Social power and innovation implementation : motivated compliance : a mediated model of power and implementation

(1)

Social power and innovation

implementation

Motivated compliance: a mediated model of power and implementation

Abstract

This thesis explores the underlying dynamics regarding a stable, negative relationship between

centralization and innovation implementation. The use of power is connected to three forms of

commitment to the implementation of organizational change, and via these forms, to compliance

towards change. A new factor structure of the interaction model of interpersonal influence is

compiled to test four overarching power factors: rationality, relationship based on social norms,

relationship based on economic norms, and the carrot and stick model. Data was collected by means

of a survey and analyzed using SPSS software, specifically the process add-in tool for regression.

Results partially support the hypotheses, with the power bases loading slightly different than

expected. Furthermore, two soft power factors were found to be indeed mediated, yet for harsh

power no such mediator effect was found.

Hugo Besselse 11420448, Final version master thesis MSc Business Administration – Entrepreneurship and innovation University of Amsterdam, supervisor: Belazs Szatmari, 23 June 2017

(2)

2

Statement of originality

This document is written by Student Hugo Besselse who declares to take full responsibility

for the contents of this document. I declare that the text and the work presented in this

document is original and that no sources other than those mentioned in the text and its

references have been used in creating it. The Faculty of Economics and Business is

responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Introduction

Newton’s first law of motion says that an objects at rest stays at rest and an object in motion stays in

motion with the same speed and in the same direction unless acted upon by an unbalanced force. It

seems that humans act in a similar manner: Nicolle et al. (2011) found that people tend to stick to a

status quo, even if this is the suboptimal rational choice. This finding has been replicated in several

studies, labelled the status quo bias and shown in MRI scans by highlighting certain regions of the

brain (Fleming et al., 2009). An explanation lies in the concept of uncertainty avoidance: abandoning

the status quo entails risk and uncertainty, whereas people are shown to be generally risk avoidant

(Kahneman & Tversky, 1979). On the other hand, organizations require an ever increasing focus on

change: short product life cycles, changing market requirements and intense domestic and

international competition due to a globalized economy. New products and processes need to be

developed and implemented to maintain a competitive advantage and continuously exploit the

changing playing field. Can men be their own unbalanced force, if circumstances require it?

So far, research tentatively suggests not. Approximately 50% of innovation implementation efforts

result in failure, or not meeting projected targets (Baer & Frese, 2003, Repenning & Sterman, 2002,

Aiman-Smith & Green, 2002). Stumbling blocks mentioned by Klein & Knight (2005) include imperfect

design of (technological) innovations, complexity of innovations and acquiring new skills, top-down

implementation decision making, change of roles, norms and routines, an initial drag of performance

and organizational norms oriented towards maintaining the status quo. For organizations, it is

essential to be able to fully realize the benefits of innovations, especially if adoption of the

innovation has substantial costs. Damanpour (1991) found an average correlation between

centralization and innovation implementation of -.173, yet a deeper understanding of the underlying

dynamics lacks: 21 studies found a negative correlation, whereas 10 studies found a positive one. The

understanding of power dynamics has evolved over time, but has not been connected to innovation.

In this thesis, I bring these concepts together to create a deeper understanding of how power affects

(4)

4

My research utilizes the interaction model of interpersonal influence developed by Raven et al. (1998)

combined with the model of Herscovitch & Meyer (2002) regarding commitment to organizational

change. I collected data by using a survey regarding people’s experiences of organizational changes

that have been implemented in a top-down manner. This way, I can connect the use of power bases,

different forms of commitment to change, and the degree of compliance towards such a change. The

main question guiding this thesis is:

How does the use of various power bases affect compliance to change via commitment to change?

In the next section, I present an overview of the relevant theories regarding power, innovation and

commitment to change. Afterwards, I formulate my hypotheses and show the conceptual model. The

third section summarizes the methods of data collection and analysis, whereas the fourth section

presents the results of the analysis. The results are discussed and the hypotheses accepted or

rejected in the fifth section and I conclude with the limitations of this study and possible future

orientations. Throughout this thesis, innovation implementation and change are used

interchangeably. Since the implementation of an innovation by its nature entails change, the

constructs can be used in such a way. The survey utilized organizational change though, since I

perceived that as being more intuitive. In table 1, an overview of definitions is presented.

