Consumer adoption of audiobook streaming services

(1)

Linköping University SE-581 83 Linköping, Sweden

Consumer adoption of

audiobook streaming

services

Casper Engström

Jonathan Nyström Müller

(2)

Acknowledgements

We wish to thank all those who contributed to the successful completion of this thesis. Firstly, we express our profound gratitude to all those who gave feedback during the pilot-study and those who participated in the questionnaire, without which this thesis would not have been possible. We also want to thank the administrators of the Facebook groups for allowing us to publish our questionnaire.

Finally, we wish to thank our supervisor, senior lecturer Aku Valtakoski, for his support and guidance throughout the research process. Aku’s expertise and insights has been a major contributing factor to the quality of this thesis.

(3)

ABSTRACT

The streaming of audiobooks is a new and fast-growing phenomenon amongst Swedish consumers. There exists a lack of knowledge on this topic as it is previously unexplored within academia. As the audiobook streaming services are growing it is important for providers and competitors to understand the key factors determining usage intention from an user’s perspective. This thesis therefore examines what factors influence end-user’s adoption of audiobook streaming-services, the examination extends both for consumers intention to begin using as well as continue using the service. Factors related to the end-user’s usage intention were explored and identified in a Swedish context. A theoretical model consisting of eight hypotheses was constructed to examine the relationships between five variables. The model is based on previous literature of technology adoption and in particular the technology acceptance model (TAM). Additional theory was gathered on previous research on similar technologies such consumers adoption of E-books and E-readers. To achieve the research purpose of the thesis, a quantitative method was used and data was gathered through an online questionnaire with 935 (769 valid) responses. Structural equation modeling (SEM) was used with RStudio to test the collected data on our theoretical model. The results showcased that our model was successful in explaining key factors influencing usage intention. The findings indicate that perceived usefulness and perceived ease of use influence usage intention positively, while end-user’s attachment to paper books and reading impact usage intention negatively. Further results and the implications of these results are discussed.

Keywords

Technology acceptance model (TAM), Diffusion of innovations (DIT), Structural equation modeling (SEM), Information and communications technology (ICT), Audiobooks

(4)

List of Figures

Figure 2.1. Technology acceptance model ... 9

Figure 2.2. Four major categories ... 9

Figure 2.3. The technology acceptance model with emotional attachment... 10

Figure 2.4. Theoretical model... 13

Figure 2.5. Theoretical model with hypotheses ... 16

Figure 4.1. Results for theoretical model in this thesis (Group A) ... 30

Figure 4.2. Results for theoretical model in this thesis (Group B) ... 31

List of Tables

Table 2.1. Summary of hypotheses. ... 16

Table 4.1. Responses. ... 24

Table 4.2. Demographics. ... 25

Table 4.3. Cronbach alpha, Mean and SD ... 26

Table 4.4. HTMT matrix for Group A... 26

Table 4.5. HTMT matrix for Group B. ... 26

Table 4.6. Correlation matrix (Pearson) for Group A ... 27

Table 4.7. Correlation matrix (Pearson) for Group B ... 27

Table 4.8. Confirmatory factor analysis. ... 28

(8)

1. Introduction

1.1. Background

Digitalization and information and communications technology (ICT) has changed how people consume books in Sweden. Literature that was once only available in printed format has now been converted into a digital format.

The digitization of books can be traced back to the 1970’s. With the project Gutenberg the world's first digital library was created, converting physical books into electronic books (E-books) (Hosch, 2019). As the internet became more widespread, E-books followed suit (Connaway and Wicht, 2008).

Another way to consume literature is through audiobooks. Audiobooks are as the name suggests books consumed by listening rather than reading. The first version of audiobooks became possible with Thomas Edison’s invention of the phonograph, he himself made the first recorded audiobook, or phonographic book as he called it (Rubery, 2011). The audiobook is a format that continues to develop since its first appearance from the phonograph to cassette tape to digital files (Rubery, 2011). The next step in the evolution of the audiobook format is that of streaming services. The streaming of audiobooks was first introduced in Sweden in the mid 2000’s (Ingelman-Sundberg, 2007).

Audiobook streaming-services, from here on referred to as “Audiobook-services”, can be divided into two fundamental aspects. The economical aspect refers to the subscription-based fixed price system which is used by the major providers of the service in Sweden such as Storytel, Nextory and BookBeat. Similarly to Spotify, a monthly fee allows the consumer to listen to thousands of available titles at will through a digital application. This allows the consumer to try out literature freely without additional costs, instead of being committed to the purchase of a specific book.

The second aspect is the technological one. Audiobook-services use data streaming to provide the content for its users. Streaming refers to the technology where media is played without being completely downloaded first (Oxford University Press, 2019).

(9)

However, Swedish audiobook-services also provide a download option which allows customers to listen to audiobooks offline. The audiobooks are available through a mobile application and are a part of the ongoing digitalization of the Swedish consumer (Davidsson and Thoresson, 2018).

Audiobook-services has in recent years become a widespread phenomenon in Sweden. In 2018 it accounted for 14,47% of the total sales of literature in Sweden, an increase from 5,41% in 2015 (Wikberg, 2017, 2019). When it comes to the total volume of literature consumed, the growth is even more significant. In 2018, 36,95% of literature consumed was through audiobook-services, an increase from 10,32% in 2015 (Wikberg, 2017, 2019).

This rapid growth demonstrates a massive change in the Swedish literature industry in a relatively short period of time. It is apparent that the consumption pattern of the Swedish consumer has changed when it comes to literature.

This shift is explained by Niclas Sandin, CEO of BookBeat (Interview, Wikberg, 2018, p. 12) [Translated from Swedish], who says that:

“The growth of in particular audiobooks [through audiobook-services] means that the literature-industry remains strong in competition with other digital media such as Spotify and Netflix. [...] Even groups whom previously did not read that much has now started to listen.” He continues: “[…] It feels like a sign of strength for the entire industry that is in the middle of a major reshaping...”

Kristoffer Lind, CEO of the Swedish publishing company Lind & Co (Interview, Wikberg, 2019,p. 13) [Translated from Swedish], further explains that:

“You could think that it would have a positive effect on reading in general [when consumers begin to listen to audiobooks] but when it becomes such an established behavior, customers will be able to forgo printed formats.” He continues: “We stand in front of a paradigm shift where we as retailers or publishers need to think more about the formats.”

(10)

These statements from authorities within the Swedish literature industry emphasize the changes that are occurring in this industry. Audiobook-services are changing the way Swedish consumers behave as they are embracing literature in a new format. This change could also transpire elsewhere as there is no reason to believe that Swedish consumers preferences are unique in this regard.

