The Importance and Effects of Different Movie Attributes
Concerning Movie Trailers and Movie Watching Intentions:
A Choice-Based Conjoint Analysis
N. J. Rottink
The Importance and Effects of Different Movie Attributes
Concerning Movie Trailers and Movie Watching Intentions:
A Choice-Based Conjoint Analysis
N. J. Rottink
Faculty of Economics & Business
Master Thesis Msc. Marketing Intelligence
June 2014
Address:
W.A. Scholtenstraat 26a
Phone:
06-51514242
Email:
njrottink@gmail.com
Student Number:
S2029499
First Supervisor:
F. Eggers
Second Supervisor:
M. J. Gijsenberg
Organization:
Rijksuniversiteit Groningen
Preface:
This report is my Master thesis for the conclusion of my Master program Marketing Intelligence at the University of Groningen. It has been a real fun and challenging time and I’m really happy with how my master thesis turned out. It was really interesting doing research about such an under investigated and entertaining field of interest, the motion picture industry.
I would like to thank my head supervisor Felix Eggers and second supervisor Maarten Gijsenberg for their time and valuable input during the master thesis period, next I would also like to thank all the participants who helped gathering data for my study.
Abstract
This empirical study looks at how different movie attributes, like actors and the genre, have an effect on peoples intention to watch the movie trailer and watch the movie itself. This is done by performing a choice-based conjoint analysis. The attributes used are divided over four categories, which are the 4 Ps as described by Porter. So this study looks at how different movie attributes concerning price, place, promotion and product have an effect on people’s intentions to watch a movie (trailer). Furthermore this study looks at the moderating effect of people’s mood, and uses gender and age to segment the different respondents. The main goal of the study is to find the most important attributes when it comes to intentions to watch a movie (trailer).
The outcomes of this study show that movie attributes concerning the product and promotion have the most effect and are the most important when it comes to movie (trailer) watching intentions. Furthermore price seems to have some small effects and no effects are found when it comes to place. Next mixed results are found when it comes to the moderating effect of mood, because mood does have some moderating effects but it doesn’t improve the overall model fit. Lastly using a latent class analysis, different segments are found for the two different dependent variables investigated in this study. The study starts with an introduction about the motion picture industry, then a literature review, followed by the methods used. After that the results are presented followed by a discussion and conclusion. The study ends with some managerial implications and directions for further research.
Table of Contents
1. Introduction ... 6
2. Literature Review ... 8
2.1 Dependent Variables ... 9
2.1.1 Movie Genre (Product) ... 9
2.1.2 Actor Power (Product) ... 9
2.1.3 Ticket Price (Price) ... 10
2.1.4 Critic Review (Promotion) ... 10
2.1.5 Distribution (Place) ... 11
2.2 Moderating Variables ... 12
2.2.1 Mood ... 12
2.2.2 Gender & Age ... 12
2.3 Independent Variables ... 12
2.3.1 Intention to watch the movie trailer ... 12
2.3.2 Intention to watch the movie ... 13
2.4 Conceptual Model ... 13
3. Study ... 14
3.1 Method & Measurement ... 14
3.1.1 Attributes & levels ... 15
3.1.2 Choice Design ... 15
3.1.4 Mood, gender & age ... 16
4. Results ... 17
4.1 Sample ... 17
4.2 Conjoint Analysis ... 17
4.2.1 Intention to watch the movie trailer ... 18
4.2.2 Intention to watch the movie ... 20
... 23
4.4 Hypotheses ... 24
4.5 The No-choice option ... 27
4.6 Willingness-to-pay ... 27
4.7 Segmentation ... 28
4.7.1 Segments for intention to watch the movie trailer ... 30
5. Conclusion & Discussion ... 34
5.1 Main effects ... 34
5.2 Moderating effects ... 36
5.3 Segmentation ... 37
6. Managerial Implications ... 38
7. Limitations & Directions for further research ... 39
7.1 Limitations ... 39
1. Introduction
Nowadays research on the motion picture industry and issues related to this big industry has become very popular. Some reasons for this rather new and interesting field of research are the fact that this industry has a highly economic importance in the global economy and is the key driver for the market of entertainment products. Especially for movies produced in the United States, more specifically, movies produced in Hollywood (Eliashberg et al., 2006).According to the Motion Picture Association of America (MPAA, 2014) the motion picture industry generates US$ 140 billion in total job wages yearly and accounts for US$ 14.3 billion in exports worldwide leading to almost US$ 45 billion in box offices worldwide. Another reason is the fact that movies attract a lot of attention and are a big part of culture. Movie critics’ sites like IMDB are visited frequently and social media sites pay a lot of attention to new trailer releases from big movie productions (Eliashberg et al., 2006). Another point is the fact that there is a lot of data to be worked with, a lot of new products (movies) are released in a very short time, and also the whole product life cycle for a movie is mostly transcendent (Eliashberg et al., 2006). While the motion picture industry generates a lot of money, the average movie loses around $17 million at the box-office, excluding DVD rentals and sales (Gazley et al., 2010). So because of this poor performance there has been a high demand for methodological assistance for predicting box-office success and the drivers for movie-going consumer behavior. Because of the fact that the production of a movie costs a company several millions of dollars, and only a small number of movies released each year actually make money during their theatrical run, it is important that a movie attracts an as large as possible audience by using the right promotional activities (Finsterwalder et al., 2012).
Although it is stated before that movie trailers are the most influential form of motion picture promotion, this does not mean a trailer really is the most influential for a person to have the intention to go watch a particular movie, and also this influence could be a negative influence, like when a person dislikes a certain actor in the movie trailer or maybe doesn’t like the genre. So considering the facts that movies cost a lot of money to produce, only a few movies actually make money, there are a lot of different approaches for the promotion of movies and people have various motives for watching a movie, there is not a lot of research within this field. There is a need for movie producers to know what really are the most important factors for consumers to watch a movie. So are movie trailers really the most important influential factor, or do other ways of promotion or movie attributes have a more influential effect on consumers and are movie trailers not as important as we think.
The primary objective of this study is to identify the main drivers for consumers to choose to watch a particular movie trailer and a particular movie, so what are the most influential variables that effect consumer choice in the motion picture industry. We thus focus on both the intention to watch the movie trailer and on the intention to watch the movie itself. More specifically this study wants to pinpoint the effect of movie attributes like actors & genre and promotional strategies on someone’s intention to go watch a movie trailer and/or watch the movie itself, while also considering psychological factors like someone’s mood. This is an important research area because as stated before, research in the motion picture industry is limited, especially when it comes to consumer choice. So due to the complications of making a movie as described before, it would be really interesting and effective for movie producers to know what really are the key motives for consumers to watch a movie trailer and/or the movie itself. So for example when it comes to the price, place or promotion of a movie what are the important things to focus on to attract more customers, or even when we look at the product itself.
