The change willingness of public professionals

The change willingness of public professionals

To what extent does operational powerlessness have an effect on the change

willingness of locomotive engineers of Nederlandse Spoorwegen (NS) and to

what extent is this effect moderated by the content of the change message?

Name: Daphne Bossink

Student number: S1693786

Master programme: Public Administration

Master track: Public Management and Leadership Supervisor: Dr. P.E.A. van den Bekerom

Second reader: Dr. J. van der Voet

Date: 04-06-2020

Abstract

The change willingness of public professionals is increasingly discussed by public change management scholars. This article studies the effect of operational powerlessness and the change message on change willingness and the moderating effect of the change message on the relationship between operational powerlessness and change willingness. These relationships are tested with statistical analysis through a survey among 441 locomotive engineers who implement a new technological device that should improve time efficiency. The locomotive engineers are employed by NS, the main train passenger service company of the Netherlands. The results show that operational powerlessness is negatively related with change willingness, but this relationship can be attenuated by a change message with frames that correspond with the professional values of client service. Such a change message also has a positive effect on the change willingness of the public professionals on its own. The confirmation of the hypotheses concerning the relationships between operational powerlessness and change willingness and the change message and change willingness reinforce the existing theory. In addition, the confirmation of the hypothesis regarding the moderating effect of the change message on the relationship between operational powerlessness and change willingness shows the need for further research as this relationship has not been examined before.

Content

Appendix 1 Survey Items ... 43

Appendix 2 Factor Analyses And Cronbach’s Alpha Test ... 44

Appendix 3 Codebook ... 47

Introduction

Nederlandse Spoorwegen (hereafter: NS) is the main passenger rail service company of the Netherlands (ProRail, n.d.). One of their most important responsibilities is to run the trains on time (Ministerie van Infrastructuur en Waterstaat, n.d.). In order to improve its achievements on this key performance indicator, NS has introduced the programme ‘Op Tijd Reizen’ (OTR, Running on Time) in 2018 (R. van Betuw, personal communication, February 17, 2020). With the use of technology, standardisation of work processes is introduced through for instance smartwatches and tablets so that the trains can depart with the precision of seconds (NS, 2019). The programme has a three-dimensional focus: on the locomotive engineers, on the conductors, and on the passengers (R. Danek, personal communication, October 10, 2019). In this thesis, the part of OTR that is directed towards the locomotive engineers will be researched. For the locomotive engineers, the programme entails an app called TimTim that should improve the time efficiency of the train journeys (Nu.nl, 2018).

In this research, public management theory on policy alienation, as well as change management theory will be used. Additionally, insights from the field of psychology will support the underlying mechanisms of the expected relationships. The mechanisms will not be measured empirically, but the literature can help to theorise about these underlying mechanisms.

According to the research of Tummers (2011), public professionals are less willing to change their behaviour in order for a new policy to succeed if this would lead to more operational powerlessness, since that causes policy alienation. Policy alienation emerges when public professionals do not feel psychologically involved in new policies (Tummers et al., 2009, p. 688). Operational powerlessness is a component of policy alienation that refers to the influence a public professional has on the actual implementation of new policies (Tummers, 2011, p. 562). When that influence is low, operational powerlessness is high. The study of Tummers (2011) among Dutch healthcare professionals has confirmed the hypothesis that operational powerlessness leads to less change willingness, wherein change willingness means having the commitment to change one’s behaviour for a successful implementation of a new policy (Herold et al., 2007, p. 943). As the programme OTR involves technologies with which standardisation of the work of locomotive engineers is realised, it is plausible that it implies more operational powerlessness, which is expected to cause less change willingness (NS, n.d.b.; ProRail, 2018; Tummers, 2011).

Andersen and Jakobsen (2017) have researched the effect of the content of the change message on the change willingness of street-level bureaucrats working in the education sector

in the United States and Denmark. With content of the change message, they refer to the use of frames and cues (Andersen & Jakobsen, 2017). In short, framing is highlighting certain ideas over others and cues are small pieces of information which people can use to quickly form a judgment (Druckman et al., 2010). The findings of the study support the hypothesis that change communication that gives attention to professional norms has a positive influence on the change willingness of street-level bureaucrats (Andersen & Jakobsen, 2017, pp. 62-63).

Based on the study of Andersen and Jakobsen (2017), one can expect a change message with frames that are in accordance with professional norms to have a positive effect on the change willingness of street-level bureaucrats. Next to this relationship, the main focus of this research will be on the interaction between operational powerlessness and the change message. A change message with frames corresponding to professional values is expected to mitigate the negative relationship between operational powerlessness and change willingness. The reason for this is that when professionals recognise the value of the new policy for their clients, and thus understand the reason for the increase in operational powerlessness, the increased operational powerlessness is expected to have a smaller negative impact on the change willingness.

All in all, operational powerlessness is expected to have a negative influence on change willingness and a change message with frames corresponding to professional values is expected to mitigate this relationship. Beside these two hypotheses, the individual expectation that a change message including frames in accordance with professional values will positively influence change willingness will also be examined.

The described programme OTR and the mentioned theories of Tummers (2011) and Andersen and Jakobsen (2017) lead to the following research question: ‘To what extent does operational powerlessness have an effect on the change willingness of locomotive engineers of Nederlandse Spoorwegen (NS) and to what extent is this effect moderated by the content of the change message?’.

Since the programme OTR has been introduced for all locomotive engineers of the main passenger rail service of the Netherlands, it is relevant to research the implications for the change willingness of locomotive engineers. But the outcomes of this research are also relevant for other street-level bureaucrats who’s work can or will become more standardised.

The outcomes of this research are not only relevant for society, but also for the current state of the academic field. The relationship between powerlessness and change willingness has been researched before (Tummers, 2011; Tummers et al., 2015), as well as the influence of the change message on change willingness (Andersen & Jakobsen, 2017). However, these

relationships have not yet been examined in combination with each other, where the change message is the moderating variable. In this way, the thesis will add to the field of change management.

Besides the contribution to change management, this research will also borrow insights from the field of organisational psychology. In this way, the study contributes to the growing field of behavioural public administration. Behavioural public administration can be defined as “the analysis of public administration from the micro-level perspective of individual behavior and attitudes by drawing on insights from psychology on the behavior of individuals and groups” (Grimmelikhuijsen et al., 2017, p. 45). It is useful to combine insights from the field of organisational psychology with change management and public management studies in this research, because it sheds light on the psychological reasons for the expected relationships between concepts. As mentioned above, the expected relationships are those between operational powerlessness and change willingness and between the change message and change willingness. In addition, the expected moderating effect of the change message on the first hypothesised relationship will also be researched.

