Improve the prediction of enterprise system acceptance with clockspeed

(1)

Master Thesis

Improve the prediction of enterprise system acceptance with

clockspeed

Jeroen de Haan

Student ID 10548246

jeroen.dehaan@student.uva.nl

MSc Information Studies: Business Information Systems

Supervisor

Dr. Dick Heinhuis

d.heinhuis@uva.nl

University of Amsterdam - The Netherlands

October 27, 2015

(2)

(3)

Improve the prediction of enterprise system

acceptance with clockspeed

Jeroen de Haan

MSc Information Studies: Business Information Systems University of Amsterdam - The Netherlands

Abstract

An extension to the Unified Theory of Acceptance and Use of Technology (UTAUT) created in 2003 by Venkatesh et al. has been proposed to integrate clockspeed into this model. Openness to change has been added which has influence on a person’s behavioural intention. This influence is moderated by age, gender, and clockspeed. To validate this extension, hypotheses have been created based on the proposed model. An anonymous survey has been conducted under working people in The Netherlands who have experience with Enterprise Systems (ES). 217 respondents from 18 to 68 years of age filled in the survey, 125 men and 92 women. Based on findings, implications to both theory and practice are discussed.

1 Introduction

There is a great interest in both the academic and practitioner oriented literature on strategic manage-ment in the speed of changes in industries [17] [20] [36] [43]. Continuous improvemanage-ments to respond quick market changes are indispensable for organisations to survive in constantly changing global markets [72] [109]. Organisations are therefore under increasing pressure from stakeholders to find new ways to compete effectively in these dynamic markets and changing customer preferences. For these business innovations, clockspeed has been considered to be a possible moderator for determinants of innovation success [45].

Clockspeed is the speed of changes in an industry, execution of product development, and manufacturing [77]. There are three facets of industry clockspeed: product, process, and organisational [38]. Product clockspeed represents new product introduction and product obsolescence rates [38]. The aircraft industry is an example of an industry with a slow clockspeed, because firms launch, on average, about two new products per decade. The motion picture industry, in contrast, is considered as an example of a fast clockspeed, which created dozens of new products a year. Process clockspeed reflects the rates at which process technologies are replaced in an industry [22]. Because of the frequent replacements of process technologies, the semiconductor industry is an example of an industry with high process clockspeed. Firms in this industry invest about a billion dollars in a wafer fabrication plant and expect that plant to be obsolete in 5 years [44]. In contrast, in the automobile industry has a slow rate of change, because replacement of process technologies is less frequent. Therefore automobile firms can invest significantly in an auto assembly plant because it provides cash flow for about 20 years [38]. Organisational clockspeed reflects the rate of change in the strategic actions and structures of firms in an industry [38]. Product, process, and organisational clockspeeds reflect together industry-level changes based on aggregate actions initiated by all the companies in the industry [82]. Qu et al. [92] states industrial clockspeed as one key environmental dimension. Industries that are characterised by a fast clockspeed usually have shorter time for development cycles and have less time between product improvements, for instance the redesign process [77]. These industries are more likely to develop processes that a cope with rapid changes in process given that the clockspeed of these organisations requires new activities such as the development of new products and other internal operations [22] [77] [82]. On the other hand, industries characterised by a slow clockspeed usually have stable technologies and markets [22]. Generally, these enterprises still require to develop new business processes, however, not that often as the fast clockspeed industries.

These days there are no stable industries anymore, all organisations have to cope with changes [82]. Be-cause organisations operate according to business processes, changes are required in a company’s business

(4)

process. These changes do not only include changes in activities, but also the introduction of new Enter-prise Systems (ES). ES is software which supports business processes, information flows, and reporting in organisations [28]. Employees who are involved in these business processes are required to cope with these changes in their daily operations. Because most business processes include an ES, it is useful to examine the influence of flexibility on the acceptance of an ES. Changes in ES require some degree of flexibility from employees, they need to be open to change, to make changes in an organisation a success [117]. Literature states that [41] flexibility may have important consequences for the operational efficiency and long-term ef-fectiveness of an ES, it is often not considered explicitly as a decision factor during system design, selection, and implementation. Organisations change to cope with changes in industries, therefore their information systems change with it. Because nowadays all organisations have ES [114], there are also requirements imposed on flexibility of employees. The implementations of an ES is usually a lengthy and complex process during which the organisation undergoes many difficulties and has to remove various barriers for the project to succeed [108]. In 2014, Qu et al. [92] published a study that sourced from a program that collects data related to the use of use of information and communication technology and e-business in enterprises by means of representative surveys. The 2007 survey they used included the chemicals, steel and furniture sector of industry. In total, representatives from 300 US enterprises (which had at least 10 employees) from the three sectors were interviewed in regard to their IT and e-business usage. Their results shows enterprise systems do not have a negative effect on process flexibility. Rather the result is consistent with the view that, while enterprise systems inhabit the flexibility of existing processes, they enable the creation of new processes to cope with environmental changes [42]. Obviously, the result of ES implementation can be var-ied, from overall success that results in an increase of a company’s profitability, to a complete failure where a company goes bankrupt. One of the aspects that determines the degree in which the implementation of an ES is successful is the willingness to accept and use the new ES [105].

Consequently, it is important to investigate aspects that influence user acceptance of an ES. Because this acceptance is an important aspect within organisations, many popular models that describe an individuals acceptance of an ES have been developed. These models are used to gain insight in the many different aspects that are involved during user acceptance and and have been applied to different technologies, e.g. word processors [19], e-mail [104], health care information systems [87], e-learning technologies [85]. These models have been improved over time, just as companies have improved their processes and systems. However, models that are widely used to gain insight in user acceptance do not mention clockspeed and related aspects, such as the flexibility and the openness of its employees to changes. Therefore, it is likely that developments in the field of user acceptance, such as clockspeed, have not been added to these models.

The focus of this study will be on whether the addition of clockspeed to the most recent frequently used model that describes an individuals acceptance of an ES improves prediction of an individuals acceptance of an ES. Based on this scope, the following research question is formulated to be answered in this thesis:

To what extent does the inclusion of clockspeed as a variable improve prediction of an individuals acceptance of an Enterprise System?

To answer this research question, first the following sub-questions must be answered in order to come up with a complete answer to the research question:

1. Which model that describes an individuals acceptance of an ES is the the leading model? 2. How could this model be extended with clockspeed?

(5)

This thesis has an academic relevance because it illustrates a possibility to improve a model that describes an individuals acceptance of an ES with clockspeed. This extension will be tested quantitatively by creating and conducting a survey, analysed using statistical tests to gain more insights into the results of this survey, and validated by creating and testing hypotheses. Besides the academic relevance, this thesis also has an theoretical relevance because it provides insights in the history of models that describe an individuals acceptance of an ES, which aspects are involved in the acceptance of an ES, and it will gain insight in ways to improve the prediction of ES acceptance by the addition of clockspeed as an variable.

In the first two subsections of the theory section, literature about aspects relevant to an users acceptance of an ES will be covered, such as the history of the models that describe an individuals acceptance of an ES, clockspeed, and openness to change. After the theory the proposed model and hypotheses will be presented. The third section will cover the methodology used to test the proposed model. Followed up by the results of the conducted survey in the fourth section. The fifth and final section of this thesis will cover the conclusion, discussion, and limitations.