This study builds on a previous study by Gazley et al.’s (2010) which aimed to identify the effects of movie attributes, information sources, promotional strategies, distribution strategies, pricing strategies and competition among different genres on consumer choice. While this study only incorporated economic variables and used a rating-based conjoint analysis this study will also incorporate psychological factors and use a choice-based conjoint analysis. This will help us not only find the most important attributes for consumer choice, but will also help us better understand the consumer choice process and why consumers choose a particular movie over another one. So this study also hopes to find the link between watching a particular movie trailer and the intention to watch that movie and if a movie trailer is really that important.
2. Literature Review
2.1 Dependent Variables
2.1.1 Movie Genre (Product)The genre of a movie denotes the type of the movie or the class the movie belongs too. People have various expectations associated with the genre the movie belongs too, which can create genre preferences and affect movie choices (Desai & Basuroy, 2005). According to previous research a movie’s genre is seen as the most important factor which influences a person to go see a particular movie (Austin & Gordon, 1987; De Silva, 1998). Next it is also pointed out that genre is one of the most important expectation influencers. People also have a strong opinion about the genre they prefer and a lot of people agree that there are genres which they are not interested in (Finsterwalder et al., 2012). So the genre of a movie has a big influence on people’s expectations and opinions about a movie, which consequently leads to an effect on someone’s intention to go watch a movie trailer and also the movie. With this information we come up with the following hypotheses.
H1a. Someone’s intention to watch a particular movie trailer will vary for different genres. H1b. Someone’s intention to watch a particular movie will vary for different genres.
2.1.2 Actor Power (Product)
Star power, or the effect of popular actors/actresses, is a recurring theme in the motion picture industry and research in this field. The direct effect of star power is the effect of featuring a highly popular star in a movie which makes consumers expect a highly entertaining (or high-quality) film, as long as the movie is of a genre with which the star is typically associated (Desai & Basuroy, 2005). Furthermore star power can also have a moderating effect between, for example the genre of the movie, and the liking of that movie by consumers (Gazley et al., 2010). So the actors who play in a particular movie have a very big influence on customers, because customers have some prior experience with actors, therefore have pre-determined expectations of these actors and therefore want these actors to live up to these expectations (Finsterwalder et al., 2012).
This leads to the next hypotheses
2.1.3 Ticket Price (Price)
There are not a lot of studies which have considered varying ticket prices for a movie as a variable. But as there are movies with a higher or lower ticket price (due to for example 3D, student or elderly discounts)and the fact that we are looking at behavioral intentions from the concept of marketing theory and the marketing mix we will include varying ticket prices into this study, even though the ticket prices do not vary much and most of the times have the same price for everyone (Orbach & Einav, 2007). It would not only be a valued contribution to see the immediate impact of price changes on someone’s intention to watch a movie (trailer), it would also be a valued point for motion picture researches to see how much someone is willing to pay for different movie attributes, for example if people are willing to pay more for a movie if it has better reviews, or is a different genre.
So as it is with most products and services we look at pricing theory, which states that when all variables are at an equal level, customers prefer a ticket price which is lower than compared to a ticket price which is higher (Begg et al., 2003), while of course this pricing theory wouldn’t have an effect on someone’s intention to watch a trailer, because trailers can ben see for free. With this information in mind the following hypotheses are formulated.
H3a. Someone’s intention to watch a movie trailer is not affected by price.
H3b. Someone’s intention to watch a movie is higher for a movie with a low ticket price than for a movie with a high ticket price.
2.1.4 Critic Review (Promotion)
The opinions of critics are very important for experience products like movies. Critical reviews give consumers some sort of indirect experience on sensory aspects, like quality of acting and music, which can’t be delivered by other attributes like the actors and music director (Desai & Basuroy, 2005). They also state that this influence from critical reviews is very substantial because these reviews are normally the first link in the diffusion of information about new products, movies, and also have a high credibility because of their professional status. Another study on this subject states that critical reviews are much more impactful when they are negative than when they are positive. So people are more influenced by negative messages than by positive messages ( d’Astous & Touil, 1999). This gets us to the next hypotheses.
H4a. Someone’s intention to watch a movie trailer is lower when the critical review about this movie is negative than when the review is positive.
H4b. Someone’s intention to watch a movie is lower when the critical review about this movie is negative than when the review is positive.
2.1.5 Distribution (Place)
By distribution we mean the place of the cinema where the trailer is shown. The location of the cinema affects the distribution strategy influence on consumer choice, meaning that having more locations to show the movie has a positive impact on the choice of the consumer to go watch the movie trailer (Gazley et al., 2010). This means that having the movie trailer shown at a lot of places increases the consumer’s intention to watch that movie. We therefore look at distribution in terms of how many places the movie is shown, more specifically if the movie is shown in major cinemas, minor cinemas or somewhere in between. Because of the fact that trailers can be watched everywhere and at every time we do not expect an effect of the distribution of the movie on the intention to watch a movie trailer.
H5a. Someone’s intention to watch a movie trailer is not affected by the number of places where the trailer is shown.
2.2 Moderating Variables
2.2.1 MoodAccording to previous studies someone’s mood has an impact on how that person experiences and evaluates an advertisement. This leads to the fact that an advertisement, or in this case a movie trailer, is evaluated more favorably when the mood someone is in is pleasant compared to unpleasant (Devlin et al., 2011). So the fact that the experience of a movie (trailer) is influenced by someone's mood leads to the point that someone’s mood will have an effect on the relation between the different movie attributes and someone’s intention to watch a movie (trailer). So we believe that someone’s mood will have a moderating effect on the relations. This leads to the following hypotheses.
H6a. The mood someone is in will have an effect on the relationship between the different movie attributes and the intention to watch a movie trailer.
H6b. The mood someone is in will have an effect on the relationship between the different movie attributes and the intention to watch a movie.
2.2.2 Gender & Age
There hasn’t been a lot of research on how a person’s age and gender affect their intention to watch movie trailers. We therefore don’t have any predetermined hypotheses regarding a person’s age and gender and how they affect the relationship between the movie attributes and someone’s intention to watch a movie trailer. Gender and age will mainly be used for segmentation uses for the different customer groups we expect to find.
2.3 Independent Variables
2.3.1 Intention to watch the movie trailer
Figure 1: Conceptual model 2.3.2 Intention to watch the movie
The intention to watch the movie is the second dependent variable in this study, we again question someone’s intention, this time the intention to watch a particular movie. We will also base this on the different movie attributes. This means that we look at how the different movie attributes affect someone’s intention to watch the movie. This variable is also based on the theory of reasoned action (Ajzen & Fishbein, 1969), which is also used to define the first dependent variable.
2.4 Conceptual Model
In figure 1 the conceptual model is shown, this model is based on the variables and the relationships between these variables discussed before.