The data necessary for answering the research question will be attained by sending a survey to all relevant locomotive engineers of NS, which limits selection bias. In this case, relevant locomotive engineers are those that use TimTim on their routes. The research design is cross-sectional large-N, using the conditioning approach in the statistical analysis in which the influence of possible confounding variables is adjusted for (Toshkov, 2016, pp. 220-224). The potential confounding variables such as age and years of employment will be selected with the use of knowledge on policy alienation and change willingness.

The outline of the research is as follows. In Chapter 1, the context of the case will be outlined. In Chapter 2, the theoretical framework will be developed, with the use of a literature review. On the basis of this theoretical framework, a conceptual framework will be developed, and hypotheses will be set up. The research design and data collection will be outlined in Chapter 3. In the same chapter, the operationalisation of the described concepts will be given as well as a reflection on validity and reliability. Chapter 4 consists of the empirical findings of the survey of which the implications for the hypotheses will be discussed. The implications for the theory will be given in Chapter 5, where attention will be given to the strengths and limitations and suggestions for further research. Chapter 6 consists of the conclusion of the research.

Context

In order to fully comprehend the application of the theories to the case of NS, it is necessary to have an understanding of the context of the case. As mentioned in the introduction, NS is the main train passenger service company of the Netherlands (ProRail, n.d.). The company used to be publicly owned but was privatised in 1995 (Keuning, 2012; NS, n.d.c.). Privatisation in this case means that it became a listed company; however, 100 percent of the shares are owned by the government (Rijksoverheid, n.d.b.). From the privatisation onwards, NS had to compete with other train transport providers and could focus solely on railway transport (NS, n.d.c.).

Because NS is now a semi-public organisation, the company is largely free to make its own decisions regarding its business operations. Nevertheless, for the right to conduct passenger services on the main railway network, NS has a contract with the Dutch government; more specifically with the Ministry of Infrastructure (Rijksoverheid, n.d.a.). In this contract, NS agrees to commit to certain arrangements concerning various subjects like sustainability and international cooperation (NS, n.d.a.). For several performance areas, key performance indicators are developed and specific performance levels are agreed upon (NS, n.d.a.).

The programme OTR (Running on Time: see introduction) is introduced to improve the performance of NS on the key performance indicators concerning punctuality (R. van Betuw, personal communication, February 17, 2020). As mentioned in the introduction, the programme focuses on three groups: locomotive engineers, conductors, and passengers (R. Danek, personal communication, October 10, 2019). For all three groups, part of the programme entails the use of technology in order to be more time efficient. Since this thesis investigates the part of the programme that is directed towards locomotive engineers, the implications for the execution of their job will be outlined in the next paragraph.

The most concrete impact of the programme on the daily operations of the locomotive engineers is an app called TimTim that is introduced to help to conduct more time efficient train journeys (Nu.nl, 2018). Concretely, the locomotive engineers use the app on a tablet during their train journeys which should help them run the train in a more efficient way (R. van Betuw, personal communication, March 16, 2020). TimTim has already been operational since 2017, but a new function called ‘RolTijdAdvies’ was only introduced in 2019 (R. van Betuw, personal communication, February 17, 2020). With this function, the app shows the locomotive engineers exactly when they can best pull up and reduce speed. It is not compulsory to use TimTim, but NS has experienced that nearly all locomotive engineers have chosen to use it and that is why circa 2500 locomotive engineers do their work with help from the app (R. van Betuw, personal communication, February 17, 2020).

Theory

Since the objective of this thesis is to study the (combined) effect of operational powerlessness and the change message on the change willingness of locomotive engineers, change management theory as well as public management theory will be used. The focus will be on the theory of Tummers (2011) that attempts to explain the willingness of public professionals to implement new policies with the concept policy alienation, and on the theory of Andersen and Jakobsen (2017) which states that communication that emphasises the coherence between a given policy and professional norms will move policy positions of bureaucrats in favour of the policy. Furthermore, studies from organisational psychology will be used to specify the underlying psychological mechanisms of the relationships that are being examined. As already stated in the introduction, by doing this, the research contributes to the upcoming field within public administration theory: behavioural public administration.

First, an overview of the existing perspectives of scholars in this field is provided, as well as a discussion on important concepts. Second, the theoretical framework is presented, in which the theories that will be used are specified and the hypotheses are developed. Third, the conceptual framework will logically follow from the described theory.

Literature review

In order for organisations to survive and remain competitive, it is often argued that they need to be able to adapt to the changing environment by implementing change when necessary (Burnes, 2004a; Cohen, 1999, p. 373). However, change is not always embraced by the employees of organisations that have to implement it. As Tummers (2011, pp. 556-557) states, the pressure on public professionals in service delivery is a hotly debated topic among public administration scholars. More specifically, more and more attention is given to the unwillingness of public professionals to implement new policies. A potential cause is the shifted emphasis of policies towards economic values since the rise of New Public Management (NPM) in the 1980s (Hood, 1991; Tummers 2011, p. 556). This emphasis would cause public professionals to have difficulties with the policies they have to implement, because the values that public professionals attach importance to often do not match with solely economic values (Tummers, 2013, pp. 3-4).

However, not only public administration literature focuses on the introduction of new policies that have to be implemented by public professionals. Change management scholars have developed many concepts that form the basis of research on the motivation to change when new policies are implemented and increasingly apply this specifically to the public sector

(Miller et al., 1994; Piderit, 2000). Therefore, some of these concepts, such as change willingness and change message, will be examined in this literature review as well. As mentioned earlier, insights from psychology are used to support the underlying mechanisms of the theoretical expectations.

Professionalism

Because the literature on change willingness and policy alienation is related to professionals, it is crucial to have an understanding of the current debate on the conceptualisation of professionalism (Andersen & Jakobsen, 2017; Tummers, 2011). Public administration scholars use different conceptualisations of professionalism; hence, the criteria on what is a professional are disputed (Evetts, 2003; Larson, 1979, p. 6).

First, it is important to recognise the existence of several connected concepts that are associated with professionals. With these concepts, scholars try to identify the features that distinguish professions from non-professional occupations. At the same time, an increasing number of researchers no longer think it is important to draw this hard definitional line (Olofsson, 2009; Svensson & Evetts, 2003). For this research however, it is helpful to have an understanding of the debate to be able to determine to what kind of work fields the theories are applicable. Evetts (2013, p. 781) distinguishes three concepts: profession, professionalisation and professionalism. Different views on these three concepts will be discussed, with an emphasis on profession.