The literature industry may be following in the footsteps of the music industry where services such as Spotify and Apple Music has had fundamental implication on the music industry. In 2008 physical revenue in music industry were US$ 11.9 billion and stood for roughly 70% of the industry’s total revenue while the streaming as a revenue source were relatively new and amounted to US$ 0.3 billion (International Federation of the Phonographic Industry, 2018). In 2017, ten years later, streaming revenues has grown from US$ 0.3 to US$ 6.6 billion and is today largest source of revenue in the music industry while physical revenue has decreased to US$ 5.2 billion(International Federation of the Phonographic Industry, 2018). If the growth of audiobook-services continues in the current pace, the Swedish literature industry may face the same development as the music industry.

However, the rise of audiobooks through streaming differ from the music industry in the sense that it is not only a new format that affect the consumption of audiobooks, but most likely also affect printed books. Listening to a book is a fundamentally different experience than reading it, and audiobooks has faced criticism as it usually is portrayed as a less legitimate and aesthetic method of literature consumption (Rubery, 2011). Streaming thus not only change the format in which literature is consumed, but also how it is consumed. As the consumer experience is substantially different between audiobook-services and reading printed books, it is of interest to study what factors influence consumers in their decision to adopt the service.

Prior research on the related topic of E-books has found that consumers experiences differ when reading an E-book compared to a printed book, which had implications on the decision to adopt the technology (Antón, Camarero and Rodríguez, 2013, 2017). The differences in experiences between reading and listening to literature and its effects on adoption have however not yet been given much attention in the academic world.

(11)

In addition, research on audiobooks through a business perspective has been done previously (Liu, Li and Yang, 2012; Pinheiro, Odete Fernandes and Teixeira, 2019), however no research on acceptance or adoption of audiobooks has been published.

1.2. Delimitation and Scope of research

The focal point of the thesis is the end-user, both present users and non-users of services. Previous users who have decided to not continue using audiobook-services are excluded from the thesis. As this group would have required a separate analysis the answers from this group were omitted. This decision is motivated by the low response-rate from this group which made the sample size inadequate for statistical analysis.

The topic of this thesis is also limited to audiobook-books and does not include E-books. E-books are also a part of the product which audiobook-services provide. However, they only account for a small part of the total sales and volume of the service. In 2018, only around 7% of sales through these services came from E-books, and only around 6% of volume consumed (Wikberg, 2019).

Furthermore, the topic of E-books is a widely researched one and the topic is relatively well saturated within the academic world. There has been plenty of research conducted regarding consumer behavior and which factors that influence usage intention for E-books and E-readers (Read, Robertson and McQuilken, 2011; Antón, Camarero and Rodríguez, 2013, 2017; Lee, 2013). Meanwhile, audiobooks in general and in particular audiobooks through digital means is a somewhat unexplored topic and there is little to no research conducted on consumer behavior and usage intention within this area. Audiobooks-services are therefore a topic that needs further attention within academia.

1.3. Research Purpose

The purpose of this thesis is to examine the phenomenon of audiobook-services. The thesis will study Swedish consumers’ perception of the phenomenon and aim to further the understanding of technology acceptance and adoption within a previously unexplored area.

(12)

The first research question concerns non-users of audiobook-services who has no previous experience with the service and thus are in the pre-acceptance stage of technology acceptance. This will provide insights into which factors have a significant impact on consumers' willingness to begin using audiobook-services. The first research question is thus as follows:

RQ1: Which factors influence Swedish non-users of audiobook-service intention to begin using the service?

The second research question concerns consumers that have adopted and use audiobook-services instead of or alongside physical books. These consumers are in a post-acceptance stage of technology acceptance as they have already adopted the technology.

This will provide insights into which factors have a significant impact on consumers' willingness to continue using audiobook-services. The second research question is thus as follows:

RQ2: Which factors influence Swedish audiobook-service users’ intention to continue using the service?

To answer these questions, the statistical method of structural equation modeling will be used and a theoretical model will be built and tested. The theoretical model is built upon previous research on the technology acceptance model (TAM) (Davis, 1989) and Rogers (1995) diffusion of innovation theory (DIT). Eight hypotheses are suggested based on earlier research on adoption of E-readers and E-books to be able to find significant factors determining consumers usage intention for audiobook-services.

1.4. Contribution to knowledge

This thesis aims to contribute knowledge to the phenomenon of audiobook-services. By examining which factors influence adoption of the service, a further understanding of what determines usage intention may be found for those in the pre- as well as post-acceptance stage of audiobook-services.

(13)

Furthermore, mediating effects are examined in order to attain a greater understanding of which factors have an indirect impact on intention to use or continue using audiobook-services. This thesis also seeks to contribute to a general understanding of technology acceptance amongst consumers for new ICT’s.

In order to understand the growth of audiobook-services and its implications on the literature industry, an understanding of consumer behavior is necessary. Since consumer behavior has a significant impact on the development of an industry, insights into consumer behavior and the factors that influence consumers usage intention for the phenomenon is crucial for a greater understanding of the digitalization of the literature industry. The further understanding attained on the usage intention of audiobook-services could be of value for both the providers of the service as well as the publishing industry which faces a newly found competition from the service.

(14)

2. Theoretical framework

The following section will outline the theoretical framework used in this thesis. The theories used will together create an analytical framework which will serve as the theoretical basis for the thesis. The theories were found through a literature review of articles related to adoption of E-books and E-readers. Chosen theories has demonstrated effectiveness in explaining usage intention for new ICT’s. The articles were provided and found through the academic search engines of Scopus and Google Scholar.

This thesis proposes a research model of the original technology acceptance model (Davis, 1989) with the external predictor of involvement with new ICT’s (Antón, Camarero and Rodríguez, 2013) and the extension of emotional attachment (Read, Robertson and McQuilken, 2011). This model is integrated with Rogers (1995) technology adoption theory, known as the diffusion of innovation theory. The model aims to identify factors influencing the usage intention of consumers for audiobook-services.

2.1. Diffusion of innovation theory

Diffusion of innovations theory (DIT) refers to the process by which an innovation is spread through a social system (Rogers, 1995). It is a well-known theory that has been used in highly influential research on technology adoption (Venkatesh et al., 2003).

The process of diffusion differs for different innovation. Some innovations get adopted amongst consumers, and some do not. There is also significant difference in the rate of which innovations get adopted, some are adopted quickly amongst consumers, and some at a much slower rate (Rogers, 1995). According to Rogers, the different rate of adoption is determined by the consumers perception of five different factors, relative advantage, compatibility, complexity, trialability and observability. For the purpose of this thesis two of these factors are of particular importance. Namely, relative advantage and complexity.