Intention to watch the movie trailer Movie Genre (Product)
Ticket Price (Price) Actor Power (Product)
Critic Review (Promotion)
Distribution (Place)
Gender Mood Age
3. Study
3.1 Method & Measurement
According to a previous study by Gazley et al. there are 2 approaches with which researches examine motion picture success. First there is the approach which is called the psychological approach, using primary data from individual customers to find out why consumers really choose for a particular movie. The second approach, which is called the economical approach uses secondary data from industry trade sources to analyze the variables that influence the financial performance of a movie (Gazley et al., 2010). The economic approaches dominate the research field, because it is much easier to collect and work with secondary data. This study follows the first approach, or the psychological approach. We will use primary data by the means of a conjoint analysis to test the hypotheses.
This study will make use of a choice-base conjoint analysis (CBC), we are looking to find people’s preferences when it comes to movie trailer and movie watching intentions. A CBC analysis repeatedly asks consumers to choose their preferred product from a set of alternatives (Eggers & Sattler, 2011). Using this study we will eventually find the most important attributes for each consumer segment and the effects of these attributes on consumers’ intention to watch movie trailers and movies. So using the CBC analysis we plan to find the most important attributes for movie (trailer) watching intentions and the differences between the intention to watch a trailer and the intention to actually watch the movie.
3.1.1 Attributes & levels
For this study we use the attributes described previously. The first attribute, which is the genre of the movie, consists of five different levels, namely action, drama, horror/thriller, comedy and sci-fi/fantasy. The next attribute, which is actor power, consists of four levels ranging from unknown actors, C-star actors, B-star actors to star actors, which is based on the popularity of actors with A-star actors being the most popular. The next attribute in this study is ticket price which again has four levels, ranging from €8, €10, €12 to €14. Then we have the attribute critic review which also has 4 levels, and ranges from 2 out of 5 stars, 3 out of 5 stars, 4 out of 5 stars to 5 out of 5 stars. The last attribute Cinema consisting of three levels ranges from minor cinemas, major cinemas to all cinemas and describes the amount of cinemas the movie is displayed in.
3.1.2 Choice Design
The attribute levels described before are combined to create 12 different choice sets where participants have to make a choice between 2 different movies which are created with the attribute levels described in the previous section. To create balanced choice sets we make sure there aren’t any dominating choice sets where one choice would be clearly preferred over the other. And furthermore attribute levels appear in the different choice sets an equal number of times. So this means that for every choice set two movies are presented which are made up of the 5 attribute levels, next to that we present a third option indicating a no-choice decision.
3.1.3 Intention to watch the movie trailer and the movie
From the following options please state which movie trailer and which movie you would want to watch.
Option 1 Option 2 Option 3 Movie Genre: Drama Sci-Fi/Fantasy If these are my
choices I wouldn't pick any of these
movies. Actor Power B-star Actors A-star Actors
Ticket Price € 14 € 8
Critic Review: 2 out of 5 stars 5 out of 5 stars Cinemas the movie is played: All Cinemas Major
Cinemas I would want to watch the trailer of this movie.
I would want to watch this movie.
3.1.4 Mood, gender & age
For the moderating variables age and gender we simply ask the participants to state their age and gender. When it comes to mood we will use a scale ranging from 0 (very negative) to 100 (very positive) where we ask the participants to state their current mood. We stated that examples for a negative mood are being sad, uninterested or indifferent and examples for a positive mood are being happy, excited or motivated.
3.2 Data collection
This study will be performed in the Netherlands, we will use email and social media to find respondents for this choice-based conjoint analysis, which will eventually be mainly students and young people. The CBC-study will be designed as a survey and we will use my.preferencelab.com as the survey software to store all the data so we can use it for the following analysis. Next we will do the analyses we need by making use of the software programs SPSS and Latent Gold.
4. Results
4.1 Sample
First we look at some statistics regarding the respondents. We have 82 respondents who’ve completed the choice-based conjoint test , and with 217 people starting the test that means a completion rate of 37,8%. Of these respondents 45 are male (54,9%) and 37 are female (45,1%). Next the age of these respondents ranges from 18 to 56 with a mean age of 26 and with 80% of the respondents having an age between 18 and 27. And lastly the mood ranges from 20 to 100 and the mean mood of the respondents is 70 out of 100, also 82% of the respondents scored higher than 50, which means that most of the respondents are fairly more happy/positive or in a good mood.
Next we will look at the results of the choice-based conjoint tests. We first look at the outcomes using the first dependent variable, which is people’s intention to watch the trailer of the movie. Later we will look at the outcomes using the second dependent variable which is people’s intention to watch the movie itself and lastly we will look at how the proposed moderating variables affect both dependent variables.
4.2 Conjoint Analysis
For the conjoint analysis and estimation purposes we will make use of the multinomial logit model which follows the following formula.
( | ) ( ) ∑ ( )
Table 2: Prediction table Table 1: Model fit/statistics 4.2.1 Intention to watch the movie trailer
First we will discuss the model fit and prediction power. Table 1 shows the most important outcomes concerning the model. These results show us that, looking at the Chi-Square, that this model outperforms the NULL-model, p(760,2927, df 16) <,001. So this means that the parameters in the model are significantly different from zero and the model fit is good. Next in table 2 we can see the observations and predictions of this model, we calculate the hit rate of the model to be (236 + 251 + 3/984) 0,50, or 50%. This means that this model predicts 50% of the respondents’ observations correctly, which is better than the 33,3% the NULL-model would have.
Knowing that this model has a good fit and predicts relatively
well we will now look at the results concerning the effects of the different attributes and the importance of the attributes and the different levels within these attributes.
What we can see from the results provided in table 3 is first the fact that the effects of movie genre (p=,000), actor power (p<,05) and
the critical review (p=,000) are all significant and the effects of the ticket price (p=,51) and the cinema (p=,24) are not significant at the
95% confidence level. This is also consistent with the importance of the attributes, we namely see that the attributes which are highly significant also are the most important for people, namely the movie genre (36,2%) and critic review (32,4%) are by far the most important for people when considering the movie trailer. Followed by the actor power with 17,5% and lastly cinemas (7,3%) and the ticket price (6,7%). This is also consistent with the predictions made, because we wouldn’t expect the ticket price of a movie and the cinemas where the movie would be played to have any effect on someone’s preference for a particular movie trailer.
