A study among non-users and users of no-checkout technologies:
A critical review of the TAM model & attitude and usage intentions in
the Netherlands toward no-checkout technologies
By
A study among non-users and users of no-checkout technologies:
A critical review of the TAM model & attitude and usage intentions in
the Netherlands toward no-checkout technologies
By
Daphne den Hartogh
Master Thesis MSc Marketing Management University of Groningen
Faculty of Economics and Business First supervisor: Dr. J. Berger Second supervisor: Dr. J.I.M. de Groot
Preface
This thesis is written to complete the MSc Marketing Management at the University of Groningen. Due to an interest of the author in the retail industry, this thesis is written within this theme. This research applies three well-established models to a new innovation in the retail industry, namely no-checkout technologies. The Technology Acceptance Model (TAM model) serves as a foundation and has been extended by using it simultaneously with the Theory of Planned Behavior (TPB) and the Expectancy Disconfirmation Theory (EDT). The extended TAM model has been applied to both non-users and users of no-checkout technologies. The main goal of this part is to identify the attitudes and usage intentions of non-users and users in the Netherlands toward no-checkout technologies.
I would like to thank those involved in my graduation period. Firstly, dr. J. Berger, who provided me with useful feedback during the process of writing this thesis. I also want to thank everyone who took time to fill in my survey. This resulted in valuable data and insights into this topic.
Table of Contents
1. Introduction ... 5
2. Literature Review... 7
2.1 A changing retail environment ... 7
2.2 Models ... 8
2.2.1 The Technology Acceptance Model ... 9
2.2.2 The Theory of Planned Behavior ... 10
2.2.3 The Expectancy Disconfirmation Theory ... 10
2.3 The extended TAM model ... 11
3. Theoretical Framework and Model Development ... 12
3.1 Conceptual models ... 12
3.2 Intention to use no-checkout technologies ... 13
3.3 Perceived usefulness ... 13
3.4 Perceived ease of use... 14
3.5 Subjective norms ... 14
3.6 Perceived behavioral control ... 15
3.7 Attitude toward no-checkout technologies ... 15
3.8 Usage experience ... 16
3.9 Disconfirmation... 16
3.10 Satisfaction ... 17
4. Research Design ... 18
4.1 Data collection ... 18
4.2 Population and sampling method ... 18
4.3 Operational definitions ... 19
4.4 Plan of analysis ... 20
5. Analysis and Results... 22
5.1 Samples ... 22
5.2 Analysis of non-users ... 23
5.2.1 Convergence check ... 23
5.2.2 Measurement model (outer model)... 23
5.2.3 Structural model (inner model) ... 25
5.2.5 Mediation analysis ... 27
5.3 Analysis of users ... 28
5.3.1 Convergence check ... 28
5.3.2 Measurement model (outer model)... 28
5.3.3 Structural model (inner model) ... 30
5.3.4 Model measurement... 31
5.3.5 Mediation ... 33
5.4 Further research into insignificance ... 33
5.5 Hypotheses ... 34
6. Discussion ... 36
7. Conclusions and Recommendations ... 40
7.1 Introduction... 40
7.2 Important findings ... 40
7.3 Implications... 41
7.4 Limitations of the study... 42
7.5 Suggestions for future research ... 42
References ... 44
Appendices ... 47
Appendix A: Survey questions... 47
Appendix B: SmartPLS output non-users ... 51
1. Introduction
The retail industry is rapidly changing due to technology. These technological changes can both benefit shoppers and retailers. One of the benefits shoppers can experience due to technology is obtaining faster service (Grewal, Roggeveen & Nordfalt, 2017). Self-service technologies are already used by shoppers for a couple of years (Meuter et al., 2000). Self-service technologies enable shoppers to provide services to themselves while they are shopping. A new type of technology is no-checkout technologies, or as it is sometimes called Just Walk Out Technologies (JOWT). This type of technology enables shoppers to shop without any human interaction during the shopping experience. Amazon was the first company who introduced this technology for brick-and-mortar stores. They created a shopping experience without human contact, due to complex technologies which track the items the shopper selects (Johnston, 2018). No-checkout technologies are recently introduced in the Netherlands by AH to go. Two stores in Amsterdam and one store in Zaandam are now in the trial phase of this technology (Albert Heijn, 2018). Is the Dutch market ready for no-checkout technologies and what drives shoppers to decide whether or not to use this technology?
The user acceptance of new technologies can be researched and determined by an often-used model, namely the Technology Acceptance Model (TAM model). The TAM model can be used to gain insights into and explain the adaption of a technology (Davis, 1989). The development of this model is based on the Theory of Planned Behavior (TPB). The TPB includes determinants of an individuals’ decision to engage in a particular behavior (Azjen, 1991). Therefore, the TAM model and the TPB are often used simultaneously. The TAM model is a strong and well-established model for predicting user acceptance, with more than 424 journal citations to the initial article that introduced the model as of 2000 (Venkatesh & Davis, 2000). However, there have been a lot of critics on this model since it is stated that the original model does not cover all factors which influence usage intention and usage behavior of technology. Previous research stated that the TAM model can be effectively combined with the Expectancy Disconfirmation Theory (EDT) to explain usage intentions over time (Venkatesh & Goyal, 2010). This research built upon and extended TAM research. The TAM model serves as a useful foundation to build further on researching user acceptance of no-checkout technologies (George, 2004).
critics. Based on this, it is decided which factors can contribute in improving this model. Based on this critical review, part two applies the extended TAM model to non-users and users of no-checkout technologies. Non-users are in a pre-adoption situation, since they have never used the technology before. On the other hand, users are in a post-adoption situation since they have used the technology at least one time before. The aim of part two is to identify the factors which drive shoppers’ usage intentions of no-checkout technologies in the Netherlands to determine whether non-users and users will use or continue to use this technology. Accordingly, the research question of this research is formulated as follows:
“Which variables can contribute to the Technology Acceptance Model in order to improve this model, and when this model is applied to no-checkout technologies to which insights does
this result about the drivers of non-users and users in the Netherlands of no-checkout technologies?”
This research provides scientific relevance by extending the original TAM model with variables of the TPB and EDT. The TAM model has proved to be suitable to identify the adoption of new technologies (Davis, 1989). Therefore, this model has been applied to the user acceptance of no-checkout technologies. Constructs of the TPB are used to identify whether shoppers’ confidence in using no-checkout technologies and their perceived social pressure influence their usage intention (Azjen, 1991). The constructs of the EDT extend the users model by providing insights into continuance usage intentions (Oliver, 1980). This research contributes in managerial relevance by applying the extended TAM model to no-checkout technologies. This have led to insights in the adoption of non-users and users in the Netherlands of no-checkout technologies.
