1
EXPLAINING READINESS FOR CHANGE
BY LOOKING AT THE PAST:
“The influence of previous change experience sentiments and
change frequency on self-efficacy and subsequently on readiness for change”
Master thesis, MSc Business Administration, specialization Change Management University of Groningen, Faculty of Economics and Business
24th June, 2013 Oeke Hettinga Student number: s2231700 Turfstraat 9 A 9712 JK Groningen Email: o.g.hettinga@student.rug.nl Words: 19,741 Supervisor at university: 1st supervisor drs. H.P. van Peet 2nd supervisor
2
EXPLAINING READINESS FOR CHANGE
BY LOOKING AT THE PAST:
“The influence of previous change experience sentiments and
change frequency on self-efficacy and subsequently on readiness for change”
ABSTRACT
When change does not lead to results the organization aimed for, failure can have disastrous consequences for the company and its employees. While failure highly depends on readiness for change, this study attempted to explain readiness for change by looking at the past. The study tested the mediating effect of self-efficacy on the influence of previous change experience sentiment (individual history of change), frequency of change, and change fatigue on affective, behavioural and cognitive readiness for change. Questionnaires were gathered among employees of a large health care organization in Northern Netherlands. The 916 valid responses were analysed with regression analysis to test the hypotheses. Self-efficacy does play a large explaining role in the assumed relations of this study, particularly between the negative relation of change fatigue and behavioural readiness for change. This relation is completely explained by self-efficacy. This is also the case with the positive relation between frequency of change and behavioural readiness for change. A surprising finding was that older employees showed lower self-efficacy, than their younger colleagues did. Employees’ who were frequently involved in change situation showed higher levels of self-efficacy and were more ready to change (behavioural and cognitive). The study emphasized the importance of higher levels of self-efficacy, because of the positive influence on readiness for change, which can be achieved by frequent changes and avoiding negative change experience sentiments and feelings of change fatigue.
Acknowledgements:
I would like to thank Drs. van Peet for her feedback, guidance and support during the writing process of this thesis and I would also like to thank Dr. van Offenbeek for her feedback. Furthermore, I would like to thank Mr. Berger for giving me the opportunity to do my research at Lentis and thank Mrs. Van Beek for her helpful thoughts.
Keywords: Readiness for change, Self-efficacy, Frequency of change, Change Fatigue, Previous change experience sentiment, History of change
3
Table of Content
1. Introduction ... 4
2. Theory ... 9
2.1 Readiness for change ... 9
2.2 Frequency of change ... 13
2.3 Previous change experience sentiment ... 14
2.4 Self-Efficacy ... 16 2.5 Conceptual model ... 19 3. Methods ... 20 3.1 Research Design ... 20 3.2 Data Collection ... 20 3.3 Measurements ... 25 3.4 Data Analysis ... 27 4. Results ... 33
4.1 Factor and reliability analyses ... 33
4.2 Descriptive analysis ... 37
4.3 Hypotheses testing ... 41
4.4 Qualitative results ... 45
4.5 Summary of quantitative results ... 46
5. Discussion ... 50
5.1 Discussion ... 50
5.2 Implications ... 56
5.3 Limitations and future research ... 59
5.4 Conclusion ... 61
References ... 62
Appendix A: Cover Letter and Questionnaire... 70
Appendix B: Original constructs ... 75
Appendix C: Factor Analysis ... 78
Appendix D: Histograms and K-S test ... 80
Appendix E: Assumption checks of regression models ... 82
Appendix F: Sobel test ... 83
Appendix G: Cross tabulations with control variables ... 84
4
1. Introduction
Change management consists of continually renewing the direction, structure and capabilities of an organization to serve the ever-changing needs of customers (Todnem, 2005). During time, the conception emerged that organizations continuously need to change to survive (Burnes, 2011b), which makes change an ever-present feature in organizations (Todnem, 2005). The speed, magnitude, unpredictability and the importance of change have increased considerably (Burnes, 2011b). Not all change initiatives produce the results the company aimed for, causing staggeringly high failure rates of change implementation between 60% and 90% (Burnes, 2011a; Burnes, 2011b; Quinn, 2004; McKinsey & Company, 2008). Considering most organizations make moderate to major changes at least every four to five years, failures might have disastrous consequences for the company and its employees (Lewis, 2000). The failure of change initiatives highly depends on the readiness of the organization and its employees. Therefore, being ready is essential for an organization and its employees before attempting to implement and manage change initiatives (Todnem, 2005).
5 Multiple reasons underline the importance of this study. The relation between frequency of change and readiness for change remains ambiguous. Also, only little research has been conducted about previous change experience on the individual level, as well as the potential mediating role of self-efficacy. Therefore, this research focuses on the influence of frequency of change and previous change experience sentiments on employees’ readiness for change mediated by self-efficacy. Therefore the following research question was postulated:
To what extent does self-efficacy mediate the influence between frequency of change and previous change experience sentiment on employees’ readiness for change?
1.1 Research Context
In order to study the relations, postulated in the research question, data of an organization dealing with organizational change was needed. Lentis is the organization in which this research took place.
The organization
6 by strengthening their autonomy, enhancing their personal functioning and social participation, as well as by optimizing their quality of life by relieving their suffering1.
In the Netherlands the accessibility of care is threatened because health care becomes priceless and the future availability of care takers becomes also problematic (Lentis, 2013). The company, and other health care providers, perceive innovation as a necessity. Innovation should enable more responsibility with citizens themselves and care provision should focus more on the support of those citizens. However, innovation within healthcare can only be successful if individuals treat their own health with more responsibility. Hence, health care providers should focus more on behaviour and health, instead of the current focus on care for illnesses (RVZ, 2010).
The company wants to move along with this shift and aims to support individuals in doing so. The shift to more focus on behaviour and health is also called Health 2.0 which is characterized by participation between patients, between professionals and between professionals and patients (Lentis, 2012B). Health 2.0 and eHealth go together in the way that eHealth enables goals of Health 2.0 like two-way communication and collaboration (Timmer, 2011). EHealth is referred to as the use of information and communication technologies, particularly internet technology, designed to support or improve health and health care (RVZ, 2010). The company considers the use of smart technology as opportunity to provide durable, affordable and available care (Lentis, 2012B). The program developed to work on this shift and the implementation of eHealth is called iLentis. This program provides the opportunity to support the move towards more two-way communication and collaboration with digital application. Within the innovation program iLentis, the organization aims to improve the quality, availability and effectiveness of healthcare, and providing the possibility for self-management and empowerment of patients (Lentis, 2012B).