Model Statistics Respondents 82 Parameters 16 Df 16 Choice Sets 12 Log-Likelihood Statistics Log-Likelihood (LL) -983,9673 BIC 2029,6287 AIC 1995,9436 AIC3 2009,9346 CAIC 2043,6287 LL(0) -1081,034 Chi-Square 194,1344 Model Fit R² 0,0898 R²(Adj) 0,0750
Prediction Table Estimated
Observed 1 2 3 Total
1 236 148 1 387
2 146 251 3 398
3 105 91 3 199
Table 3: Attribute effects/Importance
Next if we look at the differences between the levels of the attributes we see, starting at the movie genre, that trailers for comedy movies are mostly preferred, followed by action and sci-fi/fantasy movie trailers and we even see negative effects for drama and horror/thrillers. This means that people mostly prefer to watch trailers of comedy movies and have the lowest preference for watching trailers of horror/thriller movies. When it comes to actor power we see that people prefer watching trailers with A-star actors the most and preferences decline with how lower the popularity of the actor gets, except for unknown actors, we namely see a remarkable effect of this attribute level because it is actually more preferred than B- and C-star actors. So people prefer A-star actors the most, but actually like unknown actors more than actors with a lower popularity than A-star. When it comes to critic review, people really prefer watching movie trailers which score 5 out of 5 stars by a famous critic and preferences again decline with how lower the stars get.
Attributes Class1 Wald p-value Importance
4.2.2 Intention to watch the movie
We will again start with looking at the model fit, some statistics and prediction power. From table 4 we can see that this model also performs way better than the NULL-model, the Chi-Square statistic namely shows that p(760,2927) < ,001. Furthermore all Log-Likelihood statistics are all lower than the previous model, meaning that this model has an even better fit and explanation power.
This is also what table 5 shows us. Because for the model looking at the intention to watch the movie itself the hit rate is even higher, namely (100 + 96 + 318)/984 is 52,5%. This means that we can predict choices a lot better than the NULL-model, which would have a hit rate of 33,3%. So looking at someone’s intention to watch a movie we see that we have a better model fit and prediction power than looking at someone’s intention to watch the movie trailer.
Next we will again look at the effects, significance and importance of the different attributes in this study. Table 6 shows these results. The first thing we notice is that the movie genre, actor power and the critic review are all highly significant (p=,000). Furthermore the ticket price is significant at the 90% confidence interval (p<,10). And lastly the cinema is not significant (p=,24). When we look at the importance of the attributes we now see that not the movie genre is the most important factor, as it was with people’s intention to watch the movie trailer, but the critic review is by far (39,2%) the most important factor. Followed by the critic review are movie genre (26,6%), actor power (18%) ticket price (12,8%) and lastly the cinema (3,5%).
Model Statistics Respondents 82 Parameters 16 Df 16 Choice Sets 12 Log-Likelihood Statistics Log-Likelihood (LL) -959,319 BIC 1980,3321 AIC 1946,638 AIC3 1960,638 CAIC 1994,3321 LL(0) -1081,0345 Chi-Square 243,430984 Model Fit R² 0,1126 R²(Adj) 0,0978
Prediction Table Estimated
Observed 1 2 3 Total
1 100 31 142 273
2 30 96 143 269
3 54 70 318 442
Total 184 197 603 984
Table 4: Model fit/statistics
Table 6: Attribute effects/importance
We see that if we look at the differences between the levels within the attributes that again for the movie genre, comedy movies are by far the most preferred, followed by a slight positive effect of action movies. Furthermore we see that there are negative effects of sci-fi/fantasy and drama movies, but which are really small, and a huge negative effect of horror/thriller movies, so people really dislike watching horror/thriller movies. When it comes to the actor power and critic review we see the same effects as with the previous model, namely that A-star actors are really preferred over the other types of actors and preferences decrease as the popularity of the actors decreases. With the critic review we see the highest preference for movies with 5 out of 5 stars, and this also declines with how lower the number of stars gets.
Then we see a small negative price effect (-0,077). This means that when the price increases, people’s intention to watch a movie decreases with a rather small amount, but still significant at the 90% confidence interval (p<,10). Again the cinema is not significant. A last remarkable observation is
Attributes Class1 Wald p-value Importance
4.3 Mood
We will next look at the effects, and if there are any, of the moderating variable. The moderating variable we will look at is someone’s mood, which ranges from 0 (negative) to 100 (positive) as also explained before. As we did with the two aggregate models we will first look the log-likelihood statistics to see if the model including the moderating variable mood performs better than the aggregate models. Table 7 shows the log-likelihood statistics of the model including the moderator mood for both the dependent variables.
Intention to watch the movie
trailer
Intention to watch the
movie
Log-likelihood Statistics Log-likelihood Statistics
Log-likelihood (LL) -975,3173 Log-likelihood (LL) -948,7888
BIC (based on LL) 2100,4631 BIC (based on LL) 2047,406
AIC (based on LL) 2018,6346 AIC (based on LL) 1965,5775
AIC3 (based on LL) 2052,6346 AIC3 (based on LL) 1999,5775 CAIC (based on LL) 2134,4631 CAIC (based on LL) 2081,406
Looking at the log-likelihood (LL) we see that for both dependent variables the LL when including the moderating variable mood the LL becomes lower than the aggregate models. This lower LL stems only from the fact that there are more variables included in the model with the moderating variables. When we want to know which model has the best fit we need to look at the Bayesian Information Criterion (BIC). We see that when the moderating variables are included the BIC values are 2100,4631 and 2047,406 for both dependent variables respectively. Looking at the BIC values of the aggregate models we see that they are 2029,6287 and 1980,3321 respectively. So this means that including the moderating variables of mood does not increase model fit and therefore we will leave them out of the model for future model calculations.
While we will not use the moderator for future model calculations we still want to see the effects of the moderator and if there would be any significant moderating effects. Therefore we will show in table 8 the outcomes for both dependent variables when including the moderating variables.
Intention to watch the movie
trailer Intention to watch the movie
Attributes Class1
p-value Attributes Class1
p-value Mood*Action 0,0104 0,046 Mood*Action 0,011 0,068 Mood*Drama -0,009 0,097 Mood*Drama -0,0087 0,15 Mood*Horror/Thriller -0,0009 0,87 Mood*Horror/Thriller 0,0031 0,63 Mood*Comedy 0,0103 0,039 Mood*Comedy 0,0145 0,0096 Mood*Sci-fi/Fantasy 0,0007 1 Mood*Sci-fi/Fantasy 0,0021 1 Mood*UnknownActors -0,0094 0,035 Mood*UnknownActors -0,0014 0,8 Mood*C-starActors 0,0023 0,61 Mood*C-starActors -0,0063 0,23 Mood*B-starActors 0,0059 0,17 Mood*B-starActors 0,0011 0,82 Mood*A-starActors 0,001 1 Mood*A-starActors -0,0006 1 Mood*8€ -0,0017 0,61 Mood*8€ 0,0049 0,21 Mood*10€ -0,0003 0,83 Mood*10€ 0,0011 0,53 Mood*12€ 0,0009 0,54 Mood*12€ -0,0006 0,75 Mood*14€ 0,0004 1 Mood*14€ 0,0003 1 Mood*2outof5stars -0,0007 0,88 Mood*2outof5stars -0,0013 0,82 Mood*3outof5stars 0,0035 0,41 Mood*3outof5stars -0,0008 0,88 Mood*4outof5stars 0,0001 0,99 Mood*4outof5stars 0,0106 0,032 Mood*5outof5stars 0,0007 1 Mood*5outof5stars 0,0021 1 Mood*AllCinemas -0,0012 0,73 Mood*AllCinemas -0,0075 0,066 Mood*MajorCinemas -0,003 0,37 Mood*MajorCinemas -0,0016 0,68 Mood*MinorCinemas -0,001 1 Mood*MinorCinemas -0,0007 1
The highlighted variables in table 8 are the variables that are significant at the 90% confidence interval level. So looking at the dependent variable intention to watch the movie trailer, we see that mood has a significant moderating effect on action, drama and comedy movies and on having unknown actors. Looking at action movies we see that when people are in a more positive mood their preference for actions movie trailers goes up, this effect is the same for comedy movie trailers. Looking at drama movies we see that when people are in a more negative mood their preference for drama movie trailers goes up. And lastly for the actor power we see that people in a more positive mood have lower preferences for unknown actors when looking at a movie trailer.
