BEHAVIOURAL AND VIEWING PATTERNS OF

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014

BEHAVIOURAL AND VIEWING PATTERNS OF CINEMA-GOERS IN MALAYSIA Noraini Abdullah1 , Suhaimi Salleh2, Ramzah Dambul 3& Diana Hassan4 1, 4 Faculty of Science &Natural Resources, 2, 3 Faculty of Humanities, Arts & Heritage, Universiti Malaysia Sabah, 88400 Kota Kinabalu, Sabah, Malaysia Email : [email protected], [email protected]

ABSTRACT Market studies on the behavioural and viewing patterns by cinema and movie-goers have not been explored extensively. Thus, in the advancing wave of global businesses, these behavioural and viewing patterns by movie-goers have to be studied and analysed inthe Islamic perspectives, hence determining the choice of films shown. This study thus is important in helping existing and budding film producers and makers to understand the behavioural and viewing patterns of the cinema fans and movie-goers. Besides, it can strategically guide the film makers to lure cinema fans back to viewing movies in the cinemas by focussing on the significant factors affecting cinema and movie goers.The various factors affecting cinema and movie goers would include encouragement, theme, motivation,perception, gratification and genre. Significant determining factors on these patterns are mathematical modelled using the multiple regression (MR) technique, focussing on the development of model-building, multicollinearity remedials and removals of insignificant factors. Consequently, a best model with significant factors that affects the number of cinema goers is obtained. This study hence would give an insight to the sustainability factors contributing to the Malaysian film industry. Keywords:

mathematical modelling,multiple regression (MR), development,multicollinearity, sustainability factors.

model-building

INTRODUCTION Nasional Film Development Corporation Malaysia (FINAS) being Malaysia’s leading film agency has defined that Malaysian films are films that are produced in Malaysia, or in any other places, but produced by Malaysians, or by companies registered in Malaysia with majority of its shareholders are Malaysians. With its two main objectives i.e. to develop the film industry to an international level, and to boost the growth of the local film industry as a destination centre for films in this region.According to Bernama (2014), more local film producers are encouraged to produce films which portray the social and cultural aspects of the people, while Mahadi (2010),suggested moreIslamic perspectives should be decpicted in films. The management of the film development agency in Malaysia needs to set clear publishingguidelines and not confined to focussing on strategies based on increasing cinema tickets collection and number of films produced. The film industry is one of the 13 sectors comprising the creative industries(DCMS, 1998). According to DeFillippi &Arthur (1998),film making and production were very closely related to the new economic millenium whereby natural-based projects and large management scale involving the exploitation of ideas from creative individuals and counterparts.However, current research had shown that countries in Asia and less-developed 876 ISBN: 978-967-5480-10-2

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014

countries had emerged as major film producers.India is an example whereby a less-developed country has become the world’s largest film producer (UNESCO, 2006; UNCTAD, 2008), with a production of about one thousand films annually which is more than by Hollywood itself. This phenomenon can also be seen in South Asia where the film industry of Korea is amongst the most successful film industry and contributes to the country’s economic growth.The success of the Korean film industry is partly due to the support from their government. Hence, the government’s role and support are of prime importance in ensuring the success of the film industry (Rampal, 2005). LITERATURE REVIEW According to Pardoe (2005), it was difficult in the decision-making processes of deciding the award winners, especially in the main Oscar winning awards- Best Picture, Best Director, Best Male Actor and Best Female Actress which had been nominated annually.He believed that Oscar awards panel of judges had not exploited the full use of the statistical report, and hence the award winning results were not always empirical. He thus had suggested the methods on how to choose the potential candidates from each category of the academy awards (Pardoe, 2005). Discrete choice models were used to compare the Bayesian method with the maximum likelihood estimator. His research had answered the difference in opinions on the Oscar award winning results, and thus had placed the award winners in a better position of their acting career. In addition, it had also given a positive message on role of mathematics to those involve in the media and the film industry. Akira (2012)had a surprise finding in his research on forecasting the success or failure of a blockbuster box office film using a mathematical model. His research in the ‘The “Hit” Phenemonon : a Mathematical Model of Human Interactions as a Stochastic Process’had used the effects of advertisements and verbal communication to form a model that had successfully abled to predict how the outcome of each film which was shown at silver screen.The convensional model which was exponential in nature was only abled to predict the sale of tickets in viewing films by cinema goersafter being linearized using logarithmictransformation, and factors representing film qualities such as, advertising budget, main role of leading actors, strength of verbal communation, actors role, script quality and others.The expectedoutcome can be seen in the increased interest in watching films through the funds used for film advertising and verbal communication through the webpages and links. METHODOLOGY Data Collection and Factor Selection Data sampleswere collected with 121variables comprising of factors influencing viewers to watch films. Significant factors would include region, state, location, age, gender, etnicity, religion, education andsalary. There were 1,337 respondents, comprised of 647males and 688 females, and aged between7to 68 years.There were 82 variables that could be used as independent variables. However, after factor analysis was carried out only five variables were chosen. The chosen variables were genre (dependent variable) while motivation, perseption, theme and gratification were the independent variables.Development of modelling processes encompasses data sampling, data filtering and rescaling, variable selection using factor analysis, variable transformationinto dummies and finally model-building procedures to obtain the best model. Modelling Technique 877 ISBN: 978-967-5480-10-2

