong linear relationship with the other predicators. A VIF score above 3 indicates a problem that there exists strong linear relationship. As tested in the regression models, all the VIF values of variables in this study are lower than 3 thus multicollinearity is no issue for this analysis.
Table 3: Correlation Analysis Mean SD 1 2 3 4 5 6 7 1. Box Office 30936104.14 56288464.93 1 2.Meta Score 53.86205967 17.533 0.090** 1 3.Production Budget 39017907.13 42411226.24 0.661** -0.008 1 4.User Score 6.793 1.3370 0.006** 0.575** -0.097** 1 5.Time Period 1066.05 1847.533 0.108** -0.06** 0.086** -0.051** 1 6.Time Difference 1630.64 2496.904 0.019** 0.064** -0.022** 0.144** 0.005 1 7.Genre Similarity 2.3334 1.7213 0.027** 0.010 0.015 0.004 0.038** 0.089** 1
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
4.3 Regression Analyses 4.3.1 Box Office Regression
In this section, the results of the hierarchical regression of time period, production budget and user score on the independent variable - box office will be reported. This analysis is done to check whether H1a is hold. H1a expects a positive relationship between time period one movie artist has been involved in one movie genre and the financial performance of the artist’s new movie in this genre. The outcome of the regression analysis can be found in Table 4 below.
As can be seen in Table 4, the coefficients of control variables are significant in each step of the regression model. The hierarchical regression model is significant in both steps of the analysis as can be seen from the F-statistics (17904.110** and 11942.617**) in the table. R-square which stands for the explained variation is significantly high for both steps in the regression model (0.453** and 0.453**). This means that a relatively large part of the financial performance of the artist’s new movie in this genre can be explained by time period one movie
The coefficient of independent variable is significantly positive (0.012**), which is in support for H1a. From the box office hierarchical regression, it can be concluded that the time period one artist has been involved in one movie genre is positively associated with the financial performance of the artist’s new movie in this genre.
Table 4: Box Office Regression
Box Office Step 1 Step 2 Control Variable Production Budget 0.672** 0.671** User Score 0.144** 0.144** Independent Variable Time period 0.012** (constant) -40412388.20 -40939271.87 R-Square 0.453** 0.453** F-Statistic 17904.110** 11942.617** F-Change 17904.110** 11.189**
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
4.3.2 Meta Score Regression
In this section, the results of the hierarchical regression of time period, production budget and user score on the independent variable - meta score will be reported. This analysis is done to check whether H1b is hold. H1b expects a negative relationship between time period one movie artist has been involved in one movie genre and the artistic performance of the artist’s new movie in this genre. The outcome of the regression analysis can be found in Table 5 below.
As can be seen in Table 5, the coefficients of control variables are significant in each step of the regression model. The hierarchical regression model is significant in both steps of the analysis as can be seen from the F-statistics (13359.428** and 8922.27**) in the table. R-square which stands for the explained variation is significantly high for both steps in the regression model (0.382** and 0.382**). This means that a relatively large part of the artistic performance
of the artist’s new movie in this genre can be explained by time period one movie artist has been involved in one movie genre.
The coefficient of independent variable is significantly negative (-0.021**), which is in support for H1b. From the meta score hierarchical regression, it can be concluded that the time period one artist has been involved in one movie genre is negatively associated with the artistic performance of the artist’s new movie in this genre.
Table 5: Meta Score Regression
Meta Score Step 1 Step 2 Control Variable Production Budget 0.050** 0.052** User Score 0.621** 0.621** Independent Variable Time period -0.021** (constant) -4.267 -4.028 R-Square 0.382** 0.382** F-Statistic 13359.428** 8922.27** F-Change 13359.428** 30.019**
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
4.4 Moderation Market Performance
In this section it will be researched whether time difference between the artist’s two movies or the total genre similarity between the artist’s two movies is moderating the relationship between time period one movie artist has been involved in one movie genre and the market performance of the artist’s new movie in this genre. From section 4.3.1 and 4.3.2, it has become clear that time period one movie artist has been involved in one movie genre has different effects on the financial performance and the artistic performance of the artist’s new movie in this genre, therefore the tests for moderation are conducted separately for financial performance and artistic performance. Four similar models are created. The first one tests for the
moderating effect of time difference between the artist’s two movies on the relationship between time period one movie artist has been involved in one movie genre and the box office of the artist’s new movie in this genre. The second model tests the moderating effect of total genre similarity on the relationship between time period one movie artist has been involved in one movie genre and the box office of the artist’s new movie in this genre. The third model test the moderating effect of time difference between the artist’s two movies on the relationship between time period one movie artist has been involved in one movie genre and the meta score of the artist’s new movie in this genre. And the last model tests for the moderating effect of total genre similarity on the relationship between time period one movie artist has been involved in one movie genre and the meta score of the artist’s new movie in this genre. In all models, control variables production budget and user score are added.