Next when looking at the second dependent variable, the intention to watch the movie, we see that people in a more positive mood have higher preferences for watching action and comedy movies, have higher preferences for movies with a rating of 4 out of 5 stars and surprisingly have lower preferences for movies that are played in all cinemas.
So there are some significant effects of the moderator mood, but these effects are all really small and only just significant. Furthermore as state before the model fit does not increase significantly when including the moderating variables of mood and therefore we will drop the variable from any future model calculations.
4.4 Hypotheses
Next we will look at the hypotheses stated previously and which of these hypotheses we can confirm based on the results.
H1a. Someone’s intention to watch a particular movie trailer will vary for different genres. The first hypothesis is confirmed based on the results, we namely see a significant effect of the movie genre on movie trailer watching intention (p=,000), and these effects differ for the different genres provided.
H1b. Someone’s intention to watch a particular movie will vary for different genres. The second hypothesis is also confirmed, there namely is again a highly significant effect of the movie genre on someone’s intention to watch the movie (p=,000) and again the effects differ between the different movie genres.
H2a. Someone’s intention to watch a movie trailer is higher when the actors in the movie are highly popular.
This hypothesis is also confirmed, we namely see a highly significant effect of actor power on someone’s intention to watch the movie trailer (p=,001) and this effect increases when the popularity of the actor increases.
H2b. Someone’s intention to watch a movie is higher when the actors in the movie are highly popular. Also hypothesis 2b can be confirmed, as with hypothesis 2a we see the same effect. A highly significant effect of the popularity of the actor on someone’s intention to watch the movie (p=,000) and this effect increases when the popularity of the actor increases.
H3b. Someone’s intention to watch a movie is higher for a movie with a low ticket price than for a movie with a high ticket price.
Again we can confirm a hypothesis, although this effect is slightly less significant than the other effects. We namely see a negative significant effect of ticket price on someone’s intention to watch the movie (p=,08). So increasing the price decreases someone’s intention to watch the movie.
H4a. Someone’s intention to watch a movie trailer is lower when the critical review about this movie is negative than when the review is positive.
Next we can confirm hypothesis 4a, there is a highly significant effect of the critic review on someone’s intention to watch the movie trailer (p=,000), and this effect increases when the number of stars the critic gives the movie increases.
H4b. Someone’s intention to watch a movie is lower when the critical review about this movie is negative than when the review is positive.
Also hypothesis 4b can be confirmed, as with the previous hypothesis we see that there is a highly significant effect of a critic review on someone’s intention to watch the movie (p=,000), and again this effect increases when the number of stars the critic gives the movie increases.
H5a. Someone’s intention to watch a movie trailer is not affected by the number of places where the trailer is shown.
Hypothesis 5a can be confirmed, we namely see that the effect of the number of places the movie is shown has a non-significant effect on someone’s intention to watch the movie trailer (p=,24). So the number of places the movie is shown has no effect on someone’s intentions to watch the movie trailer.
H5b. Someone’s intention to watch a movie is higher the higher the number of places where the trailer is shown.
Figure 3: Utility Scores -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 Acti o n Dram a H o rro r/T h rille r C om ed y Sc i-fi/Fan ta sy U n kn o w n Ac to rs C-st ar Ac to rs B-sta r Ac to rs A-sta r Acto rs Price 2/ 5 Sta rs 3/5 St ar s 4/5 St ar s 5/5 St ar s All C in em as Ma jo r Ci n e m as Min o r Ci n e m as Uti lity Sc o re
Intention to watch the trailer Intention to watch the movie
H6a. The mood someone is in will have an effect on the relationship between the different movie attributes and the intention to watch a movie trailer.
Hypothesis 6a can only partly be accepted, and mostly rejected. While there does seem to be a small moderating effect of someone’s mood on some of the movie genre levels and on one of the actor power levels, mood further has no significant moderating effects on the relations between the different attributes and the intention to watch a movie trailer.
H6b. The mood someone is in will have an effect on the relationship between the different movie attributes and the intention to watch a movie.
For hypotheses 6b we can say the same as for hypotheses 6a, namely there are a few attribute levels mood has a small moderating effect on, like some movie genre levels as well as one of the critic review and one of the cinema levels. But mostly there are no significant moderating effects of someone’s mood on the relationship between the movie attributes and the intention to watch the movie.
4.5 The No-choice option
To see what effect the no-choice option has and how strong this effect is we will look at the average choice share for a random movie trailer and movie and compare them to the no-choice option. First we will look at the movie trailer. We are going to look at an action movie, with B-star actors, a ticket price of €12, which got 3 out of 5 stars and only plays
in major cinemas. Table 9 shows the utilities for these different attributes and the no-choice option for both watching the movie trailer and watching the movie. We can see from this table that there is a huge difference between either watching the movie trailer or watching the movie. If we put together this random set of attributes the choice share for watching the movie trailer will be 70,4% while for watching the actual movie the choice share has immensely dropped to about 16%. So there is a huge difference between people’s preferences when either watching a movie trailer, or actually attending the movie in the cinema.
4.6 Willingness-to-pay
To get a more extended view on how big the differences are between the levels of the different attributes we will look at the willingness-to-pay (WTP) of the attribute levels. We will do this by dividing the utilities of the different attribute levels by the price utility. We will only look at the WTP for our second dependent variable, intention to watch the movie, because only for this variable the effect of price is significant and thus gives us more practical WTP results.
As we can see from table 10 the WTP for comedy movies is by far the highest (6,15) so people are willing to spend much more money on a comedy movie than
the other movie genres. They are willing to spend more money on action movies also, but when it comes to all the other genres people are definitely not willing to spend more money, and this effect is even really low for horror/thriller movies with a WTP level of -6,30.