The planned transition with iLentis and the deployment of smart technology is expected to affect all aspects of the organization. The introduction of smart technology is not the most difficult part of the transition. Changing the way of working will be the main challenge for Lentis. Employees have to start interacting more; to be active instead of passive (Lentis, 2012B). The changes the
7 company aims to achieve will have significant impact on inter alia people, processes, systems, finance, relationships, marketing and communication (Lentis, 2012B). One of the already initiated projects is mobile working, which offers employees a tablet or laptop in order to access necessary information at any time and at any place, like at home or when visiting patients. In the past, the company made a structural change when the organization divided itself in the previously mentioned care groups, representing a specific type of health care. Also, the introduction of the electronic patient file altered the working procedures of many professional groups involved.
Change Process
As part of iLentis, the company aims to replace its current intranet. The current intranet has been running for eight years and its functionalities no longer support the organisation; for example the intranet is not available at home and on mobile devices. Also, the system is very slow and software updates are no longer available (Lentis, 2012A). In early 2012, the company decided it can benefit from a web based and useful intranet. A useful intranet means that it is attractive to use and that employees’ work becomes easier, clearer and more efficient. The results of a survey of 2012 about employees satisfaction with the current intranet was used to determine important features of the intranet. In May 2013, a vendor for the new Social Intranet was chosen and the company aims to have the Social Intranet running at the end of this year. With the Social Intranet, Len the company tis aims to enhance social collaboration instead of the current intranet which is more a publishing platform with especially static content. The Social Intranet will be equipped with features that enable content creation, knowledge sharing, discussion possibilities, real time communication by means of video conferencing, chat, working together in the same documentation and developing collective intelligence (knowledge) such as wiki’s. Applied research showed that the last mentioned feature is in practice the least applied one, which can be explained by the large impact of this feature. Collective intelligence features influences the organization and its culture more than other tools (Jansen, 2013).
8 the diagnosing and treatments processes (Lentis, 2012A). Patients can benefit from professionals sharing their knowledge (Lentis, 2012B). Secondly, the Social Intranet aims to improve the work and processes due to being easily accessible anywhere and offering work related information. As third, the layout of the Social Intranet aims to be structured more user friendly due to a more logical layout based on how often something is consulted instead of the current layout based on the organisation’s hierarchy and will make for example an often used travel expenses claim form easy findable (Lentis, 2012A).
Eventually, this Social Intranet will change the way the organisation is structured and managed. The semi-autonomous working traditions of professionals will change and will be replaced with more collaboration, professional profiling on intranet with fields of expertise and sharing their knowledge with colleagues (Lentis, 2012B). From the autumn of 2012, employees are notified about the Social Intranet in many ways, such as messages on Yammer, intranet and in the weekly innovation newsletter. From the summer of 2013, a targeted campaign for the Social Intranet will start. The full potential of the intranet will be implemented in stages. In the first stage of the implementation the features of the intranet are equal to all employees. In a later stage, the type of features, use and the layout of the intranet will differ between employees because for example psychologists will create content based on their specific knowledge while secretaries will not make use of those features. The implementation of the first stage is planned for the end of this year, and the implementation of the following stages is adjusted to the circumstances.
9
2. Theory
In this study, the influence of previous change experiences on employees’ readiness to participate in future changes will be addressed. In particular, whether there is a link between exposure to frequent changes, previous negative or positive experiences and their readiness for change, and a possible mediating effect of self-efficacy. This chapter provides a review of the existing literature and concludes with the conceptual model and the associated hypotheses. First of all the dependent variable readiness for change will be elaborated, followed by the independent variables, frequency of change and previous change experience sentiment and the mediator self-efficacy.
2.1 Readiness for change
10 organizational change. Scholars who believed organizational change is a continuous open-ended process criticised Lewin’s model for being too simplistic and mechanistic, for ignoring the power and politics within organizations and as only being applicable for isolated change (Burnes, 2009).
The most used definition of readiness for change is developed by Armenakis, Harris & Mossholder (1993) who stated that readiness for change consist of the individual’s beliefs, attitudes, and intentions regarding the extent to which changes are needed and the organization’s capacity to successfully undertake those changes. As well, the extent to which employees believe the change is likely to have positive implications for themselves and the wider organization (Jones, Jimmieson & Griffiths, 2005). Other definitions of readiness for change are almost all derived from the original work of Armenakis et al. (1993). Researchers have defined individual readiness for change in somewhat different ways. Although they all agree that individual readiness for change involves an individual’s assessment about the individual and organizational capacity for making a successful change, the need for a change, and the benefits the organization and its members may gain from a change (Choi et al., 2010).
Holt et al. (2007) present an instrument to measure individual readiness for change, which consist of four beliefs. This instrument measures if the change is regarded to be appropriate for the situation, if management supports the change, whether someone is capable to implement the change and whether the change is believed to be beneficial or not. When someone scores high on those four beliefs, their readiness towards change is also high (Holt et al., 2007). However, this instrument measures the extent to which the beliefs appropriateness, management support, self-efficacy and personal valence, are present with individuals and not explicitly employees’ readiness for change. Therefore, it is not an accurate instrument to accomplish what this research aims for: determining the actual individual readiness for change.
11 and cognitive components was initially postulated by Elizur and Guttman (1976). Piderit (2000) attempted to integrate the three components of attitudes based on the work of Ajzen (1984) who called it the tripartite view of attitudes. Any definition of attitudes with more focus on one component, at the expense of the other two, is incomplete. “Conceptualizing employees' responses to proposed organizational changes as multidimensional attitudes permits a richer view of the ways in which employees may respond to change” (Piderit, 2000, p. 789). The multifaceted view of responses to
change can enhance the accuracy in predicting employees’ behaviours (Piderit, 2000) because employees can have different reactions along the three components of readiness for change.
Affective readiness for change
The affective component of readiness for change refers to a set of feelings about the change (Bouckenooghe, 2010; Piderit, 2000) and consists of emotions such as hate, sadness, happiness and excitement (Rafferty, Jimmieson & Armenakis, 2013). This component of readiness represents feelings, emotions and moods that people experienced in relation to change and subsequently associate with it (Eagly & Chaiken, 1993). An employee’s affective response to organizational change can range from strong positive emotions to strong negative emotions (Piderit, 2000).
