An empirical study of users’
acceptance of MOOCs
Master Thesis Research Entrepreneurship & Innovation Faculty of Economics and Business
Master Business Administration
Date: January 26th, 2018 Supervisor: Dr. Alexander Alexiev Author: Martijn Karels (11362200)
Statement of originality
This document is written by Martijn Karels who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Preface
Before you lies the master thesis ‘’An empirical study of users’ acceptance of MOOCs’’. It has been written as a conclusion to the Master Entrepreneurship and Innovation. However, the research was challenging and demanding. Performing a comprehensive research gave me the tools me to answer the research question.
My thesis is the output of seven years of studying in this field. This thesis has been written between June 2017 and January 2018 at the faculty of Business and Economics at the University of Amsterdam.
Especially, I would like to thank my supervisor Dr. Alexander Alexiev for his valuable feedback and supervision throughout this project. Without the helpful comments and directions, the completion of this thesis would still be in progress. I also would like to thank my family and friends and all respondents who helped me to conduct this research.
Martijn Karels
Abstract
The aim of this study is to determine whether the factors that govern the influence on behavioural intention to use Massive Open Online Courses (MOOCs). Based on the strong theoretical foundation, The Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) is further modified in order to investigate the effect of performance expectancy, effort expectancy, social influence, facilitating conditions, hedonic motivation, and privacy risks on behavioural intention to use. By using a survey method the data was collected from 141 respondents. Statistical analyses found high correlations between the original UTAUT2-model determinants and significant positive effects for effort expectancy and hedonic motivation. Individuals who have the behavioural intention to use MOOCs are mostly influenced by effort expectancy and hedonic motivation. Age was found to be a statistically significant moderation effect on facilitating conditions and hedonic motivation. More specifically, the older the respondents, the more likely they will use MOOCs. A contrary result, in comparison to prior research, the negative coefficient of facilitating conditions is measured as insignificant. The new added variable privacy risk was insignificant and not found to be a strong predictor of behavioural intention. The findings in this study can provide insight into how these students’ value MOOCs and whether they are prepared to adopt this new technology. This research findings can help practitioners improve technical design and promotion, as to attract new potential users and subsequently improve students’ experience.
Key words: MOOC, behavioural intention, UTAUT2, multi-sided platform, online learning.
Table of contents
Statement of originality ... 2 Preface ... 3 Abstract ... 4 1. Introduction ... 7 2. Literature Review ... 102.1 Introduction to online learning ... 10
2.2 Introduction to multi-sided platforms ... 13
2.3 Introduction to Acceptance Theory Models ... 16
2.4 Theoretical Framework ... 21
2.4.1 Conceptual model ... 21
2.4.2 Hypotheses ... 23
2.4.3 Research hypothesis summary ... 28
3. Methodology ... 29
3.1 Strategy ... 29
3.2 Measurements ... 30
3.2.1 Pilot test ... 33
3.3 Research sampling ... 33
3.4 Data collection & analysis ... 34
4. Results ... 37 4.1 Reliability ... 38 4.2 Construct validity ... 39 4.3 Correlation matrix ... 42 4.4 Regression analysis ... 44 4.5 Moderating effect ... 46
4.6 Summary research model ... 51
5. Discussion ... 53
5.1 Contributions ... 57
5.2 Implications ... 57
5.3 Limitations & further studies ... 58
7. Reference list ... 62
Appendix ... 69
Appendix 1 Cronbach’s Alpha extended ... 69
Appendix 2 Table hypotheses summary ... 70
Appendix 3 Survey ... 71
List of tables Table 1: Survey construct descriptions ... 31
Table 2.1: Frequency table Chi Square ... 34
Table 2: Frequency table Chi Square ... 34
Table 3: Descriptive statistics ... 36
Table 4: Normal distribution ... 37
Table 5: Cronbach’s Alpha ... 38
Table 6: Pattern matrix – 1st iteration ... 41
Table 7: Means, standard deviation, correlation ... 43
Table 8.1: Model summary FC*Age ... 47
Table 8.2: Model moderator FC*Age ... 47
Table 8.3: Conditional effect FC*Age ... 48
Table 9.1: Model summary HM*Age ... 49
Table 9.2: Model HM*Age ... 49
Table 9.3: Conditional effect HM*Age ... 50
List of figures Figure 1: Technology Acceptance Model (Davis et al., 1989) ... 17
Figure 2: Unified Theory of Acceptance and Use of Technology 2 (Venkatesh et al., 2012) ... 20
Figure 3: Conceptual framework ... 21
1.
Introduction
The emergence of digital technologies and their penetration into all levels of education, from nursery schools to universities, has challenged higher education institutions to redesign their business models. One of the most important aspects of modernizing the traditional learning system is redesigning organizational infrastructures (Lee, 2010; Guri-Rosenblit, 2009). Recently, online learning has gained a lot of attention, although it is not a new phenomenon. New technologies in online education make it possible to reach more students than ever at minimal costs (Cusumano, 2013). Therefore, the online education sector is highly attractive and is growing promptly (Sun, Tsai, Finger, Chen, & Yeh, 2008).
However, new digital technologies and innovations are dealing with uncertainties. In developing new business models in the higher education sector, universities and institutions have to understand how individuals react to and adopt specific innovations and technologies (Rogers, 1995). There is enough anecdotal evidence that the higher education sector is on the verge of transforming into customizable and sustainable platform business models. Interactive open online learning systems for students based on a platform business model will lead to more open, shared data on the students overall performance (Škrinjarić, 2014; Bacow, Bowen, Guthrie, Long, & Lack, 2012). Up to now, a number of academics have already examined the acceptance of online learning, although most of these studies investigate the usage and adoption of online learning when proposed as a complement to the traditional learning system. In addition, a lot of these studies used the perspective of individual universities or educational institutions and not the students’ perspective (Allen & Seaman, 2013; Belleflamme & Jacqmin, 2015; Cusumano, 2013).
Online education, and especially highly interactive, machine-guided online instruction, is highly attractive because it can both improve learning outcomes and bend the cost curve in higher education (Bacow, et al., 2012). Further development of the e-learning experience can be realized if the students accept this new learning paradigm in an appropriate manner. Škrinjarić, (2014) proposed in his article that he foresees the future of online learning probably in a customizable and sustainable platform business model. Bowen (as cited Škrinjarić, 2014) in argues that a new
mind-set is necessary to overcome the adoption of online learning. Therefore, it seems relevant to study the behavioural intention to use online education platforms. Furthermore, Daphne Koller (as cited Škrinjarić, 2014) stated that online learning should not try to replicate the traditional learning system but should put more emphasis on developing entirely new platforms and systems of education (Škrinjarić, 2014). A relatively new technology that is transforming online education is Massive Online Open Courses (MOOCs) (Daniel, 2012). MOOCs offer high-quality short online learning courses from the world’s best universities and institutions to learners everywhere and impugn current online education challenges (Belleflamme & Jacqmin, 2015).