Intention to watch the movie trailer
Intention to watch the movie
Attribute Utility Utility
Action 0,1925 0,1241
B-star Actors -0,0703 -0,1608
Ticket Price(12€) 0,306 -0,9348
3 out of 5 stars -0,1941 -0,2718
Major Cinemas 0,0694 0,0254
Total utilty sum 0,3035 -1,2179
Table 11: Willingness-To-Pay Actor Power
Table 12: Willingness-To-Pay Critic Review
Table 13: Willingness-To-Pay Cinema Table 11 shows the WTP for the actor power, here we see that
people are willing to pay the most for A-star actors (6,40) and for all other actors the WTP is below 0.
Table 12 shows the WTP for the critic review, here we see huge differences, people are willing to pay a lot more for movies with either 4 out of 5 stars (5,89) or 5 out of 5 stars (7,97) and really don’t want to pay any money for movies with only 2 out of 5 stars (-10,38).
Lastly table 13 shows the WTP for the cinema. We see that people are willing to pay the most for the minor cinemas (0,65) and the least for movies which play in all cinemas (-0,98), but the WTP’s are really small, and the variable has no significant effect on people’s intention to watch the movie.
4.7 Segmentation
Now that we know the main effects of all the different movie attributes we will also look if there are any segments to be found within the data for both the dependent variables. To do this we will use latent gold and perform a preference based segmentation with latent classes, which is an advanced segmentation process(Vermunt & Magidson, 2002). To find the optimal amount of classes, or segments, we will compare different model fit criteria and again look at the amount of classes with the lowest Bayesian Information Criterion (BIC) which will be the optimal model and amount of segments. We will compare the outcomes of the model fit for differentiating one to six classes and find the optimal amount of classes/segments. For segmentation we make use, as described before, of the age and the gender of the respondents. So based on these personal characteristics and personal preferences we do expect to find segments, but because the sample size is not that big (N=82) we do not expect to find a lot of different segments. Lastly because we found no improvements of model fit when including the moderator mood we will leave this moderating effect out and look for segments based on the aggregate models. Table 14 shows the outcomes of the latent class analysis when it comes to model fit for both the intention to watch the movie trailer as the intention to watch the movie.
What we can see from table 14 is that when we look at the intention to watch the movie trailer the BIC-score is lowest at a 3-class model, meaning that there would be 3 different segments in this model, and as for the intention to watch the movie itself we find the lowest BIC-score at a 2-class model, so leading to only 2 different segments for the second dependent variable. We could also look at the CAIC and classification error for deciding on the number of segments. But when we first look at the classification error we see that it is very low for all segments and differences are really small that they are almost negligible. For the CAIC we see that for the intention to watch the movie it also is the lowest at 2 segments. And for our first dependent variable it is the lowest for 2 segments, but fairly close to the 3 segments class. So we stick with the 3 and 2 class segmentations concluded from the BIC scores.
Next we will discuss how the different segments look like for both the dependent variables, so how are there preferences for the different attributes, which are the most important and who the people in those segments are.
Intention to watch the movie trailer LL BIC(LL) AIC(LL) AIC3(LL) CAIC(LL) Class.Err. R²(0) Model1 1-Class Choice
-983,9673 2029,6287 1995,9346 2009,9346 2043,6287 0 0,0986 Model2 2-Class Choice
-921,5303 1979,6689 1905,0606 1936,0606 2010,6689 0,0341 0,1957 Model3 3-Class Choice
-879,9452 1971,4129 1855,8903 1903,8903 2019,4129 0,0311 0,2711 Model4 4-Class Choice
-846,2303 1978,8974 1822,4607 1887,4607 2043,8974 0,0269 0,3362 Model5 5-Class Choice
-823,5434 2008,4377 1811,0867 1893,0867 2090,4377 0,0249 0,353 Model6 6-Class Choice
-798,8343 2033,9337 1795,6685 1894,6685 2132,9337 0,0248 0,3975 Intention to watch the movie LL BIC(LL) AIC(LL) AIC3(LL) CAIC(LL) Class.Err. R²(0) Model7 1-Class Choice -959,319 1980,3321 1946,638 1960,638 1994,3321 0 0,1153 Model8 2-Class Choice
-874,9006 1886,4096 1811,8013 1842,8013 1917,4096 0,0376 0,2466 Model9 3-Class Choice -839,977 1891,4765 1775,9539 1823,9539 1939,4765 0,0468 0,3101 Model10 4-Class Choice
-811,6108 1909,6583 1753,2215 1818,2215 1974,6583 0,0466 0,3587 Model11 5-Class Choice
-781,8225 1924,9959 1727,6449 1809,6449 2006,9959 0,0206 0,389 Model12 6-Class Choice
Table 15: Segments
4.7.1 Segments for intention to watch the movie trailer
First we will say something about the segments themselves, like the medium age, the percentages male/female and the total number of people in the segment, table 15 shows these outcomes.
So what we see is that segment 1 is the largest, with a medium age of 22,4 years and are mostly females (60%). Segment 2 is a bit smaller with a higher medium age of 29,8 years and are predominantly males (76%). And lastly the smallest segment is segment 3 with an almost the same medium age as segment 2 of 29,9 years and has slightly more males than females (55% Vs. 45%). Next table 16 will show the different outcomes for these segment when it comes to attribute effects and importance.
The first thing we notice is that when it comes to the
significance of the different attributes these are the same as with the aggregate model, so the movie genre, actor power and critic review are all significant at the 99%
confidence interval level whereas the ticket price and the cinema are not significant. Looking now at the movie genre we see that segment 1 has the highest preference for sci-fi/fantasy movie trailers, segment 2 has the highest preference for comedy movie trailers and segment 3 has the
highest preference for horror/thriller movie trailers. Looking at actor power we see that segment 1 has the highest preference for C-star and unknown actors, segment 2 has the highest preference for unknown actors and segment 3 really has the highest preference for A-star actors.
Attributes Class1 Class2 Class3 Wald p-value
MovieGenre Action 0,2100 0,7828 -0,3246 105,5226 0,00 Drama -0,2310 -0,3770 -0,6175 Horror/Thriller -0,8429 -0,6191 0,8821 Comedy 0,2741 0,8926 -0,1549 Sci-Fi/Fantasy 0,5898 -0,6793 0,2150 ActorPower Unknown Actors 0,0364 0,3112 -0,4568 29,4740 0,00 C-star Actors 0,0442 -0,3888 -0,4430 B-star Actors -0,0788 -0,0484 -0,3026 A-star Actors -0,0018 0,1260 1,2023 TicketPrice 0,0672 -0,1527 -0,0792 4,9085 0,18 CriticReview 2 out of 5 stars -0,0906 -0,1715 -1,7107 74,5013 0,00 3 out of 5 stars -0,2486 0,0092 -0,4264 4 out of 5 stars 0,4216 -0,1184 0,5358 5 out of 5 stars -0,0823 0,2807 1,6012 Cinemas All cinemas -0,0378 0,1107 0,1645 7,5032 0,28 Major Cinemas -0,0257 0,2190 0,0622 Minor Cinemas 0,0635 -0,3298 -0,2267 None_option -1,6622 0,2206 -1,0732 52,9901 0,00
Segment 1 Segment 2 Segment 3
Number of people 37 25 20
Medium Age 22,4 29,8 29,9
% Male/Female 40%/60% 76%/24% 55%/45%
Table 18: Overview segments
Table 17: Attribute importance When it comes to the critic review we see somewhat of the same results, every segment has either the highest preference for 4- or 5 out of 5 stars critic reviews.
