Change
The moderating role of change-specific cynicism and diverging combinations
of readiness for change dimensions across the change process
Master Thesis MSc BA – Change Management
University of Groningen, Faculty of Economics and Business
June 06, 2013 Lisanne Meurs Studentnumber: 1785435 Tuinbouwstraat 132a 9717 JR Groningen Tel.:+ 31 (0) 6 33 78 40 64 E-mail: lisanne_meurs@hotmail.com Supervisor University Drs. J.C.L. Paul Second assessor University
Dr. J.F.J. Vos
Supervisor field of study Dhr. M.A. Schotman
Factors Influencing Individuals’ Readiness for
Change
The moderating role of change-specific cynicism and diverging combinations
of readiness for change dimensions across the change process
ABSTRACT
In this study it was attempted to retest the instrument of Bouckenooghe, Devos, & Van den Broeck (2009), that was created for assessing the context of change, process factors of change and individuals’ readiness for change. To improve the integrity of the model, the factor “self-control” was added, and the variable “cynicism” was added to the context factors. Data were collected by distributing questionnaires among employees of company XYZ. Outcomes of multiple regression analyses showed that single context, process and self-control factors could have reverse effects on the different dimensions of readiness for change. Furthermore, it resulted that cynicism was the only factor which had a positive influence on an individual’s readiness for change to put energy in the change process. Besides this, it resulted from a continuous moderator analysis that cynicism was not just an antecedent to readiness for change, but it also had a moderating effect on the extent to which a person is triggered by specific actions of management to increase individuals’ readiness for change.
The results underscore the salience of distinguishing between the different dimensions of individual’s readiness for change: cognitive, intentional and emotional readiness for change. However, it resulted that the dominant presence of a dimension of readiness for change did not differ between the different stages of implementation. That is, irrespective of the stage of implementation, intentional readiness was highest, emotional readiness was lowest, and cognitive readiness was scaled in between them. Implications for managers were also highlighted.
TABLE OF CONTENTS
1. INTRODUCTION 5
2. THEORY AND HYPOTHESES 7
2.1 The concept of readiness for change 7 2.2 Antecedents of readiness for change 7 2.3 The moderating role of change-specific cynicism 9 2.4 Diverging combinations of readiness for change dimensions across
the change process 15
3. RESEARCH METHODS 18
3.1 Data collection 18
3.1.1 Case organization 18
3.1.2 Population 19
3.1.3 Questionnaire distribution 19 3.1.4 Sample and questionnaire administration 20
3.2 Measures 21
3.2.1 Dependent variables: readiness for change variables 21 3.2.2 Independent variables: process factors of change 21 3.2.3 Independent variables: internal-context factors of change 22 3.2.4 Independent variables: self-control factors 22 3.2.5 Independent variable: stage of implementation 23
3.3 Data analysis 24 3.3.1 Data examination 24 3.3.2 Factor analysis 25 3.3.3 Deriving factors 25 4.RESULTS 34 4.1 Descriptive statistics 35
4.2 Retesting Bouckenooghe et al.’s (2009) instrument: A multiple regression
analysis 36
4.2.1 Change process factors 38
4.2.2 Internal-context factors of change 39
4.2.3 Self-control factors 40
4.3.2 Cynicism as moderator of internal-context of change factors’
effects 45
4.3.3 Cynicism as moderator of self-control factors’ effects 46 4.4 Analysis of relationship between stage of implementation and diverging
combinations of readiness for change dimensions 47
5. DISCUSSION 50
5.1 Contributions 50
5.1.1 Contrasting effects of single factors on readiness for change 50 5.1.2 Single positive effect of cynicism on intentional readiness for
change 51
5.1.3 Impact of differences in history of change 52 5.1.4 Moderating role of change-specific cynicism 53 5.1.5 Stage of implementation and readiness for change 54 5.1.6 Assessing individual’s readiness for change in a quality
improvement project 55
5.2 Limitations 56
5.2.1 Bias of cultural context 56
5.2.2 Confusions about the respective change 56 5.2.3 Instrument is limited to Lean6Sigma studies 56 5.2.4 Snapshot in a change process 57 5.2.5 No preserving of the original factor structure 57
5.3 Managerial implications 58
6. REFERENCES 60
APPENDIX A: QUESTIONNAIRE 66
APPENDIX B: BOXPLOTS 74
APPENDIX C: DESCRIPTIVE STATISTICS 76
APPENDIX D: ASSUMPTIONS CHECK OF MULTIVARIATE ANALYSIS 77
APPENDIX E: REGRESSION COEFFICIENTS 92
1. INTRODUCTION
The majority of traditional organizations has recognized that they need to be receptive and open to change in order to survive in, and adapt to today’s fast-changing economy (Beer & Nohria, 2000). Change interrupts the normal patterns of organization, and thus asks recipients to enact new patterns, which entails interplay of, ambiguously, planned and emergent
processes (Mintzberg & Waters, 1985 in Ford, Ford & D’Amelio, 2008). Appropriate alteration of individuals’ on-the-job behaviors in organizations is essential in order to realize successful change over the long term (Jones, Jimmieson, & Griffiths, 2005; Meyer, Srinivas, Lal, & Topolnytsky, 2007; Weeks, Roberts, Chonko, & Jones, 2004 in Choi, 2011). To ensure successful change, organizations need to be in a continued state of change readiness (Rowden, 2001). Since the openness, preparedness and readiness for change of employees are
perquisites to realize this organizational readiness for change (Backer, 1995; Eby, Adams, Russell & Gaby, 2000 in Madsen, Miller and John, 2005), several researchers have
underscored the value of a more person-oriented focus in organizational change literature (Soumyaja et al., 2011). Subsequently, several instruments (Cunningham et al., 2002; Jones, Jimmieson & Griffiths, 2005) have been developed in order to assess individuals’ readiness prior to the introduction to change (Holt, Armenakis, Feild & Harris, 2007a). In this study the instrument of Bouckenooghe et al. (2009) was adopted, and thus a multidimensional view of readiness for change was used in which a cognitive, intentional and emotional dimension were distinguished. To improve the integrity of the model, the factor “self-control” was added, and the variable “cynicism” was added to the context factors.
First, the study was started by examining the direct relationship between readiness for change and its antecedents by using the adapted model of Bouckenooghe et al. (2009).
Second, it was attempted to extend the current research on readiness for change by addressing the moderating impact of change-specific cynicism. In other words, it was investigated
whether an individual’s readiness for change is influenced by the same antecedents irrespective of the level of change-specific cynicism.
using stage-inappropriate antecedents to readiness could be ineffective or even counter-productive (Weinstein, Rothman & Sutton, 1998 in Abraham, 2008). By using the
multidimensional concept of readiness for change, it was examined if different combinations of the dimensions of readiness for change could be distinguished across the change process.
2. THEORY AND HYPOTHESES 2.1 The concept of readiness for change
Readiness as a concept was first introduced by Jacobson (1957), and the design of readiness as a construct is incorporated within the creation of several theoretical models of the process of change (Choi, 2011). Although several researchers have addressed the relevance of readiness for change (Beckhard & Harris, 1987; Beer & Walton, 1987; Turner, 1982 in Armenakis, Harris & Mossholder, 1993), creating readiness has been most often explained in conjunction with prescriptions for reducing resistance (Armenakis, Harris & Mossholder, 1993).
