Explaining the Acceptance of Municipal eServices in
The Netherlands: An Empirical Investigation
By
Michel Bernsen
University of Groningen, The Netherlands
Faculty of Economics & Business
Department of Business & ICT
Supervisor:
Dr. D. Seo
2
Explaining the Acceptance of Municipal eServices in The
Netherlands: An Empirical Investigation
Abstract
eGovernment becomes more and more prioritized by national and local governments. Municipalities invest heavily in supplying eServices to their citizens. Nationally, on average only 24% of all citizens in
the Netherlands use the eServices offered by the municipality they live in. This research identified factors that influence a citizen’s intention to use general municipal eServices. Based on pilot interviews with citizens and literature a research model was hypothesized. From two municipalities (Emmen and Groningen), 337 surveys on users and nonusers of municipal eServices were collected and compared to each other and tested by applying Partial Least Squares tests and Dunnett’s
two-sided t-test. It was found that nonusers and users perceive individual factors fundamentally different, while users (and nonusers) across municipalities perceive these factors quite similar. PLS results show no significant pattern in what individual factors are the most important factors that drive acceptance. However, in all four studies and in the pilot interviews perceived usefulness and perceived behavioral control were found to directly determine one’s intention to use municipal
eServices. Investments of municipalities that positively influence these factors are considered safe to increase the use of municipal eServices.
Keywords: eGovernment services, acceptance, citizen’s perception, intention to use, users and
3
Table of contents
1 Introduction ... 4 2 Literature review ... 6 2.1 IS literature ... 6 2.2 eGovernment literature ... 72.3 Dutch eGovernment adoption literature ... 8
3 Pilot interviews ... 10
3.1 Findings... 12
3.2 Conclusion ... 13
4 Research model & hypotheses ... 14
5 Applied research method ... 24
5.1 Data collection ... 25 5.2 Measurement instrument ... 25 5.3 Pretest ... 27 5.4 Field survey ... 28 6 Results ... 30 6.1 Comparing constructs ... 30 6.2 Model testing ... 31 6.2.1 Scale validation ... 31 6.2.2 Hypotheses testing ... 32 6.2.3 Post-Hoc analysis ... 38 7 Discussion ... 40 7.1 Key findings ... 40 7.2 Theoretical contributions ... 44 7.3 Managerial implications ... 45 7.4 Limitations ... 46 8 Conclusions ... 46 8.1 Further research ... 47 References ... 49
Appendix A: Interview transcripts. ... 59
Appendix B: Measurement tool. ... 75
Appendix C: Multiple comparisons. ... 79
4
1
Introduction
In 1994, The Netherlands introduced the first action program for the development of Internet
services to provide their citizens with information that previously was provided as hardcopy brochures briefing information on different services and rules of the government. The government developed more programs to exploit the Internet as a tool for service and communication (eOverheid, 2011). Electronic service delivery is seen as one of the main pillars of modernization, besides legislative changes and new arrangements between national and local government levels
(IOS press, 2005). To realize this, the Modernizing Government Program was launched in December 2003. The main objective of the program is to make government and public services simpler, more effective and more efficient for the benefit of citizens and businesses, by focusing on core competences and reorganizing service delivery around customer needs (IOS press, 2005). Then, in
2006 municipalities and government (agencies) signed a commitment to improve and expand digital service providence and to decrease operational costs by using ICT (eOverheid, 2011). The succeeding action program for e-Government was the National Execution Program for Service and eGovernment over the period of 2008-2011. Again, the emphasis is on improving digital services to citizens and businesses and providing up to 65% of municipal services online, the so-called municipal eServices.
Hence, digitizing municipal services becomes increasingly prioritized and municipalities keep on digitizing more and more services.
The efforts and financial investments stimulated by action programs led to an increase in the quality of the eServices of many municipalities (Overheidsmonitor, 2011). Yet, the actual use lagged behind as demonstrated in the research of van Dijk et al. (2007). Their research showed an average use of
24% of municipal eServices, while the intention to use the e-Services is 77%. This is exorbitant low compared to the 56% use of national government eServices (Van Dijk et al., 2007). One unanimous municipality reported, for the purpose of this research, that up to 96% of all cases was processed at the traditional service counter in 2010; meaning a 4% exploitation of their municipal eServices. The municipality of Groningen reported that the average use of their eServices is approximately 30%.
Despite quality improvements and promotional efforts, these accountable managers find themselves unable to increase the use of municipal eServices significantly.
National government eServices range from municipality eServices to reporting tax income to social services like job search or utilizing social security services. A few examples of municipal eServices are a request for a birth certificate, making an appointment to apply for a passport, a notification of
5 known as the eLoket. Every municipality has realized its own eLoket (Overheidsmonitor, 2011). There are no mandatory rules imposed by the national government for designing an eLoket. So, municipalities are free to digitize whatever service they want. This results in mostly commonalities,
but also large differences in the services provided. For instance, it is possible to apply for a building permit using eLoket at a number of municipalities, but this eService is not offered by the municipality of Muiden. Some municipalities digitized twenty of their municipal services and some only four. Therefore, it is likely that every eLoket is perceived differently by citizens.
This research is focused on identifying factors that influence the intention to use municipality
eServices as discussed in the next chapter. Fishbein and Ajzen (1975) state that behavioral intention is the main determinant for actual behavior. Therefore, if municipalities can identify what shapes a citizen’s behavioral intention, they might in all probability discover methods to increase the actual usage of municipal eServices. Based on relevant theory and interviews a research model has been developed containing both enabling factors and inhibiting factors. Moreover, this paper presents
new insights on user and nonuser perceptions and if people’s perception of municipal eServices differ per municipality.
This paper contributes to extant literature in various ways. First, it postulates a clear distinction in user and nonuser perceptions and that these should be seen as distinct groups instead of one group for further research. Second, this paper brings new insights in the field of technology acceptance,
where several established theories and definitions are altered since eGovernment context requires a perspective switch. Third, research on Dutch eGovernment acceptance is scarce and the field remains relatively unexplored. This research explores new factors that explain the intention to use municipal eServices. For example, this research is the first of its kind that takes geographical distance
into account, where the assumption is that when people that live further away of city hall develop more intention to use the services and it increases the usefulness they perceive. Fourth, this paper holds a comparison of findings across municipalities and presents new insights in the perceptions of citizens from different municipalities.
To emphasize the contributions and importance of this study within existing research, a literature
review is presented in the next chapter. The succeeding chapter provides information about the pilot interviews that were conducted to discover the main issues for not using municipal eServices. This practical foundation provides theoretical leads to design the research model and formulate the hypotheses of this research in the next chapter. Then, the method is explained of how this research
6 contributions, managerial implication and research limitations. Further, the conclusions and recommendations for further research will complete this paper.
2
Literature review
In this section the Information System (IS) literature is reviewed that is related to IS acceptance with a particular focus on acceptance of eGovernment applications. The aim of the review is to provide insight in the conducted research and unexplored areas and which areas need attention for research.