Behavioural readiness for change
The behavioural component, often called the intentional, is about the effort and energy someone is willing to invest (Bouckenooghe, 2010). The motivational factors that influence behaviour are the indicators of how hard a person is willing to try and how much effort someone is willing to exert in order to perform certain behaviour (Rafferty et al., 2013). An employee's behavioural response can range from positive intentions to support the change to negative intentions opposing it (Piderit, 2000).
12 consider the behavioural component of readiness for change to be equally important as the affective and cognitive component of readiness for change.
Cognitive readiness for change
Cognitive readiness refers to the beliefs and thoughts someone has about the outcomes of change (Bouckenooghe, 2010; Piderit, 2000). The beliefs express positive or negative evaluation in greater or lesser extremity, and sometimes beliefs are exactly neutral (Eagly et al., 1993). An employee’s response to change within the cognitive component might range from strong positive beliefs to strong negative beliefs and can be formulated as ‘this change is essential for the organization’ or ‘this change could ruin the company’ (Piderit, 2000).
Employees’ change readiness is the most prevalent positive attitude towards change. Ninety percent of all conceptual work on change attitudes is about either readiness or resistance to change (Rafferty et al., 2013). This well investigated concept is very useful in change processes because knowing the amount of readiness, and identifying relations with antecedents (e.g. change process, content and internal context) can help design the strategies to increase readiness for change. Also, the concept adds value by revealing the perception of the involved ones and can expose the possible gap between change agents expectations about change efforts and those of others in the organization (Bouckenooghe et al., 2009). In this study the multifaceted concept of readiness for change to determine employees’ affective, cognitive and behavioural readiness for change will be used because it actually measures individual readiness for change and gives insights in possible different reactions along the three components of an attitude.
13 research showed that the level of self-efficacy determined the effort someone’s is willing to put in a certain event (Bandura, 1982). Therefore, this study is interested in the question whether a higher level of self-efficacy also influences someone’s readiness for change in a positive way.
2.2 Frequency of change
Frequency of change is defined as the individuals’ perception regarding how often change has occurred in their work environment and is an important characteristic of change that is easy noted by individuals (Rafferty et al., 2006A). The literature provides mixed results regarding the relation between frequency of change and employees reactions. On one hand, results show that frequent change can have a negative influence on employees and organizations; on the other hand research shows it can have a positive influence on both. “While exposure to an increasingly frequent organizational change can lead to change fatigue and cynicism, it can also generate more positive reactions to change” (Stensaker et al., 2012, p. 106). Negative influences of frequent changes are
feeling of mentally exhaustion and high unpredictability. Also, frequently encountered changes can lead to higher turnover intentions and indirectly to lower job satisfaction (Rafferty et al., 2006A), which indicates that frequency of change has a negative influence on employee’s work happiness. However, as mentioned before, frequently encountered change can also have a positive influence on employees because it offers an arena for learning and change is perceived as less threatening. Employees who limitedly encounter change show strong emotional reactions, whereas experienced employees were not frustrated by the insecurity and uncertainty of change. Experienced employees became prepared and receptive to change. The sense of familiarity and becoming accustomed with change affected their reaction to change (Stensaker et al., 2012). Important to note is that the evidence that frequent change can lead to more positive reactions to change is limited (Thornhill & Saunders, 2003).
14 explains that employees in a relatively stable organization have more difficulties with adapting to change and perceive it as disturbing and risky. Employees who frequently encounter changes tend to find change less disturbing and hence less risky, and cope with changes more easily (Park & Krishnan, 2003).
Hence, employees’ perception about the frequency of change influences their reaction towards future changes. Employees who perceive high frequency of change get used to it and are accustomed to deal with it. Their experience with frequent changes gives them confidence about their ability to deal with changing situations. After all, self-efficacy is based on and determined by experiences and performances of the past in combination with expectations of the future (Rafferty et al., 2006A). Therefore, it is hypothesized:
H1: Higher levels of frequency of change have a positive influence on recipients’ self-efficacy
2.3 Previous change experience sentiment
15 change (Bordia, Restubog, Jimmieson & Irmer, 2007). The social learning theory suggests employees have little reason to be fearful of events in which one has been successful in the past (Bernerth, 2004), which suggest that an employee’s previous positive change experience can lead to a positive reaction to future changes. Employees learn from outcomes of previous change experiences due to the feedback loop which revises their beliefs and expectations of the future (Bernerth, 2004).
Bouckenooghe and Devos (2007) try to explain the effect of organizational change history on readiness for change by the expectancy theory of Vroom (1964) and the schema theory by Lau & Woodman (1995). When an individual is involved in change, a schema of previous change is triggered and their positive or negative sentiments about change determine the effort they will put into the change. If the experience is positive, employees will increase their effort, when the experience is negative employees will limit their effort (Bouckenooghe et al., 2007). Within this explanation, the individual and organizational levels are mixed up. Bouckenooghe et al. (2007) describe how someone’s previous change experience influences their sentiment for subsequent change and how it influences their effort. The explanation focuses on the individual level, although the label ‘organizational history of change’ tells it focuses on organizational level. The mix up of individual and organizational level becomes quite clear when looking at the scale used to measure history of change. Bouckenooghe et al. (2007) adopted the scale from Metselaar (1997) which consists of four items like ‘Our organization has always been able to cope with new situations’, ‘Announced changes usually came to nothing in the past’, ‘Past changes were generally successful’ and ‘Our company has
proven to be capable of major changes’. When reading those items carefully, it is obvious that the
scale is intended to measure history of change at the organizational level, and it is even more about the organization’s historical ability to change. When looking more closely to the research of Bouckenooghe and Devos (2007) it was remarkable to detect they measured the history of change on both organizational and individual level with the scale of Metselaar (1997).
16 that the concept of individual history of change is not measureable yet. Further research would be necessary to fill this gap. However, that is beyond the scope of this research. For this research, previous change experience will be tried to capture with an overall sentiment about previous changes. Did one perceive overall successful changes or more negative experiences with changes? This concept will be referred to as previous change experience sentiment.