Armenakis et al. (1993) attempted to clarify the readiness concept. According to Armenakis et al., (1993), “readiness is reflected in organizational members’ beliefs, attitudes, and intentions regarding the extent to which changes are needed and the organization’s capacity to
successfully make those changes. Readiness is the cognitive precursor to the behaviors of either resistance to, or support for, a change effort”. This conceptualization of readiness for change is still the most well accepted definition in literature (e.g., Bouckenooghe et al., 2009; Chonko, Jones, Roberts & Dubinsky, 2002; Eby, Adams, Russell & Gaby, 2000; Holt et al., 2007b; Madsen, Miller, John, 2005; Weeks, Roberts, Chonko, Lawrence & Jones, 2004).
Besides the instrument of Armenakis et al. (1993), several instruments (Cunningham et al., 2002; Jones, Jimmieson & Griffiths, 2005) have been developed in order to assess
individuals’ readiness prior to the introduction of change (Holt, Armenakis, Feild & Harris, 2007a). In the majority of those studies the topic is considered to be an one-dimensional construct (Cunningham et al., 2002; Jones et al., 2005 in Neves, 2009). More specifically, those studies considered readiness for change as a cognitive attitude of the change recipients towards change instruments. Besides this, those instruments appear to measure readiness from one of several perspectives, namely, change process, change content, change context, and individual attributes (Holt et al., 2007b).
process, context and individuals involved. Furthermore, besides cognitively, individuals would also be emotionally inclined to accept, embrace and adopt a particular plan to purposefully alter the status quo (Holt et al., 2007b). Holt et al. (2007a) and Armenakis, Bernerth, Pitts and Walker (2007) urged to build on the existent insights of the diverse research. They developed scales to gauge readiness for change at an individual level (Neves, 2009). Those instruments considered readiness for change as a multidimensional construct. Using a multidimensional definition to describe employees’ attitudes toward proposed changes was proposed by Piderit (2000). However, although Holt et al. (2007a) recognized that readiness for change reflects both a cognitive and emotional dimension, they did not distinguish between those components in their study.
Following Holt et al. (2007a), Bouckenooghe, Devos and Van den Broeck (2009) recognized the relevance of context of change, process of change and readiness for change in predicting change behavior and the lack of well-validated available measures to assess those
components. With the creation of their model, Bouckenooghe et al. (2009) assumed that “intentional, cognitive and affective reactions toward change come into play at different stages in the change process, and do not necessarily coincide”. Many organizational areas have traditionally been judged as an one-dimensional construct, however affect is virtually present in every aspect of organizational life (Barsade & Gibson, 2007). This
multidimensional approach of readiness for change was rated as a more accurate view of reality, since it is acknowledged that the reactions of recipients of change also involve affections and intentions besides just cognitions (Bouckenooghe et al., 2009, Oreg & Berson, 2011).
Bouckenooghe et al. (2009) assigned the following definitions to the separate dimensions of readiness for change: “Cognitive readiness for change is the beliefs and thoughts people hold about the change” (Oreg, 2006; Piderit, 2000). “Intentional readiness for change is the extent to which employees are prepared to put their energy into the change process”. “Emotional readiness for change is the affective reactions toward change” (Oreg, 2006; Piderit, 2000).
Bouckenooghe et al. (2009) follow this recognition and distinguish readiness for change in specific (i.e. emotional and intentional readiness for change) and general (i.e. cognitive readiness for change).
2.2 Antecedents of readiness for change
A wide range of factors has been identified that influence readiness for change (Oreg et al., 2011). The existing instruments measure readiness from one or several perspectives, namely change process, change content, change context and/or individual attributes (Holt et al. 2007b). In this study, the antecedents and its corresponding definitions identified by
Bouckenooghe et al. (2009) were used. The model of Bouckenooghe et al. (2009) was based on the human relations movement, and thus the interpersonal dimension was highly valued. This interpersonal dimension is represented by the context of change and process of change (Bouckenooghe et al., 2009).
Bouckenooghe et al. (2009) allocated seven dimensions to the antecedents of organizational readiness for change. Quality of change communication, participation, attitude of top
management toward organizational change, and support by supervisors pertain to the process of change. The factor “context of change” involves trust in leadership, cohesion and
politicking. Both the context and process of organizational change have been recognized to be major contributors of employees’ readiness for change (Holt et al., 2007b). Context of change was defined as “employees’ perceptions of the internal circumstances under which change occurs”, and the process of change was defined as the “actual approach of how a specific change project is dealt with” (Bouckenooghe et al. 2009).
FIGURE 1: Conceptual Model derived from Bouckenooghe et al. (2009).
The following hypotheses were derived from the study of Bouckenooghe et al. (2009):
Hypothesis 1a: The four process factors of change (quality of communication, support by supervisors, attitude of top management and participation) each have a positive influence on an individual’s readiness for change.
Hypothesis 1b: Two context-factors of change, trust in leadership and cohesion, each have a positive influence on an individual’s readiness for change.
Hypothesis 1c: Context-factor of change politicking has a negative influence on an individual’s readiness for change.
efforts fail due to underestimating the central role of individuals in the change process (Armenakis, Harris, & Mossholder, 1993; George & Jones, 2001; Greenhalgh, Robert,
Macfarlane, Bate, & Kyriakidou, 2004; Hall & Hord, 1987; Isabella, 1990; Lau & Woodman, 1995 in Choi, 2011). Besides this, some employees would favor organizational changes more than others, because of the differences between individuals (Holt et al., 2007a).
To improve the integrity of the model, self-efficacy was adopted in this study as an antecedent of readiness for change. Rotter (1966 in Metselaar, 2007) recognized that the extent to which a manager experiences control towards a change process is influenced by both external control factors and self-control factors. In this study, the related concept of Bandura (1986 in Metselaar, 2007) was used. Bandura (1986 in Metselaar, 2007) defined self-efficacy as “people’s beliefs in their ability to execute the courses of action required to attain certain levels of performance”. Self-efficacy has consistently been found to influence actions and emotions, and is one of the most widely referenced factors (Kirton and Mulligan, 1973; Paglis and Green, 2002 in Kuntz & Gomes, 2012). A positive relationship with an individual’s readiness for change was hypothesized.
Hypothesis 1d: Self-efficacy has a positive influence on an individual’s readiness for change.
In addition to self-efficacy, the role of cynicism was also incorporated into this study, since contradictory research outcomes were found regarding the relationship between cynicism and readiness for change.