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014

According to Devore (2012), multiple regression is an extension of the simple linear regression consisting of two or more independent variables (W1 ,W2 ,W3 ,...,Wk ) relating to the dependent variable (Y).Equation (1.0) shows the general multiple regression (MR)model given as: Y = Ω 0 + Ω1W1 + Ω 2W2 + ... + Ω kWk + u …(1.0)with, Y as the dependent variable, as thejth independent variable,

is the constant regression coefficient,

regression coeffient of independent variable

,

is thejth

is therandom residuals andk is the number

of independent variableswherej =1, 2, …, k. All these variables were categorical or qualitative; hence, each independent vatiable would be accompanied by a corresponding dummy variable denoting the factors affecting the behavioural and viewing patterns of the cinema goers (Noraini et al., 2014). Factor Analysis and Statistical Tests According to Huck (2012), factoranalysis is a statistical procedureperformed to reduce the number of variables in a data set besides relating their relationships. While according to Hair et al. (1995), factoranalysis is a statisticalmethod which is part of the multivariate analysis. Its main objective is to identify the basic structure in a data matrics; thus the corelations between the variables known as factors. However, Zainodin et al.(2011) has considered factors as variables. The number of possible models without interactions can be calculated q

using the following formula: N = ∑ C qj … (2.0) where ‘q’ is the number of independent j =1

variables and j=1,2,…q. In this paper, q=4, hence, the total models without interactions are: N = C14 + C24 + C34 + C44 = 15 models … (3.0).Zainodin et al. (2011) had shown the four phases in model-building while the multicollinearity remedial techniques and coefficient test on variables with absolute correlation coefficients more than 0.95 (i.e.│r│≥0.95) had been carried out as in Noraini & Zainodin (2013). The coefficient test is carried out to remove any insignificant factors that have p-values greater than 0.05 from the models as in Noraini et al.(2008), until the factors that remain have p-values less than 0.05. RESULTS AND DISCUSSIONS After initially performing the data filtering and screening, factor analysis was then done resulting in five factors being chosen for analysis, and represented by the symbols Y,A, B, C, D as shown in Table 1.The factor on film genre was chosen as the dependent variable (Y), while the other four factors were the independent qualitative variables in this paper. Each qualitative factor had its own respective dummy variable. These selected variables were also similar to past researches on film genre. Table 1: Summary and Symbols of Research Factors Factor Y

A :Theme [Y=f(W1)]

Description Genre-Film A1 :Theme- Family A2 :Theme- Love/Friendship A3:Theme- Religion A4:Theme- War A5: Theme- History A6 :Theme- Travels A7: Theme- Society

Category 1=Not Important 2=Less Important 3=Slightly Important 4=Important 5=Very Important

878 ISBN: 978-967-5480-10-2

Variable Type Quantitative

Qualitative

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014

B :Motivation [Y=f(W2)]

C:Perception [Y=f(W3)] D:Gratification [Y=f(W4)]

A8: Theme- Crime A9: Theme- Culture B1:Motivation-Theme B2:Motivation-Director B3 :Motivation-Actor/Actress B4:Motivation-Reviews/Promotion B5 :Motivation-Human Influence C1:Perception-Ticket Price C2:Perception-Location C3:Perception-Clean C4:Perception-Showing Times C5:Perception-Foreign Film D1:Gratification-Emotion/Soul D2:Gratification-Entertainment D3:Gratification-Inspiration D4:Gratification-Information D5:Gratification-Cultural/Historical D6:Gratification-Relationship

1=Not Important 2=Less Important 3=Slightly Important 4=Important 5=Very Important 1=Not Suitable 2=Less Suitable 3=Slightly Suitable 4= Suitable 5=VerySuitable 1=Not Important 2=Less Important 3=Slightly Important 4=Important 5=Very Important