When the change experiences are perceived relatively positive, employees tend to find the change less disturbing. Because of the positive experiences employees belief change is something that is necessary and are accustomed to change. Readiness for change is inter alia determined by the previous change experiences of employees (Cawsey, Deszca & Ingols, 2012). Positive change experiences can contribute to the development of individual change capabilities due to generating the ability to cope with uncertainty during change, maintaining control and increasing an employee’s market value (Stensaker et al., 2012). An employee’s self-efficacy will also increase when someone perceives to have the required skills and abilities to show the new behaviour or gained the relevant experience to execute the new tasks. Therefore, it is hypothesized:
H2: Higher levels of positive previous change experience sentiments have a positive influence on
recipients’ self-efficacy.
2.4 Self-Efficacy
Self-efficacy is the belief to possess the skills and abilities to successfully execute the behaviour required to fulfil the assigned tasks (Bandura, 1982; Caldwell, 2011). It is someone’s general sense of confidence in successfully navigating and mastering change situations (Caldwell, Roby-Williams, Rush & Ricke-Kiely, 2009). Bandura (1982) defines it as someone’s belief to be able to successfully execute the behaviour required to produce the future outcomes. “When beset with difficulties people who entertain serious doubts about their capabilities slacken their efforts or give up altogether, whereas those who have a strong sense of efficacy exert greater effort to master the challenges”
17 will convert this feelings into increased effort and motivation (Robbins et al., 2012). Self-efficacy, someone’s perception of capabilities, functions as adeterminant of how people behave, their thought patterns and emotional reaction preceding and during the change (Bandura, 1982). “Those who judge themselves inefficacious in coping with environmental demands dwell on their personal deficiencies
and imagine potential difficulties as more formidable that they really are” (Bandura, 1982, p. 123). In
this research the definition of self-efficacy by Bandura (1982) will be used.
Self-efficacy can be enlarged in order to change someone’s perception of their ability to handle a certain situation. Reinforcing self-efficacy reduces discrepancy between current and future tasks demands (Armenakis et al., 1993). Bandura (1982) proposed four ways to increase self-efficacy. The most important source is enactive mastery which comprises of gaining relevant experience with the task because when someone performed the task successfully in the past, someone is more confident he will be able to it in the future. Vicarious modeling is a way to increase confidence through seeing someone else doing the tasks. Even, the effect will be enlarged when comparing yourself similar to the person performing the task. The third way is verbal persuasion that allows to become more confident because someone convinced you of having the necessary skills to be successful. Verbal persuasion is closely related to self-fulfilling prophecy by which believing something will happen actually come true. The fourth way to increase self-efficacy is arousal by which someone gets in an energized state which drives someone to complete the task (Robbins et al., 2012).
18 not believe to possess the ability to show the required behaviour are likely to resist the change (Cummingham, Woodward, Shannon, MacIntosh, Lendrum, Rosenbloom & Brown, 2002). In a change situation, employees with lower self-efficacy might intent or act with lessen their effort or even give up (Robbins et al., 2012) and those employees belief changes are more problematic than they really are (Bandura, 1982). Due to research that self-efficacy is positively related to individual reactions to change (Herold et al., 2007), it is hypothesized:
H3A: Higher levels of self-efficacy have a positive influence on affective readiness for change
H3B: Higher levels of self-efficacy have a positive influence on behavioural readiness for change
H3C: Higher levels of self-efficacy have a positive influence on cognitive readiness for change
H4A: Self-efficacy mediates the positive relationship between frequency of change and recipients’
affective readiness for change, such that higher levels of frequency of change lead to higher levels of self-efficacy; and consequently to more affective readiness for change.
H4B: Self-efficacy mediates the positive relationship between frequency of change and recipients’
behavioural readiness for change, such that higher levels of frequency of change lead to higher levels of self-efficacy; and consequently to more behavioural readiness for change.
H4C: Self-efficacy mediates the positive relationship between frequency of change and recipients’
cognitive readiness for change, such that higher levels of frequency of change lead to higher levels of self-efficacy; and consequently to more cognitive readiness for change.
H5A: Self-efficacy mediates the positive relationship between previous change experience sentiment
and recipients’ affective readiness for change, such that more positive previous change experience sentiments lead to higher levels of self-efficacy; and consequently to more affective readiness for change.
H5B: Self-efficacy mediates the positive relationship between previous change experience sentiment
19 sentiments lead to higher levels of self-efficacy; and consequently to more behavioural readiness for change.
H5C: Self-efficacy mediates the positive relationship between previous change experience sentiment
and recipients’ cognitive readiness for change, such that more positive previous change experience sentiments lead to higher levels of self-efficacy; and consequently to more cognitive readiness for change.
2.5 Conceptual model
This research is based on a mediation model that aims to identify and explicate the mechanisms underlying a possible relation between frequency of change, previous change experience sentiment and the dependent variable readiness for change by including the explanatory variable self-efficacy. This research hypothesizes that frequency and previous experience sentiment with change influence self-efficacy, which in turn influences the three components of readiness for change. The conceptual model shows the hypotheses of this research (Figure 1).
Figure 1: Conceptual model
20
3. Methods
This chapter provides an overview of the research design and how data were collected and analysed to be able to test the hypotheses of the conceptual model.
3.1 Research Design
The theoretical discussion showed that frequency of change and positive previous change experience sentiments are likely to have a positive influence on self-efficacy and subsequently have a positive influence on readiness for change. This research was based on a mediation model that aims to identify and explicate the mechanisms underlying a possible relation between frequency of change, previous change experience sentiment and the dependent variables of readiness for change by including the explanatory variable self-efficacy. Based on what was stated in the theoretical discussion, a conceptual model was designed in which the dependent variable, the independent variables and the mediator are depicted, with the expected relations. This causal study attempted to reveal statistically significant relationships between the variables with quantitative techniques and additionally used qualitative results for clarifying reasons.
3.2 Data Collection
This section describes how the data were collected, how the survey was structured and provides information about the respondents.
Questionnaire and procedures
21 distribution are the possibility for the respondent to postpone their response to an appropriate time which can improve the quality of the response and maintaining anonymity which makes respondents feel free to give the answers they want (Blumberg, Cooper & Schindler, 2011). Because the questionnaire was send to a large group, about five thousand employees of the company, it was attractive to make use of self-administered questionnaires and distribute it by e-mail. A weakness in self-administered studies is the non-response error, which will develop when the targeted respondents refuse to participate. The non-response error can be reduced by sending a reminder to fill in the questionnaire (Blumberg et al., 2011).