(15)

2.1.1. Relative advantage

Relative advantage concerns the extent to which an innovation is perceived as being superior than the innovation it supersedes. The more advantageous the innovation is perceived to be, the more rapid its adoption will be amongst consumers. The “objective” advantage of the innovation is of little importance, instead the subjective perception of the consumer is the essential determinant of technology adoption (Rogers, 1995).

2.1.2. Complexity

Complexity refers to the extent to which an innovation is perceived as difficult to understand and use. The perception of complexity naturally differs between different individuals depending on the various characteristics of the individual. In general, however, innovations that are less complex and thus more easily understood will be adopted at a quicker rate than more complicated innovations that may require a greater commitment for the individual to adopt (Rogers, 1995).

2.2. Technology acceptance model

The technology acceptance model (TAM) is an explanatory model that concerns the information technology acceptance of the individual. It is a model that has proven to explain a significant amount of the variance in usage and behavior of individuals when it comes to technology (Venkatesh and Davis, 2000).TAM has also been proven applicable to research fields related to the adoption of audiobooks-services, such as books and E-readers (Lee, 2013; Williams, Slade and Dwivedi, 2014).

The two primary factors determining an individual’s willingness to adopt a technology in this model is the perceived usefulness and perceived ease of use. It is important to note that the measures concern individuals’ subjective perception of the technology and is not an objective measure. This does not indicate a weakness in the theory but is rather the focal point of the theory, since individuals’ perception of technology is an important factor in their willingness to adopt the technology (Davis, 1989).

(16)

Figure 2.1. Technology acceptance model. (Fred D. Davis, Bagozzi and Warshaw, 1989)

TAM has over the years developed as the theory has been refined. Researchers have found many different extensions that are significant key determinants for the model. As these extensions were made, modifications of the model were required to improve it and allow the incorporation of new accompanying factors. These modifications can roughly be divided into four categories (Marangunić and Granić, 2015), which can be seen in Figure 2.2. Two of these categories will be used in our model, namely External predictors and Factors from other theories.

Figure 2.2. Four major categories. (Marangunić and Granić, 2015)

Attitude Toward Using Perceived Ease of Use Behavioral Intention to Use Actual System Use External Variables Perceived Usefulness Intention to Use Perceived Ease of Use Actual System Usage Factors from other theories External Predictors Perceived Usefulness Usage Measures Contextual Factors

(17)

External predictors affect perceived usefulness as well as perceived ease of use and thus have an impact on the intention to use or continue using, while including factors from other theories of technology acceptance increase the predictive validity of the TAM (Marangunić and Granić, 2015). An external predictor as well as a factor from another theory will thus be included in our model.

2.3. Technology acceptance model with emotional attachment

The technology acceptance model with emotional attachment (TAME) is an extended model of TAM (Fred D. Davis, Bagozzi and Warshaw, 1989) and was developed by Read, Robertson and McQuilken (2011). The addition of emotional attachment is to address the effect that consumers’ emotions have on adoption of technology which previously has been neglected in the TAM (Kulviwat et al., 2007). Read, Robertson and McQuilken (2011) found that consumers emotional attachment to paper books had a negative effect on their attitude toward using E-reader technology when reading for pleasure. Emotional attachment might influence consumers’ intention to use or continue using audiobook services and will thus be included in our model.

Figure 2.3. The technology acceptance model with emotional attachment. (Read, Robertson and McQuilken, 2011)

Perceived usefulness Perceived Ease of Use

Attitude Behavioral intention to Adopt

Emotional Attachment

(18)

2.4. Constructs

2.4.1. Perceived usefulness

Perceived usefulness (PU) traditionally refers to the extent which “a person believes that using a particular system would enhance his or her job performance” (Davis, 1989, p. 320). However, perceived usefulness has been found to be an adequate predictor of technology acceptance outside of work-related activities, such as users’ willingness to revisit a website (Castañeda, Muñoz-Leiva and Luque, 2007) or consumers intentions to use E-readers (Williams, Slade and Dwivedi, 2014). As such, perceived usefulness does in this thesis concern the perception of usefulness from the perspective of a user, or a potential user. This construct shares similarities with Roger’s construct of relative advantage. Moore and Benbasat (1991) even argues that perceived usefulness is clearly related to relative advantage. This idea has been adopted in later research where relative advantage has been integrated with the technology acceptance model (Tung et al., 2009; Lee, 2013).

2.4.2. Perceived ease of use

Perceived ease of use (PEOU) refers to “the degree to which a person believes that using a particular system would be free of effort” (Davis, 1989, p. 320). This traditional definition is still applicable to the subject of this thesis and will thus be used. Perceived ease of use is also highly related with Roger’s construct of complexity (Moore and Benbasat, 1991). The construct has a diametrical relationship, this means that a technology with high perceived complexity also has a low perceived ease of use and vice versa.

2.4.3. Information and communications technology

The construct Information and communications technology (ICT) concern the individuals’ involvement with new ICT’s and will be the external predictor of our model. This includes the attitude towards as well as prior usage of new technology. People with a high value for this variable might feel little resistance toward audiobook-services as early adoption of new technology is compatible with their lifestyle and self-image. They are also more likely to perceive the technology as useful, as well as easy to use (Antón, Camarero and Rodríguez, 2013).

(19)

2.4.4. Attachment to paper books and reading

Attachment to paper books and reading (ATT) add an emotional element to our model as it includes the emotional attachment of paper books and reading, which may improve the model’s predictive capability (Antón, Camarero and Rodríguez, 2017). Read, Robertson and McQuilken (2011) suggest that the emotional attachment to paper books amongst some respondents is caused by the format, namely a physical book. This emotional attachment acts as a resistance mechanism for innovations which means that an E-reader cannot easily replace the enjoyment of reading a physical book. We argue that this factor is relevant for audiobook-services as well and that it is also likely that there might be an emotional attachment to reading itself, which could influence consumers’ view on audiobook-services since you do not read audiobooks. Attachment to paper books and reading is thus appropriate for our model as a factor from another theory.

2.4.5. Intention to use and Intention to continue using

Intention to use (ITU) and intention to continue using (ICU) is the dependent variable for consumers in the pre-acceptance stage and post-acceptance stage respectively. The variance in the dependent variable is presumed to be affected by the previously mentioned constructs(Bryman and Bell, 2011).