Reichers, Wanous and Austin (1997) argue that cynicism can result from an organization’s history of change attempts and the tendency to consider things from a cynical perspective. Additionally, Choi (2011) notes that cynicism can result from “blaming those responsible for one’s pessimism”. Stanley, Meyer & Topolnytsky’s (2005) definition of change-specific cynicism was used in this study: “a disbelief of management’s stated or implied motives for a specific organizational change”. Change-specific cynicism is argued to determine a significant part of the variance of resistance (Stanley et al., 2005).
al., 1997). However, Stensaker & Meyer (2012) suggested that change-experienced people react more loyal to change initiatives, and that employees with limited change experience would react emotionally to the uncertainty and insecurity of changes. The loyalty of
experienced employees could be a result of either a person’s change capabilities or cynicism (Stensaker & Meyer, 2012). Change capabilities would allow individuals to realize changes in an efficient manner, since they are prepared for and open to change, and thus only resist the change in a limited way (Stensaker & Meyer, 2012). However, Stensaker & Meyer (2012) also argued that loyal behavior could result from cynicism, since those people are convinced that this is the least conspicuous behavior, as it creates the least attention and emotion.
Furthermore, the majority of the highly cynical people identified by Reichers et al. (1997) were willing to continue trying to make change. In other words, they suggested that cynicism does not necessarily produce resistance to change. Because of the contradictions in existing literature about cynicism, the direct relationship between cynicism and readiness for change was tested here.
Hypothesis 1e: The presence of cynicism has a negative influence on an individual’s readiness for change.
The adapted model is graphically displayed below.
2.3 The moderating role of change-specific cynicism
Several researchers have considered a direct relationship between change-specific cynicism and reactions to change (Devos, Vanderheyden, van den Broeck, 2002; Devos, Buelens & Bouckenooghe, 2007; Reichers et al., 1997; Stanley et al., 2005). However, studies examining potential other effects of change-specific cynicism are scarce.
Byrne & Hochwarter (2007) were the first who examined the non-linear relationship between organizational support and performance as moderated by employee cynicism. Recently, Barton & Ambrosini (2013) built on this finding, and addressed the moderating effect of organizational change cynicism on middle managers’ strategy commitment. They defined strategy commitment as middle managers’ willingness to support an effort to continue an organization’s strategic direction. In empirical research, strategy commitment is
acknowledged as a discrete form of change commitment (Ford et al. 2003 in Barton & Ambrosini, 2013). Barton & Ambrosini (2013) examined whether the presence of
organizational change cynicism would result in a reduction of the willingness of employees to support future change initiatives. It resulted that the relationships between the antecedents “senior management support” and “information availability” with strategy commitment were moderated by the presence of organizational change cynicism. However, the relationship between participation and strategy commitment was not moderated by the presence of organizational change cynicism. It resulted that both the perceived support for strategies by senior management and the information provided by senior management for change
implementation were positively related with strategy commitment. However, if middle managers had developed organizational change cynicism, their commitment was reduced and information provided by senior management for change implementation did not seem to enhance strategy commitment (Barton & Ambrosini, 2013). In other words, it resulted that the presence of cynicism disadvantaged the positive relationship between strategy commitment and its antecedents.
Barton & Ambrosini (2013) concluded that, in order to increase the understanding of
behavior, “future research should incorporate additional contextual variables as antecedents to strategy commitment with a corresponding exploration of how organizational change
generalized to other functional areas nor to other organizational levels (Barton & Ambrosini, 2013).
Since it was recognized that people could have different reactions to organizational change (Stensaker & Meyer, 2012); the preexisting level of cynicism about organizational change would differ (Wanous, Reichers, Austin, 2000); and a moderating role of change-specific cynicism was identified at middle managers in marketing positions, this study attempts to answer the question whether an individual’s readiness for change is influenced by the same antecedents irrespective of the level of change-specific cynicism by using the identified antecedents of Bouckenooghe et al. (2009) and self-efficacy.
In line with the results reported in the study by Barton & Ambrosini (2013), we suggest that change-specific cynicism would likely diminish the positive relationship of quality of communication, support by supervisors, attitude of top management, participation, trust in leadership, cohesion and self-efficacy with an individual’s readiness for change and increase the negative relationship between politicking and an individual’s readiness for change.
On base of the above reasoning, a conceptual model was created.
Those above graphically displayed relationships can be expressed in the following hypotheses:
Hypothesis 2a: The relationship between change process factors (quality of communication, support by supervisors, trust in top management, participation) and an individual’s readiness for change is moderated by specific cynicism such that higher levels of change-specific cynicism will be associated with lower levels of an individual’s readiness for change.
Hypothesis 2b: The relationship between two internal context factors of change (trust and leadership and cohesion) and an individual’s readiness for change is moderated by change-specific cynicism such that higher levels of change-change-specific cynicism will be associated with lower levels of an individual’s readiness for change.
Hypothesis 2c: The relationship between internal context factor politicking and an
individual’s readiness for change is moderated by change-specific cynicism such that higher levels of change-specific cynicism will be associated with lower levels of an individual’s readiness for change.
Hypothesis 2d: The relationship between self-control and an individual’s readiness for change is moderated by change-specific cynicism such that higher levels of change-specific cynicism will be associated with lower levels of an individual’s readiness for change.
2.4 Diverging combinations of readiness for change dimensions across the change process
Besides the lack of an unambiguous answer regarding the role that change experience plays in organizational change, Stensaker & Meyer (2012) also pointed that empirical evidence lacks about the development of reactions to change over time.
Prochaska, DiClemente and Norcross (1992) argued that change of behavior is characterized by a pattern of stages of readiness. Therefore, stage models are based on the idea that
the Contemplation stage, when someone is wondering if he/she needs to change, and the Action stage is when a person is ready to attend the change.
In line with the identification of Prochaska et al. (1992), Armenakis et al. (1993) argue that the use of a program for creating employee readiness for change should be guided by the urgency of the change and the extent to which employees are ready for the needed change. Those authors created four generic readiness programs, which are consistent with the
combinations of the extreme conditions of readiness and urgency, and are called: aggressive, crisis, maintenance, quick response.
Both in the study of Prochaska et al. (1992) and in the study of Armenakis et al. (1993) it was recognized that it is important that interventions are adapted to a person’s stage of readiness for change, since otherwise people would not gain maximum benefit from the program (Lam, Wiley, Siu, Emmett, 2010). However, in those studies readiness for change was considered to be an one-dimensional construct: a cognitive attitude of the change recipients towards a change instruments.
Bouckenooghe et al. (2009) assumed that “intentional, cognitive and affective reactions toward change come into play at different stages in the change process, and do not necessarily coincide”. By assuming this, they refer to the previous research from George and Jones (2001) and Piderit (2000). George & Jones (2001) analyzed the ambiguous way in which cognitive and affective processes operate in the sense making process. They proposed that emotional reactions determine the extent to which managers can adapt their strategy to the diverging circumstances they face. Emotion would put an individual in a state of readiness to deal with the discrepancy and would subsequently mobilize cognitive processing and behavior (George & Jones, 2001). However, it was stated that empirical evidence is needed to test this proposition. Besides George & Jones (2001), Piderit (2000) suggested that employees’ responses to change might evolve over time. She underscored that ‘paying attention to this evolution might yield insights about how to manage change initiatives successfully’. Again, empirical evidence about the existence of a variation in the combination of emotional, cognitive and intentional readiness across the change process lacks.
