An Explanation of the Adoption of Mobile Internet: A Model Comparison Approach

An Explanation of the Adoption of Mobile Internet:

A Model Comparison Approach

by

Hylke Hoekstra

An Explanation of the Adoption of Mobile Internet:

A Model Comparison Approach

by

Hylke Hoekstra

University of Groningen Faculty of Economics and Business

Department of Marketing

January, 2012

Name Hylke Hoekstra

Student ID s1921894

Course program Marketing Management

Table of Content

2. The Mobile Internet Context ... 2

2.1 Differences Between Mobile Internet and Stationary Internet ... 2

2.2 Past Research on Mobile Internet ... 3

3. Multi-Attribute Models in the Context of Technology Acceptance ... 4

3.1 Theory of Planned Behavior ... 6

3.2 Technology Acceptance Model ... 7

3.3 Comparison of Two Theories ... 8

3.4 Hypothesis ... 9

3.5 Overview of Theoretical Framework ... 10

4. Methodology ... 10

4.1 Ensuring a Fair Comparison ... 10

4.2 Operationalization of Constructs ... 11

5. Results ... 13

5.2 Convergent Validity ... 14

5.3 Discriminant Validity ... 16

5.4 Explaining Behavioral Intention ... 17

5.4.1 The Technology Acceptance Model ... 18

5.4.2 Theory of Planned Behavior ... 19

6. Discussion ... 20

6.1 Conclusion ... 20

6.2 Theoretical Implications ... 20

6.3 Comparing the Results with Mathieson’s Three Criteria ... 22

6.4 Limitations and Directions for Further Research ... 23

List of Appendices ... 25

Appendix ... 26

Abstract

This study compared the theory of planned behavior and the technology acceptance model in the field of mobile Internet acceptance. The main goal was to examine which model performed better in an unbiased comparison study. The results indicate that both models provide a good explanation to behavioral intention to use mobile Internet although TPB stands out based on variance explained and fit to the data. The greater explanatory and predictive power of the TPB model can be attributed to the ability of that model to incorporate social influences. Thus, although TAM is slightly more easier to use, TPB performs better in the context of mobile Internet. Moreover, TPB provides a more complete understanding of behavioral intention because it incorporates more specific information.

Acknowledgements

I would like to thank my supervisors of the University of Groningen who supported me with this thesis. Especially, I would like to thank my first supervisor dr. M.C. Leliveld for her precious time, constructive feedback as well as her kind support. In addition, I would like to thank dr. ir. M.J. Gijsenbergfor his suggestions and for assessing my thesis as a second supervisor.

Hylke Hoekstra

List of Abbreviations

α: Significance Level

χ2: Chi-square Distribution

2G: Second Generation Mobile Telecommunications

3G: Third Generation Mobile Telecommunications

AGFI: Adjusted goodness-of-fit Index

AT: Attitude

AVE: Average Variance Explained

BB: Behavioral Beliefs

BI: Behavioral Intention

CB: Control Beliefs

CFA: Confirmatory Factor Analysis

CFI: Comparative Fit Index

df: Degrees of Freedom

EoU: Ease of Use

EV: External Variables

GFI: Goodness-of-fit Index

IT: Information Technology

ITU: International Telecom Union

NB: Normative Beliefs

NFI: Normed Fit Index

PBC: Perceived Behavioral Control

PU: Perceived Usefulness

R2: Coefficient of Determination

RMSEA: Root Mean Square Error of Approximation

SN: Subjective Norms

TAM: Technology Acceptance Model

TPB: Theory of Planned Behavior

List of Figures

Figure 1: Theory of Reasoned Action (Fishbein and Ajzen 1975) ... 5

Figure 2: Theory of Planned Behavior (Ajzen, 1991) ... 6

Figure 3: Technogoly Acceptance Model (Davis, 1989) ... 8

Figure 4: Standardized Solution of the TAM Model ... 18

List of Tables

Table 1: TPB Sample Demographics ... 13

Table 2: TAM Sample Demographics ... 14

Table 3: Summary of Measurement Scales for TAM ... 14

Table 4: Summary of Measurement Scales for TPB ... 15

Table 5: Correlation Matrix Between Constructs TPB ... 16

Table 6: Correlation Matrix Between Constructs TAM ... 16

Table 7: Fit Indices and Explanatory Power of Each of the Models ... 17

Table 8: Contrasting Results with Other Studies ... 21

1.

Introduction

Over the last decade, there have been dramatic advances in technologies of mobile devices and networks. As a result, mobile technologies have the potential to create new markets, change the competitive landscape of business, create new opportunities, and change existing community and market structures (Stewart and Pavlou, 2002). In addition, in recent years we have witnessed a global increase in the use of third generation (3G) mobile networks. The number of people that have access to mobile networks is increasing at an exponential rate. In 2008 over 60% of the world population had access to a wireless connection (International Telecom Union; ITU). In 2010, access to mobile networks was available to 90% of the world population and 80% of the population living in rural areas. People are moving rapidly from 2G to 3G platforms, in both developed and developing countries. In 2010, 143 countries were offering 3G services commercially, compared to 95 in 20071. ITU forecasts mobile penetration in emerging markets will grow to 96% in 2013. Most importantly, whereas most Internet-users in the old days used stationary Internet, investments and developments in mobile infrastructure made it possible to connect to the Internet virtually everywhere.

This last change is important because the manner, in which we communicate on an everyday level, has changed in quite a major way. Much of our communication today takes place through mobile technology. Devices and systems based on mobile technologies have become commonplace in our everyday lives, increasing the accessibility, frequency and speed of communication (Balasubramanian et al., 2002). The role of mobile Internet in mobile technologies has become more important over time. Mobile Internet is perhaps one of the few technologies that comes close to emulating the success of the fixed Internet. Backed by the entire telecommunication industry, coupled with the fact that it combines two of the hottest innovations in recent times (mobile phone and the Internet), mobile Internet is poised to succeed the fixed Internet as the next big thing (Jiang, 2009). Ten years ago one would not even consider the possibility of having mobile Internet while today it is considered by most consumers as an indispensable service. Hence, mobile Internet is on the rise. As a growing proportion of the hundreds of millions of mobile handsets sold globally each year are mobile Internet capable, the importance of the mobile Internet is increasing (Vatanparast and Qadim, 2009). The increasing pace at which consumers are shifting from stationary Internet to mobile Internet has led to a struggle for mobile network providers to keep up with this pace. Traditional sources of income (e.g. mobile telephone conversations and text messaging) are declining (Dutch mobile market revenues drop 4.5% in Q2 20112). The spread of the mobile Internet presents a

1 _{ICT Fact and Figures- The world in 2010, International Telecom Union.} 2

solution to falling average revenue per user (by the addition of the mobile Internet service component into current supply of services). Furthermore, mobile Internet presents many opportunities to operators, equipment manufacturers, content and service providers, new entrants from the Internet world and other mobile industry players (Vatanparast and Qadim, 2009). In sum, the introduction of mobile Internet has revolutionized the entire framework of today’s communication technology. Both consumers and mobile network providers acknowledge mobile Internet to be a vital service component in today’s mobile communication offering.