2.1 IS literature
A popular stream of IS research has focused on identifying motives that result in IS acceptance. Research in this particular stream has yielded a number of antecedents of (intention to) use. Several theories developed based on psychological beliefs have been applied to investigate IS acceptance, the major ones being Technology Acceptance Model (TAM) (Davis et al., 1989) and Theory of Planned Behavior (TPB) (Ajzen, 1991).These theories are developed to investigate the demand side; the users
of the system. Another one is the Innovation Diffusion Theory (Rogers, 1995), which is a theory developed for the supply side. The Unified Theory of Acceptance and Use of Technology (UTAUT) is a theory based on these three theories and the Theory of Reasoned Action (Fishbein & Ajzen, 1975), the Motivational Model (Davis et al., 1992), a model that combined the Technology Acceptance
Model and the Theory of Planned Behavior called TAM-TPB (Taylor & Todd, 1995), the model of PC utilization (Thompson et al., 1994) and the Social Cognitive Theory (Bandura 1986; Compeau and Higgins, 1995) and led to a unified model with significant explanatory power.
The previous mentioned theories and models are applied, validated and extended in numerous studies. A distinction in research can be made for pre-adoption, initial adoption and post-adoption
(or IS continuance). Especially the TAM is extended with new determinants many times in the field of initial adoption. Here, it is assumed that motives and beliefs of users to use a particular system would also apply to potential users (Davis et al., 1989). The second most popular stream is post-adoption. This stream is made explicit by new proposed models that explain IS continuance, such as DeLone and McLean’s model of IS success (2003), Technology Continuance Theory (Liao, Pavlia & Chen, 2009)
and the Expectation Confirmation Model of Bhattacherjee (2001).
Yet, pre-adoption and the lagging acceptance behavior of nonusers, explicitly, are little investigated by means of the mentioned acceptance theories. Reason for this might be the high frequency of validation and extension of TAM or TPB in business environments, where the target group has many
7 characteristics and stimuli, for instance greater voluntary use or less familiarization with technology by some. This idea is validated by Ramayah et al. (2002). They found huge differences in the perceptions of users and nonusers regarding a particular online banking technology. This stresses the
idea that users cannot explain the intentions of nonusers.
2.2 eGovernment literature
Research on eGovernment can be roughly categorized into three streams, pinpointed by Heeks (in press). The first stream, technological determinism, assumes that the introduction of eGovernment is mainly influenced by technological features of IT. The second stream of research, social determinism,
makes the assumption that social structures are the main determinants of the success of eGovernment. The third stream is a combination of both elements. Research conducted in the socio-technical determinism stream investigates the interaction of IT and people. Most eGovernment research has been conducted in the third stream.
As discussed before, there are some theories to use as a basis for research in the IS adoption stream.
Here, a distinction has been made between theory for demand and for supply. The same distinction can be seen in eGovernment literature. Much research is conducted on the supply side to measure quality of the supplied eServices (e-Overheid, 2010; Ernst & Young Advisory, 2009; Ernst & Young Advisory, 2010; Kunstelj et al. 2006; Thomas & Streib, 2003; Van Dijk, 2003) and get insight in the percentage of use. Olpher and Damodaran (2007) claim that eGovernment is currently characterized
by a techno centric approach with minimal engagement of citizens. Many international researchers (Anthopoulos et al., 2007; Evans & Yen, 2006; Gouscos et al., 2007; Luk, 2009; Olphert & Damodaran, 2007) have developed frameworks or models that should help governments in successfully supply IT
to citizens.
In most recent years, there’s a worldwide trend to explore the demand side and to find people’s motives to start using eGovernment. Delone and McLean (1992; 2002; 2003) developed a model for that explains IS success, where information quality, service quality and system quality determine user satisfaction and these four factors together explain the intention to use IS. According to the model, intention to use becomes actual use which influences the user’s satisfaction and net benefits. Net
benefits in turn influence intention to use and user satisfaction. Wang and Liao (2008) tested the model in an eGovernment context, but ignored the factor intention to use in their model. They validated the model, except for the relationship of system quality with use. More importantly, only 21% of the variance for use could be explained. This means that this model is considered weak for
8 As said before, intention to use is the main determinative for actual use. In different countries the acceptance of eGovernment services was empirically investigated. Researchers (Carter & Bélanger, 2005; Chang et al., 2005; Hung et al., 2006; Yaghoubi et al., 2010) mainly investigated antecedents of
intention to use based on the theories of TAM and TPB. This resulted in a variety of results which are overlapping and complementary. Yet, there are many unexplored theories and perspectives left. None of the empirical studies that investigate acceptance makes a clear distinction between users and nonusers, if acceptance among nonusers is even investigated.
2.3 Dutch eGovernment adoption literature
Van Dijk and van Deursen pleaded in 2006 for a perspective switch from supply to demand. Yet, only few researchers (Horst et al., 2007; van Dijk et al., 2006; van Dijk et al., 2007; van Dijk et al., 2008) answered this call, by examining municipality e-Services from the user’s point of view. Horst et al. (2007) and Van Dijk et al. (2008) used theoretical frameworks like combined TAM-TPB and UTAUT,
respectively. These frameworks provide a theoretical base to identifying users’ motives.
Horst et al. (2007) researched the acceptance of municipality eServices among Dutch citizens. They found that trust has a direct relationship with perceived usefulness and the risk perception of one has a direct effect on one’s trust. Further, they found that personal experience with municipality eServices is positively correlated with perceived usefulness. Remarkably, they found no correlation between perceived behavioral control and intention to use. Yet, this is a strong relationship in TPB
and verified by numerous researchers (Hung et al., 2006; Hsu & Chiu, 2004; Wu & Chen, 2005; Yaghoubi et al., 2010). Perceived behavioral control is defined as one's perception of the difficulty of performing a behavior (Ajzen, 1991). In other words, one might want or intend to do something, but there might be something keeping the person from performing the behavior. Moreover, they could
not verify the relationship of subjective norm and risk perception with intention to adopt.
Van Dijk et al. (2008) examined the acceptance of government Internet services, ranging from simple information websites to online tax service and municipal eServices. Hence, results are not entirely applicable to this research. Van Dijk et al. (2008) tested an extended UTAUT model. However, they found no or little evidence (effort expectancy determines intention to use) to support the original
posited relationships of UTAUT. The relationships of constructs that extended the original UTAUT were validated. Digital media preference, access and experience are correlated with intention to use. Moreover, they found that actual government Internet service supply is a precondition for people to develop the intention to use these services. Government Internet service supply does not lead to a
9 between digital media access and knowledge of the existence of eServices. Meaning, people that are able to access the Internet are more likely to be aware of the existence of eServices.