According to Freidson (1984), “professions have been singled out as occupations that perform tasks of great social value because professionals possess both knowledge and skills that in some way set them apart from other kinds of workers” (p. 2). This does not mean that professions are homogeneous, but they have some shared characteristics that make them distinguishable from other occupations. According to Andersen and Jakobsen (2017, p. 59), these characteristics encompass client-orientation and a foundation of knowledge. For professionals, serving clients comes down to doing what the professional thinks is best for the clients (Freidson, 2001, p. 122).

Evetts (2013) regards the concept of professionalisation as “the process to achieve the status of profession” (p. 782). The trend of categorising more and more occupations as professionalised through less strict criteria causes the concept to become less useful (Noordegraaf, 2007; Wilensky, 1964). Therefore, Wilensky advocates more traditional criteria of professionalism such as specialised knowledge and the adherence to professional norms (Wilensky, 1964, p. 138).

Professionalism used to be considered as a normative value that should be strived for in every occupation (Evetts, 2013, p. 782). Nowadays, the concept is not merely used as an ideal, but scholars use different criteria to classify occupations as professionalised or not. As mentioned above, Wilensky (1964) presses for a more traditional model of professionalism. In this model, autonomous expertise and the service ideal are considered necessary for professionalism to occur in an occupation (Wilensky, 1964). Other scholars view professionalism as a form of moral community or even as an ideology in which values like integrity and autonomy are embraced (Durkheim, 1992; Eraut, 1994).

When looking at the described debate of the different concepts, one can see that several aspects are associated with professionals by diverse authors. The most frequently mentioned aspects are autonomy and values such as client orientation and integrity. That is why in this thesis, professionals are regarded as employees with a certain amount of autonomy in performing their jobs and with a culture in which professional values are prominent.

A concept that has not yet been mentioned but that is essential in the debate about public professionals, is street-level bureaucracy. Lipsky (1980) states that street-level bureaucrats are public servants that have direct contact with citizens and experience a big amount of discretion. Lipsky (1980, pp. 201-202) also recognises serving clients as a value that is prominent for street-level bureaucrats. Because the characteristics of locomotive engineers correspond with the concept of professional as well as with the concept of street-level bureaucrat, both concepts are considered applicable towards this case.

Planned and emergent change

Before focusing on concepts that are related to change, it is essential to recognise the distinction between different forms of organisational change. The most prominent distinction is the one between planned and emergent change (Bamford & Forrester, 2003; Liebhart & Garcia-Lorenzo, 2010). Both change practices rest upon a different understanding of the way in which organisations work (Van der Voet et al., 2014, p. 173).

When organisations are seen as stable, planned change is considered the appropriate way to change an organisation (By, 2005, p. 374). Before this type of change takes place, the goals of the change are determined and the strategy and plan are developed (Van der Voet et al., 2014, p. 173; Liebhart & Garcia- Lorenzo, 2010, p. 215). Fundamental for the planned form of change is the design that is comprised of sequential phases (Bamford & Forrester, 2003; Lewin, 1951). For this type of change, planned organisational change models have been developed (Burnes, 2004b).

Critics of the planned change model argue that in order to have sustainable changes one should change in a different way, namely through emergent change (Burnes, 1996; Burnes, 2011). Emergent change as opposed to planned change is more open-ended, often unintended and less rigid (By, 2005, pp. 374-375). Emergent change is argued to be more suitable in the highly dynamic environments that organisations nowadays face (Wilson, 1992). The view of emergent change advocates on organisations is thus that they are always changing to be able to adapt to their surroundings (Burnes, 1996).

The change process that is analysed in this research can be categorised as a planned change. The goals of the change and the ways to achieve it were determined in advance of the change process and the implementation is top-down, although several representatives of the locomotive engineers have been included in the process from the beginning.

Change willingness

A main topic of research in change management literature is the change willingness of employees (Jaros, 2010; Judson, 1991). As stated in a review on public sector change management literature from 2000 to 2010, employees are generally seen as vital actors in change programmes (Kuipers et al., 2014, p. 15). This was demonstrated by the emphasis on resistance to change in the assessed literature (Kuipers et al., 2014, p. 10). Miller et al. (1994, p. 59) point out that the cooperation of employees, which is partly determined by their attitude towards the change, is essential for an organisational change to succeed. Change management scholars thus agree on the necessity of the willingness of employees to change for the realisation of new policies, although they have different ideas about the best ways to achieve it (Higgs & Rowland, 2005).

In a discussion of theory on change willingness, it is useful to compare definitions of the concept. It is important to emphasise that there are numerous concepts that relate to the concept of change willingness (Herold et al., 2007, p. 943). Since commitment to change is widely used interchangeably with change willingness by change management scholars, definitions of that term will be included in this discussion. However, Herold et al. (2007, p. 943) stress the difference between change commitment and other constructs like readiness for change and openness to change. The crucial dissimilarity lies in the element that is part of commitment to change: being willing to change behaviour (Herold et al., 2007, p. 943). This feature can indeed be found in the definition of change willingness of Metselaar (1997): “a positive intention towards the implementation of modifications in an organization’s structure, or work and administrative processes, resulting in efforts from the organization member’s side

to support or enhance the change process” (p. 34). Hersecovitch and Meyer (2002, p. 475) also incorporated in their definition of commitment to change the component of the willingness to put effort into activities that help the organisational change to be successful.

Ensuring willingness to change amongst employees is not a part of an organisational change process that is often easily accomplished or that one should neglect. As already mentioned in the introduction paragraph of this chapter, specific attention is given by scholars to the change willingness of professionals in the public sector (Tummers, 2011, pp. 556-557). Lipsky (1980) and Kaufman (2010) stress that the stance towards policies of street-level bureaucrats influences the way they implement these policies. This, together with the notion of Andersen and Jakobsen (2017) that policy positions are important indicators for the support of employees, shows that policy standpoints are important predictors for the change willingness of professional street-level bureaucrats.

Change message

Since it is increasingly recognised by organisations that employees are important determinants for the successfulness of organisational change, more attention has been paid to ways to communicate change to people (Harkness, 2000, p. 67). Qualitative as well as quantitative studies have been performed and several authors have even developed change communication models (Allen et al., 2007; Goodman & Truss, 2004; Self et al., 2007).

A common understood fact in the change management literature is that the way in which a change is communicated has an influence on the perception of the change of the employees that have to participate in the change programme (Barrett, 2002; Harkness, 2000). It is noteworthy that human resource literature focuses on the intended message to employees, the actual message and the perceived message which causes employees to behave accordingly (Wright & Nishii, 2007). This suggests that the perceived message is the indicator for employee behaviour, but two steps precede that reaction: the intended message and the actual message (Wright & Nishii, 2007, p. 10). Hence, noise in communication can have consequences for the successfulness of a change communication strategy.