All above-mentioned statements provide strong arguments for studying the underlying behavioural intentions to use MOOCs. Previous research indicates the importance of online learning in higher education. Therefore, investigation in the determinants of students’ adoption of online learning systems through using acceptance models contributes to the scientific and practical impact technologies could have (Lee, 2008).
The purpose of my study is to get richer insights in the underlying behavioural intentions to use MOOCs by using acceptance models to empirically study this phenomenon. The study’s objective is to determine which specific factors influence the behavioural intention to use MOOCs in order to elaborate on who, why, and how individuals accept MOOCs and how they intent to use it. Making use of all the above-provided information leads to the following research question:
What are the factors that influence the behavioural intention to use MOOCs? The results of this study may in turn lead to a new theoretical approach that contributes to a thorough understanding of students’ motives to use MOOCs and help to empirically test those predictors of behaviour. The combination of extensive research and empirical study will help managers to better understand potential users. And therefore, help them improve existing technological aspects of MOOCs and marketing activities. Consequently, it is hoped that our findings further improve the experience of students.
The proposed research outline of this paper is as follows. Firstly, in the literature review, an introduction to online learning, multi-sided platforms and Acceptance Theory is elaborated. Subsequently, a conceptualized model is presented including accompanying hypotheses. The third chapter explains the methodology used in this research and shows how the research design is conducted. The fourth chapter presents the empirical results. And the last two chapters are divided into respectively, discussion and conclusion.
2.
Literature Review
This chapter discusses current relevant literature to this study. First, an introduction to online learning will be discussed in order to get a general understanding of this topic. Second, the introduction to multi-sided platforms presents theoretical background information about the business models behind MOOC’s. Lastly, an introduction to Acceptance Theory & Models will be discussed. It will be followed a proposition of the theoretical framework and an extensive explanation of each construct and fourteen hypotheses.
2.1 Introduction to online learning
Online learning environments also known as electronic learning (e-learning) offer possibilities for communication, interaction and multimedia material delivery that enhance the students’ learning environment. In this research, the terms e-learning, online education, online learning courses and online learning are used interchangeably. E-learning is driven by digital technologies and is developing into an essential part of higher education institutes. (Stefanovic, Drapsin, Nikolic, Scepanovic, Radjo & Drid, 2011). Govindasamy (2001) defines the broadest definition of e-learning as: ‘’E-learning includes instruction delivered via all electronic media including the Internet, intranets, extranets, satellite broadcasts, audio/video tape, interactive TV, and CD-ROM. All efforts to implement e-learning will eventually move towards total automation of administrating the teaching and learning processes by means of a software known as Learning Management Systems (LMS)’’. Rosenberg (2001) defines online education as a process of delivering course content to the end-user by means of a computer using Internet technology. The new learning paradigm of online learning is an advantage when it comes down to cheaper financial options. Moreover, it enhances students’ learning experiences and improves their learning outcomes and abilities (Stefanovic et al., 2011). Also factors such as infrastructure, quality of support systems, quality of content and assessment, and peer support networks may influence the e-learning experience (Stefanovic et al., 2011). However, Allen & Seaman (2013) stated that academic leaders think that “students need more discipline to succeed in online courses’’.
Škrinjarić (2014) notices a new trend in higher education, ‘’blended learning’’. It combines the traditional learning system, which is mostly based on face-to-face interaction with web-based course content. A relatively new phenomenon in higher education is ‘’virtual learning’’. Universities are investing heavily in improving the learning experience by implementing new online learning technologies. The research of Concannon, Flynn, & Campbell (2005) proposes that students see e-learning as a benefit and improvement of teaching quality. This means that complete courses are only available online to students. The online learning market is growing rapidly and the percentage of higher education students taking at least 1 online course is 32% (Sun, et al., 2008; Allen & Seaman, 2013). The main contribution of e-learning is to improve the quality of learning activities by reusing and sharing information and knowledge while the student has control over its own flexibility and pace of learning (Stefanovic et al., 2011). A recent study of Park (2009) shows how students adopted e-learning in higher education by using the Technology Acceptance Model (TAM). This acceptance model and others will be discussed in paragraph 2.3.
A relatively new technology that is transforming online learning is Massive Online Open Courses (MOOCs). These platforms aim to fill the gap in distance learning by offering high-quality short online learning courses from the world’s best universities and institutions to learners everywhere (Daniel, 2012). So, instead of physically attending lectures and seminars, it is possible to complete courses online by using university materials and resources. Important characteristics in comparison to earlier attempts in online distance learning are its availability, the fact that they are free of charge, and often do not have entry requirements. Learners can self-organize their participation according to their own learning desires. In addition, learners have the possibility to receive a certification of completion, although this is often not free of charge (Belleflamme & Jacqmin, 2015). The same authors also mentioned potential monetization strategies for MOOCs. The first strategy is MOOC platforms as multi-sided platforms. Using this strategy is important to manage interactions for different network groups. The second monetizing strategy is the certification model. After successful completion of an online course the user receives a certificate. A third strategy is the Freemium model. This strategy offers the basic features for free, although a paid premium account is necessary to receive enhanced access. The fourth strategy is the advertising model, which is widely used in the internet sector for
subsidizing free content. The fifth strategy is the Job matching model. And lastly the subcontractor model implements independent platforms that are connecting different stakeholders (Belleflamme & Jacqmin, 2015). In paragraph 2.2, multi-sided platforms will be discussed. This paragraph will be linked to online learning in order understand the relevance of multi-sided platforms in the online education sector.
However, uncertainties of monetizing MOOCs and lack of coherence make that MOOCs are still being perceived as a supplement along the traditional programs (Belleflamme & Jacqmin, 2015). Nevertheless, this evolving environment of online education platforms will flourish in the upcoming years. Online and hybrid models in the higher education sector will transform the traditional systems (Belleflamme & Jacqmin, 2015). According to Cusumano (2013), these MOOCs and other open online courses that offer free learning will stay and will be mostly seen as a complement to traditional education. However, he foresees that universities will have to charge tuition fees for offering open online courses and studies. These technology developments in online education will force universities and institutions to bring down educational costs and tuition fees.