Qualitative

Qualitative

Qualitative

Refering to equation (3.0), since there were four independent qualitative/categorical variables (W1 ,W2 ,W3 ,W4 ) , hence the total number of models without interactions were 15.Taking model M5:[f ( W1 , W2 ) ] for illustration purposes: M5 : Y5 = Ω 0 + δ1A1 + δ 2 A 2 + δ 3 A 3 + δ 4 A 4 + δ 5 A 5 + δ 6 A 6 + δ 7 A 7 + δ 8 A 8 + δ 9 A 9 + λ1B1 + λ 2 B 2 + λ 3 B 3 + λ 4 B 4 + λ 5 B 5 + u

….......(4.0)

where Y as the dependent factor on genre film, Aj are the jth independent variables of factors on ‘Theme’ with j=1,2,…,9,and Bj are the jth independent variables of factors on ‘Gratification’ withj=1,2,…,6.The jth regression coefficients were given by δ j , λ j , τ j for the independent variables,Ajdan Bj respectively. The random error was given by ‘u’, and ‘k’ is the total number of independent categorical variables with respect to each factor forj =1, 2, …, k. The multicollinearity remedials and model-building phases had been carried out as in Zainodin et al. (2011) and Noraini & Zainodin (2013). In the phase 2 of the model-buiding process, the Pearson correlationmatrics of model M5 in Table 2 did not show any absolute high values more than 0.95 (i.e.│r│≥0.95).Consequently, model M5 then becomes model M5.0, meaning that model M5 is free from multicollinearity effects between the independent variables. Table 2: Pearson Correlation Matrics of Model M5 Y A1 A2 A3 A4 A5 A6 A7 A8 A9 B1 B2 B3 B4

Y 1.00 .186 .300 .211 .313 .227 .338 .186 .332 .171 .333 .235 .249 .273

A1 .186 1.00 .287 .402 -.027 .190 .052 .281 -.039 .237 .131 .075 .063 .110

A2 .300 .287 1.00 .246 -.038 .085 .096 .123 .002 .124 .144 .060 .181 .147

A3 .211 .402 .246 1.000 .106 .293 .140 .392 .036 .340 .110 .077 .049 .137

A4 .313 -.027 -.038 .106 1.000 .314 .298 .117 .344 .089 .155 .101 .097 .155

A5 .227 .190 .085 .293 .314 1.000 .342 .359 .177 .363 .133 .086 .079 .209

A6 .338 .052 .096 .140 .298 .342 1.000 .203 .309 .212 .204 .051 .116 .228

A7 .186 .281 .123 .392 .117 .359 .203 1.000 .123 .470 .081 .041 .023 .188

879 ISBN: 978-967-5480-10-2

A8 .332 -.039 .002 .036 .344 .177 .309 .123 1.000 .131 .156 .104 .131 .140

A9 .171 .237 .124 .340 .089 .363 .212 .470 .131 1.000 .064 .093 .032 .188

B1 .333 .131 .144 .110 .155 .133 .204 .081 .156 .064 1.000 .094 .286 .209

B2 .235 .075 .060 .077 .101 .086 .051 .041 .104 .093 .094 1.000 .348 .171

B3 .249 .063 .181 .049 .097 .079 .116 .023 .131 .032 .286 .348 1.000 .205

B4 .273 .110 .147 .137 .155 .209 .228 .188 .140 .188 .209 .171 .205 1.000

B5 .240 .039 .162 .004 .116 .057 .180 .086 .133 .090 .178 .112 .202 .351

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014 B5

.240

.039

.162

.004

.116

.057

.180

.086

.133

.090

.178

.112

.202

.351

1.000

Next, the coefficient test is performed on model M5.0 using the backward elimination method (Noraini et al., 2008; Noraini et al., 2012), as shown in Table 3. Table 3 shows that factor A5 has the highest p-value of 0.961. Hence, factor A5 is removed and the model is then rerun. Table 3: Pearson Correlation Matrics of Model M5 Model 5.0 Constant A1 A2 A3 A4 A5 A6 A7 A8 A9 B1 B2 B3 B4 B5

Standardized Coefficients Beta

Unstandardized B Coefficients Std. Error 0.414 0.008 0.014 0.006 0.045 0.005 0.010 0.006 0.034 0.006 0.000 0.006 0.030 0.006 0.005 0.006 0.038 0.005 0.000 0.006 0.047 0.007 0.028 0.005 0.009 0.006 0.013 0.006 0.015 0.006

0.065 0.202 0.048 0.156 -0.001 0.136 0.023 0.174 -0.002 0.163 0.128 0.036 0.059 0.066

t

p-value

54.591 2.508 8.228 1.767 6.081 -.049 5.247 .822 6.936 -.072 6.647 5.236 1.401 2.330 2.647