In the field of change communication, it is customary to make the distinction between frames and cues (Andersen & Jakobsen, 2017; Druckman et al., 2010). Framing effects can be established by emphasising a subset of potentially relevant considerations when communicating (Druckman, 2001, pp. 226-231). In this way, it is possible for the receivers of the communicated information to form their opinions while focusing on the considerations that have been highlighted (Druckman, 2001, p. 230).

One can also use cues to influence one’s opinion through communication. Cues are small pieces of information that enable people to quickly make up their mind (Druckman et al., 2010, p. 137). One prominent example of cues is expert advice (Andersen & Jakobsen, 2017, p. 59). When a cue such as expert advice is included in a message, the receivers can infer conclusions without analysing further information (Druckman et al., 2010, p. 137).

As mentioned before, some authors have developed change communication models. The model of Armenakis et al. (1999) specifies five key message components that should be included in a change message to guarantee an effective implementation of the change. These components are as follows: self-efficacy, principal support, discrepancy, appropriateness, and personal valence (Armenakis et al., 1999). Andersen and Jakobsen (2017) also focus on crucial components of a change message: frames and cues. Their main hypothesis and outcome is that bureaucrats will be more favourable towards a policy when it is communicated to them with the use of frames and cues that correspond with their policy positions (Andersen & Jakobsen, 2017). In the paragraph on professionals, it has already been pointed out that Andersen and Jakobsen (2017) state that the professional norm of serving clients is a critical attribute of professional bureaucrats. Consequently, provided that the message consists of the necessary components as described by Armenakis et al. (1999), change communication that emphasises the advantages for clients is expected to have a positive influence on the viewpoints of bureaucrats towards new policies.

Operational powerlessness

Operational powerlessness is considered a component of the concept policy alienation (Tummers et al., 2009, pp. 697-698). Seeman (1959) started distinguishing several dimensions of the general concept of alienation. Building on that, Blauner (1964) focused on the more specific concept of work alienation. Using these sociological conceptualisations, Tummers et al. (2009) distinguish five dimensions of policy alienation. Tummers et al. (2009) stress the distinction between work alienation and policy alienation, the latter being more focused on the policy than on the job, on the public sector rather than the private sector and on professionals instead of manual workers (Tummers et al., 2009, p. 688).

Having acknowledged the distinctive character of the concept policy alienation, it is worthwhile to look at the definition that scholars attribute to the concept. There is a consensus on the meaning of policy alienation; however, a distinction has been made between policy alienation towards a specific policy or towards general public policies (Tummers, 2017; Van Engen et al., 2016). In this thesis, the concept of policy alienation that is directed towards

specific policies will be used. Based on the definition of work alienation given by Kanungo (1982), Tummers et al. (2009) define policy alienation as “a general cognitive state of psychological disconnection from the policy programme being implemented, here by a public professional who regularly interacts directly with clients” (p. 688). In this definition, one can see that policy alienation is applied to professional street-level bureaucrats.

As mentioned before, the concept of policy alienation has been divided into five dimensions by Tummers et al. (2009). These dimensions can roughly be put in two broader categories: powerlessness and meaninglessness (Tummers, 2017, p. 572). The powerlessness dimensions consist of strategic, tactical, and operational powerlessness and the meaninglessness dimensions of societal and client meaninglessness (Tummers, 2011, p. 564). Lipsky (1980) already pointed to the feeling of professionals of being powerless over the implementation of policies. Powerlessness in this context is the influence one has on shaping a policy programme, as perceived by the professional (Tummers et al., 2009, p. 696). Meaninglessness refers to the inability to understand the contribution of the policy to the greater good (Seeman, 1959, pp. 786-787).

A specific powerlessness dimension of policy alienation is operational powerlessness (Tummers et al., 2009, pp. 697-698). Tummers (2011) defines operational powerlessness as “the perceived degree of freedom in making choices concerning the sort, quantity and quality of sanctions and rewards on offer when implementing the policy” (p. 562). Two concepts that are closely related to operational powerlessness are discretion and autonomy. According to Tummers et al. (2013), discretion is usually used when talking about street-level public servants and autonomy is used when referring to professionals. These terms will be used indifferently in this thesis. Discretion, and thus autonomy as well, can be seen as the opposite of operational powerlessness, meaning that when a street-level bureaucrat experiences a low level of discretion in the implementation of a policy, this goes together with a high level of operational powerlessness (Tummers, 2011, p. 563).

Theoretical framework

In this theoretical framework, the relations between the examined concepts are discussed. This will result in hypotheses that will be tested in the results chapter.

Relation between operational powerlessness and change willingness

The first relationship this study is focusing on, is based on the theories concerning the relationship between operational powerlessness and change willingness. In order to understand

the direction of the expected relationship, theories from the field of change management as well as organisational psychology are useful. Following the aforementioned change management-related conceptualisation of operational powerlessness, this variable increases when professionals feel that they have less discretion or autonomy in implementing policies. In the study of Tummers (2011), it was found that operational powerlessness has a negative influence on the change willingness of professional street-level bureaucrats. This relationship was expected, because operational powerlessness is a dimension of policy alienation, which decreases the commitment of street-level bureaucrats towards an introduced policy since it involves less influence of employees (Wanberg & Banas, 2000).

One can expect this relationship not only based on change management theory, but also on the basis of organisational psychology. In that field of study, the cause of this decreased commitment is explained with the help of a psychological concept: the sense of ownership over the implementation of a new policy (Dirks et al., 1996). Psychological ownership is broader than operational powerlessness as it is not exclusively about the implementation of public policies, but also applicable to the private sector and jobs and organisations in general (Dirks et al., 1996). Psychological ownership emerges when employees view an organisation or new policy as ‘theirs’, which happens when they feel like they have a great influence on it (Pierce et al., 2001, p. 299). When employees do not experience psychological ownership regarding a new policy, their willingness to contribute to this change is expected to be lower (Avey et al., 2009, p. 186). Psychological ownership can thus be seen as a broad concept that is related to the described policy alienation concept, where policy alienation is the opposite psychological state in which public professionals feel emotionally disconnected from the policy (Pierce et al., 2001, p. 299; Tummers et al., 2009, p. 688). Since operational powerlessness is one of the components of policy alienation, the first hypothesis can be developed in the following way:

H1: Operational powerlessness is negatively related to change willingness.