The questionnaire consisted of 46 items to assess the constructs readiness for change, resistance to change, self-efficacy, frequency of change, previous change experience sentiments and organizational history of change. The addition of the construct resistance to change was requested by the faculty of Economics and Business of the University of Groningen for further research purposes, but will not be included in the analysis of this research. Also, the construct organizational history of change by Metselaar (2007) and used by Bouckenooghe et al. (2007) was added but due to time constraints it was not possible to include it in the analysis.
The original scales of these constructs were written in English. The used items were translated to Dutch because of the native language of the respondents. For the translation multiple theses of other researchers were used. The translation was checked by the researcher on suitability of the translation with the original items. A pre-test was conducted to check the level of understanding and applicability of the questionnaire. Eight employees of the company filled in the questionnaire and provided feedback. The pilot group emphasized the importance of clarification about the Social Intranet and reciting that clarification on every page in the questionnaire. The comments on statements and the cover letter were used to improve the questionnaire.
22 of the questionnaire was about employees’ readiness to participate in that change process. The questionnaire was available for eleven days, among them eight working days. After one week an email was sent to all employees thanking them for responses and reminding others to complete the questionnaire. The questionnaire started with an introduction about the goals of the research, the content and clarification about the Social Intranet. As mentioned earlier, it is of great importance that all respondents have the same change process in mind when answering the statements. Therefore, above every page with statements respondents were reminded about what the statements were referring to.
Respondents were asked to give reactions to the statements on a Likert scale with a five point format, ranging from strongly disagree (1), disagree (2), neutral (3), agree (4) and strongly agree (5). The Likert scale is the most frequently used variation of the summated rating scale, which expresses either the favourable or unfavourable attitude towards the statements (Blumberg et al., 2011). The influence of different number of options on the Likert scale is still ambiguous. For this research a Likert scale of five points was used because respondents evaluated scales with a maximum of five as easy to rate, a way to quickly express their feelings and being complete enough to express their feeling satisfactorily (Wakita, Ueshima & Noguchi, 2012).
23 Population and responses
Eventually, 1,109 respondents filled in the questionnaire. This constituted a response rate of 22.6 %. The 1,109 participants included 348 men (31.4%) and 761 women (68.6%). The respondents had an average age of 45.5 years, with a range from 16 to 69 years. The respondents were employed for .3 to 43 years with an average of 13.85 years. The number of fulfilled positions ranged from one till eleven with an average of 2.2 positions within the company.
In 2013, the population of the company has an average age of 45.6 years, the average years of employment is 11 and approximately 4,500 employees work at the company. An overview of demographics between population, responses, and adjusted responses are provided in Table 1. The average age and years of employment are slightly higher in the adjusted responses than in the population and responses. The difference in the actual number of employees and the email database of the company, which contains 5.015 addresses, can be explained by not updating the database when people leave the company and also volunteers were included in the database. The population showed less male respondents and more female respondents than the responses and adjusted responses. The percentages of the latter two lie very close to each other.
Population Responses Adjusted Responses
Average age 45.6 45.5 47.46
Average years of employment 11 13.85 16.2
Gender Male 24 % 31.4 % 32.5 %
Female 76% 68.6 % 67.5 %
Table 1: Demographics of population, responses and adjusted responses
24
Control variables Responses Adjusted responses*
Responses 1109 22.64 % 916 of 1109 (85.6%)
Gender Male 31.4 % 32.5 %
Female 68.6 % 67.5 %
Age category 24 and younger 4.5 % 1.3 %
Between 25 and 34 15.6 % 12.4 % Between 35 and 44 22.4 % 22.1 % Between 45 and 54 32.3 % 35.2 % Between 55 and 64 24.7 % 28.4 % 65 and older .5 % .7 % Years of employment in
categories Three and less 16.2 % 0 %
(34 missing values) 4-10 31.7 % 37.7 %
11-17 18.1 % 21.7 %
18-24 16.1 % 19.2 %
25 and longer 17.9 % 21.4 %
Number of positions Less than 3 68.0 % 62.3 %
(12 values missing) From 3 till 5 28.5 % 33.6 %
From 6 till 8 3.2 % 3.9 %
More than eight .3 % .3 %
Aware of implementation
of Social Intranet Yes 37.5 % 39.0 %
No 62.5% 61.0 %
Table 2: Demographics of respondents
*In the adjusted responses, employees working 3 years or less at the company or work as pupil or intern are excluded.
Non-responses
25 reason for not finalizing the questionnaire was lacking clear knowledge what the Social Intranet was about, which made it, according to some respondents, difficult to answer the questions.
When analysing the 492 respondents who initially started the questionnaire, it is remarkable to see that 163 did not answer any question. When getting to the question if the respondent was familiar with the Social Intranet another 96 did not continue. After the personal characteristics like age, function and location another 40 quit. Another 92 respondents dropped out at the statements about readiness for change and resistance to change, which are specifically about this particular change process; the Social Intranet. Fifteen respondents completed the questionnaire till the last statement but probably did not continue till the end to finalize the questionnaire. Besides the previous mentioned signals the research received, the exact reasons for the high dropout remain unknown.
3.3 Measurements
This section provides an overview of the measurement scales used in the questionnaire, ordered as the theoretical discussion. A comprehensive overview of the original items and the translated items of the questionnaire can be found in Appendix B (Original Constructs). Table 3 provides a short overview of the origin of the measurement scales of the constructs.
Variables Measured by
Three dimensions of readiness for change (dependent) OCQ-C, P, R of Bouckenooghe (2009)
Frequency of change (independent)
Rafferty & Griffin (2006A) and Berneth, Walker, Harris (2011)
Previous change experience sentiment (independent)
OCQ-C, P, R of Bouckenooghe (2009) & developed by researcher
Self-efficacy (mediator) Holt, Armenakis, Field & Harris (2007) Table 3: Measurement scales of the constructs
Control variables
26 dependent and independent variables. Also, someone’s profession could influence how someone perceived the past changes. Not all control variables were used for this study but it might be interesting for the company.
