Service Quality, Customer Satisfaction and Brand Loyalty in the Swedish Subscription Video On Demand- Industry

A mixed methods study on what factors of service quality affect customer satisfaction and brand loyalty within the subscription based video-on- demand services amongst 18-29 year olds in Sweden.

BACHELOR DEGREE PROJECT THESIS WITHIN: Business & Administration NUMBER OF CREDITS: 15 ECTS PROGRAMME OF STUDY: Marketing Management AUTHOR: Oliver Berg, Elliot Strand, Viktor Sandell JÖNKÖPING May 2019

Bachelor Degree Project in Business Administration

Title: Service Quality, Customer Satisfaction and Brand Loyalty in the Swedish Subscription Video On Demand-Industry Authors: Oliver Berg (940624), Elliot Strand (950713), & Viktor Sandell (970708) Tutor: MaxMikael Wilde Björling Date: 2019-05-17

Key terms: Subscription video on demand (SVOD), Service Quality, Customer Satisfaction, Brand Loyalty

Abstract

The main objective of this research is to measure the relation between different dimensions, or constructs of service quality with customer satisfaction and brand loyalty; thereby identifying the most important factors of service quality that affect customer satisfaction and brand loyalty, in the context of subscription video on demand (SVOD) services in the Swedish market. This study utilised a mixed methods approach divided into two stages, whereof the first involved multiple focus groups for collecting qualitative data. The second stage consisted of quantitative data collection through online questionnaires from 122 valid respondents, ranging from the age of 18 to 29 by a non- probability sampling method. SPSS was used to analyse the data with three statistical analysis methods: bivariate correlations, for establishing evidence of construct validity; a reliability analysis to test Cronbach alpha for internal consistency; and multiple regression analysis to examine the relationship between independent and dependent variables. The research findings indicate that there is a positive, significant relationship between customer satisfaction and certain constructs of service quality. Also, findings indicate that the statistical relationship between the constructs of service quality and brand loyalty are partially mediated by customer satisfaction. As a result, companies operating in the Swedish SVOD-industry need to focus on improving the significant factors of service quality, as they may increase customer satisfaction of 18 to 29-year-old customers, and further create long-term relationships as a way of gaining competitive advantage.

Keywords: Subscription video on demand (SVOD), Service Quality, Customer Satisfaction, Brand Loyalty.

i

Acknowledgements

The authors of this thesis would like to acknowledge everyone who have been involved and contributed to the research project. First of all, the authors would like to express a special thanks of gratitude to their tutor, MaxMikael Wilde Björling, who have been helpful in providing guidance and support throughout the research project.

Secondly, the authors would also like to extend their gratitude to the individuals who participated in the focus group sessions. Your insights and opinions expressed have been invaluable for developing the online questionnaire. Also, many thanks to all respondents who completed the online questionnaire. Without your answers, there would not have been any quantitative data to analyse.

Third, the authors would like to express their appreciation to the teaching team at Jönköping International Business School for providing the authors with clear guidelines and instructions throughout the research project.

Last but not least, the authors would finally like to thank Jonas Persson for sharing his knowledge and technical expertise in questionnaire development in Qualtrics, and data analysis in SPSS.

This thesis would not have been completed without the concerned parties’ enormous help and support. Thank you indeed!

Jönköping, May 17, 2018

Oliver Berg Viktor Sandell Elliot Strand

ii

Table of Contents

1. Introduction...... 1 1.1 Background ...... 2 1.2 Problem ...... 4 1.3 Purpose ...... 5 1.4 Delimitations ...... 5 2. Frame of Reference...... 6 2.1 Literature review ...... 6 2.1.1 The Subscription Video On Demand•industry...... 6 2.1.2 Generation Y&Z ...... 7 2.1.3 Customer Satisfaction ...... 7 2.1.4 Brand Loyalty ...... 8 2.1.5 Service Quality ...... 8 2.2 Conceptual Model & Hypotheses ...... 10 3. Methodology ...... 11 3.1 Research Context ...... 11 3.2 Research Approach ...... 11 3.3 Research Design ...... 12 3.3.1 Stage 1•Exploratory Qualitative Research ...... 12 3.3.2 Stage 2•Descriptive Quantitative Research...... 13 3.4 Population & Sample ...... 14 3.4.1 Sample Size ...... 14 3.4.2 Sampling Method ...... 15 3.4.3 Sampling Error ...... 15 3.5 Data Collection Methods ...... 16 3.5.1 Focus Groups for Qualitative Data Collection ...... 16 3.5.2 Online Questionnaire for Quantitative Data Collection ...... 18 3.5.3 Questionnaire Development ...... 18 3.5.4 Conceptualisation ...... 19 3.5.5 Questionnaire Design ...... 19 3.5.6 Questionnaire Translation ...... 21 3.5.7 Questionnaire Testing & Revision ...... 21 3.5.8 Quantitative Data Collection ...... 21 3.6 Data Analysis Methods ...... 21 3.6.1 Correlational Research Design & Multiple Regression ...... 21 3.6.2 Multiple R & Adjusted R2 ...... 22 3.6.5 Standardized Beta Coefficient ...... 23 3.6.3 Multicollinearity...... 23 3.7.1 Reliability ...... 23 3.7.2 Construct Validity ...... 24 3.7.3 Convergent Validity ...... 24 3.7.4 Discriminant Validity ...... 25 3.8 Ethical Considerations ...... 25

iii

4. Measurement Instruments ...... 26 4.1 Constructs of Service Quality ...... 27 Efficiency… ...... 27 System Availability ...... 28 Privacy..…...... 28 Fulfilment • Library of Content ...... 29 Recommendation System ...... 30 Quality of Experience ...... 31 Pricing…… ...... 31 4.2 Additional Variables ...... 32 4.2.1 Overall Customer Satisfaction ...... 32 4.2.2 Brand Loyalty ...... 32 4.3 Classification Variables ...... 32 5. Statistical Analysis & Results ...... 33 5.1 Reliability & Validity...... 34 5.1.1 Cronbach Alpha ...... 34 5.2 Convergent & Discriminant Validity ...... 34 5.2.1 Convergent Validity ...... 35 Efficiency… ...... 35 System Availability ...... 35 Privacy..…...... 35 Recommendation System ...... 36 Fulfilment • Library of Content ...... 36 Quality of Experience ...... 36 Pricing…… ...... 37 5.2.2 Discriminant Validity ...... 37 5.3 Multiple Regression Analyses ...... 38 Efficiency… ...... 38 System Availability ...... 39 Privacy..…...... 40 Fulfilment • Library of Content ...... 41 Recommendation System ...... 42 Quality of Experience ...... 43 Pricing…… ...... 44 5.3.1 Hypotheses Testing ...... 45 Hypothesis 1 ...... 46 Hypothesis 2 ...... 47 Hypothesis 3 ...... 48 Hypothesis 4 ...... 49 6. Conclusion ...... 51 7. Discussion ...... 53 7.1 Theoretical Implications ...... 56 7.2 Practical Implications ...... 56 8. Limitations...... 58 8.1 Suggestions for Further Research ...... 59 9. References ...... 60

iv

10. Appendices...... 65 10.1 Questionnaire ...... 65 10.1.1 Information Sheet ...... 65 10.1.2 Form of Consent...... 66 10.1.3 Questionnaire in English ...... 67 10.1.4 Questionnaire in Swedish...... 70 10.1.5 Absolute Frequency ...... 74 10.2 Convergent Validity ...... 87 10.2.1 Efficiency ...... 88 10.2.2 System availability ...... 88 10.2.3 Privacy ...... 89 10.2.4 Recommendation system ...... 89 10.2.5 Fulfilment • Library of Content ...... 90 10.2.6 Quality of Experience ...... 91 10.2.7 Pricing ...... 91 10.3 Discriminant Validity ...... 92 10.4 Model Summaries and Coefficients ...... 93 10.4.1 Efficiency Construct ...... 93 10.4.2 System Availability Construct ...... 93 10.4.3 Privacy Construct ...... 94 10.4.4 Fulfilment•Library of Content Construct ...... 94 10.4.5 Recommendation System Construct ...... 95 10.4.6 Quality of Experience Construct ...... 95 10.4.7 Pricing Construct...... 96 10.5 Hypotheses Testing ...... 97 10.5.1 Hypothesis 1 ...... 97 10.5.2 Hypothesis 2 ...... 98 10.5.3 Hypothesis 3 ...... 99 10.5.4 Hypothesis 4 ...... 100

v

1. Introduction

______

In this section, an introduction followed by a background on the research topic is given. A discussion of the problem and the purpose of this research is later covered. The research question is presented, along with the research objectives. Moreover, the last section discusses the delimitations of the research. ______

About a decade ago, smartphones by the standard as we know them by today did not exist. Three decades earlier, most people did not even own a personal computer. Nowadays, people are spending their time looking at bright screens, whether it is a smartphone screen, a computer screen or a television screen. The progression of technology is moving at a faster pace than ever before and the advances in technology have paved the road for new business opportunities to emerge. During this development, the opportunity to exploit the systems arose, and piracy became increasingly popular by consumers of online entertainment (Poort, Quintais, van der Ende, Yagafarova, & Hageraats, 2018). Subjected to many challenges, consumers’ willingness to pay for entertainment decreased, and the danger of piracy harmed businesses worldwide, facing economic losses estimated to over $200 billion in 2005, as a result of counterfeit and pirated goods (Organisation for Economic Co-Operation and Development, 2007). In recent years, rapid advances in technology have enabled users to access increased bandwidth and speed in their internet connections. Such improvements in combination with modern operating systems and optimized personal computers have created new, lucrative business opportunities.

Through several business ideas that have risen, subscription based video on demand streaming is an example of a service that has made its mark on the modern media industry. The phenomenon is more commonly known as subscription video on demand (SVOD). The services include all forms of video services where users have unrestricted access to a frequently updated variety of movies and series, at a fixed monthly rate, and where they have the ability to play, pause, fast forward and rewind the content as they please (Grece, 2014). New industries are argued to be important for economic growth, as they may transform mature industries and drive innovation, which accurately describes what has happened within the entertainment industry (Feldman & Tavassoli, 2014). Furthermore, for businesses to flourish, understanding potential customers is of high importance in order to generate a

1

superior offering. Absolute determinants of customer satisfaction and brand loyalty within the industry of SVOD are yet to be established, due to the limited research conducted in the field. The positive relationship between the two concepts, however, has been made apparent by previous research (Oliver, 1999; Ahmed, Chandio, & Quresh, 2015).

1.1 Background

Two highly dominant marketing topics of today are customer satisfaction and brand loyalty. The actuality of the topic has led to extensive research in the field, and vast advances for both practice and academia (Schirmer, Ringle, Gudergan, & Feistel, 2016). Satisfaction can be defined as a person’s enjoyment or disappointment resulting from a performance specific expectation, in comparison to the expectancy disconfirmation, which implies diminishment of prior expectations (Oliver, 1980; Kotler, 2000). Also, some researchers argue that disconfirmation of expectations has an effect on perceived quality, together with various other factors, such as product features and attributes (Bolton & Drew, 1991; Parasuraman, Zeithaml, & Berry, 1988). Perceived quality is also proposed to be similar to attitude, and that the process shaping perceived quality is similar to the journey of attitude creation. (Gotlieb, Grewal, & Brown, 1994; Dodds, Monroe, & Grewal, 1991; Parasuraman et al., 1988).

Furthermore, brand loyalty is based on the fundamentals of customers seeking the best value, in which customer satisfaction is a predominant driver (Reichheld, 1994; Anderson & Sullivan, 1993). The link between satisfaction and loyalty has been evident in previous research, though reports have also indicated that satisfaction alone is not always necessarily a recipe for loyalty (Reichheld & Sasser, 1990). The two concepts can be examined through the utilization of the Net Promoter Score (NPS), which is used in management as a loyalty metric to predict firm growth (Keiningham, Aksoy, Cooil, Wallin Andreassen, & Williams, 2008). Even though the NPS has been widely criticized for its simplicity (Kristensen & Westlund, 2004; Morgan & Rego, 2004; Kristensen & Eskildsen, 2014), it has still captured a lot of interest and is frequently applied in marketing research (Kristensen & Eskildsen, 2014).

Aided by technological enhancements, the media industry has recognised the capability of evolving the internet into a video outlet. The vast increase in online media has brought to question the viability of cable television and further motivated new companies to arise, as well as transition, from traditional media to online platforms. This has allowed for consumers to access media content considerably quicker. The fast access, along with the simplicity of modern media sharing aided by internet

2

development may, according to Cha and Chan•Olmstead (2012), have been contributing to the debated cannibalising effect on cable television. As mentioned by Trowbridge (2013), “A paradigmatic shift has been occurring in terms of how film and media content at large is distributed, spread, and shared between people and organisations...” (p. 228). The presence of a shift has also been noticeable in the Swedish market (Mydigheten för Press, Radio och TV, 2017). For many individuals in Sweden under the age of 30, screens have been a part of their daily life since childhood, along with the internet. Individuals in Sweden, younger than 30 years of age, are among those who watch cable TV the least amount of time each day, while being the heaviest users of SVOD•services (Myndigheten för press, radio och TV, 2017).

The Swedish market consists of both subscription-based video on demand services as well as services that rely on advertisements as their main source of revenue. In a study conducted by Audience Project (2017), 59% of the Swedish population use the internet in order to access TV shows or movies on a weekly basis. The most popular sources for online media in Sweden are YouTube, SVT Play, Netflix, TV4 Play, Viaplay, Cmore, and HBO Nordic, whereof YouTube, SVT Play, and TV4 Play do not require a subscription. Instead, SVT•play is free for users to watch, whereas YouTube and TV4 Play rely on advertisements. Contrarily, Netflix, Viaplay, Cmore, and HBO Nordic provide subscribers with on demand content, in exchange for a monthly fee.

Since April 2018, new rules have made it possible for consumers to stream online content all across the EU (European Parliament, 2018). Naturally, this includes videos from the previously mentioned service providers. The reasoning behind the decision is partially to discourage piracy, as some content was geographically limited to only some countries within the EU before the new regulation. The regulation aims to counter copyright related issues, as copyrights were previously acquired and enforced on a country•to•country basis. Together with new roaming regulations, travelling consumers within the EU now have the same access to the content they usually watch at home (European Parliament, 2018).

3

1.2 Problem

In recent years, a dramatic switch from analogue home entertainment towards internet•based behaviour has been apparent, which has allowed for video streaming services to thrive (Rodríguez, Rosa, Costa, Abrahão, & Bressan, 2014). A survey conducted by Swedish quality index (2018), showed that 70% of Swedish video on demand subscribers believe that they would remain loyal to their current streaming service providers for at least one year. This entails a 30% gap in consumers who believe that they would not remain loyal to their current service provider. The reasons for why the doubtful 30% question their loyalty remains to be determined. Furthermore, the reason for why the Swedish population within the age category 18•29 is of interest to this study, is since people of these demographics are among those who are watching cable TV for the least amount of time, and spend the most time using SVOD•services instead (Mydigheten för Press, Radio och TV, 2017).

As new business opportunities emerge through the change in consumer preferences, the matter of investigating factors that allow businesses to thrive in the focal industry becomes critical. In order for businesses in the subscription•based video streaming industry to grow, returning customers are of high importance. As proven by Reichheld and Sasser (1990), customer loyalty may lead to increased profits by 25 to 85 percent. In order to maintain subscribers, companies need to understand which factors are satisfying their customers to the extent where they perceive the service to provide a higher value than what they are paying. To date, research regarding this issue remains limited.