Relation between change message and change willingness

The literature review has shown that according to change management literature, the way in which a change is communicated has an influence on the attitudes of professionals towards this change. More specifically, Andersen and Jakobsen (2017) state that frames in the change message that match with professionals’ values have a positive influence on professionals’ perception on the policy that they have to implement. As mentioned in the paragraph on professionals, Anderson and Jakobsen (2017) state that the professional norm of serving clients

is a critical characteristic of professional bureaucrats. Therefore, it is expected that change communication that emphasises the advantages for clients has a positive influence on the viewpoints of bureaucrats towards new policies. Borrowing from the field of psychology, it follows that attitudes influence behaviour (Ajzen & Fishbein, 1980; Glasman & Albarracín, 2006). Considering that the change message is expected to influence the attitudes of public professionals towards new policies, the second hypothesis can be developed:

H2: Communication frames emphasising professional values of client service are positively related to change willingness.

Interaction effect of operational powerlessness and change message

Apart from the direct relationship between the change message and change willingness, the change message can also be expected to have a moderating effect on the hypothesised relationship between operational powerlessness and change willingness. This moderating effect has not yet been researched, but when looking at the theories that have led to the previous two hypotheses, one can expect this moderating relationship. As explained, operational powerlessness is expected to lead to less change willingness because it entails that street-level bureaucrats feel like they have little influence on the implementation of the change (Tummers, 2011). Since a change message containing frames in accordance with client orientation is expected to positively turn the attitudes of professionals towards an introduced policy (Andersen & Jakobsen, 2017), such a change message can be expected to mitigate the relationship between operational powerlessness and change willingness. Professionals are expected to understand the reasons behind the initiated policy that causes more operational powerlessness and therefore the negative influence of operational powerlessness on change willingness is expected to decrease.

H3: Communication frames emphasising professional values of client service have an attenuating effect on the negative relationship between operational powerlessness and change willingness.

Change willingness Change message

The developed theoretical framework and the corresponding hypotheses are represented by the following conceptual framework:

Operational powerlessness

-

-

Research design

In this chapter, the research design will be outlined, with attention to the methods of data collection and analysis. Further, the operationalisation of the concepts that are described in the chapter on theory will be provided. Finally, the implications of the design for the validity and reliability will be discussed. The data management plan can be found in Appendix 4.

This research is conducted on the basis of large-N statistical data acquired through a survey. As the data collection is performed at a single point in time, across multiple cases and the output is quantitative, the research design can be categorised as cross-sectional (Bryman, 2012, p. 59).

Data collection

The data needed for testing the developed hypotheses are acquired with the use of a survey. A survey is a structured interview that is self-administered (Bryman, 2012, p. 232). Survey questions can be closed or open-ended. In the survey of this research, the questions are closed. In this way, the answers are standardised, which makes them comparable across a large sample of respondents (Bryman, 2012, pp. 249-250). The survey is distributed with the use of the internal digital communication system of NS. Because the common survey distributing method of NS is performed with the programme ‘Google Forms’, this programme is also used for this research.

Case selection

As outlined above, the quantitative data are acquired with the use of a survey. In total, NS employs circa 3,500 locomotive engineers. The survey is distributed among 1,989 locomotive engineers of NS that make use of the TimTim app. The 1,989 locomotive engineers are selected based on the routes on which they operate for the reason that some routes are less appropriate for TimTim (R. van Betuw, personal communication, March 16, 2020). Since the population of this research consists of the locomotive engineers of NS that use TimTim, there was no need to select a sample. Consequently, the survey has been sent to the entire relevant population. In this way, the risk of sample bias is absent (Bryman, 2012, pp. 186-190). Nevertheless, there is always the risk of non-response error that should be kept in mind when analysing the data. This non-sampling error emerges when the group of respondents differs from the non-respondents (Bryman, 2012, pp. 199-200). In order to reduce this risk, one should strive for a high response rate. This also increases the chance of generalisability of the findings (Toshkov, 2016, p. 248). In this research, 442 locomotive engineers have completed the questionnaire, of which one

indicated that he does not work with TimTim. After excluding that respondent from the data, the response rate is 22 percent. When looking at comparable research, one can see that this response rate is relatively similar; 25 percent in the study of Allen et al. (2007) and 29 percent in the study of Tummers (2011).

Data analysis

The survey answers are examined with statistical analysis, using the conditioning approach. With a statistical conditioning approach, one can adjust for the influence of possible confounding variables (Toshkov, 2016, pp. 220-224). A confounding variable can be defined as a variable that is related to both your independent and dependent variable(s) and is also called a control variable (Toshkov, 2016, pp. 221-222). With the help of multiple regression analysis, the effect of possible confounding variables can be measured, removed and in this way the residual association between the independent and dependent variable(s) can be determined (Toshkov, 2016, p. 220). In this research, the possible confounding variables are selected with the help of theory on operational powerlessness and change willingness. The control variables that are included in the statistical analysis are gender, age, and years of tenure at NS. All three control variables are individual characteristics that can shape workers’ judgments concerning organisational change (Caldwell et al., 2009, pp. 438-439). The variable gender is measured with four possible answer categories: ‘man’, ‘woman’, ‘I prefer not to say’, and ‘other’. The variables age and years of tenure are measured with the help of an open text box in which the respondents can indicate the number of years.

Validity

Validity refers to whether an indicator really measures the concept it is supposed to measure (Bryman, 2012, p. 171). In order to determine whether the developed tests for the three concepts are valid, the construct validity of these tests is measured by using factor analysis. The output of the factor analyses can be found in Appendix 2, and it will briefly be presented in the operationalisation section of this chapter.

External validity is a condition for the generalisability of results (Toshkov, 2016, p. 248). When a sample is randomly selected, this improves the external validity and thus the generalisability of the results (Toshkov, 2016, pp. 247-251). Since the survey in this research is sent to the relevant population as a whole and the response rate is acceptable, it is not expected that there will be problems with the external validity of the results. However, one cannot be certain as the characteristics of the respondents and non-respondents cannot be compared.

Internal validity relates to the issue of causality (Bryman, 2012, p. 47). Is the variation in the dependent variable really explained by the independent variable(s)? To increase the internal validity, the described conditioning approach is used in the statistical analysis. In this way, the possibility that a relationship between independent variable(s) and the dependent variable is found due to the influence of confounding variables will be controlled for. Apart from the risk of not identifying all confounding variables, there is always the possibility that a demonstrated relationship is in actuality not in the direction that was hypothesised; this is called reversed causality (Toshkov, 2016, pp. 222-223). In order to reduce this risk, the expected causal directions of the relationships are based on theory.