O'Donoghue, Singh, & Dorward (2001) proposed the extreme ideology that universities can become obsolete and scattered around the world. Students could in their reasoning acquire degrees without attending classes. However, this inevitably change of digitalization also has its downsides. Group and face-to-face social interaction is an important element to the current educational system, although new technologies will keep improving the current learning experience and will aim to improve these problems. The research of Bacow et al. (2012) sheds light on barriers of adoption of online learning systems from an individual educational perspective. The authors proposed that development in online education will continue because it offers potential that improves learning outcomes and cost reductions in higher education. However, how to design ‘’online education platforms’’ for higher education is far from obvious. A fact is that most universities do not offer distance learning by making use of e-learning (Guri-Rosenblit, 2009). Insights into platform business models will help to develop, distribute, maintain, and upgrade these new forms of online learning (Bacow et al., 2012).
2.2 Introduction to multi-sided platforms Network effects & externalities
First, starting with the theory about network effects and externalities of Katz & Shapiro (1985), which presents some general foundations that provide theoretical background information to platform business models. Katz & Shapiro (1994) define network effects or network externalities as “the value of membership to one user is positively affected when another user joins and enlarges the network” (p. 94). Network effects are one of the success factors behind every platform. The more platform participants, the more valuable the platform becomes. This is due to a larger network that contains more suitable matches, complementary innovations, and more available data (Gawer & Cusumano, 2014; Alstyne, Parker & Choudary, 2016). Network effects are highly related to economies of scale and show similar results in fluctuations. Therefore, platforms should understand their community and its financial value (Alstyne et al., (2016). Network effects are exogenous and are affected by the design and investments. Examples are the quality of technologies, the offered services, the working mechanism, and the interaction among platforms’ sides (Bakos & Katsamakas, 2008). However, Boudreau (2014) shows in his research that too many complementors can also have negative effects on network, which results in negative feedback loops as well.
Gawer & Cusumano (2014) argued that network effects could be very vigorous in magnitude. Network effects can be direct or indirect. Direct network or same side effects are the interactions between the platform and the user without any involvement of non-users of the specific platform. Indirect network or cross-side effects arise when a new platform side becomes attracted to the platform because of the current users of the platform, which is also often accompanied by a large amount of users. Caillaud and Jullien (2003) studied imperfect-competition among intermediaries with indirect network externalities. Their study shows that low participation fees and maximal feasible transaction fees are the key to become market leader and protect this position. Tucker & Zhang (2010) further extended network effects and presented that displaying the amount of users and sellers influence new attractions to the platform. They came to the conclusion that displaying the number of users to potential sellers is the most effective way to attract
new sellers. To attract new buyers the most effective strategy is to show both the amount of buyers and sellers.
‘’Marquee buyers‘’ influence the price structure of a platform. In a follow-up study from 2006, Rochet and Tirole referred in their article to ‘’marque buyers’ as a group that is extremely valuable to influencing the other side of the market, namely its potential users. These exclusive buyers make the platform more attractive, which could result in higher revenue (Bakos & Katsamakas, 2008). These so-called ‘’marquee buyers’’ can be described in my research as “marquee students” and an important issue to this study is to look at which students intend to use MOOCs. So that these can be targeted first in order to make use of the above described network effects.
Multi-sided platforms
To date, a lot of academic research has been done on the economic perspectives behind multi-sided platforms. These economic theories are included in this paper and present contextual background about business models of multi-sided platforms. Rochet & Tirole (2003) are the founders of multi-sided platforms and introduced them as “a market in which one or several platforms enable interactions between end-users and try to get the two (or multiple) sides “on board’’ by appropriately charging each side” (Rochet & Tirole, 2006, p. 645). Later on, more authors elaborated on this theory and Armstrong (2006) added that both groups benefit from the size of the platform and the accompanied positive network externalities in order to create value.
An important issue to a platform is to get one side on board. This so-called chicken-and-egg problem is addressed in the articles of Rochet & Tirole, (2003, 2006, in press). This theory states that one side of the platform has to be “on board” before it can attract the other side (or multiple sides). In addition, platforms should be able to affect the transactions’ volume on both sides of the market. This implies that the price-structure is highly important to the design of the platform to get both (or multiple) sides on board. Evans (2003) describes that the way to get one side of the platform “on board” is to give service away for free or paying early adopters to gain the critical mass. Rysman (2009) mentioned that a lot of companies such as
amazon.com initially started as a one-sided model. Amazon’s business model was focusing only on online book retailing. Later on, they changed its business model and introduced the marketplace, the second-side. By changing from a one-side to a two-sided market they had to overcome the chicken-and-egg problem. Caillaud & Jullien (2003) proposed another strategy that does invest in one side of the market in order to lower the consumer’s costs of participating. A third strategy to solve this chicken & egg problem is the “divide and conquer strategy”. One side of the market is subsidizing the other side of the market, which is making profits. But once a pricing strategy had been used to get both sides on board, a new pricing strategy should be implemented to maintain both sides on board and stimulate the retention rate. Using above-mentioned strategies will increase value for all involved stakeholders (Evans, 2003).
Traditional markets and industries evolve rapidly and becoming more dynamic. Alstyne et al. (2016) propose that platforms always win when they enter a traditional pipeline firm’s market. This has resulted in “platform envelopment”. This term represents a shift to a market that consist of platform providers. This economic shift brings along blurry market boundaries and threats of becoming obsolete (Eisenmann, Parker, & Van Alstyne, 2006). Li, Liu & Bandyopadhyay (2010) conducted research on the importance of online platform differentiation. Online platforms have often no or less switching costs. Therefore, it is easier to attract new consumers to the platform because no registration fees have to be paid. However, consumers can easily use other platforms as well. The phenomenon that occurs when users or sellers use several platforms at the same time is called multi-homing (Li et al, 2010). However, the results of this study shows that online competing platforms should differentiate from each other in order to increase profits by raising higher prices for example. Furthermore, Cennamo & Santalo (2013) elaborate in their article about platform distinction and propose to choose a distinct market positioning in comparison with rivals. Structuring and assembling a platform’s ecosystem that is different is called “distinctive positioning” and is aiming for market niches.