4

1.3 Purpose

To address the mentioned gaps in current literature, this study aims to explore what factors of service quality affect customer satisfaction and brand loyalty in the SVOD•industry. This is important as the SVOD•industry is subject to change in both consumer taste and technology. Thus, SVOD•service providers must focus their efforts toward factors of service quality that are important drivers of customer satisfaction, which in turn, may facilitate the creation of long-term relationships. In order to do so, companies must be able to measure users’ perceptions of service quality (Juluri, Tamarapalli, & Medhi, 2016). Accordingly, the question that this study will aim to answer is:

“What is the relation between service quality, customer satisfaction, and brand loyalty in the Swedish SVOD•service industry, amongst 18 to 29 year-olds?”

In order to answer the proposed research question, the following objectives are presented:

➢ To identify what factors of service quality affect customer satisfaction and brand loyalty with regards to SVOD•services. ➢ To determine the importance of the derived factors of service quality in relation to customer satisfaction and brand loyalty. ➢ To examine the mediating effect of customer satisfaction between service quality and brand loyalty.

1.4 Delimitations

This study seeks to solely focus on services which exclusively offer video streaming, more specifically, on demand video streaming services that rely on subscription fees. Thus, other streaming services that do not solely rely on subscriptions are disregarded. As there has been a finite amount of studies conducted with focus on the Swedish market, this study aims to make a contribution to the scarcely researched field.

In Sweden, the major market shareholders are Netflix, Viaplay, C More and HBO Nordic (Mydigheten för Press, Radio och TV, 2017), and will for this reason be included in this study. This entails disregarding services such as TV4 Play, and Dplay, as they can be used without a subscription, and the concept of loyalty could be interpreted differently depending on the revenue model utilised by the service provider. This may therefore contribute to misleading results.

5

2. Frame of Reference

______

The following section will review the work of literature on the subscription video on demand•industry, generation Y and Z, customer satisfaction, brand loyalty, and service quality. This is followed by a presentation of a conceptual model and the developed research hypotheses. ______

2.1 Literature review

2.1.1 The Subscription Video On Demand•industry

With increased quantity of user ratings from online streaming services, more studies in recent years have focused on the motivation for why users view content from online media services instead of cable television (Bondad-Brown, Rice, & Pearce, 2012). Studies have also covered the various impacts that piracy has had on the industry (Hui & Png, 2003; Borja, Dieringer, & Daw, 2014), whether the cannibalization of online video platforms on cable television indeed takes place (Cha & Chan-Olmsted, 2012), and to what extent changes in consumer behaviour can be caused by fluctuations in video quality (Krishnan & Sitaraman, 2013). However, research on SVOD•services have mainly focused on what factors impact an individual when deciding whether to choose online streaming over cable TV (Lee, Nagpal, Ruane, & Lim, 2018).

The SVOD industry has been evolving naturally, as new technologies become available, and consumer preferences and demands change. This further emphasises the importance of providing relevant content for their viewers through well-crafted search engines and recommendation systems (Hasan, Jha, & Liu, 2018). All SVOD•services investigated in this study utilise a recommendation system and a search engine. The latter is a tool for users to locate specific information sources, however, requires the user to know a significant part of the information which they are searching for, in order to be successful in finding the correct result (Hasan et al., 2018). Recommendation systems display suggestions for users about the next activity to indulge in, which are derived from the users preferences, history, and several other factors (Park, Choi, Kim, & Kim, 2011). Thus, recommendation systems are not limited to the memory of the user (Hasan et al., 2018). A well- crafted recommendation system works towards attracting consumers to continue watching, and remain on the site by recommending relevant content to the user (Li, Zheng, Yang, & Li, 2014).

6

According to Häubl and Trifts (2000), a recommendation system possesses the capability to significantly increase user’s experience and exposure of the particular platform.

2.1.2 Generation Y&Z

It is agreed that individuals born around 1980, up until the mid, or late ‘90s belong to Generation Y (Nagy, 2012; Shatto & Erwin, 2016). However, there is no agreed definition of the age category Generation Z, as some treat this age category as logically following Generation Y, including individuals born from the mid­1990’s (Shatto & Erwin, 2016; Southgate, 2017; Montana & Petit, 2008). Meanwhile, Caroline Geck (2006) treat individuals born in or after the year of 1990 as Generation Z, as these share more characteristics with regards to internet knowledge and habits, since they were all born into a world where internet existed. As this study aims to investigate the age category of 18 to 29 year-olds, it can be agreed that the target population of this study consist of young individuals belonging to Generation Y, as well as the oldest ones of Generation Z. With this in mind, the target population is not represented by the entirety of generation Y and Z. Thus, some characteristics may not be shared between individuals belonging to the older tier of Generation Y, or the younger tier of Generation Z.

Moreover, individuals of Generation Y are considerably more skilled in multitasking than previous generations, while they generally are more technologically knowledgeable (Nagy, 2012). Also, they often have difficulties in concentrating on one issue for a long period of time, which is even more apparent when considering Generation Z (Nagy, 2012; Shatto & Erwin, 2016). According to Shatto and Erwin (2016), Generation Z is also the first generation to watch programming or streaming services when they find it convenient, as smart devices and sophisticated services allow for them to watch content anywhere, and at all times.

2.1.3 Customer Satisfaction

Customer satisfaction has for decades been a highly researched topic within marketing, and shown to be a viable predictor of future consumer purchasing intentions (McQuitty, Finn, & Wiley, 2000; Cardozo, 1965; Oliver 1980; Bloemer & Kasper, 1995). Marit G. Gundersen, Morten Heide, and Ulf H. Olsson (1996) define satisfaction as “a postconsumption evaluative judgment concerning a specific product or service” (p.74). Oliver (1980) argues that the determinants of satisfaction include factors such as consumers pre•purchase expectations and post•purchase evaluation. This, in turn, means that disconfirmation resulting from the consumption of a good, should ultimately decrease as consumers’ knowledge of the product experience increases, hence a decrease in pre•purchase expectations

7

(McQuitty et al., 2000). Customer satisfaction is argued to increase a firm’s profitability through the development of customer retention (Hennig-Thurau & Klee, 1997). Furthermore, Patterson, Johnson, and Spreng (1996) have discovered customer satisfaction as the primary predictor of repurchase intent.

2.1.4 Brand Loyalty

The concept of brand loyalty is highly connected to customer satisfaction, as it often is the ultimate goal towards which businesses strive, and where customer satisfaction is one of the predominant drivers, mainly together with brand trust (Oliver, 1999; Ahmed et al., 2015). However, there are contradicting studies which argue that the certainty of the relation between the two concepts is not as straightforward as suggested. Instead, factors such as demographics, education, and age also play major roles in the process of attaining loyal customers (Kumar, Pozza, & Ganesh, 2013; Keiningham, Aksoy, Cooil, & Hsu, 2007; Homburg & Giering, 2001). Although there is no fine line regarding what a loyal customer is, Newman and Werbel (1973) define loyal customers as those who repurchased a brand and only considered that particular brand without engaging in any brand related information search.

In a study conducted by Hellier, Carr, Geursen, and Rickard (2003), with support from empirical evidence, customer satisfaction was found to have a direct positive effect on repurchase intention (Anderson & Sullivan, 1993; Bolton 1998; Cronin Jr. & Taylor, 1992; Fornell, 1992; Oliver, 1980). Gupta, Lehmann and Stuart, in their article “Valuing Customers” (2004), argue that an improvement in customer retention by 1% may lead to an increase in firm value with up to 6,75% if managed correctly. The article, however, does not focus on SVOD•services, but rather measures the value of subscriber retention in general.

2.1.5 Service Quality

In previous studies, results have shown that service quality is a significant predictor of customer satisfaction (Gorondutse & Hilman, 2014). Service quality is a measure of how well the service is performed, according to a customer’s subjective assessment, however, it is not only an evaluation of a services’ outcome as it also involves evaluations in the way that the service is delivered (Parasuraman, Zeithaml, & Berry, 1985). Based on insights from exploratory research, Parasuraman et al. (1985) have developed a model which summarises ten dimensions of service quality as consumers perceive it. In later studies, however, a strong correlation among the ten dimensions was found. The ten dimensions were, therefore, combined into five dimensions of reliability,

8

responsiveness, assurance, empathy, and tangibles which constituted as the basis of a tool known as SERVQUAL, which is used for measuring service quality (Parasuraman, Zeithaml, & Berry, 1996).

While the SERVQUAL instrument has been extensively used in various contexts for measuring service quality, the rise of the internet as a channel for selling goods and services demanded a refinement of the SERVQUAL instrument (Parasuraman, Zeithaml, & Malhotra, 2005). Accordingly, Electronic Service Quality (E•SERVQUAL) was developed and designed to measure the service quality of electronic commerce websites. It was developed in 2005 by A. Parasuraman together with Valarie A. Zeithaml and Arvind Malhotra. The scale consists of 22 items categorised into four dimensions: efficiency, system availability, fulfilment and privacy (Parasuraman et al., 2005).

Several other scales such as WebQual (Loiacono, Watson, & Goodhue, 2007), WebQual (Barnes & Vidgen, 2002), and SITE•QUAL (Yoo & Donthu, 2001) have been developed for evaluating and measuring website quality. The WebQual scale developed by Barnes and Vidgen (2002) has been criticised for lacking aspects of the purchasing process, as it only measured the consumer’s website experience until the point of purchase (Parasuraman et al., 2005). Also, the WebQual designed by Loiacono, Watson and Goodhue (2007) was, in the same article criticised for generating information for website designers instead of measuring customer experiences, as well as for not including the fulfilment dimension. Parasuraman et al. (2005) argue that the customer’s evaluation of the quality of a website involves both experiences during website interaction, as well as service related experiences following interaction with the website. Therefore, E•SERVQUAL is argued to be a superior satisfaction measurement, as it takes into account each stage of a customer’s interaction with a website (Parasuraman et al., 2005).

Moreover, a significant amount of research has also facilitated the development of metrics that attempt to measure the quality of service, or quality of experience perceived by the user (Juluri, Tamarapalli, & Medhi, 2016; Su et al., 2016; Rodríguez et al., 2014; Torres Vega, Perra & Liotta, 2018; Maia, Yehia, de Errico, 2014). Furthermore, research has shown a link between quality of service and repurchase intent, and how it is mediated through customer satisfaction (Brady & Robertson, 2001).

9

2.2 Conceptual Model & Hypotheses

The proposed conceptual model (figure 1.) for this study assumes that the best way to explain the effect of the independent variable, service quality, on the dependent variable, brand loyalty, is by using a third variable, customer satisfaction, as a mediator. That is, instead of the independent variable causing the dependent variable directly, the dependent variable is caused by a mediator, which in turn is caused by the independent variable. The conceptual model for this research, therefore, assumes that service quality, consisting of several dimensions, influences customer satisfaction, which in turn influences brand loyalty. The constructs of service quality were derived partly from the multiple•item scale of E•SERVQUAL (Parasuraman, Zeithaml, & Malhotra, 2005), studies by Hasan, Jha and Liu (2018), Krishnan and Sitaraman (2013), Maia, Yehia and Errico (2014) and MUX (2017), as well as from qualitative, primary findings gathered from focus groups. The final constructs included efficiency, system availability, fulfilment•library of content, privacy, quality of experience, pricing and recommendation system.

Figure 1.

From the conceptual model, the following hypotheses were developed:

H1: At least one construct of service quality is significantly, positively associated with customer satisfaction

H2: At least one construct of service quality is significantly, positively associated with brand loyalty

H3: Customer satisfaction is significantly, positively associated with brand loyalty

H4: Customer satisfaction mediates the relationship between service quality and brand loyalty

10

3. Methodology

______

The following section presents the research context, the research approach, and the research design, where the two stages of this study are explained. This is followed by a description of the target population, sample size, and sampling method. Lastly, the methods for data collection and analysis are presented and discussed, as well as ethical considerations. ______

3.1 Research Context

The study aims to research the Swedish SVOD sector, due to the previously mentioned changes with regards to consumer behaviour in the industry (Mydigheten för Press, Radio och TV, 2017; Svenskt Kvalitetsindex, 2018). As both domestic and global players compete for market share in Sweden, understanding how consumers value factors of quality is of high importance.

3.2 Research Approach

In quantitative studies, a deductive research approach is widely applied when testing a conceptual or theoretical structure, and when attempting to establish, confirm, or validate relationships, in order to make generalisations that add value to theory (Churchill & Brown, 2004; Collis & Hussey, 2014). A deductive approach was therefore considered to be an appropriate, logical process for exploring the phenomena from a top•down approach, where the findings would provide indications and generalisations, and which in turn could contribute to theory. This was utilised through developing a conceptual model, deriving constructs from existing research which served as foundations for the remaining parts of data collection and analysis.

The study was executed in two steps, where both qualitative and quantitative research methods were utilised. This course of action is known as mixed methods research and aims to draw from the strengths as well as minimize the weaknesses associated with the two approaches (Creswell, 2013). For this research, the utilisation of mixed research methods allowed the researchers to explore the problem, and obtain a deeper understanding before conducting the descriptive stage of the research.

11

3.3 Research Design

Research designs are generally classified into three broad categories: descriptive, exploratory and causal•comparative (Burns, Veeck, & Bush, 2017). The descriptive research approach is a method that focuses on describing the nature of a particular condition as it exists during the time of the study. It involves an observational basis where attributes of a particular phenomenon are identified, or where the correlation between phenomena are explored (Leedy & Ormrod, 2001). Exploratory research is unstructured and informal research, and is generally conducted when the researcher has limited knowledge about the research problem, and where there is a need for additional and more recent information. Causal•comparative research is on the other hand undertaken to measure the causality in relationships between variables (Burns et al., 2017), and will therefore not be included in this study.

Based on the research question, this study was conducted in two stages: an exploratory qualitative research stage, followed by a descriptive quantitative research stage. The exploratory research was undertaken in the outset of the research project to efficiently review previous research through library and online sources. Also, it assisted in the collection of current information about the general nature of the research problem, which was necessary in order to conduct a descriptive research design. The results of the exploratory approach could then be quantified and described with more detail, thereby reflecting the application of a descriptive approach. Conducting research in this manner allowed for the exploratory qualitative phase to serve as a foundation for the descriptive quantitative phase, as well as providing first-hand knowledge for the research questions.

3.3.1 Stage 1•Exploratory Qualitative Research

In the exploratory qualitative stage data was collected from focus groups, which is a useful technique for gaining individual insights on the research problem and to generate fresh ideas from a limited sample of respondents (Burns et al., 2017). Thus, a total number of three focus groups were conducted to ensure that the results were consistent and rigorous. The findings from the focus groups were then used to guide the second phase of the research where a questionnaire was developed and distributed.

In the development of the questionnaire, findings from the focus groups were combined with the findings obtained from academic papers and journals. The secondary sources were used in order to take advantage of questions appearing in questionnaires of prior research and to ensure that the requirements of face validity were met to a reasonable extent. Face validity, in this case, ensures that the questions measure what they are supposed to through high relevance and that desired topics are covered (Fallowfield, 1995; Rowley, 2014). Hence, the purpose of the focus groups was to generate

12

fresh ideas and to supplement findings obtained from research papers in order to manifest a degree of contextualisation.