2 Theory

To answer the first sub-question it is essential to determine the leading model that describes an individuals acceptance of an ES. To do this, the history of models that describe an individuals acceptance of an ES will be covered by providing a brief summary of this history. To provide a good impression of history, the models are ordered chronologically.

2.1 Models that predict an individuals acceptance of a system

In 1962, Rogers states in the Innovation Diffusion Theory (IDT) [96] that a person has the will to perform or not perform the behaviour to accept an innovation. The IDT is one of the oldest social science theories. Its goal was to explain how an idea or product over time gains momentum and diffuses through a specific group or social system. The diffusion could have a wide variety of consequences, positive and negative [95]. Rogers [96] states that the adoption of a new idea, product, or behaviour is a process whereby some people are more willing to adopt the innovation than others. Rogers [96] found out people who adopt an innovation early have different characteristics than people who adopt later on. It is important, when promoting an innovation to a target group, to understand the characteristics of the target group. In 2008, Zhou [134] reconfirmed that the perceived attributes are the most powerful predictors of innovation adoption with his study.

Because there was a vast amount of research however, little agreement about what an attitude is and what role it plays in influencing or determining behaviour [39], the Theory of Reasoned Action (TRA) was proposed by Fishbein and Ajen in 1975 [39] [6]. It provides a coherent and systematic conceptual framework applied to the diverse literature on attitudes. In TRA, there is a central role for the Behavioural Intention, because intentions are assumed to capture the motivational factors that influence behaviour [39]. Therefore, a major assumption of TRA is that most human social behaviour is under volitional control and can be predicted from people’s intentions [4]. According to Ajzen [5], intentions are indications of how hard people are willing to try, of how much of an effort they are planning to exert, in order to perform the behaviour. Ajzen [5] states the following general rule: the stronger the intention to engage in a behaviour, the more likely should be its performance. The Behavioural Intention (BI) is jointly determined by the person’s attitude (A) and subjective norm (SN) concerning the behaviour in question, with relative weights typically estimated by regression: BI = A + SN [123].

The Technology Acceptance Model (TAM), introduced in 1986 by Davis [30], is an adoption of TRA. The goal of TAM was to explain an individuals computer use and acceptance. TAM assumes that an individual’s information systems acceptance is determined by two major variables: Perceived Usefulness (PU) and Perceived Ease of Use (PEOU). The information systems community considered TAM a parsimonious and powerful theory [123] [73] [97] [68]. Despite TAM’s popularity, it has its limitations. The most common

(6)

reported limitation is self-reported usage [68]. Respondents may exaggerate symptoms in order to make their situation seem worse or they may under-report the severity or frequency of symptoms in order to minimise their problems. This distorts and exaggerates the causal relationship between independent and dependent variables [2] [91]. The second most cited limitation of the studies is the tendency to examine only one information system with a homogeneous group of subjects on a single task at a single point of time. This displays the generalisation problem of any single study [68].

Also in 1986, Bandura [14] proposed the social foundations of thought and action, where he explains the Social Cognitive Theory (SCT). SCT states people are neither driven by inner forces nor automatically shaped and controlled by the environment, however they function as contributors to their own motivation, behaviour, and development within a network of reciprocally interacting influences [14].

The Theory of Planned Behaviour (TPB) is an extension of TRA [6] [39]. Ajzen [5] proposed TPB in 1991 and stated this extension was required because TRA has limitations in dealing with behaviours over which people have incomplete volitional control. Ajzen states in TPB that Behavioural Intention only can occur when the person can decide at will to perform or not perform the behaviour [5]. Therefore Ajzen states, the performance depends at least of some non-motivational factors such as the availability of requisite opportunities and resources [5]. Because these factors represent a person’s actual control over their behaviour, a person should have the required opportunities, resources, and intentions to perform the behaviour in order to succeed in accepting that [5].

To help to gain a better understanding of factors that influence the use of personal computers, Thompson et al. [113] created, in 1991, the Model of PC Utilisation (MPCU) and confirmed the importance of the expected consequences of using PC technology, suggesting that training programs and organisational policies could be instituted to enhance or modify these expectations. The relationship between social factors and utilisation is significant [113]. This is consistent with the theory of reasoned action proposed by Fishbein and Azjen [39].

In 1992 Davis et al. [29] created the Motivational Model (MM) based on two studies concerning the rela-tive effects of usefulness and enjoyment on intentions to use, and usage of, computers in the workplace where they compared the influence of perceived usefulness and enjoyment on intentions to use computers in the workplace. In both studies there is a positive interaction observed between usefulness and enjoyment, which implies that enjoyment has a greater positive effect on intentions when the computer system is perceived to be more useful [29]. Because Davis et al. [29] found a inverse relationship between productivity and satisfaction mention, they suggest that insufficient satisfaction and enjoyment can undermine the adoption of otherwise productive computer systems.

In 1995, Taylor and Todd [112] combined TAM and two variations of the TPB (The Theory of Planned Behaviour and the Decomposed Theory of Planned Behaviour) to assess which model best helps to under-stand usage of information technology. C-TAM-TPB adapted its attitude towards behaviour, subjective norm and perceived behaviour control from TPB and perceived usefulness from TAM.

Venkatesh and Davis [123] proposed an extension of TAM in 2000, called TAM2. They have improved TAM by identifying and theorising about the general determinants of perceived usefulness (subjective norm, image, job relevance, output quality, result demonstrability, and perceived ease of use) user acceptance.

Based on a review of literature, Venkatesh et al. [124] developed, in 2003, the Unified Theory of Acceptance and Use of Technology (UTAUT) as a comprehensive synthesis of prior technology acceptance research (see figure 1). They compared and tested the variables of eight different models for user technology acceptance (TRA, TPB, TAM, MM, C-TPB-TAM, MPCU, IDT and SCT). UTAUT identifies the key factors in acceptance of ICT as measured by behavioural intention to use the technology and actual usage. The four determinants of ICT acceptance are performance expectancy, effort expectancy, social influence, and facilitating conditions. Performance expectancy was defined as the degree to which an individual believes that using the system will help him or her to attain gains in job performance [124]. Effort expectancy was defined as the degree of ease associated with the use of the system [124]. Social influence was defined as the degree to which an individual perceives that important others believe he or she should use the new system [124]. Facilitating conditions was defined as the degree to which an individual believes that an organisational

(7)

and technical infrastructure exists to support use of the system [124]. The main advantage of the UTAUT model is its predictive efficiency, this is 70%, much higher than other models [86]. The main limitation of UTAUT is that it does not include cultural factors. Im et al. [57] stated it is crucial to obtain a better understanding of the impact of culture on technology adoption because technology is used all over the world. Despite its limitations, UTAUT is an important concept because it integrated eight major theories and was tested on a large real world data set [125].

Figure 1: The Unified Theory of Acceptance and Use of Technology

The Technology Acceptance Model 3 (TAM3), proposed in 2008 by Venkatesh and Bala [122], combined TAM2 and and the model of the determinants of perceived ease of use [121]. In TAM3, shown in figure 2, experience has gained a larger role. Venkatesh and Bala [122] state that by increasing hands-on experience with a system, a user will have more information on how easy or difficult the system is to use. Experience will moderate the effect of computer anxiety on perceived ease of use. The effect of computer anxiety on perceived ease of use will diminish. That experience will moderate the effect of perceived ease of use on behavioural intention.