Reliability

Reliability refers to the consistency of measures or concepts (Bryman, 2012, p. 169). Internal reliability is applicable to multiple-indicator measures (Bryman, 2012, p. 170). The measures are internally reliable if the results on each indicator point to the same direction for the same respondent (Bryman, 2012, p. 170). The internal reliability is tested with the Cronbach’s alpha and the results of these tests are provided for all three concepts in the operationalisation section of this chapter and in Appendix 2. In order to make sure that respondents have not just chosen the same value for every question, reversed items are included in the survey questions. External reliability refers to the possibility of replicating the research (Bryman, 2012, p. 390). With quantitative research, replicability is accomplished more easily. This method section, together with Table 5 in Appendix 1 which includes the survey questions, should ensure the replicability of this research.

Operationalisation

Three different concepts are measured with the survey: operational powerlessness, change willingness and change message. In this section of the research, the operationalisation of these variables will be discussed.

Operational powerlessness

Since the definition of the concept operational powerlessness from the study of Tummers (2011) is used in this research, the same operationalisation is adopted. In the article of Tummers (2012) on the measurement of the subdimensions of policy alienation, the development of the measurement scales is described. First, they have generated 10 items for each subdimension of policy alienation (Tummers, 2012, p. 519). After that, 21 experts have examined these items

(Tummers 2011, p. 568). This resulted in the six best items for each subdimension (Tummers, 2012, p. 519). These items have been included in a survey that was sent to 1,800 mental health care professionals (Tummers, 2012, p. 519). As a result of the analysis of that survey, the best items for each subdimension of policy alienation have been selected (Tummers, 2012, pp. 519-521). Finally, different types of validity of the items have been tested (Tummers, 2011, pp. 568-569). For operational powerlessness specifically, this has resulted in a 6-item scale (Tummers, 2011, p. 569). The answer options for these items are divided into a 5-point Likert scale, with responses ranging from strongly disagree to strongly agree (Tummers, 2011, p. 568). The 6 items are adopted in the survey of this research, with slight changes to specific terms so that they are applicable to the TimTim case and they have been translated into the Dutch language. The questions are included in Table 1 and Appendix 1.

To ensure the validity and reliability of the questions concerning operational powerlessness, a factor analysis and a Cronbach’s alpha test have been performed, which can be found in Table 6 in Appendix 2. The factor analysis found two factors in the items of operational powerlessness. In the rotated component matrix, one can see that the item ‘When I work with TimTim, I have to adhere to tight procedures’ loads on another factor than the other items. Therefore, this item is excluded from the construct variable operational powerlessness in the analysis. Excluding this item also has a positive effect on the Cronbach’s alpha: that value increases from .71 to .77. For the construct variable operational powerlessness, the Eigenvalue is 2.711 and the factor for operational powerlessness explains 45.2 percent of the variance of the model.

Change willingness

Change willingness is also measured in the study of Tummers (2011). The items used to measure it, are based on the validated five-item scale that is outlined in the research of Metselaar (1997). Not all items are adopted in this survey, because of the applicability to the specific case of TimTim. After consultation with NS, the item concerning the intention to make time to implement the change has been removed from the survey, since locomotive engineers do not need to invest extra time in TimTim (R. van Betuw, personal communication, 16 January 2020). In order for the measurement to be thorough, two items are added to the survey based on the study of Miller et al. (1994). These items are about openness towards and the implementation of the new way of working because of the policy change (Miller et al., 1994, p. 68). To ensure consistency, the answer options to all questions are divided into the 5-point Likert scale described above. Again, the survey questions can be found in Table 1 and Appendix 1.

The validity of the scale for change willingness is measured using factor analysis. The analysis only found one factor, which means that no item has been excluded from the construct variable in the analysis. The Eigenvalue is 3.110 and the factor for change willingness explains 62.2 percent of the variance. The reliability is tested with the Cronbach’s alpha test, of which the value is .85. The results of the factor analysis and Cronbach’s alpha test are shown in Table 7 in Appendix 2.

Change message

This study also examines the effect of the part of the change message that consists of frames corresponding with professional values. Since Andersen and Jakobsen (2017) have identified client orientation as the most important value of street-level bureaucrats, the concept change message is measured on the basis of the correspondence of the frames in the message with client values. Just like the other two variables, this variable is also measured according to the perceptions of the locomotive engineers. In addition to this, it is determined if the necessary components of a change message according to Armenakis et al. (1999) were present in the change message concerning TimTim. Therefore, five items that correspond with the components of Armenakis et al. (1999) are added to the survey, namely: self-efficacy, principal support, discrepancy, appropriateness, and personal valence. Just like the other two variables, the response options are limited to the 5-point Likert scale in order to standardise the answers to increase comparability. The seven items in total that measure the variable change message are included in Table 1 and Appendix 1.

The validity and reliability of the items for change message are again tested with a factor analysis and Cronbach’s alpha test, which are shown in Table 8 in Appendix 2. The factor analysis only found one factor of which the Eigenvalue is 3.894 and which explains 55.6 percent of the variance. The Cronbach’s alpha for this scale is .87.

Table 1

Items survey with corresponding concepts

Concept Item

Operational powerlessness

I have freedom to decide how to use TimTim. ®

While working with TimTim, I can be in keeping with the traveller’s needs. ®

Working with TimTim feels like a harness in which I cannot easily move.

When I work with TimTim, I have to adhere to tight procedures.

While working with TimTim, I cannot sufficiently tailor to the needs of the travellers.

While working with TimTim, I can make my own judgments. ®

Change willingness

I intend to try to convince other locomotive engineers of the benefits TimTim will bring.

I am willing to contribute to the success of TimTim.

I intend to try to reduce resistance among other locomotive engineers about TimTim.

I consider myself to be open to the changes TimTim will bring to my profession.

Since the introduction of TimTim, I perform my job in a different way.

Change message

TimTim has been introduced because it is advantageous for the traveller.

When NS introduced TimTim, it was clearly explained which advantages this entails for the traveller.

When I heard that we would work with TimTim, I had confidence in the ability of locomotive engineers to successfully implement the change.

My most important colleagues are committed to the successful implementation of TimTim.

It has been made clear to me that TimTim is necessary for performing a change.

TimTim helps us to improve our performance.

TimTim also has advantages for me personally.

Control variables

What is your gender? What is your age?

For how many years have you been working for NS? Note. ® = reverse item.

Results

In this results section, the outcomes of the statistical analysis will be presented, and the analysis of the results is provided. The chapter is divided into three sections: descriptive statistics, correlations, and the regression results.