3.3.2 Stage 2•Descriptive Quantitative Research

For the quantitative descriptive stage, a cross•sectional sample survey was conducted using structured computer•administered questionnaires. A cross•sectional survey design is one of the most common designs applied in research (Olsen & St. George, 2004). Its purpose is to measure units from a representative sample of the population, only at one point in time (Salkind, 2010, Burns et al., 2017). Based on the work of dos Santos Silva (1999), cross•sectional surveys can be appropriate for measuring a populations’ attitudes, beliefs, practices, and knowledge concerning a certain phenomenon. Even though it is not possible to obtain information about changes in behaviour over a period of time, cross•sectional studies offer cost efficiency, as the method does not require a lot of time and can be used to examine several outcomes (Salkind, 2010).

Moreover, the purpose of using a structured computer•administered questionnaire was due to its major advantages: ease of administration, standardisation and the reduction of interview evaluation apprehension, i.e. when respondents feel that they have to provide answers that are “right” or “desirable”, and also to facilitate the process of analysing and keeping the study within budget (Burns et al., 2017).

To summarise, this study can be considered to use a combination of exploratory and descriptive research design. In the exploratory stage, an extensive literature review and multiple focus groups were conducted. While in the descriptive stage, a cross•sectional computer•administered questionnaire was used to collect quantitative data.

13

3.4 Population & Sample

A population can be defined as the total number of objects in a dataset, which shares a specific characteristic and can be analysed through collecting and analysing data. In most cases, collecting data from an entire population is simply not possible, while in other cases, vastly resource demanding. Samples of the entire population are therefore used, which can be described as a representable portion, or a subset of the population. When studying a sample, it is important to ensure that it is representable of the entire population, which is determined by the suitability of the used sampling method (Burns et al., 2017). However, a sample can always be regarded as inaccurate. Thus, the aim should be to eliminate the significance of the inaccuracy, which can be accomplished by increasing the sample size (Burns et al., 2017).

The population targeted for this study is Swedish individuals, both male and female, within the age gap of 18 to 29 year-olds. This target population amounts to almost 1.4 million individuals (Valmyndigheten, 2018). This population is amongst the heaviest users of the four SVOD•services included in this study (Mydigheten för Press, Radio och TV, 2017), and constitutes around 18,4% of the total Swedish population over 18 years of age. (Valmyndigheten, 2018). Out of the total Swedish population, more than 50 percent have access to, and use SVOD•services on a daily basis (Mydigheten för Press, Radio och TV, 2017).

3.4.1 Sample Size

Determining a suitable sample size may for any researcher be challenging, as it is generally a settlement between what is perfect in theory and what is manageable in practice (Burns et al., 2017). Even though there are no uniform rules of how a sample size should be calculated, several “rules of thumb” have been brought forward. For example, in the relationship between the number of respondents per variable, Nunnally (1978) recommend a 10:1 ratio, while Stevens (1996) suggests a ratio of 15:1. Green (1991) and Maxwell (2000) independently, found no support for this method and concluded that using this method may lead to a non•sufficient sample size. Maxwell instead suggested between 70 and 100 observations per variable in order to establish correct results. On the other hand, Green (1991), as well as Tabachnick and Fidell (2007), advocated using a method based on 50 observations, plus an additional 8 observations for each independent variable, when testing multiple correlation. Based on the latter, this would entail that this research would require a sample size of 106 respondents, as the measures of service quality consisted of seven constructs, or independent variables. It may be considered naive to believe that all distributed questionnaires will be answered.

14

Thus, 180 potential respondents were asked to answer the survey of which a total of 122 provided complete responses, which translates to a response rate of 67,78%.

3.4.2 Sampling Method

As a master list of the studied population is not possible to obtain, a non•probability sampling method was to be used. A non•probability sampling method utilizes subjective techniques, which entails an unknown probability of participants being selected (Burns et al., 2017). Purposive sampling is a method categorised as non•probability, which implies that the researchers invite certain individuals to participate, based on their shared characteristics, and where the probability of being selected is unknown (Burns et al., 2017).

3.4.3 Sampling Error

It is, however, important to note that due to the sampling method applied in this research, the representativeness of the sample being studied may be questionable. Hence, the ability to make generalisations from the sample to the population is limited. Even though the researchers were confident that the sample selection would yield knowledgeable respondents about the given issue and would be qualified to participate, the inherent bias in the purposive sampling should be considered.

In order to obtain a sample as representative as possible of the target population, despite the chosen sampling method, respondents were drawn from a national homogenous sample on the basis of their age, and existing prior experience of using SVOD•services. Both criteria were of high importance in order to avoid sampling error, since the survey was of post•evaluative nature and that respondents had to fit within the targeted age category. To further improve the representativeness and reduce bias, selecting participants from only one particular subset of the population was avoided. For example, invitations to participate were sent electronically to individuals from diverse parts of the country, both larger cities, as well as a range of smaller towns. Individuals were also selected based on other demographics, such as occupation, as well as parental and relationship status.

15

3.5 Data Collection Methods

In this research, both primary data and secondary sources were used to address the research problem. The secondary sources used constituted of scholarly journal articles that were extensively reviewed, as well as data from governmental institutions. There are several advantages that justify the utilization of secondary data. For example, the data can be inexpensive to collect, obtained quickly, and it may also enhance the primary data by providing supporting evidence (Burns et al., 2017). However, the research may not be up-to-date nor match the researcher’s need. It should also be handled carefully, as previous researchers have analysed and interpreted the data, and subjectiveness may alter the findings.

Primary data refers to information that is being collected for the first time and specifically for the research project (Burns et al., 2017). The primary data collected for this study was done partly through multiple focus groups, where relevant, qualitative data contributed to the second stage of the research process. The findings were, together with theories collected from secondary sources, used to develop a cross•sectional sample survey, which was conducted using structured computer•administered questionnaires. A major drawback with using a survey is that respondents may not answer all questions accurately, due to misinterpretations (Cooper & Schindler, 2011).

3.5.1 Focus Groups for Qualitative Data Collection

A focus group is a frequently used method for collecting qualitative data and consists of a few participants that is guided by a moderator, through an unstructured discussion which can provide the researcher with relevant information for the research problem (Burns et al., 2017).

The objective of the focus groups was to collect consumers’ perceptions, evaluations and feelings towards the four major SVOD•services in Sweden, and to discover plausible drivers of customer satisfaction and brand loyalty. The discussions, in turn, would also be helpful for developing dimensions and question items for the online questionnaire. The ideal size of a traditional focus group is considered to be six to twelve people (Burns et al., 2017). Although there is no correct answer to how many focus groups should be conducted, small projects generally consist of three to four focus groups in order to ensure that results are consistent and rigorous (Burns et al., 2017).

Considering the selection of participants for the focus groups, purposive sampling was used. Purposive sampling refers to when the selection of individuals is based on the judgement of whom the researcher feels represent the population (Burns et al., 2017). As the purpose was to discover

16

plausible drivers of customer satisfaction and loyalty in SVOD•services, it was necessary that the participants were consumers of SVOD•services, and purposive sampling we therefore considered appropriate.

In order to guide the discussion and maintain focus on the topic, a semi•structured interview technique was applied. Such a technique can uncover valuable descriptive data regarding participants’ perspectives, and personal experiences (Schensul & LeCompte, 2013). Therefore, this technique was considered appropriate for the development of variables and items of service quality, that previous research failed to recognise, and which may contribute to customer satisfaction and brand loyalty.

In the finishing stages of each focus group, the discussion leader conducted a member check, where the main topics brought up during the discussion were summarized, for the focus group members to approve, explain or rephrase. This was done in order to ensure that the points were interpreted as intended by the participants (Harper & Cole, 2012).

After each session, the transcripts were analysed through a thematic analysis, where the researchers partly were searching for repeated points and themes not examined by previously reviewed literature. Analysing focus groups thematically can result in great advances since the method allows for flexible, yet richly detailed conclusions to be drawn (Braun & Clarke, 2006). However, researchers should utilize this method in a careful manner since flexibility may lead to inconsistency when deriving appropriate themes from the focus groups (Holloway & Todres, 2003).

Moreover, the focus groups were conducted by sending out requests in student groups and to specific individuals who are familiar with the research topic. The sessions took place in group rooms at Jönköping University, as this was considered to be a familiar setting for all participants. Beverages and snacks were provided for the participants. In total, three focus groups of eight participants were carried out over one week. All participants of the three groups were within the targeted age category, however, the youngest individual attending was 19 years old, and the oldest 27. Thus, the entire target population was not represented during the conducted focus groups. Two researchers acted as moderators during the discussions in order to guarantee the coverage of all questions formulated prior to the meetings. The remaining researcher made sure that the discussions were recorded and that main points were transcribed at the time of the discussion. Before initiating conversation, all participants signed a form of consent which clearly stated why their opinions were desired, the purpose of the study, and information regarding their rights. The information sheet and the form of consent are available in appendix (10.1.1; 10.1.2).

17

The discussions were initiated with an explanation of the aim and purpose of the study and why the input from the participants was needed, as well as a brief introduction to the SVOD•industry. Later, a few situational questions were asked to make the participants feel comfortable and further promote conversation, before they were faced with questions regarding their habits, preferences, and feelings with regards to the SVOD•services included in this study.

3.5.2 Online Questionnaire for Quantitative Data Collection

In the quantitative stage of the research, data was collected by the use of an online questionnaire consisting of close ended questions. Questions of this nature can be helpful for increasing the response rate of the questionnaire, as they are quick for respondents to complete, and facilitates the process of coding and analysing (Rowley, 2014). Thus, this research adopted a structured questionnaire with close•ended questions for collecting quantitative data.

3.5.3 Questionnaire Development

The questionnaire development is an important activity in the research process since a poorly developed questionnaire can result in less valid and reliable data. Contrarily, a rigorously designed questionnaire may yield valuable information. Therefore, the questionnaire development used for collecting the quantitative data of this research was linked with relevant literature and guided through the steps presented by Brancato et al. (2006), in “Handbook of Recommended Practices for Questionnaire Development and Testing in the European Statistical System”. According to Brancato et al. (2006), there should be a consistent strategy that can guide the designing and testing of a questionnaire. In questionnaire design and testing, five mains steps need to be covered by the strategy. These steps include conceptualisation, design, testing, revision and data collection (Brancato et al., 2006).

18

3.5.4 Conceptualisation

The first step in designing a questionnaire is to specify the conceptual basis before considering the formulation of questions. In this phase, the objectives and concepts of the research, and the information needed to address it must be determined. In addition, specifying what is to be measured is essential for designing a survey instrument of high standard. Thus, basic concepts such as the target population, sampling, preferable data collection method, and the variables to be measured need to be clarified and defined (Brancato et al., 2006; Harlacher, 2016). Therefore, for this research, the phase included defining the research objectives, developing the conceptual model, and determining measurable variables by conducting an extensive review of relevant literature related to service quality, customer satisfaction, and brand loyalty.

3.5.5 Questionnaire Design

After the conceptual basis was defined, designing a first draft of the questionnaire including wording of concrete questions was formulated, and the process of structuring of the questionnaire took place. Based on the research design, data collection method requirements, and the questionnaire content, the questionnaire was organized into three sections, and is available in appendix (10.1.3).

In the first section, potential respondents were given an introduction to the questionnaire. This was followed by a few screening questions to determine whether the respondent qualified for participation. In addition, a few so-called “warm­up questions” were used at the beginning of the survey to demonstrate the ease of completing the survey and generating the respondent’s interest (Burns et al., 2017). The warm•up questions concerned whether the respondent had any experience with SVOD•services, if they had access to an SVOD•service account, and which service they were most familiar with.

In the second section, a transition statement was given to make the respondent aware of a change in format. In this section, the intention was to capture the respondents’ attitudes, opinions, evaluations, impressions, perceptions, feelings, and intentions toward multiple items representing each underlying constructs. In order to do so, rating scale formats were adapted to allow respondents to translate their feelings and opinions in a clear and convenient way. A suitable tool for this purpose is using a Likert scale, which is an interval scale that can be used to capture the direction, and the degree to which respondents agree or disagree with various statements or questions (Burns et al., 2017). All questions of this type used a 5•point Likert scale, in which only the endpoints were labelled. The reasoning for choosing a 5-point scale rather than providing more options, is to obtain a response rate as high as

19

possible. A scale with more options or not being able to answer neutrally, may lead to the respondent becoming frustrated when answering the survey (Revilla, Saris, & Krosnick, 2013). It is also argued by Revilla et al. (2013) that a 5-point scale provides a higher mean quality than a 7•point scale.

The respondents were also given the alternative to answer the questions of this type with an “I do not know” option. This was included in order to facilitate the process of answering the questionnaire truthfully, however, was removed for the final question of each construct, since they measure the overall satisfaction connected to the respective construct. The measure will in this study be referred to as construct satisfaction. Denying the option to answer “I do not know” for these questions will, at minimum, result in receiving the respondents’ overall impression of each construct. This is important if the respondent chooses the “I do not know” option in several other questions from that construct. For some constructs, the labelling was changed to very satisfied to very dissatisfied, as it was better suited, while still measured on the same range scale.

In the third section, respondents were asked to indicate on a 5•point scale, their overall satisfaction regarding the chosen SVOD•service, as well as to what extent the service has met their expectations. An average of these measures was later used as a dependent variable, since a regression analysis cannot be performed with two dependent variables. This measure will be referred to as overall customer satisfaction. The Net Promoter Score (NPS) was also included in the third section and was used in order to measure the user’s loyalty intentions with their chosen SVOD•service.

Lastly, the end of the final section was reserved for a set of classification questions that were used for classifying respondents into groups based on gender, age, income, and subscription length. The reason for why these classification questions were placed at the end of the survey, was since respondents may have refused to answer them if they would have occurred in an earlier section, as they can be considered personal. Placing the classification questions in the final section, therefore, would allow any refusal to happen at the end of the survey process. This placement is also important as it could remove a potential negative tone towards the survey (Burns et al., 2017).

20

3.5.6 Questionnaire Translation

Since the target population consists of only Swedish individuals, aged 18 to 29, the questionnaire was tailored to suit the knowledge and understanding of the potential respondents in regard to language. Therefore, once all questions were developed, they were carefully translated into Swedish to eliminate potential language barriers and minimize the number of “I do not know” •answers, resulting from language capabilities. The translated version of the questionnaire is available in appendix (10.1.4).

3.5.7 Questionnaire Testing & Revision

When a questionnaire has been designed, a constituting problem remains that one cannot be entirely confident whether the respondents understand the questions, and answer them accurately (Cooper & Schindler, 2011). Before distributing the questionnaire, it is recommended to conduct a pre-test of the questionnaire on a set of representative respondents in order to discover any questionnaire errors. By testing the questionnaire in the early process of developing it, words, phrases, instructions, the flow of questions, or other questionnaire aspects that appear to be confusing, difficult to comprehend, or problematic to the respondent, can be identified and revised (Burns et al., 2017).

3.5.8 Quantitative Data Collection

For collecting the quantitative data, respondents were directed to the questionnaire page, generated through Qualtrics, which is a software for creating surveys and recording survey results. The respondents were presented with information about the study, such as the aim of the research, the included streaming services, as well as contact information to the researchers. The collected data was thereafter exported from Qualtrics to SPSS for further analysis.