(8)

Figure 2: The Technology Acceptance Model 3

In 2012, Venkatesh et al. [125] proposed extensions to the original UTAUT and named it Unified Theory of Acceptance and Use of Technology 2 (UTAUT2), shown in figure 3. It focuses on the consumer technology use context instead of workplace technology use context. Therefore, factors important for a consumer are included. One of these factors is Hedonic Motivation. It is defined as the fun or pleasure derived from using a technology, and it has been shown to play an important role in determining technology acceptance and use [53]. In a consumer technology use context, price is also an important factor as, unlike workplace technologies, consumers have to bear the costs associated with the purchase of devices and services. The third and last added factor is Habit. Venkatesh et al. stresses the habits and experience not the same thing in this model. Experience reflects an opportunity to use a target technology and is typically mentioned as the passage of time from the initial use of a technology by an individual [124]. Habit has been defined as the extent to which people tend to perform behaviours automatically because of learning [71], while Kim et al. [63] equate habit with automatism.

(9)

Figure 3: The Unified Theory of Acceptance and Use of Technology 2

2.2 The leading model that predicts an individuals acceptance of a system

The history of the models that describe an individuals acceptance of an ES has been covered. Therefore, the leading model can be determined. Based on the history of models that describe an individuals acceptance of an Enterprise System, UTAUT, UTAUT2 or TAM3 should be selected as the most recent and most used model. UTAUT2 is the most recent model, however, it is the only one which focuses on the consumer market, while UTAUT and TAM3 focuses on the businesses, therefore the UTAUT2 is not the most appropriate model. When looked at the emergence of TAM 3, it is a combination of TAM2 and the model of the determinants of perceived ease of use [122]. These both models are extensions of TAM [123], while UTAUT reviewed and compared eight extant user acceptance models to formulate an unified model. Both models have used TAM as a source, however, because UTAUT used more user acceptance models which provides a unified view of the user acceptance models. Therefore, UTAUT has been chosen to be the leading model for this thesis.

The first sub-question has been answered because the leading model has been determined. Thus, the second sub-question can be covered. The following two subsections will demonstrate how UTAUT could be extended with clockspeed. Although clockspeed has been introduced in the introduction, it is essential to gain more insights into an individuals openness to change because this may have an influence on clockspeed, before UTAUT can be extended with clockspeed and the proposed model can be introduced.

2.3 Clockspeed and openness to change

Organisations do not exist by themselves in a vacuum, they are a part of interactive and dynamic en-vironments. In today’s highly turbulent, extremely competitive environment, fast communications and technology developments, change of laws, and globalisation, the modern organisation faces considerable pressure to meet or exceed the expectation of customers, beneficiaries and stakeholders by delivering prod-ucts and services that are of the highest quality. As organisations attempt to cope with a progressively more turbulent economic, technological, and social environment they rely increasingly on their employees to adapt to change [13]. However, political behaviour inevitably accompanies organisational change [21] [90] because employees often resist change [110]. Organisations need to cope with resistance to make a successful

(10)

change. Openness of employees towards organisational change has long been identified as an underlying trait of flexibility and contrasted with intolerance, rigidity, dogmatism, and premature closure [76]. In re-cent years, psychologists have suggested that an individual’s capacity to be cognitively and behaviourally flexible in dealing with new situations is one of the key factors in personality structure [34]. Miller et al. [78] and Devos et al. [33] have defined openness to change as the (a) willingness to support the change and (b) positive affect about the potential consequences of the change, e.g. the feeling that the changes will be beneficial in some way (p. 609).

There are aspects that increase the openness to change, such as an increase in job security [59]. This produces greater organisational commitment [79], which is associated with favourable attitudes to organi-sational change [59]. Other aspects decrease the openness to change, such as a membership to an union. Union members were found to be less accepting of change than non-union members [59].

Based on the information provided in the previous subsections about models that predict an individual acceptance of a system, clockspeed, and openness to change, the following subsection will propose an extension of UTAUT, which provides an answer for the second sub-question.

2.4 Proposed model

To show how UTAUT could be extended with clockspeed, openness to change and clockspeed have been added to UTAUT. This proposed model is introduced in figure 4.

Figure 4: The proposed model.

Because openness to change increases due to the increased clockspeed of organisations [92], it plays an important role in determining technology acceptance and use. Therefore, it is essential to add openness to change as an factor of influence and clockspeed as moderator to the proposed model.

The second sub-question has been answered by presenting the proposed model (figure 4) and motivating the model. In the following subsections the third sub-question will be covered.

(11)

2.5 Hypotheses

To answer the third sub-questions, which requires for a empirically validated test of the proposed model, hypotheses are drawn up. This section will provide the hypotheses and motivation.

Literature states good and bad experiences with changes in an organisation have an effect on open-ness to change. From the perspective of social exchange theory, organisational cynicism is generated from the formation of expectations, the experience of disappointment at failing to meet those expectations, and subsequent disillusionment [11]. Organisational cynicism results in real consequences such as emotional fatigue, burnout, lower organisational commitment, and lower intention to engage in organisational citizen-ship behaviour [1] [12] [32] [62]. Therefore, experience has a moderating effect on the openness to change. Therefore, the following hypothesis is stated:

Hypothesis 1 : Openness to change has no influence on a person’s behavioural intention.

For decades, surveys in enterprise have shown that older employees are often associated with the stereo-type of less open to change. By 1953, for instance, Tuckman and Lorge had already found older people to be subjected to the prejudice from young graduate students in that they were less adaptable to changing environments [116]. Similar findings of managers’ and employers’ age versus openness to change stereotypes have been reported in more recent studies in different cultural contexts [24] [98] [127] [118]. On the other hand, Mirvis and Hall [47] argued that “there is no physiological and scant psychological evidence that ageing is in any way related to personal adaptability and resistance to change” (p. 285). Recent empirical results have supported this assertion by reporting no significant effects between self-reported individual adaptability and age [83] as well as between age and self-reported acceptance of organisational change [59]. Furthermore, in research on personal initiative and pro activity, no effects or even positive relationships were found with regard to age [119] [126] [65].

Given the contradicting theoretical arguments, the limited number of empirical studies and the mixed results, it is clear there is a relationship between age and openness to change, however it is not clear whether age has a positive or negative effect to the openness to change, therefore the following hypothesis is stated: Hypothesis 2 : The influence of a person’s openness to change on behavioural intention is not moderated

by age.

Gender related differences in responses towards new technologies have been acknowledged by literature. Some literature states that men have more positive attitudes towards new technologies than women [66] [103]. Other literature states that women have more positive attitudes towards new technologies than men [35] [111] [106]. Because of the mixed results, there is a reasonable chance there could difference in openness to change between the genders. Therefore the following hypothesis is stated:

Hypothesis 3 : The influence of a person’s openness to change on behavioural intention is not moderated by gender.