Next we will look at the importance of the different attributes for the different segments, table 17 shows how important the different attributes are.
So we see that for segment 1 the movie genre is by far the most important and the critic review is somewhat important, segment 2 also has the movie genre as the highest importance with all other attributes being somewhat at the same level. For segment 3 the critic review is the most important followed by almost the same importance for the movie genre and the actor power.
Table 18 shows an overview of how the different segments look like and what their preferences are and what their most important attributes are. The numbers between brackets indicate the importance of the different attributes, with 1 being the most important and 3 being the least important.
So as we can see segment 1 are young females, with a preference for sci-fi/fantasy movie trailers with C-star actors and which have a critic rating of 4 out of 5 stars and this segment values the movie genre as the most important attribute. Segment 2 are slightly older males with a preference for comedy trailers, starring unknown actors and having a critic rating of 5 out of 5 stars and valuing the genre as the most important attribute. And finally segment 3 is a mixed group and are the oldest people, they have a preference for horror/thriller movie trailers with A-star actors and receiving a
Segment 1 Segment 2 Segment 3
MovieGenre 56,6% 42,1% 21,1%
ActorPower 4,9% 18,8% 23,4%
TicketPrice 8,0% 12,3% 3,3%
CriticReview 26,5% 12,1% 46,7%
Cinemas 4,0% 14,7% 5,5%
Segment 1 Segment 2 Segment 3
Medium Age 22,4 29,8 29,9
Gender Mostly Female Mostly Male Mixed
Table 20: Attribute effects
4.7.2 Segments for intention to watch the movie
For the second dependent variable we found not three but two segments at the optimal model. We will again start with some general information about the people within these two segments, table 19 shows this
information. So we can see that segment 1 is almost twice as big as segment 2. Furthermore segment 1 consists of mostly men (60%) with a medium age of 27,3 years, whereas segment 2 consists of mostly females with a medium age of 24,9 years. But the percentages are quite close as are the medium years so we don’t really have clear distinctions between the different segments.
Therefore we will now look at the different attribute effects for both of these segments, table 20 shows the different attribute effects and significances for both segments.
We again start with the significances, as with the aggregate model the movie genre, actor power and critic review are again significant at
the 99% confidence interval, the cinema is again non-significant but this time the ticket price is also insignificant, which is different from the aggregate model. Looking at the different attributes we see that both segments highly prefer comedy movies, but for segment 1 there is also quite a preference for action movies whereas for segment 2 there is also quite a preference for drama movies. When it comes to the actor power we again see the same thing, namely that both segments really highly prefer movies with A-star actors. And lastly also for the critic review we see the same, both
segments really prefer movies with a rating of 5 out of 5 stars.
Attributes Class1 Class2 Wald p-value
Table 22: Overview segments
Table 21: Attribute importance So far the differences between the two segments are really
small, therefore we will next look at how important the attributes are for the two segments.
Table 21 shows the difference in importance.
So here we finally see the big difference between the 2 segments. Segment 1 namely lies the most importance on
the critic review, whereas segment clearly beliefs the movie genre is the most important.
So lastly for the segmentation part of this study table 22 shows an overview of the two segments when it comes to the intention to watch the movie.
So while there are not really big differences between the two segments in terms of their most preferred attribute levels we see that segment 1 is slightly older and are mostly males, whereas segment 2 is slightly younger and are mostly females. Furthermore segment 1 beliefs the critic review is the most important and the genre the least important whereas segment 2 really beliefs the movie genre is of the highest importance.
Segment 1 Segment 2
Medium Age 27,3 24,9
Gender Mostly Male Mostly Female
Movie Genre Comedy/Action (3) Comedy/Drama (1) Actor power A-star Actors (2) A-star Actors (3) Critic review 5 out of 5 stars (1) 5 out of 5 stars (2)
5. Conclusion & Discussion
5.1 Main effects
The primary objective of this study was to find the most important factors for consumers when they choose which movie (trailer) to watch. So what specific attributes of a movie trailer or movie have the most impact on consumer choice, and what are the differences between the trailer and the movie itself, while also looking at differences in mood between these consumers. Based on distinguished academic literature hypotheses with regard to different movie attributes were formulated and almost all of these hypotheses could be confirmed with high significance levels.
First we will discuss the intention to watch the movie trailer, which was the first dependent variable we investigated. Doing the analysis we found that the attributes movie genre, actor power and critic review all had significant effects on people’s intention to watch the movie trailer, while the ticket price and the cinema had no significant effects. This was of course expected due to the fact that when only watching the movie trailer there will be no ticket price to pay or cinema to watch the trailer in, which relates back to H3a and H5a.
Looking at the three different variables that did have a high significant effect on people’s intention to watch the movie trailer we saw that the movie genre was of highest importance, closely followed by the critic review and then the actor power. When watching a movie trailer people are most influenced by the genre of the movie and critic reviews, while the popularity of the actor is of a far less influence. Knowing that genre is of the highest importance we looked at the differences between the genres we examined. We found that looking at the genre comedy is by far the most preferred genre, followed by action and sci-fi/fantasy movie trailers which were really close together, and we even saw negative preferences for drama and horror/thriller movie trailers. People really like to watch trailers of comedy movies and somewhat trailers of action and sci-fi/fantasy movies while they dislike watching trailers of drama and horror/thriller movies. We can relate these outcomes back to the first hypotheses which stated that there are indeed differences between the effects of movie genres on people’s intention to watch the movie trailer.
For the last attribute which had a significant effect on peoples intention to watch the movie trailer, which is the actor power, we see that people only prefer movie trailers with A-star actors, or the most popular actors, and when the actors displayed in the movie trailer are any lower than A-star actors the preference for watching the trailer drops, only A-star actors have a positive effect on the intention to watch the movie trailer, and the lower the popularity of the actors in the movie trailer gets, the lower the intention to watch the trailer gets, which was also displayed in H2a. So for people to have the highest intentions to watch a particular movie trailer it should be a comedy movie, with A-star actors and getting a critic review worth 5 out of 5 stars.