Construct Definition Example Source

Innovation A product or practice that is new to its developers and/or potential users.

A new year’s resolution to become more healthy by exercising.

Klein & Knight, 2005.

Innovation adoption

The decision to use an innovation. Buying an exercise machine for in-home use.

Innovation implementation

The transition period during which individuals ideally become increasingly skillful, consistent and committed in their use of the innovation.

The actual use of the exercise machine.

Power The potential or ability of an agent to bring change in attitudes, behavior or belief by using resources available to him or her.

Being informed by a doctor that one’s health is at risk and exercising is required.

Raven, 2008.

Affective commitment to change

A desire to provide support for the change based on a belief in its inherent benefits.

Truly believing that exercising will be beneficial for one’s health.

Herscovitch & Meyer, 2002.

(5)

5 Continuance

commitment to change

A recognition that there are costs associated with failure to provide support for the change.

Health risks resulting from not exercising. Herscovitch & Meyer, 2002. Normative commitment to change

A sense of obligation to provide support for the change.

An obligation towards friends who have always been supportive.

Herscovitch & Meyer, 2002.

Table 1: Definitions of constructs

Theory

In this section, I describe relevant theories regarding innovation, power and commitment to change.

Three main streams of literature are covered in the field of innovation: organizational characteristics,

innovation characteristics, and implementation. For power, the interaction model of interpersonal

influence is described, and studies testing the model evaluated. Lastly, the study of Herscovitch &

Meyer (2002) is explained, which incorporates three forms of commitment to change.

Innovation

I start with the meta-analysis conducted by Damanpour (1991), who analyzed various studies

concentrating on innovation and organizational characteristics. Table 2 gives an overview of all

variables analyzed over time. Formalization, managerial tenure and vertical differentiation are

denoted with an asterisk, since the hypothesized relationship was rejected.

Independent variable

Expected relation

Argumentation Moderator

Specialization Positive A variety of specialists provides a broader knowledge base, increasing the cross-fertilization of ideas.

Type of organization

 Profit/non profit

 Manufacturing/ service Functional

differentiation

Positive Coalitions of professionals in differentiated units elaborate on and introduce changes.

Type of innovation

 Administrative/ technical

 Product/process

 Radical/ incremental Professionalism Positive Increase of boundary-spanning activity, self-confidence,

and commitments to move beyond the status quo.

Stage of adoption

 Initiation/ implementation Formalization* Negative Flexibility facilitates innovation, whereas formalization

inhibits it.

Scope of innovation

Centralization Negative Concentration of decision-making authority prevent innovation; dispersion is required.

Managerial attitude towards change

Positive Managers’ attitude towards change creates an internal climate conductive to implement innovation.

Managerial tenure*

Positive Longevity of managers provides legitimacy and knowledge of tasks and political processes.

(6)

6 Technical

knowledge resources

Positive Higher levels of technical knowledge facilitates understanding of new technical ideas.

Administrative intensity

Positive A higher proportion of managers facilitates innovation with more leadership, support and coordination.

Slack resources Positive Slack resources allow organizations to explore new ideas, absorb failure and bear the costs of implementing innovations.

External communication

Positive Exchanging information with the environment facilitates innovation.

Internal

communication

Positive Facilitates dispersion of ideas, resulting in higher levels of cross-fertilization. Increases favorability to the survival of new ideas.