Reviewing the results of these studies, both research teams did not find significant evidence to support core relationships from original models. The results present no understanding in what
motives are essential for people to use municipality eServices. Further, the results from the analytical tests do not provide directions on how to increase the use of municipality eServices. Therefore, it is important to gain a deeper understanding on what triggers or drives people to use municipal eServices. A goal in the most recent action program for advancing eGovernment in whole Europe is
(1) to examine past relevant studies and whether or not these are applicable and valuable and (2) to put the citizen in the center of attention (European Commission, 2011). Hence, research within this field to support the goals would be of value. The call for research in Dutch e-Government and the lagging use of it, results in the next research question:
What are the underlying motives and incentives that shape people’s intention to use eServices
of Dutch municipalities?
As mentioned before, Ramayah et al. (2002) measured huge differences in the perceptions of users and nonusers towards online banking. For the sake of this research, users are defined as ‘citizens that have used municipal eServices and are able to recall this experience’. Nonusers are the people that
do not meet these two conditions. Ramayah et al. (2002) found that TAM could explain 39% of the variance for intention to use among users, but only 5.2% of the variance of intention to use for nonusers. Therefore, this research doubts an equal treatment for users and nonusers. More specifically, it’s expected to find fundamental differences between the perceptions and key motives for users and nonusers and therefore it would take different kind of action to transform nonusers
into users than maintaining users. This is argued since nonusers have to rely on expectations and users rely on experience. Expectations are not always formed by deliberation, but may also be developed based on experience with alternative technology or on imagination. Hence, users’ and nonusers’ judgment and perceptions may differ heavily from each other. This results in the following
sub question:
Are there fundamental differences in the perceptions of users and nonusers regarding municipal eServices and in the factors that determine their intention to use?
10 which are potentially associated with desired goal states and which thus stimulate goal-directed behavior (Beckmann & Heckhausen, 2008; McClelland, 1985; Schmalt, 1996; Schneider & Schmalt, 2000; Schüler, 2010). Motives can be stimulated intrinsically and extrinsically. In case of the latter, strengthening the incentives for (non)users would possibly increase a person’s intention to use municipality e-Services. For some outcomes the (local) government could perhaps play an essential role, which could lead to a more widely use and acceptance of municipality eServices. Therefore, it is of importance to identify constituent drivers and also to investigate which are of key importance to future use. The next sub question will be investigated:
What determinants of intention to use municipal eServices are most important to users and nonusers?
These three research questions cover a wide range of important research. However, this study focuses on municipal eServices, which are not uniformly developed among municipalities. So, one might argue that the findings in this research might not be applicable to other municipalities. It is
therefore of importance to determine the applicability of the findings. The following sub question is formulated:
To what extent do the findings represent the reality and are these applicable for other municipalities?
3
Pilot interviews
Ultimately, this research would change the low usage ratio. Change management literature prescribes examining the areas or individuals that are unknown or resistant (Luecke, 2003). Hence, qualitative research was conducted among nonusers to identify the main reasons for them not using
municipal eServices. A semi-structured interview technique was applied, which started with a few specific questions and then follows the individual’s tangents of thought with interviewer probes (Cooper & Schindler, 2008). Questions in this interview where open-ended and intended to reveal the main reason for not using municipality eServices. Other questions where focused on the influence of their own ability, the influence of their environment and their perception of the eLoket
and its offered services. The questions in this interview where based on grounded theory, being TAM and TPB, as discussed in the literature review section. Some questions were intended to discuss their attitude towards the technology. The interview transcripts, written in Dutch, can be found in
11 Interviewees where selected on four criteria. The first condition was that they do business with the municipality regularly. To define regularly, one should do business with the municipality once a year. This is based on research conducted by Van Duivenboden and Lips (2001) where they conclude that
citizens contact the municipality on an average of 2.6 times per year. In some cases, one’s life partner never deals with the government. These individuals are not suited as participants in this research.
Second, the interviewees must be aware of the existence of eLoket. If people are not aware of an existing technology, a lack of awareness would be their main reason for not using the system. A 5-year old study conducted in The Netherlands concludes that awareness of the existence of municipal
eServices is a precondition of actual use (Van Dijk et al., 2006). In 2004, Bongers et al. found that 50% of the citizens were awareness of the existence of these services. Later research conducted in Rotterdam revealed a higher percentage of 57% (De Graaf, 2005). One can only assume this percentage grew over the last few years. Notwithstanding, the latter research found that 22% of the respondents that were aware of the existence of eGovernment service actually used it. Hence,
awareness would not be the main driver for using or not using eLoket or its services.
Third, to have a broad distribution of civil opinions, interviews were conducted in different villages and municipalities. Five interviews were conducted in Emmen, four in Groningen and one in Oosstellingswerf. Finally, the participants must own a computer and have an Internet connection. The group of interviewees is heterogeneous, allowing this research to have a broad perspective input
for further theory building. The interviews were conducted face-to-face. Table 1 presents demographic information of the participants.
Interviewee Municipality Age Gender Distance to city hall Job title
1 Emmen 49 Female 15 km Production employee
2 Ooststellingswerf 34 Female 7 km School teacher
3 Emmen 52 Male 13 km Construction worker
4 Groningen 20 Female 4 km Housewife
5 Groningen 26 Male 2 km Dockyard employee
6 Groningen 21 Female 1 km Student
7 Emmen 49 Female 15 km Cleaning lady
8 Emmen 56 Female 6 km Housewife
9 Emmen 38 Female 10 km School teacher
10 Groningen 29 Male 5 km Student
12 The transcripts were analyzed according to the content analysis method and a thematic unit of measurement. The latter indicates the topics contained within texts (Cooper & Schindler, 2008). This represents a higher level of abstraction, which is in line with this research; the root of causes is
investigated in this research.
3.1 Findings
Some of the interviewees had more than one main reason for not using the technology. Four out of ten interviewees thought a lack of usefulness was their main reason for nonuse. Also, four out of ten claimed they thought they are unable to use, because they expect to have a lack of knowledge to
understand the online procedures (2 interviewees) or they expected to have too little computer skills to handle the functionality of municipal eServices. Two interviewees mentioned a lack of awareness would be their main reason. The interviewees were not fully informed of the possibilities of eLoket and mentioned it is not an active topic in their mind and therefore did not think of the technology when they had to deal with the government. From this, the factor habit can be inferred. Limayem et
al. (2007) defined habit within the IS context as ‘the extent to which people tend to use information systems automatically because of learning’. Reflecting the definition to this research, it is for these citizens a habit to go to city hall and they therefore do not think of alternatives like online municipal services. Two subjects indicated to be influenced by the opinion of others. The opinions or actual use of others simulate them in making their choice of using of not using the eServices themselves. One
participant claimed to consciously resist to the use of municipal eServices, because success of the system would cost people their jobs according to her.
