DETERMINANTS OF ARMENAKIS AND READINESS FOR
CHANGE; MODERATION?
Master thesis, Msc Business Administration; specialization Change Management University of Groningen, Faculty of Economic and Business
August, 2011 CAROLINE VAN VLIET Student number: 1799401 Emmalaan 29 1862 ES Bergen (NH) Tel: +31 (0)6 46077836 Email: [email protected] Supervisor/University Dhr. Dr. C. Reezigt Co-supervisor/University Mevr. Dr. J.F.J. Vos Supervisor/field of study Dhr. Ing. J. van Helvoirt
ABSTRACT
Readiness is the most important determinator for successful change (Smith, 2005). Employees are the key drivers of change; they are the ones who need to be willing to make the change. As the “ Medisch Centrum Alkmaar” is planning to make a change from the old hospital beds to new ones, the actual readiness for this change among nurses of the 6 largest departments, is examined. Are the nurses able to see whats in their for them? Do they have the idea that the change is supported by their managers? Do they have the belief that they can do it? In order to get answers to these questions, the influence of appropriateness, self efficacy, management support and personal valence (independent variables) on readiness for change is investigated. An extra investigation was done on `moderating` or `mediating` effects between each of the four determinants. Furthermore, the role of gender, age and educational level was explored. Are they moderators? The data for this research is collected by means of a questionnaire among the nurses of these 6 largest departments. The results showed that each single independent variable has a direct causal relationship with readiness for change. Self efficacy stays an important determinant in the multiple regression analysis. An extra investigation on “moderating” or “mediating” effects among the four determinants revealed that there are mediating effects present in the relationship between the determinants and readiness for change.
Key-words: readiness for change; appropriateness; self efficacy; management support; personal valence; mediating effects
TABLE OF CONTENTS
ABSTRACT ... 2
TABLE OF CONTENTS ... 3
INTRODUCTION ... 4
1.1 Research motive & objectives ... 4
1.2 Change project – proposed change ... 4
1.3 Organization ... 5
1.4 Research question ... 6
THEORETICAL FRAMEWORK AND CONCEPTUAL MODEL ... 7
2.1 Definitions ... 7
2.2 Timeline readiness for change - resistance ... 8
2.3 Model of Armenakis, Harris & Mossholder - 1993 ... 9
2.4 Model of Armenakis & Harris - 2002 ... 10
2.5 Model of Metselaar (1997/2005) ... 11
2.6. Holt et al. (2007) ... 13
2.7 Description four determinants ... 15
2.7.1 Appropriateness ... 15
2.7.2 Principal Support (management support) ... 15
2.7.3 Self efficacy ... 15 2.7.4 Personal valence ... 16 2.8 Personal characteristics ... 16 2.8.1. Gender ... 17 2.8.2. Age ... 18 2.8.3. Educational level ... 18 METHOD ... 20 3.1. Data collection I ... 20
3.2. Data collection II – quality of the scales ... 21
3.2.1. Factor & reliability analyses ... 21
3.2.2. Extreme values ... 21
3.3. Sample size ... 22
3.4. Actual response ... 22
3.5. Consequence for the research ... 22
3.6. Representativity ... 23
3.6. Data analysis ... 24
3.6.1 Recoding & transformations ... 24
3.6.2 Normal distribution ... 24
3.7. Correlation & regression analysis ... 25
RESULTS ... 27
4.1. Correlation analysis ... 27
4.2. Regression analysis ... 30
4.3 Moderating effects of the demographic variables; age & educational level ... 31
4.4. Further implications from the multiple regression analysis ... 32
4.5. Change readiness at “Medisch Centrum Alkmaar” ... 36
CONCLUSION ... 37
5.1. Acceptation of hypotheses ... 37
5.1.1. Appropriateness & readiness for change (H1) ... 37
5.1.2. Management support & readiness for change (H2) ... 37
5.1.3. Self efficacy & readiness for change (H3) ... 37
5.1.4. Personal valence & readiness for change (H4) ... 37
5.2. Hypotheses demographic variables ... 38
5.2.1 Age ... 38
5.2.2. Educational level ... 38
5.3. Fisher’s Z scores educational level ... 38
5.4. Mediating effects of the determinants ... 38
DISCUSSION ... 39
REFERENCES ... 43
APPENDIX 1: INVITATION LETTER QUESTIONNAIRE TO NURSES ... 47
APPENDIX 2: QUESTIONNAIRE NURSES ... 48
APPENDIX 3: POSTER ... 52
APPENDIX 4: FACTOR ANALYSIS ... 53
APPENDIX 5: EXTREME VALUES – BOXPLOTS ... 55
APPENDIX 6: NORMAL DISTRIBUTION ... 56
APPENDIX 7: CORRELATION ANALYSIS AGE AND EDUCATIONAL LEVEL & READINESS FOR CHANGE ... 57
AGE ... 57
EDUCATIONAL LEVEL ... 57
APPENDIX 8A: MODERATOR ANALYSIS AGE & EDUCATIONAL LEVEL ... 59
APPENDIX 8B: MODERATOR ANALYSIS OF THE DETERMINANTS ... 61
APPENDIX 9: MEAN SCORES & STANDARD DEVIATION ... 65
INTRODUCTION 1.1 Research motive & objectives
This research is primarily conducted in the interest of the project group at Medisch Centrum Alkmaar which is carrying out the project, “Replacement hospital beds”. As the former hospital beds were no longer in line with current quality requirements, it became necessary to replace the previous type of bed. This lack of quality regarding the hospital beds was highlighted by the publication of the “Algemeen Dagblad Ziekenhuis Top 100” on 7th September 2010. In this publication, Medisch Centrum Alkmaar did not score sufficiently on “percentage of patients with bedsores”. However, before this fact on bedsores was publicised, a project group had already been created to take charge of the project, “Replacement hospital beds”. As the Medisch Centrum Alkmaar has a capacity of around 700 hospital beds (formal recognition of 900 hospital beds), a totality of 700 new hospital beds need to be implemented. Since it is not possible to provide all the 700 new hospital beds at once, the implementation consists of two phases:
Phase 1 (end of 2009 and beginning of 2010): the installation of four hundred new hospital beds. Phase 2 (end of 2010 and beginning of 2011): the installation of the remaining three hundred new hospital beds.