Platform strategy
Eisenmann et al. (2006) proposed three strategic goals to pursue as a platform. The first strategic goal is to get the price right. It is important that the price on all participating sides asks appropriate fees and or free access. The second strategic goal is the winner-takes-all competition. All platforms are trying to achieve the position of a monopoly-firm and platforms have to deal with this. The third strategic goal is to avoid envelopment. This term is already earlier-mentioned in this thesis and can be briefly described as non-differentiation of platforms. Platforms should focus on differentiation from its competitors and aim for uniqueness. Bakos & Katsamakas (2008) added a fourth strategic goal, namely network design. The design of a network can be explained by the value proposition the network contains such as potential users. The design of multi-sided platforms is defined as “the architecture of the services offered and the infrastructure that facilities the interaction between participating sides, and a set rules, such as pricing terms and the rights and obligations of the participants’’ (Eisenmann et al. (2006). Furthermore, the study of Bakos & Katsamakas (2008) shows that the most profitable strategy for platforms is to minimally invest on the seller or producer side. In other words, invest on the side where revenues are generated, often the user’s side. Appropriate pricing structures and design strategies are the key to success in multi-sided platforms. According to the authors, the network benefits are internal in each side of the platform and established by the investment strategy.
2.3 Introduction to Acceptance Theory Models
For more than two decades, different theoretical approaches are formed in order to explain the acceptance of information systems and new technologies (Venkatesh, Morris, Davis, & Davis, 2003). The study of Social Cognitive Theory (SCT) focuses on the human behaviour and the sequences of this behaviour by observing others (Bandura 1986). Davis’ (1989) theory approach, which is called Technology Acceptance Model (TAM), is based on the Theory of Reasoned Action (TRA) and the Theory of Planned behaviour (TPB). TRA and TPB are both focusing on the attitude towards behaviour. TAM is proposing that behavioural intention is determined by two cognitive beliefs: perceived usefulness and perceived ease-of-use. According to Davis (1989) this is a powerful model to predict user acceptance The Innovation
Diffusion Technology (IDT) model of Rogers (1995) explains how new ideas and technologies are spread through different cultures. Venkatesh et al., (2003) proposed the theoretical framework of the Unified Theory of Acceptance and Use of Technology (UTAUT) to explain behavioural intentions to use new technologies. In a more recent study by Venkatesh Thong, & Xu, (2012), the UTAUT2 was extended to a customer’s perspective. Accordingly, the TAM, IDT, UTAUT-model, and UTAUT2-model are further elaborated.
Technology Acceptance Model (TAM)
The TAM is proposed by Davis (1989) in order to investigate the impact of technology on user behaviour. The model is an extension of the theory of reasoned action (TRA) and is the most widely applied model of users acceptance and usage of technology. TAM provides a framework and shows how external variables influence belief, attitude, and intention to use. Originally, TAM focuses on two major cognitive beliefs: perceived usefulness, which focuses on outcome expectancy and perceived ease of use, which focuses on process expectancy.
Perceived usefulness (PU) is explained as to what extent an individual believes that using a particular technology will enhance his job performance (Davis, 1989).
Perceived ease-of-use (PEOU) means to what extent an individual believes that using a particular technology would cost less or not effort (Davis, 1989).
Figure 1: Technology Acceptance Model (Davis et al., 1989)
TAM specifically explains computer usage behaviour using TRA as a theoretical foundation for specifying the causal links among five key constructs: perceived usefulness, perceived ease of use, attitude toward use, behavioural intention to use,
and actual system usage. In addition to TAM, perceived usefulness is also influenced by perceived ease-of-use because a system that is easier to use can also be more useful (Venkatesh & Davis, 2000). TAM may be especially well suited for modelling computer acceptance, including internet services in which it explains the social influence and cognitive instrumental processes. (Venkatesh et al, 2003).
Innovation Diffusion Theory (IDT)
The Innovation Diffusion Theory (IDT) is also a widely applied model in technological adoption decision-making and has been used to study a variety of innovations. Moore and Benbasat (1991) adapted the model of information systems and improved some constructs to focus specifically on individual technology acceptance. IDT includes five significant innovation characteristics: relative advantage, compatibility, complexity, trial ability, and visibility. It has been widely applied in disciplines such as education, sociology, communication, marketing, etc. (Rogers, 1995).
Unified Theory of Acceptance and Use of Technology (UTAUT)
The Unified Theory of Acceptance and Use of Technology also known UTAUT-model is a comprehensive theory based on eight UTAUT-models of acceptance and adoption of technologies. The models and theories that are taken into account by the researchers: Venkatesh, Davis, Davis, and Morris, (2003) are the Theory of Reasoned Action (TRA, Fishbein & Ajzen, 1975), Technology Acceptance model (TAM, Davis, 1989), Motivational Model (MM, Davis, Bagozzi, and Warshaw, 1992), Theory of Planned Behaviour (TPB, Ajzen, 1991), Combination of TAM and TPB (Taylor & Todd, 1995), Model of PC utilization (MPCU, Thompson, Higgins, and Howell, 1991), Innovation Diffusion Theory (IDT, Moore & Benbasat, 1991), Social Cognitive Theory (SCT, Compeau & Higgins,1995).
UTAUT has four key constructs: performance expectancy, effort expectancy, social influence, and facilitating conditions. Those constructs influence behavioural intention to use new a technology. The constructs and definitions as described in the research of Venkatesh et al. (2003) are accustomed to consumer technologies and use of context.
Performance expectancy (PE) is defined as the degree to which using a technology will provide benefits to consumers in performing certain activities. Performance Expectancy is moderated by the individual factors age and gender.
Effort expectancy (EE) is the degree of ease associated with consumers‘ use of technology and is moderated by age, gender and experience. The concept of effort expectancy is similar to the constructs perceived ease of use of the TAM and the construct complexity of the IDT.
Social influence (SI) is the extent to which consumers perceive that important others (e.g., family and friends) believe they should use particular technology and is moderated by age, gender and experience.
Facilitating conditions (FC) refer to consumers‘ perceptions of the resources and support available to perform behaviour. The facilitating conditions have direct effect on use behaviour and are moderated by age and experience.
The model explains that performance expectancy, effort expectancy, and social influence are theorized to influence behavioural intention to use technology, while behavioural intention and facilitating conditions determine technology use (Venkatesh et al., 2003).
Figure 2 on the following pages presents the extended version of the UTAUT-model, UTAUT2-model. New variables and relationships will be further detailed in the additional paragraph.
Figure 2: Unified Theory of Acceptance and Use of Technology 2 (Venkatesh et al., 2012)
Unified Theory of Acceptance and Use of Technology 2 (UTAUT2)
Venkatesh et al. (2012) extended the original UTAUT-model to UTAUT2-model. This extended model is developed from a customers’ perspective instead of the organisational perspective in the first model, which makes it a suitable model to measure behavioural intention to use MOOCs from a student’s perspective. The UTAUT2-model incorporates three constructs that are focused on the customer’s perspective. The first construct is hedonic motivation and is moderated by age, gender and experience. This construct is focusing on the degree to which a user experiences enjoyment of using it. The second construct is price value and is moderated by age and gender. This is the degree to which the technology is in line with the price. A trade-off between perceived benefits in relation to monetary costs. A negative price value is the case when the technology’s value does not meet price’s value. The last added construct in the UTAUT2-model is habit, which is moderated by age, gender and experience. This construct does not only have a direct effect on
behavioural intention but also on use behaviour. The three added constructs are considered as complex, although it provides the model with a customer’s perspective. The UTAUT2-model also proposes a new effect between facilitating conditions and behavioural intentions that is influenced by age, gender and experience.