Asking the subjects for secondary causes for not using the technology, one person believes she lacks computer skills to operate the system and one individual claimed he would value the ease of use of
the system. Six other interviewees thought they would have enough skills to operate the system without help from others. The opinions about having the necessary knowledge to understand the procedures were equally distributed: two expect to have enough knowledge and two expect not to have enough knowledge to understand these procedures. Generally, interviewees see the benefits of using municipality eServices offered by the eLoket. However, for five this cannot outweigh the
different disadvantages such as an effort to learn to operate the system or the uncertainty that data is processed correctly. Four respondents find the opinions and support of others important in making the decision to use or not to use the system. They expect it would influence their habits if eLoket is a subject matter in their personal environment, which it isn’t in their case. The interviews indicate that trust is not really an issue. Only one respondent claims not to trust the technology to work
13 interviewees expect no or little risk by using municipality eServices. Three interviewees regarded the risks of data distortion to be high and their uncertainty was related to their nonuse.
These findings are presented in table 2 to provide an overview. The findings are categorized on main reasons for not using, secondary reasons for not using and factors that influence their decision to use
or not to use the municipal eServices. The actual findings are listed together with the number of times they were mentioned by the interviewees.
Reason Indicated
Main reasons Lack of usefulness 4
Not able to: lack of computer skills 2
Not able to: lack of knowledge on the service 2
Not a habit 2
Opinion of others 2
Resistant to use 1
Secondary reasons Too much risk of data distortion 3
Not able to: lack of knowledge on the service 2 Not able to: lack of computer skills 1
Lack of trust 1
Influencers Opinion of others 2
Perceived disadvantages 4
Table 2. Overview findings interviews.
3.2 Conclusion
The perceived ability of using the system for some of the interviewees is low. This phenomenon is better known in IS literature as perceived behavioral control. Seven interviewees indicated that perceived behavioral control is determinative for their nonuse; four indicated to expect to have too little knowledge to understand the procedures and three people doubted their computer skills
heavily.
The second most important factor determining intention to use of the nonusers is subjective norm. They claim it is important for them to know what their environment thinks of the system and argue that they would be stimulated to use the system if their environment has a positive judgment.
14 has a few advantages. However, the presence of various perceived disadvantages outweighs the advantages for some.
Habit, ease of use and perceived risk are of equal importance in withholding people from using the system. Each factor was mentioned trice to be determinative. Only one interviewee claimed not to
trust municipal eServices and therefore rather goes to city hall.
Resistance is only mentioned twice as being the key factor for nonuse. However, it is for every interviewee implicitly applicable to a certain extent. Not using a system despite being aware of the
possibility to do so is a form of passive resistance, namely ignoring (Lapointe & Rivard, 2005).
Habit
Going to city hall is for some a natural way of doing business with the municipality and at the moment they claim to not think of possible alternatives. Despite their habit they might still have an intention to use, but when it comes to actual use their habit brings some of them to city hall instead of eLoket. In existing research, habit is established as a determinant of actual system use, which is in
some cases referred to as actual behavior (Bergeron et al., 1995; Limayem et al., 2007; Limayem & Hirt, 2003). As this research focuses on people’s intention to use, this finding will not be used in building a research model.
4
Research model & hypotheses
Based on the conclusions, segmentations of the obtained information and previous, sometimes arguable, findings on Dutch local eGovernment, a research model and hypotheses are developed. This research model, presented in figure 2, is supported by international eGovernment research using technology acceptance theories and findings from the pilot interviews discussed in the previous
chapter. The model is extended with original hypotheses discussed in the subsequent paragraphs.
The Theory of Reasoned Action (TRA) developed by Fishbein and Ajzen (1975) is a social psychology theory explaining one’s behavior. TRA posits that individual behavior is driven by behavioral intentions, where a behavioral intention is the consequence of an individual's attitude toward the behavior and subjective norms surrounding the performance of the behavior (Ajzen & Fishbein,
15 The technology acceptance model (TAM) is specialized for the prediction of IS adoption and is derived from TRA. TAM is applicable in the pre-adoption stage and in the repurchase stage (Bhattacherjee, 2001; Thong et al., 2006).TAM explains user acceptance of a technology based on
one’s perception of the technology (Davis, 1989; Davis et al., 1989). Davis et al. (1989) assert, “A key purpose of TAM is to provide a basis for tracing the impact of external factors on internal beliefs, attitudes and intentions”. As a pioneer researcher in this field, Davis argued that TAM is the most powerful and valid tool to analyze technology acceptance (Davis, 1989; Davis et al., 1989; Davis &
Venkatesh, 1996). TAM is widely accepted in the research field of IS adoption (Hsiao & Yang, 2011).
Initially, Davis (1986) added two predictors of attitude: perceived usefulness and perceived ease of use. Recalling the findings of the pilot interviews, usefulness and ease of use are important factors in having an accepted system. Perceived usefulness is defined as “the degree to which a person believes that using a particular system would enhance his or her job performance” and perceived
ease of use is defined as “the degree to which using the technology will be free of effort” (Davis, 1986; Davis, 1989). Moreover, Davis (1986) dropped subjective norms, the opinion of others, from his model arguing that subjective norms are context-driven. In his research, system usage was not likely to be influenced by social influences. This resulted in the original technology acceptance model
presented in figure 1.
Figure 1. Original Technology Acceptance Model. Source: Davis et al. (1989).
Through the years, TAM experienced a great deal of modifications, extensions and successors. The
first accepted modification of TAM was in 1989 by Davis et al. They eliminated the intervention of ‘attitude towards using’ in the original TAM. Numerous studies, conducted by for example Straub et al. (1997) and Gefen et al. (2003), validated that the constructs perceived usefulness and perceived
ease of use are directly correlated with intention to use.
16 2003), and intention to use. Within the context of this topic, the importance of perceived usefulness is emphasized by the interviewees that participated to this research. Therefore, this study hypothesizes a positive relationship of perceived usefulness with intention to use resulting from the
theory of TAM:
H1 A citizen’s perceived usefulness of municipal eServices has a positive relationship with its intention to use.
The successor of the Technology Acceptance Model is TAM2 as proposed by Venkatesh and Davis (2000). TAM2 ignores ‘attitude towards using’ and hypothesizes a direct correlation between
perceived usefulness and perceived ease of use and Intention to use. Furthermore, Venkatesh and Davis identified five antecedents of perceived usefulness, namely image, job relevance, output quality, result demonstrability and, the most salient ingredient of this model, subjective norm having a relationship with intention to use, perceived usefulness and image. From this the importance of perceived usefulness in determining one’s intention to use may be inferred. Aside from these five
factors, there are undoubtedly many others since the model explains only 40 to 60% of the variance for perceived usefulness. Addressing these (remaining) factors is not within the scope of this research. However, there is an interesting new relationship posed by Van Dijk et al. (2006).
Van Dijk et al. (2006) conclude from their study that geographical distances may encourage citizens
to use eGovernment services. Yet, they have not performed an empirical research to establish any relationship. A logical effect of an individual that has to travel a few kilometers to deal with local government is that it increases one’s perceived usefulness and, ultimately, one’s intention to use. However, economical reasons are not always enough motive for individuals to act. See, for example, heavy smokers with low income; it is far more economical to quit smoking, though, they continue.