3.6 Data Analysis Methods

3.6.1 Correlational Research Design & Multiple Regression

The correlational research design is a type of non•experimental research in which the statistical relationship between two or more variables are measured by using correlational statistics (Creswell, 2013). The reason for applying a correlational research design is to determine if two or more variables of interest are related. The statistical analysis of the correlation can be analysed through a multiple regression analysis, which is a useful tool for examining what factors are related to the dependent variable, as well as how the dependent variable is influenced by each of the factors, and to what

21

degree it is influenced (Burns et al., 2017). Thereby, it is possible to inspect the strength of a linear relationship between variables of interest. Multiple regression analysis was therefore considered as an appropriate statistical method for analysing the relation between service quality, customer satisfaction, and brand loyalty.

This was done by first conducting a multiple regression analysis for each of the seven constructs of service quality, in order to determine the amount of variation of each construct that could be explained by all the items within the constructs explained together. In each of the seven multiple regression models, the items underlying its construct were considered as independent variables, whereas the measure of construct satisfaction was considered as the dependent variable.

Four additional multiple regression models were then conducted in order to test the research hypotheses stated earlier. In the first of these four regression models, all measures of construct satisfaction were considered as independent variables, while the overall customer satisfaction was considered as the dependent variable. This was done in order to test hypothesis 1, and determine if any service quality constructs were significantly, positively associated with overall customer satisfaction. In the second test, the independent variables were unaltered, however, brand loyalty was considered as the dependent variable. This was done in order to test hypothesis 2, and determine if service quality was positively associated with brand loyalty. This was followed by a third regression model to test hypothesis 3, and examine whether overall customer satisfaction was positively associated with brand loyalty. Therefore, overall customer satisfaction was considered as an independent variable, whereas brand loyalty was considered as the dependent variable. In the fourth and last regression model, all measures of construct satisfaction, together with overall customer satisfaction were considered as independent variables, whereas brand loyalty was considered as the dependent variable. This procedure allowed the researchers to test hypothesis 4, and examine the relation between service quality, overall customer satisfaction, and brand loyalty.

3.6.2 Multiple R & Adjusted R2

A useful measure for determining the strength of a linear relationship between variables is the coefficient of determination (R2), also known as multiple R. It ranges from 0 to 1 and refers to the amount of variation in the dependent variable that can be explained by all the dimensions of the independent variable(s). The higher the fraction, the more the dependent variable can be explained by the combined dimensions of the independent variables (Burns, et al., 2017). However, in sampling distribution, multiple R has a tendency to overestimate the true value in the population as the sample size decreases (Tabachnick & Fidell, 2007). In order to provide a better estimate of the true

22

population, adjustments must be made to the multiple R. Hence, the adjusted R square is a modified version of the multiple R which provides an adjustment of this value. Therefore, with respect to the relatively small sample size used in this research, the adjusted R square may be a better indicator to report, concerning the amount of variation in the dependent variable associated or explained by all the independent variables explained together.

3.6.5 Standardized Beta Coefficient

If the sample correlation is statistically significant, the strength of the relationship can be determined based on the size of the correlation, and therefore also the importance of the independent variables with regard to each other. However, as these variables are regularly measured by the use of different units, they must therefore be standardised by normalising the independent variables through obtaining a standardised beta coefficient. A larger absolute value of a standardised beta coefficient indicates a greater relative importance of the value when considering the dependent variable (Burns et al., 2017). A negative number indicates a negative correlation, which entails that investigated variables move in opposite directions. Thus, correlation only assumes a linear relationship between the variables and does not imply causation (Simon & Goes, 2012).

3.6.3 Multicollinearity

Multicollinearity is another important measure to examine as it refers to the existence of either moderate or strong correlations among the independent variables. If multicollinearity is present, the results of the independence assumption of multiple regression analysis will be negatively affected. Therefore, testing and potentially removing multicollinearity is crucial. A way to remove multicollinearity is by using the variance inflation factor (VIF), which determines if multicollinearity is present. VIF is a single number which is generally considered to reject multicollinearity if the number is lower than 10, and therefore if the VIF of a variable is greater than 10, it should be removed and the multiple regression test must be revisited (Burns et al., 2017).

3.7.1 Reliability

When selecting scales to be included in a study, one concern is to test the internal consistency, or reliability of each construct. That is, the extent to which items underlying a construct, are measuring the same construct. In order to measure the reliability of each construct, the Cronbach’s alpha test was run in the analysis. This is a common indicator for internal consistency were a Cronbach’s alpha value equal to or above .7 is required for establishing reliability (Pallant, 2007).

23

3.7.2 Construct Validity

According to Peter and Churchill (1986), the interpretation of the calculated relationship and its strength between latent variables is only relevant if construct validity can be demonstrated. Because construct validity is an assessment of how well a study’s measurement model measures what is it supposed to measure, researchers must apply various subcategories of construct validity, including convergent & discriminant validity, in order to assess study results (Henseler, Ringle, & Sarstedt, 2014).

3.7.3 Convergent Validity

Convergent validity pertains the extent to which measures that theoretically should be measuring the same construct, are in fact more correlated with each other than with measures of other constructs. Evidence of correspondence between measures of constructs can be established by conducting a bivariate correlation test, where the correlation coefficient (represented by the letter r for Pearson’s correlation) can be interpreted as representing an effect size, which is an indicator of the correlation strength between two variables. A correlation coefficient value of approximately .5 is an indicator for evidence of a strong correlation, whereas a value of approximately .3 indicates a moderate correlation, and a value of approximately .2 indicates a weak correlation (Swank & Mullen, 2017). Hence, this assumes that the greater the correlation coefficient between measures of the construct, the better the evidence of correlation. On the other hand, if two items are completely correlated, they would have a correlation coefficient of 1, which would suggest that the two questions have been answered identically. This is not an optimal result, as one question could be seen as overabundant if there are no changes to Cronbach’s alpha after removing one of the questions.

It is, however, essential that the bivariate correlation coefficient is statistically significant, or meaningful. Otherwise, no evidence of convergent validity can be established. To test for statistical significance, the p•value is examined and compared to the level of significance for the analysis. In addition, the hypothesis related to the correlational analysis needs to be tested (Swank & Mullen, 2017).

24

3.7.4 Discriminant Validity

Contrary to convergent validity, discriminant validity reflects the extent to which measures of constructs that theoretically are supposed to be dissimilar, are in fact observed to have a low or weak correlation between one another. To establish evidence of discriminant validity, a common norm is to seek for low correlation between measures, or items that are not supposed to measure the same construct (Swank & Mullen, 2017; Henseler, Ringle & Sarstedt, 2014). As with convergent validity, a bivariate correlation test was conducted in order to establish evidence of discriminant validity.

3.8 Ethical Considerations

In any research study, it is important to implement appropriate ethical principles in order to provide protection for human subjects (Orb, Eisenhauer, & Wynaden, 2001). In this study, potential participants of the focus groups were approached individually and were adequately informed about the research purpose, the research process, and the methods of data collection. It was explained that the participants had the freedom of choice to participate, decline, or withdraw from the study at any time during the process. After a thorough explanation of the research, potential participants were provided with an information sheet concerning the study. Participants were given the appropriate time to carefully read through the study purpose, think of any questions or concerns, and to decide whether they would like to participate or not.

The identities of the participants were not disclosed in any section of the study, thus the anonymity and confidentiality of the participants was maintained. Also, privacy and confidentiality of the focus group environment was managed cautiously over the course of the focus group sessions, data analysis, and dissemination of the research findings. The same applied for the online survey, where respondents were provided with a description of the study, an indication of what kind of questions would be asked, that respondents had the opportunity to decline or withdraw from the online survey at any time, and that confidentiality and anonymity of the respondents was reassured.

25

4. Measurement Instruments

______

In this section, a presentation of each construct and its underlying items representing service quality is given. Also, instruments for measuring overall customer satisfaction and brand loyalty and its underlying items are presented. ______

The purpose of this study was to explore what factors of service quality affect customer satisfaction and brand loyalty in the SVOD•industry. Therefore, it was necessary to obtain multiple observable items from each dimension of service quality, representing relevant aspects of SVOD•services for measuring customer satisfaction indirectly as a latent variable. However, the customer’s attitudes, opinions, evaluations, impressions, and perceptions of different quality variables for SVOD•services (e.g. features, reliability, performance or efficiency) are subjective properties, and cannot be observed directly as they are mental fabrications. As each quality variable cannot be fully covered by a single question, they were considered as latent constructs and were therefore measured indirectly by using multiple observed items, or questions, for each construct. The questions were derived from previous literature, and qualitative findings from the previously conducted focus groups.

Based on the E•SERVQUAL framework, together with focus group findings, seven independent variables, or constructs were identified, representing different quality aspects in the context of SVOD•services. Of the seven constructs, four of them were obtained from the E•SERVQUAL scale (Parasuraman et al., 2005). Although E•SERVQUAL was designed for measuring the service quality of e•commerce websites, it was still considered as an appropriate measurement for this research as the dimensions correlate with the research context. However, some essential items had to be thoughtfully rephrased to conform to the research context.

26

4.1 Constructs of Service Quality

Efficiency

The efficiency construct in E•SERVQUAL is defined as the easiness for the customer to effectively access the website, and the speed of using the site (Parasuraman et al., 2005). Based on the E•SERVQUAL•framework, several items have been adapted to fit the context of SVOD•services. Respondents were asked to indicate on a Likert scale ranging from 1=“strongly disagree” to 5=”strongly agree”, including an option to answer “I do not know”, to what extent their current streaming service fulfilled their demands of a good streaming service when it comes to the following statements:

➢ This SVOD•service makes it easy to find what I am searching for ➢ This SVOD•service is simple to use ➢ This SVOD•service website is well organized ➢ The response time between my actions and the results is fast

One frequently mentioned topic during the focus group sessions was the search function of different streaming services, hence the following item was added to the construct:

➢ The search function is satisfactory

Respondents were also asked to rate their overall customer satisfaction of each construct in order to gather the general attitude towards each construct as a whole. Adding this item to each construct would later be helpful in the regression analysis where the relative importance of the constructs to customer satisfaction were analysed.

➢ Overall, I am satisfied with the [construct] of the SVOD•service

27

System Availability

The system availability construct is referred to as the technical function of the website, especially the extent to which it is available and functioning correctly (Parasuraman et al., 2005). In SVOD•services, this translates to the customers’ ability to access the content whenever and wherever they desire. Due to the need for adaptation, the items were modified to fit the context of SVOD•services. The items were partially derived from themes gathered during the focus group sessions. Respondents were asked to indicate on a scale from 1=“strongly disagree” to 5=”strongly agree”, with an option to answer “I do not know”, to what extent they would disagree or agree with the following statements:

➢ This SVOD•service allows me to stream content on multiple devices at the same time ➢ This SVOD•service allows me to share my account with other users ➢ This SVOD•service lets me access content wherever I want ➢ This SVOD•service lets me access content whenever I want

Privacy

The privacy construct refers to the extent to which customers believe that their personal information is being protected, and the site is secured from intrusion. Based on the E•SERVQUAL•framework (Parasuraman et al., 2005), respondents were asked to indicate on a scale from 1=“strongly disagree” to 5=”strongly agree”, including a “I do not know” option, to what extent they disagreed or agreed with the following statements:

➢ I feel that the SVOD•service ensures that my transactions are safe ➢ I feel that the SVOD•service protects any information regarding my payment ➢ I feel that the SVOD•service protects my personal information ➢ I feel that the SVOD•service does not misuse any of my personal information

28

Fulfilment • Library of Content

The fulfilment construct is the degree to which the sites’ promises are actually performed (Parasuraman et al., 2005). In the context of SVOD•services, this could be considered to be the library of content. The reasoning behind the alteration is due to the fact that the customer signs up for a subscription plan where they pay a certain price in exchange for access to a library of content (Grece, 2014). Since access to content is what the streaming services are promising to deliver, the fulfilment construct would therefore be relevant for measuring how well the streaming services fulfil their promise. Based on the qualitative findings from the focus groups, respondents were asked to indicate on a scale ranging from 1=”very dissatisfied” to 5=“very satisfied”, including an option to answer “I do not know”, on how dissatisfied or satisfied they were with the following:

➢ Selection of content available ➢ Quality of content ➢ Diversity of genres ➢ Selection of original content available ➢ Quality of original content ➢ Actuality of content available ➢ Update frequency of new content ➢ Ability to watch sporting events ➢ Range of sporting events offered ➢ Ability to watch trailers

Beyond the four constructs obtained from E•SERVQUAL, three additional service quality constructs, including recommendation system, quality of experience, and pricing were added. The items underlying these constructs were obtained from further literature sources and focus group findings, in order to reveal respondents’ overall customer satisfaction and brand loyalty towards SVOD•services.

29

Recommendation System

The recommendation system construct was adopted from a previous study by Hasan, Jha, and Liu (2018). The authors explain a recommendation system as a program or system that is designed to recommend content that the user might find interesting and relevant on a platform, based on factors such as the user’s history and preferences (Hasan, Jha, & Liu, 2018). Hence, the following items have been adopted to measure the recommendation system construct on a scale ranging from 1=”strongly disagree” to 5=”strongly agree”, with an additional option to answer “I do not know”.

➢ I frequently watch content that is recommended to me by the SVOD•service ➢ The recommendation system helps me to find relevant content to watch ➢ The recommendation system helps me to save time when searching for content to watch

The following topics were frequently mentioned during the focus group sessions, and regarded the accuracy of the recommendation systems, and whether the user would have found the content without the help from such systems.

➢ The recommendation system makes accurate predictions based on my preferences ➢ The recommendation system is useful as it introduces me to content that I am interested in which I may not have found without it

30

Quality of Experience

The quality of experience (QoE) is referred to as an assessment of the users’ satisfaction regarding the content displayed and the visual and auditory experience. For example, factors subjectively perceived by the user, which may include the quality of video and audio streaming, playback interruption, or noticeable delay after a video has been clicked (Maia, Yehia, & Errico, 2014). The items were developed based on previous literature on QoE (Krishnan & Sitaraman, 2013; Maia, Yehia, & Errico, 2014; MUX, 2017). Respondents were asked to indicate on a scale ranging from 1=”very dissatisfied” to 5=“very satisfied”, including an option to answer “I do not know”, on how dissatisfied or satisfied they were with the following:

➢ The picture quality of videos ➢ The reliability of the SVOD•service ➢ The initial loading and start•up time of videos ➢ The frequency of video freezing and re•buffering ➢ The ability to stream content in 4k Ultra HD resolution

Pricing

The pricing construct was added as the findings from the focus groups indicated that price is an important factor to consider when measuring overall customer satisfaction. Respondents were asked to indicate on a scale ranging from 1=”very dissatisfied” to 5=“very satisfied” with an option to answer “I do not know”, on how dissatisfied or satisfied they were with the following:

➢ The value•for•money of the SVOD•service ➢ Ability to select between various subscription plans ➢ Terms and cancellation policy of subscription ➢ The ability to select between various payment methods

31

4.2 Additional Variables

4.2.1 Overall Customer Satisfaction

In order to measure the overall customer satisfaction of SVOD•services, the following two questions were derived from the American Customer Satisfaction Index and the work of Angelova and Zeqiri (2011). Respondents were asked to indicate their answer on a scale with responses options ranging from 1=”very dissatisfied” to 5=”very Satisfied” for the first question item, and response options 1=”worse than expected” to 5=”better than expected” for the second question item.

➢ What is your overall satisfaction with your SVOD•service delivery? ➢ To what extent has the service met your expectations?

