Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

The Journal of Arts Management, Law, and Society

ISSN: 1063-2921 (Print) 1930-7799 (Online) Journal homepage: http://www.tandfonline.com/loi/vjam20

The Demand for Cinema in Sweden: An Application


of Proportional Odds Model

Manuchehr Irandoust

To cite this article: Manuchehr Irandoust (2017): The Demand for Cinema in Sweden: An
Application of Proportional Odds Model, The Journal of Arts Management, Law, and Society, DOI:
10.1080/10632921.2017.1377658

To link to this article: http://dx.doi.org/10.1080/10632921.2017.1377658

Published online: 12 Oct 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=vjam20

Download by: [UNIVERSITY OF ADELAIDE LIBRARIES] Date: 13 October 2017, At: 07:28
THE JOURNAL OF ARTS MANAGEMENT, LAW, AND SOCIETY
https://doi.org/./..

The Demand for Cinema in Sweden: An Application of


Proportional Odds Model
Manuchehr Irandoust
Kristianstad University, Kristianstad, Sweden
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

ABSTRACT KEYWORDS
This article examines the determinants of demand for cinema using a Advertising; demand; film
randomly drawn sample of the movie-going population in Sweden. A industry; nonlinear;
proportional odds model is applied to capture the natural ordering of proportional odds model;
willingness to pay
dependent variables and any inherent nonlinearities. The findings show
that individual demand for cinema depends on gender, age, educational
attainment, income, marital status, critical reviews, word of mouth, and
willingness to pay. The fact that cinema demand is correlated with eco-
nomic and socio-demographic variables has important implications for
theoretical and empirical research in cultural economics and human
decision-making processes.

Introduction
Despite the increasing opportunities to see movies at home—via the Internet, through buy-
ing or renting movies, and not least by the large selection of movies on TV—interest in going
to the movies continues to increase in Sweden. According to Statistics in Sweden (Statistiska
Centralbyrån, SCB 2015), sixty-three percent of the population between sixteen and eighty-
four attended the cinema at least once during 2015. From that, about ninety percent of young
people between sixteen and twenty-four go to the cinema at least once a year; the proportion
gradually decreases with age so that, among those who are 65–74 years old, it is thirty-seven
percent. In higher ages, 75–84, twenty percent go to the cinema at least once a year. The avail-
ability of a diverse range of films naturally plays a certain role. In the large cities, the percentage
of moviegoers is also greater than in the rest of the country. Overall, the Swedish people went
to the cinema 17 million times in 2015. That’s about 1.7 cinema visits per inhabitant.1
While Fernandez-Blanco and Banos-Pino (1997) and Dewenter and Westermann (2005)
report that the number of filmgoers and the average attendance per person all dropped con-
siderably in Spain and Germany, respectively, the growth in cinema attendance in Sweden has
increased since 1995. The main objective of this article is to study the major factors driving
cinema attendance in Sweden.
Previous literature which examined demand for cinema or movies can be classified into
three groups. The first group studies the impact of film-specific determinants such as movie
stars, awards/nominations, the movie’s country of origin, and the popularity of directors (e.g.,
Albert 1998; Bagella and Becchetti 1999; DeVany and Walls 1999; Elberse and Eliashberg

CONTACT Manuchehr Irandoust manuchehr.irandoust@hkr.se; manuch.dost@gmail.com Kristianstad University,


College of Business Studies, Department of Economics and Finance, Elmetorpsvägen ,  Kristianstad, Sweden.
©  Taylor & Francis
2 M. IRANDOUST

2003; Walls 2005; Deuchert et al. 2005; Hennig-Thurau et al. 2006; Elberse 2007; d’Astous
et al. 2007; McKenzie 2009). The second group investigates the rational addiction hypothe-
sis (e.g., Cameron 1999; Yamamura 2009). The third group estimates demand function (e.g.,
Cameron 1990; Fernandez-Blanco and Banos-Pino 1997; MacMillan and Smith 2001; Hand
2002; Collins and Hand 2005; Dewenter and Westermann 2005; Kim 2009).
Most cinema demand studies have been conducted either by time-series analysis (cointe-
gration) or cross-sectional approach in the U.K., Spain, Korea, U.S., and Germany. Further-
more, they have not considered any nonlinearities in their estimations. To the best of the
author’s knowledge, there has not been any econometric study of cinema demand in Sweden.
This article aims to fill these gaps by modeling the Swedish cinema market. A proportional
odds model (POM) is used here, which appropriately captures the natural ordering of depen-
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

dent variables and any inherent nonlinearities (Agresti 2007).