0.000 0.012 0.000 0.077 0.000 0.961 0.000 0.411 0.000 0.943 0.000 0.000 0.161 0.020 0.008

Four removals of insignificant variables (A5, A9, A7 and B3) had been carried on model M5.0. Referring to model labelling (Zainodin et al., 2011; Noraini & Zainodin, 2013), model M5.0 thus becomes M5.0.4 indicating there are zero multicollinearity variables and four insignificant variables had been removed, and the elimination process can be given as: M5.0.0=>M5.0.1=>M5.0.2=>M5.0.3=>M5.0.4. It can be seen in Tables 3 that the significant factors that remained in the model had all the p-values less than 0.05. These modelling procedures are illustrated in detail (Zainodin et al., 2011; Noraini et al., 2008; Noraini & Zainodin, 2013). Table 3: Pearson Correlation Matrics of Model M5.0.4 Model 5.0.4 Constant A1 A2 A3 A4 A6 A8 B1 B2 B4 B5

Unstandardized B Coefficients Std. Error .414 .008 .014 .006 .045 .005 .010 .006 .034 .006 .030 .006 .038 .005 .047 .007 .028 .005 .013 .006 .015 .006

Standardized Coefficients Beta .065 .202 .048 .156 .136 .174 .163 .128 .059 .066

t

p-value

54.591 2.508 8.228 1.767 6.081 5.247 6.936 6.647 5.236 2.330 2.647

0.000 .012 .000 .077 .000 .000 .000 .000 .000 .020 .008

880 ISBN: 978-967-5480-10-2

Correlations Zero-order Partial .186 .300 .211 .313 .338 .332 .333 .235 .273 .240

.071 .227 .050 .170 .147 .193 .185 .147 .066 .075

Part .057 .186 .040 .137 .119 .157 .150 .118 .053 .060

Collinearity Statistics Tolerance VIF .763 .843 .700 .775 .763 .812 .848 .855 .787 .825

1.311 1.186 1.428 1.291 1.311 1.232 1.179 1.170 1.271 1.212

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014

All the possible 15 models had undergone the statistical tests in the methodology section. Finally the best model was chosen based the eight selection criteria (8SC) (Zainodin et al., 2011). Model M15.0.12 was then found to be the best model given by equation (5.0): M15 .0.12 = Ω 0 + δ1 A 1 + δ 2 A 2 + δ 4 A 4 + δ 6 A 6 + δ 8 A 8 + ω1 B1 + ω 2 B 2 + ω 4 B 4 + π1C1 + λ 2 D 2 + λ 3 D 3 + λ 5 D 5 + λ 6 D 6

………………….(5.0)

Substituting the values of the regression coefficients, equation (5.0) then becomes:

M15.0.12 = 0.391+ 0.12A1 + 0.38A2 + 0.35A4 + 0.25A6 + 0.32A8 + 0.36B1 +

……………..(6.0) 0.29B2 + 0.12B4 + 0.16C1 + 0.19D2 + 0.14D3 + 0.14D5 + 0.35D6 Substituting the original representation of the research factors into equation (6), the best model M15.0.12 is thus given by: GenreFilm= 0.391+ 0.12Family + 0.38Love/ Friendship+ 0.35War + 0.25Travels+ 0.32Crime + 0.36Motivation(Theme) + 0.29Director+ 0.12 Re views/ Pr omotion ………….(7.0) + 0.16TicketPr ice + 0.19Entertainment + 0.14Inspiration + 0.14Cultural/ Historical+ 0.35 Re lationship

Equation (7.0) implies that the increase in the number of cinema-goers and watching movies based on film genre is determined by the significant research factors in the best model of M15.0.12. The relevant factors are those that are concerned with family, love and friendship, war, travels, crime, theme, director, reviews and promotion. In addition, the price of tickets, types of entertainment, inspiration inculcated, either on cultural or historical perspectives and relationships involved have direct relationships to the genre of the film shown. The highest contribution on the genre film firstly comes from themes on love and friendship (0.38), followed by motivation to watch films because of the theme itself (0.36), gratification from roles of relationships in the script (0.35) and finally themes based on crime (0.32).All these factorsare contributing positively to the increase in the number of cinema-goers and watching movies.However, in comparison to Noraini et al. (2014), the director’s motivation, perception on the film’s location, gratification and inspiration from film-viewing besides relationships involved in the script, gave significant positive impacts on film viewers in East Malaysia. These differences can be said due to the geographical location, demographic factors, income levels, facilities available, cultural influence and infrastructure development. CONCLUSIONS This paper introduces the concept and procedures in mathematical modelling using the multiple regression technique so as to identify significant factors that affect the sustainability of the Malaysian film industry. Significant factors are as represented in the best model of M15.0.12: GenreFilm = 0.391 + 0.12Family + 0.38Love / Friendship + 0.35War + 0.25Travels + 0.32Crime + 0.36Theme + 0.29Director + 0.12 Re views / Pr omotion + 0.16Ticket Pr ice + 0.19Entertainment + 0.14Inspiration + 0.14Cultural / Historical + 0.35 Re lationship