Next we will discuss the second dependent variable, namely people’s intention to watch the actual movie. The first relevant issue we notice is that while for watching the trailer the most important attribute was the movie genre, for watching the movie itself the most important attribute by far is the critic review. Followed at a large distance by the genre, actor power, price and with a tiny importance the cinema. We also noticed that while for this dependent variable price is just significant, the cinemas the movie is displayed in is again not significant at all and of a really low importance. So people really don’t care at all where the movie is displayed. Furthermore the first three most important attributes were again highly significant.
The outcomes we found regarding the genre of the movie are in really close line with one of the studies this study was based on, namely the study by Gazley et al. also showed significant differences between the genres, and showed that comedy was the most popular genre and horror was the least popular, or most disliked, genre (Gazley et al., 2011). Also the outcomes regarding the strong influences of actor power and critic reviews are in line with many other studies (Gazley et al., 2011; Dellarocas et al., 2007; Finsterwalder et al., 2012). While we didn’t find any significant effect of the cinemas the movie is displayed in this is again in line with previous research stating: ‘’Neither wide nor narrow distribution strategies have any influence on respondents…’’ (Gazley et al., 2011). While there wasn’t any research looking at prices specifically in the motion picture industry, the findings regarding price are in line with pricing theory stating that increasing prices cause people to have lower preferences for a specific product (Begg et al., 2003).
So when it comes to the motion picture industry we can state that an increasing price seems to have a negative effect on moviegoers’ preference for watching a movie, although this effect is really small and only barely significant. People don’t seem to care much about the price in relation to their much stronger preferences for the movie genre and critical reviews. We also see that people’s probabilities for watching a movie trailer are way higher than probabilities for watching the actual movie. Therefore movie trailers seem to be a very important tool for companies to persuade customers to watch a movie, by presumably increasing probabilities for customers’ intention to watch the movie itself after watching the movie trailer.
5.2 Moderating effects
Next, including this moderator in the model made the model fit and prediction power not significantly better than the model fit and prediction power of the aggregate model, therefore we don’t believe the small significant effects we found are of much impact, and therefore someone’s mood will probably have small to zero effect on people’s intentions to watch either the movie trailer or the movie when it comes to the different attributes we’ve included in the model.
5.3 Segmentation
When we did a segmentation with latent classes we found that for both dependent variables there was more than 1 segment when it comes to the respondents. When we investigated the intention to watch the movie trailer actually 3 different segments were found. The first one consisted of young females with a preference for sci-fi/fantasy movie trailers starring somewhat less popular actors and having a somewhat high critic rating. The second segment consisted of somewhat older males with a preference for comedy movie trailers and unknown actors with the highest critic rating. While the last group preferred horror/thriller movie trailers with A-star actors and also the highest critic rating. The first 2 groups considered the genre to be the most important attribute while the last group considered the critic rating to be the most important. When it comes to someone’s intention to watch a trailer it is seen that we can divide people based on their genre preference.
When looking at the intention to watch the movie actually only 2 segments were found, which were not that different from each other, what was different was that the first group were slightly more males and considered the critic rating the most important attribute, while the second group were slightly more females and considered the movie genre the most important attribute.
6. Managerial Implications
We base the implications on the marketing mix we’ve used in the study to decide on the movie attributes we included in our choice based conjoint analysis. We will give some advice on the 4 P’s as described by Porter, namely the price, product, place and promotion. Starting with the price, this attribute seems to have little to no effects when it comes to our study, the effects we found were small and the importance of this attribute is very low for our participants. So this means that customers don’t really seem to care that much about the movie ticket price, it should of course not exceed certain price thresholds, and movie theaters therefore could raise their ticket prices and definitely not decrease prices.
Next the place, this part of the marketing mix has no effects at all on movie (trailer) watching intentions, it also is of very low importance for the participants in our study. This means that where the movie is displayed doesn’t matter to customers, and movie producers/studios could therefore choose to sell the movie to those movie theaters that will be the most profitable for the movie producer/studio. Because where the movie is displayed is not important at all for customers.
Then we look at the product, and more specifically the genre and the actor power. Different genres have different significant effects on both the intentions to watch the movie trailer and the movie. When it comes to the trailer the genre is most important attribute and for the movie itself it is the second most important attribute. So the movie genre is really important when it comes to customer behavior in this motion picture industry. For movie theaters it would be really profitable to show as much comedy and action movies as possible, because these two movie genres result in the highest intentions for customers to go watch the movie. When it comes to the actor power movie
Lastly we get to the promotion part of the marketing mix. In our study we used a critic review to represent promotion. For movie trailer watching intentions this was the second most important attribute and for movie watching intentions it even was the most important attribute. Getting a good rating for a movie results in the highest intentions for people to watch a movie, and getting a bad rating results in really low intentions, because the effect of the critic review is really strong. So for movie producers/studios it is really important to get this high rating, this simply means making good movies is really important. So while the critic review is the most important factor for movie going intentions, and therefore really important for movie producers/studios, it also is the factor the movie producer/studio has the least control over. Thus making good movies and getting good critic reviews is of utmost importance for the profitability of a movie, and also being and remaining friends with the critics and review sites would be good thing.
A final note would be that movie producers/studios/theaters have to keep in mind that there are of course different groups of customers, especially when it comes to preferences regarding the
different movie attributes. While comedy seems to be the most preferred movie genre and the place the movie is shown doesn’t seem to be of any influence, there will be a small group of people who really prefers to watch horror movies in small movie theaters.
7. Limitations & Directions for further research
7.1 Limitations
For the segmentation part we only used age and gender to segment the respondents on, and again the sample size was relatively small, which could be the reason that we didn’t find clear distinctions between the segments that were found for peoples intention to watch the movie. Furthermore we used 12 choice sets for the respondents to choose their preferred movie (trailer) structure, while this is a pretty standard amount of choice sets, we used 2 dependent variables (intention to watch the trailer and intention to watch the movie), so basically people had to choose their preferred option 24 times, which could have led to potential fatigue effects stemming from monotonous answering and cause biased outcomes.
The last thing to discuss is that the effect of the no-choice option was really strong, especially for the intention to watch the movie variable, meaning that a lot of times people chose the no-choice option over the 2 alternatives provided, meaning that information regarding these other 2 alternatives, and therefore information regarding the other attributes, was lost (Eggers & Sattler, 2011).
7.2 Directions for further research
The main effects found in this study were all pretty clear and showed helpful outcomes when it comes to the importance and effects of the different movie attributes. But as discussed in the limitations there is clearly room for further research. First of all there will be probably be more moderating effects on the relationships between the different attributes and the two intentions, there is definitely some effect of peoples mood and using more items to measure mood would be a great starting point. Also looking at more moderators would be a nice option for further research. Next a larger sample size and more demographic information could really help doing a better segmentation resulting in more distinctive segments. Which would be a great extension to the basis laid here when it comes to importance and effects of different movie attributes?