So, the relationship is not obvious in every case and this makes it an interesting relationship to test. Relative advantage is identified as being a major influencer for one’s behavioral intent and is acknowledged in theories as UTAUT and Innovation Diffusion Theory. Therefore, geographical distance is hypothesized as having a relationship with both perceived usefulness and intention to use. So, the further away someone lives from city hall, the more intentions it develops to use
municipal eServices or the more usefulness it perceives. The following hypotheses are formulated:
H2 A citizen’s geographical distance from its residence to city hall will have a negative relationship with one’s intention to use municipal eServices.
H2A A citizen’s geographical distance from its residence to city hall will have a negative
17 The other key factor in TAM is perceived ease of use. Van Dijk et al. (2008) found little evidence to support the relationship of perceived ease of use with the intention to use Dutch eGovernment. Though, in the third version of TAM perceived ease of use is acknowledged as being a key factor for
determining intention to use. Six antecedents of perceived ease of use are established through empirical testing (Venkatesh and Bala, 2008). Since one target group is nonusers, they have no actual perception of the complexity and usability of the system. Yet, they may have expectations of the degree of difficulty to operate municipal eServices and may expect to have the necessary computer skills to operate the system.
Some studies found little (Van Dijk et al., 2008; Carter & Bélanger, 2005) or no (Bhattacherjee, 2007; Tseng & Lo, 2011) evidence to support a relationship of perceived ease of use with intention to use. Yet, the importance of perceived ease of use cannot be ignored. Interviewee 3 is clear about the importance of ease of use “I have to say it is really important to me” and asking if he perceived no usefulness at all he denies by saying “No, I just thought it would be too hard for me to take care of it
online”. Some studies (Muthitcharoen, 2011; Tseng & Lo, 2011) found more evidence for the relationship of perceived ease of use with perceived usefulness than for perceived ease of use with intention to use. Therefore the importance of perceived ease of use is hypothesized as an interrelationship between perceived ease of use and perceived usefulness and a relationship of
perceived ease of use with intention to use. Two hypotheses are formulated:
H3 A citizen’s perceived ease of use of municipal eServices has a positive relationship with its intention to use.
H3A A citizen’s perceived ease of use of municipality eServices has a positive relationship with its perceived usefulness.
TRA, and also its technological successor TAM, have some limitations (Hale et al., 2003; Sheppard et
al., 1988) including the assumption that when someone forms an intention to act, they will be free to act without limitation. In practice, constraints such as limited ability, time, environment and resources will limit the freedom or ability to act.
The Theory of Planned Behavior (TPB) was an improvement of the Theory of Reasoned Action. Ajzen
18 (Ajzen, 1991). Mathieson (1991) validated the TPB model in an IT context and concluded that TPB is a powerful model to predict the usage of IT.
The Theory of Planned Behavior posits that perceived behavioral control plays an important role in determining one’s behavioral intention. Horst et al. (2007) found no evidence to support the relation
of perceived behavioral control with the intention to use of municipal eServices in The Netherlands. This is remarkable since the relationship is verified by various researchers (Hung et al., 2006; Wu & Chen, 2005; Yaghoubi et al., 2010). Moreover, the findings of the pilot interviews pinpoint perceived behavioral control as a heavyweight in determining one’s intention to use the system. Therefore, the
relationship is reexamined:
H4 A citizen’s perceived behavioral control of municipality eServices has a positive relationship with its intention to use.
The Unified Theory of Acceptance and Use of Technology (UTAUT) model (Venkatesh et al., 2003) is a
more evolved model than TAM or TAM2. Eight different models and theories were investigated and the components of those models were integrated into a single unified model. UTAUT resulted in three antecedents of intention to use, namely performance expectancy, effort expectancy and social influence. These three relationships were moderated by gender and age. The relationship effort expectancy with intention to use is also moderated by experience and the relationship of social
influence with intention to use is also influenced by experience and voluntariness of use. Facilitating conditions, moderated by age and experience, and intention to use had a direct relation to use behavior. Performance expectancy and effort expectancy are defined exactly the same as perceived usefulness and perceived ease of use, respectively. The UTAUT model is considered more predictive than any of the individual models alone.
Venkatesh et al. (2003) state facilitating conditions are similar to perceived behavioral control. Yet,
there’s a fundamental difference. Perceived behavioral control is defined relatively global, while facilitating conditions is defined as being specific to the support of organizational and IT infrastructure: “the degree to which an individual believes that an organizational and technical infrastructure exists to support use of the system”. Yet, eGovernment is for citizens to use in private
or public environments and thus organizational support is not entirely applicable. For the sake of the context of this research, the original definition of Venkatesh et al. is redefined to “the degree to which an individual believes that technical infrastructure exists to support use of the system”. Comparing the adjusted definition to the original definition of perceived behavioral control, it may be inferred that facilitating conditions is a determinant of perceived behavioral control. In concrete, if
19 expect not to have the necessary facilities to operate the system it will limit their ability to do so. Knowing this should lower one’s intention to use the municipal eServices. This results in the following hypothesis:
H4A Perceived facilitating conditions of municipality eServices have a positive relationship with
intention to use.
Yet, if by redefining the original definition of facilitating conditions, organizational support is ignored. In private and public environments, organizational support is hard to measure and to coordinate, because this ‘organization’ is without boundaries. However, the support delivered from the system
and organization itself can be measured to means of people’s perception. If the municipality designs the system so that it becomes easy to operate and its procedures and terminology become easy to understand, one will perceive the system as something that is within their control.
A term that is frequently used to explain one’s technological knowledge of a system is self-efficacy.
Self-efficacy is hypothesized often as being an antecedent of perceived behavioral control (Hsu & Chiu, 2004; Hung et al., 2006; Yaghoubi et al., 2010). Bandura's (1982) extensive research on self-efficacy, defined self-efficacy as "judgments of how well one can execute courses of action required to deal with prospective situations". According to Davis (1989), self-efficacy is similar to perceived ease of use. If one expects or perceives the use of a system as being free of effort, it does not
withhold the person from using the system, which is a form of behavioral control. Nonetheless, the definition of self-efficacy is not IT limited. Compeau and Higgins (1995), therefore, coined the spin-off definition computer self-efficacy. This definition is in fact similar to perceived ease of use. Hence, perceived ease of use determines perceived behavioral control.
Reconsidering the original definition of self-efficacy and the context of this study, it can be hypothesized that people are just too unfamiliar with and insecure about certain procedures and
therefore prefer the traditional service counter over eService, because they are guided through the process by a municipal employee. Some people may experience the procedure for tax returns as being hard to understand and do not recognize terms from everyday life. Therefore, the term perceived required knowledge is coined and defined as “the knowledge one perceives to be required
for following a procedure or understanding a terminology which is designed in the information system”. Interviewee 4 states: “I did it (went to city hall) because I expect to have too little knowledge on the subject and the person behind the counter desk understands it in all probability better than I do”. So, if someone expects to have the knowledge to understand the system, the
20 In short, if someone expects to have enough knowledge of the regarding procedure and terminology and expects to have the skills to operate the system, using the system will be perceived as within their control. The next hypotheses are formulated:
H4B A citizen’s perceived ease of use of municipality eServices has a positive relationship with
its perceived behavioral control.