These factors give positive and direct contribution to the increasein the number of cinema-goers and movie fans based on the genre of the films shown. It can be concluded 881 ISBN: 978-967-5480-10-2

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014

thatfilms genres of various themes would certainly boostthe Malaysian film industry while providing various choices to film-fans. Since Malaysia is designated as an Islamic country, it is only appropriate that all genres and themes of films being produced herewith must be compliant to Islamic perspectives, while not the neglecting factors which will ensure their survival and viewability appeals. For Malaysian audience, based on the researched findings, it is recommended that films which portray family, love/friendship, war and travel will captivate the fans. Films theme, director, reviews/promotion and ticket price would attract viewers as well as values of entertainment, inspiration, culture/history and relationship. These considerations will also form part of the strategy to promote indigenous and Shariahcompliant content so as to counter possible negative impacts of foreign films brought in from the outside of Malaysia. ACKNOWLEDGEMENT The authors would like to thank FINAS and Universiti Malaysia Sabah which had funded this research. REFERENCES Akira Ishii. 2012. The “Hit” Phenomenon: A Mathematical Model of Human Dynamic Interactions as a Stochastic Proces. New Journal of Physics, 14(063018):3-7. Bernama. 2014. Available from: www.utusan.com.my/ …/Banyakkan_filem_berunsur_islam. Arkib date: 25/04/2014. DCMS. 1998. Creative Industries Mapping Document. Department of Culture, Media and Sport. United Kingdom. DeFillippi, R.J. & Arthur M.B. 1998. Paradox in project-based enterprise: The case of film making. California Management Review, 40 (2): 125 – 139. Devore, J.L. 2012. Probability and Statistic for Engineering and the Science. 8th edition. Canada: Brooks/Cole, Cencage Learning. Huck, S.W. 2012. Reading Statistic and Research. 6th edition. New Jersey: Prentice Hall. Hair, J.F., Anderson, R.E., Tatham, R.L. and Black, W.C. 1995. Multivariate data analysis with reading. 5th edition. New Jersey: Prentice Hall inc. Mahadi J Murad. 2010. Cara hidup Islam perlu diperbanyakkan. www.sinarharian.com.my/ …/filem_tonjol_cara_hidup_islam_.... Noraini Abdullah, Zainodin H.J. & Nigel Jonney J.B. 2008. Multiple Regression Models of the Volumetric Stem Biomass. WSEAS Transaction on Mathematics, 7(7): 492-502. Noraini Abdullah, Zainodin H.J. &Amran Ahmed. 2012. Comparisons Between Huber’s and Newton’s Multiple Regression Models for Stem Biomass Estimation. Malaysian Journal of Mathematical Sciences, 6(1): 1-28. Noraini Abdullah & Zainodin H.J. 2013. Multicollinearity Remedial Techniques in ModelBuilding. Matematika (UTM), 29(1b): 105-115. Noraini Abdullah, Suhaimi Salleh, Zainodin H.J. and Ramzah Dambul. 2014. Faktor-faktor Signifikan KearahKemapanan Industri Perfileman di Malaysia Timur. Journal of Social Sciences :Kinabalu. Universiti Malaysia Sabah. (In Press). Pardoe, I. 2005. Applying Discrete Choice Models to Predict Academy Awards Winners. Journal of Royal Statistical Society, 375-394. Rampal, K.R. 2005. Cultural Imperialism or Economic Necessity: The Hollywood Factor in the Reshaping of the Asian Film Industry. Global Media Journal, 4(6). 882 ISBN: 978-967-5480-10-2

2nd World Conference on Islamic Thoughts & Civilisation 18-19 August 2014

UNESCO. 2006. Trends in Audiovisual Markets: Regional Perspectives from the South. Paris, France. UNCTAD. 2008. Creative Economy Report 2008: The Challenge of Assessing the Creative Economy: Towards Informed Policy-making. UNDP. New York, NY. Zainodin, H.J., Noraini, A. & Yap, S.J. 2011. An Alternative Multicollinearity Approach in Solving Multiple Regression Problem. Trends in Applied Science Research, 6 (11): 1241 1255.

883 ISBN: 978-967-5480-10-2