4.2.2 Brand Loyalty

In order to measure the user’s loyalty intentions with their chosen SVOD­service, the Net Promoter Score was applied (Reichheld, 2003). Respondents were asked to answer how likely they are to recommend a company’s product or service to a friend or colleague, by rating it on a scale ranging from 0 to 10. The higher the rating, the more inclined the subject is to recommend the company to a friend or colleague.

➢ How likely is it that you would recommend this specific SVOD•service to a friend or colleague?

4.3 Classification Variables

The first and last sections of the questionnaire consisted of a set of classification questions, with the purpose to ensure that all respondents were included in the target audience of this study. Also, these questions allowed for the researchers to obtain additional information on characteristics which may be of high value, in order to be able to understand and explain the cause and presence of certain deviations. The respondents were asked to indicate how many different SVOD•service accounts they have access to, as well as for how long time they have been a subscriber to the chosen SVOD•service. Respondents were also asked to answer questions regarding gender, age, and annual income. However, the disclosure of personal information was optional.

32

5. Statistical Analysis & Results

______

In this section, the data from a sample of 122 respondents is presented and analysed through three statistical methods in SPSS. First, a bivariate correlation analysis was executed, with the purpose of establishing evidence for convergent- and discriminant validity. Second, reliability analysis was used to test Cronbach’s alpha for internal consistency. Third, multiple regression was used to analyse the data, in order to measure the linear relationship between the variables presented in previous sections. ______

Figure 2. visualises the demographic profile of the 122 questionnaire respondents. As can be interpreted from the characteristics description table below, the majority of the respondents had access to between one and three SVOD•service accounts. A majority of the respondents were within the age of 22•25 and were fairly distributed between male and female. Also, a majority of respondents indicated a monthly income below 15 000 SEK, and most respondents had been a subscriber to the chosen SVOD­service for more than one year, or uses someone else's’ account.

Demographic Profile of Respondents (n=122)

Figure 2.

33

5.1 Reliability & Validity

5.1.1 Cronbach Alpha

The Cronbach Alpha reliability measure for each construct is presented below (figure 3.), as well as how many items were measured within each construct, excluding the respective item measuring construct satisfaction. The fractions portrayed under Cronbach Alpha can range from 0 to 1, and a general rule of thumb for acceptance of reliability is values exceeding .7. When considering this assessment, all constructs except for system availability could be considered reliable, indicating that respondents have given similar responses to questions of the same nature. In order to increase the reliability of the construct system availability, the test was run multiple times, removing items that may have affected the outcome. However, the original test proved to be the most reliable and was therefore considered for further analysis.

5.2 Convergent & Discriminant Validity

The bivariate correlation test was conducted in order to provide evidence of convergent validity for the underlying items of each construct. That is, ensuring that items which were supposed to be related, in fact, were observed to be related to one another. Values above zero indicate some degree of positive correlation, whereas exactly 1 indicate a perfect correlation. If two questions were to be perfectly correlated, this would indicate that they may measure the same thing. Therefore, the maximum acceptable level of correlation for this study was set to .8. The test was run within each of the seven constructs of service quality. Results from the test are presented for one construct at a time. The reason why the number of observations differs slightly between each item is because respondents have chosen the “I do not know”­option for some of the questions.

Figure 3.

34

5.2.1 Convergent Validity

Efficiency (Appendix 10.2.1)

For the efficiency construct, all questions were, to some extent, empirically related to each other as values derived from Pearson Correlations test ranged between .203 and .570. Notable is that the observed number for each question is equal to, or above 118, entailing a maximum of four respondents choosing the “I do not know” •option for any specific item. This may be due to the clarity of the questions, or to the high subjective knowledge of respondents.

System Availability (Appendix 10.2.2)

For the system availability construct, a strong correlation was apparent between question one and two, as well as three and four respectively. The fourth question, on the other hand, indicated a non•significant, negative correlation with question one, while the first question also showed a strong, significant relation (.424) with question three. Question one measured how satisfied the respondents were with the ability to stream content on multiple devices simultaneously, while the third and fourth measured the satisfaction connected to whether or not the users were allowed to access content wherever, and whenever they desired respectively. The frequency of respondents choosing the “I do not know” •option for questions within this construct was considered satisfactory since each question received at least 106 accepted answers.

Privacy (Appendix 10.2.3)

The construct of privacy shown to be interesting from various perspectives. The correlation analysis indicated that all questions were strongly correlated with one another, thus ensuring convergent validity, while on the other hand, the number of observations decreased drastically. The questions in this construct measured the respondents’ perceptions toward how the SVOD•service providers handle and protect personal and payment information. This topic may be difficult to answer as the general knowledge regarding how these issues are handled can be considered to be rather low, in comparison to the knowledge regarding other constructs. This may therefore be one factor contributing to the increased usage of the “I do not know” •option.

35

Recommendation System (Appendix 10.2.4)

When considering correlation, the recommendation system construct indicated satisfactory values, ranging from .443 to .739, and thus, ensuring evidence of convergent validity. Interestingly, question two, which was derived from existing literature, and question four, derived from the focus groups, were the two questions showing the highest correlation. No respondents answered with the “I do not know” •option for question one, while the other questions received only one answer of that nature.

Fulfilment • Library of Content (Appendix 10.2.5)

Even though all questions had to be altered somewhat to fit the study context, they all indicated a satisfactory level of correlation, with the exception of question eight and nine. These questions measured the ability to watch sporting events, and the range of sporting events offered. The correlation between question eight and nine was remarkably strong (.923) meaning that the questions measured similar perceptions. All four SVOD•services included in the questionnaire do not offer sports, which may be a reason for the lower correlation with other questions. The frequency of users choosing the “I do not know” •option was also high for these questions, while only one response of this nature was given for question one and two.

Quality of Experience (Appendix 10.2.6)

All questions of this construct were indicated to be highly correlated, however, not excessively, as no value reached a correlation coefficient of .8. Noteworthy is that question five, measuring the satisfaction connected to the ability to watch content in 4k Ultra HD resolution, indicated a satisfactory value of statistical significance, while half of the respondents from the sample chose the “I do not know” •option for this question. This may indicate that the SVOD•services do not provide clear enough information of video quality, or that the respondents do not possess a device capable of viewing content at 4K Ultra HD resolution, resulting in them not paying attention to the feature.

36

Pricing (Appendix 10.2.7)

Also for the construct of pricing, a clear correlation between the items was indicated, ranging between .340 and .720, which further ensures convergent validity. The number of observations was remarkably low, as almost half of all respondents chose the “I do not know” •option for questions three and four. The questions measured the satisfaction connected to the terms and cancellation policy, as well as the user’s ability to select between various payment methods. An explanation to this could be that many users choose the option to add their credit card while signing up for the service, and have the service provider make an automatic transaction every month. Users may find this convenient and are happy with this setup. They may not care or be aware whether the provider offers alternative payment methods or not. Regarding the terms and cancellation policy, users may not be aware of these, since they might not have read the terms upon signing up, and may not have tried to cancel their subscription yet.

5.2.2 Discriminant Validity (Appendix 10.3)

All measures of construct satisfaction were shown to be correlated with each other to a certain extent, however, the strongest correlation found was between efficiency and system availability. The correlation between the two items was .589, which is considered a very high correlation. The remaining items presented a correlation between .21 and .40, which is considered moderate and acceptable. Therefore, evidence of discriminant validity was established for all constructs with the exception of efficiency and system availability (Swank & Mullen, 2017).

37

5.3 Multiple Regression Analyses

After observing the correlation of each item, and the reliability of each construct, the items were all tested against the respective measure of construct satisfaction. This was done in order to establish the effect of which the independent variables, in this case, the different items, influence the dependent variable, construct satisfaction. In other words, it is possible to determine the degree to which the different questions influence construct satisfaction, as well as to what level of certainty. It should be clearly stated that this stage of the analysis only considers the construct satisfaction, and not overall customer satisfaction. Items are categorised into two significance levels, depending on their “sig.­value”. Items with a sig.­value of p < 0,01 are labelled with two asterisks “**” indicating “very significant”, whereas a sig.­value of p < 0,05 is considered “significant”, indicated with only one asterisk “*”. Additionally, all items were checked for multicollinearity, and analysed through the variance inflation factor (VIF). All items in each construct presented a VIF value below 10, resulting in no removal of items due to the absence of multicollinearity.

Efficiency (Appendix 10.4.1)

Figure 4.

For this construct, the response time of the website was indicated to be the most important factor of customer satisfaction in regard to efficiency, with a beta value of β=.283. This was also the item with the highest significance, depicted with two asterisks. The second significant item within this construct is the extent to which the SVOD•service makes it easy for users to find what they need. The three remaining items did not appear to be significant. The adjusted R2 obtained for the construct was 0.362, indicating that only 36,2% of the variance can be explained by all the items within the construct explained together. Despite the fact that two items indicated significant values, none of the values stood out as remarkably high, and the construct variance cannot be considered to be fully explained by the five items explained together. This may be due to the fact that over half of the respondents had been subscribers for more than a year, and can therefore be considered to be familiar with the service and its systems.

38

System Availability (Appendix 10.4.2)

Figure 5.

Both “Accessing content wherever” as well as “Accessing content whenever” indicated to have a very significant effect on system availability, considering their Sig. values: .000 and .001. Of the two measures, the highest standardised coefficient beta value was present in the variable “accessing content wherever” (β=.372), whereas “Accessing content whenever” presented a slightly lower beta value of β=.328. Furthermore, the obtained adjusted R2 of the construct was 0.356 which signified that only 35.6% of the construct variation can be explained by all four items explained together.

The remaining two variables showed weak Sig. values of .421 and .588. The reason for low significance when considering the ability to stream on multiple devices may be a result of the respondents’ lack of variance in device use. For example, some users may only watch content on their TV, deeming the possibility of watching content on their phone insignificant. The low significance in the variable “Sharing Account” may be explained by some users not wishing to share their account, due to security reasons regarding their personal information, such as passwords. Furthermore, 65.32% of the respondents strongly agreed that their SVOD•service allowed them to share their account, which perhaps indicates that all SVOD•services facilitate the process of sharing accounts.

39

Privacy (Appendix 10.4.3)

Figure 6.

For the construct of privacy, protection of payment information showed to be the most significant item. The measure also indicated the largest impact on the construct satisfaction of privacy, with a standardised beta value of β=.395. The item measuring the extent to which the respondents felt that the SVOD•service did not misuse their personal information, also showed to be significant. This measure further indicated to be the second most impactful predictor of the satisfaction of privacy. The two remaining items, however, did not present significant values. The adjusted R2 value of privacy amounted to 0.586, entailing that 58,6 percent of the privacy construct variation can be explained by the four questions explained together.

The questionnaire responses indicate that the respondents are mostly satisfied or do not know whether they are satisfied, concerning the items within the construct of privacy. For example, protection of payment information is the most important and significant item, which may entail that the respondents are more concerned about financials than personal information with regards to privacy. They have a greater ability to monitor that their payments are protected, as it otherwise would be visible on their bank statement. However, it is difficult for users to control whether or not their personal information is protected or misused; they simply have to trust the service provider.

40

Fulfilment • Library of Content (Appendix 10.4.4)

Figure 7.

Regarding fulfilment, multiple variables were found to be insignificant, resulting in only two significant variables, of which one was considered very significant. Quality of original content was considered to be significant, and quality of content was considered very significant, hence its Sig. value of .003. Additionally, quality of content had a standardised coefficient beta value of β=.291, which was the highest of the construct. Furthermore, the adjusted R2 of the fulfilment construct was 0.642, meaning that 64.2% of the construct variation can be explained by all items explained together. The variable “Diversity of genres” obtained a negative coefficient beta value, however, the item was insignificant. Thus, it cannot be concluded that the item “Diversity of genres” has a negative impact on its construct satisfaction. Moreover, the survey results indicate that most respondents are satisfied with the diversity of genres of their chosen SVOD•service, perhaps meaning that SVOD•services are expected to offer varying content, and that the services indeed meet the users’ expectations in this regard.

41

Recommendation System (Appendix 10.4.5)

Figure 8.

Finding relevant content and accuracy of predictions both presented significant values, however, time•saving and usefulness were the only two variables to present Sig. Values below .01, deeming them very significant. Furthermore, Usefulness presented the highest standardised coefficient beta value of β=.333. The adjusted R2 of recommendation system was .702, which defines the construct variables to explain 70.2% of the construct. The frequency of watching content recommended to the user was not deemed significant, as its Sig. value was .686. This may be due to the fact that users are not always aware of whether or not the content that they watch is recommended to them, as this is not always explicitly stated.

42

Quality of Experience (Appendix 10.4.6)

Figure 9.

The measures indicating the highest contribution towards the construct satisfaction proved to be service reliability and the ability to stream content in 4K Ultra HD resolution. This is noteworthy since half the sample chose to answer with the “I do not know”­option for the latter. The vast majority of respondents who gave an acceptable answer to the question of 4K streaming, indicated a high, or very high satisfaction. This indicates that the measure of 4K streaming is vastly divided, which may be a result of the device issue explained earlier. Clearly, streaming content in 4K is an important feature for the users who usually do so, as it impacts their satisfaction with regards to quality of experience. Moreover, reliability of the service may be thought of as an obvious predictor of satisfaction, and the measure indeed proved to be the most impactful. Overall, the items proved to measure the construct rather well, as the adjusted R2 was 0.709. However, since the number of observations for the measure of 4K resolution was rather low compared to most other items, it should be stated that the values for the analysis are not as rigor as they could have been. However, it allowed the researchers to observe the mentioned pattern.

43

Pricing (Appendix 10.4.7)

Figure 10.

Within the pricing construct, value•for•money was the only variable to present very significant results, with a Sig. value of .000, as well as a standardised coefficient beta value of β=.650. As presented in the model above, the adjusted R2 of the construct pricing was 0.542, indicating that 54.2% of the construct variation can be explained by all its items explained together. The three remaining variables may be deemed insignificant due to the fact that at least 33,6% of respondents chose to answer “I do not know” on each of the three questions. This may be since users have signed up a long time ago, and that they no longer remember what subscription plans, terms and cancellation policy, or payments methods are offered.

44

5.3.1 Hypotheses Testing

In order to examine the amount of relation between all constructs of service quality with customer satisfaction and brand loyalty, four multiple regression models were conducted. A summary of each of the four multiple regression analyses are presented in figure 11.

Figure 11.

45

Hypothesis 1 (Appendix 10.5.1)

➢ H0: No constructs of service quality are significantly, positively associated with customer satisfaction ➢ H1: At least one construct of service quality is significantly, positively associated with customer satisfaction

Figure 12.

In the first regression model, the seven constructs of service quality were considered as independent variables whereas the mediator, overall customer satisfaction, was considered as the dependent variable. As shown in figure 12, a value of adjusted R2= 0.581 was obtained in this model. Thus, it could be noted that 58,1% of the variation in overall customer satisfaction could be explained by all seven constructs, explained together in the first regression model.

46

Also, the results of the multiple regression analysis showed that three out of seven constructs of service quality were significantly, positively correlated to overall customer satisfaction. Assessing from significant standardized beta•values, the construct of fulfilment•library of content indicated the highest contribution to overall customer satisfaction with β=.455 and p < 0,01. Pricing showed the second highest contribution with β=.234 and p < 0,01, whereas system availability had the third highest contribution to overall customer satisfaction with β=.183 and p < 0,05. Although the privacy construct had a negative beta value of β= ­.069, the measure was insignificant. Thus, it cannot be concluded that privacy has a negative impact on overall customer satisfaction.