Intention to use refers to non-users’ intention to begin using audiobook-services or in other words, the pre-acceptance stage of a new technology. Intention to use is a construct derived from the original technology acceptance model (Davis, 1989) and has been used regularly in research related to adoption of E-books and E-readers (Lee, 2013; Williams, Slade and Dwivedi, 2014). Intention to continue using is a construct for consumers who use audiobook-services, or in other words, consumers who is in the post-acceptance stage of a new technology. The construct refers to their intention to continue using the service and has been used in a similar fashion in earlier research on E-books (Jin, 2014) and other new technologies (Castañeda, Muñoz-Leiva and Luque, 2007).

(20)

2.5. Hypotheses and theoretical model

In order answer the research questions, an extensive literature review was made, and a theoretical model was hypothesized. The theoretical model is based upon the hypothesized relationship between the different constructs.

Figure 2.4. Theoretical model

Antón et al (2013) found a relationship between respondents’ involvement with information and communications technology and perceived ease of use for E-books. In addition, Lu, Yao and Yu (2005) found that “Personal Innovativeness”, as an external variable in the technology acceptance model, has a significant positive relationship for perceived usefulness and perceived ease of use of adoption of wireless internet services via mobile technology. These relationships are also anticipated for the construct ICT in our model, where perceived usefulness and perceived ease of use instead concern audiobook-services. Thus, the following two hypotheses are proposed for our model:

H1: Involvement with new information and communication technologies has a positive effect on perceived usefulness.

H2: Involvement with new information and communication technologies has a positive effect on perceived ease of use.

PU

ICT

PEOU

(21)

2.5.2. Perceived usefulness & Perceived ease of use

The technology acceptance model (Davis, 1989; Fred D. Davis, Bagozzi and Warshaw, 1989) has established the relationship between perceived usefulness, perceived ease of use and usage intention. This relationship has been confirmed in many previous articles (Read, Robertson and McQuilken, 2011; Lee, 2013) and is thus hypothesized to also be valid for audiobook-services:

H7: Perceived ease of use has a positive effect on perceived usefulness.

H4a: Perceived usefulness has a positive effect on intention to use.

H4b: Perceived usefulness has a positive effect on intention continue using.

H6a: Perceived ease of use has a positive effect on intention to use.

H6b: Perceived ease of use has a positive effect on intention to continue using.

It is also predicted that perceived usefulness and perceived ease of use has a negative relationship for attachment to paper books and reading. Lee (2013) found that perceived usefulness and perceived ease of use had a negative impact on innovation resistance.

We argue that attachment to paper books and reading act as a type of innovation resistance. As previously established, a high perceived usefulness indicates a high perceived relative advantage for audiobook-services compared to other alternatives of literature consumption. This may cause a lower level of attachment to paper books and reading as the individual is willing to see the benefits of the new technology. Ram (1989) claimed that perceived complexity can increase the resistance towards an innovation. We hence argue that a high perceived ease of use decreases the emotional attachment to paper books and reading. Based on this, the following two hypotheses are proposed for our model:

H3: Perceived usefulness has a negative effect on attachment to paper books and reading.

H5: Perceived ease of use has a negative effect on attachment to paper books and reading.

(22)

Antón, Camarero and Rodríguez, (2017) and Read, Robertson and McQuilken (2011) showcased the negative relationship between attachment to paper books and usage of E-readers. This factor is as previously mentioned potentially also relevant for audiobook-services and might act as form of resistance to audiobook-audiobook-services, and therefore have a negative impact on the intention to use/continue using the service. Thus, a final hypothesis is proposed:

H8a: Attachment to paper books and reading has a negative effect on intention to use. H8b: Attachment to paper books and reading has a negative effect on intention to continue using.

(23)

2.5.4. Summary of theoretical model and hypotheses

Figure 2.5. Theoretical model with hypotheses.

H1: Involvement with new information and communication technologies has a positive effect on perceived usefulness.

H2: Involvement with new information and communication technologies has a positive effect on perceived ease of use.

H3: Perceived usefulness has a negative effect on attachment to paper books and reading. H4a:

H4b:

Perceived usefulness has a positive effect on intention to use.

Perceived usefulness has a positive effect on intention to continue using.

H5: Perceived ease of use has a negative effect on attachment to paper books and reading. H6a:

H6b:

Perceived ease of use has a positive effect on intention to use.

Perceived ease of use has a positive effect on intention to continue using. H7: Perceived ease of use has a positive effect on perceived usefulness. H8a:

H8b:

Attachment to paper books and reading has a negative effect on intention to use. Attachment to paper books and reading has a negative effect on intention to continue using.

Table 2.1. Summary of hypotheses.

H4 H6 H8 H5 H3 H7 H2 H1

PU

ICT

PEOU

ATT

ITU/ICU

(24)

3.Method

3.1. Research design

Different approaches can be used to connect theory and empiricism. Quantitative research strategy emphasizes quantification of data and with the gathered data test existing theory (Bryman and Bell, 2011). Quantitative studies primarily have a deductive approach, which in social sciences represent the most common perception on the relationship between theory and practice (Bryman and Bell, 2011). Deductive reasoning essentially means that conclusions on a specific phenomenon can be drawn from the basis of general assumptions (Olsson and Sörensen, 2011). Observations derive from theory in deductive theory which is the opposite to an inductive approach, where theory derive from observations and results. (Bryman and Bell, 2011).

The intent of this thesis was to test existing theories on technology adoption on a phenomenon that is previously unexplored, as such our aim was not to construct new theories, but rather confirming existing ones. The thesis aim was to confirm a number of hypotheses which if proven significant would affirm the theories of which the hypotheses is built upon. This research purpose called for a deductive quantitative approach which was used, as the aim was to confirm existing theories and test hypotheses. This research methodology has been used on topics of similar research such as (Lee, 2013; Antón, Camarero and Rodríguez, 2017) on the subjects of E-books and E-readers.

A cross-sectional study was used as research design for the thesis. The design of this method is to gather quantifiable data from a wide range of individuals at a certain point in time. The variation in the data is then studied in order to examine possible relationships between the different variables chosen (Bryman and Bell, 2011).

A weakness with this methodology is that information is only gathered at one specific point in time. Changes in attitude is thus not accounted for and cannot be studied. Consequently, the psychometric values of the data may have a certain bias from for instance seasonal differences. It is possible that for example the perception of usefulness may differ between the month of march (the month in which the data was gathered) and other months of the year.