The focus of this study will be on Phase 2, because Phase 1 was already executed (end 2009/beginning 2010). For Phase 2, the research objectives will be two-fold. The first research objective is to determine the level of readiness for change among the hospital employees involved in phase 2 of the project. The change will involve the entire personnel of the MCA but the focus will be on the six largest departments. These departments include 291 nurses. Are they ready for the proposed change? With the information obtained on the level of readiness for change for phase 2, the project group can be advised on the optimization of this readiness level before phase 2 will be given its definite go-ahead. The determination of readiness for change holds scientific value in itself. However, emphasis will be placed on a further and deeper understanding of the factors influencing readiness for change and what characteristics relating to personnel should be taken into consideration.
1.2 Change project – proposed change
demonstrate their products. After a trial period and intensive use of different beds, one bed was chosen. That particular bed was the best option in terms of quality and economic aspects. The new beds come with more functions and gadgets that will make the work of the nurses easier. Nevertheless, they need to learn a new routine. Employees will be faced with a new product which has different possibilities and requires different instructions for the patient. However, before instruction of the patient may take place, the nurses and other personnel themselves need to be instructed by so-called “bed coaches”. These “bed coaches” are better known in the hospital as the “ergo coaches”. The “ergo coaches” are specialised in ergonomics. One of their most important duties is to notice bottlenecks in the area of physical effort and strain caused to the personnel. These “ergo coaches” will be instructed by the supplier on the usage of the new bed in order to be able to instruct the personnel in the best possible manner. It can be said that a different way of working will be needed for the personnel involved when the new beds are in full operation. For example, the new bed is operated electrically, so less effort is required from the personnel in controlling the bed. The patient has his own remote control to raise and lower the bed. Another example is the nurse being able to move the new bed by her/himself because the reaction speed is much quicker in comparison with the old bed. To move the old bed still requires two persons. Both examples do have a negative and positive aspect concerning the personnel. The negative aspect is approaching the proceedings and duties that nurses have been used to for many years. “I have always done it like this”, being part of their habits. The positive aspect about the new bed is that duties previously performed by the nurses will now be supported by the new hospital bed.
1.3 Organization
1.4 Research question
To follow up what has been described in section 1.2, “Change project – proposed change”, the employees of the MCA are confronted with the replacement of the old hospital beds and have to deal with the introduction of new hospital beds. For the nurses working in the six largest departments of the MCA, another way of working is required and different routines should be adapted. The nurses of the six largest departments are ready for the proposed change when they have the feeling the change of old to new hospital bed is needed and that themselves and the organization have the ability to enact the change. This level of readiness can be influenced by four factors; (1) the feeling that the change is relevant (appropriateness) (2) there is sufficient support from management (management support) (3) there are personal benefits (personal valence) (4) having the belief that they are able to succeed in executing the tasks and duties associated with the change (self efficacy). Additionally, the readiness for change might also depend on gender, age or educational level. A nurse who achieved a higher education, might be closer to the “ right” information about the change and in most cases might have has a greater ability to make a sound consideration. The main objective of this research is to find out how the readiness for change for phase 2 of the project “Replacement hospital beds” can be influenced and what personnel characteristics should be taken into account.
The main research question is therefore formulated is as follows:
“How can readiness for change be influenced and which personnel characteristics should be taken into consideration?”
In finding a response to this main research question, two sub-questions will be posed at the same time:
“Which factors influence readiness for change?” &
“ Which personnel characteristics should be taken into consideration, both when measuring and when influencing readiness for change?”
To extend the above formulation, a data-related question can be created:
“ To what extent are the nurses of the six largest departments of the MCA in the project “Replacement hospital beds” ready for the last phase of the implementation of the 300 new hospital
In order to provide answers to these questions, a literature study was performed, followed by data collection and analysis. The next chapter “Theoretical Framework & Conceptual Model”, presents this literature review, in which the main streams in “readiness for change” and their authors are described. Additionally, a description of different personnel characteristics is presented.
THEORETICAL FRAMEWORK AND CONCEPTUAL MODEL
The topic readiness for change has been extensively examined in the literature throughout the years (Lewin (1951), Armenakis et al. (1993, 2002), Holt et al. (2007)). The fact that so many authors wrote about readiness for change, confirms that it is an important issue which cannot be ignored in the arena of “change management”. Employees should be treated well, because they are the key drivers of change. They are the ones that need to be willing and motivated to make the change. In order to make the execution possible, employees need to be ready. Without their cooperation, it will be almost impossible to implement the change and resistance will occur. Jones et al. (2005) stated that in most cases “organizations move directly into change implementation before the individual or group to be changed is “psychologically” ready .” These organizationas look at the capacity of the organization and if it’s a go, they move into the implementation of the change. It is also very important to look at the feelings and beliefs the nurses have towards the change. They need to feel that the change is relevant and necessary.
2.1 Definitions
2.2 Timeline readiness for change - resistance
The exact origin of the term “readiness for change” is not easy to trace, because in the early literature the term is not mentioned very extensively, whereas the relatively “negative” association “resistance” is mentioned more frequently. Resistance is often mentioned in the sense of the “unavoidable behavioral response” towards a change initiative (e.g. Lewin, 1951; Coch & French, 1948; Gray & Stark, 1984). Different views dominated the literature of resistance and explained this phenomenon (Metselaar, 1997). According to the political view, resistance to change occurs when there are fluctuations in the existing balances of power between interest groups. The social view explains resistance as the consequence of social interaction due to different norms and attitudes. In the psychological stream, resistance is clarified as the outcome of balancing between change and stability and is referred to as a “tendency towards homeostasis” (Olthof, 1985). The focus of a variety of research is working against and overcoming resistance by following different strategies and methods (Metselaar, 1997). These different strategies and methods can be demonstrated by Kotter & Schlesinger (1979) with examples such as communication, participation facilitation and negotiation, extended by manipulation and coercion by King & Anderson (1990). For example, in earlier years, Coch & French (1948) did the first experiment in reducing resistance to change by regulating employee participation in change projects to different extents. It was proven that this type of participation led to reduced resistance to organizational change. Some years after the experiment of Coch & French (1948), Lewin (1951) identified the “unfreezing” phase in his three-step model of change as “readiness for change”. Unfreezing means that before new behavior can be accepted, there should be destabilization of the forces of inertia and the estrangement of old behavior. This was actually a positive view at that time when literature was definitely dominated by the term “resistance” as a reaction towards change.
organization. Resistance is often already recognized when employees are expressing their opinions and views of the truth and of the organization they want to work in.
In the years between 1993 and 2009, the literature about readiness for change was dominated by a few key authors, namely Armenakis et al. (1993, 2002), Metselaar (1997/2005) and Holt et al. (2007). The focus of the following section will be on these three key authors and their models.