The results indicate that when the construct satisfaction of fulfilment•library of content, pricing and system availability is increased, overall customer satisfaction is also increased. Hence, three constructs of service quality are significantly, positively associated with overall customer satisfaction, and H0 is rejected.

Hypothesis 2 (Appendix 10.5.2)

➢ H0: No constructs of service quality are significantly, positively associated with brand loyalty

➢ H1: At least one construct of service quality is significantly, positively associated with brand loyalty

Figure 13.

47

In the second regression model, the seven constructs of service quality were still considered as independent variables, whereas brand loyalty was considered as the dependent variable. As shown in figure 13, a value of adjusted R2=0.487 was obtained. Thus, it can be noted that 48,7% of the variation in the dependent variable could be explained by the second model.

Considering the standardized beta•values, four of seven constructs of service quality were observed to be significantly, positively correlated to brand loyalty. Judging from figure 13, fulfilment•library of content could be interpreted as having the highest impact on brand loyalty with β=.407 and p < 0,01. The second highest impact on brand loyalty was efficiency with β= .205 and p < 0,05. Quality of experience had the third highest impact on brand loyalty with β= .190 and p < 0,05, whereas recommendation system indicated to have the fourth strongest contribution with β= .156 and p < 0,05.

Because the results indicate that four of seven constructs of service quality have a significantly, positive correlation with brand loyalty. Therefore, with its increase, brand loyalty will also be increased and H0 is rejected.

Hypothesis 3 (Appendix 10.5.3)

➢ H0: Customer satisfaction is not significantly, positively associated with brand loyalty

➢ H1: Customer satisfaction is significantly, positively associated with brand loyalty

Figure 14.

In the third regression model, the mediator, overall customer satisfaction was considered as the independent variable whereas brand loyalty was considered as the dependent variable. As visualised in figure 14, a value of adjusted R2=0.596 was obtained. Thus, it can be noted that 59,6% of the variation in the dependent variable can be explained by the third model.

48

A standardized beta value of β=.774 and p < 0,01 was obtained, indicating that customer satisfaction has a strong positive impact on brand loyalty. Judging from the results, brand loyalty is considerably increased when customer satisfaction is increased, and H0 is rejected.

Hypothesis 4 (Appendix 10.5.4)

➢ H0: Customer satisfaction does not mediate the relationship between service quality and brand loyalty

➢ H1: Customer satisfaction mediates the relationship between service quality and brand loyalty

Figure 15.

The fourth and final regression model treated all measures of construct satisfaction for service quality as independent variables along with overall customer satisfaction, whereas brand loyalty was considered as the dependent variable. In this model, the adjusted R2 obtained was 0.641, indicating that 64,1% of the variation in the dependent variable can be explained by the fourth model. In comparison to the previous regression models, a higher percentage of the variation in the dependent

49

variable can now be explained.

Although brand loyalty has a meaningful linear relation with both overall customer satisfaction and at least certain constructs of service quality, brand loyalty is significantly increased when overall customer satisfaction is increased, when considering the positive beta value of β=.612 and p < 0,01. The regression model for testing hypothesis 2 allowed for 48,7% of the variance in brand loyalty to be explained. However, when adding the overall customer satisfaction variable in the fourth regression model, 64.1% of the variance in brand loyalty could be explained. Hence, the results indicate that the statistical relationship between the constructs of service quality and brand loyalty are partially mediated by customer satisfaction, and H0 is rejected.

Considering the appropriateness of the four regression models discussed above, the highest adjusted R2 is derived from the fourth regression model. In addition, overall customer satisfaction and all constructs of service quality have been taken into consideration as independent variables in the fourth regression model. The results, therefore, indicate that the fourth regression model can be considered as most appropriate, compared to the previous three regression models.

50

6. Conclusion

Based on the online questionnaire that was distributed to 180 potential respondents, whereof 122 completed the questionnaire, research results indicate that service quality has a significant impact on overall customer satisfaction for the sample. In fact, findings indicate that approximately 58% of overall customer satisfaction variation is explained by all seven constructs of service quality explained together. Based on the findings, three constructs of service quality including fulfilment•library of content, pricing, and system availability indicated a significantly positive correlation with overall customer satisfaction. Therefore, H0 for hypothesis 1 was rejected, as there is indeed at least one construct of service quality that is significantly, positively associated with customer satisfaction.

Moreover, research findings also indicated a direct relation between four constructs of service quality and brand loyalty. These constructs were: fulfilment•library of content, efficiency, quality of experience and recommendation system. H0 for hypothesis 2 was, therefore, also rejected; there is indeed a significantly positive association between at least one construct of service quality and brand loyalty.

Concerning hypothesis 3, findings indicated that overall customer satisfaction is positively associated with brand loyalty. Based on the research findings, brand loyalty is considerably increased when overall customer satisfaction is increased. Hence, H0 for hypothesis 3 was rejected and the alternative hypothesis could be accepted, as overall customer satisfaction is significantly, positively associated with brand loyalty.

Furthermore, in the fourth regression model, brand loyalty was considered as the dependent variable and overall customer satisfaction was added to the model as an independent variable together with the satisfaction constructs of service quality. The adjusted R2 then increased from a value of 0.487 to 0.641. This indicated that brand loyalty has a meaningful linear relation with both overall customer satisfaction and at least certain constructs of service quality. Also, brand loyalty significantly increased when overall customer satisfaction was increased. As a result, the relationship between constructs of service quality and brand loyalty is partially mediated by customer satisfaction.

Therefore, H0 for hypothesis 4 was rejected.

51

Lastly, the results of the regression analyses conducted in this research do not show a cause•and•effect relationship between independent and dependent variables. The regression analysis is merely a tool that assumes a statistical relationship and explains how each factor influences the dependent variable, and to what extent. Hence, it is not possible to conclude that improved service quality will result in increased customer satisfaction and brand loyalty. Taking the multiple regression findings into consideration, however, research findings indicate that SVOD•services in the Swedish market should emphasise their efforts toward the significant service quality factors that were identified in this study, as it may improve overall customer satisfaction amongst 18 to 29-year-old users. It may also facilitate the creation of long-term relationships as a way of gaining a competitive advantage.

52

7. Discussion

______

This section presents and discusses implications for theory and practice, as well as how the researchers of this study recommend future research on the subject to be conducted. Limitations for the study are presented along with reflections of how they may have affected the results.

______

Discussion

When the constructs of service quality were considered as independent variables, and overall customer satisfaction as the dependent variable in the first multiple regression analysis, the fulfilment•library of content construct indicated to be the most important driver of overall customer satisfaction. This can be thought of as obvious, since as presented earlier, access to a library of content is what SVOD•services are promising to deliver, and may be the main reason for why a user initiated their subscription. Furthermore, certain items within the construct such as the ability to watch sporting events or trailers received a high number of “I do not know” answers. This may further have affected the significance of the construct’s relation to customer satisfaction, as some of the studied SVOD•services do not offer sporting events or trailers, hence the uncertainty of the respondents.

The pricing construct was indicated to be the second most important driver of overall customer satisfaction. This may be due to the fact that over half (51%) of the respondents claimed to have a monthly income below 15,000 SEK. Hence, this may indicate that price is an important factor when considering a low income target population, as they may be more price sensitive. If respondents would have indicated a higher monthly income, the pricing construct may have had a lower importance. Lastly, one third of the respondents claim to use someone else's account when watching content from a SVOD•service. This may entail that respondents have less knowledge about questions of transaction security, personal information, and the ability to choose between different payment methods, which are included in the pricing and privacy constructs. Furthermore, the fact that multiple respondents use someone else's account may also affect the system availability construct, as it measures aspects such as the ability to share accounts.

53

System availability, the third most significant construct when evaluating customer satisfaction, consisted of two significant items. These items include the ability to access content wherever and whenever the consumer desires. Due to the increase in internet access through the past decades, as well as the development of devices capable of streaming video content, it has allowed for SVOD•services to exploit the opportunity, and build a borderless service. A reason for why consumers of this target audience find these questions important, may be linked back to the generation traits and habits presented earlier. Most users within this age category have grown up with internet, which may have shaped their habits. With portable devices capable of streaming being available, users want to take advantage of the opportunity and become less restricted.

Recommendations system, privacy, quality of experience, and efficiency are the four constructs deemed insignificant when measuring the relation to overall customer satisfaction. Many SVOD•services use complicated algorithms to recommend content to users. Therefore, it may be difficult for consumers to know what content is recommended to them, which may lead to inaccurate interpretations of recommendation systems. Privacy, quality of experience, and efficiency are constructs measuring properties of SVOD•services that the user may expect to perform at a certain level. Additionally, the four services included in this study are ranked as the top four SVOD•services in Sweden. Hence, consumers may therefore perceive these features to perform at a satisfactory level.

Moreover, when the constructs of service quality were considered as independent variables and brand loyalty as the dependent variable in the second multiple regression analysis, the fulfilment•library of content construct was once again indicated to be the most important driver. Repeatedly, the construct may be significant because users may expect that the service provides a large selection of quality content in varying genres in order for the user to remain loyal; otherwise the user would probably cancel their subscription and switch to another service provider, which offers a more desirable library of content.

Efficiency was also indicated to be an important driver of brand loyalty. An explanation for its significance may be that the construct items consist of basic features that a SVOD•service needs in order to function consistently. If the SVOD•service does not offer these features at a satisfactory level, users will probably not continue to use the service for a long period of time.

54

Furthermore, the construct of quality of experience was, amongst other features, measuring the users’ satisfaction regarding the ability to view content at 4k Ultra•HD resolution. However, 50% of the respondents answered with the “I do not know” option. As mentioned previously, 51% of the respondents had a monthly income below 15,000 SEK, which may be an indicator of that users cannot afford, or have access to devices capable of viewing content in 4k Ultra•HD. However, as new technologies develop, prices for these devices decrease, and more individuals have the opportunity to access the feature of 4k Ultra•HD content. As this feature indicated to have a high importance for those who are using it, SVOD•services should promote the high resolution content to users with capable devices, to a greater extent, as the feature indicated to be a driver for both satisfaction and loyalty.

The recommendation system construct was also significantly, positively associated with brand loyalty. One reason for its significance, may be that a recommendation system with a strong algorithm is able to find relevant content for the users, which enables them to discover new content, and continue to use the SVOD•service; hence remain loyal.

However, some constructs of service quality also proved to be insignificant when measuring the correlation with brand loyalty. For example, the pricing construct was not significantly associated with brand loyalty, which may be due to the possibility of lower price elasticity for loyal customers. This entails that the purchasing intentions of loyal customers may be less affected by changes in price. Furthermore, the privacy construct was also indicated to be insignificant. This may be due to users having limited information about how the service provider handles payment information; they expect their information to be safe, unless something suspicious regarding payments occurs. On the other hand, if a SVOD•service were found to be guilty of sharing or misusing users’ private information, the results of these questions would most likely be strongly affected. However, many of the respondents answered similarly to question one and two, which are measuring perceived transaction safety, and payment information protection. Understandably, users may not have different perceptions of the two issues, resulting in a high correlation coefficient. However, the question regarding protection of payment information proved to be significantly more impactful on construct satisfaction, compared to the measure of transaction safety.

55

7.1 Theoretical Implications

Several constructs and items were derived from the framework of E•SERVQUAL, which was originally developed for measuring electronic service quality of websites. Constructs and items were, therefore, added and carefully altered to be more suitable for the research context. Fulfilment•library of content was one of the most altered constructs, as it was originally called fulfilment and regarded a websites delivery of its promise. Library of content was added to the construct, as it is a fundamental promise of a SVOD•service, and thus all questions were tailored. The construct indicated the strongest impact on overall customer satisfaction, and should therefore be included in a future framework which has the intent to measure SVOD•services. However, the items included can, and should be further refined as they for this study managed to explain 64,2 percent of the construct variance. Moreover, the entire pricing construct was added from the qualitative stage of this research. The construct showed to be the second most important satisfaction driver, with value for money as the strongest predictor, which may be due to the aforementioned reasons. However, the construct should be considered for further studies on the subject. Furthermore, this research managed to explain an acceptable degree of overall customer satisfaction for this sample (58,1%), through the seven constructs of service quality. However, through rigorous refinement, this value can certainly be increased as only three of seven constructs were found to be significant. Refinement, removal, and addition of new items and constructs is needed in order to be able to make more generalisable claims.

7.2 Practical Implications

The findings of this study contribute to the understanding regarding what factors of service quality have an impact on customer satisfaction, and brand loyalty of Swedish 18 to 29 year-olds, when considering SVOD•services. The findings indicate that individuals with these characteristics value the library of content highly, where the quality of the offered content showed to be most important. Moreover, value for money indicated to be of severe importance, which may be linked to the general financial state of 18 to 29 year-olds. The numbers indicate that the target population does not value quantity of content, but rather quality. Also, privacy, website organisation, frequency of videos freezing, or easiness to use do not contribute to their satisfaction of a SVOD•service. However, if a service providers were to remove all content of low quality, share users’ personal information, make their site complicated to use, and freeze videos randomly, the likelihood of a customer becoming less satisfied with the service is rather obvious. This exaggeration is made to illustrate the fact that even if values in the models indicate that some factors have a low impact on customer satisfaction, they still carry importance, and should not be disregarded. Users may assume these features to carry a

56

certain standard of quality, and may prove what is suggested by previous research, which is that satisfaction is related to expectations. The findings, or proposed additions and alterations to previously developed frameworks, may be utilised by SVOD•service managers for further research, or in order to facilitate the process of resource allocation. The methods used in the study, the utilised measurement instruments, or parts of the findings may also be of interest for businesses within sectors of media streaming, such as music, podcasts and video.

57

8. Limitations

Despite the potential contribution this research could bring to existing studies of SVOD•services, there are some anticipated, and unanticipated flaws and barriers which present suggestions for further research. The reader should note that one important limitation which may have influenced the analysis of the collected data is the sample bias in the chosen sampling method. Because the study is based on data from a national sample that was selected based on a non•probability sampling method, the representativeness of the sample may be questionable. However, the sampling method was selected due to the limited ability to gain access to a master list. Although efforts were made to enhance the representativeness of the sample being studied, care should still be exercised in generalising the findings to the entire population as well as to other populations and countries. Also, studying a greater sample size would have enhanced the ability to generalise the findings.

Another important limitation of this research is the concern for non•response surveys. The appropriate sample size was determined to include 106 samples. However, as it was anticipated that not everyone would respond to the questionnaire, a total of 180 questionnaires were sent out. This resulted in a total of 122 completed questionnaires and a response rate of 67,78%. Hence, approximately 28% did not respond to the questionnaire. This was considered as a limitation because there might be important differences between those who completed the survey and those who did not, which could be significant for the results of the study. For example, the study failed to reasonably represent the age category in an equal manner, as 75% of respondents were between 22 and 25 years old. This captures the midway of the target population in terms of age, however limits the information gathered from ages 18 to 21 and 26 to 29. Optimally, the respondents should have been distributed equally between all ages within the targeted age category.

A third limitation concerns the measures that were adopted to the research. In order to comply with the research context, some of the items had to be carefully rephrased and many items had to be developed. Hence, the validity and reliability of some measurements may have been reduced. Also, the items added from the focus groups indicated varying levels of importance and significance. Since these had to be developed by the researchers, some may have failed to capture the entirety of the respondents’ attitudes.