H4C A citizen’s perceived required knowledge of municipality eServices has a positive relationship with its perceived behavioral control.
Various researchers (Ang et al., 2001; De Ruyter et al., 2001; Gefen, 2002; Horst et al., 2007; Mayer
et al., 1995; Wang & Emurian, 2005) have addressed the importance of trust as a factor that determines the intention to use an electronic innovation. Trust is a complex and abstract construct, which is difficult to define. The definition, within the field of psychology, of Rotter (1967) is adopted for this paper: “expectancy held by individuals or groups that the word, promise, verbal or written
statement of another can be relied on.” Translating this definition to the eGovernment context, it can be said that this definition assumes trust between people, while trust in eGovernment is about trust between an individual and the government and its IT facilities. One should have a certain level of trust in the municipality and its eServices in order to interact through these eServices. This is confirmed by Hoffman et al. (1999); they stated “consumers simply do not trust most Web providers
enough to engage in ‘relationship exchanges’ involving money and personal information”. If not redressed, a lack of trust may eventually have disastrous ends to the successful conduct of eServices (Bhattacherjee, 2002; Lee & Turban, 2001; McKnight & Chervany, 2001; Schneiderman, 2000). One might expect government agencies in the Western world to be trustworthy by dealing in the citizen’s best interest. Yet, this is not the only factor that determines trust. According to Bhattacherjee (2002),
trust in an online service provider is based on a person’s perception of the provider’s ability to deal with the transaction, the integrity of the service provider and the provider’s benevolence to deal with the transaction in the best interest of the consumer or user.
While trust is about relying on an organization, risk perceptions may be formed by the presentation of an organization, by the environment it acts in, the way it provides its products or its services in
specific. One’s perception of a service being risky may withhold the person from engaging in transactions. Pavlou (2003) pinpoints two forms of risk: (1) a risk of monetary loss, since one has to rely on electronic information and thus becomes vulnerable to incomplete or distorted information provided by Web retailers and third parties (Lee, 1998) and (2) the risk of loss of privacy associated with providing personal information to online service providers (Culnan & Armstrong, 1999).
21 for myself that the transaction is confirmed, (…) If I want to change my address I want to make sure it went okay and not find out otherwise”.This stresses the importance of perceived risk in government-to-citizen relationships and transactions.
When people trust others, they assume that those they trust will behave as expected, reducing the
complexity of the interaction. Consumers tend to assume that a trusted Web retailer will not engage in opportunistic behavior (Gefen, 2000) and, thus, trust reduces the perceived risk (Lewis & Weigert, 1985; Luhman, 1979; Mayer et al., 1995). So, risk perception is a term which is often associated to trust. Mayer et al. (1995) claim there is only a longing for trust in a risky environment. However,
theorists do not unanimous agree about the relationship between risk perception and trust. Also, they disagree about risk perception being the antecedent of trust or that trust is the antecedent of perceived risk. Horst et al. (2007) suggest further research to establish the antecedent and the consequence. Since establishing this relationship is not the goal of this research, the dissonance on this matter will be respected. Notwithstanding, results from the pilot interviews present a telling fact
for this context specifically. It reveals that trust is not seen as an influencer of their intention to use, while their perception of risk is in few cases. Based on the discussed theories and empirical findings, two hypotheses are formulated:
H5 A citizen’s perceived risk of municipality eServices has a negative relationship with its intention to use.
H5A A citizen’s trust in municipality eServices has a negative relationship with its perceived risk.
Equity theory suggests that in every exchange relationship, individuals are constantly concerned about their inputs, outcomes, and the fairness of the exchange. Individuals are also constantly comparing themselves with others in their reference group to assess whether the relative gains are the same (Adams, 1963; Adams, 1965; Walster et al., 1978). So, if their direct environment is using
municipal eServices and they experience that those people have relatively higher gains than when using other contact channels, they might be stimulated to use the system their selves.
Another theory that acknowledges the effect of social influence on human behavior is the Social Cognitive Theory (Bandura, 1977; Bandura, 1982; Bandura, 1986). Social Cognitive Theory is a widely
22 affected by environmental or situational characteristics, which are in turn affected by behavior. Finally, behavior is influenced by cognitive and personal factors, and in turn, affects those same factors. To simplify the previous, one influences the environment one acts in and the environment
influences the people that are part of the environment. So, if one’s environment, consisting of friends, family, neighbors and/or co-workers, is positive and supportive of the use of municipal eServices, according to the Social Cognitive Theory, the individual in question would be influenced by this and might even change its behavior towards using these online services.
This phenomenon can also be traced back to the Theory of Reasoned Action. In this theory,
subjective norm is defined as “the person's perception that most people who are important to him think he should or should not perform the behavior in question” (Fishbein & Azjen, 1975). Where TAM did not recognize the importance of subjective norm, TAM2 promoted subjective norm as a key factor in explaining the intention to use an information system. In TAM2 subjective norm has a direct effect on the perceived usefulness and the image of a technology, another antecedent of perceived
usefulness. Moreover, subjective norm is correlated with intention to use. Hence, subjective norm is an important construct in this model. Yet, many researchers choose to base their IS adoption model on the original TAM (Carter & Bélanger, 2005; Phang et al., 2005; Phang et al., 2006; Van Dijk et al., 2008; Wu & Chen, 2005) and did not include subjective norm as a construct in the model. Having in mind that equity theory, Social Cognitive Theory, Theory of Reasoned Action and TAM2 all stress the
importance of social influence as a factor, subjective norm can only be embraced as a factor that determines one’s behavior. To sum up, the more positive the environment is about municipal eServices and the more pressure they put on the nonusers, the higher their intention to use will be.
The following is hypothesized:
H6 Subjective norm has a positive relationship with one’s intention to use.
As mentioned before, not using a system despite being aware of the possibility to do so, is a form of passive resistance. However, it is not necessarily so that they purposely resist the change.
First, resistance behaviors are discussed to emphasize its harmful nature. Resistance behaviors are characterized by low levels of use, by a lack of use or by dysfunctional, harmful use. Such behavioral
reactions are well documented in the literature (Martinko et al., 1996). For example, active resistance which took the form of sabotage, union resistance and threats followed the introduction of new technology into the US Post Office in a study conducted by Dickson et al. (1974). Passive resistance in the form of low system usage and affective reactions has been well documented (Baroudi et al., 1986; Igbaria & Parasuraman, 1989; Meyer & Goes, 1988). In practice, user resistance
23 the multi-level resistance of Lapointe and Rivard (2005). They distinguish four levels of resistance: apathy, passive resistance, active resistance, and aggressive resistance. Examples of these four levels are inaction, refusal, voicing dissatisfaction and rebellion, respectively.