2.3 Model of Armenakis, Harris & Mossholder - 1993
Armenakis et al. (1993) state that readiness for change is one of the factors contributing to the effectiveness with which a change plan is implemented in the organization. The main purpose of the model is explaining how change agents can influence employees’ readiness during organizational change. In the theoretical basis of the model, emphasis is placed on the fact that readiness for change does not only involve convincing individual beliefs, but goes much further than that. It is seen as a social event. According to the social-information processing model of Griffin (1987), individual readiness may also be shaped by the readiness of others.
Figure 1: Creating readiness for change
The contextual factors are the reasons why the organization started thinking about change; change in governmental regulations, increased competition etc. To communicate this change to the employees, a message need to be sent out. According to Armenakis et al. (1993) this message should contain two important aspects; the need for change and the individual and collective efficacy. The need for change is the gap between the actual situation and the desired end situation. Katz & Kahn (1978) describe
Contextual Factors
The message: discrepancy & efficacy
individual and collective efficacy as creating the belief that change is needed by showing how the current performance of the organization differs from some desired end-state.
The message sent out to the employees depends on the choice of the influence strategies (active participation, persuasive communication and management or external information). The interpersonal and social dynamics, together with the change agent attributes (experience, trustworthiness, reputation), influence the outcome of these strategies. It is important for a change agent to be aware of the distinction between individual and collective readiness for change. As mentioned before, it is not only a matter of influencing individual beliefs, but rather of convincing a collection of socially-interacting individuals according to discrepancy and efficacy of the message (Armenakis et al., 1993).
The last item in the model is called the readiness assessment. In simple words this means that the system in the organization should be ready to embrace change. “An assessment of the perceived discrepancy and efficacy of the target would be performed in gauging the state of readiness” (Armenakis et al., 1993 p. 690). In order for the right readiness for change programs to be implemented, they should follow the guideline of the level of urgency of change and the extent of readiness among employees (Armenakis et al., 1993). The model emphasizes that the area of necessary change should be identified first, followed by the design of the readiness program to influence the appropriate solution (individually/collectively) and finally to successfully implement the change
2.4 Model of Armenakis & Harris - 2002
Figure 2: Model of Armenakis & Harris (2002)
Armenakis et al. (1993) proposed, in the fundamentals of creating readiness for change, that readiness is “a precursor of resistance in and adoption of behaviors”. One of the steps in the model is assessment – “an assessment of the perceived discrepancy and efficacy of the target” (Armenakis et al., 1993). This step is intended to determine just how ready for change employees are before organizational changes are implemented. The assessment enables managers to identify gaps that may exist between their own expectations about the change initiative and those of other members of the group. If significant gaps are observed and no action is taken to remedy this discrepancy, resistance would be expected, and therefore, change implementation would be threatened. To take a critical view of the theory of Armenakis & Harris (2002), it can be said that both researchers, both in their model and in their questionnaire, assume that the five determinants of the change message (self efficacy, principal support, discrepancy, appropriateness and personal valence) determine the level of readiness for change. This assumption is not proven by Armenakis & Harris (2002). Additionally, readiness for change is not measured as an independent variable. However, research by De Wagt (2010) and Garcha (2010) indeed proved that these five determinants of the change message do have an influence on the level of readiness for change.
2.5 Model of Metselaar (1997/2005)
The diagnostic model for measuring readiness for change was designed by Metselaar & Cozijnsen (2005) and is better known as the DINAMO-model. This abbreviation stands for: Diagnostics Inventory for the Assessment of the willingness to change among Management in Organizations. As the name already explains, the model offers the opportunity to assess the willingness among both employees and managers. Here, the term “willingness to change” is used instead of “readiness for change”. Metselaar (1997, p.34) defines “willingness to change” as “a positive behavioural intention towards the implementation of modifications in an organization’s structure or work and administrative
Self efficacy Principal support Discrepancy Appropriateness Personal valence
processes, resulting in efforts from the organization’s side to support or enhance the change process”. Both definitions are very similar, therefore it should be appropriate to use them interchangeably.
The diagnostic model is based on the “Theory of planned behaviour” created by Ajzen (1991). Ajzen’s model consists of three variables (attitude, subjective norm & behaviour control) which influence both the intention of behaviour and behaviour itself. Attitude can be explained as the “outcomes the employee expects from the change process”. Metselaar & Cozijnsen (2005) define this “attitude” as wanting to change, divided into three more specific explanations: expected consequences for the work of the employee, the emotions that change roused in the employee (affective response) and additional value of the change for the organization. The subjective norm is defined as the “attitude of colleagues and managers towards the organizational change” and means the same as having to change. People in the employee’s circle are putting pressure on the decision that he must make (peer pressure). Last, perceived behavioural control can be described as the “amount of control the employee experiences over the change process” and is the same as being able to change. Factors including the experience the individual has of previous change projects and the availability of time and resources to invest in the change determine this perceived behavioural control.
Figure 3: DINAMO model of Metselaar
the willingness to change, such as experience, knowledge and skills. Both definitions overlap in that they are about being able to change.
2.6. Holt et al. (2007)
Holt et al. (2007) conducted a research about readiness because of several reasons. Firstly, to evaluate the existing instruments which assessed readiness for change prior to the introduction of the change. Secondly, according the outcomes of the evaluation, build a new instrument which “ measures readiness at individual level because change activities are initiated and carried out by individuals within organizations (Holt et al., 2007 p. 251). The evaluation of the existing instruments concluded that only two of the 32 instruments satisfied the standards concerning validity and reliability. As a result, a complete measurement model was suggested based on four factors already observed in the existing instruments; change process (the steps taken during the implementation), change content (the particular change initiative introduced and its attached chacateristics), change context (conditions and environment within which employees function) and individual attributes (differences between individuals in their reactions towards the change).
Figure 4: The relationship between content, process, context, and individual attributes with readiness As shown in figure 4 it can be said that “ readiness for change is a comprehensive attitude that is influenced simultaneously by the content, the process, the context and the individuals involved”. It is basically the beliefs which causes the readiness for change and consequently create the attached behaviors. In order to use the complete measurement model, change themes were identified aligned with the subjects in the model; self efficacy and personal valence (individual attributes), organizational valence (content), discrepancy (context) and management support (process). Formal tests were done on the items per factor and different adoptions and changes were made. One of these changes was with the factor of discrepancy. In the research of Holt et al. (2007) respondents found it difficult to
distinguish between discrepancy (need for change) and organizational valence. They saw it as a unitary construct. Therefore, the new factor of appropriateness was created.