2. Literature Review
This chapter reviews exiting literature and identifies important findings and gaps within this research theme.
2.1 A changing retail environment
New opportunities are created by the Internet of Things (IoT), which drives innovation by bringing objects, activities and consumers into the digital domain. This movement is especially disruptive in the retail industry. The way in which retailers are operating has dramatically changed in the last couple of years. A number of retailers already started with changing their strategy into a digital one. However, the implementation of the IoT is still in its early stage. Due to changing customer expectations and high industry rivalry, retailers are forced to adapt their current strategy to the digital environment in order to survive in this competitive market (Gregory, 2015).
Due to these developments in the digital environment, the in-store shopping experience of consumers is changing. Self-service technologies which enable shoppers to provide services to themselves while they are shopping are already common for a couple years. Self-service technologies are defined as “technological interfaces that enable customers to produce a service independent of direct service employee involvement” (Meuter et al., 2000, p.50). No-checkout technologies are a relatively new type of self-service technologies. This technology enables shoppers to shop without any human interaction in the store. It provides shoppers a contactless checkout by automatic scanning as the shopper walks out of the store. This automatic scanning is controlled by complex technologies in-store while simultaneously using the shoppers’ smartphone (Johnston, 2018).
thing the shopper needs to do is to download the Amazon Go app in order to scan their personal QR code which identifies him when entering the store (Polacco & Backes, 2018).
No-checkout technologies are recently introduced in the Netherlands by AH to go. Two stores in Amsterdam and one store in Zaandam are now offering the payment method “Tap To Go”. In comparison with the Amazon Go system, Dutch shoppers have to use the AH to go app or their personal Tap To Go card to scan the price tag on the shelves. The shopper can pick the desired product, scan the corresponding price tag and walk away with it immediately. After leaving the store the shopper will be charged within ten minutes through his bank account (Albert Heijn, 2018). Is the Dutch market ready for no-checkout technologies and what drives shoppers to decide whether or not to use this technology?
2.2 Models
Previous research tried to understand the determinants of technology usage. A very common used model in this type of researches is the Technology Acceptance Model, shortly written as the TAM model (Demoulin & Djelassi, 2016). The TAM model can be used to gain insights into and explain the adaption of a technology (Davis, 1989). The TAM model is based on the Theory of Planned Behavior (TPB). The TPB is a psychological theory which states that an individuals’ attitude toward a behavior is a predictor of the actual usage behavior (Azjen, 1991). Previous research showed that the TAM model, and thus the TPB, can be effectively combined with the Expectancy Disconfirmation Theory (EDT) to explain usage acceptance of new technologies (Venkatesh & Goyal, 2010). The research of Bhattacherjee (2001) integrated the TAM model and EDT to gain insights in usage intentions over a longer time. By integrating both models he gained insights in underlying continuance usage intentions. Continuance usage intentions are the intentions to continue to use a technology over time (Bhattacherjee, 2001). Therefore, this theory is relevant for the research among users of no-checkout technologies.
non-users and non-users of no-checkout technologies. Therefore, the conceptual model can be focused on the specific group and results of both groups can be compared. Third, this research applies the extended TAM model to a new type of technology. To date, this is the first research that uses an extended TAM model to predict usage intentions of no-checkout technologies in the Netherlands. This research tried to identify whether shoppers in the Netherlands will adopt no-checkout technologies in their shopping behavior. In order to do this, these three well established models are used. Each model will be individually introduced in the following sub paragraphs.
2.2.1 The Technology Acceptance Model
The first version of the Technology Acceptance Model has been developed in 1989 by Davis. The TAM model has been widely used to explain the adaption of information technology and information systems. This model states that technology acceptance is influenced by the strength of attitudes toward and the usage intention of that technology. The TAM model consists of two main constructs, namely perceived usefulness and perceived ease of use. Perceived usefulness and perceived ease of use are the basis for an individuals’ attitude toward using a particular technology or system (Davis, 1989). Since the publishing of the first version of the TAM model, many researches have tried to extend or improve the original model due to critics and experienced limitations. Below some of these researches are highlighted.
As well as perceived usefulness, perceived ease of use is also an important driver of the acceptance of technologies. Based on behavioral decision theories, Venkatesh states that anchoring and adjustment are important predictors of perceived ease of use. Anchoring and adjustment are important heuristics that individuals use. The model of determinants he used explained almost 60 percent of the variance in perceived ease of use. The determinants are control, intrinsic motivation and emotion. The findings state that there should be a focus on individual difference variables to enhance usage intentions and actual usage (Venkatesh, 2000).
2.2.2 The Theory of Planned Behavior
The TAM model is developed and based on the Theory of Planned Behavior. This theory includes determinants of an individuals’ decision to engage in a particular behavior. The TPB can serve as useful framework in the models of this research to cope with human social behavior. The TPB includes three determinants of an individuals’ decision to engage in a particular behavior. These determinants are attitude toward the behavior, subjective norms, and perceived behavioral control. The more favorable the attitude toward the behavior, the subjective norm with respect to that and the perceived behavioral control, the stronger one’s intention to engage in that particular behavior. The importance and influence of attitude, subjective norms, and perceived behavioral control varies across different behaviors and different situations (Azjen, 1991).
2.2.3 The Expectancy Disconfirmation Theory
usage intentions over time (Bhattacherjee & Premkumar, 2004). Therefore, this theory is added to the model to identify continuance usage intentions of users of no-checkout technologies. There have been some attempts in previous research to extend the EDT. The goal of the research of Meuter et al. (2000) was to identify the factors which influence the satisfaction or dissatisfaction of shoppers with the usage of self-service technologies. Their findings show that satisfaction of a technology is increased by three factors:
1. The ability of the self-service technology to avoid troubling situations.
2. The advantage shoppers perceive to get from using the self-service technology.
3. The degree to which shoppers are surprised by the capabilities of the self-service technology and the degree to which it influences satisfaction.
On the other hand, dissatisfaction of a technology is increased by the following three factors: 1. Technology failure or process failure.
2. A poor technology design. It is important that the design meets shopper needs.
3. Mistakes of shoppers which lead to dissatisfying outcomes, which they named customer-driven failures (Meuter et al., 2000).
2.3 The extended TAM model
H2
H4
3. Theoretical Framework and Model Development
This chapter gives an overview of the conceptual model of both non-users and users of no-checkout technologies. It will discuss the variables and identify the complementing hypotheses.