The behavior on any of these levels is undesirable for a system to get accepted. Interviewee 9 makes
it clear: “I think it’s ridiculous that municipalities cooperate to create high levels of unemployment. If everyone in The Netherlands is going to use this eLoket, then service desk employees will be dismissed”. Grounded or not, this attitude will lead to resistance behavior. This emphasizes the importance of identifying if there is resistance to the change. The use of municipal eServices is
entirely voluntary and inaction can therefore be mistaken for apathy resistance. Inaction can also be product of not being aware of the existence of municipal eServices. It is therefore of paramount importance that the presence or absence of resistance to change is established.
Kim and Kankanhalli (2009) define user resistance as the opposition of a user to change associated with a new IS implementation. Information systems implementations have historically been plagued
by failures for which user resistance has consistently been exposed as the main driver behind this failure. A survey of ITtoolbox in 2004 (Kim & Kankanhalli, 2009) indicated that user resistance is the first-ranked challenge for the implementation of large-scale IS.
Bhattacherjee and Hikmet (2007) point out that in IS acceptance enabling factors are examined
thoroughly, but that there is little if any consideration of inhibiting factors. Based on Cenfetelli’s (2004) dual factor model and Lewin’s (1947) notion of opposing forces, Bhattacherjee and Hikmet propose, and validated, the dual factor model for IT usage where intention to use and resistance to change coexist as antecedents of IT usage and resistance to change is an antecedent of intention to use.
It is important to address resistance and how strong the impact of resistance is within a certain
situation in order to treat if necessary. This results in the formulation of the following hypothesis:
H7 A citizen’s resistance to change will have a negative relationship with one’s intention to use municipal eServices.
Figure 2 visualizes the thirteen hypotheses introduced and explained in this chapter. The research
24
Figure 2. Research model.
5
Applied research method
The research model and hypotheses discussed in the previous chapter can be measured by applying numerous methodologies. Though, Heeks and Bailur (2007) called for an increase in the use of quantitative analysis with strict collection and analysis of data in eGovernment research. To answer this call and in order to answer the research questions of this study, data is collected and separated
25 5.1 Data collection
To test the research model, a survey on municipal eServices was conducted among users and nonusers from the municipalities of Groningen and Emmen. To answer the fourth research question,
if the research model is applicable to various municipalities, two samples from two municipalities were taken. Groningen and Emmen were selected because they both score relatively high on e-Overheid monitor (2011) with rankings of 70 and 8, respectively, out of 418 municipalities based on quality criteria for designing eLoket. From this it may be inferred that both municipalities invest heavily in their eLoket and thus in their eServices. The eLoket websites of the municipality are not
very different in de sense of the online services they offer. The distance between both municipalities is relatively short with approximately 70 kilometers. Yet, the demographics, presented in table 3, both municipalities are fundamentally different and herein lays the challenge.
Demographic information Groningen Emmen
Population 187.197a 109,493b
M/F distribution 49.5% / 50.5%a 49.4% / 50.6%b
Avarage age 36.4c 41.7b
Total surface in m2 83,690,000a 346,240,000b
Student percentage of total population 28.2%a 2%d*
a
Gemeente Groningen (2010), b Gemeente Emmen (2010), c Groningen Toerisme (2010), d Kiesjestudie (2011)
*
No exact indication of publication date. Associated population is similar to data from 2010. Table 3. Demographic information on Groningen and Emmen.
Analyzing the information presented in table 3, it can be said that the average geographical distance from the homes of citizens of Emmen is longer compared to the citizens of Groningen, since the
surface of Emmen is more than four times as large as Groningen. There’s approximately a five year difference in the average age and Groningen has a much larger population. Just two percent of the population in Emmen is student, while 28.2% of the population of Groningen is a student. So, there are fundamental differences in the demographics between the two. One may state that Emmen is
more rural and Groningen is more urban.
5.2 Measurement instrument
26 Because this is a direct measure, and no construct like the other factors, the factor distance is visually distinct in figure 2 on page 24.
Lehmann and Hulbert (1972) state in their work “if the focus is on individual behavior, five to seven point scales should be used.” So, for the measurement items, except for demographics, a seven-point
Likert scale is adopted. These scales range from (1) strongly agree to (7) strongly disagree.
Nunnally (1978) points out, “rather than test the validity of measures after they have been constructed, one should ensure the validity by the plan and procedures for construction". Therefore, validated measures are adopted from earlier work as presented in table 4. Measures are
reformulated to fit the context and redundant or irrelevant measures in the original work are omitted. The survey items for users are formulated to address their experience with municipal eServices and the survey items for nonusers are formulated to address their expectations of municipal eServices. The measurement items that were included in the survey is presented in appendix B. Since, this study has a specific focus on establishing the difference between users and
nonusers, items of both surveys are listed in the appendix. The items were randomized in the actual survey to avoid creating a bias and to test consistency.
Construct Indicator Source
Intention to use (INT) INT1 / INT4 Venkatesh et al., 2003
INT2 / INT5 Venkatesh et al., 2003
INT3 / INT6 Taylor & Todd, 1995b*
Perceived usefulness (PU) PU1 / PU4 Venkatesh et al., 2003
PU2 / PU5 Koufaris, 2002
PU3 / PU6 Pavlou, 2003
Perceived ease of use (PEOU) PEOU1 / PEOU4 Venkatesh et al., 2003
PEOU2 / PEOU5 Venkatesh et al., 2003
PEOU3 / PEOU6 Venkatesh et al., 2003
Perceived behavioral control (PBC) PBC1 / PBC4 Taylor & Todd, 1995b
PBC2 / PBC5 Taylor & Todd, 1995b*
PBC3 / PBC6 Taylor & Todd, 1995b
Perceived required knowledge (PRK) PRK1 / PRK4 New developed item
PRK2 / PRK5 New developed item
PRK3 / PRK6 New developed item
27 FC2 / FC5 New developed item**
FC3 / FC6 (reversed) New developed item**
Trust (T) T1 / T4 Pavlou, 2003
T2 / T5 Bhattacherjee, 2002*
T3 / T6 Bhattacherjee, 2002*
Perceived risk (PR) PR1 / PR4 Pavlou, 2003*
PR2 / PR5 Malhotra et al., 2004
PR3 / PR6 Malhotra et al., 2004
Resistance to change (RC) RC1 / RC4 Bhattacherjee & Hikmet, 2007
RC2 / RC5 Bhattacherjee & Hikmet, 2007
RC3 / RC6 Bhattacherjee & Hikmet, 2007
Subjective norm (SN) SN1 / SN4 Venkatesh et al., 2003
SN2 / SN5 Venkatesh et al., 2003*
SN3 / SN6 Thompson et al., 1991*
* statement adopted from noted source but altered to fit the context. ** statement is based on items developed by Taylor &Todd (1995b). Table 4. Construct development for measurement instrument.