For the sake of clarity, no attempt was made to combine the models, because this would have exceeded the reach of the current research. Instead, one model was chosen to measure the readiness for change. The model of Armenakis & Harris (2002) was selected because of several reasons. Firstly, the model is extensive and internationally recognised. This international recognition of the model is a positive attribute in comparison with the model of Metselaar (1997/2005), which only has achieved national recognition. Lastly, Holt et al. (2007) was used because they developed a scale to measure readiness for change. The accompanied questionnaire is widely accepted and valid.
By choosing the model of Armenakis & Harris (2002) for measuring readiness for change, both the independent and dependent variables were presented at the same time.
The factors of self efficacy, principal support, appropriateness and personal valence are the independent variables in this conceptual model, and the dependent variable is the “readiness for change” studied here. As already stated before, discrepancy is included in “ appropriateness” so will therefore not be an item to investigate and will not be included in the research.
The first part of the conceptual model is outlined in figure 4.
Figure 4: Part one of the conceptual model.
Appropriateness
Management support
Self efficacy
Personal valence
2.7 Description four determinants
This section will contain a more detailed description of the four determinants and how they influence readiness for change. Moreover, as presented in the previous discussion, the element of the conceptual model presented on the left-hand side will be explained, followed by the hypotheses.
2.7.1 Appropriateness
According to Armenakis & Harris (2002), appropriateness can be explained in terms of an employee agreeing with the need for change but not agreeing with the proposed change. Thus, the definition of appropriateness highlights the importance of the type of change matching with the organization. It is important to convince the employees that the proposed change is relevant. A future perspective should be drawn in which the appropriateness for the organization is shown. By convincing the employees that the proposed change is appropriate, the levels of readiness will increase.
Hypothesis 1: The more positive the appropriateness, the higher the levels of readiness for change. 2.7.2 Principal Support (management support)
Principal support can be explained as the extent of commitment of the principal towards the proposed change (Armenakis & Harris, 2002). This definition will be used in the research. If the commitment is not shown by the key players of the organization, the employees become sceptical and are unwilling to make the change. Kotter & Schlesinger (1997/2008) also show the relevance of principal support by stating that to overcome resistance, change managers need to be supportive. Metselaar emphasized this and used the term subjective norm in referring to principal support. Subjective norm is “the likelihood that important referent individuals or groups approve or disapprove of performing a given behavior” (1997 p. 47) In other words, when referent individuals or groups approve a proposed change initiative, it will have a positive influence on readiness for change. Therefore, principal support will positively influence readiness for change.
Hypothesis 2: Higher levels of principal support will lead to higher levels of readiness for change. 2.7.3 Self efficacy
to was described later on as the confidence of an employee to succeed in the execution of tasks and duties associated with the proposed change (Armenakis & Harris , 2002). This definition will be used in this research. Additionally, Bandura (1997) even suggested that “self-confidence lies at the heart of an individual’s incentive to act or to be proactive”. Self control factors is the term used by Metselaar (1997) referring to self efficacy. These self control factors are experience, knowledge and skills. The level of these self control factors, in combination with other factors, will determine the ”willingness to change”. Logic reasoning may say that the higher the belief in the ability to succeed in the execution of the tasks and duties which are linked to the change, may result in higher levels of readiness. Therefore, the following hypothesis can be formulated.
Hypothesis 3: A higher level of self efficacy will lead to higher levels of readiness for change. 2.7.4 Personal valence
Together with self efficacy, personal valence can be covered by the concept of individual attributes in the research of Holt et al. (2007). He says that “because of the differences between individuals, some employees are more inclined to favor organizational change than others may be” (Holt et al., 2007 p.234). In most cases individuals question themselves: “What’s in it for me?”. Therefore, the personal benefits should be clearly highlighted when making the decision to respond positively to proposed change. In this research the definition of Armenakis et al.“ reflects the belief that the change is beneficial to the change recepient; there is something of benefit in it for them” (2009 p. 129) will be used. Both the models of Armenakis et al. (1993) and Metselaar (1997) recognized the presence of personal valence being connected to the readiness for change/willingness to change. Metselaar (1997) assumed that positive work-related outcomes will positively influence the willingness to change. Furthermore, when the affective response is more positive (level of excitement), the level of readiness for change will increase.
Hypothesis 4: The more positive the personal valence, the higher the levels of readiness for change. 2.8 Personal characteristics
literature. Research is performed on age, gender and educational level in combination with readiness for change. However, gender, age and educational level are never treated as moderators which influence the relationship between the four determinants and readiness for change. Secondly, the hospital as company type has not yet been used (again in combination with the four determinants and readiness for change). The three demographics can be labeled as moderators. According to Baron & Kenny a moderator can be defined in general terms as “a qualitative (e.g. sex, race, class) or quantitative (e.g. level of reward) variable that affects the direction and/or strength of the relation between an independent or predictor variable and a dependent or criterion variable” (1986 p. 1174).
The following three paragraphs will contain the description of gender, age and educational level also followed by their hypothesis.