(Armenakis et al., 1993; Lam et al., 2010). Since it is recognized that readiness for change can be considered as a multifaceted concept (Bouckenooghe, 2009; Piderit, 2000), it is interesting to know if diverse combinations of readiness components in the different stages of
implementation can be distinguished. Although Piderit (2000) and George & Jones (2001) already suggested the expanding of the combination of readiness for change over time, empirical evidence about the existence of diverse combinations across the different stages of implementation lacks.
On base of the above reasoning, the following conceptual model was created:
FIGURE 4: Conceptual Model Correlation between Degree of Implementation and Readiness for Change
Those relationships can be expressed in the following hypothesis:
3. RESEARCH METHODS 3.1 Data collection
3.1.1 Case organization
This study was performed at a division of listed company XYZ, which is active in property development, residential building, non-residential building, technical services and
infrastructure. XYZ arose from a merger of company A and B, which core jobs were
respectively non-residential building and technical services. Organization XYZ operates in the Netherlands, Belgium and Germany. However, this case-study was based on its eight Dutch offices.
Change project
At the time of the merger in 2007, the technical services company (A) was a stable player and created positive results. However, the building company (B) experienced problems then. The organization was unstructured and employees preferred their department and/or office above company interests. After the merger the problems at company B continued, and it resulted that the executives of company A gradually became board members of the umbrella organization XYZ. Since the board assigned the positive business results of company A to the use of Six Sigma, it was argued that a similar program should be started at company B. This would optimize the different processes, and would result in cost-reduction and increasing customer value.
3.1.2 Population
Since this study is about individuals’ readiness for organizational change and the Lean6Sigma program is a companywide initiative, the relevant population consisted of all people working at those offices of company XYZ. Besides this, researchers have suggested that common factors that emerge from heterogeneous samples tend to provide a more general and complete understanding of a phenomenon (Sutton, 1987 in Bouckenooghe et al., 2009).
In sum, 1800 individuals are working for company XYZ. 600 employees originate from company B and 1200 individuals were employed by company A. It was attempted to represent all employees of company XYZ in this study. In order to differentiate between the functional levels, one of the general questions considered the functions respondents were in.
3.1.3 Questionnaire distribution
Before the questionnaire was distributed, a pilot-test was executed to ensure that the final version of the survey was clear and unambiguous. Firstly the chairman of the Lean6Sigma team, a green belt, and a HR manager were asked to revise the questionnaire individually. Based on this, the following actions were performed: The introductory text was rewritten neutrally without scientific terms; the purpose of the study was described more clear; and the used terms were aligned with the language use at company XYZ. Finally, a pilot-test was held among four operating managers and five mechanics on two sites. Those people tested the questionnaire on duration, formulation and understandability. Although they noticed that some questions were repeated in the questionnaire, no confusion was expressed about e.g. the difficulty or irrelevance of questions.
Data collection was carried out in two ways, namely 1. the distribution of a self-administered internet survey using Thesistools.com and e-mail, and 2. by distributing printed
questionnaires using regular mail.
With respect to the Thesistools.com questionnaire, the accompanying e-mail consisted of an introductory text to inform about the goal and relevance of the survey. Besides this, the e-mail contained the link which provided access to the questionnaire, and guaranteed, in order to prevent socially accepted answers, the anonymity of participation in this study. One week after the questionnaire was first distributed, employees received a reminder. Subsequently, one week later another reminder was distributed, and the first week after the end of the summer holidays a last call was done to complete a questionnaire.
employees working on the site. Those project leaders were asked to authorize the contribution of their workforce during work time. In addition to a customized amount of questionnaires, the package contained a manual for the project leader and a self-addressed envelope. The project leaders were requested to distribute the questionnaire five minutes before the end of the lunch break, and to ask people to resume their work after finishing the questionnaire. The survey comprised nine general questions and 68 closed questions. The closed questions were drawn in a random order. Participants were asked to spend only 10 à 15 seconds on each question, since people’s first impression was expected to be most valuable. The average time to complete the questionnaire was suggested to be approximately 15 minutes (see Appendix A for the questionnaire).
3.1.4 Sample and questionnaire administration
According to Hair, Black, Babin & Anderson (2010a), an ideal sample size would yield five respondents for each questionnaire item. The questionnaire (see appendix A) consisted of 77 items. Therefore, it would have been ideal to collect 385 correctly completed questionnaires. Since the organization gave permission to distribute only 450 questionnaires using the
organization’s email system, a random selection among the 1350 employees using business e-mail was made. 176 questionnaires were sent to different construction projects on different locations, and thus over one third of all mechanics got a questionnaire. In sum, 626 out of 1800 employees successfully received a questionnaire.
3.2 Measures
Multi-item measures were used to ensure adequate measurement of each variable. Except of the moderator “stage of implementation”, all variables were measured by pre-existing scales. The studies of Bouckenooghe et al. (2009); Stanley et al. (2005), and Metselaar (1997) each show sufficient evidence that their item-scales are valid and reliable. With respect to the variable “stage of implementation” a new measure was developed.
While the closed questions were imitated from existing studies, the open questions were specifically designed for this study. This latter kind of questions was created to distinguish between individuals based on gender, age, work experience, function, office, company and experience with Lean6Sigma. The closed questions had a 5-point Likert-scale ranging from “strongly disagree” to “strongly agree”. In order to create an accurate view of the current situation, all items were formulated in Dutch. Moreover, they were formulated in a way which would enable any employee to answer the questionnaire. That is, some items were adapted to the language use at company XYZ.
3.2.1 Dependent variables: readiness for change variables
The readiness for change variables were imitated from the study of Bouckenooghe et al. (2009). Subsequently, this study’s readiness for change factor comprises three dimensions: cognitive, intentional and emotional readiness for change. Each variable was measured by three questionnaire items. Based on the Exploratory Factor Analysis (EFA) and inter-item analysis, Bouckenooghe et al. (2009) concluded that the scale showed acceptable internal consistency. Based on a content validity test that both the emotional and intentional variables consisted of change-specific variables, and that the cognitive variable consisted of items that refer to change in general. Although the items that related to the emotional and intentional dimension were all positively formulated, the items that refer to the cognitive dimension of readiness were negatively formulated.
3.2.2 Independent variables: process factors of change
3.2.3 Independent variables: internal-context factors of change
The variables that Bouckenooghe et al. (2009) used for measuring the context of change were extended by one factor, namely change-specific cynicism. Change-specific cynicism was defined as “a disbelief of management’s stated or implied motives for a specific
organizational change” (Stanley et al., 2005). This definition is compatible to Bouckenooghe et al.’s (2009) definition of context of change: “employees’ perceptions of the internal circumstances under which change occurs”, since the actions of management are part of the internal circumstances wherein change occurs. Subsequently it was decided to classify change-specific cynicism as a context of change factor.
Stanley et al. (2005) validated eight questions that proved to be reliable for measuring change-specific cynicism. After an evaluation of those questions it was decided to omit items 4, 6 and 8 from this study’s questionnaire, since those were very similar to other items of Stanley et al. (2005). Subsequently, item number 1, 2, 3, 5 and 7 were used for measuring change-specific cynicism. Item 3 was recoded, because this item was stated in a positive way, which is in contrast to the hypothesized negative relationship between change-specific cynicism and readiness for change.