Vertical

differentiation*

Negative Hierarchical levels inhibit communication between levels and flow of innovative ideas

Table 2: meta analysis organizational characteristics and innovation

From this meta-analysis, I concentrate on the dynamics of centralization and innovation. In image 1

below, the correlations between centralization and innovation are shown, with all moderators. The

results of a T-test of differences between mean correlations indicates that only profit/nonprofit (T

value 2.016, significant at p<.05), and low/high scope (T value 5.659, significant at p<.01) are

significant moderators. Damanpour (1991) set the cutoff point for the low/high group at five

innovations. Overall, centralization seems to have a stable, negative mean correlation with

innovation regardless of organizational type, innovation type or stage of adoption. Interestingly, a

low scope of innovation flips the relationship to a positive one (.148, versus -.206), which might

explain positive correlations found in other studies. Since my thesis concentrates on one single

innovation, this finding is essential to highlight since it could inflate the results to a positive

relationship between power and innovation implementation. Intuitively, one could argue that the

use of power to implement a single or few innovations could indeed have a positive effect, but that

(7)

7

Image 1: mean correlations amongst moderators

Although organizational characteristics are not the scope of my research, it makes sense to evaluate

the meta-analysis to evaluate potentially important control variables or moderators and present a

basic overview of variables that have been studied. Another stream of literature reflects on the

characteristics of innovations and their effect on adoption or implementation of an innovation, which

is discussed next.

In this framework, adoption describes the process of deciding to acquire an innovation, whereas

implementation concentrates on the actual usage. Tornatzky & Klein (1982) found 10 attributes of

innovations which have consistently been tested in adoption research: 1) compatibility, 2) relative

advantage, 3) complexity, 4) cost, 5) communicability, 6) divisibility, 7) profitability, 8) social approval,

9) trialability, 10) observability. However, several points of critique are mentioned by Tornatzky &

(8)

8

Most importantly, the issue of subjectivity is raised: innovation characteristics are potentially

evaluated based on the adopter’s framework. An example would be two organizations who are

considering adopting the same innovation and incur the same costs. If we assume that these

organizations have different revenues (e.g. one 10,000 dollars, the other 20,000 dollars), the

evaluation of the adoption costs of the innovation would also differ: the one making more revenue

would perceive the costs as smaller. For a characteristic such as complexity, this argumentation

makes even more intuitive sense: an innovation that is seen as highly complex for one organization,

could be seen as little complex by another. Tornatzy & Klein (1982) posed the question whether the

explanatory power of innovation characteristics is completely dependent on situational factors: ‘How

diverse are perceptual judgements about the same innovation?’ (Tornatzy & Klein, 1982, p.2). This

question seems unanswered as of today.

Furthermore, in the studies analyzed by Tornatzy & Klein (1982), methodological issues arise in

generalizing the results. These issues include studying one innovation characteristic (46.7% of

studies), studying one innovation (53.3% of studies), studying only adoption as a dependent variable

(93.3% of studies) and having a sole, non-organizational person as the adopter of an innovation (57.3%

of studies) (Tornatzy & Klein, 1982). If one innovation is studied, the results could potentially reflect

attributes solely applicable to that innovation and thus limit the generalizability. Concentrating on a

single characteristic does not allow research regarding potential interrelationships between

characteristics and does not allow tests regarding the predictive power of each attribute (e.g. costs

being a stronger determinant than complexity). Additionally, in line with the first point, comparing

the characteristics of different studies is difficult since the evaluation could be subjective. The lack of

implementation as a dependent variable poses another problem: the relationship between

innovation characteristics and adoption might be different than their relationship with

implementation (Tornatzy & Klein, 1982). In practice, this would mean that organizations could use

(9)

9

lack of implementation. Finally, the adoption process of an individual might not be generalizable to

the adoption process of an organization.

This concludes the chapter regarding innovation adoption. It can be concluded that although many

studies are performed in this field, the results cannot be aggregated to provide a clear (quantitative)

overview such as the meta-analysis of Damanpour (1991). The last chapter of innovation research

concentrates on the implementation of innovations. The model of Klein & Sorra (1996) is described

and evaluated based on several empirical studies of innovation implementation.