(25)

3.2. Survey design

In order to gather the desired data, a survey was conducted. A survey is a method of sampling data from a population. It gathers quantitative information through a self-administeredquestionnaire and with that sample makes statistical inferences for the population as a whole (Dillman, Hox, de Leeuw, 2012).

An internet survey also generally results in quicker response times than other alternatives, as delivery of the survey is instantaneous, and the response time is also generally very swift (Denscombe, 2010). According to the research of Denscombe (2008), the usage of online questionnaires does not appear to have any significant impact on the respondents in comparison to the more traditional paper questionnaires. Denscombe’s research (2008) also gives a general assurance as to the methodological adequacy of online-based questionnaires.

A potential downside to a survey-design is the inherent inflexibility of the method. The inadequate phrasing of a question, such as spelling errors, may have unforeseen consequences which cannot be adjusted for in retrospect. To prevent this, an initial pilot study was conducted after which some minor adjustments were made to make the questions more comprehensible. After this, the questionnaire was made available in several different Facebook groups. A reminder was also posted on the Facebook group-pages roughly one week after the initial post. The questionnaire was created with the software Google Forms. The advantages of an internet-based survey are that they are cheap to conduct and easy to code as the data is automatically downloaded into a database. It also allows for “skip logic” where a specific question can be used as a filter and take the respondent to different question sections depending on their answer (Bryman and Bell, 2011). This is especially important as the thesis concerns two distinct groups and the questions are adapted to each group.

Even though the questionnaire was sent out through one link, the skip logic question acts as a separator for the respondents between those who use audiobook-service and those who do not. Thus, two questionnaires were in reality created, and they will be treated as two different, albeit similar questionnaires. Questionnaire A concern non-users of audiobook-services and questionnaire B concern users of audiobook-services.

(26)

3.3. Questionnaires

The questionnaires were primarily based upon the questions used in previous similar research and is thus a theory driven survey. After an extensive literature review, six different articles were eventually used for questionnaire A and five for questionnaire B. The questions are gathered from research on the relationship between physical books, E-readers and E-books.

The topic of E-books is wide-studied compared to the topic of this thesis but is a similar area of research. A firm selection process for the questions combined with some minor adjustments as well as addition of supplementary questions, made the questions applicable for the thesis. The questions were also translated from English to Swedish as the thesis address Swedish consumers and the likelihood of receiving answers from this group would most likely increase if the survey was translated to their native tongue.

The translations were not made completely literal but rather an attempt to translate the essence of the question in English to Swedish. Thus, possible confusion from respondents caused by linguistic issues were limited. There is still however a risk that the Swedish version of the questions may be interpreted differently than the English, thus somewhat limiting the comparability between this thesis and the previous research of which the questions were gathered.

The questionnaire was developed with a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree) where 3 can be seen as a neutral response. The Likert scale have been used in similar research related to the E-book and E-readers (Lai and Chang, 2011; Lee, 2013) and is an index which purpose is to measure attitudes (Olsson and Sörensen, 2011).

All items are considered to be equal and parallel instruments, which means that random errors in an individual item partly is averaged away (van Alphen et al., 1994). Even though Likert scales strictly are ordinal, meaning that the distance between different categories are unequal (Bryman and Bell, 2011), parametric methods (which is used on Likert scale data) turns out be “incredibly versatile, powerful and comprehensive” (Norman, 2010, p. 627).

(27)

Two questionnaires with 19 items each compromising five dimensions was constructed. The first questionnaire, which can be seen in the Appendix B, was constructed for respondents who are in the pre-acceptance stage of audiobook-service adoption and was compromised by five dimensions -   information and communication technologies, perceived ease of use, perceived usefulness, attachment to paper book and reading and intention to use. The second questionnaire, which can be seen in Appendix C, was constructed for respondents who are in the post-acceptance stage of audiobook-service adoption and was also compromised by five dimensions - information and communication technologies, perceived ease of use, perceived usefulness, attachment to paper books and reading, and intention to continue using.

3.3.1. Items

Items selected for measuring the constructs was recast from previous inventories to confirm the scales content validity, since items must describe the concepts about which generalizations are to be made. Items ICT1 to ICT4, used for measuring interest in or involvement with new information and communication technologies, were revised from Antón, Camarero and Rodríguez (2013) and Lee (2013). Items PEOU1 to PEOU4 is used for measuring perceived ease of use and is referenced from Porter & Donthu (2006), Williams, Slade and Dwivedi (2014) and Read, Robertson and McQuilken (2011). The dimension of perceived usefulness is measured by four items (PU1 to PU4) and were derived from Porter & Donthu (2006) and Williams, Slade and Dwivedi (2014). Items ATT1 to ATT4 were adapted from Antón, Camarero and Rodríguez (2013, 2017) and is used for measuring attachment to paper books and reading. Item ATT4 was reworded to avoid a double negative wording. The values gathered from this item was reversed in Excel. Intention to use and intention to continue using is measured by three items each (ITU1 to ITU3 and ICU1 to ICU3), items two and three for both constructs were derived from Read, Robertson and McQuilken (2011) while ITU1 and ICU1 were adapted from Lai and Chang (2011) and Lee (2013) respectively.

(28)

3.4. Hypothetical model

A model and set of hypotheses were made for the two distinct groups. The first (group A) concerns non-users of audiobook-services. The second (group B) concern respondents who use audiobook-services. The groups are not completely comparable as the questions asked are not the same, but rather are formulated to suit the characteristics of each group. The questions are very similar with only minor linguistic adjustments. As the questions are not exactly the same, comparisons between the groups must be made with care and with appreciation for the limitations of their comparability. The software used for the analysis of the model was RStudio version 3.5.3 and SPSS version 25.

3.5. Sampling

A non-probability sampling was used for the survey. The research questions concern a distinct population, namely the users of audiobook-services in Sweden. This population is a relatively small part of the overall Swedish population. Unfortunately, time restraints and lack of sufficient information regarding the population of audiobook-service users had an impact on the sampling-method.

In order to get an adequate amount of responses from this group, the questionnaire was as previously mentioned sent out to several different Facebook groups namely, “Storytel-vi som gillar Ljudböcker”, “Vi som älskar att läsa böcker” and “Vi som älskar ljudböcker”. These groups had a connection to the topic of the research as they were Facebook groups created for the discussion of both printed books and audiobooks-services. This would according to Denscombe (2010) be considered a purposive sample. Since we assumed that a significant portion of the people that are a part of these groups are users of audiobook-services, the groups and thus the respondents, were chosen purposefully based on a known characteristic relevant to the topic of research. This method for sampling was used to reach a population group that would have been difficult to reach otherwise.