3.1 Conceptual models
Since no-checkout technologies are still in their trial phase in the Netherlands, there are not many people with usage experience yet. Therefore, this research will focus on shoppers in the pre-adoption and post-adoption situation. Non-users have never used no-checkout technologies before, and therefore they are in a pre-adoption situation. On the other hand, users are in a post-adoption situation and have some usage experience with no-checkout technologies. Therefore, this research contains two conceptual models. The first is applicable to non-users of no-checkout technologies. The second conceptual model is focused on users. The conceptual model for users includes three more variables, since those are only relatable to respondents with usage experience. Figure 1a and 1b summarizes the conceptual models of this research. Each factor is supported and described in the following paragraphs.
Figure 1a. Conceptual model non-users of no-checkout technologies.
Figure 1b. Conceptual model users of no-checkout technologies.
3.2 Intention to use no-checkout technologies
Usage intention is defined as an individuals’ decision to put effort in performing a behavior. It contains the motivational factors that influence a particular behavior. The intention to use no-checkout technologies indicates how hard someone is willing to try and how much effort he wants to exert in order to perform this specific behavior. According to Azjen, a general rule is that “the stronger the intention to engage in a behavior, the more likely should be its performance” (Azjen, 1991, p. 181). Previous research showed that when an individual does not encounter serious problems of control while engaging in a specific type of behavior, this behavior can be predicted of his usage intentions. Therefore, usage intention is considered as an indication of behavior (Sheppard, Hartwick & Warshaw, 1988).
3.3 Perceived usefulness
Perceived usefulness is one of the basis constructs in the TAM model. This construct defines an individual’s attitude toward using a particular technology or system. Perceived usefulness is
H9 H3 H7 H5 H 1 H4 H6 H8 Disconfirmation
Satisfaction experience Usage
defined as the extent to which an individual thinks that using a particular technology will lead to a better performance (Davis, 1989). Therefore, it is expected that perceived usefulness will have a direct effect on both non-users’ and users’ attitude toward no-checkout technologies. Accordingly, the following hypothesis is formulated:
H1: Non-users and users’ perceived usefulness of no-checkout technologies will have a positive
influence on their attitude toward them.
3.4 Perceived ease of use
Perceived ease of use is the second construct in the TAM model. People outweigh the benefits of using a technology by the effort it costs them to use it. Perceived ease of use is defined as the extent to which an individual think that using the technology will not take too much of his effort (Davis, 1989). It is assumed that when everything else stays equal, perceived ease of use will positively influence non-users and users’ attitude toward no-checkout technologies. Accordingly, the following hypothesis is formulated:
H2: Non-users and users’ perceived ease of using no-checkout technologies will have a positive
influence on their attitude toward them.
According to the TAM model, perceived usefulness will be influenced by the perceived ease of use as well. The theory behind this is, when all other variables are being equal, the easier the technology is to use the better its performance will be, i.e. the usefulness (Davis, 1989). Accordingly, the following hypothesis is formulated as well:
H3: Non-users and users’ perceived ease of use will have a positive influence on perceived
usefulness.
3.5 Subjective norms
According to the TPB, subjective norms have a direct effect on the intention of usage behavior. Accordingly, the following hypothesis is formulated:
H4: Non-users and users’ subjective norms will have a positive influence on their intention to
use no-checkout technologies.
3.6 Perceived behavioral control
According to the TPB, an individuals’ perceived behavioral control will predict whether he has an intention to engage in specific usage behavior. However, it has to be considered that no-checkout technologies are relatively new. Individuals may have little knowledge, or they are not familiar with the usage behavior yet. Therefore, it assumed that perceived behavioral control will have less effect on usage intention for non-users, compared to users of no-checkout technologies. Perceived behavioral control can vary across different situations and behaviors. This means that one can have a high perceived behavioral control in general, but a low perceived behavioral control for a specific action or behavior. Perceived behavioral control refers to the perceived ease or difficulty of performing the behavior (Azjen, 1991). Accordingly, the following hypothesis is formulated:
H5: Non-users and users perceived behavioral control over no-checkout technologies will have
a positive influence on their intention to use it.
3.7 Attitude toward no-checkout technologies
Attitude is an important construct in both the TAM model and the TPB. Attitude is defined as the degree to which an individual has a favorable or unfavorable evaluation of a specific behavior (Azjen, 1991). It is assumed that a positive attitude toward the usage of a particular technology will lead to a higher intention of using it. It is expected that attitude will have a mediating role between perceived usefulness or perceived ease of use and intention to use. Attitude determines the behavioral intention to use a technology and leads to the actual usage behavior (Davis, 1989). Accordingly, the following hypothesis is formulated:
H6: Non-users and users’ attitude toward no-checkout technologies will have a positive
3.8 Usage experience
Previous research by Morris and Venkatesh (2000) indicated that age is an important factor influencing technology acceptance. Their research studied age differences in technology adoption and usage by using the TPB. Their results showed that attitudes are considered more important for younger people, whereas subjective norms and perceived behavioral control are more important for older people. This might be caused by the fact that younger people have more experience with using technologies (Morris & Venkatesh, 2000). Therefore, it is assumed that usage experience will influence future usage intentions. Usage experience entails how many times an individual has used no-checkout technologies before. Therefore, the following hypothesis is formulated:
H7: Usage experience with no-checkout technologies will have a positive influence on users’
intention to use it.
3.9 Disconfirmation
Disconfirmation is one of the constructs of the EDT. Disconfirmation is defined as the difference between the initial expectations of the user and the actual performance of the technology. Disconfirmation can be positive or negative, depending on whether the initial expectations are met. Positive disconfirmation occurs when the actual outcome is greater than the initial expectations of the technology. On the other hand, negative disconfirmation occurs when the outcome is less than the initial expectations. Both positive and negative disconfirmation leads to psychological discomfort of users. Psychological discomfort likely results to a negative effect on continuance usage intentions. However, in case of positive disconfirmation the greater outcome is expected to reduce the negative consequences of psychological discomfort. In case of negative disconfirmation, these negative consequences are expected (Oliver, 1980). Furthermore, it is expected that usage satisfaction increases as the initial expectations are met (Rust & Oliver, 2000). Accordingly, the following hypothesis is formulated:
H8: Positive and negative disconfirmation of a users’ initial expectations will have a negative
3.10 Satisfaction
Satisfaction is the second construct of the EDT. Satisfaction is an affect-based factor and defined as an affective state which represents an emotional reaction to the usage experience (Oliver, 1980). Satisfaction is considered as an important factor, since it influences someone’s’ attitude and therefore measures his future usage intentions (McKinney, Yoon & Zahedi, 2002). It is assumed that satisfaction positively influences a users’ attitude toward no-checkout technologies (Bhattacherjee & Premkumar, 2004). Accordingly, the following hypothesis is formulated:
H9: Satisfaction will have a positive influence on users’ attitude toward no-checkout
4. Research Design
This chapter first covers the method of data collection, followed by the population and sampling method of this research. Furthermore, it will give the operational definitions of all variables. Finally, an analysis plan will be provided which will be used to analyze the collected data.