The article is organized as follows. The next section describes theoretical considerations
and hypotheses; then I introduce data and methodology. The fourth section presents estima-
tion results, followed by a conclusion.

Theoretical considerations and hypotheses


There are several factors which influence cinema demand. The most important variables cited
in the literature are as follows.

Critical reviews
The influence of film reviews is outlined in many studies (e.g., Eliashberg and Shugan 1997).
It is suggested that critical reviews could act as opinion leaders because the reviewers are
considered more experienced and more knowledgeable regarding the quality of movies. The
impact of critical reviews has been found to be positive in many studies, including Elberse and
Eliashberg (2003), Basuroy et al. (2006), Hennig-Thurau et al. (2006), Boatwright et al. (2007),
Gemser et al. (2007), and Elliott and Simmons (2008). In a study conducted by Reinstein and
Snyder (2005), it has been shown that a critical review does not have a significant impact
on movie admissions when such reviews are revealed by television talk shows. However, the
following hypothesis (H1) can be proposed as a summary:
H1: Critical reviews have a significant effect on movie admissions.

Word of mouth (WOM)


Word of mouth (WOM) has been shown to have a significant effect on movie admissions.
Basuroy et al. (2006) measure WOM as the cumulative number of screens since a movie’s
release, finding a positive effect. Neelamegham and Chingagunta (1999) find no significant
results between weekly revenue and WOM measured as cumulative viewership. Elberse and
Eliashberg (2003) used the previous week’s average revenue per screen as a proxy for WOM
and report significant positive results. Liu (2006) proposes that the volume of WOM (from the
Yahoo movies website) offers significant explanatory power. WOM is more trustworthy than
advertising or critical reviews because it comes from other moviegoers. Duan et al. (2008)
also show that WOM has a significant impact on movie admission. DeVany and Lee (2001)
argue that WOM can be a credible means to share information about good and bad movies.
The second hypothesis (H2) is therefore proposed as follows:
H2: WOM has an important impact on movie admissions.
THE JOURNAL OF ARTS MANAGEMENT, LAW, AND SOCIETY 3

Education
Several studies have verified the important role of education in explaining cultural consump-
tion (Seaman 2006). As cinema is considered a cultural product, education can be relevant
to cinema demand (Frey 2000; Levy-Garboua and Montmarquette 2002; Rodondo and
Holbrook 2010). People with higher education levels have on average fever leisure activities
than those with lower education levels (Suominen 2013). At the same time, highly educated
people have a larger variety of leisure activities and therefore more variety in cultural con-
sumption (Ruuskanen 2004). Thus, a negative relationship between education and demand
for cinema is expected. The third hypothesis (H3) is therefore proposed as follows:
H3. Cinema attendance is negatively related to education.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

Income
Because cinema is usually considered a normal good (normal goods are any goods for
which demand increases when income increases and falls when income decreases), income
is a relevant explanatory variable. However, with rising income the variety of other affordable
options increases as well. Therefore, increasing opportunity costs (the opportunity cost is
the value of the choice of a best alternative cost while making a decision) of going to the
cinema, due to its time intensiveness, may lead to a reduction in consumption. However,
several studies show that there is a positive income elasticity of demand for cinema (e.g.,
Fernandez-Blanco and Banos-Pino 1997; Dewenter and Westermann 2005; Kim 2009), and
so we expect a positive relationship between income and demand for cinema. The fourth
hypothesis (H4) is therefore suggested as follows:
H4: Cinema demand is positively related to income.

Age
MacMillan and Smith (2001), Collins and Hand (2005), Yamamura (2008), and Rodondo and
Holbrook (2010) have indicated that young people go to the cinema more than any other age
group. Using cluster analysis with pooled data from 1998 and 2002, Lizardo (2006) reveals that
the following four genres comprise the “highbrow cluster”: arts consumption, ballet atten-
dance, theatre attendance, and classical music or opera concert attendance. The “lowbrow
cluster” consists of attending a popular live music concert, going to see a movie in the cinema,
or reading a novel, poem, or play. Younger people appear to favor lowbrow culture, whereas
older people favor highbrow culture. This implies that willingness to attend the cinema is
decreasing at an increasing rate as reported age increases. Thus, our fifth (H5) hypothesis
is:
H5: Cinema demand has a nonlinear negative relationship with reported age.