Literature states that an employees habits can be influenced by the culture in an organisation. Because there will be different culture within an organisation with a high clockspeed than in an organisation with a low clockspeed, clockspeed has an moderating effect on the openness to change. Because it could be expected employees who work in an high clockspeed industry will be more used to change, the following hypothesis is stated:

Hypothesis 4 : The influence of a person’s openness to change on behavioural intention is not moderated by clockspeed.

(12)

3 Methodology

Now the hypotheses are introduced and motivated, the hypotheses need to be tested by conducting a survey. This section provides details of the survey, such as the location of the survey, survey items, and measurement scales.

User acceptance of many different technologies has been tested using UTAUT, such as online learning [99], bulletin boards [75], enterprise systems [124], and desktop computing [7]. Table 1 provides an overview of a sample of studies that used UTAUT to test user acceptance of IT related topics and how they tested user acceptance using UTAUT. Study number one in table 1 is a study conducted by Venkatesh et al. to validate UTAUT. Because most studies in this sample used the same methodology as Venkatesh et al [124] did, the same methodology as Venkatesh et al [124] used will be used to validate the proposed model.

# Topic Method Measurement scale N Respondents Ref 1 Enterprise systems Survey 7-point Likert scale 215 Employees [124] 2 Online learning Survey 5-point Likert scale 102 Students [99] 3 Bulletin boards Survey 5-point Likert scale 132 Students [75] 4 Desktop computing Survey 7-point Likert scale 722 Employees [7] 5 Family dispute resolution system Survey 7-point Likert scale 127 Employees [23] 6 Use of internet technologies Survey 7-point Likert scale 102 Employees [46] 7 E-learning Survey 7-point Likert scale 261 Employees [131] 8 E-government Survey 7-point Likert scale 229 Employees [133] 9 ERP system Survey 5-point Likert scale 271 Employees [88] 10 Smart boards Survey 7-point Likert scale 68 Employees [94] 11 Healthcare Information System Survey 7-point Likert scale 315 Students [56] 12 Online information services Survey 5-point Likert scale 121 Students [84] 13 RFID Survey 5-point Likert scale 252 Employees [25] 14 Web-based training system Survey 7-point Likert scale 290 Employees [10] 15 Web-based data sharing system Survey 7-point Likert scale 282 Employees [120]

Table 1: Details of the sample of studies that used UTAUT to test user acceptance

This study was conducted in The Netherlands in the context of organisational flexibility of employees. An anonymous questionnaire has been created in English. As the questionnaire was administered in Dutch, the language used predominantly by the local residents in The Netherlands, the questionnaire was translated from English to Dutch and then back to English to ensure translation equivalence [18]. The methodology used for this research is based on the methodology used by Venkatesh et al. in 2003 [124] to validate UTAUT. Venkatesh et al. [124] conducted longitudinal field studies at four organisations among individuals being introduced to a new technology in the workplace. By conducting field studies at multiple organisations, Venkatesh et al. [124] tried to gain consistent results of the effects of his model in business from these field studies. They administered a questionnaire containing items measuring constructs from all eight models at three points in time: post training, one month after implementation, and three months after implementation. The questions used in the UTAUT survey were reused and questions regarding openness to change and clockspeed were added. By using the UTAUT questions, a baseline can be established and the questions about openness to change and clockspeed can be used to test hypotheses. Personal questions about age, gender and the sector in which the respondent works were asked at the end of the questionnaire to minimise the risk of stereotype threat [102] [54]. The performance expectancy, effort expectancy, social influence, facilitating conditions and habit questions used in the survey are the same questions as used by Venkatesh et al. [124]. The questions about hedionic motivation are similar to the questions used by Venkatesh et al. [125]. The personal questions and questions about openness to change were created to validate the proposed model.

(13)

3.1 Survey statements

The surveys consisted of statements. The following statements were presented during the conducted surveys. Performance expectancy

PE1 I would find the system useful in my job.

PE2 Using the system enables me to accomplish tasks more quickly. PE3 Using the system increases my productivity.

PE4 If I use the system, I will increase my chances of getting a raise. Effort expectancy

EE1 My interaction with the system would be clear and understandable. EE2 It would be easy for me to become skillful at using the system. EE3 I would find the system easy to use.

EE4 Learning to operate the system is easy for me. Social influence

SI1 People who influence my behaviour think that I should use the system. SI2 People who are important to me think that I should use the system.

SI3 The senior management of this business has been helpful in the use of the system. SI4 In general, the organisation has supported the use of the system.

Facilitating conditions

FC1 I have the resources necessary to use the system. FC2 I have the knowledge necessary to use the system. FC3 The system is not compatible with other systems I use.

FC4 A specific person (or group) is available for assistance with system difficulties. Openness to change

OC1 Change within organisational processes is good. OC2 I am open to changes in the way that I work. OC3 I adjust quickly to me when I have to do other work. OC4 I am negative towards changes in the organisation. Behavioural intention

BI1 I intend to continue using the system in the future. BI2 I will always try to use the system whenever possible. BI3 I plan to continue to use the system frequently. Clockspeed

CL1 I work at a dynamic and flexible company.

CL2 The company where I work expects from me that I take a positive view to change. CL3 The company where I work innovates regularly.

Personal

PI1 In which sector do you work? PI2 What is your age?

PI3 What is your gender?

Many studies used UTAUT to gain insight into user acceptance of IT related topics used the Likert scale. Table 1 provides an overview of a sample of these studies. In this sample the five and seven point Likert scale was frequently used. Because this research is based on the methodology used by Venkatesh et al. in 2003 [124] to validate UTAUT, the 7-point Likert scale is used as grading scale, with the anchors

(14)

being “strongly disagree” and “strongly agree.”. The following subsection will provide more information about the Likert scale.

The personal questions were not measured using the Likert scale. For the sector question, the respondent could choose a sector from a list consisting sectors from the Global Industry Classification Standard (GICS) [58]. In 1999, Standard & Poor’s and MSCI Barra jointly developed the GICS to establish a global standard for categorising companies into sectors and industries. GICS was developed in response to the global fi-nancial community’s need for one complete, consistent set of global sector and industry definitions, thereby enabling asset owners, asset managers and investment research specialists to make seamless company, sector and industry comparisons across countries, regions, and globally. GICS has become an industry model widely recognised by market participants worldwide [58]. It sets a foundation for the creation of replicable, custom-tailored portfolios and enables meaningful comparisons of sectors and industries globally [58]. The GICS classification system currently consists of the following 10 sectors: Consumer Discretionary, Consumer Staples, Energy, Financials, Health Care, Industrials, Information Technology, Materials, Telecommunica-tion Services, Utilities. Each sector was given a number, corresponding to the following sector:

1 Consumer Discretionary 2 Consumer Staples 3 Energy 4 Financials 5 Health Care 6 Industrials 7 Information Technology 8 Materials 9 Telecommunication Services 10 Utilities

Age was measured in years. Gender was coded using a 0 or 1 dummy variable, where 0 represented women.

Before the results of this survey will be presented, the target population for this questionnaire will be covered in the following subsection.