By having stressed the impact of mobile Internet, the topical and essential question that arises is what makes mobile Internet different from stationary Internet. With mobile Internet basically offering the same service as stationary Internet, one could at first sight argue that there is no difference. Nevertheless, a substantial number of papers hold a different view, namely that there are significant differences between mobile Internet and stationary Internet (e.g. Chae and Kim, 2003; Vatanparast and Qadim, 2009 and; Chae, Kim, Kim and Ryu, 2002). Note that, there is little knowledge about how consumers react to the mobile Internet (Vatanparast and Qadim, 2009), which makes mobile Internet consumer behavior an interesting research topic. To develop a deeper understanding of the relation between consumers’ beliefs and mobile Internet adoption, the next section discusses in more detail the context of mobile Internet.

2.

The Mobile Internet Context

In general, mobile Internet can be defined as the convergence of mobile communications and the Internet, providing the Internet through wireless connections (International Telecommunication Union Internet Reports, 2002). Hong and Tam (2006) defined mobile Internet as a collection of mobile data services accessed only through a mobile communication network. Mobile Internet thus, is a form of wireless Internet which can be obtained through mobile devices not attached to a specific location. Despite the convergence of mobile communications and the Internet, there are significant differences between mobile Internet and stationary Internet (Chae and Kim, 2003). Knowledge about consumer Internet behavior cannot instantly be transferred to mobile Internet consumer behavior without considering the differences in characteristics of the service.

2.1 Differences Between Mobile Internet and Stationary Internet

are usually more personal and individual than stationary Internet devices. It is not uncommon for people to share their desktop computers, whereas it is very rare for them to share mobile Internet phones. Therefore, the mobile device always carries its user identity. Mobile devices are usually setup in such a way that users have direct access to their own personalized content. Advanced mobile technologies, such as 3G services, allow people to build and maintain their social relationships trough co-creation and joint usage of content (Ghose and Han, 2011). For example, social media offer users the possibility to directly interact with friends and family through mobile devices, usually without authentication. With stationary Internet, users first need to authenticate to verify the user’s identity. Another example is Internet banking, a service that most banks nowadays also offer through mobile devices. The verification process with stationary Internet is more extensive than with mobile Internet because mobile devices are already setup for the user’s own identity. Consequently, the services offered on mobile Internet devices contain more personalized aspects. Mobile Internet services are more intended for personal and individual needs and expectations (Vatanparast and Qadim, 2009).

Secondly, from an environmental perspective, mobile Internet provides direct access to the Internet anywhere and anytime (Lamming, Eldrigde, Flynn, Jones, Pendlebury and Satchel, 2000). A mobile Internet system is portable and always available. By contrast, stationary Internet systems are usually not movable and require long pre-processes, such as booting up, which usually take more than a few minutes (Chea and Kim, 2003). And finally, from a system’s perspective, most mobile Internet systems have lower resources compared to those provided by stationary Internet (Chae, Kim, Kim and Ryu, 2002). Mobile Internet systems and in particular mobile Internet phones generally have a smaller screen, less processing power and have a battery that is running out at some point. In contrast, stationary Internet offers more resources. There usually is a constant power supply, processing capabilities are higher and screens are more convenient in usage.

2.2 Past Research on Mobile Internet

decades, will be explained in the next chapter. Vatanparast and Qadim (2009) conclude that perceived usefulness and ease-of-use are significant predictors of a consumer’s intention to use mobile Internet and of consumers’ actual behavior. The adoption behavior of early adopters of mobile commerce services is researched by Pedersen (2005). An extended and modified model was developed to explain the early adopters’ intention to use mobile commerce services. Lu, Yu, Liu and Yao (2003) in their paper seek to explain the factors influencing user acceptance of mobile Internet. The authors found that gender had a moderating effect in the technology acceptance process, females have the tendency to have more concerns about others’ opinions and interaction with others, females may form their attitude toward mobile technology with more reliance on social influence than males do.

Given this research on the differences between mobile Internet and stationary Internet, together with research on technology acceptance, the objective of this thesis is to provide better understanding of mobile Internet consumer behavior. I argue that, the most important difference between mobile- and stationary Internet is the presence of a social component in mobile Internet. Mobile Internet is more intended for personal usage than stationary Internet. The studies undertaken in consumers’ adoption of mobile Internet are briefly discussed in this section. To perform research on mobile Internet behavior, a framework is needed that is capable of explaining the adoption of mobile Internet. In the next section, an overview is presented of leading theories in the field of new technology adoption behavior. These frameworks will be studied in the field of mobile Internet usage.

3.

Multi-Attribute Models in the Context of Technology Acceptance

concerned with the determinants of consciously intended behaviors. TRA (illustrated in Figure 1) has received considerable attention from within the field of consumer behavior. TRA is clear in its applicability to volitional behaviors, and a vast body of empirical findings supports this model (e.g., Ajzen and Fishbein, 1980). The model provides a relatively simple basis for identifying where and how target consumers’ behavioral change attempts (Sheppard, Hartwick and Warshaw, 1988). TRA assumes that individuals are considering the consequences of their actions prior to deciding whether to perform a given behavior. Individuals thus, act rationally according to TRA. Actual behavior of individuals is directly being influenced by behavioral intention.