2.4 Theoretical Framework 2.4.1 Conceptual model
Figure 3: Conceptual framework
As previously mentioned, a variety of acceptance theories have been studied in order to find the most appropriate model to explain the behavioural intention to use MOOCs. The empirical research of Chang & Tung (2008), which investigates the behavioural intention to use online learning course websites, is a nearly equivalent study. The research methodology of Chang & Tung (2008) is a combination of the TAM and IDT. Those theories will be used as background information to study my research objective. The research of Tarhini et al. (2016) investigates students’ usage behaviour of e-learning systems in Lebanon. Their research makes use of an
adapted version of the UTAUT2-model and includes trust as a primary independent variable to explain the behavioural intention. This research is the closest equivalent study and will be used to provide valuable information to conduct my research model.
In figure 3, the proposed conceptual model of the behavioural intention to use MOOCs is presented. The UTAUT2-model is the most recent model, which incorporates all important attributes and elements of the previous theories about acceptance and adoption models. In addition, the UTAUT2-model is widely used by academics and researchers and is tested in related fields (Tarhini, Mohammed, & Maqableh, 2016). Therefore, this model is the most applicable to use in this research. The former UTAUT-model has a predictive efficiency of 70%, which is a substantial improvement compared to previous models used in researching technology acceptance (Venkatesh et al., 2003). The UTAUT2-model of Venkatesh (2012) is further extended with new constructs and variables in order to fit the research objective and to find new insights of students in the behavioural intention to use MOOCs. Therefore, utilizing the extended model is the most appropriate to investigate behavioural intention and adoption of this new technology.
This model proposes that the four initial constructs: performance expectancy (PE); effort expectancy (EE); social influence (SI); facilitating conditions (FC) of the UTAUT-model have significant positive influence on the behavioural intention to use MOOCs (Venkatesch et al., 2003). Furthermore, the construct hedonic motivation (HM) of the UTAUT2-model takes the students’ perspective into account and has significant positive influence on the behavioural intention to use MOOCs (Venkatesch et al., 2012). Additionally, privacy risk (PR) was added to the existing model. It is assumed that this construct has a significant negative influence on the behavioural intention (Martins, Oliveira, & Popovič, 2014). The moderating variables age, gender, education, and experience were incorporated to explain the variance on facilitating conditions and hedonic motivation. All above-mentioned independent variables were selected to explain and predict the influence on behavioural intention.
In this research, the dependent variable use behaviour (UB), as shown in Figure 2: Unified Theory of Acceptance and Use of Technology 2 (Venkatesh et al., 2012) has
been excluded from the theoretical framework. It is difficult to measure technology which is still being developed and not yet widely used by students. MOOCs are still in the early stage of their lifecycle (Christensen et al., 2013). Hence, this research is mainly focusing on the behavioural intention to use this new technology and therefore the actual use behaviour is valued less important than other constructs.
Due to former justification, the construct facilitating conditions, only measures the behavioural intention and not the direct relationship with use behaviour. The original UTAUT2-model constructs price value and habits are also not incorporated in this research. The research objective is focused on MOOCs, which are often free of charge. Therefore, price is considered irrelevant. In addition, the novelty of MOOCs combined with the main interests in the behavioural intention leads to the fact that the variable habit is also of less importance to this research. This allows this research to put emphasis on six main independent variables without losing the accuracy and strength of the original model.
2.4.2 Hypotheses
Performance expectancy (PE)
Performance expectancy is defined as “the degree that enhances the performance when using new technology.” The performance expectancy is the most important determinant of the UTAUT2-model and is consistent as predictor of the behavioural intention to use new technologies. This construct has been originated by incorporating multiple constructs of different acceptance theories. An important input is the TAM model of Davis (1989). The construct Perceived Usefulness states that using a particular system would enhance the performance. The constructs: extrinsic motivation (MM), job-fit (MPCU), relative advantage (IDT) and outcome expectations (SCT) are important predictors of previous models, which are also incorporated into performance expectancy (Venkatesh et al., 2003). Because of the consistency of this strong predictor in the UTAUT2-model it is expected that it will have a positive influence on the behavioural intention to adopt online education platforms in higher education (Venkatesh et al, 2012)
Hypothesis 1: Performance expectancy will have a significant positive influence on the behavioural intention to use MOOCs.
Effort expectancy (EE)
Effort expectancy is defined as “the degree to how much effort users should put into new technology.” This second determinant is an important predictor of the behavioural intention to use new technologies. This construct is based on three root constructs of previous studies within the context of acceptance models. Perceived ease of use of the TAM measures the belief that using a system would be free of effort. The complexity of MPCU model relates to the degree that a system is perceived as difficult to understand. The last construct ease of use of IDT defines the perception of difficulty when using an innovation. Logical reasoning would assume that the more effort is needed, the less attractive the new technology will be and vice versa. Therefore, the hypothesis can be concluded as:
Hypothesis 2: Effort expectancy will have a significant positive influence on the behavioural intention to use MOOCs.
Social influence (SI)
Social Influence is defined as “the degree to which the consumers perceive that important others have impact on their decision making.” This determinant is based upon three constructs, respectively subjective norm (TAM), social factors (MPCU), and image (IDT). Previous acceptance models express an individual’s behaviour by how others will judge them as a result of using the technology (Venkatesh et al., 2003). Therefore, it is assumed that behavioural intention to use MOOCs in higher education is influenced by others.
Hypothesis 3: Social influence will have a significant positive influence on the behavioural intention to use MOOCs.
Facilitating conditions (FC)
Facilitating Conditions is defined as “the degree of consumer perceptions of the resources and support available to perform a behaviour.” This determinant removes barriers to use new technologies and is based upon three constructs of previous
acceptance theories: The perceived behaviour control (TPB), facilitating conditions MPCU) and compatibility (IDT). Facilitating conditions encompasses items such as resources, knowledge, compatibility, and external help (Venkatesh et al., 2003. In this research only the relationship on behavioural intention will be tested and not the relationship on use behaviour. The focus will be mainly on potential users who have little experience with MOOCs. Therefore, the following hypothesis is proposed:
Hypothesis 4: Facilitating conditions will have a significant positive influence on the behavioural intention to use MOOCs.