Gender
It is widely known that gender has an impact on cultural consumption (Kracman 1996;
Borgonovi 2004; Montgomery and Robinson 2006; Rodondo and Holbrook 2010).
Researchers report that single males prefer sports and single females prefer arts. DiMaggio
and Mukhtar (2004) find that feminization persists for all cultural activities, despite changes
in labor market participation and in traditional household roles.
4 M. IRANDOUST

Gray (2003) argues that the positive effect on participation of being a woman, even after
other variables are controlled for, can be explained in terms of early socialization on cultural
activities (more exposure to art, while males are socialized through sports). Warde and
Gayo-Cal (2009) find mixed evidence concerning the gender effect on cultural consumption
using British 2002–03 survey data. Women appear to be more active in “legitimate” culture.
Bourdieu defines legitimate as being related to dominant classes and powerful social groups
and being aesthetically the most valuable. However, the sixth hypothesis (H6) is suggested as
follows:
H6: Being female increases the probability of being a heavy cinema attendee.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

Willingness to pay (WTP)


Cameron (1990), Fernandez-Blanco and Banos-Pino (1997), Collins and Hand (2005), Walls
(2005), Dewenter and Westermann (2005), Yamamura (2008), Kim (2009), and Rodondo
and Holbrook (2010) have shown that an increase in admission price is negatively associated
with cinema demand. Davis (2002) estimates that the theatre price elasticity of demand is
approximately −3. The six theatres in the sample displayed different numbers of movies,
ranging from two to nine during a six-week period. Davis (2006) also presents similar con-
sumer price sensitivity results. In a study conducted by Dewenter and Westermann (2005),
the price elasticity is approximately −2.5 using German long-term (1950–2002) annual data.
Akerlof (1979) refers to issues that arise due to asymmetric information possessed by
the buyer and the seller of an investment or product regarding its value. In many markets,
sellers of a product are better informed about the good’s quality than buyers, resulting in the
well-known lemon problem. Because it is impossible for many consumers to distinguish high
quality from low quality, goods of each type sell for the same price. A growing body of liter-
ature has shown that the lemon problem is an important issue for experience and credence
goods for which quality cannot be ascertained before consumption. However, producers can
signal quality, and high quality can be signaled by advertising (Milgrom and Roberts 1986).
One behavioral explanation, supported by empirical evidence (e.g., Rao and Monroe 1988;
Stiving 2000) suggests that consumers infer information (e.g., quality) from price. Conse-
quently, price appears to play two opposite roles (allocative and informational) in consumer
demand: (1) higher price decreases consumer utility, because they must pay more for the
product; and (2) higher price may create higher-quality perceptions, which increase utility.
Intuitively, this complex relationship may lead to a non-monotonous (individual) utility
function over price, which then should induce an (aggregate) demand function that is not
necessarily downward sloped, as assumed ubiquitously in the literature. As a result, consumers
may be willing to pay above a certain level of price difference with the original ticket price if
they perceive that their utilities are still increasing. Therefore, a nonlinear relationship would
realistically account for the effects of price on demand. Thus, our seventh hypothesis (H7) is:

H7: Cinema attendees are willing to pay above a certain level of price difference with the existing
ticket price, but the effect is nonlinear.

Marital status
As culture activities are time-intensive, household variables such as marital status and family
responsibilities are likely to influence the rates of participation in various leisure activities.
THE JOURNAL OF ARTS MANAGEMENT, LAW, AND SOCIETY 5

Lee and Bhargava (2004) have investigated the different leisure activities of single and married
individuals. According to their research, married people are more likely to participate in
active sports, whereas single people prefer cultural activities. Our eighth hypothesis (H8) is:
H8: Being married or cohabiting decreases the probability of being a heavy cinema attendee.