4.1 Data collection
Data is collected by using surveys. This qualitative research method is most appropriate to gain enough data within the time and resource limitations of this research (Blumberg, Cooper & Schindler, 2014). Since the appliance of no-checkout technologies in the Netherlands is still in its trial phase, it is not possible to gain enough respondents with usage experience. Therefore, the data collection looked as follows.
The first part of the data is collected through an electronically administrated survey. These surveys were spread using social media platforms. The first part of the survey is equal for both non-users and user of no-checkout technologies. The first part is closed with the question whether the respondent has experience using no-checkout technologies. A user is directed to a few more questions, and a non-users’ survey is closed after this question (see Appendix A).
The second part of the data collection is focused on gathering respondents with usage experience. Therefore, physical surveys were spread on AH to go locations which make use of no-checkout technologies (i.e. Amsterdam and Zaandam). A QR code is developed which directs the respondent to the online survey. AH to go shoppers can scan this QR code after shopping so they can fill in the survey while they wait or travel.
4.2 Population and sampling method
Population. The population contains both non-users and users of no-checkout technologies. For
the electronically administered surveys there is no need for respondents to be familiar with the technology. Since the survey is spread through social media, anyone can participate in this research. On the other hand, AH to go shoppers should have some experience in using no-checkout technologies in order to have respondents with usage experience in the sample. There are no demographic limitations in this research.
Sample. Besides that there are no demographic limitations, it is preferred to have a deviation in
and snowball sampling method is chosen. A disadvantage of this method is that this results in less control over the sample (Blumberg, Cooper & Schindler, 2014).
Since SmartPLS is used to analyze the data, which will be further explained in paragraph 4.4, there are two conditions of the sample size. First, the sample size should be ten times the number of indicators of the scale with the largest number of formative indicators. In this model there are no formative variables and therefore this condition does not apply. The second condition is that the sample size should be ten times the largest number of structural paths directed to the dependent variable. The largest number of structural paths in the conceptual model of non-users is three, therefore the sample size of non-users should be at least 30. In case of the conceptual model of users, the largest number of structural paths is four. This means the sample size of users should be at least 40 (Chin, 1998).
4.3 Operational definitions
This paragraph gives the definitions of the constructs and how they were measured within this research (see table 1). These constructs are operationalized based on prior research. The complementing survey questions are shown in Appendix A. The original statements are in English. Since respondents in this research are both Dutch and international, backward translation is used. This means the original statements are translated in Dutch by a native Dutch speaker. Accordingly, the Dutch translations are translated by a well-spoken English speaker in English again. Both the original English statements and the translated English statements are compared. This has been done until both statements were identical (Bjorner et al., 1998).
Table 1. Operational definitions.
Variable Operational definition
Intention to use
no-checkout technologies
The intention to use no-checkout technologies is measured according to the article of Davis (1989). Usage intentions are measured with two statements.
Attitude toward
no-checkout technologies
Perceived usefulness & perceived ease of use
The perceived usefulness and perceived ease of use of no-checkout technologies is measured according to the article of Davis (1989). Both constructs are measured with four different statements with the same Likert scale.
Subjective norms Subjective norms are measured with two statements, based on the article of Taylor and Todd (2005).
Perceived behavioral control
Perceived behavioral control is measured according to the article of Taylor and Todd as well (2005). This construct is measured with three different statements with the same Likert scale. Usage experience Usage experience with no-checkout technologies is measured
with a 6-point Likert scale from ‘never’ to ‘more than 4 times (Raju, Lonial & Mangold, 1995).
Disconfirmation Disconfirmation is measured with three statements, based on the article of Oliver (1980).
Satisfaction Usage satisfaction is measured with four items, based on the article of Oliver (1980) as well.
4.4 Plan of analysis
separately for both non-users and users of no-checkout technologies. Therefore, both models were created within the program and analyzed separately.
5. Analysis and Results
This chapter describes the analysis and results of this research. First, the sample of both non-users and non-users is described. After this, the analysis and results of non-non-users and non-users is covered separately. The chapter finishes with analyzing the hypotheses of this research.
5.1 Samples
The sample of this research is divided in two groups, namely non-users and users of no-checkout technologies. Since both groups have a different conceptual model, the results within this chapter are separated as well. Both samples will be highlighted below.
Non-users. The sample of non-users contains 76 respondents, which meets the requirements of
at least 30 respondents (see paragraph 4.2). Of this sample, 37 percent is male, and 63 percent is female. The average age of this sample is 39 years old (see table 2a).
Table 2a. Sample non-users. Sample size N = 76
Age Mean: 39
Min: 18 - max: 72
Gender Males: 28 (37%)
Females: 48 (63%)
Users. The sample of users contains 47 respondents, which meets the requirements of at least
40 respondents (see paragraph 4.2). This relatively low number of respondents, compared to non-users, is due to the fact that no-checkout technologies are still in their trial phase in the Netherlands. Therefore, there are not many users yet. This sample contains 45 percent males and 55 percent females. Most users have used the technology for only one or two times. A few are more experienced and have used it for three or even more than four times (see table 2b).
Table 2b. Sample users. Sample size N = 47
Age Mean: 32
Min: 18 - max: 74
Females: 26 (55%) Usage experience 1 time: 16
2 times: 19 3 times: 9 4 times: 0
More than 4 times: 3
5.2 Analysis of non-users
The model of non-users is created within SmartPLS (see Appendix B: 1). All latent variables of the conceptual model are included with their complementing indicators. The indicators are the statements which have been used in the survey questions. First, the measurement model will be covered, followed by the structural model. Based on these analyses the significant results will be analyzed.
5.2.1 Convergence check
The first step is to check for convergence. Convergence is not very often an issue within SmartPLS, however if it fails to converge the coefficients are unreliable. A convergence check is done by using the matrix “Stop Criterion Changes” (see Appendix B: 3). If the number of listed iterations is below the maximum, the solution converged. The default for the maximum is 300. In this case, convergence was reached in eight iterations.