The paired indicators in table 4 are adopted from the same source. The series 1 to 3 are indicators in
the user sample and the series 4 to 6 are indicators in the nonuser sample. As explained in the previous section, the definition of facilitating conditions was stripped down and ended up in two constructs. Hence, new scale items had to be developed. Indicator FC3 (and also FC6) is a reversed item. Meaning if one agrees to FC1 and FC2, the person should disagree with FC3, where it states the
opposite.
5.3 Pretest
The survey was qualitatively and quantitatively tested on a small scale. Citizens from Emmen and Groningen were asked to take part in the pretest. The test was conducted face-to-face. The goal was to test the clearness of the survey and reliability of the measurement items, not to create preliminary
results. The pretest resulted in 23 respondents. Thirteen of them were nonusers and ten are users. It took them an average of 4 minutes and 32 seconds to complete the test.
The questionnaire was accurately translated into Dutch based on the measurement items shown in appendix B. An independent individual, fluent in both Dutch and English, translated the survey back
28 Then the survey was sent to the participating municipalities. They commented to change the order of the survey; to first test the measurement items and then gain demographic information. Moreover, they felt not comfortable asking about the citizen’s income. Therefore the income question, as part
of the demographic information, was dropped. Further, they had difficulties with the term municipal eServices; they rather saw it replaced by eLoket. Yet, municipal eServices is far more specific than the term eLoket, which also contains downloading information or forms. Examples of these eServices are included in the introduction of the survey.
The third part of qualitatively testing was to gauge the citizens if they comprehend the statements
and words of the test. Ten respondents mentioned they had trouble grasping the Dutch term ‘in mijn macht’, which means in my control. This concerned the items PBC2 and PBC5. Eight respondents asked what is meant with the Dutch term ‘steunen’, which is a translation for support. This comment was directed at the items SN3 and SN6.
From the factor loadings in the quantitative part of the pretest, the items PEOU1, PBC2, SN3, PU4,
INT6 and T6 scored inconsistently with the other items in their construct. PBC2, SN3 and T6 were therefore reformulated. To these constructs minor changes were applied, like using synonyms or adding ‘I think’ to ease down a statement. Due to these results and results from the qualitative pretest, PBC5 and SN6 were reformulated consistently to PBC2 and SN3, respectively. PU4 and PEOU1 are not altered despite inconsistent factor loadings. These items were copied straightforward
from previous validated studies and therefore no consistency is expected in a larger sample. INT6 was left in its original state, since there was a minor inconsistency in the nonuser sample and was considered consistent in the intention to use construct in the user sample. Hence, no major
adjustments are made due to the low reliability of this small scale pretest.
5.4 Field survey
A field survey was conducted for five weeks. Participations were solicited through emails exploiting the personal network of the researcher, by publishing invitations in several groups on social media like LinkedIn, Facebook and Hyves and an invitation was published on the students’ learning system of the University of Groningen. In total 628 individuals started the online survey and 292 individuals
completed all stages of the online survey.
Hardcopies of the survey were distributed in Emmen at city hall, various neighborhoods and among family and acquaintances. In Groningen the survey was distributed in the researcher’s personal network and in various neighborhoods. This resulted in 91 returned surveys. Among them, 73 were
29 In total 365 surveys were completed. Among them, 337 are useful and 28 surveys that were filled out inconsistently were excluded from the sample. As discussed before, the 337 surveys are divided in four samples. Demographic information on these four samples is depicted in table 5. The results are
quite similar to the actual profile per municipality as illustrated in table 3, where citizens from Groningen live more close by city hall, are younger and are more educated. The percentage of gender participants is also representative. The samples nonusers Emmen, nonusers Groningen, users Emmen and users Groningen in table 5 are abbreviated as NE, NG, UE and UG, respectively.
Demographic variable Category UE
(N=70) UG (N=83) NE (N=111) NG (N=73) Age 18-29 31.4% 50.6% 25.2% 45.2% 30-49 38.6% 25.3% 30.6% 31.5% 50-64 22.9% 20.5% 33.3% 15.1% 65+ 5.7% 3.6% 6.3% 8.2% Unwilling to disclose 1.4% - 4.5% - Gender Male 44.3% 54.2% 50.5% 50.7% Female 55.7% 45.8% 49.5% 49.3% Unwilling to disclose - - - - Education Academic 17.1% 45.8% 4.5% 27.4% Polytechnic 35.7% 37.3% 18% 40.1% Secondary vocational 25.7% 14.5% 43.2% 21.9% Primary vocational 15.7% 2.4% 21.6% 5.5% Secondary school 5.7% - 12.6% 2.7% Unwilling to disclose - - - 1.4% Distance 0 - 5 km 11.4% 56.6% 12.6% 45.2% 6 - 10 km 20% 26.5% 23.4% 28.8% 11 - 15 km 30% 8.4% 36.9% 13.7% 15 - 20 km 34.3% 8.4% 26.1% 9.6% 21+ 4.3% - 0.9% - Unwilling to disclose - - - 2.7%
30
6
Results
This chapter is dedicated to present the results of the statistical analysis. Two different methods
were applied for statistical testing: Dunnett’s 2-sided t-test and Partial Least Squares.
6.1 Comparing constructs
The constructs of the research model are compared across the samples for two reasons. First, to establish the need for four samples. In case there’s no difference between two or more samples, there’s no need to distinct the two from each other and the samples may be merged. Second, if there
is a difference, it’s necessary to provide insight on how big this difference is in order to answer the research questions.
To establish whether there’s a difference or not, Dunnett’s 2-sided tests is selected. Dunnett’s t-test treats one group as a control group and compare all other groups against it. Here, the means per construct are compared to the mean of the same construct in a different sample. This test was
conducted using IBM SPSS Statistics 19. The results of the ten constructs and the factor distance are produced in appendix C.
To compare the samples to each other, six possible comparisons can be made. However, comparing nonusers from Emmen to users from Groningen and nonusers from Groningen to users from Emmen
seems irrational and in case these groups are similar it would be mere coincidence. Relevant comparisons are summed up based on details from appendix C.
Comparing users from Groningen and Emmen to each other, there is evidence (ρ < 0.05) to state that two constructs are significantly different. Perceptions of these groups on trust (-0.475) and facilitating conditions (-0.38) are different from each other. The results also indicate that the distance
(-1.31) among the groups is different to each other.
Comparing nonusers from Emmen and Groningen to each other, there is evidence to state that again two constructs are different. Respondent’s resistance to change (-0.66) and intention to use (-0.57) differ slightly from each other. Similar to the users sample comparison, distance (1.00) is considered
to be different across the samples.
The comparison of users and nonusers from Emmen present clear differences. Eight out of eleven constructs are significantly different. No evidence was found to support a difference for trust, distance and subjective norm. Almost equal results were produced in comparison of users and nonusers from Groningen. Here, nine out of ten constructs are significantly different. No evidence