Data and methodology

Data
The sample was drawn from individuals in Malmö, which is located in the south of Sweden.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

The data for our variables were collected from responses to a questionnaire during April
2016. A number of representative cinemas were chosen. A systematic stochastic sample of
210 individuals was chosen in these cinemas. The questionnaire was distributed to randomly
selected individuals who were ticket buyers. Face-to-face interviews were conducted during
the process.
Respondents were asked questions about their demand for cinema. Economic and socio-
demographic characteristics were also included. The observations on our dependent variable
take three different forms: individuals with heavy cinema attendance (HCA), individuals with
moderate cinema attendance (MCA), and individuals with low cinema attendance (LCA).
LCA means that a person goes to the movies less than five times per year; MCA indicates that
he or she is a moviegoer between five and ten times per year; and HCA represents a person
who is a moviegoer more than ten times per year. Participants are ranked according to their
willingness to attend the cinema.

Independent variables
The independent variables are: Income, which indicates the level of income among individu-
als (1 if they earn more than 25,000 SEK or 2700 Euros per month, 0 otherwise)2 ; Education,
which denotes the level of education among individuals (1 if they have completed higher
studies, 0 otherwise); Age, which indicates age of respondent; Age square, which is the square
of age and measures any nonlinearities; Gender, which shows gender among participants
(1 if the respondent is female, 0 otherwise); Marital status, which indicates the marital status
among respondents (1 if they are married legally or de facto, 0 otherwise); Critical reviews,
which shows critical reviews for a specific movie (1 if the respondent considers it in deciding
to attend, 0 otherwise); WOM, which indicates word of mouth regarding a specific movie (1
if the respondent pays attention to it in deciding to attend, 0 otherwise); WTP, which shows
an individual’s willingness to pay above the existing price for cinema tickets (the magnitudes
of the WTP); and WTP square, which is the square of WTP and measures any nonlinearities.

Descriptive statistics
The gender distribution across the sample was fifty-four percent female and forty-six percent
male. The age distribution was sixty percent (aged from sixteen to thirty years old), twenty
percent (aged from thirty to forty-five years old), fifteen percent (aged from forty-five to
sixty years old), and five percent (aged from sixty to eighty). With regard to the education
and income variables, thirty-three percent of respondents had higher studies at university
or college levels and fifty-eight percent of respondents had a monthly income below 25,000
SEK (2,700 Euros). The percentage of married/cohabiting people was thirty-two percent
6 M. IRANDOUST

of respondents. The percentage of those who consider critical reviews or WOM in their
decision-making process was forty-six percent and seventy-eight percent, respectively.
Regarding WTP, seventy percent were willing to pay above the existing ticket price and
the magnitudes of the WTP were ranged from ten percent up to 100 percent. The share of
low, medium, and high cinema consumers was forty-two percent, thirty-three percent, and
twenty-five percent, respectively. Appendix A, Table A1, shows the correlation matrix for the
explanatory variables.

Data analyses
Ordered response variables are very common in social science research. Before methods such
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

as ordinal logistic regression gained popularity in the 1980s and 1990s, researchers would
often estimate models with ordinal dependent variables using OLS regression. The unrealistic
assumption of equal spacing between the categories of ordinal variables (e.g., Likert scales
numbered 1–5) and the potential for misleading results eventually led researchers to consider
methods designed explicitly for ordinal outcomes (Long 1997). When the ordinal categorical
responses are treated as ordinal categorical variables and the frequencies of the categories of
the responses are of interest, proportional odds models are appropriate (McCullagh 1980).
Another advantage of proportional odds models is invariance to selection of response cate-
gories (Agresti 2007). This means that when nine categories and three categories which were
collapsed from the nine categories are used in parallel, similar conclusions will be reached
(Agresti 2007).
A proportional odds model (POM) is used here, which appropriately captures the natural
ordering of dependent variables and any inherent nonlinearities (Agresti 2007).3 This means
that not only is the effect of a regressor non-constant on the probability of a given attitude,
but also its influence varies across different attitudes. This model, which is based on a con-
strained ordinal model, has better statistical power than simple logistic regression (Capuano
2012). Furthermore, since our response (the observations on our dependent variable) is a
three-level ordinal variable, it is wise to consider the natural ordering to the response levels
when modeling the effects of the explanatory variables on the consumer behavior (Agresti
2007). Appendix B provides the description of the proportional odds model in more detail.