In order to conduct the moderating test, mean centering method is used in the regression analysis by transforming the independent variables, dependent variables, control variables and moderators into Z value respectively in order to avoid the collinearity among the variables. According to Aiken, West and Reno (1991), centering is the process of selecting a reference value for each predictor and coding the data based on that reference value so that each regression coefficient that is estimated and tested is relevant to the research question.
4.4.1 Moderation time difference on time period one movie artist has been involved in one movie genre and the financial performance of the artist’s new movie in this genre
In this section, we will look at whether time difference between the artist’s two movies is negatively moderating the relationship between time period one movie artist has been involved in one movie genre and the financial performance of the artist’s new movie in this genre when measuring financial performance by box office. This is done to check whether H2a can be supported. A moderation analysis using PROCESS in SPSS is conducted. Since this study only
involves one moderating variable (time difference), a single moderation analysis is done. The results are reported in Table 6.
The regression coefficient for X*M is -0.0090 and is not statistically different from zero, t= -0.0119, p=0.9905. Thus, the effect of time period one artist has been involved in one movie genre on the box office of the artist’s new movie in this genre does not depend on the time difference between the artist’s two movies. Moreover, this interaction effect accounts for 0% of variance in box office. As it can be seen from probing the interactions, the slope linking time period and box office is negative among all groups of time difference but not significant. In other words, although time difference does not significantly moderate the relationship between time period and box office, such trend has the same direction within all groups. However, the direction of the moderating effect is the same as expected in H2a where we expect that the time difference between the artist’s two movies will have a negative moderating effect on time period and box office. Thus H2a is partially supported.
Table 6 Coefficient SE t Constant 23405981.3000 617794.8310 37.8863 *** Time Period(X) -129.9927 216.9367 -0.5992 Time Difference(M) -42.2531 235.2645 -0.1796 Time Period*Time Difference(X*M) -0.0090 0.0765 -0.0119 R-Square=0.4891 F=3357.2344 p<0.001
Time Difference Unstandardized boot effects Boot SE Boot LLCI Boot ULCI Conditional effect at levels of time difference -1629.6400 -128.5034 256.0053 -630.2991 373.2923 5.6718 -129.9978 216.9137 -555.1702 295.1745 2485.1978 -132.2638 280.6457 -682.3573 417.8298 R-Square change=0.000 p=0.9905
**. Correlation is significant at the 0.01 level (2-tailed). *.
4.4.2 Moderation time difference on time period one movie artist has been involved in one movie genre and the artistic performance of the artist’s new movie in this genre
In this section, we will look at whether time difference between the movie artist’s two movies is negatively moderating the relationship between time period one movie artist has been involved in one movie genre and the artistic performance of the artist’s new movie in this genre when measuring artistic performance by meta score. This is done to check whether H2b can be supported. A moderation analysis using PROCESS in SPSS is conducted. Since this study only involves one moderating variable (time difference), a single moderation analysis is done. The results are reported in Table 7.
The regression coefficient for X*M is 0.0000, which is not statistically different from zero, t=1.1899, p=0.2341. Thus, the effect of time period one artist has been involved in one movie genre on meta score does not depend on the time difference between the artist’s two movies. Moreover, this interaction effect accounts for 0% of variance in meta score. As it can be seen from probing the interactions, the slope linking time period and meta score is positive among all groups of time difference and only significant when the levels of time difference are middle and high (time difference=1.1407/2476.8436). In other words, although time difference significantly moderates the relationship between time period and meta score only when levels of time difference are middle and high, such trend has the same direction within all groups. However, the direction of the moderating effect is different from what H2b has expected that there exists a negative moderating effect on time period and meta score. Thus H2b is not supported.