3.2.4 Independent variables: self-control factors
3.2.5 Independent variable: stage of implementation
Although seven general stages of Six Sigma implementation have been identified (Moosa, 2007b in Moosa & Sajid, 2010), it resulted from a literature review that no tool is present for calculating or measuring the degree of implementation of Six Sigma, since the
implementation of Six Sigma differs between companies, countries, locations and even from time to time (Moosa & Sajid, 2010). In order to assess the stage of implementation of change, it was thus needed to develop a specific measurement scale for this study.
This study’s scale consists of an objective and a subjective part. With respect to the subjective part, two members of the Lean6Sigma taskforce were asked to judge the extent to which each office makes use of Lean6Sigma. The objective scale was based on the company’s roadmap with respect to the Lean6Sigma program. In determining the degree of implementation at an office, the subjective and objective measurements of implementation were combined. The objective determination of implementation at an office was calculated as follows: the number of certified yellow belts and current & finalized Lean6Sigma projects were each divided by the number of employees at an office. Subsequently, those amounts were summed up and divided by four, since it covered the realized trainings and projects for both company A and B, and thus an average amount followed.
With respect to the subjective determination of implementation, two project members were asked to assess independently from each other the implementation rate of each office on a scale from one to three. This measurement-scale was related to Lewin’s three stages (1947): unfreezing, moving and freezing. Subsequently, if a project member assessed an office to be in the “unfreezing” stage, a score of one was granted, the “moving” stage was associated with a score of two, and if an office was assessed to be in the final “freezing” stage a score of three was granted. In order to calculate a comparative measure with respect to the subjective
determination of implementation, the ratings of the two project members were averaged. Subsequently, this outcome was divided by three (the maximum subjective determination score) and a ranking of the subjective implementation in percentage ratio resulted.
At the next step the objective and subjective determinations were combined by multiplying. However, since offices were characterized by a different number of employees and projects, comparisons between the offices were not possible yet. Subsequently, there was controlled for the size of the organization by dividing the “objective/subjective-combination score” of an office by the score which would have resulted if Lean6Sigma would have been fully
office; 2. using the maximum number of finished projects based on the company’s roadmap (16 for each office); and 3.using a score of three with respect to the subjective determination. A division of an office’s “objective/subjective-combination score” by its unique maximum score resulted in a relative degree of implementation.
3.3 Data analysis
In order to analyze the data gathered from the questionnaires, IBM SPSS Statistics 20 was used.
3.3.1 Data examination
The data analysis was started by identifying the cases that were characterized by too much missing values. In doing this, the following criterion of Hair et al. (2010a) was used: “variables with as little as 15 percent missing data are candidates for deletion, but higher levels of missing data, that is 20% to 30%, can often be remedied”. In sum, 101 cases out of 375 respondents were deleted, because of the refusal of answering statements. In addition to this, another 7 cases have been deleted from the data base, because those respondents had answered all questions about the change with the same answer, namely ‘neutral’. In order to ascertain the reliability of the ultimate, it was decided to delete those related cases. In addition to the cases, it was also evaluated if there were many missing values with respect to the variables. This was not the case, and thus no variables were omitted from the analyses. In addition to those missing values, several outliers were identified by graphically examining the data using a boxplot. With respect to cohesion and intentional readiness, striking boxplots resulted. However, since cohesion comprised only one item, and it seemed that the
questionnaires had been completed in a conscious way, the outliers were kept. See appendix B for those graphical representations.
After the deletion of those respective cases, 15 items were recoded. This means that these item-scores were adapted in order point in the same direction as all other items. Items 4 (quality of communication), 10 and 18 (participation), 25 and 26 (support by supervisors), 32 (trust in leadership), 36, 37 and 38 (cohesion) 44 and 45 (emotional readiness), 46, 47 and 48 (cognitive readiness) were all stated negatively, which was in contrast to the positive
item was stated positively, whilst change-specific cynicism had a hypothesized negative relationship with readiness for change.
3.3.2 Factor analysis
An exploratory factor analysis (EFA) was performed in order to examine the underlying patterns and relationships between the items and to determine whether the information could be summarized in a smaller set of factors (Hair, Black, Babin and Anderson 2010b). An absolute sample size of 100 or more observations is preferred to perform a factor analysis. Since this study comprised 267 completed questionnaires, this criterion was amply satisfied.
With the development of their instrument, Bouckenooghe et al. (2009) also performed a factor analysis. It resulted that the item-loading pattern endorsed the conceptual distinction between the three major categories: context, process and readiness for change factors (Bouckenooghe et al., 2009). The factors and items that were retained after EFAs; inter-item analyses; and content adequacy evaluation in the study of Bouckenooghe et al. (2009) were used here as a starting point.
This study’s conceptual model comprises three separate categories, namely process of change, context of change and readiness for change. A separate factor analysis was performed for each category. Process of change initially consisted of 28 items. Based on Bouckenooghe et al. (2009), it was hypothesized that this category would comprise four factors. Change-specific cynicism was added to the context of change, and it thus resulted that this category consisted of 16 items. Based on the study of Bouckenooghe et al. (2009), 13 items were related to readiness for change.
3.3.3 Deriving factors
cross-loadings, it was a candidate for deletion (Hair et al., 2010b). Besides this, the criterion was used that variables should have communalities of greater than 0,40.
Rather than the structure matrix, the pattern matrix was used for interpretation, since the loadings of this latter matrix represent the unique contribution of each variable to the factor. In order to ascertain the reliability of the factors, the internal consistency among the variables relating to the same factor was tested by using the Cronbach’s alpha. In general, 0.70 is determined as being the lower bound for it. However, in exploratory research 0.60 is allowed as well (Hair et al., 2010b), and thus this latter limit was used in this study. Since the model of Bouckenooghe et al. (2009) was used as a starting point, the validity of the measures was already assessed.
The final step in this process of factor extraction was the computation of factor scores. “The factor score is computed based on the factor loadings of all variables on the factor” (Hair et al., 2010b). The results of the performed EFA’s are discussed below.
3.3.3.1 Results Exploratory Factor Analysis Process of Change Factors.
Regarding the 29 process-of-change items, 21 items were retained, and thus 8 items were omitted in an one by one process. It is interesting to note that initially all items had
communalities of more than 0.40 except of only two items: I4r and I29. With respect to the significance of this study, a minimal factor loading of 0.35 was used as a perquisite. A first factor analysis showed a Measure of Sampling Adequacy of 0.892 and the Bartlett’s test was significant (0.000). Rotation converged in 30 iterations, which resulted in seven factors, based on the eigenvalue greater than 1 criterion. The cumulative % of variance resulted to be 55,848 for those 7 factors. Following the model of Bouckenooghe et al. (2009) it was hypothesized that the process factor would consist of four variables. However, it resulted from a primary analysis that the items were distributed among 7 factors. Due to the separation of the items it proved difficult to categorize clear and delineated themes. Items were deleted on an one by one basis until all requirements were met, and it resulted that four factors were approved based on their face validity.
TABLE 1: Exploratory Factor Analysis of Change-Specific Process Factors Factor Mutual Communication Support of Supervisors Attitude of Top Management Participation Item α = 0,811 α = 0,667 α = 704 α = 714
I6a: De directie houdt alle afdelingen op de hoogte wat betreft haar beslissingen.