3.2 Respondents

The target population for this questionnaire were people who work within an organisation. In order to make the questionnaire easily accessible, the online questionnaire was created in Google Forms and the questionnaire was anonymous. Google Forms provides tools that made it possible for respondents to fill in the questionnaire not only on computers but also on mobile devices, such as smart phones. Because the survey was conducted anonymously, no steps have been taken to prevent people from completing the survey multiple times. After the questionnaire was created in Google Forms and published, it was important to find as many respondents in two weeks. Therefore, a link to the online questionnaire was spread threw social media, such as Facebook and LinkedIn. Because after two weeks there were not enough respondents, several companies in Amsterdam area were visited with a request to ask employees to complete this survey. A few days after the company visits, a sufficient number respondents had completed the survey.

Based on the information about the stated hypothesis, used methodology, and the possible respondents, the third sub-question has been answered. In the following section, the results of the conducted survey will be presented.

(15)

4 Results

In this section the results of the conducted survey will be covered in this section. However, before conducting any tests the data from the conducted survey will be summarised. Hereafter, there will be found out which test can be performed to confirm or reject the stated hypotheses before tests that will confirm or reject the null hypotheses will be conducted.

In total 217 Dutch respondents filled in the survey, 92 women and 125 men. The youngest respondent was 18 and the oldest 68. The SPSS Statistics software package has been used for all statistical analysis. Table 2 provides additional information about the respondents. As in every research, it is important to find out whether data is reliable. In the methodology section already stated how the survey was created, and conducted. Every respondent was given the same set of questions. A summary of the respondents answers on the survey is presented in table 3.

Respondents (n) 217 Men [n(%)] 125(57.6) Women [n(%) 92(42.4) Consumer Discretionary [n(%)] 29(13.4) Consumer Staples [n(%)] 21(9.7) Energy [n(%)] 3(1.4) Financials [n(%)] 37(17.1) Health Care [n(%)] 43(19.8) Industrials [n(%)] 53(24.4) Information Technology [n(%)] 12(5.5) Materials [n(%)] 0(0.0) Telecommunication Services [n(%)] 1(0.5) Utilities [n(%)] 18(8.3) Minimum age (year) 18 Maximum age (year) 68 Average age (year) 41.06 Standard deviation (year) 13.38 Table 2: Details about the survey participants.

(16)

N Mean Median Mode Min Max 1 2 3 4 5 6 7 PE1 217 5.43 6 6 1 7 4 6 15 20 38 91 43 PE2 217 5.17 6 6 1 7 6 4 14 36 48 82 27 PE3 217 5.02 5 6 1 6 5 10 11 34 66 71 20 PE4 217 2.43 2 1 1 7 90 38 23 46 11 9 0 EE1 217 4.58 5 5 1 7 5 14 27 46 65 50 10 EE2 217 5.00 5 6 1 7 3 13 21 33 50 67 30 EE3 217 4.72 5 6 1 7 5 12 28 39 58 59 16 EE4 217 5.02 5 6 1 7 5 9 25 29 50 68 31 SI1 217 4.57 5 6 1 7 13 16 18 56 35 60 19 SI2 217 4.62 5 4 1 7 10 9 23 64 35 57 19 SI3 217 4.16 4 3 1 7 14 21 44 42 42 41 13 SI4 217 4.88 5 6 1 7 7 11 27 31 48 65 28 FC1 217 5.12 5 6 1 7 3 9 23 28 47 75 32 FC2 217 5.02 5 6 1 7 4 16 11 34 51 77 24 FC3 217 3.81 4 4 1 7 15 41 38 50 32 31 10 FC4 217 5.01 5 6 1 7 8 11 23 24 51 62 38 OC1 217 5.32 6 6 1 7 6 7 8 28 48 81 39 OC2 217 5.70 6 6 1 7 3 3 11 11 44 84 61 OC3 217 5.37 6 6 1 7 3 5 16 20 53 81 39 OC4 217 2.11 2 2 1 6 71 96 23 15 7 5 0 BI1 217 5.33 6 6 1 7 6 7 11 27 45 76 45 BI2 217 5.28 6 6 1 7 4 9 14 25 52 70 43 BI3 217 5.29 6 6 1 7 5 6 13 29 48 75 41 Note:

PE: Performance Expectancy; EE: Effort Expectancy; SI: Social Influence; FC: Facilitating Conditions; OC: Openness to Change; BI: Behavioural Intention.

Table 3: Details of the survey answers.

The goal of the survey was to provide sufficient information to confirm or reject the null hypotheses. When looked at the content aspect, for the first hypothesis the survey provided data by the answers on the openness to change related questions and the answers to the behavioural intention questions. For the second hypothesis, the respondents age has been added in order to gain insight in whether age moderates the influence openness to change has on behavioural intention. For the third and last hypothesis, clockspeed related questions have been added to the survey to gain more insights in whether an organisations clockspeed moderates the influence openness to change has on behavioural intention.

Because reliability of the data is important to test before conducting any experiments, Cronbach’s alpha has been used to test the reliability. Cronbach’s alpha is used to estimate reliability with little consideration of the assumptions required for the sample coefficient to be accurate [107]. Although Cronbach’s alpha does not assume normality, it may be sensitive to test score distributions. Literature recommends [107], a sample sizes of at least 100 when dealing with not normal distributed data. Because this survey had 217 respondents, Cronbach’s alpha can be used to estimate reliability. A value of 0.70 or greater suggests that the scales used in tests are reliable [16]. The Cronbach’s alpha of all questions related to the constructs was 0.93. Cronbach’s alpha has also been calculated per construct. After excluding PE4, FC3, and BI4 all values were greater than 0.70. Table 4 provides the Chronbach’s alpha for the constructs. Based on these calculations, the scales used in the test were reliable.

(17)

Cronbach’s Alpha PE EE SI FC OC BI N 217 217 217 217 217 217 Alpha 0.89 0.90 0.79 0.77 0.86 0.92 Note:

PE: Performance Expectancy; EE: Effort Expectancy; SI: Social Influence; FC: Facilitating Conditions; OC: Openness to Change; BI: Behavioural Intention; AGE: Age; GDR: Gender.

Table 4: The Cronbach’s alpha of the constructs.

Before any tests can be conducted, it is important to enquire whether parametric or non-parametric tests should be performed. Non-parametric tests can generally be performed easier and quicker than parametric tests because calculations of non-parametric tests are generally easier to perform and apply [69]. Litera-ture also states [69] non-parametric tests are relatively robust and can effectively be used for determining relationships and significance of differences using behavioural research methods. However, parametric tests are more powerful than non-parametric tests [74]. Parametric tests can also deal with continuous variables while non-parametric tests generally deals with discrete variables [74]. Although both tests have both their advantages and disadvantages, there is a preference for parametric tests because non-parametric methods may lack power as compared with more traditional approaches [128]. However, whether parametric tests can be performed depends on the data. Parametric tests assume that the populations from which samples are drawn have specific characteristics, and that samples are drawn under certain conditions. Therefore, it must be tested whether the data covers the parametric assumptions.