(29)

This type of sampling does however not fulfill the requirements of obtaining a representative sample as the sample is not random excerpt from the studied population (Denscombe, 2010). The members of the different Facebook groups may have certain characteristics that makes an impact on the data but is not something representative to the general population. This also made the sample limited since in order to be a member of the groups, it is obviously a prerequisite to have an account on Facebook.

As the data is not entirely representative, generalizations of the results would be unreliable. Members of the chosen Facebook groups may for instance have a stronger attachment to audiobooks or reading as they have chosen to join a discussion group on the topics. They may also have a higher involvement with new ICT’s since they are active on a platform such as Facebook. This may skew the results and the sample may not be representative for the average user of audiobook-services or the average book reader.

A further concern is the matter of population size. No official nor reliable information on how many Swedish consumers use audiobook-services is available. The only provider of audiobook-services who publishes its number of subscribers is the public company Storytel who reveal that they have roughly 345 000 (Storytel, 2019) paying subscribers in December 2018. As the topic of this thesis concern users of audiobook-services, not paying customers, this creates further issues. These numbers concern active accounts, but several users may use the same account at once which would indicate that the number of users is higher than reported.

(30)

3.6. Structural equation modeling

Structural equation modeling (SEM) is a statistical model that has been widely used for psychometric research. It has been proven as an applicable model for research on similar topics such as E-books (Lee, 2013) and E-readers (Read, Robertson and McQuilken, 2011). Structural equation modeling is not one, but rather many different statistical techniques used in combination (Nachtigall, Ulf Kroenhe, Steyer, 2003). The overarching purpose is to test a theoretical model by examining how variables create constructs, and the relationship between these constructs (Schumacker and Lomax, 2010). These constructs are latent, meaning they are not explicitly measured but rather inferred from our observable data (Schumacker and Lomax, 2010). At least three different items for every construct is included in the questionnaire.

The advantage of SEM is that it enables an analysis of the relationships of the latent variables and the dependencies of the constructs without measurement errors (Nachtigall, Ulf Kroenhe, Steyer, 2003). A general “rule of thumb” is that a SEM should have a sample size of at least 200 (Kline, 2013).

3.7. Ethics

The research was conducted in accordance with the Swedish research council principles for ethical research. Information concerning the topic of the research was given to participants of the questionnaire and it was also mentioned that the questionnaire is voluntary (Swedish Research Council, 2012). No ethically sensitive information was gathered, and no individual participant of the questionnaire was identifiable since no names, e-mail addresses or other information that can be used for the identification of an individual was gathered. The information collected was only used for research purposes and not for any commercial reason, which we also informed the respondents (Swedish Research Council, 2012). Permission was asked and granted by the chosen Facebook-groups administrators before the link was made available for the members of the Facebook-groups.

(31)

4. Results

4.1. Data preparation

The total number of responses for the questionnaire was 935. 80 of these were initially removed as they were not applicable to the purpose of this study. These were the answers of those who have previously used subscription-based service for audiobooks but does no longer do so. As the research questions only concern users and non-users with no previous experience of audiobook-services, the responses of those who have previously used the service but no longer do so was deemed irrelevant for the thesis. This brought the number of responses down to 855. Further removals were made for those responses with missing values. This brought the number of answers down to 469 for users of audiobook-services and 300 for those who have never used it. In total 769 answers were used. The sample size for both groups is substantially higher than Kline’s (2013) rule of thumb that requires a sample size of 200 respondents for structural equation modeling.

Initial responses Responses after removal of answers with missing values

Users of audiobook-service Non-users of audiobook-service Total responses 500 355 855 469 300 769 Table 4.1. Responses.

(32)

4.2. Respondents demographics

Measure Profile Category Group A (N = 300) Group B (N = 469) Frequency Percentage Frequency Percentage Gender Age Education Income (SEK /month) Female Male Other/n. a <18 18-29 30-49 50-64 65+ Other/n.a Elementary school High school Some collage College graduate Other/n. a < 10 000 10 000 – 19 000 20 000 – 29 000 30 000 – 39 000 40 000 – 49 000 > 50 000 Other/n.a 252 46 2 0 61 74 100 63 2 8 62 71 155 4 33 78 64 50 24 19 32 84.00 15.33 0.67 0 20.33 24.67 33.33 21.00 0.67 2.67 20.67 23.67 51.67 1.33 11.00 26.00 21.33 16.67 8.00 6.33 10.67 436 30 3 1 19 179 214 56 0 16 117 44 287 5 11 70 108 131 71 34 44 92.96 6.40 0.64 0.21 4.05 38.17 45.63 11.94 0 3.41 24.95 9.38 61.19 1.07 2.34 14.92 23.03 27.94 15.14 7.25 9.38 Table 4.2. Demographics.

The demographics of the respondents showcases a massive majority of women. It also appears to be a generally highly educated group of respondents. According to Nordicom-Sweden's media Barometer (2019) both women and the highly educated in Sweden does in general consume both physical books and audiobooks at a higher rate and for longer periods of time than men and people with less education. This may explain some of the high frequency of these two groups amongst the respondents. As the population size of both groups are unknown it is difficult to say with certainty that the sample is not representative, however the vast majority of women in the sample makes it seem very likely. This is most likely due to an overrepresentation of women in the Facebook-groups in which the questionnaire was posted.

(33)

4.3. Reliability and Validity

4.3.1. Reliability

A Cronbach alpha test was conducted for the variables. Cronbach alpha is a measure of the internal reliability of the variables. According to Bryman and Bell (2011) a Cronbach alpha value of 0.8 is generally considered to be an acceptable level as a rule of thumb. However, Kline (2013) claim that a Cronbach alpha of 0.8 is a very good value and 0.7 is considered adequate. As seen in Table 4.3., most values are very good, while some are adequate.

Group A (N=300) Group B (N=469)

Variable Cronbach Alpha (standardized) Mean Std. Deviation Cronbach Alpha (standardized) Mean Std. Deviation ICT PEOU PU ATT ITU/ ICU 0.858 0.891 0.813 0.761 0.848 2.775 3.657 2.248 3.883 2.098 0.846 0.960 0.913 0.982 0.999 0.883 0.925 0.788 0.796 0.926 3.252 4.651 4.163 2.677 4.687 0.878 0.533 0.745 1.085 0.649

Table 4.3. Cronbach alpha, Mean and SD.

Note: Since the values are in a range of 1-5 on a Likert scale, the minimum value is 1 and the maximum value is 5 for all variables

4.3.2. Validity

Table 4.4. HTMT matrix for Group A.