Table 7 Coefficient SE t Constant -0.8642 0.1592 -5.4270 *** Time Period(X) 0.0020 0.0001 2.7720 *** Time Difference(M) -0.0003 0.0001 -5.2346 ** Time Period*Time Difference(X*M) 0.0000 0.0807 1.1899 R-Square=0.3837 F=2192.3447 p<0.001
Time Difference Unstandardized boot effects Boot SE Boot LLCI Boot ULCI Conditional effect at levels of time difference -1629.6400 0.0001 0.0001 0.0000 0.0002 † 1.1407 0.0002 0.0001 0.0000 0.0003 ** 2476.8436 0.0002 0.0001 0.0001 0.0004 ** R-Square change=0.000 p=0.2341
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
†. Correlation is significant at the 0.100 level (2-tailed).
4.4.3 Moderation total genre similarity on time period one movie artist has been involved in one movie genre and the financial performance of the artist’s new movie in this genre
In this section, we will look at whether total genre similarity between the same artist’s two movies is positively moderating the relationship between time period one artist has been involved in one movie genre and the financial performance of the artist’s new movie in this genre when measuring financial performance by box office. This is done to check whether H3a can be supported. A moderation analysis using PROCESS in SPSS is conducted. Since this study only involves one moderating variable (total genre similarity), a single moderation analysis is done. The results are reported in Table 8.
The regression coefficient for X*M is 249.6442 and is statistically different from zero, t= 1.8766, p=0.0606. Thus, the effect of time period one artist has been involved in one movie
genre on box office depends on the total genre similarity between the artist’s two movies but the moderating effect is only significant at the 0.100 level (2-tailed). Moreover, this interaction effect accounts for 0.01% of variance in box office. As it can be seen from probing the interactions, the slope linking time period and box office is negative when total genre similarity is low (genre similarity=-1.6972, effect=-377.7280), but positive when total genre similarity is middle and high though not significant. In other words, total genre similarity does not significantly moderate the relationship between time period and box office and such trend has different directions within all groups. This is different from what H3a has expected. Thus H3a is partially supported when the level of total genre similarity is middle and high though the moderating effect is not significant.
Table 8 Coefficient SE t Constant 22407054.5 645701.23 34.7019 *** Time Period(X) 45.9589 226.3177 0.2031 Genre Similarity(M) 56244.9022 386070.709 0.1457 Time Period*Genre Similarity(X*M) 249.6442 133.0289 1.8766 † R-Square=0.4873 F=3076.4358 p<0.001
Genre Similarity Unstandardized boot effects Boot SE Boot LLCI Boot ULCI Conditional effect at levels of genre similarity -1.6972 -377.7280 324.1055 -1013.0107 257.5547 0.0245 52.0641 226.2503 -391.4115 495.5397 1.7461 481.8562 319.7482 -144.8856 1108.5979 R-Square change=0.0001 p=0.0606
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
4.4.4 Moderation total genre similarity on time period one movie artist has been involved in one movie genre and the artistic performance of the artist’s new movie in this genre
In this section, we will look at whether total genre similarity between the artist’s two movies is negatively moderating the relationship between time period one movie artist has been involved in one movie genre and the artistic performance of the artist’s new movie in this genre when measuring artistic performance by meta score. This is done to check whether H3b can be supported. A moderation analysis using PROCESS in SPSS is conducted. Since this study only involves one moderating variable (total genre similarity), a single moderation analysis is done. The results are reported in Table 9.
The regression coefficient for X*M is 0.0000, which is not statistically different from zero, t= 0.8854, p=0.3759. Thus, the effect of time period one artist has been involved in one movie genre on meta score does not depend on the total genre similarity between the artist’s two movies. Moreover, this interaction effect accounts for 0% of variance in meta score. As it can be seen from probing the interactions, the slope linking time period and meta score is positive among all groups of time difference and only significant when the level of total genre similarity is middle and high (total genre similarity=0.0243/1.7478). In other words, although time difference significantly affects meta score only among middle and high levels of total genre similarity, such trend has the same direction within all groups. However, the signs of the coefficient are positive, which is contrary to what H3b has expected. Thus H3b is not supported.