Estimation results
Table 1 reveals the relative importance of the explanatory variables outlined earlier. All
variables have been found to be statistically significant at conventional levels with expected

Table . Results of proportional odds model.


Variable Parameter Estimate p-value Odds Ratio

Constant  − . . —


Constant  − . . —
Gender . . .
Income . . .
WTP . . .
WTP square − . . .
Education − . . .
Marital status − . . .
Age − . . .
Age square − . . .
Critical reviews . . .
Word of mouth . . .
THE JOURNAL OF ARTS MANAGEMENT, LAW, AND SOCIETY 7

signs except education, which is not significant. The likelihood of being a heavy moviegoer
depends on age. Thus, young individuals are more likely to be heavy attendees but the effect
is nonlinear. This implies that willingness to attend the cinema is decreasing at an increasing
rate as reported age increases.
The influence of marriage decreases the likelihood of being a heavy cinema consumer.
Having a university education decreases the likelihood of being a heavy cinema attendee, but
the effect is not significant. Also, having a high income increases the likelihood of being a
heavy cinema attendee. The probability of being a heavy versus a moderate or low cinema
consumer is also determined by gender. Being female increases the probability of being a
heavy cinema consumer. The likelihood of being a heavy cinema consumer is also determined
by critical reviews and WOM. The effect of WOM is stronger than critical reviews on cinema
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

attendance. This implies that consumers place more trust in recommendations from other
consumers (e.g., word-of-mouth recommendations or blogs) than on paid advertising or
critical reviews. This also stems from the fact that young audiences are frequent cinema
attendees and they read fewer newspapers than older citizens. Finally, willingness to pay is
positive and significant, which shows that consumers are willing to pay above a certain level
of price difference with the existing ticket price, but the effect is nonlinear.
Note that Table 1 also shows odds ratios. For example, in the case of income and WOM,
the estimated odds ratios of being a heavy versus moderate or low cinema attendee are 3.136
and 4.554, respectively. As a test for the proportional odds assumption, we perform a score
test. The resulting p value is 0.387, implying that we cannot reject the null hypothesis of the
proportional odds assumption.

Conclusion
This study attempts to shed light on a number of determinants explaining demand for cinema
in Sweden. The novel contribution of this study stems from its used methodology, which takes
into account any nonlinearities in cinema demand. Based on the estimation of a proportional
odds model, the results indicate that the probability of being a heavy cinema consumer is
affected by marital status, age, income, gender, willingness to pay, critical reviews, word of
mouth, and education.
The fact that cinema demand is correlated with economic and socio-demographic variables
has important implications for theoretical and empirical research in industrial organization
and managerial economics. These traits are usually assumed to be relatively independent in
economic models and could be a potentially important source of model misspecification.
Results from this study provide some evidence and support to the notion that people can
generally be characterized by certain factors related to cinema demand. This knowledge can
help policymakers and planners when they discuss concepts related to cinema demand. One
limitation of this study stems from the fact that the sample is missing non-moviegoers.

Notes
1. The data come from SCB, Sweden (2015).
2. If the variable in question has an exactly linear relationship with the outcome, you do lose infor-
mation by making a continuous variable into a categorical one. Treating the variable as continuous
allows the linear component of the relationship to be estimated, but the categorical version allows
much more complicated relationships to be captured.
3. Gray (1995) also applies logit models to arts participation.
8 M. IRANDOUST

Acknowledgments
The author would like to thank two anonymous referees for valuable and helpful comments. Any
remaining errors are the author’s.

References
Akerlof, G. A. 1970. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.” Quar-
terly Journal of Economics 84 (3):488–500.
Albert, S. 1998. “Movie Stars and the Distribution of Financially Successful Films in the Motion Picture
Industry.” Journal of Cultural Economics 22 (4):249–70.
Agresti, A. 2007. An Introduction to Categorical Data Analysis, 2nd ed. New York, NY: John Wiley &
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