ICT PEOU PU ATT ICU ICT -

PEOU 0.127 -

PU 0.138 0.643 -

ATT 0.117 0.188 0.430 -

ICU 0.079 0.598 0.830 0.403 -

Table 4.5. HTMT matrix for Group B.

ICT PEOU PU ATT ITU ICT -

PEOU 0.419 -

PU 0.254 0.441 -

ATT 0.274 0.300 0.512 -

(34)

The heterotrait-monotrait ratio of correlations (HTMT) is a test for discriminant validity. Discriminant validity, introduced by Campbell and Fiske (1959), measures whether the constructs are distinct from each other or not (Kline, 2013). Measuring discriminant validity is in general considered a requirement for the analysis of relationships between latent variables for structural equation modeling (Henseler, Ringle and Sarstedt, 2014). The most conservative threshold for HTMT is according to Henseler, Ringle and Sarstedt (2014) below 0.85. This would imply that the correlation between ITU and PU for group A would be insufficient. However, the threshold is debatable and 0.90 is also considered acceptable (Henseler, Ringle and Sarstedt, 2014).

4.4. Correlation (Pearson)

Table 4.6. Correlation matrix (Pearson) for Group A. Correlations significant on 0.01 level (**)

Table 4.7. Correlation matrix (Pearson) for Group B. Correlations significant on 0.01 level (**)

Correlations significant on 0.05 level (*)

ICT PEOU PU ATT ITU ICT -

PEOU 0.373** -

PU 0.218** 0.380** -

ATT -0.225** -0.245** -0.392** -

ITU 0.187** 0.391** 0.717** -0.392** -

ICT PEOU PU ATT ICU ICT -

PEOU 0.109* -

PU 0.118* 0.536** -

ATT 0.024 -0.154** -0.333** -

(35)

4.5. Goodness of Fit

In order to test the fit of the model, a structural confirmatory factor analysis (CFA) was conducted to attain several goodness of fit measures. These values were used to measure how well the model fit the data. This tells us how applicable the model is to the gathered data (Kline, 2013). Three specific measures are further analyzed below. Additional measures can be found in the confirmatory factor analysis in Table 4.8.

The first one is the root mean square error of approximation (RMSEA) which is an absolute fit index. What is considered an acceptable RMSEA is generally reported as a maximum of 0,08 (Hooper, Coughlan and Mullen, 2008). The RMSEA value for users of audiobook-services was 0,079 and for non-users 0,073 and thus at an acceptable level. Hu and Bentler (1999) however, argue that a RMSEA value lower than 0,06 is required for an adequate model fit.

Comparative fit indices (CFI) are an incremental fit index with a value between 0-1 where values closer to 1 indicate a better fit. A good fit is generally considered to be higher than 0,90 (Bagozzi and Yi, 1988), Hu and Bentler (1999) claim that a 0,90 is acceptable but 0,95 is required in order for the fit to be considered good. The CFI for those who use audiobook-services was 0,934 and for those who do not 0,932. This indicate an acceptable fit.

According to Hu and Bentler (1999) a Tucker-Lewis index (TLI) of close to 0,95 is required to confirm a good fit between the hypothesized model and the data. The TLI value for users of audiobook-services was 0,921 and for those who do not 0,919. The TLI-values are thus not quite high enough to ensure a good fit, but relatively close.

Measures Group A (N=300) Group B (N=469)

RMSEA 0.073 0.079 SRMR 0.087 0.081 CFI 0.932 0.934 TLI 0.919 0.921 ECVI 1.557 1.391 AIC 14421.992 17463.855 X2 (df) 371.015 (142) 556.404 (142)

(36)

Overall, the measurement model showed a decent model fit for both groups and the model fit is thus deemed adequate. The goodness of fit measures did not differ substantially between the groups as can be seen in Table 4.8. Which indicate that the model is applicable for both users and non-users of audiobook-services.

4.6. T-test

Independent samples tests were done for the model’s constructs. The T-tests tested if there was a significant difference in the mean values of the constructs for those who do not use audiobook-services (group A) and those who do (group B). The differences between all the constructs were significant at a confidence interval of 99% (Appendix A). This implies that it is reasonable to assume that there is a significant difference between the constructs of the two groups.

4.7. Path analysis

A path analysis is a test of the structural model and its assumptions (Kline, 2013). The method extends the possibilities of multiple regression as it allows for the testing of more complex models (Streiner, 2005). However, since we do not know the true causal model between the variables and instead hypothesize the model, we cannot with certainty prove causal relationships. As such, the path analysis only proves that our hypothesized model fits with the data, not that the model corresponds with reality (Kline, 2013).

Hypothesis Group A Group B

Standardized coefficients Supported Standardized coefficients Supported H1: ICT → PU 0.088 (1.537) NO 0.060 (1.538) NO H2: ICT → PEOU 0.373 (6.972)*** YES 0.109 (2.369)** YES H3: PU → ATT -0.349 (-6.122)*** YES -0.351 (-6.816)*** YES H4: PU → ITU/ICU 0.623 (13.905)*** YES 0.485 (12.319)*** YES H5: PEOU → ATT -0.112 (-1.969)* YES 0.035 (0.675) NO H6: PEOU → ITU/ICU 0.125 (2.946)** YES 0.273 (7.262)*** YES H7: PEOU → PU 0.348 (6.063)*** YES 0.530 (13.542)*** YES H8: ATT → ITU/ICU -0.117(-2.738)** YES -0.139 (-4.136)*** YES

Table 4.9. Path analysis.

(37)

4.7.1. Test of hypotheses (pre-acceptance stage)

For group A were seven of eight paths significant at a p level of .05. Hence were seven hypotheses (H2, H3, H4, H5, H6, H7 and H8) supported and one (H1) rejected. ICT did not have a significant positive effect on perceived usefulness (H1) (β = .088, ns). ICT did however have a highly significant effect on perceived ease of use (H2) (β = .373, p < .001).

As hypothesized, perceived usefulness (H4) (β = .623, p < .001) and perceived ease of use (H6) (β = .125, p < .01) have a positive and significant effect on the intention to use. Moreover, did perceived ease of use have a highly significant positive influence on perceived usefulness (H7) (β = .348, p < .001). Perceived usefulness (H3) (β = -.349, p < .001) as well as perceived ease of use (H5) (β = -.112, p < .05) did have a significant negative influence on attachment to paper books and reading.

Furthermore, the standardized path coefficient of attachment to paper books and reading to the intention to use was highly significant in a negative way (H8) (β = -.117, p < .001).