Hedonic motivation (HM)
Hedonic motivation is defined as “the fun or pleasure derived from using a technology.” In the context of students, hedonic motivation is an important determinant to technology acceptance and the use of it. This construct is rather conceptualized as perceived enjoyment, which is a direct predictor of the behavioural intention to use MOOCs of students in higher education (Venkatesh et al., 2012). Therefore, we assume that:
Hypothesis 5: Hedonic motivation will have a significant positive influence on the behavioural intention to use MOOCs.
Privacy Risk (PR)
The first one who introduced perceived risk is Bauer (1960) and defined it as “a combination of uncertainty plus seriousness of outcome involved.” Furthermore, Ostlund (1974) elaborated further on this by stating that negative consequences may arise from consumers’ action that can lead to particular consumer behaviour: perceived risk. More recently, Featherman & Pavlou (2003) tested seven risk facets of perceived risk including: performance, financial, time, psychological, social, privacy, and overall risk in the field of internet service adoption. In this study, privacy risk will be used as predictor to measure behavioural intention to use MOOCs. According to Luo, Li, Zhang, and Shim (2010), perceived risk is a significant determinant to explain innovative technology acceptance. Furthermore, Featherman & Pavlou (2003) proposed in their research that privacy risk negatively affects the performance expectancy. However, this relationship is not incorporated because we
were interested in direct effects on behavioural intention. The main indicators of privacy risk are uncertainty and potential loss. Hence, the influences of the latter two indicators are reasonable to explain the relationship on the usefulness of MOOCs. In addition, Luo et al. (2010) mentioned that risk perception is a prominent antecedent to innovative technology acceptance. Therefore, the following hypothesis is designed:
Hypothesis 6: Privacy risk will have a significant negative influence on the behavioural intention to use MOOCs.
Moderating variables: age, gender, education & experience
According to UTAUT2-model, there are three demographical moderating factors, which are age, gender, and experience. The variable education has been added to the existing model in order to get a better comprehensive understanding of the research sample. As shown in figure 2 and 3, moderating variables influences the relationship of facilitating conditions and hedonic motivation on behavioural intention (Venkatesh et al., 2012). In addition, Tarhini et al. (2016) found that demographic moderators such as age, gender, and experience have influence on the relationship between facilitating conditions and hedonic motivation on usage behaviour. Note that notion, the research of Liu et al. (2010) also stated the importance of classifying demographic variables in order to compare differences of categories within the research objective. Therefore, in this research facilitating conditions and hedonic motivation will also be investigated on possible moderating effects. Starting with the following hypotheses that every moderator has potential influence on the relationship between facilitating conditions and behavioural intention to use MOOCs.
Hypothesis 4-1: Age is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
Hypothesis 4-2: Gender is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
Hypothesis 4-3: Education is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
Hypothesis 4-4: Experience is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
The following hypotheses are designed to find for every moderator a potential influence between hedonic motivation and behavioural intention to use MOOCs.
Hypothesis 5-1: Age is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
Hypothesis 5-2: Gender is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
Hypothesis 5-3: Education is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
Hypothesis 5-4: Experience is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
2.4.3 Research hypothesis summary
H 1: Performance expectancy will have a significant positive influence on the behavioural intention to use MOOCs.
H 2: Effort expectancy will have a significant positive influence on the behavioural intention to use MOOCs.
H 3: Social Influence will have a significant positive influence on the behavioural intention to use MOOCs.
H 4: Facilitating Conditions will have a significant positive influence on the behavioural intention to use MOOCs.
H 4-1: Age is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
H 4-2: Gender is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
H 4-3: Education is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
H 4-4: Experience is a moderating relationship between facilitating conditions and behavioural intention to use MOOCs.
H 5: Hedonic Motivation will have a significant positive influence on the behavioural intention to adopt online education platforms.
H 5-1: Age is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
H 5-2: Gender is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
H 5-3: Education is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
H 5-4: Experience is a moderating relationship between hedonic motivation and behavioural intention to use MOOCs.
H 6: Privacy Risk will have a significant negative influence on the behavioural intention to use MOOCs.
3.
Methodology
As explained in the previous section, this study utilizes the Unified Theory of Acceptance and Use of Technology model 2 (UTAUT2-model). This study employed a quantitative research approach to test the proposed research model. The empirical study mainly imitates the prior research of (Tarhini et al. 2016; Chang & Tung, 2008; Martins et al. 2014). The gathering of data for this study was carried out by a survey approach (Pinsonneault & Kraemer, 1993) states that surveys are widely used in management information systems research. This chapter provides a detailed explanation of the research approach and strategy, measurements, pilot test, research sampling, and data collection and analysis.
3.1 Strategy
From a pragmatic philosophy perspective, a deductive approach was used, whereby existing theory is used to conduct a comprehensive research plan. This research plan includes the conceptualization and operationalization of definitions and gathers empirical data (Bryman & Bell, 2015). The theory and hypotheses of previous chapters are the underlying mechanism and drivers of gathering data. This structured methodology allows me to test hypotheses that explain relationships between variables. Subsequently, it measures the expectation of the empirical data to be found (Saunders, 2011). Existing literature was used to give directions and insights throughout the whole project, an iterative process. In favour of my research plan the decision was made to conduct the research at a particular time. This so-called ‘’snapshot’’ time horizon and is better known as a cross-sectional study, which is perfect for quantitative research methods such as surveys (Saunders, 2011). The argumentation to choose a measurement method at one point in time is build on the tight time-constraint of this research project.
A quantitative survey is conducted in order to measure which factors have influence on the behavioural intention to use MOOCs from a student’s perspective. The conceptual model in figure 3 and the associated hypothesizes are used as a starting point of gathering data first hand. A structured online survey was administered in
order to collect numerical data (Saunders, 2011). The online questionnaire was created with www.qualtrics.com, and multiple online channels were utilized to deploy this survey. Social network websites such as Facebook and LinkedIn, mailing lists, survey websites, but also WhatsApp were used. Furthermore, in university halls and libraries students were approached in order to gather more responses. The survey was accessible to everybody with a laptop or Smartphone. No further exceptions were made based on age limitations, native country or whether you are currently a student.