Sons.
Bagella, M., and L. Becchetti. 1999. “The Determinants of Motion Picture Box Office Performance:
Evidence from Movies Produced in Italy.” Journal of Cultural Economics 23 (4):237–56.
Basuroy, S., K. K. Desai, and D. Taluktar. 2006. “An Empirical Investigation of Signaling in the Motion
Picture Industry.” Journal of Marketing Research 63:287–95.
Boatwright, P., S. Basuroy, and W. Kamakura. 2007. “Reviewing the Reviewers: The Impact of Individual
Film Critics on Box Office Performance.” Quantitative Marketing and Economics 5 (4):401–25.
Borgonovi, F. 2004. “Performing Arts Attendance: An Economic Approach.” Applied Economics
36:1871–885.
Cameron, S. 1990. “The Demand for Cinema in the United Kingdom.” Journal of Cultural Economics 14
(1):35–47.
Cameron, S. 1999. “Rational Addiction and the Demand for Cinema.” Applied Economics Letters 6
(9):617–20.
Capuano, A. W. 2012. “Constrained Ordinal Models with Application in Occupational and Environ-
ments Health.” Ph.D dissertation, University of Iowa.
Collins, A., and C. Hand. 2005. “Analyzing Moviegoing Demand: An Individual Level Cross-Sectional
Approach.” Managerial and Decision Economics 26 (5):319–30.
d’Astous, A., F. Colbert, and V. Nobert. 2007. “Effects of Country-Genre Congruence on the Evaluation
of Movies: The Moderating Role of Critical Reviews and Moviegoers’ Prior Knowledge.” Interna-
tional Journal of Arts Management 10 (1):45–51.
Davis, P. 2002. “Estimating Multi-Way Error Components Models with Unbalanced Data Structures.”
Journal of Econometrics 106:67–95.
Davis, P. 2006. “Spatial Competition in Retail Markets: Movie Theaters.” Rand Journal of Economics 37
(4):964–82.
Deuchert, E., K. Adjamah, and F. Pauly. 2005. “For Oscar Glory or Oscar Money?” Journal of Cultural
Economics 29:159–76.
DeVany, A., and W. D. Walls. 1999. “Uncertainty in the Movie Industry: Does Star Power Reduce the
Terror of the Box Office?” Journal of Cultural Economics 23 (4):285–318.
DeVany, A., and C. Lee. 2001. “Quality Signals in Information Cascades and the Dynamics of the Distri-
bution of Motion Picture Box Office Revenue.” Journal of Economic Dynamics and Control 25:593–
614.
Dewenter, R., and M. Westermann. 2005. “Cinema Demand in Germany.” Journal of Cultural Economics
29:213–231.
DiMaggio, P., and T. Mukhtar. 2004. “Arts Participation as Cultural Capital in the United States, 1982–
2002: Signs of Decline?” Poetics 32:169–94.
Duan, W., B. Gu, and A. B. Whinston. 2008. “The Dynamics of Online Word-of-Mouth and Product
Sales: An Empirical Investigation of the Movie Industry.” Journal of Retailing 84 (2):233–42.
Elberse, A. 2007. “The Power of Stars: Do Star Actors Drive the Success of Movies?” Journal of Marketing
71 (4):102–20.
Elberse, A., and J. Eliashberg. 2003. “Demand and Supply Dynamics for Sequentially Released Prod-
ucts in International Markets: The Case of Motion Pictures.” Marketing Science 22 (3):329–
54.
Eliashberg, J., and S. M. Shugan. 1997. “Film Critics: Influencers or Predictors.” Journal of Marketing 61
(2):68–78.
THE JOURNAL OF ARTS MANAGEMENT, LAW, AND SOCIETY 9

Elliott, C., and R. Simmons. 2008. “Determinants of UK Box Office Success: The Impact of Quality
Signals.” Review of Industrial Organization 33 (2):93–111.
Fernandez-Blanco, V., and J. Banos-Pino. 1997. “Cinema Demand in Spain: A Cointegration Analysis.”
Journal of Cultural Economics 21 (1):57–75.
Frey, B. S. 2000. Arts & Economics: Analysis & Cultural Policy. Berlin Heidelberg, Germany: Springer
Verlag.
Gemser, G., M. Van Oostrum, and M. Leenders. 2007. “The Impact of Film Reviews on the Box Office
Performance of Art House Versus Mainstream Motion Pictures.” Journal of Cultural Economics
31:43–63.
Gray, C. M. 1995. Turning On and Tuning In: Media Participation in the Arts. Research Division Report
#33, National Endowment for the Arts, Seven Locks Press, Washington DC.
Gray, C. M. 2003. “Participation.” In A Handbook of Cultural Economics, edited by R. Towse, 326–332.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

Cheltenham, UK: Edward Elgar.