TRA is a general model, and as such, it does not specify the beliefs that are operative for a particular behavior. With TRA, one must first identify the beliefs that are salient for participants concerning the behavior under investigation. TRA is very general, “designed to explain virtually any human behavior (Ajzen and Fishbein, 1980). TRA is criticized for neglecting the importance of social factors that in real life could be a determinant for individual behavior (Grandon and Mykytyn, 2004; Werner 2004). Social factors mean all the influences of the environment surrounding the individual (such as norms) which may influence the individual behavior (Ajzen 1991). To overcome this weakness in TRA, TPB was introduced by Ajzen (1991). In TPB an additional factor in determining individual behavior is added, which is Perceived Behavioral Control. The social factors are likely to play a role, especially in the field of mobile Internet behavior. I therefore, in this paper, further concentrate on using TPB instead of TRA.

Figure 1

Figure 2

Theory of Planned Behavior (Ajzen, 1991)

3.1 Theory of Planned Behavior

The theory of planned behavior (Ajzen, 1991) is an extension of TRA. TRA does not incorporate the function in dealing with behaviors in which individuals have incomplete volitional control. This is considered to be a missing element in TRA. According to TPB there is another factor that influences behavioral intention apart from attitude and subjective norms, that is perceived behavioral control (PBC). PBC directly influences behavioral intention and actual behavior (Figure 2). The theory of planned behavior assumes three independent determinants of behavioral intention: (a) attitude toward behavior - behavioral beliefs about the likely outcomes of the behavior and the evaluations of these outcomes; (b) subjective norm - normative beliefs about the normative expectations of others and the motivation to comply with these expectations; and (c) perceived behavioral control - control beliefs about the resources and opportunities possessed (or not possessed) by the individual and also the anticipated obstacles or impediments toward performing the target behavior (Ajzen, 1991).

question”. For example, when close friends encourage you to use social media via mobile Internet devices. However, you might not be inclined to use mobile Internet. The beliefs of your friends, weighted by the importance you attribute to each of their opinions, will influence your behavioral intention use mobile Internet, which will lead to your behavior to use mobile Internet or not to use mobile Internet. Subjective norm can be obtained by multiplying normative beliefs and motivation to comply. Perceived behavioral control refers to people's perceptions of their ability to perform a given behavior. It is assumed that perceived behavioral control is determined by the total set of accessible control beliefs, i.e., beliefs about the presence of factors that may facilitate or impede performance of the behavior.

The stability of intention is influenced by many extraneous factors. The observed relation between intention and behavior depends on two factors: First, the measure of intention has to correspond to the behavioral criterion in action, target, context and time; second, a measure of intention will predict behavior only if the intention does not change before the behavior is observed. As attitude, subjective norms and perceived behavioral control increase, the greater behavioral intention will be to perform the given behavior.

3.2 Technology Acceptance Model

In addition to TPB, another theory has come forward that is successful in explaining and predicting technology acceptance behavior; the technology acceptance model (TAM; Davis, 1989). TAM has emerged from TRA as a powerful model to specifically represent the antecedents of technology use. TAM (Davis, 1989) is one of the most influential extensions of TRA. TAM (Figure 3) replaces many of TRA’s attitude measures with two technology acceptance measures - perceived ease of use and perceived usefulness. TAM adapts the framework of TRA and states that an individual’s acceptance of a technology is determined by his or her voluntary intention to use that technology. Intention is influenced by attitude, which in turn is being influenced by perceived usefulness and perceived ease of use.

TPB and TAM it is important to pay attention to measurement accuracy due to the fact that, the strength of the belief-attitude-intention-behavior relation is determined by the measurement accuracy (Ajzen & Fishbein, 1980).

3.3 Comparison of Two Theories

There are a number of studies in literature that have compared the theoretical models used in technology acceptance research (e.g. Chau and Hu, 2001; Davis et al., 1989; Gentry & Calantone, 2002; Mathieson, 1991; Plouffe, Hulland and Vandenbosch, 2001; Riemenschneider and Hardgrave, 2000; Taylor and Todd, 1995). Literature shows that the two models in general are perfectly capable of predicting and explaining the adoption of new technology use although there are differences in the application of research. In a comparison study between TAM and TPB to study the adoption of an information system, Mathieson (1991) found that although TAM was a slightly better predictor of intention, TPB showed better explanatory power because of its incorporating specific beliefs, rather than generic beliefs. Taylor and Todd (1995) concluded that all three models (TRA, TPB and TAM) performed well in terms of fit and were roughly equivalent in terms of their ability to explain behavior. Note that, the majority of papers that investigated the acceptance of new technology during the last decade used TAM rather than TPB as their framework. TAM has been the dominant framework for explaining the acceptance and use of IT for nearly 20 years (Keil, Beranek and Konsynski, 1995). A major reason to choose for TAM is because TAM is believed to be most robust, parsimonious and influential in explaining information technology/ information system technology adoption behavior (Venkatesh and Davis, 2000). This study compares TPB and TAM in the context of

Figure 3

mobile Internet. The uniqueness of the current study is that it attempts to seek answers to the question whether existing behavioral models will work in an environment (mobile Internet) that they were not developed to describe.

The most important difference between TPB and TAM is the incorporation of social variables. TAM does not explicitly include any social variables. Davis et al. (1989) state that social norms are not independent of outcomes. For example, an individual might perceive pressure from his or her friends to use mobile Internet. Davis et al. (1989) argues that this effect will then already have been taken into account to some extent in the evaluation of outcomes. The social variables may, however, still capture unique variance in intention that is not already explained by other variables in the model. There could be social effects that are not directly linked to job-related outcomes such as usefulness (Mathieson, 1991). It might well be that subjective norms are an important driver for consumers to choose mobile Internet. These social variables are more likely to be captured by TPB than by TAM.

Another difference between TAM and TPB is their treatment of behavioral control, referring to the skills, opportunities and resources needed to use the system (Mathieson, 1991). Ajzen (1991) defined internal control factors as characteristics of the individual and external control factors as factors that are specific to each situation (e.g. time, opportunity and cooperation of others). Perceived-ease-of-use describes the internal control factors within TAM model but TAM does not incorporate the external control factors explicitly. Therefore, TAM is less likely to identify idiosyncratic barriers because of its robustness; it is designed to work in many situations. In contrast, TRA and TPB are more likely to capture the situation specific factors as these models first identify the important control variables for each situation.

3.4 Hypothesis

acceptance research over the last decades, I expect that, in the context of mobile Internet, TPB will outperform TAM. It is my understanding that the social variables are of crucial importance in explaining the adoption and use of mobile Internet. Therefore, I hypothesize that TPB is superior to TAM in explaining mobile Internet acceptance:

H1: TPB will explain more of the variance in behavioral intention towards the adoption and usage of mobile Internet than TAM.