5.2.2 Measurement model (outer model)
The measurement model is the outer model. This includes the indicators and the arrows connecting them to the corresponding latent variable (Garson, 2016). As indicated in chapter 4, the model of non-users consists of only reflective variables. Therefore, the direction of all arrows goes from the latent variable to the indicators. Only after having a satisfactory measurement model, the assessment of the structural model can be done (i.e. the inner model). Therefore, only the significant indicators and latent variables will be used in further analyses.
Significance. The significance of PLS coefficients is assessed by using Bootstrapping within
show that five out of six variables have significant paths with all their indicators. The variable subjective norms has one indicator which is not significant as indicated with an asterisk. Subjective norms are a reflective variable. This means its indicators are representative for this variable and therefore it is allowed to drop indicators which are not significant. However, since there are only two indicators measuring this variable dropping one of them is not appropriate. Since the T statistic of this indicator is also very close to being significant, it is decided to use all indicators in further analyses.
Table 3. T Statistics outer model non-users.
Latent variables Indicators T Statistics
Perceived usefulness Indicator 1 9.777
Indicator 2 30.621
Indicator 3 20.729
Indicator 4 25.986
Perceived ease of use Indicator 1 11.887
Indicator 2 4.870 Indicator 3 8.445 Indicator 4 5.169 Attitude Indicator 1 49.289 Indicator 2 20.483 Indicator 3 36.547 Indicator 4 47.283 Intention Indicator 1 94.401 Indicator 2 72.821
Subjective norms Indicator 1 1.729*
Indicator 2 3.045
Perceived behavioral control Indicator 1 8.327
Indicator 2 11.450
Indicator 3 5.699
Composite reliability. Composite reliability is used as a test of convergent validity in this
adequacy of the outer model. Composite reliability can vary between 0 and 1, in which a value of 1 means perfect estimated reliability. Since this research is exploratory, the composite reliability should be equal or greater than .60 (Chin, 1998). Table 4 shows the values of composite reliability for this research. Every variable has a composite reliability greater than .60. As indicated, three variables have a composite reliability higher than .90 which is considered as very high. There are two reasons which could have led to this result. First, this may indicate that the indicators are too similar wording variants instead of truly representative measures. It could also be an indication that the indicators are very representative of the variable and therefore correlate highly. In case of the variable attitude, the high composite reliability is expected to be due to similar wordings. However, since this is a validated scale based on prior research it is allowed to assume this scale measures what it should measure. The same explanation applies for the variable intention. The variable perceived usefulness has indicators which ask for different opinions, which indicate that this is not due to similar wordings. Therefore, it can be concluded that these indicators are very representative of this variable.
Table 4. Composite reliability non-users.
Variables Composite reliability
Attitude 0.948
Intention 0.972
Perceived behavioral control 0.894
Perceived ease of use 0.817
Perceived usefulness 0.908
Subjective norms 0.808
5.2.3 Structural model (inner model)
Since the measurement model is satisfactory, the structural model can be assessed. The structural model is the inner model. The inner model includes all latent variables and the arrows connecting them (Garson, 2016).
Significance. The results of the inner model show four out of six paths are significant. The
Table 5. T Statistics inner model non-users.
Paths T Statistics P-value
Perceived usefulness > Attitude 6.014 0.000
Perceived ease of use > Attitude 2.150 0.035
Perceived ease of use > Perceived usefulness
4.152 0.000
Attitude > Intention 6.089 0.000
Subjective norms > Intention 0.306 0.760
Perceived behavioral control > Intention 0.060 0.952
5.2.4 Model measurement
Based on the measurement and structural model, it can be concluded which variables and paths are satisfactory and therefore which paths can be calculated in further analyses. The estimation method which is chosen is PLS Algorithm. PLS Algorithm is the default standard partial least squares modelling procedure (Garson, 2016). Both the outer and inner model results will be highlighted below.
Outer model measurement. Since this model contains only reflective variables, the output of
the outer model are loadings. Outer loadings represent the absolute contribution of an indicator to the latent variable. The loadings are the standardized paths connecting the factor to the corresponding indicator variables. They vary from 0 to 1. The closer the loading to 1, the more reliable the measurement model is. Path loadings should be above .70 for a well-fitting reflective model (Henseler, Ringle & Sarstedt, 2012). Indicators which have a loading between .40 and .70 should be dropped if this will improve reliability. The results show that four out of six variables contain only high path loadings (see Appendix B: 4). On the other hand, perceived ease of use has one out of four indicators with a path loading lower than .70. However, this loading is still close to .70, namely .624, and therefore it is decided to keep all indicators. Furthermore, subjective norms have one out of two indicators which have a path loading lower than .70. Since this path loading is still close to .70 as well, namely .617, it will be kept.
Inner model measurement. The inner model is assessed by estimating the path coefficients. Path
mean the coefficients vary from -1 to +1. Coefficients close to -1 reflect strong negative paths, close to +1 reflect strong positive paths, and close to 0 reflect weak paths. Table 6 shows the results of each significant path. The results show that two paths are strong, namely the effect of perceived usefulness on attitude and the effect of attitude on intention. This means an individuals’ perceived usefulness of no-checkout technologies has a strong effect on his attitude toward them, which will have a strong effect on his intention to use it. The other two paths have a moderate influence. First, the effect of perceived ease of use on perceived usefulness is moderate. This means that an individuals’ perceived ease of using no-checkout technologies, meaning his feeling that using the technology will not take too much effort, will moderately influence his believe that using this technology leads to better performance. Perceived ease of use has a moderate influence on atttidue as well.
Table 6. Path coefficients non-users.
Paths Path coefficient Strength
Perceived usefulness > Attitude 0.645 Strong
Perceived ease of use > Perceived usefulness 0.468 Moderate
Perceived ease of use > Attitude 0.245 Moderate
Attitude > Intention 0.698 Strong
The numbers indicated within the blue circles are the R-square coefficients for the endogenous latent variables (see Appendix B: 1). Endogenous latent variables are the variables in the conceptual model which are an effect of at least one other latent variable (Garson, 2016). In this case, there are three endogenous latent variables. First, perceived usefulness has an R-square of 0.219 meaning that 22 percent of the variance in perceived usefulness is explained by the model. Attitude has an R-square of 0.559 meaning that 56 percent of the variance in this variable is explained by perceived ease of use and perceived usefulness. Lastly, intention has an R-square of 0.509 meaning that 51 percent of the variance in this variable is explained by the model.
5.2.5 Mediation analysis
mediated path to see the consequences in the corresponding correlations. There are two mediation paths in the non-users’ model. The correlations of both paths are shown in table 7.