Dependent variables
Readiness for change was measured as a part of the validated “Organizational Change Questionnaire- Climate of Change, Processes and Readiness” (OCQ-C, P, R) scale of Bouckenooghe,
Devos and Van den Broeck (2009), which measures the degree of willingness to cooperate in a change project. The original scale consists of nine items which represent the three components of readiness for change; affective, behavioural and cognitive. An item of affective readiness for change is “I experience the change as a positive process”. With inter alia the item “I am willing to make a
significant contribution to the change” behavioural readiness for change is tested. An item of
cognitive readiness for change is “Plans for future improvement will not come too much”. The items of affective and behavioural readiness for change were positively formulated statements and focused specifically on the Social Intranet, while the three item statements of cognitive readiness for change were negatively formulated and focused on changes in general (Bouckenooghe et al., 2009)
Resistance to change was measured by the validated 15-items scale of Oreg (2006). The scale is divided among the three components of resistance to change; affective, behavioural and cognitive which are all measured by five items. An example item of affective resistance to change is “I was afraid of the change”. With inter alia the item “I presented my objections regarding the change to management” behavioural resistance to change is tested. An example item of cognitive resistance to
change is “I believed that the change would make my job harder”. The addition of the construct was requested by the University of Groningen for further research purposes. The construct and belonging items of resistance to change were not used in this research.
Independent variables
27 & Harris (2011) about change fatigue. This scale is based on six change specific items and measures the extent to which employees feel somewhat tired as a consequence of frequent changes. Examples of this scale are “I am tired of all the changes in this company” and “It feels like we are always being asking to change something around here”.
Positive previous change experience sentiments was a difficult concept to measure because an already verified and tested scale to measure it on an individual level does not exist. Therefore, one item was adopted from the scale of Bouckenooghe, Devos & Van den Broeck (2009) and other items were developed by the researcher. From the scale of Bouckenooghe et al. (2009) about organizational climate this study adopted one item: “I am confident that changes can lead to the desired outcomes” To capture the overall sentiment employees hold about previous change experience the researcher added three items, of which these are examples: “I generally have positive experiences with previous change projects” and “I generally have an unpleasant feeling about previous change projects”.
Mediator
Self-efficacy was measured by the validated scale of Holt, Armenakis, Field & Harris (2007). This scale consists of six items of which “I have the skills that are needed to make this change work” is an example.
3.4 Data Analysis
This section describes which methods were used for the data analysis and the way the analysis was performed.
Exploring data
28 Three items of the OCQ-C, P, R of Bouckenooghe et al. (2009) where negatively formulated, while the other six where positively formulated. Therefore, the three items for cognitive readiness for change were reverse-coded (CREA1, CREA2, CREA3). Also, the second item of the scale of Holt et al. (2007) (SE2) was negatively formulated and was reverse-coded. The scale to measure previous change experience sentiment contains one negatively formulated item (CE4), which was reverse-coded. The item scores from strongly disagree to strongly agree were changed in order, to make sure that high scores on an item are always related to a positive or higher levels of that item.
Factor analysis
As guidance for performing a principal component analysis (PCA), the precisely described procedure by Pallant (2002) was used. To determine if a data set is suitable for factor analysis, two issues should be considered: sample size and the strength of the relationship among the variables. For the sample size the easiest directive is the larger, the better (Pallant, 2002). Nunally (1978) and Hair, Anderson, Tatham and Black (1998) recommend ten cases for each item to be factor analysed. In this study, 27 items were factor analysed and with almost thousand respondents this condition was satisfied. The strength of the relationship between the items is of equal concern. The factorability of the data can be assesed with the Kairser-Meyer-Olkin (KMO) measure and the Bartlett’s test of sphericity. The KMO measure should be above .6 and the Bartlett’s test of sphericity should show significance for the p-value < .05 (Pallant, 2002). In the forthcoming chapter it will become clear that both conditions were satisfied.
The factor analysis was performed twice. The items of the three components of readiness for change were factorized together and the items of the independent variables and the mediator were factorized together. This separate approach, instead of factorize all items together, was guided by theory. The used measurement scales were already validated by other researchers and the factor analysis was performed to confirm the existing scales. For example, Bouckenooghe et al. (2009) performed the factor analyses separately for the different categories in their OCQ-C, P, R scale.
29 the Oblimin rotation technique. Due to the aim to obtain several meaningful constructs (Hair et al., 1998), the oblique solution was most appropriate and applied in the factor analysis.
Reliability analysis
One of the main concerns of reliability is the internal consistency, which is commonly indicated by Cronbach’s alpha coefficient. Schmitt (1996) describes that reliabilities between .7 and .8 are acceptable and between .8 and .9 are good. Reliabilities above .9 reliabilities are described as excellent. However, reliabilities between .5 and .6 are identified as poor and below .5 are perceived as unacceptable. So ideally, the Cronbach’s alpha coefficient of a scale should be above .7 (Pallant, 2002; Nunnally, 1978). According to Henson (2001, p. 178) internal consistency is desirable because it ‘speaks directly to the ability of the researcher to interpret the composite score as a reflection of the test’s items’. However, a very high Cronbach’s alpha is actually not desirable because very high
intercorrelations among the items mean that the items are overly redundant and the construct measures too specific (Briggs & Cheek, 1986; Steiner, 2003). Very highly correlated items are usually avoided because they limit the breadth of scope of the construct instrument due to being too narrowly focused (Steiner, 2003). According to Steiner (2003) an alpha over .90 is not a desirable level of internal consistency because it indicates that items are measuring the same. It is acknowledged that the higher the correlation among the items, the higher the value of alpha. Although, a high value of alpha does not necessarily imply a high degree of internal consistency. Therefore, another explanation for too high alpha can be proposed, namely that an alpha is strongly affected by the length of the scale (Cortina, 1993).
Testing assumptions
30 centred in the middle and smaller frequencies to the extremes, indicating normality. However, in the K-S test the variables showed all significant relations, indicating of not normally distributed data (Appendix D: Histogram and K-S test). An additional method to check normality is the Normal Probability Plot, in which the point should be lying on a reasonable straight diagonal line from the bottom left to the top right, indicating no major deviation from normality (Pallant, 2002). The plots in Appendix E (Assumption checks for regression models) show no major deviation indicating normality. Also, the sample size is used in the discussion about normally distributed data: “The Central Limit Theorem reassures that, with sufficiently large sample sizes, sampling distributions of means are normally distributed regardless of the distribution of variables” (Tabachnick & Fidell,
2001, p. 72). Many researchers provided rules of thumbs to determine sufficient sample sizes; with four independent variables 60 (=15*4) cases are sufficient according to Stevens (1996) and Tabachnick et al. (2001) prefer 82 cases (=50+8*4). Hair et al. (1998) argues a sample should be at least five times the number of variables to be analysed; which in this research should be a minimum of 350 cases (7*50). The current sample size of 916 seems to be a sufficient large size. Due to the histograms, the Normal Probability Plot and the sample size, it was assumed that the data of this study was normally distributed.