Table 9 Coefficient SE t p Constant -1.0033 0.1662 -6.0371 *** Time Period(X) 0.0002 0.0001 2.5931 ** Genre Similarity(M) -0.0016 0.0992 -0.0161 Time Period*Genre Similarity(X*M) 0.0000 0.0000 0.8854
R-Square=0.3779 F=1974.2699 p<0.001
Genre Similarity Unstandardized boot effects Boot SE Boot LLCI Boot ULCI Conditional effect at levels of genre similarity -1.6992 0.0001 0.0001 -0.0001 0.0003 0.0243 0.0002 0.0001 0.0000 0.0003 ** 1.7478 0.0002 0.0001 0.0000 0.0004 * R-Square change=0.000 p=0.3759
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
5. Discussions
5.1 Summary
There exists a well-developed stream of literature on how movie genre, movie star and critics review influences either the financial and the artistic performance of a movie (e.g., Reddy, Swaminathan and Motley, 1998; Desai and Basuroy, 2005; Austin and Gordon, 1987; Ravid, 1999; Levin, Levin and Heath, 1997). This study contributes to these literatures in three different ways. First, it combines the effect of movie artist’s power and movie genre, rather than focusing only on star power’s or movie genre’s influence on the market performance of a movie. Second, it shows how the movie artist’s power and movie genre interact with each other and decide the market performance of the movie. Third, it distinguishes between two indicators of the movie’s market performance (the financial performance and the artistic performance) with prior literature solely focusing on either box office or critics’ reviews.
Following the literature on star power and genre in the motion picture industry and category spanning, it was hypothesized that the longer time period one movie artist has been involved in one movie genre would have positive effect on the financial performance of the movie artist’s new movie in this genre and negative effect on the artistic performance of the movie artist’s new movie in this genre. This study indeed finds these relationships and the effect
is significant. When one movie artist has been involved in one movie genre for a long time period, audiences could easily associate him/her with movies in this genre and thus cast a positive influence on the financial performance of the artist’s new movie in this genre. However, when one movie artist has been involved in one movie genre for a long time period, the critics, who hold a professional and artistic standard for evaluation, will take this as a sign of lacking innovativeness and devalue the artistic performance of the artist’s new movie in this genre.
Moreover, the time difference between the artist’s two movies has a moderating effect on the relation between the time period one movie artist has been involved in one movie genre and the financial performance of the artist’s new movie in this genre. When there is longer time difference between the artist’s two movies, the audiences’ memory of this movie artist’s been in this movie genre gets faded and couldn’t easily associate this movie artist to this movie genre, thus results in worse financial performance. However, this moderating effect is not very significant. It is a widely accepted phenomenon that information temporarily maintained for immediate use or recall rapidly vanishes from memory. A venerable tradition attributed forgetting only to the passage of time and assumes that memory traces suffer from a temporal decay (Brown. J, 1958; Conrad, 1967; Peterson and Peterson, 1959). Nevertheless, further findings on which this hypothesis was based came to the conclusion that, there is no evidence for temporal decay in working memory so far (Lewandowsky and Oberauer, 2009; Lewandowsky, Oberauer and Brown, 2009). Therefore, modern psychology has rejected this hypothesis and commonly assumes that forgetting is not because of decay (Brown and Lewandowsky, 2010), but because of representation-based interference created by the intervening events occurring between encoding and retrieval (Nairne, 1990; Oberauer and Kliegl, 2006; Oberauer and Lewandowsky, 2008). This could be another argument to explain the insignificant moderating effect of time difference on the relation between time period and financial performance as well as the results from the analysis that the moderating effect of time difference on the relation
between time period and artistic performance is positive instead of what we have hypothesized, negative.