2.8.1. Gender
The demographic variable gender is the first moderator under investigation. Gender can be defined as the “properties that distinguish organisms on the basis of their reproductive roles” (synoniemen.net). Literature concerning the role of gender in attitudes towards change is quite inconsistent. Many researches, like the one from Cordery et al. (1991) or Decker et al. (2001) all investigated the direct relationship between gender and change. However, these researches do not count as evidence for gender having a moderator role. The researches which does show gender as a moderator role is the one from Ong & Lai (2004). Ong & Lai (2004) were exploring gender differences in perceptions and relationships among dominants affecting e-learning acceptance (self efficacy, perceived usefulness, perceived ease of use and behavioral intention to use e-learning). It was shown that the men’s ratings on these determinants for e-learning acceptance were all higher than the women’s ratings. It can be said that a positive/negative attitude towards a certain change depends on the subject of the change (men’s ratings are higher in e-learning acceptance because in general they possess more skills, knowledge of and interest in the topic) and on the consequences it brings along with the change. The above explanation can be emphasized by Bussy et al. (1984), who indicated that the level of self efficacy a man or woman experiences in relationship to the readiness for change depends on the possession of personal attributes, the type of tasks and duties that need to be performed and what is socially acceptable. As a result, the following hypothesis can be formulated:
2.8.2. Age
Much research is done in the arena of the demographic variable age in relation to the participation in developmental activities. An example of where age is the moderator can be found in the research of Netz & Raviv (2004). A moderator effect can be defined as “influencing the strength of the relationship between two variables” (Barron & Kenny, 1986). Netz & Raviv (2004) investigated the individual’s motivation to engage in physical activity by examining age, gender and level of education of 2298 Australians aged 18 to 78 in relation to three postulates (self efficacy, outcome expectations and self evaluated satisfaction or dissatisfaction). The results showed that older individuals felt lower self efficacy in relation to physical activity and expected fewer benefits from participating in physical activity. Other researchers did investigate on a direct relationship between age and learning (Martocchio, 1994) or the effect of age, gender and tenure on readiness for change (Rhodes, 1983). Even Weber & Weber (2001) did not find a relationship between readiness for change and age. However, these investigations on direct relationships with age do not count as evidence for age as a moderator. Therefore, taking the research of Netz & Raviv (2004) as the source of information, it can be said that age will have a negative effect on the relationship between the four determinants and readiness for change. The following hypothesis can be defined:
Hypothesis 6: Age moderates the relationship between the four determinants and readiness for change: the relationship is less positive when age increases, as opposed to when age decreases 2.8.3. Educational level
was that readiness for change is not only influenced by contextual and process factors of the employees’ work environment, but that individual-level qualities also play a role. As a result, it might be said that the relationship between the four determinants and readiness for change is not moderated by educational level, but by self efficacy. In other words, self efficacy (the confidence to have power over the duties and tasks that need to be executed because of the change) influences the level of education and at the same time influences the level of readiness for change. Taking into consideration the above mentioned arguments, the following hypothesis can be formulated:
Hypothesis 7: Educational level moderates the relationship between the four determinants and readiness for change; the relationship is less positive when the educational level is low, as apposed to when the educational level is high.
With this last hypothesis, the middle element of the conceptual model is presented, which leads to the overall presentation of the entire conceptual model.
Figure 5: The total conceptual model.
The following section will be the “Method” section, which describes in which way the data were gathered and how these data were analysed afterwards.
Self efficacy
Principal support
Appropriateness
Personal valence
Readiness for change
H5: Gender (=) H6: Age (–)
METHOD
The model of Armenakis & Harris (2002) was used in order to measure the readiness for change of the nurses of the six largest departments at the “Medisch Centrum Alkmaar” before the second phase of project (implementation of the remaining 300 new hospital beds). Again, the model does not measure readiness for change as an independent variable but as a dependent variable. As a result, four items from the research of Metselaar (1997/2005) were added to the already existing items of Holt et al. (2007) to represent readiness for change as an dependent variable (see Appendix 2). Furthermore, discrepancy was included in the item “appropriateness” (see Holt et al. 2007 in the theoretical framework). As a result, the five determinants (of Armanakis) will be “the four determinants” (of Armenakis). The majority of the statements in the questionnaire (translated into Dutch) created by de Wagt (2010) were used (see Appendix 2). Adaptations were made in order to suit this specific research. The control questions about gender, age and educational level were added to represent the role of the moderators.
3.1. Data collection I
The questionnaire consisted of 29 statements, which covered the range of the subjects: self efficacy, appropriateness, management support, personal valence and readiness for change. The items about readiness for change were, as already said, taken from Metselaar (1997/2005). On each item the respondent was able to answer using a 7-point Likert scale, varying from 1= strongly disagree and 7=
strongly agree. The website of Explora (research program of MCA,
to the location of hanging up the poster in order for them to be spotted immediately by the nurses. A month after the distribution of the questionnaire the response rate was still low. Therefore, an extra month was taken to collect more questionnaires. Within this additional time close contact was held with the departmental supervisors and nurses.
3.2. Data collection II – quality of the scales 3.2.1. Factor & reliability analyses
To determine whether statements could be combined into one concept, a factor analysis and a reliability analysis were performed (see 3.6.1 & Appendix 4). When the total variance percentage and Cronbach’s Alpha were satisfactory, the items could be joined. In the following table, an overview is presented of the findings of the factor analysis and the reliability analysis.
Factor analysis Total variance (%) explained Reliability analysis Cronbach’s Alpha Appropriateness 72,592 0,935 Management Support 69,777 0,888 Self efficacy 66,592 0,864 Personal valence 77,291 0,848 Readiness 78,984 0,911
Table 2: Overview of the findings of the factor analysis and the reliability analysis
The values of the total variance and Cronbach’s Alpha were satisfactory, so we can say that all items of the concepts were combined successfully.
3.2.2. Extreme values
3.3. Sample size
Many different groups in and around the hospital Medisch Centrum Alkmaar will be affected by the proposed change – replacement of the old hospital beds by the new ones – because it will require new ways of working. In total, 3500 employees are working at the MCA. However, in order to narrow down the research, the focus will be on the nurses in the six largest departments of the hospital: Surgery, Neurology, Cardiology & Long, Orthopaedics, Intensive intake and Geriatrics. The reason for this choice is that nurses are the ones who work in high frequency with the hospital beds. They are the ones most affected by the change. Additionally, these six largest departments are the areas in which the new hospital bed was first introduced. As already mentioned earlier, the total amount of nurses (291) from six departments received a digital invitation through their team leader to fill in the questionnaire. Initially it was calculated what minimum number of questionnaires should be filled in to get an acceptable accuracy (0,05) and reliability (95%). We calculated that 146 nurses should have responded to the questionnaire (formula Foreest Medical School, 2011).
3.4. Actual response
So, if 146 nurses filled in the questionnaire, an acceptable accuracy (0.05) and reliability (95%) could be achieved. However, the actual response was 89 nurses (30.5% response rate). This actual response was reached after the three initiated actions for getting more respondents; reminders through emails, informative speeches on every department and an informative poster. The time frame for these former mentioned actions was almost two months. A reason for this time frame was to be able to get a response as high as possible in order to be able to clearly distinguish (in demographic variables) between the different sub-groups. However, the amount of respondents could not be increased so a choice was made to keep the reliability level of 95% and make an adaption in the accuracy level. The calculated accuracy appeared to be 0.087 instead of 0.05. As a consequence, the results should be interpreted with care and be read with caution. For example, if 50% of the nurses are ready for the proposed change, it actually means that this percentage lays between 41,3% and 58,7% with a reliability of 95%. The accuracy level of 0.087 only counts for the whole group of 89 respondents.