,772 -,129 -,096 ,076 I13: Afdelingen worden voldoende geraadpleegd over de verandering. ,690 -,075 ,129 ,089 I12: De afdelingsleiding houdt rekening met de opmerking die het
personeel maakt.
,615 ,256 ,043 -,234 I3: De verstrekte informatie over veranderen is helder. ,549 -,138 ,317 ,110 I7: De communicatie over en weer tussen de directie en de afdelingen is
heel goed.
,545 ,108 ,277 -,001 I4r: Informatie over de veranderingen bereikt ons meestal via geruchten. * ,529 ,040 -,002 ,153 I19: Medewerkers worden door onze afdelingshoofden voldoende
betrokken bij de invoering van de veranderingen.
,484 ,252 -,001 ,129 I15: Het management moedigt projectleiders, operationele
leidinggevenden en montagepersoneel (ons) aan om onderwerpen voor discussie aan te dragen.
,430 ,213 -,087 ,180 I27: Als ik enige problemen ervaar, kan ik altijd bij mijn leidinggevende
aankloppen voor hulp.
,183 ,716 -,095 -,100 I26r: Mijn leidinggevende lijkt niet erg enthousiast om mij te helpen met het
vinden van een oplossing als ik een probleem heb. *
-,061 ,710 ,233 -,013 I28: Mijn leidinggevende kan zichzelf/haarzelf in mijn positie plaatsen. -,006 ,696 ,056 ,033 I29: Mijn leidinggevende moedigt me aan om dingen te doen die ik nog
nooit eerder heb gedaan.
-,062 ,541 ,113 ,050
I16: Onze afdeling biedt voldoende tijd voor overleg. ,200 ,428 -,237 ,177 I22: De directie ondersteunt het veranderingsproces onvoorwaardelijk. ,150 -,016 ,713 -,135 I21: De directie is actief betrokken bij de veranderingen. ,005 ,089 ,631 ,226 I20: De directie heeft een positieve kijk op de toekomst. -,079 ,187 ,630 ,063 I8: De directie legt de noodzaak van de verandering duidelijk uit. ,218 -,003 ,574 ,235 I1: Ik word regelmatig geïnformeerd over hoe het met de verandering gaat. ,028 -,150 ,084 ,751
I9: Veranderingen worden altijd besproken met alle mensen die er mee te maken hebben.
-,044 ,153 -,065 ,740
I14: Medewerkers zijn geraadpleegd over de redenen van de verandering. ,091 ,117 ,087 ,602
I5: We worden voldoende geïnformeerd over de voortgang van de verandering.
,253 -,072 ,113 ,541
Extraction Method: Principal Component Analysis. Rotation Method: Oblimin with Kaiser Normalization.
a. Rotation converged in 7 iterations. b. * = recoded items
α Process factors = 0,874
Although the loadings are distributed across four factors now, the themes of
Bouckenooghe et al. (2009) are not totally separated. That is, both the first and the fourth factor consist of items originally belonging to quality of communication and
originally referring to the quality of communication (I8). In order to assess the reliability of the composition of each factor, the Cronbach’s alphas were computed. It resulted that the Cronbach’s alphas of the four factors were respectively 0,811; 0,667; 0,704; 0,714, and the Cronbach’s alpha of all process factors resulted to be α = 0,874. Subsequently, it could be concluded that there is sufficient consistency among the variables relating to the process factors in general and the specific process factors.
Zooming in on the items belonging to the first and fourth factor, it resulted that both groups of items are about the communication within the organization. However, with respect to the first factor, the focus is more on mutual communication, and item I1 and I5 (now grouped in the fourth factor, but originally belonging to communication) , are about the extent to which people feel their concerns are respected. Those latter items can thus be assigned to participation, while the participation items belonging to the first factor were typified here as mutual communication. Subsequently, the following labels were assigned to the factors: mutual communication, support by supervisors, attitude of top management and participation.
3.3.3.2 Results Exploratory Factor Analysis Internal Context or Climate of Change Factors.
As mentioned before, five items relating to change-specific cynicism were added to the context factors identified by Bouckenooghe et al. (2009). Subsequently the factor analysis was started by factors relating to trust in leadership, politicking, cohesion and change-specific cynicism. Regarding the 16 internal-context items, only 6 items were retained, and thus 10 items were omitted in an one by one process. It resulted from an initial scan that four items had cross-loadings. With respect to this it is noteworthy that three of them, namely item 62, 63b and 64 were related to change-specific cynicism. The other one, item 37r, referred to cohesion. All items had communalities of more than 0.40, the
Measure of Sampling Adequacy was 0.826 and the Bartlett’s test proved to be significant (0.000). Rotation converged in 24 iterations, which resulted in five factors, based on the eigenvalue greater than 1 criterion, which had a cumulative % of variance of 57,926. Since cross loadings of more than 0.35 are not allowed in creating factor scores, the creating process was started with deleting cross-loaded items on an one by one process. It followed that the maintenance of four different factors seemed impossible due to the high presence of cross loadings. Furthermore, the use of a high number of items was
data relating to the items of ‘trust in leadership’ did not enable to create a factor. The combination that resulted consisted of the following items I64, I61 and I62 (cynicism), I39 (cohesion), and I35 and I33 (politicking). Based on a fixed number of 3, the items communalities were relatively high, and the Measure of Sampling Adequacy resulted to be 0.803. Besides this, the Bartlett’s test of sphericity proved significant, and the
cumulative % of the factors was 77,042. Rotation converged in 5 iterations. The pattern matrix which is related to this diversion is shown below.
TABLE 2 Exploratory Factor Analysis of Context or Internal-Climate Factors
Item
Factor
Cynicism Cohesion Politicking α = 0,794 α = 1 α = 0,626 I64: Ik geloof dat de directie haar bedoelingen met de introductie van
de verandering heel anders zijn dan dat zij het personeel vertelt.
,909 -,067 ,137 I61: 61Ik denk dat de directie in werkelijkheid andere redenen heeft om
de verandering in te voeren dan zij zegt.
,805 -,043 -,090 I62: De directie probeert de reden voor deze verandering achter te
houden.
,773 ,100 -,128 I39: Ik heb vertrouwen in mijn collega’s. -,017 ,984 ,032 I35: In onze organisatie is vriendjespolitiek een belangrijke manier om
iets te bereiken.
-,030 -,081 -,858
I33: Binnen onze organisatie spelen machtsspellen tussen de verschillende afdelingen een belangrijke rol.
,062 ,050 -,824
Extraction Method: Principal Component Analysis. Rotation Method: Oblimin with Kaiser Normalization. a. Rotation converged in 5 iterations.