3.5 Overview of Theoretical Framework

So far I have illustrated the advances in mobile technology and the increasing use of mobile Internet by consumers. There are considerable differences between stationary Internet and mobile Internet. The most important difference is the social aspect in mobile Internet. Both TPB and TAM have been discussed in the light of new technology acceptance. Both are good models to predict the acceptance of mobile Internet, however TAM does not incorporate social variables, while on the contrary TPB does. There have been a few studies that compare TPB and TAM, but there has been no study that compares TPB with TAM in the context of mobile Internet. I have argued that TPB is superior to TAM in explaining mobile Internet acceptance. This study compares TPB with TAM in explaining the adoption of mobile Internet behavior. In the next section, methodology will be discussed.

4.

Methodology

There is varying degree of generality within TPB and TAM (Mathieson, 1991). TPB methodology incorporates identifying salient beliefs specific to each situation while for TAM methodology it is not essential. TAM assumes that beliefs about usefulness and ease-of-use are always the primarily determinants of use decisions. TPB uses beliefs that are specific to each situation, the model does not assume that beliefs apply in one context also apply in other contexts (Mathieson, 1991). In addition, TPB is more difficult to apply across diverse user contexts than TAM. TAM constructs are measured in the same way in every situation. Therefore TPB requires a pilot study to identify relevant outcomes, referent groups and control variables in every context in which it is used.

4.1 Ensuring a Fair Comparison

comparison possesses; (a) procedural equivalence and; (b) distributional equivalence (Cooper and Richardson, 1986). The first condition for procedural equivalence is that the boundary conditions of both theories should be observed. TAM is a model that is especially designed to deal with technology acceptance, whereas TPB is more of a general model that has proven its explanatory power in a large number of domains. From this perspective TAM is more specific than TPB because TAM is already designed for technology acceptance research. Nevertheless, when both TPB and TAM are applied to technology acceptance research (which is the case in the current study) TPB is more specific than TAM. In sum, when applied to technology acceptance research the most important difference in boundary conditions is that TPB is more specific than TAM.

Furthermore, equal attention should be given to measurement. The same measures were used for all models for the constructs attitude, behavioral intention and actual behavior. For both TPB and TAM the same measurement procedure was followed as described by the authors (TPB: Fishbein and Ajzen, 1975; TAM: Davis, 1989). A limitation to the current study is that the relation between behavioral intention and actual behavior is not objectively measured. Literature shows that actual behavior should be measured objectively and unobtrusively, without signaling in any way its connection to the prior intention measurement phase (Davies, Foxall and Pallister, 2002). Since data on actual mobile Internet behavior is not at my disposal, the relation between behavioral intention and actual behavior can not be tested. One way of obtaining data on behavior would be to ask participants to estimate their own mobile Internet behavior. However, these data should then obviously be regarded as subjective measures. Research that relied on subjective measures for both independent, such as perceived usefulness, and dependent variables, such as system usage, may not be uncovering true, significant effects, but mere artifacts (Straub, Limayem, and Karahanna, 1995). Therefore, the relation between behavioral intention and actual behavior in both TPB and TAM will not be examined. Note that the relation of behavioral intention and actual behavior is the same for both TPB and TAM. The goal of this study is to compare both models in the context of mobile Internet. Therefore, it is less important to examine relationships that are the same across both models.

4.2 Operationalization of Constructs

discussed in Ajzen and Fishbein (1980). A group of 14 participants were asked to respond to the open-ended questions shown in Appendix 1. To make the pilot study specific enough (context) participants were confronted with a specific problem; viewing weather forecast either by stationary Internet or by mobile Internet. The participants then had to answer which salient beliefs came to mind. The participants that were in the pilot study did not participate in the main study.

A content analysis of the responses was then required to create a list of modal salient outcomes, referents and control factors. These lists are used to construct items to be included in the final questionnaire. For the identification of salient beliefs, the same procedure as in Mathieson (1991) was used; for outcome- and control-related issues, only beliefs mentioned by at least 50% of the subjects were retained. The relevant outcome beliefs were (1) accessibility of mobile Internet and (2) size of screen mobile Internet. These are similar to beliefs elicited by Davis et al. (1989). The 50% criterion was not reached for any control beliefs, so the two most frequently mentioned control beliefs were included. The control beliefs were: (1) ease of accessibility mobile Internet (mentioned by 5 subjects and (2) connection issues (mentioned by 4 subjects). The 50% criterion was reached for one referent group, which were friends.

5.

Results

This section presents the results of the study. First, the demographics and mean statistics of the research are briefly discussed. Second, convergent validity and discriminant validity are checked. Finally, conducting a confirmatory factor analysis, the degree to which TAM and TPB explain behavioral intention is discussed.

5.1 Participants

The questionnaire was distributed among a network of friends, students, family, colleagues and other acquaintances. In the distribution process I made sure that half of the participants received a TPB questionnaire and the other half the TAM questionnaire in a random order. In this way, no selection bias occurred. A total number of 192 participants responded to either the TPB questionnaire or TAM questionnaire. For my main statistical purpose, which is structural equation modeling, the sample size is large enough. Similar TPB-TAM comparison studies show sample sizes of approximately 150 respondents (e.g. Gumussoy and Calisir, 2011). Of those 192 participants, there were 99 participants in the TAM questionnaire and 93 participants in the TPB questionnaire. The 192 participants consisted of 114 men (59%) and 78 women (41%). The average age among participants was 26 years. The TPB sample demographics (Table 1) and the TAM sample demographics (Table 2) were tested to show if significant differences existed among demographic variables. The results of the post-hoc test revealed no significant differences, which makes an empirical comparison justified.

The age variable has been recoded into ordinal scale to show an age breakdown of the participants. Note however that further analysis was based on the original interval age variable. Of all the participants, 88% had more than 5 years of experience in using stationary Internet. For mobile

Table 1

TPB Sample Demographics

Gender Age Experience

Stationary Internet Mobile Internet

Male 63% <20 14% None 0% None 17%

Table 3

Summary of Measurement Scales for TAM

Measure #Items Mean Standard Deviation Composite Reliability Average Variance Explained (AVE) External Variables 2 3.85 0.65 0.045 0.529 Perceived Usefulness 4 4.17 1.56 0.84 0.675 Ease of Use 4 5.49 1.25 0.903 0.785 Attitude 4 5.09 1.34 0.895 0.748

Internet, the average experience is somewhat less, which is of course quite self-evident considering the fact that mobile Internet for consumers is only widely available since a few years. Most of the participants had a few years of experience in mobile Internet.