Descriptive statistics

The descriptive statistics of all variables in the analysis are presented in Table 2. One can see that 441 respondents have returned valid questionnaires, of which 430 have indicated their age and 436 their years of tenure at NS. The means and standard deviations of the two independent and one dependent variable are reported in Table 2. The average score on operational powerlessness is approximately 2 on the 5-point Likert scale, which indicates that the respondents experience a level of operational powerlessness below the median. The average scores on change message and change willingness are around 3.5, which is higher than the median. The descriptive statistics of the control variable ‘gender’ show that 90 percent of the respondents is masculine, while only 6 percent is feminine. This is not surprising, as the profession of locomotive engineer is known for being a “masculine job” (Don, 2017). The average age of the respondents is approximately 49.5 years. The minimum age is 21 years and the maximum age is 77 years. In Table 2, one can also see that the average years of tenure at NS of the respondents is roughly 21 years. The minimum years of tenure is 1 and the maximum years of tenure is 60. Apart from the described descriptive statistics, the Cronbach’s alpha of the three main variables is also indicated in Table 2. These values have already been interpreted in the research design chapter, and all three variables have an acceptable Cronbach’s alpha. The values for skewness and kurtosis in Table 2 show that one can assume that the scales for the three main variables are normally distributed.

Tab le 2 D es cr ipt iv e s ta ti st ic s K urt os is St. Error .232 .232 .232 S ta ti st ic 1.573 .005 .269 S ke w ne ss _St. _Error _{.116 .116 .116} S ta ti st ic 1.222 -.627 -.672 C ronba c h 's a .772 .846 .866 M ax 5.00 5.00 5.00 77 60 1.00 1.00 1.00 1.00 M in 1.00 1.00 1.00 21 1 0.00 0.00 0.00 0.00 S t. D ev. .799 .955 .871 11.537 13.960 .294 .240 .156 .095 M ea n 1.962 3.475 3.481 49.48 21.12 .90 .06 .02 .01 N 441 441 441 430 436 441 441 441 441 V ari abl es O pe ra ti ona l pow erl es sne ss Cha nge w il li ngne ss Cha nge m es sa ge A ge T enure S ex (1 = m an) S ex (1 = w om an) S ex (1 = I do not w ant t o te ll ) S ex (1 = ot he r)

Correlations

The bivariate correlations between the variables of the analysis are shown in Table 3. The possibility of multicollinearity has been tested. The VIF-values were below 10 and the tolerance values above .10 so no multicollinearity has been detected. All hypothesised bivariate correlations of the variables are statistically significant at the 0.01 level and in the anticipated direction. Operational powerlessness is negatively related to change willingness and change message is positively related to change willingness. The control variable gender in the category ‘I don’t want to tell’ has a significant but small negative relation with change willingness. The control variable age does not have any statistically significant correlations with the other variables. However, the control variable tenure is negatively correlated with change willingness and change message. Tenure has a positive correlation with age, which is not surprising as in general people with a higher age have been part of the working population for a longer period of time.

Table 3

Correlations between all variables in the analysis (N = 441)

(1) (2) (3) (4) (5) (1) Operational powerlessness 1.000 (2) Change willingness - .680** 1.000 (3) Change message -.699** .795** 1.000 (4) Age .039 .010 -.055 1.000 (5) Tenure .092 -.098* -.124** .767** 1.000 (6) Sex (1 = man) -.041 .074 .045 .067 .051 (7) Sex (1 = woman) .000 .002 .013 -.076 -.070

(8) Sex (1 = I do not want to tell) .048 -.095* -.050 -.058 -.027

(9) Sex (1 = other) .047 -.078 -.088 .076 .065

Note. The correlations between the control variables about gender are excluded from this table. ** Correlation is significant at the 0.01 level (2-tailed).

* Correlation is significant at the 0.05 level (2-tailed).

Regression results

The results of the hierarchical multiple regression analysis are presented in Table 4. In all models, change willingness is the dependent variable. In the first model, operational powerlessness is entered into the regression. In the next models, the following variables are added: change message (model 2), the interaction variable of operational powerlessness and

change message (model 3), and the control variables (model 4). For each model, the Rsquared is reported at the bottom of Table 4. One can see that the Rsquared increases with every step in the hierarchical regression. In the first model, operational powerlessness explains the variation in change willingness for 46.2% and in the last model, all variables explain 67.4% of the variation in change willingness. This shows that all independent variables are important predictors of the dependent variable change willingness. The results also show that the two independent variables and one interaction variable all have a statistically significant effect on change willingness. Furthermore, age has a small positive effect on change willingness while tenure has a very small negative effect on change willingness. Hereafter, the implications of the regression analysis results for the individual hypotheses will be considered.

Hypothesis 1 predicts that operational powerlessness is negatively related to change willingness. Table 4 shows that operational powerlessness indeed lowers the change willingness of locomotive engineers. This relationship is statistically significant (p < .01). The B-value indicates that when operational powerlessness increases with 1 point on the 5-point Likert scale, change willingness decreases with .463 on the same scale. This effect is greater than the measured effect of the same relationship in the article of Tummers (2011), where the effect was -.15. The empirical results thus support the first hypothesis.

The second hypothesis predicts that communication frames emphasising professional values of client service are positively related to change willingness. The communication frames emphasising professional values of client service are called change message in the analysis. Including this relationship in the regression analysis leads to an increase of the R squared with .200. This shows that the change message contributes to the explanation of the variance in change willingness for 20 percent in this regression analysis. The results show that change message is positively related to change willingness and that this relationship is statistically significant (p < .01). When change message increases with 1 point on the 5-point Likert scale, change willingness increases with .546 point on this scale. Hypothesis 2 can therefore be accepted.

Hypothesis 3 examines the moderating effect of the change message on the relationship between operational powerlessness and change willingness. Expected was an attenuating effect of the change message on the negative relationship between operational powerlessness and change willingness. Table 4 indeed shows that the interaction variable has a statistically significant effect on change willingness (p < .05). Including the interaction variable in the regression analysis increases the R squared with .004. Hence, it increases the degree of explanation of the variance in change willingness with 0.4 percent. This implies that the change

message has an effect on the negative relationship between operational powerlessness and change willingness. When the change message scores higher on communication frames emphasising professional values of client service, operational powerlessness has a less negative effect on the change willingness of locomotive engineers. The B-value of the interaction effect is .064. This means that with a level of 3 for operational powerlessness (the median of the 5-point Likert scale), the value for change willingness is .54 when change message is 1. However, with the same level of operational powerlessness, the value for change willingness increases to .67 when change message has a value of 5. This effect is shown visually in Figure 1, where the value for low change message is 2.611 (the mean minus the standard deviation) and for high change message 4.351 (the mean plus the standard deviation). The values for low and high operational powerlessness are respectively 1 and 5 (the range of answer possibilities). Here, one can see that the negative effect of operational powerlessness on change willingness is attenuated by a change message with a high score on frames emphasising the value of client service. This result supports hypothesis 3.