The relationships among independent variables cannot be higher than .9 due to existence of multicollinearity (Pallant, 2002). The correlation analysis showed that the correlation between change fatigue and previous change experience sentiment is .588, hence below .9. For the regression models of this study the Variance Inflation Factor (VIF) values were all below 10 and the tolerance statistics all above .1 (Pallant, 2002). Therefore, the possibility of multicollinearity in the data was ruled out (Appendix D: Histogram and K-S test).
31 Correlation analysis
To determine the extent of correlation between variables, a bivariate correlation analysis was performed. The Pearson’s correlation coefficient was used since the data were interval and assumed to be normally distributed (Pallant, 2002). This correlation can be interpreted using the guidelines of Cohen (1988), who states that .10 to .29 are small correlations, .3 to .49 are medium and .5 to 1.0 are large correlations. Those categories are the same for negative numbers, but represent a negative correlation.
Regression analysis
The multiple regression analysis was used to test the hypotheses. One of the basic assumptions before using parametric test is that the data should be normally distributed (Baron & Kenny, 1986). For the execution of the multiple regression analysis the procedure of Pallant (2002) was used as guidance. Baron et al. (1986) wrote a clarifying article about mediator variables and determined three steps and three conditions to determine when a variable functions as a mediator. The first step is regressing the mediator on the independent variables, to subsequently regress the dependent variable on the independent variables. The last step was regressing the dependent variables on both the independent variables and mediator. The outcomes of the steps should satisfy three conditions. Firstly, the variations in levels of the independent variable significantly account for variations in the mediator (path a), and secondly the mediator significantly accounts for variations in the dependent variable (path b). Lastly, the independent variable must affect the dependent variable in the second step (path c). When path c reduces to zero, one single, dominant mediator is active which is called complete mediation. However, when path c significantly reduces, multiple mediating factors are present, which is called partial mediation (Baron et al., 1986). To determine the significance of the mediation effect, the Sobel test was used
32 Qualitative results
The survey finished with an optional open question, which 336 respondents used to express their opinion. The question was: Thinking about change projects you experienced in the past at the company; what would you like to see differently? The results were manually analysed, with inductive reasoning, on repetition of certain terms (see Appendix H: Overview of Qualitative results).
33
4. Results
This chapter provides the quantitative results of the factor, reliability, correlation and regression analysis, and the results of a qualitative analysis of the open question in the questionnaire. This chapter concludes with a summary of the rejected and accepted hypotheses and a visualization of the supported relationships in a model.
4.1 Factor and reliability analyses
Dependent variable
34 Items Affective readiness for change Behavioural readiness for change
Cognitive readiness for change AREA2 .943 AREA3 .917 AREA1 .889 CREA1 .925 CREA2 .911 IREA2 .952 IREA3 .922 IREA1 .132 .831 Eigenvalues 4.594 1.046 1.279 % of variance 51.040 11.626 14.208 Cronbach’s α .912 .905 .816
Table 4: Rotated Factor Loadings: Readiness for change components
Independent variables and mediator
35 both extracted from further analysis. The principal component analysis was performed with Oblimin rotation. FOC7 is deleted because of relatively high cross loading on frequency of change (FOC5 till FOC8) and change fatigue (FOC1 till FOC3).
Items Change Fatigue Self-efficacy Frequency of change Previous change experience sentiment SE1 .895 SE3 .889 SE4 .793 SE5 .785 SE6 .732 FOC1 .815 FOC3 .634 FOC4 .850 FOC5 .827 FOC6 .715 FOC8 .621 CE1 .895 CE2 .888 CE3 .747 CE4 -.635 Eigenvalues 5.537 3.057 1.627 1.046 % of variance 30.759 16.982 9.040 5.813 Cronbach’s α .836 .874 .544 .846
Table 5: Rotated Factor Loadings: Independent variables and Mediator
The construct frequency of change was aimed to be measured by two concepts, namely frequency of change and change fatigue. However, the factor analysis clearly revealed loadings as two components en therefore these independent variables are handled as two separate constructs. Consequently, additional hypotheses needed to be formulated and the conceptual model needed to be adjusted.
Change fatigue is considered to be a negative reaction (Oreg, Vakola & Armenakis, 2011) which occurs when employees experience too much change. Change fatigue ultimately leads to the development of exhaustion and other negative consequences (Berneth et al., 2011). Furthermore, change fatigue influences employees’ psychological uncertainty which has consequences for their adaptive resources (Bernerth et al., 2011). When employees find it difficult to adapt to new situations, it is expected that employees do not have the belief to possess the capabilities to perform the required tasks. Therefore, it is hypothesized:
36 H7A: Self-efficacy mediates the negative relationship between change fatigue and recipients’ affective
readiness for change, such that higher levels of frequency of change lead to higher levels of self-efficacy; and consequently to more affective readiness for change.
H7B: Self-efficacy mediates the negative relationship between change fatigue and recipients’
behavioural readiness for change, such that higher levels of frequency of change lead to higher levels of self-efficacy; and consequently to more behavioural readiness for change.
H7C: Self-efficacy mediates the negative relationship between change fatigue and recipients’ cognitive
readiness for change, such that higher levels of frequency of change lead to higher levels of self-efficacy; and consequently to more cognitive readiness for change.
Reliability analyses
37 reduced construct validity. Frequency of change will be measured with ‘Change frequently occurs in my unit’ (FOC1) because of the high loading in the factor analysis and the items is a good
representation of the meaning of the construct.
Items Cronbach’s α Meaning
Affective readiness for change .912 Excellent
Behavioural readiness for change .905 Excellent
Cognitive readiness for change .816 Good
Change fatigue .836 Good
Previous change experience sentiment .846 Good
Self-efficacy .874 Good
Frequency of change - -
Table 6: Summary of Cronbach's alpha coefficients for all constructs
The meanings of the levels of reliability in Table 6 are assigned according to the classification of Schmitt (1996). But as mentioned before, an alpha above .90 is not a desirable level of internal consistency because it indicates that items are measuring the same (Steiner, 2003) or are affected by the length of the scale (Cortina, 1993). However, the latter is not presumably because affective and behavioural readiness for change were both measured by three items. It might indicate that affective and behavioural readiness for change are focused too narrowly. Although, both scales have been developed to solely measure one dimension of the tripartite attitude readiness for change. For this reason and due to only fractionally exceeding the limit, the current compositions of the constructs were retained.