α Context Factors = 0,648
3.3.3.3 Results Exploratory Factor Analysis Self-Efficacy
Regarding the seven self-efficacy items, five items were retained. A first factor analysis showed a Measure of Sampling Adequacy of 0,726 and the Bartlett’s test was significant. Rotation converged in six iterations, which resulted in two factors, based on the
eigenvalue greater than 1 criterion. The cumulative % of variance resulted to be 50,038 for those two factors. However, the communalities of I54 and I60 were respectively 0.316 and 0.338. Due to this, I54 was deleted from further analyses. However,
subsequently I60 still had a communality of only 0.377. The deletion of this item would create a diversion that would meet the definition of Metselaar (1997), namely a diversion of self-control between knowledge and experience. Because of the usability of a clear distinction between those concepts I60 was deleted, and I55, I56, I57r, I58 and I59 were retained. A Measure of Sampling Adequacy of 0,673 resulted. Bartlett’s test was
significant, and all items had communalities of 0.491(I58) and higher. Rotation converged in five iterations. The cumulative % of variance resulted to be 61,934. The pattern matrix which is related to this diversion is shown below:
TABLE 3 Exploratory Factor Analysis of Self-Control Factors
Item Factor Self-Control (source: knowledge) Self-Control (source: experience) α = 0,605 α = 0,571 I55: Ervaringen uit het verleden met
organisatieveranderingsprocessen geven mij vertrouwen.
-,115 ,895
I56: Door mijn eerdere ervaringen met veranderingen, verwacht ik succesvol met de verandering te kunnen gaan werken.
,175 ,767
I57r: Ik denk dat ik bepaalde taken, die vereist zijn voor deze verandering niet goed kan uitvoeren.
,752 -,017 I58: 58Ik heb de vaardigheden die nodig zijn om
deze verandering te laten slagen.
,627 ,185 I59: Ik heb de benodigde kennis om deze
verandering in mijn werk door te voeren.
,793 -,098 Extraction Method: Principal Component Analysis.
Rotation Method: Oblimin with Kaiser Normalization. a. Rotation converged in 5 iterations.
α Context Factors = 0,625
self-control factors resulted to be α = 0,625. Although the Cronbach’s alpha’s are relatively low, the combination is comparable with the solution that resulted in the study of Metselaar (1997), and thus it was decided to accept it.
3.3.3.4 Results Exploratory Factor Analysis Readiness for Change Factors.
It resulted from an initial scan that the Measure of Sampling Adequacy was high, namely 0.882, Bartlett’s test of sphericity was significant, and all communalities were higher than 0.40. Besides this, based on the eigenvalue greater than 1 criterion, three factors were created which had a cumulative percentage of variance of 59,526. It resulted from the pattern matrix that there were no cross-loaded items. Further consideration showed that both factor one and three were a combination of items relating to the cognitive and emotional dimension. Besides this, the number of cognitive and emotional items were evenly distributed across those factors. In order to study both the moderating impact of cynicism and the correlation between stage of implementation and readiness in a careful way, a clear distinction between the three dimensions of readiness would be valuable. However, it seemed impossible to respect the combination of items created by Bouckenooghe et al. (2009). Nevertheless a combination of factors resulted with the same labels as Bouckenooghe et al. (2009) used, namely cognitive readiness for change (I48r, 47r and 46r: α = 0,738), intentional readiness for change (I53, I52, I51: α = 0,640) and emotional readiness for change (I50, I49, I43: α = 0,640).
With respect to the factor intentional readiness it resulted that this factor only consisted of items originally relating to intentional readiness. This phenomenon is in line with what
happened in the study of Bouckenooghe et al. (2009). They pointed that “intentional readiness for change was a perfect reflection of the content adequacy test”. However, it resulted to be impossible to distinguish between cognitive and emotional readiness for this database. Strikingly, the items relating to cognitive readiness that remained in the study of
readiness” would lower the Cronbach’s alpha to α = 0.551. The highest possible Cronbach’s alpha (α = 0.681) would result if I50 was deleted. However, especially this item was very high loaded on the respective factor. Moreover, deletion would only evoke a small increase in the Cronbach’s alpha. Subsequently it was decided to create an emotional readiness for change factor which consisted of I50, I49 and I43, since each item was loaded sufficiently high on the factor and the combination met the perquisites with respect to the Cronbach’s alpha.
It resulted, based on the eigenvalue greater than one criterion, that all item communalities were larger than 0.40, the Measure of Sampling Adequacy resulted to be 0.882; the Bartlett’s test of sphericity proved significant, and the cumulative % of the factors was 59,526. Rotation converged in seven iterations. The Cronbach’s alpha of all readiness for change factors resulted to be 0.804. The pattern matrix which is related to this diversion is shown below.
TABLE 4 Exploratory Factor Analysis of Readiness for Change Factors
Item
Factor
Cognitive dimension Intentional Dimension Emotional Dimension α = 0,738 α = 0,789 α = 0,644 I48r: De meeste veranderingsprojecten die er toe
dienen om problemen op te lossen, zullen niet veel goeds doen. *
,869 ,075 ,013 I47r: Er zal niet veel terecht komen van
toekomstplannen om het beter te doen. *
,825 -,063 -,060 I46r: Ik denk dat de meeste veranderingen een negatief
effect zullen hebben op onze klanten. *
,743 ,030 ,025 I53: Ik ben bereid om energie te stoppen in het
veranderingsproces.
-,098 -,904 -,002
I52: Ik ben bereid om een betekenisvolle bijdrage aan de verandering te doen.
,057 -,840 -,112
I51: Ik wil mij inzetten voor het veranderingsproces. ,011 -,781 ,106 I50: De verandering zal het werk vereenvoudigen. -,099 ,127 ,927
I49: De verandering zal het werk verbeteren. ,255 -,223 ,589
I43: Ik vind de verandering verfrissend. ,171 -,316 ,471
Extraction Method: Principal Component Analysis. Rotation Method: Oblimin with Kaiser Normalization.
a. Rotation converged in 8 iterations. b. * = recoded items
The table below shows the labeled factors with their related items that resulted from this study’s EFA:
TABLE 5: Summary of created factors
Factor label Items
Change-Process Factors
Mutual communication α = 0,811 I6a; I13; I12; I3; I7; I4r; I19; I15 Support by supervisors α = 0,667 I27; I26r; I28; I29; I16
Attitude of top management α = 0,704 I22; I21; I20; I8 Participation α = 0,714 I1; I9; I14; I5
Change-Context Factors
Cynicism
α = 0,794 I64; I61; I62 Cohesion
α = 1 I39
Politicking α = 0,626 I35; I33
Self-Control Factors
Self-Control (knowledge source) α = 0,605 I57r; I58; I59
Self-Control (experience source) α = 0,571 I55; I56
Readiness for Change Factors
4. RESULTS
In order to test the above formulated hypotheses, several analyses had to be performed. Since Bouckenooghe et al. (2009) addressed the need for the cross-validation of their instrument, first a multiple regression analysis was performed. Although it was consciously decided to involve all employees from the overarching company XYZ, a separate regression analysis was also performed for company A and B separately, since Stensaker & Meyer (2012) pointed that limited research exists regarding the role that change experience plays in organizational change. As mentioned before, employees originating from company A were characterized by a history of smooth operations without impressive changes, while employees from company B have experienced poor business results for many years, many organizational changes, and are moreover less experienced with using Six Sigma. This study attempted to find out if there were differences between the companies with respect to the degree to which the divergent dimensions of readiness were explained by the identified antecedents, since they are characterized by different histories.