Table 2

TAM Sample Demographics

Gender Age Experience

Stationary Internet Mobile Internet

Male 56% <20 5% None 1% None 13%

Female 44% 20-29 79% < 1 year 0% < 1 year 28% 30-39 9% 1-2 years 2% 1-2 years 27% 40-39 1% 2-5 years 9% 2-5 years 27% >50 6% >5 years 88% >5 years 5%

5.2 Convergent Validity

The belief items of the TPB questionnaire were combined with the evaluative component using the expectancy-value approach (Ajzen, 1985, 1991). For example, individuals hold normative beliefs about the normative expectations (belief items) which then in turn are multiplied by the motivation to comply to these expectations (evaluative component) of that individual. The scale for each of the composite variables was developed by averaging responses to the individual items. The composite scales were created to accommodate the estimation technique employed in the analysis. An overview of the scale characteristics is presented in Table 3 and 4.

Table 4

Summary of Measurement Scales for TPB

Measure #Items Mean Standard

Deviation Composite Reliability Average Variance Explained (AVE) Behavioral Beliefs 2 5.76 0.92 0.174 0.621 Attitude 4 5.09 1.37 0.905 0.772 Normative Beliefs 2 4.22 1.74 0.765 0.551 Subjective Norm 3 4.17 1.90 0.941 0.805 Control Beliefs 2 5.31 1.36 0.808 0.842

Perceived Behavioral Control 3 5.68 1.26 0.841 0.893

Behavioral Intention 3 5.39 1.82 0.933 0.886

reliabilities above 0.75. The latent construct external variables consists of the experience in mobile Internet and experience in stationary Internet. Apparently experience in stationary Internet does not produce similar scores as experience in mobile Internet. This implicates that the latent construct external variables would be indicative of bad internal consistency. However, recall that mobile Internet fundamentally differs from stationary Internet (chapter 2). In their evaluation therefore it is most likely that individuals consider experience not to be a comprehensive construct, but rather form thoughts about both mobile Internet and stationary Internet separately. In fact, the results strengthen the case that there indeed is a difference between stationary Internet and mobile Internet.

As for the behavioral beliefs construct, which deals with importance of accessibility of Internet, low internal consistency was measured too. First, respondents were asked to state whether accessibility is higher at mobile Internet and second whether in their view it is important to have Internet always accessible. Bring to mind that the behavioral beliefs construct was the result of an extensive pilot-test (Paragraph 4.2; Operationalization of Constructs) based on well-constructed methodology (Ajzen, 1991). Moreover, as Ryan and Bock (1990) indicate it is difficult to set one task that is appealing to the whole target group. No single task with a single group of users can fully represent this diversity. This clarifies why those two constructs were not found to be internal consistent. Again, both the TPB and the TAM are theoretically sound models which have shown more than respectable results in explaining IT adoption over the last 20 years. Omitting those constructs would feel like a missed opportunity in explaining the variance. Therefore it makes logical sense to maintain the current external variables and behavioral beliefs constructs.

that the variance extracted value should exceed 0.50 for a construct. If the average variance extracted (AVE) is less than 0.50, then the variance due to measurement error is greater than the variance due to the construct. If the values were less than 0.50, convergent validity would be questionable. In the current study, all AVE measures fulfill the suggested levels with variance extracted value ranges from 0.529 to 0.893.

5.3 Discriminant Validity

In addition to convergent validity, also discriminant validity was checked. Discriminant validity was confirmed by examining correlations among the constructs. As a rule of thumb, I used a 0.85 correlation as a maximum because a higher value indicates poor discriminant validity in structural equitation modeling. The correlation matrixes for both TPB (Table 5) and TAM (Table 6) indicate that none of the correlations present values above 0.85. The results suggest an adequate discriminant validity of the measurement.

Table 5

Correlation Matrix Between Constructs TPB

Constructs BI AT SN PBC BB NB CB

Behavioral Intention (BI) 1.00

Attitude (AT) 0.79 1.00

Subjective Norms (SN) 0.48 0.44 1.00

Perceived Behavioral Control (PBC) 0.52 0.54 0.28 1.00 Behavioral Beliefs (BB) 0.63 0.64 0.21 0.51 1.00 Normative Beliefs (NB) 0.45 0.41 0.69 0.25 0.26 1.00

Control Beliefs (CB) 0.67 0.69 0.35 0.44 0.59 0.46 1.00

Table 6

Correlation Matrix Between Constructs TAM

Constructs PU EoU AT BI EV

Perceived Usefulness (PU) 1.00

Ease of Use (EoU) 0.52 1.00

Attitude (AT) 0.67 0.71 1.00

Behavioral Intention (BI) 0.34 0.61 0.58 1.00

5.4 Explaining Behavioral Intention

To access construct validity a confirmatory factor analysis (CFA) was performed involving all of the measures the TPB model and the TAM model. The CFA was conducted with LISREL 8.80. To access the model, multiple fit indices are presented. The literature suggests that, for a good model fit, chi-square/ degrees off freedom (χ2 / df) should be less than 3, adjusted goodness-of-fit index (AGFI) should be larger than 0.8, goodness-of-fit index (GFI), normed fit index (NFI), and comparative fit index (CFI) should all be greater than 0.9, and root mean square error of approximation (RMSEA) should be less than 0.10 (Henry and Stone, 1994). For each model, overall fit, predictive power and the significance of paths were considered. R2 for each depended construct was examined to access explanatory power, and the significance of individual paths was assessed. The fit statistics are shown in table 7. Path coefficients for each model and their significance are shown in figure 4 and 5. In the next paragraph, both the TAM model and the TPB model will be individually evaluated based on above mentioned criteria.

Table 7

Fit indices and Explanatory Power of Each of the Models

TAM TPB Recommended Value (Henry and Stone, 1994)

5.4.1 The Technology Acceptance Model

Overall, the fit statistics indicate that TAM provides a reasonably well fit to the data (χ2 = 230.55, p < 0.0001; df = 114; AFGI = 0.68; GFI = 0.76; NFI = 0.91; CFI = 0.95 and RMSEA = 0.11, p = 0.00). The model accounts for 43% of the variance in behavioral intention, 68% of the variance in attitude, 27% in perceived usefulness and 34% of the variance in ease of use (table 7).