3.5. Consequence for the research
years old, 34-49 years old, 50-65 years old and > 65 years old. As 10 nurses were in the category of 50-65 years old, no nurses above 65 years old, 27 nurses between 34-49 years old and eventually 44 nurses between 18-33 years old, I have chosen to make two sub-groups out of them. A reason for this is that a better comparibility can be made between a “younger” group of nurses between the 18-33 years old and a “older” group of nurses > 33 years old (three other sub-groups are included). In total the “younger” group contained 44 nurses in the age category of 18-33 years old and the “older” group of nurses contained 37 nurses > 33 years old. So two sub-groups containing more or less the same amount of respondents. Hypothesis 7 is about educational level moderating the relationship between the four determinants and readiness for change: the relationship is less positive when the educational level is low, as opposed to when educational level is high. The same removement, as was done at the demographic variable age, counted for the educational level. Also the nurses could choose from four educational level sub-groups; MBO (V), HBO (V), Post HBO (V) and Master. In total an amount of 6 nurses achieved the Post HBO (V), 1 nurse a Master, 32 nurses HBO (V) and 50 nurses MBO (V). As the four groups did not contain equal amounts of respondents, there was chosen to make two sub-groups out of them. The first group which covered the MBO (V) education with 50 nurses and the second group named MBO (V) PLUS included; HBO (V) with 32 nurses, Post-HBO (V) 6 nurses and Master 1 nurse (in total 39 nurses).
3.6. Representativity
The population was defined as the total amount of nurses working in the six largest departments in the MCA. Furthermore, the sample was defined as the group of nurses who responded to the questionnaire. Population data and sample data were compared with respect to the representativity. The data concerning the educational level of the population was not available at the human resource department of the MCA.The only information which could be obtained is that generally nurses either are MBO (V) or HBO (V) educated. Unfortunalely the distinction for every nurse separately was not registrated.
Representativity
Population Sample P-value
Gender (Male/Female) 12 (4,1%);279 (95,9%) 7 (7,9%);82 (92,1%) 0,076
Age (18-33, 34-49, 50-65)
135 (46,4%);109 (37,5%);47
(16,2%) 48 (53,9%);30 (33,7%);11 (12,4%) 0,330
Educational level (MBO (V),HBO
(V),Post HBO (V) & Master) Unknown 50(56,2%);32(35,9%);6(6,7%);1(1,1%)
Not applicable
Table 3: Representativity
3.6. Data analysis
Version 16.0 of SPSS was used to investigate whether the aforementioned hypotheses in the literature section were supported by statistical evidence.
3.6.1 Recoding & transformations
With the use of different statements in the questionnaire the subjects appropriateness, principal support, self efficacy, personal valence and readiness for change were measured. To be able to investigate whether the different statements can be combined into one concept, several steps were taken. Firstly, reverse coding was applied; the negatively formulated statements were recoded (statements 2, 8 & 9 of appropriateness, statement 5 of management support, statement 2 of self efficacy and statement 1, 2 & 3 of personal valence). After this, a factor analysis was executed on all items, which resulted in the distinction of 6 components. However, as there are five factors, SPSS was forced to divide the items into five components. The factor analysis showed that the 10 items of appropriateness could not be combined into one component. Instead, the rotated component matrix divided the items into 2 components. Therefore, the not-correlating items (2, 8 & 9) were eliminated and it was possible to combine the remaining items into one variable using sum scales. The same held for management support and self efficacy. On the basis of the sum scales, a reliability analysis was done. Detailed information of the factor and reliability analysis can be found in Appendix 4.
3.6.2 Normal distribution
In order to investigate whether the sum scales of the concepts are normally distributed, the Kolmogorov – Smirnov test and the q-q plots were performed. Depending on the findings of these tests and plots, a specific type of test was used (nonparametric or parametric). Firstly, the Kolmogorov- Smirnov Z test was executed. The following table shows the findings for every concept of this test. The q-q plots can be found in Appendix 6.
Kolmogorov Smirnov test Kolmogorov-Smirnov Z
Asymp. P-value (2-tailed) Appropriateness 1,358 0,050 Management Support 1,239 0,093 Self efficacy 1,524 0,019 Personal Valence 2,174 0,000 Readiness 1,366 0,048
Table 4: Findings of the Kolmogorov – Smirnov test
the null-hypotheses was rejected. The conclusion is that nonparametric tests should be used for the analysis of the concepts.
3.7. Correlation & regression analysis
Two types of analysis will be performed to analyze the stated hypotheses in the theoretical framework. To start, the bi-variate correlation analysis is used (Spearman’s Rank Correlation test) to check for correlations between the four determinants and readiness for change. Furthermore, to examine if age and educational level have influence on the relationship between each of the four determinants and readiness for change, the Kruskal-Wallis test needed to be executed. The second type of analysis that will be performed is the (simple and multiple) regression analysis. The simple regression analysis to examine the causal relationship between one of the independent variables (appropriateness, management support, self efficacy or personal valence) and the dependent variable (readiness). By the execution of the multiple regression analysis the causal relationship between multiple independent variables (appropriateness, management support, self efficacy and personal valence) and the dependent variable (readiness) is investigated. The multiple regression analysis is also used for the detection of three aspects; the moderating effects of the determinants (and if necessary the mediating effects), the moderator role of age and educational level and when the results show substantial impact of the distinction into sub-groups on the assumed hypothesis, the above described analysis will be performed again to compare those sub-groups.
For moderation as well as mediation, Baron & Kenny (1986) described conditions which should be met. Figure 6 presents the conditions that should be met for moderation and Figure 7 for mediation.
Model 1: Y = α1 + bX b must be significant
Model 2: Y = α2 + bX + cZ b and c must be significant Model 3: Y = α3 + bX + cZ + dXZ b and d must be significant, c not.
Independent variable (X) Moderator (Z) Dependent variable (Y)
Figure 7: Conditions for mediation
In order to test the moderator role of age and educational level on the relationship between each of the four determinants and readiness for change, interaction terms were created. The same was done for testing the moderating effects of appropriateness, management support, self efficacy and personal valence (Table 5 & 6).