3.2 Measurements
All survey measurement constructs are adapted from earlier literature findings. Existing literature models, which have been validated by many authors, have been used and extensively tested. Especially, research concerning the UTAUT2-model was used as a starting point (Venkatesh et al., 2012). The advantage of extensively pre-tested measurement frameworks will likely help to score higher on reliability, validity, and overall score of the obtained dataset (Bulmer, Gibbs, & Hyman, 2006). From previous chapters, constructs and variables related and possible connections have been established within the hypotheses proposed. The paragraphs below will elaborate further on allocating the determined variables towards related measurement constructs in order to proceed with empirical research.
The variables used, were mainly adopted from research of Venkatesh et al., 2003; Venkatesh et al., 2012; Tarhini et al. 2016; Chang & Tung, 2008; Martins et al., 2014 and Luo et al., 2010. Former studies were used to modify certain variables in an appropriate manner for this research objective. The variables price value and habit were excluded from this research. Price value is irrelevant at this moment because MOOCs is free of charge for users, although users can choose to buy a certificate of completion at the end of their course, but this is not compulsory. As such, some MOOC platforms are shifting their business model into a membership model. In this model you pay a monthly subscription fee and you are able to use all courses on its website. This could be due to the platform’s pricing strategy to maintain both sides on board and stimulate retention rate as discussed by Evans (2003). However, in
this research price value is considered as irrelevant and not taken into account. In the following table (1) the survey’s construction and definitions are explained.
Constructs Definition References
Performance Expectancy The possible increase of study performance for the user when using MOOCs.
Venkantesh et.al., 2003
Effort Expectancy The amount of effort you should put into a MOOC.
Venkantesh et.al., 2003
Social Influence The degrees to which you think that people who are important to you have an impact on your decision-making.
Venkantesh et.al., 2003
Facilitating Support The degree of resources and support available when using MOOCs.
Venkantesh et.al., 2003; Venkatesh, Thong, & Xu, 2012
Hedonic Motivation The degree of fun or pleasure you derive from MOOCs.
Venkatesh, Thong, & Xu, 2012
Privacy Risk The privacy and security risks you might experience while using MOOCs.
Featherman & Pavlou, 2003; Luo et al., 2010; Martins et al., 2014) Behavioural Intention The measurement of
intention to use MOOCs.
Venkantesh et.al., 2003; Venkatesh, Thong, & Xu, 2012
Table 1: Survey construct descriptions
The underlying assumption of the UTAUT2-model will be used to test the influence on the behavioural intention to use MOOCs. The relationship of performance expectancy, effort expectancy, social influence, facilitating conditions, hedonic motivations, and privacy risk on behavioural intention to use MOOCs will be measured. Moderating variables: age, gender, experience, and education were
included into this research model for testing the influence of facilitating conditions and hedonic motivation on behavioural intention to use MOOCs. Two items (of constructs: effort expectancy and behavioural intention) in the survey were reversed. After recoding the values, boundaries of 2 points or more deviation were considered as invalid answers.
As mentioned earlier, Qualtrics was used to create the online survey and SPSS 22.0 was used to perform the statistical analyses. Respondents were forced to fill in all survey questions in order to avoid incomplete surveys. The survey was available online for exactly two weeks. Real-time insight into the collected data made it possible to keep track of the recorded responses and filter out incomplete questionnaires. The estimated time of completing the survey was about 5 minutes. Respondents who had a completion time less than 1 minute were considered invalid because of inadequate understandings of the provided reading materials. Most data was gathered at interval level, except gender and education were measured on nominal level.
All constructs on interval level were measured by using 7-point Likert scales ranging from (1) strongly disagree to (7) strongly agree. Robson (2011) states that Likert scale measurement is often used. The data can be used when sufficient recorded responses are collected and a normal distribution originates. Jamieson (2004) adds that parametric tests can be executed based on the data collected by Likert scale measurements. In total the survey contains 25 items that originate from 8 constructs. These constructs were accompanied by five basic demographic questions. First, the dataset was tested on scale reliabilities, descriptive statistics, skewness, kurtosis, and normality. Subsequently, internal consistency and reliability of all constructs were tested with Cronbach’s alpha and the construct validity of each item was done with factor analysis. Afterwards, correlation analysis was done to explore possible relationships between constructs. Followed by a multiple linear regression analysis that was conducted to test whether each independent construct can predict the dependent variable. Finally, the moderation analyses were conducted by using the process macro of Andrew F. Hayes for SPSS.
3.2.1 Pilot test
Before the main survey rollout commenced, a pilot study was conducted to ensure clarity and validity of the measurement variables. The pilot study was sent out to eleven Dutch persons. The respondents’ chosen had different background demographics but also a variety of background disciplines had been taken into account. After analysing the results, survey adjustments were made. Starting with the result of omitting one construct out of the survey. Furthermore, the language used was rendered in order to reach a larger audience. This included an understandable explanation of the variables being tested, condensed items, question reformation, and alteration on the attractiveness of the survey introduction.
In line with the research of Teijlingen & Hundley (2001)) the pilot testers have identified all possible flaws and misunderstandings, which improved the effectiveness and efficiency of the survey. Moreover, functionality tests were done on multiple mobile devices such as Android, iOS and Window phones but also on different operating systems such as Windows and Apple combined with several internet browsers.
3.3 Research sampling
The research sample in this study focuses on students in higher education. The data was collected from a sample of users and non-users. The aim of choosing students in higher education is to generate reliable and valid outcomes in order to generalize a representative population. The non-probability judgement sampling technique was utilized to reach the particular research sample of this study, which is students who are currently studying in higher education in the Netherlands (Saunders, 2011). Convenience sampling was used to collect data within my own social network. Accordingly, snowball sampling was used as a method to reach more respondents. This method is used to recruit new samples within the population. In other words, the network of acquaintances of the existing sample was used to quickly gather new valuable data. Existing respondents build up the research sample by sharing the survey and giving referrals and recommendations within their own network (Heckathorn, 1997). The advantages of this type of research sampling are that it is a fast method to locate potential respondents with the same characteristics’ of the
research population. However, Applying these methods deal with a number of biases. Starting with the absence of random
probability. For example, many respondents of the same sample will be recruited to participate in this survey. This will result in equivalent data and possible bias in generalization of the sample. Oversampling the research objective can also cause more bias. A method to reduce biasness is to select efficient initial points of referral. This results in respondent-driven sampling, which recruits respondents rather than search for them (Heckathorn, 1997).