Klein & Sorra (1996) define implementation as the process of gaining targeted organizational

members’ appropriate and committed use of an innovation. In most studies, implementation is

assumed to be preceded by a formal adoption decision, since in organizations the success of an

innovation depends on the coordinated, committed use of various members (Klein & Sorra, 1996). In

this research stream, innovation is seen from the user-based stage model, which indicates that

innovation is defined as a new technology or practice to be utilized by organizational members for

the first time, regardless of other organizations’ potential usage of the innovation beforehand (Klein

& Sorra, 1996).

The main model of innovation implementation is conceptualized by Klein & Sorra (1996), which has

(10)

10

Image 2: conceptual model of innovation implementation

Firstly, climate for implementation can be seen as a broad conceptualization of organizational

policies and practices (e.g. training in innovation use, user support services, time to experiment,

praise from supervisors for innovation use, job reassignment for noncompliance, budgetary

constraints and financial incentives). It involves aggregated employees’ perceptions of the extent to

which usage of an innovation is rewarded, supported and expected (Klein & Sorra, 1996). It is

proposed to mediate the implementation effectiveness via skills, (dis)incentives and obstacles.

The second component of innovation implementation is innovation-values fit and describe the

degree to which the innovation’s values fit with key employees’ values. Again, this construct is

aggregated at the group or organizational level and closely reflects organizational values or culture,

meaning that if an innovation matches the organizational values, it is expected to match many

individual members’ values of that organization. Differences in values between groups, such as

higher and lower management, could result in mismatches of innovation values, which might explain

(11)

11

Sorra, 1996). Innovation-values fit is expected to be mediated by commitment to explain

implementation effectiveness, such that high value fit increases commitment.

The interaction between implementation climate and innovation-values fit is summarized in the table

below. Within this model, I expect power and its resulting policies to be a part of the implementation

climate (e.g. reward systems based on innovation use). My conceptualization of commitment could

be seen as motivated compliance, and potentially be summarized incentives and disincentives (e.g.

an incentive to comply based on normative commitment, resulting from a power base).

Table 3: interaction of implementation climate and innovation-values fit (Klein & Sorra, 1996).

Dong et al. (2008) showed support for the antecedents and mediators of the model, as shown in

image 3 below: implementation climate was found to affect skills, incentives and the absence of

obstacles and innovation-values fit correlated with user affective commitment. Affective

commitment, skills, incentives and absence of obstacles correlated with implementation

(12)

12

Image 3: conceptual model tested (Dong et al., 2008)

Implementation climate was measured by seventeen items, capturing four dimensions; mean

emphasis, goal emphasis, task support and reward emphasis (Dong et al., 2008). Mean emphasis had

three items and describes the extent to which managers make known the methods and procedures

that employees need to utilize to perform their jobs (Dong et al., 2008). Goal emphasis also

incorporated three items and was defined as the extent to which managers exemplify the desired

outcomes and standards that employees need to accomplish (Dong et al., 2008). Task support

involved six items and is described as the perception of employees that they are adequately supplied

with required material, equipment, services and resources (Dong et al., 2008). Lastly, reward

emphasis had five items and described the perception of employees to which degree organizational

rewards are allocated based on their performance in using the new system (Dong et al., 2008).

Innovation-values fit was measured by three characteristics: quality of work output (six items),

information locatibility and accessibility (four items) and collaborative flexibility and cooperation

(three items) (Dong et al., 2008). The scales were adapted or utilized from earlier research,

(13)

13

Results from another test of the model are summarized in image 4 below. The model is revised in

the sense that management support and financial resource availability are antecedents of

implementation climate and implementation policies and practices, respectively.

Image 4: results from the revised model (Klein et al., 2001)

In this research, management support and implementation policies and practices were separated.

For management support, six items were constructed, whereas implementation policies and

practices were measured with a 36-item scale, including (1) seven items regarding training, (2) eight

items involving software and hardware quality, (3) five items about rewards, (4) five items measuring

the availability of user support, (5) five items concerned with the time to experiment and (6) six

items regarding communication to employees (Klein et al., 2001). Implementation climate was

measured with a seven item scale assessing the aggregated perceptions of support for the innovation

(Klein et al., 2001). Within this adapted model, it is less intuitive to see the role of power based on

the face validity of the constructs. My expectation would be that power influences the aggregated

perceptions of support for the innovation, thus implementation policies and practices, specifically via

(3) rewards and (6) communication.