Overview interaction terms
Age Educational level
Appropriateness Ageclassapp Edulevelapp
Management support Ageclassms Edulevelms
Self efficacy Ageclassse Edulevelse
Personal valence Ageclasspv Edulevelpv
Table 5: Overview interaction terms demographic variables age & educational level
Overview interaction terms
Appropriateness Management support Self efficacy Personal valence
Appropriateness apms Apse appv
Management support msap Msse mspv
Self efficacy seap Sems sepv
Personal valence pvap pvms pvse
Table 6: Overview interaction terms determinants
Model 1: Z= a1 + bX b must be significant Model 2: Y= a2 + cX c must be significant Model 3: Y= a3 + dX + eZ d & e must be significant
RESULTS 4.1. Correlation analysis
The bi-variate correlation analysis (Spearman’s Rank Correlation test) was used to investigate the correlations of the four determinants with readiness for change. From the following table it can be said that each determinant (appropriateness, management support, self efficacy and personal valence) correlate with readiness for change. All P-values < 0,05 and the correlation coefficients are between the 0,371 & 0,656. This was also expected as all determinants contribute to the concept of readiness for change as described by Armenakis (1993).
Correlations Appropriateness Managementsuppor t Self efficacy Personal valence Readiness Spearman's rho Appropriateness Correlation Coefficient 1,000 ,571(**) ,650(**) ,358(**) ,656(**) Sig. (2-tailed) . ,000 ,000 ,001 ,000 N 89 89 89 89 89 Managementsuppor t Correlation Coefficient ,571(**) 1,000 ,398(**) ,044 ,509(**) Sig. (2-tailed) ,000 . ,000 ,682 ,000 N 89 89 89 89 89 Selfefficacy Correlation Coefficient ,650(**) ,398(**) 1,000 ,470(**) ,649(**) Sig. (2-tailed) ,000 ,000 . ,000 ,000 N 89 89 89 89 89 Personalvalence Correlation Coefficient ,358(**) ,044 ,470(**) 1,000 ,371(**) Sig. (2-tailed) ,001 ,682 ,000 . ,000 N 89 89 89 89 89 Readiness Correlation Coefficient ,656(**) ,509(**) ,649(**) ,371(**) 1,000 Sig. (2-tailed) ,000 ,000 ,000 ,000 . N 89 89 89 89 89
** Correlation is significant at the 0.01 level (2-tailed).
Table 7: Findings of the Spearman’s Rank Correlation test
Both P-values can be found back in Appendix 7: readiness for change not significantly correlating with ageclasses but significantly correlating with educational level. The P-value of 0,045 assumes that the zerohypothesis – that the two sample sizes came from identical populations – was rejected; the readiness for change level of the two educational groups (MBO (V) and MBO (V) PLUS were not equal.
As a result, further research should be performed to be able to indicate where the differences lay between both educational groups because a substantial impact was detected in the level of readiness for change. Another correlation test will be performed for each group MBO (V) & MBO (V) PLUS individually (table 8 & 9). For the two different age groups no further action will be taken because the P-value was 0,625 > 0,05 and this means that the zero-hypothesis – that the two sample sizes came from identical populations – was accepted; the readiness for change level of the two age groups (18-33 years & > 33 years) were equal. Extended results on the two different age groups can be found in Appendix 7. Correlations Appropriateness Managementsupp ort Selfefficacy Personalvalenc e Readiness Spearman's rho
Appropriateness Correlation Coefficient 1,000 ,470(**) ,605(**) ,613(**) ,610(**)
Sig. (2-tailed) . ,001 ,000 ,000 ,000 N 50 50 50 50 50 Managementsuppor t Correlation Coefficient ,470(**) 1,000 ,320(*) ,257 ,390(**) Sig. (2-tailed) ,001 . ,024 ,072 ,005 N 50 50 50 50 50
Selfefficacy Correlation Coefficient ,605(**) ,320(*) 1,000 ,718(**) ,648(**)
Sig. (2-tailed) ,000 ,024 . ,000 ,000
N 50 50 50 50 50
Personalvalence Correlation Coefficient ,613(**) ,257 ,718(**) 1,000 ,464(**)
Sig. (2-tailed) ,000 ,072 ,000 . ,001
N 50 50 50 50 50
Readiness Correlation Coefficient ,610(**) ,390(**) ,648(**) ,464(**) 1,000
Sig. (2-tailed) ,000 ,005 ,000 ,001 .
N 50 50 50 50 50
** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).
Table 8: Correlation analysis for the educational group MBO (V).
Correlations
Appropriateness Managementsupport Selfefficacy Personalvalence Readiness Spearman's
rho
Appropriateness Correlation Coefficient 1,000 ,590(**) ,665(**) ,032 ,627(**)
Sig. (2-tailed) . ,000 ,000 ,845 ,000
N 39 39 39 39 39
Managementsupport Correlation Coefficient ,590(**) 1,000 ,443(**) -,221 ,424(**)
Sig. (2-tailed) ,000 . ,005 ,176 ,007
N 39 39 39 39 39
Selfefficacy Correlation Coefficient ,665(**) ,443(**) 1,000 ,127 ,566(**)
Sig. (2-tailed) ,000 ,005 . ,441 ,000
N 39 39 39 39 39
Personalvalence Correlation Coefficient ,032 -,221 ,127 1,000 ,208
Sig. (2-tailed) ,845 ,176 ,441 . ,204
N 39 39 39 39 39
Readiness Correlation Coefficient ,627(**) ,424(**) ,566(**) ,208 1,000
Sig. (2-tailed) ,000 ,007 ,000 ,204 .
N 39 39 39 39 39
** Correlation is significant at the 0.01 level (2-tailed).
Table 9: Correlation analysis for the educational group MBO (V) PLUS.
The above correlation analysis for the educational group MBO (V) PLUS, shows that appropriateness correlates with readiness for change with a correlation coefficient of 0,627, management support with 0,424 and self efficacy with 0,566. Personal valence does not significantly correlate with readiness for change (P-value of 0,204 > 0,05). Between the four determinants, almost all combinations correlate significantly except every combination with personal valence; personal valence & appropriateness (P-value of 0,845), personal valence & management support (P-(P-value of 0,176) and personal valence & self efficacy (P-value of 0,441).