As shown in Figure 4, most path coefficients were congruent to the proposed Technology Acceptance Model. The path from external variables to perceived usefulness and to ease of use were both significant. The path from ease of use to attitude was also found significant, as was the path from attitude to behavioral intention. Note that, the path from perceived usefulness to attitude was not significant. Behavioral intention is well explained by attitude in the TAM model (R2 = 0.68). However, there are differences in the constructs of Perceived Usefulness (PU) and Ease of Use (EoU). Although the relationship between External variables and EoU is significant (α = 0.05) the variance explained in EoU is low (R2 _{= 0.34). Also note that, the relationship between PU and Attitude is} negative. However, the relationship is not found significant and therefore irrelevant. In conclusion, the TAM model reasonably well predicts the behavioral intention of people to use mobile Internet. However, the PU construct appears to be less strong in the case of mobile Internet than previous literature presumes.

Figure 4

5.4.2 Theory of Planned Behavior

The fit statistics (Table 6) show that TPB provides a good fit to the data (χ2 = 249.83, p < 0.0001; df = 143; AFGI = 0.7; GFI = 0.78; NFI = 0.92; CFI = 0.96 and RMSEA = 0.087, p = 0.0022). Note that, all fit indices of the TPB model are closer to the recommended value (Henry and Stone, 1994) than TAM. The better RMSEA value for TPB is noteworthy, which indicates that even when the increased complexity of the TPB model is taken into consideration, the fit of the TPB model is at least equivalent to TAM. Based on the criteria of Henry and Stone (1994) the fit indices of the TPB model show a good fit to the data.

The predictive power of the TPB model was roughly equivalent to TAM. The model accounts for 76% of the variance in behavioral intention, 95% of the variance in attitude. The addition of behavioral beliefs, normative beliefs and control beliefs helped to successfully predict attitude, subjective norms and perceived behavioral control (R2 A= 0.95, R2 SN = 0.66 and R2 PBC= 0.4). The addition of these variables resulted in an increase in predictive power of behavioral intention for the TPB model (R2 BI = 0.76) versus the TAM model (R2 BI = 0.68). However, subjective norms and perceived behavioral control have a somewhat smaller effect on behavioral intention than attitude. Both relationships were found to be not significant (α = 0.05). However, if the alpha level is raised to α = 0.10 instead of the current 0.05 (which is generally accepted in the field of science, especially in replication studies such as the current study) there is a significant relation between subjective norm and behavioral intention. Although it should be noted that subjective norm had only a minor effect on behavioral intention (β = 0.13) indicating that while subjective norms do play an important role in the current TPB research model (R2 = 0.66), the effect smaller is than I hypothesized.

Figure 5

Turning to comparison of TPB and TAM, it appears that TPB (R2 = 0.95) explains attitude much better than TAM (R2 = 0.68). Overall, the TPB model in the field of mobile Internet acceptance performed better than TAM in terms of fit to the data. Moreover, TPB’s behavioral intention showed a better coefficient of determination and the attitude construct in TPB is stronger than with TAM.

6.

Discussion

The goal of the current study was to compare the Theory of Planned Behavior to the Technology Acceptance Model in terms of their contribution to the understanding of mobile Internet adoption behavior. This final chapter will first sum up the most important conclusions and implications. Then, the theoretical implications are discussed. Finally, the potential limitations and directions for further research are discussed.

6.1 Conclusion

Overall, this study compared the theory of planned behavior and the technology acceptance model in the field of mobile Internet acceptance. The main goal was to examine which model performed better in an unbiased comparison study. The results indicate that both models provide a good explanation to behavioral intention to use mobile Internet although TPB stands out based on variance explained and fit to the data. The greater explanatory and predictive power of the TPB model can be attributed to the ability of that model to incorporate social influences. Thus, although TAM is slightly more easier to use TPB performs better in the context of mobile Internet. Moreover, TPB provides a more complete understanding of behavioral intention because it incorporates more specific information.

6.2 Theoretical Implications

Table 8

Contrasting Results with Other Studies

R2 Behavioral Intention TPB R2 Behavioral Intention TAM

Mathieson (1991) 0.62 0.70

Taylor and Todd (1995) 0.57 0.52

Current study 0.76 0.68

TPB = 0.76). Note that TPB’s explanatory power of this study outperforms all other studies. In explaining the variation between those results one should of course account for differences in the

contexts which were studied. It might be that the target behavior (mobile Internet in this case) can be better explained by theoretical models than the conventional new technology in other studies.

Furthermore, it is interesting to note that both Davis et al. (1989) and Mathieson (1991) did not find a significant influence of subjective norm on behavioral intention. The subjective norm-behavioral intention link was also in this paper found to be not significant. Taylor and Todd (1995) on the other hand, do find such significant influence in this subjective norms-behavioral intention link. Recall that setting one task (target behavior) for all participants to perform is difficult. The task to perform for participants is a strong determinant for significance of the subjective norms-behavioral intention relationship. This finding is supported by Taylor and Todd (1995) who argue that the difference in their result may be due to differences in the nature of target behavior between studies.

Internet. Second, the effects of perceived usefulness could also be attributed to varying differences between countries. Studies of Lee et al. (2002) proved that perceived usefulness is a significant determinant of user’s satisfaction in Korea, Hong Kong and Taiwan, however they have also proven that the results are different through different cultural lenses. In addition, Vatanparast and Qadim (2009) found the relationship between perceived usefulness and behavioral intention to use to be only significant for a number of countries (Russia, Japan and China) whereas the perceived usefulness-behavioral intention link in other countries (United Kingdom and United States of America) was not significant. Hofstede’s (1980) dimensions of national culture are often used to help sort out differences between national cultures. The values for the cultural dimensions of the above mentioned countries reveal that The Netherlands are more likely to have similarities with the UK and the USA than with Russia, Japan and China (Table 9). Taking into that the vast majority of the sample population were Dutch participants it might not come as a complete surprise that perceived usefulness is not significant in the current study.