Hand, C. 2002. “The Distribution and Predictability of Cinema Admissions.” Journal of Cultural
Economics 26 (1):53–64.
Hennig-Thurau-Thurau, T., M. B. Houston, and G. Walsh. 2006. “The Differing Roles of Success Drivers
Across Sequential Channels: An Application to the Motion Picture Industry.” Journal of the Academy
of Marketing Science 34 (4):559–74.
Kim, S. 2009. “Cinema Demand in Korea.” Journal of Media Economics 22 (1):36–56.
Kracman, K. 1996. “The Effect of School-Based Arts Introduction on Attendance at Museums and the
Performing Arts.” Poetics 24:203–18.
Lee, Y. G., and V. Bhargava. 2004. “Leisure Time: Do Married and Single Individuals Spend it Differ-
ently?” Family and Consumer Sciences Research Journal 32 (3):254–74.
Lévy-Garboua, L., and C. Montmarquette. 2002. The Demand for the Arts. Working Papers 2002s-10,
CIRANO, Montreal, Canada.
Lizardo, O. 2006. “The Puzzle of Women’s ‘Highbrow’ Culture Consumption: Integrating Gender and
Work into Bourdieu’s Class Theory of Taste.” Poetics 34:1–23.
Liu, Y. 2006. “Word of Mouth for Movies: Its Dynamics and Impact on Box Office Revenue.” Journal of
Marketing 70 (3):74–89.
Long, J. S. 1997. Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks,
CA: Sage Publications.
McKenzie, J. 2009. “Revealed Word-of-Mouth Demand and Adaptive Supply: Survival of Motion Pic-
tures at the Australian Box Office.” Journal of Cultural Economics 33:279–299.
MacMillan, P., and I. Smith. 2001. “Explaining Post-War Cinema Attendance in Great Britain.” Journal
of Cultural Economics 25 (2):91–108.
McCullagh, P. 1980. “Regression Models for Ordinal Data.” Journal of the Royal Statistical Society Series
B 42 (2):109–42.
Milgrom, P., and J. Roberts. 1986. “Price and Advertising Signals of Product Quality.” Journal of Political
Economy 94 (4):796–821.
Montgomery, S. S., and M. D. Robinson. 2006. “Take Me Out to the Opera: Are Sports and Arts Com-
plements? Evidence from the Performing Arts Research Coalition Data.” International Journal of
Arts Management 8 (2):24–39.
Neelamegham, R., and P. Chingagunta. 1999. “A Bayesian Model to Forecast New Product Performance
in Domestic and International Markets.” Marketing Science 18:115–36.
Rao, A. R., and K. B. Monroe. 1988. “The Moderating Effect of Prior Knowledge on Cue Utilization in
Product Evaluations.” Journal of Consumer Research 15 (2):253–64.
Redondo, I., and M. B. Holbrook. 2010. “Modeling the Appeal of Movie Features to Demographic Seg-
ments of Theatrical Demand.” Journal of Cultural Economics 34:299–315.
Reinstein, D., and C. M. Snyder. 2005. “The Influence of Expert Reviews on Consumer Demand For
Experience Goods: A Case Study of Movie Critics.” The Journal of Industrial Economics 53:27–52.
Ruuskanen, O. P. 2004. An Econometric Analysis of Time Use in Finnish Households. Helsinki, Finland:
School of Economics.
Seaman, B. A. 2006. “Empirical Studies of Demand for the Performing Arts.” In Handbook of the Eco-
nomics of Art and Culture, edited by V. A. Ginsburgh and D. Thorsby, 415–472. Amsterdam, The
Netherlands: North Holland.
10 M. IRANDOUST