4.1 Descriptive statistics
Table 6 shows a summary of the distribution of respondents’ company background, age, working experience, function, and the year when they first got acquainted with Six Sigma.
TABLE 6 Descriptive statistics
Company Age
Company A 191 20 years and younger 3
Company B 60 21-35 years old 90
36-45 years old 69
46-55 years old 68
56-65 years old 36
66 years and older 0
Table 7 shows a summary of the means and standard deviations of the process, internal-context and readiness for change variables for company XYZ:
TABLE 7: Means and standard deviations process, context and readiness for change variables company XYZ
Variable M SD
1. Quality of communication 3,102 0,118
2. Support by supervisors 3,597 0,224
3. Attitude of top management 3,481 0,219
4. Participation 3,006 0,114
5. Cynicism 2,528 0,089
6. Cohesion 3,85 0,575
7. Politicking 2,874 0,077
8. Self-control from knowledge 2,55 0,36 9. Self-control from experience 2,749 0,467 10. Cognitive readiness for change 3,635 0,071 11. Intentional readiness for change 3,892 0,045 12. Emotional readiness for change 3,244 0,285
Employed years Functional level First Lean6Sigma moment 0-5 years 60 Staff-function 51 Before the merger 96
5-10 years 57 Mechanic 58 2008 34
10-15 years 48 Operational leader 96 2009 25
15-20 years 26 Projectleader 45 2010 29
20-25 years 25 Manager 14 2011 31
25-30 years 24 Board-member 3 2012 6
30-35 years 15 Not yet 43
4.2 Retesting Bouckenooghe et al.’s (2009) instrument: A multiple regression analysis To retest the model of Bouckenooghe et al. (2009), a multivariate linear regression needed to be performed, which concerns hypotheses 1a; 1b; 1c; 1d and 1e. Those hypotheses were about the direction and extent to which the different dimensions of readiness for change are
influenced by process-factors of change; context-factors of change and/or self-control factors. The regression results of company XYZ are presented first, followed by the separate results of company A and B. A comparison of the separated regression analyses would enable to find out if there were differences between company A and B with respect to the degree to which the divergent dimensions of readiness were explained by the identified antecedents. Possible differences in regression results are suggested to be a consequence of the difference in history between the companies.
As a reminder: company XYZ is the umbrella organization, employees originating from company A are characterized by a history of smooth operations without impressive changes, while employees from company B have experienced poor business results for many years, many organizational changes, and are much less experienced with using Six Sigma.
There are four main conditions that justify the use of linear regression models (Hair, Black, Babin and Anderson, 2010c). In appendix D “Assumptions check of multivariate analysis” it is clarified that the major part of this study’s data set is appropriate for doing the respective multiple regression analyses. With respect to company XYZ the relationship between self-control by knowledge and emotional readiness for change deviates from this standard. It resulted that this relationship is characterized by non-linearity. Corrective action was taken by transforming the data values of the independent variable “self-control by knowledge” using the logarithm. However, some more corrective action was needed for the separated regression analyses. With respect to company A, neither the correlation between cohesion and
intentional readiness did not prove to be significant, nor the correlative relationship between 1.support by supervisors, 2.participation, 3.cynicism, 4.cohesion, 5.self-control by knowledge and emotional readiness for change was significant.
taken by using the logarithm of those factors for company A and B in the subsequent multiple regression analyses.
Within each regression model a distinction was made between process, internal context and self-control factors of change. Attention was paid to both the variance explained by the variate and the standardized beta coefficients of the single variables. The testing of regression
between both 1. process factors and 2. self-control factors and the dependent variable “readiness for change” was corrected for the context of change, since Bouckenooghe et al. (2009) pointed that the “internal-context factors” are general, and process factors change-specific. Moreover, Metselaar (1997) stated that self-control factors concern the knowledge, experience and skills an individual has to successfully implement the change (Rotter, 1966; Kren, 1992; Bandura, 1986 in Metselaar, 1997). The extent to which his/her experience and knowledge are useful at the moment is dependent on the characteristics of the prospective change. An individual’s readiness for change is thus influenced by context factors irrespective of the level and/or existence of process factors and/or self-control.
Due to the fact that conclusions were based on only one sample, the F-ratio was calculated to determine if the regression model would be able to represent the whole organization rather than just this one sample. As can be concluded from the tables below, the F-ratio was significant for each dimension of readiness for change by using α < 0,001. Subsequently H0 (all regression coefficients are equal to zero) could be rejected, which means that there is heterogeneity between the variances, and that the models can represent the whole population.
TABLE 8: F-ratio cognitive readiness for change odel summar
Model R Square Adjusted R Square Durbin-Watson F Sig.
1 0,367 0,360 49,563 0,000
2 0,475 0,456 1,906 25,088 0,000
a. Predictors: (Constant), Politicking, Cohesion, Cynicism
b. Predictors: (Constant), Politicking, Cohesion, Cynicism, Mutual Communication, Support by supervisors, Attitude of top management, Participation, Self-Control by Knowledge, Self-Control by Experience
Context, process and self-control factors are all argued to be antecedents of readiness for change (Bouckenooghe et al., 2009, Holt et al., 2007). As can be concluded from the tables above, those predictors are responsible for the explanation of 45,6% of the variance in cognitive readiness for change (adjusted R² = 0,456; p < 0.001); 22,5% of the variance in intentional readiness for change (adjusted R² = 0,225; p < 0.001) and 21,9% of the variance in emotional readiness for change (adjusted R² = 0,219; p < 0.001).
With respect to the multiple regression analyses, the Enter method was used. By using this method, the regression of both the common and each single process, context and self-control factor could be calculated. Since the model of Bouckenooghe et al. (2009) was used here as a starting point, the regression results of the identified factors will be discussed in the following order: change process factors, internal-context factors and self-control factors. A significance level of α < 0,1 was used.
4.2.1 Change process factors
It resulted that not all four process factors were related to the three readiness for change dimensions. With respect to company XYZ, both “mutual communication” and
“participation” were not significant related to one of the dimensions of individual’s readiness
TABLE 9: F-ratio intentional readiness for change Model summary
Model R Square Adjusted R Square Durbin-Watson F Sig.
1 0,149 0,139 14,913 0,000
2 0,252 0,225 1,794 9,365 0,000
a. Predictors: (Constant), Politicking, Cohesion, Cynicism
b. Predictors: (Constant), Politicking, Cohesion, Cynicism, Mutual Communication, Support by supervisors, Attitude of top management, Participation, Self-Control by Knowledge, Self-Control by Experience
c. Dependent Variable: Intentional readiness
TABLE 10: F-ratio emotional readiness for change odel summar
Model R Square Adjusted R Square Durbin-Watson F Sig.
1 0,075 0,064 6,952 0,000
2 0,246 0,219 2,017 9,058 0,000
a. Predictors: (Constant), Politicking, Cohesion, Cynicism
b. Predictors: (Constant), Politicking, Cohesion, Cynicism, Mutual Communication, Support by supervisors, Attitude of top management, Participation, Log Factor score Self-Control Knowledge, Factor score Self-Control Experience