The first assumption covered is the measurement scale. The Likert scale was used for almost all questions in conducted survey, except for the personal information questions. The Likert scale is named after the man who developed the technique in 1932 and published the original report in the Archives of Psychology, Dr. Rensis Likert, a sociologist at the University of Michigan [70]. His goal was to create a scientific way of measuring psychological attitudes [70]. He was looking for a method which would create attitude measures that could be interpreted reasonably as measurements on a specific metric scale, such as degrees Celsius is considered a true measurement scale. The technique presents respondents with a number of statements, and for each of the statements respondents are asked how strong, they agree or disagree, by selecting one of the five positions on the five-point scale. Each statement would represent a different aspects of the same attitude [70]. The full application of the Likert scale is the sum of scores, this provides an overall attitudinal score for each individual. Today’s surveys that covers user acceptance usually include multiple topics by which scores are calculated [84] [25]. Both of the ends of the scale are often increased to create a seven-point scale by adding the term “very” to the beginning and the end of the five-point scale. The seven-point scale increases the reliability of the scale [70].

Although the Likert scale is one of the most popular scales [9], many argue it has it shortcomings [132]. Respondents who have no knowledge about a statement usually rate the middle value, which is defined as a neutral position, as neither agree nor disagree, because there is no category for respondents who have no knowledge about the topic [8]. Therefore, researchers do not know whether the respondent had a honest neutral stand, an uncaring attitude or insufficient knowledge on the topic. When a respondent rates the middle value for all the questions the problem increases, a researcher should argue whether all the responses can be considered to be valid responses [8]. Other state [129] that the semantic meanings and positions of the scale labels influence respondents’ ratings. They also may have different levels of feelings, attitude and understanding to word choices and therefore they will feel and think differently. However, they rate it the same value. Therefore, the number of categories assigned to an item in the scale may not be appropriate [37] [81]. Scherpenzeel and Saris [101] argued that there is no optimal number of response alternatives. There is also the discussion whether the Likert scale measures should be treated as ordinal or interval data. Literature states the Likert scale falls within the ordinal level of measurement [89] [15] [48] [37] [81].

(18)

Therefore, the mean and standard deviation are not appropriate even though authors frequently describe their data using these statistics. The response categories have a rank order, however the intervals between values should not be presumed equal. Although, Blaikie [15] states researchers frequently assume they are, Cohen et al. [27] states that it is illegitimate to conclude the intensity of feeling between strongly disagree and disagree is equivalent to the intensity of feeling between other consecutive categories on the Likert scale. Methodological and statistical literature states that for ordinal data someone should use the median or mode as the measure of central tendency [26], because the arithmetic manipulations required to calculate the mean and standard deviation are inappropriate for ordinal data because the numbers usually represent verbal statements [15] [26] [3]. Furthermore, ordinal data may be described using frequencies or percentages of response in each category [15]. Literature also advises non-parametric tests should be conducted, such as chisquared, Spearman’s Rho, or the Mann–Whitney U-test, because parametric tests demand interval or ratio level data [89] [26]. However, these rules are generally ignored by authors [60]. In an example stated by Jamieson [60], authors of two papers from 2003 and 2004 used Likert scales and described their data using means and standard deviations and performed parametric analyses such as ANOVA [100] [55]. This is consistent with observations of Blaikie [15] who state it has become common practice to assume that Likert-type categories consist of interval level measurement. This has also been done in the surveys in the papers mentioned in table 1. Generally, authors did not made statements about assumption of the interval status for Likert data. Therefore, it is not clear they were aware that some would consider this as illegitimate. Treating ordinal scales as interval scales has long been controversial and it still is [64]. While Kuzon Jr et al. [67] states that using parametric analysis for ordinal data is the first of the seven deadly sins of statistical analysis, Knapp [64] states that sample size and distribution are more important than level of measurement in determining whether it is appropriate to use parametric statistics. Even when it is accepted that it is valid to treat Likert scale data as interval data, it happens often that data sets generated with Likert scales have skewed or polarised distribution [60].

The second assumption is that the data is a random sampling from the defined population. The defined population in this case are workers who have experience with an ES. Table 2 provides details about the details about the survey participants. In order to gain a random sampling of working people, multiple organisations in multiple industries have been asked to complete the survey. However, because it was an anonymous survey, it is impossible to say who filled in the survey. Although the survey was anonymous, all respondents filled in their age, gender, and in which industry they worked. The minimum age of the respondents was 18, the maximum age was 68. The average age was 41 years. 57.6% of the respondents were men and 42.4% of the respondents were women. Employees working in 9 of the 10 industries filled in the survey.

The third assumption is the independence of the sample, in other words, did the respondents provided honest answers and was he or she not influenced by others. To increase chances of honest responses of the respondents, emphasis has been placed on the fact that the test was anonymous. To increase chances of honest answers from the respondents, personal questions about age, gender and the sector in which the respondent works were asked at the end of the questionnaire to minimise the risk of stereotype threat [102] [54].

The fourth and last assumption is the normally distributed population. The Kolmogorov-Smirnov Test is used to determine how likely it is that the sample came from a population that is normally distributed. Therefore, a Kolmogorov-Smirnov Normality Test has been conducted on the data. The results of this test are presented in table 5. As shown in table 5, all significant correlations were below 0.00. This suggests the data is not normally distributed.

(19)

Static Significance PE1 0.28 0.00 PE2 0.23 0.00 PE3 0.22 0.00 EE1 0.20 0.00 EE2 0.20 0.00 EE3 0.19 0.00 EE4 0.19 0.00 SI1 0.17 0.00 SI2 0.17 0.00 SI3 0.14 0.00 SI4 0.19 0.00 FC1 0.22 0.00 FC2 0.21 0.00 FC4 0.19 0.00 OC1 0.24 0.00 OC2 0.26 0.00 OC3 0.24 0.00 OC4 0.31 0.00 BI1 0.23 0.00 BI2 0.21 0.00 BI3 0.23 0.00 Note:

Table 5: Results of the Kolmogorov-Smirnov Normality Test.

In order to determine whether the data after transformation is normally distributed, the data has been transformed creating variables by adding up the respondents scores per construct. The Kolmogorov-Smirnov Normality Test has been conducted on the transformed data and presented in table 6. The results are in this case also all significant correlations were below 0.00. Therefore, the data used for this thesis is not normal distributed. Static Significance PETOTAL 0.16 0.00 EETOTAL 0.11 0.00 SITOTAL 0.09 0.00 FCTOTAL 0.12 0.00 OCTOTAL 0.18 0.00 BITOTAL 0.17 0.00 Note:

(20)

There has been tested whether the data covers the parametric assumptions. Therefore, it is now pos-sible to motivate whether parametric or non-parametric tests should be performed. The four parametric assumptions test clarified that the data is measured on an interval scale, the data is a random sampling of the defined population, the data is an independent sample, and the data is not normal distributed. Litera-ture states [69] non-parametric tests should be used in the behavioural sciences when there is no basis for assuming certain types of distributions.

Before covering the hypotheses, a graphical display of the proposed model will be made, including relevant correlations. This graphical display will help understand data related to the stated hypotheses. Figure 5 provides all correlations relevant to the stated hypotheses. How these correlations have been established will be covered in the following subsections. Figure 5 shows that there are only significant correlations established. This result indicates the presence of the relationships presented in the proposed model.

Figure 5: The proposed model including correlations relevant to the stated hypotheses.

4.1 Hypothesis 1

This subsection will cover the data relevant to reject or confirm the first null hypothesis. The first hypothesis stated in this thesis was the following:

H10: Openness to change has no influence on a person’s behavioural intention.

H11: Openness to change has influence on a person’s behavioural intention.