To test whether the correlation coefficient of the MBO (V) group significantly differed from the correlation coefficient of the MBO (V) PLUS group, a Fisher Z test (two independent samples) had been executed. The calculation of this test was done using the webpage:
Educational level
Variable Group R N Variable Group R N
Appropriateness MBO (V) 0.610 50 Management support MBO (V) 0.390 50
MBO (V) plus 0.627 39
MBO (V)
plus 0.424 39
Fisher Z score: -0.12 Fisher Z score: -0.18
1tail P: 0.4522 1tail P: 0.4286
2tail P: 0.9045 2tail P: 0.8572
Self efficacy MBO (V) 0.648 50 Personal valence MBO (V) 0.464 50
MBO (V) plus 0.566 39 MBO (V) plus 0.208 39
Fisher Z score: 0.59 Fisher Z score: 1.32
1tail P: 0.2776 1tail P: 0.0934
2tail P: 0.5552 2tail P: 0.1868
Table 10: Fisher’s Z scores for educational level
From the above table it can be said that appropriateness and management both have a minus Fisher Z score of -0.12 and -0.18 respectively. The P-values for both determinants are > 0,05 (0,4522 & 0,4286) Self efficacy has a Fisher’s Z score of 0.59 and P-value of 0,2776. Last, personal valence shows a Fisher’s Z score of 1.32 and P-value of 0,0934. The idea of the Fisher’s Z score is that the Fisher’s Z score should be above the 1.96 and the P-value < 0,05 in order to say whether both groups significantly differ from eachother.
4.2. Regression analysis
The results of the simple regression analysis showed that each of the independent variables had a causal relationship with readiness for change. All P-values < 0,05.
Simple regression Readiness (dependent
variable) Appropriateness (Predictor) ANOVA Regression coefficient P-value 0,000 0,875 0,000 Management support (Predictor) ANOVA Regression coefficient P-value 0,000 0,651 0,000 Self efficacy (Predictor) ANOVA Regression coefficient P-value 0,000 0,794 0,000
Personal valence (Predictor) ANOVA
Regression coefficient P-value
0,001 0,425 0,001
Multiple regression Readiness (dependent variable) Appropriateness (Predictor) ANOVA Regression coefficient P-value (1-tailed) 0,262 0,073 Management Support (Predictor) ANOVA Regression coefficient P-value (1-tailed) 0,195 0,054 Self efficacy (Predictor) ANOVA Regression coefficient P-value 0,471 0,000 Personal valence (Predictor) ANOVA Regression coefficient P-value 0,106 0,224
Table 12: Multiple regression
Table 12 represents an overview of the results of the multiple regression analysis and shows that when all four independent variables (determinants) are put in the analysis at the same time, 65,5% of the variane can be explained. As you can see in the table above, self efficacy is the only determinant showing a causal relationship with readiness (P-value of 0,000 & regression coefficient of 0,471). Appropriateness, management support and personal valence show expected positive influences, however with P-values > 0,05 (ranging from 0,073 for appropriateness and 0,054 for management support to 0,224 for personal valence). This is quite surprising because all three variables showed strong and significant correlations with readiness for change. Management support and appropriateness are the closest to significance in comparison with personal valence. For this last variable the correlation analysis showed a correlation coefficient of 0,371 in comparison with the others while being significant (P-value of 0,000).
From the multiple regression analysis performed for the two sub-groups of age (18-33 years & >33 years) and educational level (MBO (V) & MBO (V) PLUS) results indicated that within the age group of 18-33 years only self efficacy was significant. In the age group of > 33 years, both management support and self efficacy were significant. Additionally, in the MBO (V) group self efficacy was significant and for the MBO (V) PLUS group both self efficacy and personal valence showed a significant contribution to readiness for change. Appendix 10A shows the detailed analysis.
4.3 Moderating effects of the demographic variables; age & educational level
as moderators in the relationship between each of the four determinants and readiness for change. The extended results can be found in Appendix 8A. For every determinant standardized values were computed, however these values did not show significant differences in the P-values in comparison with the already found P-values.
4.4. Further implications from the multiple regression analysis
Different results arise from the correlation analysis (Table 7) and the multiple regression analysis (Table 12). In order to find out which effects causes these different results, a moderator or mediator analysis should be performed. It would be interesting to find out which independent variable(s) influence(s) the model. For example, does it mean that self efficacy works as a moderator or mediator in the relationship between each of the determinants and readiness for change?
Logically wise I would label for example self efficacy as a moderator. Self efficacy might influence the strength of the relationship between each of the determinants and readiness for change. For example, the relationship between appropriateness and readiness for change might be stronger if the level of self efficacy would be higher. The strength of this relationship would be less if the level of self efficacy is lower. Self efficacy being a mediator would imply that self efficacy would explain the relationship between appropriateness and readiness for change. So when the effect of self efficacy would be removed, the relation between appropriateness and readiness disappears?
To test for moderating effects, the following conditions in the presented Figure 6 need to be met (Baron & Kenny, 1986). The standardized values for appropriateness, management support, self efficacy and personal valence were computed together with the products (interaction terms, see Table 6) of the possible combinations of these standardized values.
The performed multiple regression analysis did not show a significant contribution of the product variables on the dependent variable (readiness for change).
X*Z P-value
Appropriateness*management support 0,053
Appropriateness*self efficacy 0,251
Appropriateness*personal valence 0,580
Management support*appropriateness 0,053
Management support*self efficacy 0,174
Management support*personal valence 0,951
Self efficacy*appropriateness 0,251
Self efficacy*management support 0,174
Self efficacy*personal valence 0,153
Personal valence*appropriateness 0,580
Personal valence*management support 0,951
Personal valence*self efficacy 0,513
Extended results can be found in appendix 8A & 8B.
As there are no moderating effects, mediating effects could be present and should be tested for. According to Baron & Kenny (1986 p.1177) mediation is about the following;
“To test for mediation, one should estimate the three following regression equestions: first, regressing the mediator on the independent variable; second, regressing the dependent variable on the independent variable; and third, regressing the dependent variable on both the independent variable and on the mediator. To establish mediation, the following conditions must hold: First, the independent variable must affect the mediator in the first equation; second, the independent variable must be shown to affect the dependent variable in the second equation; and third, the mediator must affect the dependent variable in the third equation. If these conditions all hold in the predicted direction, then the effect of the independent variable on the depedent variable must be less in the third equation than in the second. Perfect mediation holds if the independent variable has no effect when the mediator is controlled.”
This explanation is presented in figure 7.
Figure 7: Conditions for mediation
In order to start looking for mediating effects, it would be helpful to do again some logical reasoning. Appropriateness might be a mediator in the relationship between self efficacy and readiness for change because the belief that you are able to execute the tasks and duties linked to the change will lead to a higher level that you find the change appropriate and this leads ot higher levels of readiness for change. Appropriateness might also be a mediator in the relationship between management support and readiness for change and in the relationship between personal valence and readiness. The higher
Model 1: Z= a1 + bX b must be significant Model 2: Y= a2 + cX c must be significant Model 3: Y= a3 + dX + eZ d & e must be significant