Figure 4.1. Results for theoretical model in this thesis (Group A). Note: *p < .05, **p < .01, ***p < .001 (H4) 0.623*** (H6) 0.125** (H8) -0.117** (H5) -0.112* (H3) -0.349*** (H7) 0.348*** (H2) 0.373*** (H1) 0.088

PU

ICT

PEOU

ATT

ITU

(38)

4.7.2. Test of hypotheses (post-acceptance stage)

The result of the path analysis (Figure 4.2.) shows that six (H2, H3, H4, H6, H7 and H8) out of eight hypotheses were supported for group B since their paths were significant at a p level of .01, which can be seen in Table 4.9. Thus, were H1 and H5 rejected. This means that ICT, had a positive and significant effect on perceived ease of use (H2) (β = .109, p < .01). ICT did however not have a significant positive effect on perceived usefulness (H1) (β = .060, ns).

Both perceived ease of use (H6) (β = .273, p < .001) and perceived usefulness (H4) (β = .485, p < .001) proved to have a highly significant positive influence on the intention to continue using. Perceived usefulness did also have a highly significant negative influence on attachment to paper books and reading (H3) (β = -.351, p < .001), while perceived ease of use did not have a significant effect on attachment to paper books and reading (H5) (β = .035, ns).

The standardized path coefficient of attachment to paper books and reading to the intention to continue using was highly significant in a negative way (H8) (β = -.139, p < .001). Our data also support H7 which means that perceived ease of use has positive and highly significant effect on perceived usefulness (β = .530, p < .001)

Figure 4.2. Results for theoretical model in this thesis (Group B). Note: *p < .05, **p < .01, ***p < .001 (H4) 0.485*** (H6) 0.273*** (H8) -0.139*** (H5) 0.035 (H3) -0.351*** (H7) 0.530*** (H2) 0.109** (H1) 0.060

PU

ICT

PEOU

ATT

ICU

(39)

5. Discussion of Results

The overarching goal with this thesis was to test if the technology acceptance model with our modifications and extensions was applicable to the relatively new and unexplored area of audiobook-services. With seven out of eight hypotheses confirmed as significant for group A and six out of eight hypotheses confirmed as significant for group B, this goal was in large achieved. The theoretical model that was built has a relatively strong explanatory capability for adoption, both for consumers in the pre- and post-acceptance stage of audiobook-services adoption.

5.1. Discussion of group A (pre-acceptance stage)

H2 was significant with a standardized coefficient of 0.373. This implies, that people's involvement with new ICT’s has a positive effect on the perceived ease of use of audiobook-services. Greater involvement with new ICT’s could therefore contribute to greater confidence in using new technology, including technology that the consumer has no previous experience in using. This is coherent with previous findings on innovativeness (Lu, Yao and Yu, 2005; Lee, 2013)which suggests that involvement with new ICT’s effect perceived ease of use in similar way.

Involvement with ICT did however not have a significant effect on perceived usefulness (H1). This indicate that non-users of audiobook-services with confidence in new ICT’s does not necessarily perceive audiobook-services as useful. The perception of usefulness is thus not dependent on previous experience of new ICT, but rather on other factors.

5.1.2. Perceived usefulness

Perceived usefulness had a highly significant and negative effect on attachment to paper books and reading (H3). This indicate that a perceived relative advantage of audiobook-services for non-users decreases the emotional attachment to paper books and reading and thus decrease the resistance towards the innovation.

(40)

This phenomenon could be explained by Rogers (1995) theory of relative advantage. Potential consumers who perceive audiobook-services as highly useful will likely perceive it to have a relative advantage in comparison to paper books and reading. As the perceived usefulness and thus perceived relative advantage of audiobook-services increases, the emotional attachment to paper books and reading decreases. It thus seems like perceived usefulness of new ICT’s decreases the user’s innovation resistance.

Perceived usefulness has as predicted a positive influence on non-user's intention to use audiobook-services (H4). This means that non-users who perceive audiobook-services as useful has a greater intention to adopt the service which is in accordance with Davis (1989) original theory of technology acceptance. Despite having no experience in using audiobook-services, consumers naturally have a higher intention to begin using the service if they perceive it as useful and advantageous in comparison to paper books and reading.

5.1.3. Perceived ease of use

Perceived ease of use had significant effect on attachment to paper books and reading (H5), although the relationship was considerably weaker than all the other supported hypothesizes. Nonetheless, perceived ease of use appears to have a negative impact on non-users’ emotional attachment to paper books and reading. This implies that low perceived complexity amongst non-users of audiobook-services reduces their resistance toward the innovation. This relationship has also been found for E-books by Lee (2013). Perceived ease of use has similarly to perceived usefulness a positive effect on intention to use audiobook-services (H6) in congruence with the original TAM (Davis, 1989). Even though perceived ease of use does not have as strong of an effect as perceived usefulness on the intention to use, it is still significant. Hence, consumers' willingness to adopt a new technology is dependent on their perception of the complexity of the service. A service that is perceived as complex is most likely also perceived as complicated and thus may create a threshold for consumers intention to adopt the service.

(41)

Perceived ease of use’s effect on perceived usefulness was as hypothesized strong and highly significant (H7). Even for a consumer that has never used an audiobook-service, it appears that they have a higher perception of usefulness if their perception of complexity is low. Despite having no experience with the service, consumers tend to consider ease of use as a prerequisite for usefulness in new technology. If non-users of audiobook-services perceive the product as easy to begin using, they may more easily recognize the advantages of the innovation.

A general understanding of how audiobook-services works is therefore naturally also of importance in order to perceive the service as advantageous. It appears that in order to perceive audiobook-services as useful, one must perceive the complexity of the service as manageable.

As hypothesized, attachment to paper books and reading had a negative effect on intention to use for non-users of audiobook-services (H8). The emotional attachment to paper books and reading appears to be a hindrance to the adoption of audiobook-services and thus act as a type of resistance mechanism. This resistance is most likely due to the consumers’ attachment to reading instead of listening, as well as the emotional attachment and intrinsic value of a physical paper book.

5.2. Discussion of group B (post-acceptance stage)

H2 was significant with a standardized coefficient of 0.109. This implies that people's involvement with ICT has a positive effect on the perceived ease of use of audiobook-services. This is consistent with Lu et al. (2005) and Lee’s, (2013) findings on innovativeness which implies that the construct ICT works in a similar fashion. This suggests that users with greater involvement with new ICT tend to be more confident in using new technology, as such they perceive new ICT as less complicated and hence find audiobook-services easier to use.

Consumer adoption of audiobook streaming services