58

Also, as the questionnaire was translated from English to Swedish, some translations may have affected the respondents’ interpretation of the questions. Since the questions were not translated word•to•word, but instead to match what the English questions were intended to investigate, the translation process may have resulted in minor differences in terms of question interpretation from the perspective of the respondent.

8.1 Suggestions for Further Research

Further research on the subject should adopt, and refine constructs as well as items utilized in this study. As services which do not rely on subscriptions were disregarded from this study, future research could investigate this area to compare the two. A qualitative study with a company perspective could be conducted, to identify how executives manage these questions. The research could focus on which factors of service quality the executives believe are of most importance, and investigate whether resources are allocated accordingly. Perhaps other methods to achieve loyal customers are utilized, that this study failed to recognise.

59

9. References

Ahmed, S., Chandio , Z. U., & Quresh, M. A. (2015). Brand trust, customer satisfaction and Brand Loyalty - A cross Examination. Journal of Business Strategies(09), 62-81. Anderson, E. W., & Sullivan, M. W. (1993). The Antecendents and Consequences of Customer Satisfaction for Firms. Marketing Science, 12(2), 125-143. Angelova, B., & Zeqiri, J. (2011, October). Measuring Customer Satisfaction with Service Quality Using American Customer Satisfaction Model (ACSI Model). International Journal of Academic Research in Business and Social Sciences, 1(3), 232-258. AudienceProject. (2017). Insights 2017 - Traditional TV & Streaming in the Nordics, UK & US. Copenhagen: AudienceProject. Barnes, S. J., & Vidgen, R. T. (2002). An Integrative Approach to the Assessment of E-Commerce Quality. Journal of Electronic Commerce Research, 3(3), 114-127. Bloemer, J. M., & Kasper, H. D. (1995, July). The Complex Relationship Between Consumer Satisfaction and Brand Loyalty. Journal of Economic Psychology, 16(2), 311-329. Bolton, R. N. (1998). A Dynamic Model of the Duration of the Customer's Relationship With A Contiunuous Service Provider: The Role of Satisfaction. Marketing Science, 17(1), 45-65. Bolton, R. N., & Drew, J. H. (1991). A Longitudinal Analysis of the Impact of Service Changes on Customer Attitudes. Journal of Marketing, 55(1), 1-10. Bondad-Brown, B. A., Rice, R. E., & Pearce, K. E. (2012). Influences on TV Viewing and Online User-shared Video Use: Demographics, Generations, Contextual Age, Media Use, Motivations, and Audience Activity. Journal of Broadcasting and Electronic Media, 56(4), 471-493. Boonghee, Y., & Donthu, N. (2001). Developing a Scale to Measure the Perceived Quality of an Internet Shopping Site. Quarterly Journal of Electronic Commerce, 2(1), 31-46. Borja, K., Dieringer, S., & Daw, J. (2014, December 18). The Effect of Music Streaming Services on Music Piracy Among College Students. Computers in Human Behavior, 45(1), 69-76. Brady, M. K., & Robertson, C. J. (2001). Searching for a Consensus on the Antecedent Role of Service Quality and Satisfaction: An Exploratory Cross-National Study. Journal of Business Research, 51(1), 53-60. Brancato, G., Macchia, S., Murgia, M., Signore, M., Simeoni, G., Blanke, K., . . . Hoffmeyer-Zlotnik, J. (n.d.). Handbook of Recommended Practices for Questionnaire Development and Testing in the European Statistical System. 2006: European Commission Grant Agreement 2004103000002. Braun, V., & Clarke, V. (2006, July). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101. Burns, C. A., Veeck, A., & Bush, R. F. (2017). Marketing Research (Vol. 8). Harlow, England: Pearson Education Limited. Cardozo, R. N. (1965). An Experimental Study of Customer Effort, Expectation, and Satisfaction. Journal of Marketing Research, 3, 244-249. Cha, J., & Chan-Olmsted, S. M. (2012). Substitutability between Online Video Platforms and Television. Journalism & Mass Communication Quarterly, 89(2), 261-278. Churchill, G. A., & Brown, T. J. (2004). Basic Marketing Research. Ohio: South-Western College Pub.

60

Collis, J., & Hussey, R. (2014). Business Research: A Practical Guide for Undergraduate and Postgraduate Students. Houndmills, Basingstoke, Hampshire: Red Globe Press. Cooper, D. R., & Schindler, P. S. (2011). Business Research Methods. Boston: McGraw Hill. Creswell, J. W. (2013). Research Design. Thousand Oaks, California, United States of America: SAGE Publications. Cronin Jr., J. J., & Taylor, S. (1992, July). Measuring Service Quality - A Reexamination And Extension. Journal of Marketing, 56(55-68), 55-68. Dodds, W. B., Monroe, K., & Grewal, D. (1991). Effects of Price, Brand, and Store Information on Buyers' Product Evaluations. Journal of Marketing Research, 28(3), 307-319. dos Santos Silva, I. (1999). Cross-sectional surveys. Lyon: France International Agency for Research on Cancer. European Parliament. (2018, 03 26). Streaming without borders: EU rules to allow using online subscriptions abroad . Retrieved from News - European Parliament: http://www.europarl.europa.eu/news/en/headlines/economy/20170515STO74834/streaming-without-borders- eu-rules-to-allow-using-online-subscriptions-abroad Fallowfield, L. (1995). Questionnaire Design. Archives of Disease in Childhood, 76-79. Feldman, M. P., & Tavassoli, S. (2014). Something New: Where do new industries come from? Karlskrona: Center for Innovation and Technology Research. Fornell, C. (1992, January). National Customer Satisfaction Barometer: The Swedish Experience. Journal of Marketing, 56, 6-21. Geck, C. (2006, February). The Generation Z Connection: Teaching Information Literacy to The Newest Net Generation. Teacher Librarian, 33(3), 19-23. Gorondutse, A. H., & Hilman, H. (2014). Mediation Effect of Customer Satisfaction on the Relationships Between Service Quality and Customer Loyalty in the Nigerian Foods and Beverages Industry: Sobel Test Approach. International Journal of Management Science and Engineering Management, 9(1), 1-8. Gotlieb, J. B., Grewal, D., & Brown, S. W. (1994). Customer Satisfaction and Perceived Quality: Complementary or Divergent Constructs? Journal of Applied Psychology, 79(6), 875-885. Grönroos, C. (1988). Service Quality: The Six Criteria of Good Perceived Service Quality. Review of Business, 9(3), 10- 13. Grece, C. (2014). On-Demand Audiovisual Markets in the European Union. Strasbourg: European Audiovisual Observatory. Green, S. B. (1991). How Many Subjects Does It Take To Do A Regression Analysis. Multivariate Behavioral Research, 499-510. Gundersen, M. G., Heide, M., & Olsson, U. H. (1996, April). Hotel Guest Satisfaction Among Business Travelers: What are the Important Factors. The Cornell Hotel and Restaurant Administration Quarterly, 37(2), 72-81. Gupta, S., Lehmann, D. R., & Stuart, J. A. (2004, February). Valuing Customers. Journal of Marketing Research, 41(1), 7-18. Häubl, G., & Trifts, V. (2000, February). Consumer Decision Making in Online Shopping Environments: The Effects of Interactive Decision Aids . Marketing Science, 19(1), 4-21. Harlacher, J. (2016). An educator’s guide to questionnaire development. Institute of Education Sciences, U.S. Department of Education. The National Center for Education Evaluation and Regional Assistance. Harper, M., & Cole, P. (2012). Member Checking: Can Benefits Be Gained Similar to Group Therapy? The Qualitative Report, 17(2), 510-517.

61

Hasan, M. R., Jha, A. K., & Liu, Y. (2018, March). Excessive Use of Online Video Streaming Services: Impact of Recommender System Use, Psychological Factors, and Motives. Computers in Human Behavior, 80, 220-228. Hellier, P. K., Carr, R., Geursen, G. M., & Rickard, J. A. (2003, December). Customer Repurchase Intention: A General Structural Equation Model. European Journal of Marketing, 37(11/12), 1762-1800. Hennig-Thurau, T., & Klee, A. (1997, December). The Impact of Customer Satisfaction and Relationship Quality on Customer Retention: A Critical Reassessment and Model Development. Psychology & Marketing, 14(8), 737- 764. Henseler, J., Ringle, C. M., & Sarstedt, M. (2014, August). A new criterion for assessing discriminant validity in variance- based structural equation modeling. Journal of the Academy of Marketing Science, 43, 115-135. Holloway, I., & Todres, L. (2003, December). The Status of Method: Flexibility, Consistency and Coherence. Qualitative Research, 3(3), 345-357. Homburg, C., & Giering, A. (2001, January). Personal Characteristics as Moderators of the Relationship Between Customer Satisfaction and Loyalty - An Empirical Analysis. Psychology & Marketing, 18(1), 43-66. Hui, K.-L., & Png, I. (2003). Piracy and the Legitimate Demand for Recorded Music. Contributions to Economic Analysis & Policy, 2(1), 1-22. Juluri, P., Tamarapalli, V., & Medhi, D. (2016). Measurement of Quality of Experience of Video-on-Demand Services: A Survey. IEEE Communication Surveys & Tutorials, 18(1), 401-418. Keiningham, T. L., Aksoy, L., Cooil, B., & Hsu, M. (2007). A Longitudinal Analysis of Customer Satisfaction and Shade of Wallet: Investigating the Moderating Effect of Customer Characteristics. Journal of Marketing, 71(1), 67-83. Keiningham, T. L., Aksoy, L., Cooil, B., Wallin Andreassen, T., & Williams, L. (2008, March). A Holistic Examination of Net Promoter. Journal of Database Marketing & Customer Strategy Management, 15(2), 79-90. Kenton, W. (2018, April 18). Freemium. Retrieved from Investopedia.com: https://www.investopedia.com/terms/f/freemium.asp Kotler, P. (2000). Marketing Management: The Millenium Edition. Upper Saddle River: Prentice Hall. Krishnan, S. S., & Sitaraman, R. K. (2013, December). Video Stream Quality Impacts Viewer Behavior: Inferring Causality Using Quasi-Experimental Designs. IEEE/ACM Transactions on Networking, 21(6), 2001-2014. Kristensen, K., & Eskildsen, J. (2014). Is the NPS a trustworthy performance measure? Bingley, 26(2), 202-214. Kristensen, K., & Westlund, A. (2004). "The One Number You Need To Grow: (Comment On Reichheld)". Harvard Business Review, 82(4), 136-137. Kumar, V., Pozza, I. D., & Ganesh, J. (2013, September). Revisiting the Satisfaction-Loyalty Relationship: Empirical Generalizations and Directions for Future Research. Journal of Retailing, 89(3), 246-262. Lee, C. C., Nagpal, P., Ruane, S. G., & Lim, H. (2018). Factors Affecting Online Streaming Subscriptions. Communications of the IIMA, 16(1), 1-24. Leedy, P. D., & Ormrod, J. E. (2001). Practical Research: Planning and Design (9th Edition ed.). Upper Saddle River, NJ: Pearson Education, Inc. Li, L., Zheng, L., Yang, F., & Li, T. (2014). Modeling and Broadening Temporal User Interest in Personalized News Recommendation. Expert Systems with Applications, 41, 3168-3177. Loiacono, E., Watson, R. T., & Goodhue, D. (2007, April). WebQual: A Web Site Quality Instrument. International Journal of Electronic Commerce, 11(3), 51-87. Maia, O. B., Yehia, H. C., & de Errico, L. (2014, February). A Concise Review of the Quality of Experience Assessment for Video Streaming. Computer Communications, 57, 1-12.

62

Maxwell, S. E. (2000). Sample size and multiple regression analysis. Psychological Methods, 5(4), 434-458. McQuitty, S., Finn, A., & Wiley, J. B. (2000, October). Systematically Varying Consumer Satisfaction and its Implications for Product Choice. Journal of the Academy of Marketing Science, 10, 1-16. Montana, P. J., & Petit, F. (2008). Motivating Generation X and Y On The Job And Preparing Z. Global Journal of Business Research, 2(2), 139-148. Morgan, N. A., & Rego, L. L. (2004). "The One Number You Need To Grow: (Comment On Reichheld)". Harvard Business Review, 82(4), 134-136. MUX. (2017). 2017 Video Streaming Perceptions Report. San Fransisco: MUX. Mydigheten för Press, Radio och TV. (2017). Mediekonsumtion. Stockholm: Myndigheten för Press, Radio och TV. Nagy, R. (2012). Teaching Generation Y. Acta Universitatis Sapientiae. Social Analysis, 2(1), 97-110. Newman, J. W., & Werbel, R. A. (1973, November). Multivariate Analysis of Brand Loyalty for Major Household Appliances. Journal of Marketing Research, 10(4), 404-409. Nunnally, J. C. (1978). Psychometric Theory. New York: McGraw-Hill. Oliver, R. L. (1980). A Cognitive Model of the Antecendents and Consequenses of Satisfaction Decisions. Journal of Marketing Research, 460-469. Oliver, R. L. (1999). Whence Consumer Loyalty? Journal of Marketing, 63, 33-44. Olsen, C., & St. George, D. M. (2004). Cross-Sectional Study Design and Data Analysis. Young Epidemiology Scholars Program. Orb, A., Eisenhauer, L., & Wynaden, D. (2001). Ethics in Qualitative Research. Journal of Nursing Scholarship, 33(1), 93-96. Organisation for Economic Co-Operation and Development. (2007). The Economic Impact of Counterfeiting and Piracy. Paris: OECD. Pallant, J. (2007). SPSS Survival Manual: A Step by Step Guide to Data Analysis Using SPSS (3rd Edition ed.). Maidenhead, Berkshire, England: Open University Press. Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1985). A Conceptual Model of Service Quality and Its Implications for Future Research. Journal of Marketing, 49(4), 41-50. Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1988). SERVQUAL: A Multiple-Item Scale for Measuring Consumer Perceptions of Service Quality. Journal of Retailing, 64(1), 12-40. Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1996, April). The Behavioral Consequences of Service Quality. Journal of Marketing, 60(2), 31-46. Parasuraman, A., Zeithaml, V. A., & Malhotra, A. (2005, February). E-S-QUAL: A Multiple-Item Scale for Assessing Electronic Service Quality. Journal of Service Research, 7(3), 213-233. Park, D. H., Choi, I. Y., Kim, H. K., & Kim, J. K. (2011). A Review and Classification of Recommender Systems Research. International Proceedings of Economics Development and Research, 5, 290-294. Patterson, P. G., Johnson, L. W., & Spreng, R. A. (1996, December). Modeling The Determinants Of Customer Satisfaction For Business-To-Business Professional Services. Journal of the Academy of Marketing Science, 25(1), 4-17. Peter, P. J., & Churchill, G. A. (1986). Relationships among Research Design Choices and Psychometric Properties of Rating Scales: A Meta-Analysis. Journal of Marketing Research, 23(1), 1-10. Poort, J., Quintais, J. P., van der Ende, M., Yagafarova, A., & Hageraats, M. (2018). Global Online Piracy Study. Amsterdam: Institute for Information Law.