Table 9

Hofstede’s Cultural Dimensions

Country Power Distance Individualism versus Collectivism Masculinity versus Femininity Uncertainty Avoidance Long-term versus Short-term Orientation Russia 93 39 36 95 - Japan 54 46 95 92 80 China 80 20 66 40 118 Netherlands 38 80 14 53 44 United Kingdom 35 89 66 35 25 United States 40 91 62 46 29

6.3 Comparing the Results with Mathieson’s Three Criteria

attitude towards using mobile Internet much better than TAM. If, as a researcher, you are interested in explaining attitude on mobile Internet, this study provides compelling evidence to choose TPB instead of TAM. In sum, the TPB model outperformed TAM in almost every aspect. Not only did it show a better coefficient of determination of the behavioral intention construct but it also had a better fit with the data.

The second criterion was the value of information provided by the models (Mathieson, 1991). TAM is providing only basic information on usage intentions. The constructs of TAM are quite general and can be applied in broad domain of services. Recall that TPB needed a pilot study to develop the final questionnaire. TPB is therefore much more specific than TAM, in that it establishes its constructs based on qualitative information rather than generalized constructs. TPB delivers more specific information, measuring the system’s performance on various outcomes, and identifies factors that respondents feel might be barriers to system use. It also identifies groups whose opinions might be important to potential users (Mathieson, 1991). For example, in the current study it identified a control barrier in the ability to use mobile Internet, a construct which is not incorporated in the TAM. TPB in this case provides more specific information. Moreover, TAM would for example identify that a system was not easy to use, whereas TPB would identify possible sources of resistance. Hence, TPB provides more valuable information than TAM.

Third, how difficult are the models to apply? TAM is easier to apply than TPB. With TPB it is necessary to conduct a pilot study. TAM has as standard set of instruments, which saves time and effort. Another argument that is important in this respect is parsimony. Taylor and Todd (1995) argue that, if the variance explained and fit to the data is taken out of the equation, parsimony is an important aspect to consider in the selection of which model to choose. In trying to obtain the most complete understanding of a phenomena, a degree of parsimony may be sacrificed. In my case, both TAM and TPB are relatively parsimonious although the 5-variable TAM model is more parsimonious than the 7-variable TPB model.

6.4 Limitations and Directions for Further Research

Given the objective of the current study, a comparison study between two models in the field of mobile Internet adoption, there is no direct need to examine relationships that are the same across models. Second, there is overwhelming evidence in literature that clearly points out the relationship between behavioral intention and actual behavior. Davis et al. (1989) and Davis (1989) found that behavioral intention is linked to actual behavior. In addition, Ajzen and Madden (1986) reported a close link in this construct. Moreover, in an extensive meta-analysis, Sheppard et al. (1988) reported a correlation of 0.53 between behavioral intention and actual behavior. We can assume that the strength between the two constructs in both models is equal.

The second limitation concerns sampling issues. First, approximately 60% of the respondents in the current study are male. Research has shown that gender difference could cause discrepancies in the effects of attitude, perceived behavioral control and subjective norm on the relation with behavioral intention (Venkatesh and Morris, 2000; Armitage et al., 2002). Please note that the effects of gender on non-response bias were tested but no significant effects were found. Although a population that accounts for 60% of one gender should not be regarded as a serious divergence, the results in the current study might be little biased for not reflecting the population distribution of gender. Second, age discrepancies in this respect also deserve attention. The population of the sample is somewhat overrepresented by youthful respondents. 72% of the sample population was between 20 and 29 years old. Over 80% was under the age of 30. Since the main sample population consisted of young adults, one could argue that the results of this paper can only account for individuals that are under the age of 30. This imposes a limitation of generalizability to the population.

List of Appendices

Appendix

Appendix 1. Questions used in the Pilot study (in Dutch) Mobiel Internet

In deze korte vragenlijst wil ik je een aantal vragen stellen over het gebruik van mobiel Internet (via smartphone) ten opzichte van vast Internet (via PC of laptop). In deze vragenlijst zijn geen goede of foute antwoorden, ik ben alleen geïnteresseerd in jouw eigen mening. Uiteraard zullen de antwoorden anoniem worden behandeld.

Stel je nu de volgende situatie voor. Je staat op het punt om op vakantie te gaan. Bij het inpakken van je koffer twijfel je of je die lange broek nou wel of niet mee zult nemen. Je beseft je namelijk dat je niet precies weet wat voor weer het is op je vakantiebestemming. Enthousiast besluit je de weersvoorspelling op te zoeken via het Internet.

1. Wat zijn volgens jou de voordelen van mobiel Internet ten opzichte van vast Internet in het opzoeken van deze weersvoorspelling.

2. Wat zijn volgens jou de nadelen van mobiel Internet ten opzichte van vast Internet in het opzoeken van deze weersvoorspelling.

3. Als je het gebruik van mobiel Internet vergelijkt met vast Internet, heb je nog andere gedachten? Zo ja, welke?

In je keuze om mobiel Internet te gebruiken ten koste van vast Internet is het mogelijk dat er mensen of groepen mensen zijn die van invloed zijn op jouw keuze. Aan de hand van de volgende drie vragen probeer ik te achterhalen welke (groepen) mensen (bijvoorbeeld vrienden, collega’s, medestudenten, docenten, verkoper van telecomwinkel, familie, etc. etc.) van invloed zouden zijn op jouw keus voor vast dan wel mobiel Internet. Je hoeft dus geen namen te noemen van personen maar wel tot welke relatie zij met jou staan.

1. Welke (groepen) mensen zouden je keuze voor mobiel Internet goedkeuren? 2. Welke (groepen) mensen zouden je keuze voor mobiel Internet afkeuren?

3. Komen er nog andere (groepen) mensen bij je op, die mogelijk een oordeel hebben over je keuze voor mobiel Internet? Zo ja, welke?

De volgende twee vragen gaan over andere factoren en omstandigheden die van invloed kunnen zijn op je keuze om mobiel Internet te gebruiken.

1. Welke factoren of omstandigheden maken het je gemakkelijk of bevorderen je om mobiel

Internet te gebruiken?

References

Ajzen, I., and Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall.

Ajzen, I. (1985). From intentions to actions: a theory of planned behavior, in Kuhl, J. and Beckmann, J. (eds), action control: from cognition to behavior, 11-39, New York: Springer.

Ajzen, I.(1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211.

Ajzen, I. and Madden, T.J. (1986) Prediction of Goal-Directed Behavior: Attitudes, Intention and Perceived Behavioral Control. Journal of Experimental Social Psychology, 22, 453-474.

Armitage, C.J., Norman, P. and Conner, M. (2002). Can the theory of planned behavior mediate the effects of age, gender and multidimensional health locus of control? British Journal of Health Psychology, 7, 299-316.