The results suggest that the total genre similarity between the artist’s two movies has a positive moderating effect on the relation between time period one artist has been involved in one movie genre and the financial performance of the artist’s new movie in this genre. This supports the research of Austin and Gordon (1987) that genre labeling can help to clarify or elaborate on the movie story and the genre preferences built among consumers before can affect their choices on movies. Thus, when a movie artist has been involved in one movie genre for a long time and then when he takes a new movie that is very similar to his previous movie genre, audiences could associate him/her with the new movie as well and thus could lead to a better financial performance for the new movie. However, the moderating effect of total genre similarity on the relation between time period and artistic performance is not in line with what we have expected. As demonstrated in prior research, experts don’t rate the sequels, which may be deemed as “un-innovative”, low ( Situmeasng et al., 2014b). However, when companies continue to develop “related products”, it may backlash and turn positive evaluation from experts negative due to a perceived lack of innovativeness of companies. Thus in the hypothesis, we assumed a negative moderating effect of total genre similarity, that is, when there is larger total genre similarity between the artist’s two movies, critics may take this as a sign of lacking innovativeness and thus results in a worse artistic performance of the artist’s new movie in this genre. However, the suggestions from prior research on companies may not apply to movie artists that we are studying in this research. Another explanation for this phenomenon is that the standards critics hold when valuing movies can be various and complicated and that creativeness may only be one small part of the valuing system. Thus, total genre similarity can hardly cast effect on the relation between time period one movie artist has been involved in one movie genre and the artistic performance of the artist’s new movie in this genre.
5.2 Theoretical implications
This study contributes to the stream of research in category spanning in the movie industry and focuses on the interactive effect of movie star and movie genre on the market performance of the movie. This study’s insights provide support for some of the prior research on movies, which have treated movie stars as high-equity brands that enjoy both name recognition and positive image, and an association with particular movie types (Levin, Levin and Heath, 1997). Research on brand equity has shown that consumers favorably evaluate extensions of brands enjoying good reputation in categories that are similar to the parent categories (Aaker and Keller, 1990; Park, Milberg and Lawson, 1991).
This study furthermore provides support for the idea that even though there exist overlaps between experts’ evaluations and ordinary consumers’ evaluations (Holbrook, 1999), the critic standards of experts do seem to differ from ordinary consumers’ (Hirschman and Pieros, 1985; Holbrook, 1999; Plucker et al., 2009). In mainstream movies, evaluations of consumers tend to be based on the entertainment value of the movie whereas critics tend to analyze the artistic value of the movie according to a set of formal and artistic standards or norms (Hirschman and Pieros, 1985).
5.3 Managerial implications
As illustrated in the examples of Jim Carrey and Emma Watson in the introduction part of this thesis, when to enter a new movie genre and which movie genre to take are critical for movie artists. The results of this study help to suggest that movie artists should be very conscious about which movie genre to enter and when to enter and they also have to balance the movie’s performance from both financial and artistic aspects. In general, the longer one movie artist has been involved in one movie genre or switched only between very similar movie genres, audiences could easily associate him/her with one specific movie genre and thus would favorably
evaluate this artist’s new movie in this genre and lead to a better financial performance. However, when one movie artist has been involved in one movie genre for two long and did not make any innovation for his/her career, the critics may devalue this artist’s new movie in this genre.
Moreover, this study could have potential contribution for movie companies as well. As shown in prior research, movie artists are critical for movies’ market performance. Thus, it is very essential for movie companies when they are making decisions on which movie artist to pick for the new movie. For example, should they pick Jim Carrey, who is well-known for being an excellent comedy movie artist, for the company’s next comedy movie? Or should they use some other movie artists, like a big movie star but usually shows up in drama movies? Being more explicitly clear of the different effects movie artists being involved in movie genres have on financial and artistic performance of the new movies can assist move companies in making decisions on casting.
5.4 Limitations and recommendations for future research
The use of objective data gathered from third-party online sources instead of self-reported data from large scale surveys is one strength point of this research. However, in this study, there is a number of data missing for some movies thus the sample numbers for variables vary across different variables. Further studies may, however, be more cautious on choosing the data set that contains movies from a certain time period for study and make sure that all the samples are consist across different variables. Moreover, the choice of movie artists in this study includes all the casts in the movies. However, prior research has suggested that the big influence of movie casts on movie performance mainly applies to movie stars. Thus when choosing movie artists for study, further research may investigate the time period one movie star has been involved in the movie genre. In this way, the effect may be more significant.
Another limitation of this study is that it only takes meta score as the indicator for artistic performance of the movie. Though meta score is an objective and authoritative indicator to measure the artistic value of movies, we did not dig into the details regarding the grading system and measurement of meta score. There are some other indicators for artistic performance as well, for example, the awards winning records, the IMDb Score and the Rotten Tomatoes Score. Further research may benefit by exploring a more complete, developed and objective way to measure the artistic performance of movies.