63

Reichheld, F. F. (1994). Loyalty and the Renaissance of Marketing. Marketing Management, 2(4), 10-20. Reichheld, F. F. (2003). The One Number You Need to Grow. Harvard Business Review, 81(12), 46-54. Reichheld, F. F., & Sasser, E. (1990). Zero Defections: Quality Comes to Services. Harvard Business Review, 68, 105- 111. Revilla, M. A., Saris, W. E., & Krosnick, J. A. (2013, December). Choosing the Number of Categories in Agree–Disagree Scales. Sociological Methods & Research, 43(1), 73-97. Rodríguez, D. Z., Rosa, R. L., Costa, E. A., Abrahão, J., & Bressan, G. (2014). Video Quality Assessment in Video Streaming Services Considering User Preference for Video Content. IEEE Transactions on Consumer Electronics, 60(3), 436-444. Rowley, J. (2014). Designing and using research questionnaires. Management Research Review, 37(3), 308-330. Salkind, N. J. (2010). Cross-Sectional Design. Encyclopedia of Research Design, 314. Schensul, J. J., & LeCompte, M. D. (2013). Essential Ethnographic Methods. A mixed methods approach. Lanham, Maryland: Rowman & Littlefield Publishers, Inc. Schirmer, N., Ringle, C. M., Gudergan, S. P., & Feistel, M. S. (2016). The link between customer satisfaction and loyalty: the moderating role of customer characteristics. Journal of Strategic Marketing(26), 1-20. Shatto, B., & Erwin, K. (2016, June). Moving on From Millennials: Preparing for Generation Z. The Journal of Continuing Education in Nursing, 47(6), 253-254. Shunmuga, K. S., & Sitaraman, R. K. (2013, December). Video Stream Quality Impacts Viewer Behavior: Inferring Causality Using Quasi-Experimental Designs. IEEE/ACM Transactions on Networking, 21(6), 2001-2014. Simon, M. K., & Goes, J. (2012). Correlational Research. Dissertation and Scholarly Research. Southgate, D. (2017, June). The Emergence of Generation Z And Its Impact in Advertising. Journal of Advertising Research, 57(2), 227-235. Stevens, J. (1996). Applied multivariate statistics for the social sciences. Mahwah, New Jersey: Lawrence Erlbaum Associates. Stuart, B. J., & Vidgen, R. T. (2002). An Integrative Approach to the Assessment of E-Commerce Quality. Journal of Electronic Commerce Research, 3(3), 114-127. Su, G.-M., Su, X., Bai, Y., Wang, M., Vasilakos, A. V., & Wang, H. (2016, July). QoE in Video Streaming Over Wireless Networks: Perspectives and Research Challenges. Wireless Networks, 22(5), 1571-1593. Swank, J. M., & Mullen, P. R. (2017). Evaluating Evidence for Conceptually Related Constructs Using Bivariate Correlations. Measurement and Evaluation in Counseling and Development, 50(4), 270-274. Svenskt Kvalitetsindex. (2018). Strömmad TV Gillas Mest. Stockholm: SKI. Tabachnick, B. G., & Fidell, L. S. (2007). Using Multivariate Statistics (5th Edition ed.). Boston: Pearson Education, Inc. Torres Vega, M., Perra, C., & Liotta, A. (2018, June). Resilience of Video Streaming Services to Network Impairments. IEEE Transactions on Broadcasting, 64(2), 220-234. Trowbridge, H. (2013). Contemporary Film Distribution and Exhibition: A Review of Recent Studies. New Review of Film and Television Studies, 11(2), 224-234. Valmyndigheten. (2018, September 14). Val till riksdagen - Ålder och kön. Retrieved from val.se: https://data.val.se/val/val2018/alkon/R/rike/alderkon.html Yoo, B., & Donthu, N. (2001). Developing a Scale to Measure the Perceived Quality of an Internet Shopping Site (Sitequal). Quarterly Journal of Electronic Commerce, 2(1), 31-46.

64

10. Appendices

10.1 Questionnaire

10.1.1 Information Sheet

65

10.1.2 Form of Consent

66

10.1.3 Questionnaire in English

Have you previously used a subscription based video-on-demand service(SVOD-service) for streaming online content? (Netflix, HBO, Viaplay, Cmore) • Yes • No

Do you have a registered account with more than one SVOD-service? If so, please indicate how many. • 1 • 2 • 3 • 4 • I do not have access to any SVOD-service

Which of the following SVOD-services do you have most experience with? • Netflix • HBO • Viaplay • Cmore

We will now ask you a couple of questions about your chosen SVOD-service. Below are a number of statements and questions, please read each one and indicate on a scale 1-5 (1=”strongly disagree” to 5=”strongly agree”) to what extent your current streaming service fulfils your demands of a good streaming service when it comes to the following statements:

Efficiency • This SVOD-service makes it easy to find what I need • The response time between my actions and the results is fast • This SVOD-service is simple to use • This SVOD-service website is well organized • The search function is satisfactory

• Overall, I am satisfied with the efficiency of the SVOD-service

System availability • This SVOD-service allows me to stream content on multiple devices at the same time • This SVOD-service allows me to share my account with other users • This SVOD-service lets me access content wherever I want • This SVOD-service lets me access content whenever I want

• Overall, I am satisfied with the system availability of the SVOD-service

Privacy • The SVOD-service ensures that my transactions are safe • The SVOD-service protects any information regarding my payment • The SVOD-service protects my personal information • The SVOD-service does not misuse any of my personal information

• Overall, I am satisfied with the privacy of the SVOD-service

67

Recommendation System • I frequently watch content that is recommended to me by the SVOD-service • The recommendation system helps me to find relevant content to watch • The recommendation system helps me to save time when searching for content to watch • The recommendation system makes accurate predictions based on my preferences • The recommendation system is useful as it introduces me to content that I am interested in which I may not have found without it

• Overall, I am satisfied with the recommendation system of the SVOD-service

The following section will ask about your satisfaction towards your SVOD-service. Please indicate the degree to which you are satisfied or dissatisfied with each of the following statements on a scale from 1=Very dissatisfied to 5=Very satisfied:

Fulfilment - Library of Content • Selection of content available • Quality of content • Diversity of genres • Selection of original content available • Quality of original content • Actuality of content available • Update frequency of new content • Ability to watch sporting events • Range of sporting events offered • Ability to watch trailers

• Overall, how satisfied are you with the library of content of the SVOD-service?

Quality of Experience • The picture quality of videos • The reliability of the SVOD-service • The initial loading and start-up time of videos • The frequency of video freezing and re-buffering • The ability of the SVOD-service to stream in 4k Ultra HD resolution

• Overall, how satisfied are you with the quality of experience of the SVOD-service?

Pricing • The value-for-money of the SVOD-service • Ability to select between various subscription plans • Terms and cancellation policy of subscription • The ability to select between various payment methods

• Overall, how satisfied are you with the value-for-money that the SVOD-service offers?

68

Customer Satisfaction

• What is your overall satisfaction with your SVOD-service delivery? • To what extent has the service met your expectations?

Customer Loyalty Intentions Please indicate your response to the following statement on a 1-10-point scale that best reflects your opinion. (1 = Not at all likely, 10 = Extremely likely)

• How likely is it that you would recommend this specific SVOD-service to a friend or colleague?

Please indicate your age in years: • Younger than 18 • 18-21 • 22-25 • 26-29 • Older than 29

Please indicate your gender: • Male • Female • Other

Please indicate your monthly income: • Under 15,000SEK • 15,000SEK-23,000SEK • 23,000SEK-30,000SEK • Over 30,000SEK

How long have you been active as a subscriber of your current SVOD-service? • Less than one month • 1-3 months • 3-6 months • 6-12 months • Over 1 year

69

10.1.4 Questionnaire in Swedish

Välkommen till undersökningen!

Vi är en grupp studenter från Jönköping University som för närvarande undersöker vilka kvalitetsfaktorer som bidrar till kundnöjdhet och lojalitet, inom svenska TV-streaming-tjänster, som enbart använder sig av prenumerationer som inkomstmodell. (Netflix, Viaplay, HBO Nordic och Cmore)

Enkäten kommer att ta cirka 7 minuter att genomföra, och vi är oerhört tacksamma för din tid. Att delta i undersökningen är frivilligt, och dina svar är anonyma. Du har även rätt att avbryta undersökingen när som helst.

Om du har några frågor kring enkäten är du varmt välkommen att kontakta Elliot Strand via emailadressen: [email protected]

I enkäten används termen “SVOD-tjänster” istället för abonnemangsbaserade streamingtjänster.

Tack för att du bidrar till vår studie! ______

Använder du eller har du tidigare använt någon av följande streamingtjänster, som denna studien klassar som SVOD-tjänster? (Netflix, Viaplay Cmore eller HBO Nordic) • Ja • Nej

Har du tillgång till ett konto hos någon SVOD-tjänst? I så fall, hur många? • 1 • 2 • 3 • 4 • Jag har inte tillgång till någon SVOD-tjänst

Vilken av följande SVOD-tjänster har du mest erfarenhet av? • Netflix • Viaplay • HBO Nordic • Cmore

70

Du kommer nu att få besvara några frågor om [din valda SVOD-tjänst]. Du ska få ranka ditt svar på en skala från 1 till 5. Du ska få bedöma hur väl [din valda SVOD-tjänst] uppfyller dina krav på en bra tjänst, baserat på följande frågor:

Systemeffektivitet • Det är enkelt för mig att hitta det jag söker på [din valda SVOD-tjänst]. • [din valda SVOD-tjänst] är enkelt att använda. • Hemsidan hos [din valda SVOD-tjänst] är välorganiserad. • Sökfunktionen på [din valda SVOD-tjänst] är tillfredsställande • Det går snabbt, från det att jag klickar, till dess att något händer på [din valda SVOD-tjänst].

• Överlag är jag nöjd med systemeffektiviteten hos [din valda SVOD-tjänst].

Systemtillgänglighet • [din valda SVOD-tjänst] tillåter mig att streama innehåll på flera enheter samtidigt. • [din valda SVOD-tjänst] tillåter mig att dela kontot med andra. • [din valda SVOD-tjänst] tillåter mig att komma åt innehåll var jag än befinner mig. • [din valda SVOD-tjänst] tillåter mig att komma åt innehåll när jag vill

• Överlag är jag nöjd med systemtillgängligheten hos [din valda SVOD-tjänst].

Integritet • Jag upplever att [din valda SVOD-tjänst] säkerställer att mina transaktioner är säkra. • Jag upplever att [din valda SVOD-tjänst] skyddar min betalningsinformation. • Jag upplever att [din valda SVOD-tjänst] skyddar min personliga information. • Jag upplever att [din valda SVOD-tjänst] inte missbrukar min personliga information.

• Överlag är jag nöjd med hur [din valda SVOD-tjänst] hanterar min integritet.

Rekommendationssystem • Jag tittar ofta på innehåll som [din valda SVOD-tjänst] rekommenderar. • Rekommendationssystemet hos [din valda SVOD-tjänst] hjälper mig att hitta relevant innehåll. • Rekommendationssystemet hos [din valda SVOD-tjänst] hjälper mig att spara tid när jag letar efter innehåll. • Innehållet jag blir rekommenderad av [din valda SVOD-tjänst] är passande. • Rekommendationssystemet hos [din valda SVOD-tjänst] är användbart, eftersom det introducerar mig till innehåll som jag kanske inte hittat annars.

• Överlag är jag nöjd med rekommendationssystemet hos [din valda SVOD-tjänst].

71

I den kommande sektionen vill vi att du svarar på hur nöjd du är med de efterfrågade aspekterna hos [din valda SVOD-tjänst].

Innehåll • Urvalet av tillgängligt innehåll hos [din valda SVOD-tjänst]. • Innehållets kvalitet hos [din valda SVOD-tjänst]. • Mängden olika genrer hos [din valda SVOD-tjänst]. • Urvalet av tillgängligt orginalinnehåll hos [din valda SVOD-tjänst]. • Kvaliteten på orginalinnehåll hos [din valda SVOD-tjänst]. • Hur aktuellt innehållet är hos [din valda SVOD-tjänst]. • Hur ofta [din valda SVOD-tjänst] uppdaterar innehållet. • Möjligheten att titta på sportevenemang hos [din valda SVOD-tjänst]. • Utbudet av sportevenemang hos [din valda SVOD-tjänst]. • Möjligheten att titta på trailers hos [din valda SVOD-tjänst].

• Överlag, hur nöjd är du med innehållet [din valda SVOD-tjänst] erbjuder?

Kvalitetsupplevelse • Bildkvaliteten hos [din valda SVOD-tjänst]. • Pålitligheten hos [din valda SVOD-tjänst] (att tjänsten fungerar som den ska). • Start och buffringstid av videos hos [din valda SVOD-tjänst]. • Hur ofta uppspelningen fryser eller hackar hos [din valda SVOD-tjänst]. • Möjligheten att spela upp innehåll i 4K Ultra-HD hos [din valda SVOD-tjänst].

• Överlag är jag nöjd med kvalitetsupplevelsen hos [din valda SVOD-tjänst].

Pris • Värde för pengarna hos [din valda SVOD-tjänst]. • Möjligheten att välja mellan olika typer av abonnemang hos [din valda SVOD-tjänst]. • Villkoren, till exempel för uppsägning av mitt abonnemang hos [din valda SVOD-tjänst]. • Möjligheten att välja mellan olika betalningsmetoder hos [din valda SVOD-tjänst].

• Överlag är jag nöjd med prissättningen hos [din valda SVOD-tjänst].

Kundnöjdhet • Hur nöjd är du överlag med [din valda SVOD-tjänst]? • Till vilken grad har [din valda SVOD-tjänst] mött dina förväntningar?

Lojalitetsavsikter • Var god ange på en skala från 1 till 10, hur troligt det är att du skulle rekommendera [din valda SVOD-tjänst] till en vän eller kollega?

72

Var god ange din ålder • Yngre än 18 • 18-21 • 22-25 • 26-29 • Äldre än 29

Kön

• Man • Kvinna

Var god ange din månadsinkomst • Under 15000 SEK • 15 - 23000 SEK • 23 - 30000 SEK • Över 30000 SEK • Jag önskar att inte besvara denna frågan

Hur länge har du varit prenumerant hos [din valda SVOD-tjänst]? • Mindre än en månad • 1-3 månader • 3-6 månader • 6-12 månader • Mer än ett år • Jag använder någon annans konto

73

10.1.5 Absolute Frequency

74

75

76

77

78

79

80

81

82

83

84

85

86

87

10.2 Convergent Validity

10.2.1 Efficiency

10.2.2 System availability

88

10.2.3 Privacy

10.2.4 Recommendation system

89

10.2.5 Fulfilment • Library of Content

90

10.2.6 Quality of Experience

10.2.7 Pricing

91

10.3 Discriminant Validity

92

10.4 Model Summaries and Coefficients

10.4.1 Efficiency Construct

10.4.2 System Availability Construct

93

10.4.3 Privacy Construct

10.4.4 Fulfilment•Library of Content Construct

94

10.4.5 Recommendation System Construct

10.4.6 Quality of Experience Construct

95

10.4.7 Pricing Construct

96

10.5 Hypotheses Testing

10.5.1 Hypothesis 1

Independent Variable: Constructs of Service Quality

Dependent variable: Overall Customer Satisfaction

97

10.5.2 Hypothesis 2

Independent Variable: Constructs of Service Quality

Dependent Variable: Brand Loyalty

98

10.5.3 Hypothesis 3

Independent Variable: Overall Customer Satisfaction

Dependent Variable: Brand Loyalty.

99

10.5.4 Hypothesis 4

Independent Variable: Constructs of Service Quality; Overall Customer Satisfaction

Dependent Variable: Brand Loyalty

100