3.4 Data collection & analysis
Main survey rollout was held on Thursday 30thof November 2017 until Thursday 14th of December 2017. A structured online questionnaire collected the data. In total 155-recorded responses were collected. 141-valid 155-recorded responses and 14 responses had either missing data or were not suitable for analysis. According to the general rule of thumb, less than 10% missing data is negligible if the condition is met that the missing data is randomly distributed (Field, 2013). In this research 9% of missing data was randomly distributed across the data list and therefore it had been removed from the data set in Qualtrics. 4 out of 14 invalid responses had a completion time of less than 1 minute. These data were removed because of the credence of inadequate understandings from the reading materials. This comes down to a total research sample size of 141 respondents, which is deemed statistically valid according to Green (1991). In chapter 4.3 the sample size will be explained according to Green’s rule of thumb.
Table 2: Frequency table Chi Square
Test Statistics What is your gender? Chi-Square 10,952a df 1 Asymp. Sig. ,001
Table 2.1: Frequency table Chi Square
What is your gender?
Observed N Expected N Residual
Male 83 63,5 19,5
Female 58 77,6 -19,5
In the upper tables table 2 and 2.1 are the results presented of the chi square tests presented in order to measure validity of the research sample. The test is statistically significant with F (1, 2) = 10,952, p < 0,01. In other words, comparing the observed frequency with the expected frequency illustrates a representative population with a good fit.
Table 3: descriptive statistics provides basic demographic information about the respondents of this research. The variables gender, age, educational, nationality, and whether they are a student or not are presented. The first variable is approximately equally distributed between genders. Due to the fact that the survey was open and available to everybody, the distribution of age ranges from of 18 to 59 years old. Numerical questioning with age was probed to receive more detailed information about the exact age of each respondent. Most respondents were ranged between 21 and 27 years old. In terms of highest completed education more than 80% of all respondents had at least a bachelor degree of applied science. Moreover, 84,4% of the respondents was Dutch and 82,3% of the respondents was currently student. As already mentioned, most results and possible outliers were already filtered out of Qualtrics before the data was transferred to SPSS. However, the dataset still contains some outliers but those are not perilous to the outcome.
Demographic profile N= 141 Frequency Percentage Gender Male Female 83 58 58. 9 % 41.1%
Age Frequency Percentage Age Frequency Percentage
18 2 1.4 % 28 9 6.4 % 19 2 1.4 % 29 3 2.1 % 20 3 2.1 % 30 2 1.4 % 21 9 6.4 % 31 2 1.4 % 22 18 12.8 % 36 1 0,7 % 23 21 14.9 % 38 1 0.7 % 24 14 9.9 % 47 1 0.7 % 25 24 17.0 % 57 1 0.7 % 26 17 12.1 % 58 1 0.7 % 27 9 6.4 % 59 1 0.7 % Educational level Elementary school 2 1.4 %
High school (VMBO/HAVO/VWO) 14 9.9 %
Intermediate vocational education (MBO) 10 7.1 %
Bachelors degree of applied science (HBO) 41 29.1 %
Bachelors degree of university (WO) 37 26.2 %
Masters degree (WO) 36 25.5 %
Doctoral degree (PHD) 1 0.7 % Nationality Dutch 119 84.4 % Other, 22 15.6 % Student Yes 116 82.3 % No 25 17.7 %
4.
Results
Preliminary analysis
The normality of the distribution is shown in table 4. Additionally, plots and histograms were used to check the normality of the data. The minimum, maximum, mean, std. deviation, skewness and kurtosis were processed. The norm used for skewness: Skew/ SE skew< 2 and for Kurtosis: Kurt/ SE skew < 2. Four constructs were skewed, respectively, performance expectancy (-.524/.204), facilitating conditions .891/ .204), hedonic motivation .450/ .204) and behavioural intention (-.415/ .204). Root square resulted in the best solution for all four constructs.
Table 4: Normal distribution
Upper table was only conducted to use as foundation in order to check for outliers. Outliers were identified when z > 3.29 or z < 3.29. In total five outliers were discovered (Barnett, Lewis, 1994). Item 2.2 “using MOOCs increase my chances of achieving things that are important to me” had only one response, which was answered with (1) strongly disagree. The detected outlier was measured at z = 3.78. For question 11 “What is your age?” four outliers were discovered, which are measured at z = 5.64, z = 5.47, z = 5.30 and z = 3.63. Respectively the responses to age were answered with 59, 58, 57, and 47 years old. After deleting the outliers from the analysis the age spread in the normal distribution runs from 18 to 38. Subsequently, the data was tested on reliability, validity, and correlation.
Mean Std. deviation Skewness Kurtosis
Statistic Statistic Statistic Std. Error Statistic Std. Error
Performance expectancy 1.7155 .27459 .067 .204 -.060 .406 Effort expectancy 4.8387 1.04503 .047 .204 -.351 .406 Social influence 3.7116 1.24034 -.223 .204 -.200 .406 Facilitating conditions 1.6848 .30778 .277 .204 .370 .406 Hedonic motivation 1.8338 .31841 -.183 .204 .470 .406 Privacy risk 3.7069 1.30108 .043 .204 -.366 .406 Behavioural intention 1.7952 .32674 -.131 .204 .063 .406 Valid N (listwise)
Furthermore, a multiple regression analysis was conducted followed by a moderation analysis that was executed with the process macro of Andrew F. Hayes for SPSS.
4.1 Reliability
Reliability is the extent to which data collection techniques or analysis procedures yield consistent findings. Consistency can only be determined through multiple measures (Saunders, 2011). Reliability measurements were performed to test accuracy and precision of the measurement instruments. The Cronbach’s Alpha values of each variable were computed. The value of this output is between 0 and 1. According to the literature Field (2013) a value between 0.70 and 0.90 is considered as ideal. Values lower than 0.70 are considered as inconsistent. Values higher than 0.90 could imply resemblance within the constructs, which results in the possibility of redundancy between items. The corrected item – total correlation is valued as a good if the item is higher as 0.30. Lastly, Cronbach’s Alpha if item deleted should be less than 0.10. In the following table (5) the Cronbach’s Alpha for every construct is processed. The complete table can be found in the appendix (1) with the corrected item – total correlation and Cronbach’s Alpha if item deleted of each variable.
Construct Cronbach’s Alpha Number of items
Performance expectancy 0.781 4 Effort expectancy 0.800 4 Social influence 0.892 3 Facilitating conditions 0.790 4 Hedonic motivation 0.902 3 Privacy risk 0.863 3 Behavioural intention 0.698 3
Table 5: Cronbach’s Alpha
Most determinants have a good score on the Cronbach’s Alpha. The construct hedonic motivation has a relatively high Cronbach’s Alpha value with 0.902, which means that variables within this construct look very similar to each other. Through analysing the 3 items, the conclusion can be made that the items fun, enjoyable, and very entertaining on the whole are rather equivalent. In favour of keeping as close as