To find whether the null hypothesis will be rejected or confirmed, the relation between openness to change and behavioural intention should be tested. A correlation test provides more insights in the relation and correlation of constructs. Because the data is not normally distributed, a Spearman’s rank correlation test [80] has been conducted on all constructs that, according to the proposed model, influence behavioural intention. The sum of the scores of a respondent have been used per topic. The scores on questions PE4

(21)

and OC4 have been removed because these questions used reverse questioning. Table 7 shows correlations between the constructs. This shows significant correlations between all constructs. The highest correlation (0.68) is between Facilitating Conditions and Effort Expectancy. The lowest correlations (0.39) is between Social Influence and Performance Expectancy.

PE EE SI FC OC BI PE 1 EE 0.56** 1 SI 0.39** 0.51** 1 FC 0.42** 0.68** 0.63** 1 OC 0.48** 0.59** 0.48** 0.60** 1 BI 0.42** 0.48** 0.52** 0.52** 0.57** 1 Notes:

1. PE: Performance Expectancy; EE: Effort Expectancy; SI: Social Influence; FC: Facilitating Conditions; OC: Openness to Change; BI: Behavioural Intention.

2. **p <0.01.

Table 7: Results of the Spearman’s rank correlation test

When focused on the proposed extension and the hypothesis, the correlation between openness to change and behavioural intention was 0.57. This is a significant correlation, which suggests openness to change has influence on a person’s behavioural intention. Based on this result, H10 should be rejected. Venkatesh et

al. [124] already confirmed the the relations between the other constructs.

4.2 Hypothesis 2

The second hypothesis stated in this thesis was the following:

H20: The influence of a person’s openness to change on behavioural intention will not be moderated

by age.

H21: The influence of a person’s openness to change on behavioural intention will be moderated by

age.

To gain insight into the influence of age on the constructs, there will be a focus on the difference in relations between the constructs based on age. Because the data is not normally distributed, a Spearman’s rank correlation test [80] has been conducted on all constructs that, according to the proposed model, influence behavioural intention. The data was divided into two halves, based on age. The first group consists of respondents from 18 to 42 years, the second group consists of respondents from 43 to 68 years. The first group consisted of 111 respondents, the second group consisted of 106 respondents. Table 8 shows the correlations between the constructs. All correlations in this table are significant. Based on the results of the survey there are different correlations between the constructs among the two age groups.

(22)

AGE PE EE SI FC OC BI PE 1 1 2 1 EE 1 0.64** 1 2 0.48** 1 SI 1 0.52** 0.59** 1 2 0.24* 0.40** 1 FC 1 0.52** 0.72** 0.69** 1 2 0.31** 0.66** 0.57** 1 OC 1 0.54** 0.67** 0.60** 0.67** 1 2 0.43** 0.50** 0.36** 0.52** 1 BI 1 0.49** 0.51** 0.54** 0.59** 0.65** 1 2 0.36** 0.52** 0.56** 0.58** 0.47** 1 Notes:

2. Age group 1: 18-42 years; group 2: 43-68 years. 3. *p <0.05; **p <0.01.

Table 8: Results of the Spearman’s rank correlation test, split in two groups based on age

When focused on the correlation between behavioural intention and openness to change, there is a clear difference in correlation between the two groups. The first group had a correlation of 0.65, the second group had a correlation of 0.47. As shown in figure 5, these are both significant correlations. Because there is a difference between the two groups, where the first group (18-42 years) had a higher correlation than the second group (43-68 years), this would suggest the influence of a person’s openness to change on behavioural intention will be moderated by age. Based on this result, H20 should be rejected.

When looking at all correlations related to the constructs, in all constructs the respondents between 18 and 42 years of age have a higher correlation than the respondents between 43 and 68 years of age. This suggests younger employees are more positive towards a new ES than older employees. This is consistent with literature [116] [24] [98] [127] [118].

4.3 Hypothesis 3

The third hypothesis stated in this thesis was the following:

by gender.

gender.

To gain insight into the influence of gender on the constructs, there will be a focus on the difference in relations between the constructs based on gender. Because the data is not normally distributed, a Spearman’s rank correlation test [80] has been conducted on all constructs that, according to the proposed model, influence behavioural intention. The respondents are divided in two groups based on gender. The first group consists of women, the second group consists men. The first group consisted of 92 respondents, the second group consisted of 125 respondents. Table 9 shows the correlations between the constructs. All

(23)

correlations in this table are significant. Based on the results of the survey there are different correlations between the constructs among the two age groups.

AGE PE EE SI FC OC BI PE W 1 M 1 EE W 0.61** 1 M 0.52** 1 SI W 0.45** 0.61** 1 M 0.35** 0.42** 1 FC W 0.50** 0.78** 0.70** 1 M 0.34** 0.58** 0.55** 1 OC W 0.66** 0.68** 0.64** 0.71** 1 M 0.33** 0.51** 0.35** 0.51** 1 BI W 0.53** 0.59** 0.60** 0.66** 0.67** 1 M 0.34** 0.39** 0.46** 0.52** 0.49** 1 Notes:

2. W: Women; M: Men 3. **p <0.01.

Table 9: Results of the Spearman’s rank correlation test, split in two groups based on gender

When focused on the correlation between behavioural intention and openness to change, there is a clear difference in correlation between the two groups. Women had a correlation of 0.67, men had a correlation of 0.49. As shown in figure 5, these are both significant correlations. Because there is a difference between the two groups, women have a higher correlation than men, this would suggest the influence of a person’s openness to change on behavioural intention will be moderated by gender. Based on this result, H30should

be rejected.

When looking at all correlations for the constructs in table 9, in all constructs women have higher correlations than men. This would suggest that women are more positive towards a new ES than men. However, because of the contradict in literature, it would be premature to state that women in general are more positive towards a new ES than men.

4.4 Hypothesis 4

The fourth and final hypothesis stated in this thesis was the following:

by clockspeed.

clockspeed.

To gain insight into the influence of clockspeed on the constructs, there will be a focus on the difference in relations between the constructs based on clockspeed. Because the data is not normally distributed, a Spearman’s rank correlation test [80] has been conducted on all constructs that, according to the proposed model, influence behavioural intention. The respondents have been dived in two groups based on the total

(24)

scores of the questions related to clockspeed. The first group has an score of 16.00 or less for the sum of clockspeed related question. The second group has a score higher than 16.00 for the sum of clockspeed related questions. The first group consisted of 97 respondents, the second group consisted of 120 respondents. Table 10 provides an overview of the results.

AGE PE EE SI FC OC BI PE 1 1 2 1 EE 1 0.65** 1 2 0.41** 1 SI 1 0.58** 0.58** 1 2 0.12 0.38** 1 FC 1 0.58** 0.72** 0.65** 1 2 0.16 0.60** 0.48** 1 OC 1 0.69** 0.70** 0.55** 0.73** 1 2 0.24** 0.46** 0.30** 0.37** 1 BI 1 0.69** 0.57** 0.57** 0.56** 0.65** 1 2 0.07 0.35** 0.41** 0.52** 0.41** 1 Notes:

2. Group 1: Clockspeed =<16; group 2: Clockspeed >16. 3. **p <0.01; all other correlations are insignificant.

Table 10: Results of the Spearman’s rank correlation test, split in two groups based on clockspeed

Three of the correlations in the second group were insignificant, all other correlations were significant. There were insignificant correlations for social influence and performance expectancy, facilitating conditions and performance expectancy, and behavioural intention and performance expectancy. The first group had higher correlations for constructs than the second group. The first group has the highest correlation for openness to change and facilitating conditions, with a value of 0.73.The lowest correlation of the first group was for openness to change and social influence, with a value of 0.55. The highest correlation in the second group was for facilitating conditions and effort expectancy, with a value of 0.60. The lowest correlation in the second group was for behavioural intention and performance expectancy, with a value of 0.07.