Suominen, S. 2013. Essays on the Demand for Cultural Performances. PhD dissertation, Acta Wasaensia,
287, University of Vaasa.
Statistiska Centralbyrån. 2015. Statistics in Sweden. Stockholm, Sweden. www.scb.se
Stiving, M. 2000. “Price-Endings When Prices Signal Quality.” Management Science 46 (12):1617–29.
Walls, W. D. 2005. “Modeling Movie Success when ‘Nobody Knows Anything’: Conditional Stable-
Distribution Analysis of Film Returns.” Journal of Cultural Economics 29:177–90.
Warde, A., and M. Gayo-Cal. 2009. “The Anatomy of Cultural Ominousness: The Case of the United
Kingdom.” Poetics 37:119–45.
Yamamura, E. 2008. “Socio-Economic Effects on Increased Cinema Attendance: The Case of Japan.”
Journal of Behavioral and Experimental Economics 37 (6):2546–555.
Yamamura, E. 2009. “Rethinking Rational Addictive Behaviour and Demand for Cinema: A Study Using
Japanese Panel Data.” Applied Economics Letters 16 (7):693–97.
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

Appendix A.

Table A. Correlation matrix for the explanatory variables.


Age Age Square Education Income WTP WTP Square Gender Marital Status WOM

Age square .


Education . .
Income . . .
WTP . . . .
WTP square . . . . .
Gender . . . − . − . .
Marital status . . . . − . . .
WOM . . . . . . . .
Critical reviews . . . . . . . . .

Appendix B.

Proportional odds model (POM)


Since our response (the observations on our dependent variable) is a three-level ordinal
variable, it is wise to consider the natural ordering to the response levels when modeling the
effects of the explanatory variables on the consumer behavior (Agresti 2007). Let:
θ1 (xi ) = π1 (xi ), (1)
and
θ2 (xi ) = π1 (xi ) + π2 (xi ), (2)
where π1 (xi ) is the probability of being a heavy cinema attendee (HCA) at the ith setting
of values of k explanatory variables xi = (x1i , . . . , xki ) , while π2 (xi ) and π3 (xi ) are the
probabilities of being a moderate cinema attendee (MCA) and being a low cinema attendee
(LCA), respectively. Thus, θ1 (xi ) and θ2 (xi ) represent cumulative probabilities: θ1 (xi ) is the
probability of being a HCA and θ2 (xi ) is the probability of being MCA or even being a HCA.
Let us define the two cumulative logits:
 
π1 (xi )
logit [θ1 (xi )] = log , (3)
π2 (xi ) + π3 (xi )
and
 
π1 (xi ) + π2 (xi )
logit [θ2 (xi )] = log . (4)
π3 (xi )
THE JOURNAL OF ARTS MANAGEMENT, LAW, AND SOCIETY 11

The first cumulative logit should be interpreted as the log odds of being a HCA as com-
pared with being a MCA or a LCA, and the second logit is the log odds of being a MCA or
a HCA as compared with a LCA. By assuming that the log odds are linear functions of the
explanatory variables, we can write:
logit[θ j (xi )] = α j + xi β j , j = 1, 2. (5)
We maximize the log of the likelihood function in order to obtain maximum likelihood
estimates:
n
L= [π1 (xi )]d1i [π2 (xi )]d2i [π3 (xi )]d3i , (6)
i=1
Downloaded by [UNIVERSITY OF ADELAIDE LIBRARIES] at 07:28 13 October 2017

subject to Equation (5), where


dhi = 1 if the ith individual gets the hth purchasing option
i = 1, 2 …, n, h = 1, 2, 3
dhi = 0 otherwise.
However, this approach does not take into account the ordinal scale of the response
variable. Thus, we suggest using the following, more parsimonious, model:
logit[θ j (xi )] = α j + xi β, j = 1, 2. (7)
Hence, the effect of an explanatory variable on the log odds of being a HCA as compared
with being a MCA or LCA is the same as the log odds of being a MCA or HCA as compared
with being a LCA. Furthermore, to better grasp the consequences implied by the restriction,
let x1 and x2 be two different settings of the explanatory variables. Then, we have the following
result:
logit[θ j (x1 )] − logit[θ j (x2 )] = (x1 − x2 )β, j = 1, 2. (8)
The log cumulative odds ratios are proportional to the distance between the values of the
explanatory variables. This feature has also given the model its name as the proportional
odds model. However, maximizing the log of the likelihood function given by Equation (6)
subject to the constraints in Equation (7) yields the parameter estimates of α 1 , α 2 , and β.

You might also like