Table 7. Mediation analysis non-users.
Path Direct paths Mediated paths
1. Perceived usefulness > Intention: 0.705
Perceived usefulness > Attitude > Intention: 0.702
2. Perceived ease of use > Intention: 0.453
Perceived ease of use > Attitude > Intention: 0.443
The results show that the correlations of both paths drop a bit toward 0 after including the mediator attitude. This means both the direct and indirect paths exists. Therefore, there is partial control by the mediating variable attitude in both paths.
5.3 Analysis of users
The model of users of no-checkout technologies is created within SmartPLS as well (see Appendix C: 1). The model includes all latent variables and their complementing indicators. In comparison with non-users, this model contains three more latent variables namely disconfirmation, satisfaction and usage experience. First the measurement model will be covered, followed by the structural model. Based on these analyses the significant results will be analyzed.
5.3.1 Convergence check
The convergence check is done by using the matrix “Stop Criterion Changes” (see Appendix C: 3). If the number of listed iterations is below the maximum, the solution converged. The default for the maximum is 300. In this case, convergence was reached in seven iterations.
5.3.2 Measurement model (outer model)
Significance. The significance of the PLS coefficients in the users’ model is assessed by using
Bootstrapping within SmartPLS as well (see Appendix C: 2). Running a Bootstrap analysis shows that all seven variables have significant paths with all of their indicators (see table 8). The variable disconfirmation has one indicator which is not significant, as indicated with an asterisk. However, since the T statistic of this indicator is very close to being significant, it is decided to use this indicator in further analyses. Therefore, all indicators are used in further analyses.
Table 8. T Statistics outer model users.
Latent variables Indicators T Statistics
Perceived usefulness Indicator 1 9.065
Indicator 2 13.281
Indicator 3 11.509
Indicator 4 35.816
Perceived ease of use Indicator 1 29.349
Indicator 2 4.846 Indicator 3 58.056 Indicator 4 52.211 Attitude Indicator 1 14.103 Indicator 2 6.521 Indicator 3 8.528 Indicator 4 15.098 Intention Indicator 1 70.636 Indicator 2 60.877
Subjective norms Indicator 1 6.270
Indicator 2 3.456
Perceived behavioral control Indicator 1 10.815
Indicator 2 3.593
Indicator 3 8.375
Indicator 4 23.662
Composite reliability. Table 9 shows the values of composite reliability for all variables. Each
variable has a composite variability higher than the required .60. As indicated, seven of these variables have a composite reliability higher than .90 which is considered as very high. The same explanation applies here for the variables attitude, intention and perceived usefulness, as indicated in the analysis of non-users in sub paragraph 5.2.2. The indicators of perceived behavioral control, perceived ease of use and subjective norms are similar wording variants but proved to be valid based on prior research. Usage experience has a composite reliability of 1.000, this is due to the fact that this variable only has one indicator. Disconfirmation has a relatively low composite reliability of 0.696. This is due to indicator 2, since this indicator has a low outer loading (see Appendix C: 4). However, since this composite reliability is still above the required .60 it is decided to continue with all variables.
Table 9. Composite reliability users.
Variables Composite reliability
Attitude 0.940
Intention 0.987
Perceived behavioral control 0.932
Perceived ease of use 0.933
Perceived usefulness 0.938
Subjective norms 0.951
Disconfirmation 0.696
Satisfaction 0.892
Usage experience 1.000
5.3.3 Structural model (inner model)
Significance. The results of the inner model show four out of nine paths are not significant, as
indicated in red in table 10. Therefore, these four effects cannot be interpreted and will not be covered in subsequent analyses.
Table 10. T Statistics inner model users.
Paths T Statistic P-value
Perceived usefulness > Attitude 0.824 0.414
Perceived ease of use > Attitude 3.433 0.001
Perceived ease of use > Perceived usefulness 2.934 0.005
Attitude > Intention 3.130 0.003
Subjective norms > Intention 2.044 0.046
Perceived behavioral control > Intention 0.825 0.414
Disconfirmation > Satisfaction 9.136 0.000
Satisfaction > Attitude 0.487 0.629
Usage experience > Intention 1.389 0.171
5.3.4 Model measurement
Based on the measurement model and structural model, it can be concluded which variables and paths are satisfactory and therefore which paths can be interpreted in further analyses. The estimation method which is chosen is similar to the research of non-users, namely PLS Algorithm (Garson, 2016). Both the outer and inner model results are highlighted below.
Outer model measurement. Since the model of users contains only reflective variables as well,
Inner model measurement. The inner model is assessed by estimating the path coefficients. The
PLS Algorithm model indicates the path coefficients which connect the latent variables (see Appendix C: 1). Table 11 shows the results of each significant path. The results show that there are two strong paths, the effect of attitude on intention and the effect of disconfirmation on satisfaction. This means an individuals’ attitude toward no-checkout technologies strongly and positively influences his usage intentions. The perceived difference between the initial expectations of the user and the actual performance of no-checkout technologies, which is called disconfirmation, strongly influences his satisfaction with the technology. The effect of perceived ease of use on attitude and perceived usefulness is moderate. This means that an individuals’ feeling that using the technology will not take too much effort will moderately influence his believe that using the technology leads to better performance and his attitude. The effect of subjective norms on intention is negative. This means that the perceived social pressure to use no-checkout technologies negatively influences an individuals’ usage intentions.
Table 11. Path coefficients users.
Paths Path coefficient Strength
Perceived ease of use > Attitude 0.361 Moderate
Perceived ease of use > Perceived usefulness 0.441 Moderate
Attitude > Intention 0.739 Strong
Subjective norms > Intention - 0.196 Moderate
Disconfirmation > Satisfaction 0.654 Strong
5.3.5 Mediation
There are three mediation paths in the users’ model. The same analysis has been done as in the non-users’ model. The correlations of all paths are shown in table 12.
Table 12. Mediation analysis users.
Path Direct paths Mediated paths
1. Perceived usefulness > Intention: 0.251
Perceived usefulness > Attitude > Intention: 0.238
2. Perceived ease of use > Intention: 0.453
Perceived ease of use > Attitude > Intention: 0.450
3. Disconfirmation > Attitude: 0.175
Disconfirmation > Satisfaction > Attitude: 0.173
The results show that the correlations of all paths drop a bit toward 0 after including the mediators. This means both the direct and indirect paths exist in all relations. Therefore, there is partial control of the mediating variables attitude and satisfaction in this model.