Figure 1

Interaction effect operational powerlessness and change message 1 1.5 2 2.5 3 3.5 4 4.5 5 Low operational powerlessness High operational powerlessness C ha ng e w ill in g nes s Low change message High change message

Table 4

Regression analysis with change willingness as dependent variable Model 1 B/se Model 2 B/se Model 3 B/se Model 4 B/se Intercept 5.069*** (.090) 1.654*** (.227) 2.069*** (.296) 1.737*** (.331) Operational powerlessness -.813*** (.042) -.290*** (.047) -.474*** (.097) -.463*** (.097) Change message .686*** (.043) .548*** (.077) .546*** (.077) Operational powerlessness x Change message .068* (.031) .064* (.031) Control variables Sex (1 = woman) -.039 (.112)

Sex (1 = I do not want to tell) -302

(.171) Sex (1 = other) -.160 (.282) Age .010** (.004) Tenure -.006* (.003) N 441 441 441 430 R2 .462 .662 .666 .674 Note. + p <.10; * p <.05; ** p<0.01; *** p<0.001 N = 441

Discussion

In this chapter, the implications of the results of the research for the theory will be examined. In addition, the limitations of the study will be discussed and suggestions for future research will be made. The conceptual model has been tested with a survey among 441 locomotive engineers implementing a new technological device. The results contribute to the existing literature on public change management in different ways, which will be discussed in this chapter and in the conclusion.

The results of this quantitative study are in line with the theoretical expectations. First of all, the hypothesis that operational powerlessness has a negative influence on change willingness has been confirmed by the empirical data. This means that when locomotive engineers feel that they have less influence on the implementation of the TimTim app, their willingness to change decreases. However, when they feel that the change has been communicated while emphasising the value of client service, this positively influences the locomotive engineers’ willingness to change. On top of that, the moderating effect of the change message on the relationship between operational powerlessness and change willingness is also supported by the evidence. This implies that the negative effect of operational powerlessness on change willingness can be attenuated by a change message that emphasises professional values of client service.

There are a number of limitations of this study that give rise to recommendations for future research. First, since the study is based on one context, the generalisability of the results towards other public domains is limited. Generalising results of change management research is challenging in general because of differences between organisations, public domains, and countries (Van der Voet, 2014, p. 379). Therefore, testing the theoretical expectations among public professionals working in different occupations may improve the generalisability of the results.

Moreover, since the study is cross-sectional, the issue of reversed causality remains a risk while interpreting the results. Although the causal inferences are based on theory, one cannot know with certainty that the relationships are in reality directed in the expected ways. Hence, a second recommendation for future research is to use a longitudinal research design while testing the theoretical expectations. This allows for examination of the temporal order of cause and effect (Bryman, 2012, pp. 63-65). To improve the internal validity further, a (quasi-)experimental design can also be valuable. A suggested direction for a quasi-experiment is to make two random groups of public professionals that have to implement a new policy which will likely increase their operational powerlessness. A pretest for change willingness can

help to control for pre-existing different prepositions towards change willingness in the two groups. Of these groups, one group does not receive a change message and one group receives a message that corresponds with the professional values of client service. Then, a survey in which the relationships are tested can be distributed among both groups. In this way, the hypothesised effects can be tested and compared with the values of the control group. Hence, both a longitudinal research design and an experimental research design can contribute to the internal validity of the results.

Additionally, this was the first study to examine the moderating effect of the change message on the relationship between operational powerlessness and change willingness. Although much research has been performed on the direct relationships of both independent variables, the moderating relation has not been researched before. As the interaction effect has been confirmed by the analysis in this study, but the measured effect was relatively small, the interaction between operational powerlessness and the change message provides an interesting direction for further research.

Finally, a recommendation for further research can be made concerning the psychological mechanisms about psychological ownership and the influence of attitudes on behaviour that have supported the formulation of the hypotheses in this research. While these insights from organisational psychology have contributed to the understanding of the mechanisms underlying the hypotheses, they have not directly been tested or measured in this study. In order to contribute even more to the upcoming field of behavioural public administration, testing the psychological mechanisms can be worthwhile. This can be done with qualitative research in which rich, deep data often contributes to the understanding of the meaning of action of people or with quantitative research in which scales are developed with which the underlying mechanisms can be measured (Bryman, 2012, p. 408).

Conclusion

The aim of this study was to examine the effect of operational powerlessness on the change willingness of locomotive engineers and the possible moderating effect of the change message. This research has used theories from the academic fields of public administration, change management, and organisational psychology. Based on the statistical analysis, conclusions can be drawn concerning the hypotheses and the answer to the research question. On the basis of these conclusions, points for practitioners and for NS specifically can be provided.

First, the empirical data has supported the hypothesis that more operational powerlessness leads to a decrease in the change willingness of locomotive engineers. This corresponds with the theoretical expectation that a lower perceived autonomy or sense of ownership leads to less willingness to change (Avey et al., 2009; Tummers, 2011). This outcome does not only have implications for the theory, but also for practitioners. Policymakers should keep in mind that the perceived autonomy of public professionals plays an important role in the degree in which they are willing to change.

Second, the way in which an organisational change is communicated also affects the change willingness of locomotive engineers. When the change message is perceived as one that emphasises the professional values of client service, this increases the change willingness of locomotive engineers. This is in line with the hypothesis that was developed based on the psychological insight that attitudes influence behaviour and that the norm of serving clients is a critical characteristic of public professionals (Ajzen & Fishbein, 1980; Andersen & Jakobsen, 2017; Glasman & Albarracín, 2006). When policymakers introduce a new policy, they should therefore consider the way in which the change communication is perceived by the public professional.

Third, the content of the change message can have an attenuating effect on the negative relationship between operational powerlessness and change willingness, when perceived by the locomotive engineers as emphasising the professional values of client service. The theories that led to the above-mentioned hypotheses have contributed to the formulation of this hypothesis. This moderating effect has not been researched before. Therefore, the confirmation of this hypothesis means that this outcome of the quantitative study has a direct implication for public change management theory. Since the exposed moderating effect has not been researched theoretically and empirically apart from this thesis, the field of public administration and organisational psychology should perform more extensive research to unravel the relationship including psychological mechanisms, as has been mentioned in the discussion. Additionally, policymakers may try to attenuate the negative effect of operational powerlessness on the