This concludes my literature overview of innovation. We have seen that the field is broad, and

studies are often dispersed. Damanpour (1991) provides a solid starting point in the direction of

(14)

14

pinpoint the top ten characteristics studied. Lastly, innovation implementation is discussed and the

model of Klein & Sorra (1996) was presented. Several studies have been discussed to show empirical

evidence for the antecedents, mediators and outcomes. I theorized how my application of power

bases would fit within the model. The next section dives into power.

Power

The first conceptualization of the model of social power is developed by French & Raven (1959) and

incorporated five power bases: reward, coercion, legitimate, referent and expert. Several decades of

refining studies and empirical tests later, fourteen power bases emerged (Elias, 2008). Reward and

coercive had been split into impersonal and personal, legitimate in position, equity, reciprocity and

dependence and information power has been added. Additionally, expert and referent power have

been split in positive and negative forms and information power has been classified as direct or

indirect (Elias, 2008). I built my thesis and research on the model incorporating eleven bases of

power though. Empirical work has not caught up with all theoretical developments (Elias, 2008).

Scales measuring the eleven bases are available, whereas the additional three power bases seem to

miss empirically tested and validated scales (Elias, 2008, Raven et al., 1998, Pierro et al., 2008).

The individual bases can be summarized in two overarching factors: harsh and soft. In empirical

studies, power bases are often categorized as such (Pierro et al., 2008). Harsh power can be seen as

leaving less freedom to subordinates, compared to soft power. Raven et al. (1998) found that

generally people are more willing to comply to soft power tactics than harsh ones. Pierro et al. (2008)

replicated this finding in different organizational and national contexts, which increases the external

validity. The stream of literature investigating the perspective of the target who is being influenced,

is a relatively new one. Beforehand, much attention has been given to the viewpoint of the

influencing agent (Pierro et al., 2008).

Table 4 summarizes the power bases and an example of the overarching bases in an American

(15)

15

determined: position power was seen as soft in the Israeli sample, compared to harsh in the

American sample, yet personal reward power was seen as harsh in the Israeli sample and soft in the

American one.

Harsh Legitimate power of reciprocity Obligation to comply based on earlier favors by a supervisor.

Impersonal coercive power Threat of impersonal punishment (e.g. supervisor would fire a subordinate).

Legitimate power of equity Supervisor demands compliance to compensate for his hard work or sufferance because of the subordinate.

Impersonal reward power Promise of impersonal compensation (e.g. promotion).

Personal coercive power Threat of personal punishment (e.g. a supervisor dislikes a subordinate).

Legitimate power of position Supervisor’s right or norm to prescribe behavior for a subordinate

Soft Expert power Subordinate relies on the supervisor’s perceived superior knowledge.

Referent power Subordinate complies because of positive identification with supervisor (e.g. liking, looking up to).

Informational power Supervisor presents persuasive material or logic.

Legitimate power of dependence Subordinate’s social responsibility to assist those in need. Personal reward power Promise of personal compensation (e.g. liking of supervisor).

Table 4: individual bases of power and overarching bases

Now, I expand on the eleven bases of power. Afterwards, the interaction model is introduced and

shortly described. This section finishes with insights involving socially (in)dependent change and the

requirement of surveillance.

First is the legitimate power of reciprocity, which draws on the norm of repaying earlier favors

(Raven et al. 1998). The reciprocity norm is widely recognized and willingly or unwillingly utilized to

pressurize people to comply (Cialdini, 2007). A similar power base is legitimate power of equity,

which also draws on an obligation of compensation for earlier favors or hardship. The difference lies

in the balance between the two: whereas reciprocity favors are balanced (e.g. working one hour

extra tomorrow for leaving one hour earlier today), equity favors are not (e.g. working one hour

extra tomorrow, because one’s supervisor has worked for months to get you at your current

(16)

16

refers to the right of a supervisor to request compliance, based on an obligation towards hierarchy:

those higher in ranking may ask certain things from those lower in ranking. Dependence power,

interestingly, is sometimes also referred to as the power of the powerless and is based on the social

norm to help others in need: a supervisor who informs subordinates he is unable to finish a task

without other employees’ compliance draws on the dependence power form.