5.4 Further research into insignificance
perceived behavioral control on intention (p-value = 0.046, T Statistic = 2.050), however this effect was insignificant in the complete model. The effect of perceived behavioral control is moderate, with a path coefficient of .328. Since insignificant effects are significant in the smaller model, it can be concluded that the insignificant effect in the complete model might be due to a too small sample size. Therefore, this significant effect is used in further analyses. The effect of subjective norms is significant and the effect of usage experience is insignificant, similar to the results of the complete model.
Figure 2. User model split.
5.5 Hypotheses
Based on the analyses within SmartPLS it can be determined which hypotheses are accepted and rejected within this research. The analysis of the hypotheses is divided in non-users and users of no-checkout technologies. Figure 3a and 3b show both conceptual models including the significant path coefficients. Rejected hypotheses are indicated with dotted arrows. As indicated, four out of six hypotheses are accepted in the research of non-users. In the research of users six out of nine hypotheses are accepted.
H2: 0. 245
H5: 0.328 Figure 3a. Results non-users.
Figure 3b. Results users.
H4: - 0.196 H6: 0.739 H3: 0.468 H6: 0.698 H 1: 0.645 H8: 0.654 Attitude toward no-checkout technologies Perceived usefulness Perceived ease of use Subjective norms Perceived behavioral control Intention to use no-checkout technologies Disconfirmation
Satisfaction experience Usage
6. Discussion
This chapter covers the discussion of the results. The results show that four out of six hypotheses of the non-users’ model are accepted. In the model of users, six out of nine hypotheses are accepted. Each hypothesis will be covered and discussed below.
H1: Non-users’ and users’ perceived usefulness of no-checkout technologies will have a
positive influence on their attitude toward them.
Based on the TAM model, it is assumed that perceived usefulness has a positive influence on both non-users’ and users’ attitude toward no-checkout technologies. The extent to which a user thinks that using no-checkout technologies will lead to a better performance influences his attitude (Davis, 1989). The results of non-users show their perceived usefulness of no-checkout technologies have a strong, significant effect on their attitude toward them. Therefore, this hypothesis is accepted. This means that when a non-user believes that using no-checkout technologies will lead to a better performance, this will positively influence his attitude toward the technology. On the other hand, the results of users show that the effect of perceived usefulness on attitude is not significant. Therefore, this hypothesis is rejected. This result is initially very unlikely, since the effect proved to be significant in the research of non-users. This insignificant effect could be due to two reasons. First, users are not very experienced yet. Of the sample, 74 percent have used no-checkout technologies for only 2 times or even less. Therefore, they might not be able to judge whether they think the technology increases their performance. A second explanation could simply be that users do not think no-checkout technologies improve their performance.
H2: Non-users’ and users’ perceived ease of using no-checkout technologies will have a positive
influence on their attitude toward them.
H3: Non-users’ and users’ perceived ease of use will have a positive influence on perceived
usefulness.
The positive influence of perceived ease of use on perceived usefulness is assumed based on the theory behind the TAM model (Davis, 1989). The results of non-users show that there is a moderately strong and significant effect of perceived ease of use on perceived usefulness. Therefore, this hypothesis is accepted. The results of users confirm this hypothesis as well, since this effect is significant. This means that when both non-users and users think that no-checkout technologies will not take too much of their effort, this will result in an increase of their perceived usefulness, i.e. their performance.
H4: Non-users’ subjective norms will have a positive influence on their intention to use
no-checkout technologies.
H5: Non-users perceived behavioral control over no-checkout technologies will have a positive
influence on their intention to use it.
Based on the TPB, it is assumed that perceived behavioral control predicts usage intentions of no-checkout technologies. However, since non-users might be unfamiliar with the technology the effect of non-users is expected to be less compared to users (Azjen, 1991). The results of the research among non-users show that the effect of perceived behavioral control over no-checkout technologies did not have a significant effect of non-users’ usage intentions. Therefore, this hypothesis is rejected. The non-users in this research have little or no knowledge of no-checkout technologies yet. Therefore, it is not realistic to all of them to decide whether they think the usage of it would be easy or difficult to them. This could explain the insignificant result. Furthermore, the results of the research among users initially showed that the effect of perceived behavioral control on usage intention is insignificant. Further analysis into insignificant hypotheses revealed that this insignificant effect is due to a small sample size (see chapter 5, paragraph 5.4). This research with a relatively larger sample size resulted in a significant effect of perceived behavioral control on usage intentions. Therefore, this hypothesis is accepted. This means that users’ perceived behavioral control influence their usage intentions of no-checkout technologies.
H6: Non-users’ attitude toward no-checkout technologies will have a positive influence on their
intention to use them.
Based on the TAM model and the TPB, the attitude toward no-checkout technologies has a positive influence on usage intentions (Azjen, 1991 and Davis, 1989). The results of the research among non-users show that their attitude toward no-checkout technologies do have a positive effect on their usage intentions. Therefore, this hypothesis is accepted. This effect is very strong with a path coefficient of .698. This means there is a strong effect of a non-users’ attitude on his usage intentions. The research among users show this effect is significant as well, and therefore this hypothesis is accepted. The effect for users is even stronger with a path coefficient of .739.
H7: Usage experience with no-checkout technologies will have a positive influence on user’
intention to use it.
usage experience with no-checkout technologies does not have an effect on usage intentions of current users. Therefore, this hypothesis is rejected. This could be due to the fact that the usage experience of 74 percent of the users in this research is very minimal. Therefore, the influence of usage experience is minimal as well. It is assumed that when users will gain more experience, their experience will have influence on their future usage intentions.
H8: Positive and negative disconfirmation of a users’ initial expectations will have a negative
influence on satisfaction.
This hypothesis is only applicable to the research among users of no-checkout technologies. Based on the EDT, it is assumed that both positive and negative disconfirmation of a users’ initial expectations of no-checkout technologies has a negative influence on their satisfaction with it (Oliver, 1980). The results of this research show that disconfirmation has a strong and significant, but positive effect on satisfaction. The effect is significant, and therefore this hypothesis is accepted. However, the result of disconfirmation on satisfaction is positive instead of negative as initially expected. A positive effect of disconfirmation on satisfaction does not make sense, and therefore it is assumed that there is a measurement error. This is assumed to be due to misinterpreted questions in the survey. Since the original English statements were translated in Dutch, this could have led to misinterpretation of the respondents.
H9: Satisfaction will have a positive influence on users’ attitude toward no-checkout
technologies.
7. Conclusions and Recommendations
This chapter first introduces this research, followed by the main findings. After this the scientific and managerial relevance is described. This chapter finishes with the limitations of this research and future research directions.