Bhattacherjee, A. (2002). Acceptance of e-commerce services: the case of electronic brokerages. IEEE transactions on System, Man and Cybernetics-Part A: Systems and Humans, 20(4), 411-420. Balasubramanian, S., Peterson, R. A. and Järvenpää, S. L.(2002). Exploring the Implications of

m-Commerce for Markets and Marketing. Journal of the Academy of Marketing Science, 30(4), 348 - 361.

Chae, M. and Kim, J. (2003). What’s so different about the Mobile phone Internet?. Communications of the ACM, 46(12), 240-247.

Chae, M., Kim, J., Kim, H. and Ryu, H. (2002). Information quality for mobile Internet services. A theoretical model with empirical validation. Electronic Markets, 12, 38-46.

Chau, P., and Hu, P. (2001). Information technology acceptance by individual professionals: A model comparison approach. Decision Sciences, 32, 699–710.

Cooper, W. and Richardson, A. (1986). Unfair comparisons. Journal of Applied Psychology, 71, 179-184.

Davis, F.D. (1989). Perceived Usefulness, Perceived Ease of Use and User Acceptance of Information Technology. MIS Quarterly, 13, 319-340.

Davies, J., Foxall, G.R. and Pallister, J. (2002). Beyond the intention – behavior mythology: an integrated model of recycling. Marketing Theory, 2(1), 29-113.

Fishbein, M. and Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley.

Ghose, A. and Sang, P.H. (2011). An Empirical Analysis of User Content Generation and Usage Behavior on the Mobile Internet. Management Science, 57(9), 1671-1691.

Grandon, E. and Mykytyn, J. (2004). Theory-Based Instrumentation to Measure The Intention to Use Electronic Commerce in Small and Medium Sized Businesses. The Journal of Computer Information Systems, 44(3), 44-57.

Gumussoy, C.A. and Calisir, F. (2011). Acceptance of e-reverse auction use: An empirical comparison of models. Proceedings of the 2011 International Conference on Industrial Engineering and Operations Management. Kaula Lumpur, Malaysia, 22-24.

Henry, J.W. and Stone, R.W. (1994). A structural equation modeling of end-user satisfaction with a computer-based medical information system. Information Resources Management Journal, 7(3), 21-33.

Hofstede, G. (1980). Culture’s consequences: International differences in work related values. Newbury Park, CA: Sage.

Hong, S. J. and Tam, K.Y. (2006). Understanding the adoption of multipurpose information appliances: The case of mobile data services. Information Systems Research, 17(2), 162–179. Jiang, P. (2009). Consumer Adoption of Mobile Internet Services: An exploratory Study. Journal of

Promotion Management, 15, 418-454.

Keil, M., Beranek, P.M., Konsynski, B.R. (1995). Usefulness and ease of use: field study evidence regarding task considerations. Decision Support Systems. 13(1), 75-91.

Lamming, M., Eldridge, M., Flynn, M., Jones, C. and Pendlebury, D. (2000). Satchel: providing access to any document, any time, anywhere. ACM Transactions on Computer-Human Interaction. 7(3), 332-352.

Lee, Y., Kim, J., Lee, I., Kim, H. (2002). A Cross-cultural Study on the Value Structure of Mobile Internet Usage: Comparison between Korea and Japan. Journal of Electronic Commerce Research, 3(4), 227-239.

Lu, J., Yu, C.S., Liu, C. and Yao, J.E. (2003). Technology Acceptance Model for Wireless Internet. Internet Research, 13(3), 206 – 222.

Mathieson, K. (1991). Predicting user intentions: Comparing the technology acceptance model with the theory of planned behavior. Information Systems Research, 2, 173-191.

Nataraajan, R. and Warshaw, P. (1991). Are the models compatible for empirical comparison? An illustration with an intentions model, an expectations model and traditional conjoint analysis. Advances in Consumer Research, 19, 472-482.

Plouffe, C., Hulland, J., and Vandenbosch, M. (2001). Richness versus parsimony in modelling technology adoption decisions: Understanding merchant adoption of a smart-cart-based payment system. Information Systems Research, 12, 208–222.

Riemenschneider, C. and Hardgrave, B. (2000). Explaining software development tool use with technology acceptance model. Journal of Computer Information Systems, 41(4), 1–8.

Ryan, T. and Bock, D.B. (1990). Comparability of Ratings in MIS Survey Research. Presented at the Annual Meeting of the Decision Sciences Institute, San Diego, CA.

Sheppard, B. H., Hartwick, J. and Warshaw, P. R. (1988). The theory of reasoned action: A Meta analysis of past research with recommendations for modifications and future research. Journal of Consumer Research, 15(3), 325- 343.

Stewart, D. and Pavlou, P.A. (2002). From Consumer Response to Active Consumer: Measuring the Effectiveness of Interactive Media. Journal of the Academy of Marketing Science, 30(4), 376-396. Straub, D.W., Limayem, M. and Karahanna, E. (1995). Measuring System Usage: Implications for IS

Theory Testing. Management Science, 41(8), 1328-1342.

Taylor, S., and Todd, P. (1995). Understanding information technology usage: A test of competing models. Information Systems Research, 6, 144–176.

Vatanparast, R. and Qadim, H.Z. (2009). A Cross-Cultural Study on Mobile Internet Usage. International Journal of Mobile Marketing, 4(2), 14-27.

Venkatesh, V. and Davis, F. D. (1996). A model of the antecedents of perceived ease of use: development and test. Decision Sciences, 27(3), 451-481.

Venkatesh, V. and Davis, F.D. (2000). A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies. Management Science, 46(2), 186-204.

Venkatesh, V. and Morris, M.G. (2000). A longitudinal field investigation of gender difference in individual technology adoption decision-making process. Organizational Behavior and Human Decision Process, 83(1), 33-60.

Venkatesh, V., Morris, M.G. and Davis, F.D. (2003). User acceptance of information Technology: Toward a unified view. MIS Quarterly, 27(3), 425-478.

Werner, P. (2004). Reasoned Action and Planned Behavior, in Peterson, S.J. and Bredow, T.S., Middle range Theories: Application to Nursing Research, Lippincott Williams and Wilkins, 125-147 Zeithaml, V.A. (1988). Consumer perceptions of price, quality and value: a means-end model and

synthesis of evidence. Journal of Marketing, 52(3), 2-22.

Other publications