When focused on the correlation for behavioural intention and openness to change, there is a clear difference in correlation between the two groups. The first group had a correlation of 0.65, the second group had a correlation of 0.41. As shown in figure 5, these are both significant correlations. There is a difference for the two groups in general, where the first group always had a higher correlation than the second, this would suggest the influence of a person’s openness to change on behavioural intention will be moderated by clockspeed. Therefore, based on this result, H40 should be rejected.

It is interesting to see that the respondents with a clockspeed =<16 had higher correlations between constructs than the respondents with a clockspeed >16. This suggest the opposite than expected based on literature, that employees who work in organisations with a low clockspeed are more positive towards a new ES than employees who work in organisations with a high clockspeed.

By providing details about the data from the conducted survey, perform statistical tests on this data, and rejecting the stated hypothesis, the analysis of the results is completed. The following section will provide an answer to the research question of this thesis and its limitations.

(25)

5 Conclusion, discussion, and limitations

Based on the information provided in this thesis, this section will conclude the findings, discuss them and covers limitations of the conducted research.

5.1 Conclusion

The following research question has been stated for this thesis:

To what extent does the inclusion of clockspeed as a variable improve prediction of an individuals acceptance of an Enterprise System?

To answer the stated research question, the data from the conducted survey has been analysed and tested in the previous section. The conducted tests have shown openness to change has a influence on a person’s behavioural intention. Because it had the highest correlation of the constructs with behavioural intention, it would suggest openness to change would predict a person’s behavioural intention the best.

The conducted survey provides information which proved the influence of openness to change on be-havioural intention was moderated by three moderators, age, gender and clockspeed. Based on the findings, the younger half of the respondents had a higher correlation between openness to change and behavioural intention than the older half. This suggests there is a higher probability of predicting the response of the younger half of the respondents to behavioural intention based on openness to change then of the older half of the respondents.

The second moderator, gender, has shown that women had a higher correlation between openness to change and behavioural intention than men. This suggest there is a higher probability of predicting the response of women to behavioural intention based on openness to change then of men.

The data provided by the added moderator in the proposed model, clockspeed, the respondents were divided in two groups. The first group had an clockspeed score of 16.00 or lower, the second group had a clockspeed score of more than 16.00. The first group group had a higher correlation than the second, which suggest that there is a higher probability of predicting the response of respondents with a lower score for clockspeed to behavioural intention based on openness to change then respondents with a higher score.

5.2 Discussion

In this subsection, other possible explanations for the findings in this thesis will be covered by retracing the reliability and validity of the conducted survey.

This thesis contributed to user acceptance research by examining the relation between the leading model which predicts an individuals acceptance of a system and clockspeed. Based on literature, UTAUT has been determined to be the leading model in an employee acceptance and use setting because it has been created based on the eight extant user acceptance models, including TAM, and it focuses on user acceptance in a business setting.

The theoretical contribution is in extending UTAUT with the clockspeed context. By doing so, UTAUT has been shifted its main focus from performance expectancy as the main driver of employees technology use intentions and behaviours. Based on a literature review, openness to change was added as a construct to UTAUT, which influences behavioural intention. Clockspeed has been added as an moderator which, together with age and gender moderated the influence openness to change has on behavioural intention. Literature states there are three facets of industry clockspeed [38] (product, process, and organisational), however, during this thesis there has not been tested which of the facets is the best moderator.

An individuals openness to change will depend on many aspects. Aspects that increase the openness to change produces greater organisational commitment, which is associated with favourable attitudes to

(26)

organisational change. While aspects which decrease the openness to change decrease organisational com-mitment. The collected data from the survey provided evidence openness to change has an influence on behavioural intention. Which suggests employees who have higher organisational commitment, will have a more positive attitude towards using a new ES.

Resistance to change has been considered as negative behaviour in this thesis. This is also generally the case in the business, where managers perceive resistance negatively, and the employees that resist are considered to be disobedient and obstacles organisations must overcome to achieve their goals. However, in some cases resistance of employees may be positive and useful for organisational change. Disagreements, insightful debates, and criticism may be intended to produce improved understanding and additional solu-tions. Literature states preconception that anyone who doubts the need for change has an attitude problem is incorrect because organisations will be more vulnerable to indiscriminate and ill-advised change [31]. Organisations should be aware that they might change the wrong thing or do it wrong way. Legitimate resistance might provide additional organisational change [40]. Employee resistance may force management to rethink or reevaluate a proposed change initiative. Literature states [31] resistance is a very effective, very powerful, very useful survival mechanism which can be used to help organisations select from all possible options the one that is most appropriate to the current situation.

The conducted survey confirmed the influence of openness to change and clockspeed as a moderator. However, the lower half of the respondents, which scored 16 or lower on the clockspeed related questions, had a higher correlation than the higher scoring half. Although both correlations were significant, this may suggest other aspects moderate the influence of openness to change on behavioural intention than the already confirmed moderators.

When looked at the consistency between the findings of this thesis and literature, there is consistency with findings of the introduces of UTAUT, Venkatesh et al. [124]. As shown in figure 5, all correlations between UTAUT’s original constructs and behavioural intention are significant and therefore consistent.

5.3 Limitations

Although this research provides an exploratory investigation on user acceptance differences in technology systems adoption under the working people context, there are several limitations that should be addressed. First of all, be careful to generalise the findings from this study to all employees who have experience with ES. While the sample used was a large-scale one with employees from several firms in the Netherlands, they may not definitely represent all Dutch employees. Future study may examine the findings with samples from other parts of the Netherlands.

In addition, since the clockspeed and openness to change related questions were created by the author himself, it is may be that in this study, certain degree of subjectivity can be found. In fact, it would have been more objective if the questionnaire was created by two or more people. Therefore, when additional clockspeed research will be conducted, the survey should be created by two ore more people to increase the objectivity.

It may be interesting to extend the research into the influence of culture on the wiliness of users to accept an ES. Because there are other work values among other cultures, it may be that employees from other cultures response differently to an ES [61] [51] [52] [50] [49] [93]. There are many studies [115] [130] [93] that argue that the individualism-collectivism continuum may be the best way to measure values differences between cultures. Besides global differences, it may be interesting to measure difference in the willingness to acceptance between countries within the same continent, for example the willingness to accept an ES in the UK and Greece.

Other aspects may also have influence on acceptance of an ES, e.g. workload. However, these aspects were not covered during this research because of the scope of this research. Therefore, there is a possibility there are other aspects that significantly influence behavioural intention which are not stated in this thesis. Followup research should aim on other aspect that influences behavioural intention.