7.1 Introduction
This research built upon and extended TAM research. The TAM model serves as a useful foundation to build further on researching user acceptance of technologies (George, 2004). The aim of this research is divided in two parts. Part one critically reviews the appliance of the TAM model on the user acceptance of technologies by extending the original model. Based on this critical review, the second part applies the extended TAM model to non-users and users of no-checkout technologies. Accordingly, the research question of this research is formulated as follows:
“Which variables can contribute to the Technology Acceptance Model in order to improve this model, and when this model is applied to no-checkout technologies to which insights does
this result about the drivers of non-users and users in the Netherlands of no-checkout technologies?”
No-checkout technologies are recently introduced in the Netherlands and still in their trial phase. Therefore, there are not many shoppers in the Netherlands with usage experience yet. This research is divided in non-users and users of no-checkout technologies. The original TAM model has been extended by including variables of the Theory of Planned Behavior (TPB) and the Expectancy Disconfirmation Theory (EDT). The variables of the EDT are only included in the research of users, since this model is only applicable to respondents with usage experience. Data has been collected of 76 non-users and 47 users of no-checkout technologies via an electronically administered survey. Accordingly, the data has been analyzed separately for both groups by using SmartPLS. After this, the results of both groups are compared.
7.2 Important findings
First, the attitude of non-users is proved to be influenced by perceived usefulness and perceived ease of use. On the other hand, the attitude of users is only influenced by perceived ease of use. However, this is expected to be due to the lack of experience current users have. Both groups confirm the theory behind the TAM model by showing that the effect of perceived usefulness is increased by the perceived ease of use. The research among users show there is a significant effect of disconfirmation on satisfaction. However, the result of disconfirmation on satisfaction is positive instead of negative as initially expected. This is assumed to be due to a measurement error in the survey. This also led to an insignificant result of satisfaction on attitude.
The intention to use no-checkout technologies is not significantly influenced by other variables in the research among non-users. This is expected to be due to the fact that non-users are not familiar with, but also lack knowledge of no-checkout technologies. On the other hand, the research among users shows that both subjective norms and perceived behavioral control influence their future usage intentions. An unexpected finding is that subjective norms negatively influence users’ future usage intentions. This might be due to the fact that only a few people in the Netherlands are familiar with no-checkout technologies. Therefore, users might not feel pressure from others but actually give pressure to others themselves as they consider themselves as the early adopters.
7.3 Implications
Scientific relevance. This research was an initial attempt to extend the original TAM model for
appliance to no-checkout technologies. This research showed that the TAM model can be combined with the TPB to include perceived behavioral control and social pressure. However, the results showed that social pressure is not as important in this early phase of no-checkout technologies as initially expected. This variable might be of more value when no-checkout technologies are more familiar in the Netherlands. The EDT gave valuable insights in continuance usage intentions. However, further research needs to be done into the effect of disconfirmation on satisfaction.
Managerial relevance. The results of this study help practitioners by providing insights in
potential users. The results of this research indicated that a positive attitude is influenced by perceived usefulness and perceived ease of use. This means that if the Dutch retail market wants to succeed with no-checkout technologies, it should focus on user friendliness. The results also show that there is a lack of knowledge and experience with no-checkout technologies. To increase usage intentions in the Netherlands, shoppers’ knowledge of this technology is required.
7.4 Limitations of the study
This study has a few limitations. A first limitation of this study is the fact that there are not many users of no-checkout technologies yet. However, this was solved to split the model for both non-users and users. Therefore, this research was able to provide insights in the attitudes and usage intentions of both non-users and users toward no-checkout technologies. Secondly, this research had to deal with unfamiliarity of no-checkout technologies. This could have led to incorrect insignificant results. Thirdly, the average age of the sample of users is relatively low. However, early adopters are relatively young according to the theory of Rogers (Rogers, 2010). Since no-checkout technologies are still in their trial phase, they are mainly focused on early adopters. Therefore, this relatively low average age is appropriate. Lastly, while writing this thesis unfortunately a similar research has been published in the second volume of the Journal of Public Sector Performance Management (Chuawatcharin & Gerdsri, 2019). However, this research differs from the published research since this research threats non-users and users as different groups. This research is able to compare results of non-users and users while the research of Chuawatcharin and Gerdsri (2019) gathered results for both groups simultaneously. Therefore, this research gives more specific insights in the relatively largest group in the Netherlands, namely non-users. Furthermore, this research includes the EDT for the users group to gain insights into continuance usage intentions. The results of both the published research and this research can be used to identify future research areas to expand current research within no-checkout technologies.
7.5 Suggestions for future research
References
Azjen, I. (1991). The theory of planned behavior. Organizational behavior and human decision
processes, 50(2), 179-211.
Albert Heijn. (2018). AH to go heeft Europese primeur met pin- en kassaloos shoppen. Retrieved from: https://nieuws.ah.nl/ah-to-go-heeft-europese-primeur-met-pin--en- kassaloos-shoppen/. Accessed on 8 March 2019.
Bhattacherjee, A. (2001). Understanding information systems continuance: an expectation- confirmation model. MIS quarterly, 351-370.
Bhattacherjee, A., & Premkumar, G. (2004). Understanding changes in belief and attitude toward information technology usage: A theoretical model and longitudinal test. MIS
quarterly, 229-254.
Bhattacherjee, A., & Premkumar, G. (2008). Explaining information technology usage: A test of competing models. Omega, 36(1), 640-75.
Bjorner, J. B., Thunedborg, K., Kristensen, T. S., Modvig, J., & Bech, P. (1998). The Danish SF-36 Health Survey: translation and preliminary validity studies. Journal of clinical
epidemiology, 51(11), 991-999.
Blumberg, B. F., Cooper, D., & Schindler, P. (2014). Business Research Methods. Mcgraw- Hill Education.
Chin, W. W. (1998). The partial least squares approach to structural equation modeling. Modern methods for business research, 295(2), 295-336.
Chuawatcharin, R., & Gerdsri, N. (2019). Factors influencing the attitudes and behavioural intentions to use just walk out technology among Bangkok consumers. International
Journal of Public Sector Performance Management, 5(2), 146-163.
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of
information technology. MIS quarterly, 319-340.
Demoulin, N. T. M., & Djelassi, S. (2016). An integrated model of self-serivce technology (SST) usage in a retail context. International Journal of Retail and Distribution
Management, 44(5), 540-559.
Garson, G. D. (2016). Partial least squares: Regression and structural equation models. Asheboro, NC: Statistical Associates Publishers.
George, J. F. (2004). The theory of planned behavior and Internet purchasing. Internet