4.2 Descriptive analysis
This section provides the results of the Pearson’s correlation analysis. Table 7 shows a summary of the number of cases, means, standard deviation and two-tailed correlations. All correlations (r > .1) are significant at the level of .01.
38 positive nor negative. The respondents ranked their previous change experience sentiments rather positively (3.21). Respondents feel that changes occurred frequently in their units (3.66). Though, frequency of change showed a large standard deviation (SD=.91), indicating that the respondents differed widely in their experience.
In Appendix G (Cross tabulations with control variables) cross tabulations of age, gender, and function category against all components of readiness for change are added. Between the recipients’ age categories there are no large differences in their level of readiness for change, they ranked their readiness for change equally high. Though, the dispersion in an age category is larger at cognitive readiness for change than at affective and behavioural readiness for change. When splitting the cases at gender, no real differences occur at affective and behavioural readiness for change. Though, at cognitive readiness for change the percentage of female respondents choosing neutral was much higher than the percentages male respondent. Also, the percentage male respondents choosing ‘agree’ or ‘strongly agree’ is higher than the percentage of female respondents. When looking at the function in categories it became obvious that those in General Functions ranked their affective, behavioural and cognitive readiness for change higher than their colleagues in Direct and Supportive Care.
Age, gender and employment
A predictable result is that age and years of employment correlates highly and positively with each other (r = .551). Also, recipients working longer with the company (years of employment) have fulfilled more positions (r = .375). The number of fulfilled positions showed small, positive correlation with age (r = .159) and behavioural readiness for change (r = .117), and negatively with gender (r = -.167) and change fatigue (r = -.104). Self-efficacy showed also to be negatively correlated with age (r = -.232) and years of employment (r =.-183). It indicates that the older an employee is and the more years of employment an employee has, the lower someone’s self-efficacy.
Previous change experience sentiment, Change Fatigue & Frequency of Change
39 cognitive readiness for change (r = .550). These correlations indicated that the more positive the previous change experience sentiments were, the more ready someone is to change, on all components of the attitude. PCE showed also a positive significant, though small, relation with self-efficacy (r = .237), implying that when someone remained a positive sentiment of previous change experiences, the higher someone’s self-efficacy will be.
Change fatigue showed negative significant correlations with all components of readiness for change. The construct correlates largely significant with cognitive readiness for change (r = -.521), and small significantly with affective and behavioural readiness for change (r = -.304, r = -.302). These correlations indicated that the higher someone’s change fatigue, the less ready someone is to change on all components of the attitude. Change fatigue also correlates largely significant, negative with PCE (r = -.588) and showed a negative, small significant correlation with self-efficacy (r = -.259). Frequency of change showed a negative, small significant correlation with previous change experience (r = -.129) and positive correlation with change fatigue (r = .284).
Self-efficacy
Self-efficacy showed positive significant correlation with all components of readiness for change, a large significant relation with affective readiness for change (r = .60), a medium significant relation behavioural readiness for change (r = .469) and a small significant relation with cognitive readiness for change (r = .256). These correlation indicate that the higher the level of self-efficacy, the higher the level of readiness for change.
Readiness for change
40
Table 7: Bivariate Pearson’s correlation analysis including means and standard deviations Correlations Constructs N Mean SD 1 2 3 4 5 6 7 8 9 10 1. Years of employment 894 16.20 9.77 - 2. Age 916 47.46 10.16 .551** - 3. Gender 916 1.67 .47 -.215** -.234** - 4. Number of positions 906 2.43 1.50 .375** .159** -.167** -
5. Affective Readiness for Change 916 3.65 .76 -.071 -.053 .029 .054 -
6. Behavioural Readiness for Change 916 3.45 .86 -.021 .018 -.019 .120** .614**
7. Cognitive Readiness for Change 916 3.49 .76 -.015 -.002 -.090 .069 .365** .386** -
8. Previous Change Experience Sentiment 916 3.21 .68 -.002 .007 -.034 .065 .333** .385** .550** -
9. Self-Efficacy 916 3.88 .73 -.183** -.232** .023 .064 .600** .469** .256** .237** -
10. Change Fatigue 916 2.98 .81 .022 .000 .076* -.104** -.304** -.302** -.521** -.588** -.259** -
11. Frequency of Change 916 3.66 .91 .049 -.012 -.046 .037 -.063 -.004 -.034 -.129** .003 .284**
**. Correlation is significant at the .01 level (2-tailed). Small significant correlation (.1-.29)
Medium significant correlation (.3-.49)
41 4.3 Hypotheses testing
For testing the hypotheses, a multiple regression analysis was performed. The results of the regression analysis are depicted in Table 8 till 10 and provide the unstandardized beta-coefficients, the significance indication, the standard error and the explained variance of the dependent variable. The unstandardized beta-coefficient was used because all constructs were tested on the same scale, which made conversion to the same scale needless (Pallant, 2002).
The first step of mediation was regressing the mediator self-efficacy on the independent variables previous change experience sentiment, change fatigue and frequency of change. The results are depicted in Table 8. The analysis showed that the three independent variables explained 8.4 % of the variance in self-efficacy (R2=.084; p < .05). Previous change experience sentiment, change fatigue and frequency of change are significantly related with self-efficacy (β = .134 and β = -.187, and β = .063, p < .05). Though, change fatigue and self-efficacy showed a negatively relation. The first condition of Baron et al. (1986) about a significant relation between the independent variables and the mediator holds. The previous correlation analysis determined a correlation between self-efficacy and the components of readiness for change (Table 7), which satisfies the second condition of Baron et al. (1986).
Predictor variables Self-efficacy β Significance Standard Error
Previous change experience .134 .001* .042
Change Fatigue -.187 .000* .035
Frequency of Change .063 .019* .027
R2 .084 - -
Adjusted R2 .081 - -
*p < .05, N = 916
Table 8: Step 1: Multiple regression of independent variables on mediator