The model of interaction is shown in image 5 below. Supervisors generally start with a motivation to

influence, which could be rise based on various reasons. One classification of this motivation is

instrumental or intrinsic, in which instrumental refers to the motivation to utilize power to achieve a

certain goal and intrinsic lies in a need to seek power, or willingness to have influence (Elias, 2008).

This motivation partially influences the assessment of available power bases, during which a

supervisor selects appropriate power bases (Elias, 2008). This can be done based on internal factors,

such as motivation, factors relating to the target, such as its openness to certain power bases and

organizational factors, such as experienced backlash for harsh power bases (Elias, 2008). Preparation

for influence attempts refers to setting the stage to amplify the power tactic’s effect, for example

doing favors to invoke reciprocity. The choice of power base and influence attempt refers to the

actual influencing and the received feedback: success or failure to influence a target, and outcomes

related to the influence attempt (Elias, 2008). This feedback can be incorporated in future influence

(17)

17

Image 5: the power/interaction model

Additionally, Raven (2008) mentions power sources in context of socially independent change,

socially dependent change with surveillance necessary and socially dependent change without

surveillance necessary. This distinction is interesting for managers, since a change without

surveillance would imply that employees internalize a new norm and managers do not require

surveillance to check for compliance.

Socially independent change is said to be preceded by information power: providing persuasive or

logic material to explain why a certain change would be beneficial (Raven, 2008). Socially

independent refers to the supervisor not required to be present after utilizing the information power

tactic (Raven, 2008). Socially dependent change with surveillance, on the other hand, corresponds

with reward and coercive power, both impersonal and personal. In this case, compliance requires the

presence of the supervisor, since a subordinate would only comply if he or she believes that the

supervisor is capable of determining whether compliance has occurred (Raven, 2008). Coercive

power requires even more surveillance than reward power, since for reward power the subordinate

would benefit by letting the supervisor know that compliance has occurred (Raven, 2008). The last

form is power leading to socially dependent change, without surveillance, which corresponds to

(18)

18

because of the hierarchy norm, because he or she has faith in the knowledge of the supervisor or he

or she looks up to and wishes to identify with the supervisor (Raven, 2008).

This concludes the section regarding power. The eleven bases of power have been described, and the

model of interpersonal influence is explained. Although this model will not be used in my thesis, it

provides background information in which the results can be grounded. The last part of the theory

section refers to commitment to change, a model that is connected to the power bases.

Commitment to change

Herscovitch & Meyer (2002) extended their existing model of organizational commitment to include

commitment towards organizational change and tested it with three studies. By including

commitment to change, the model improved significantly in its predicting power in explaining

compliance to change (∆R2=.16). Commitment to change was defined as affective, normative and

continuance: affective commitment refers to an individual’s belief in the value of a change, whereas

normative relates to an obligation and continuance refers to the costs associated with

noncompliance (Herscovitch & Meyer, 2002).

Compliance was measured on a continuum scale with anchor points 0-20 active resistance

(demonstrating overt behavior to ensure failure of the change), 21-40 passive resistance

(demonstrating subtle behavior to prevent success of the change, 41-60 compliance (demonstrating

minimum cooperation by going along, but reluctantly), 61-80 cooperation(demonstrating support by

exerting effort, going along and making modest sacrifices and 81-100 championing (demonstrating

extreme enthusiasm by going above and beyond formal requirements and promoting the change)

(Hescovitch & Meyer, 2002). Additionally, in two of the three studies, a scale for compliance

(three-item), cooperation (eight-item) and championing (six-(three-item), was utilized. In the third study, only the

continuum was used as an outcome variable, to maintain a reasonable length of the survey and

(19)

19

Image 6 below summarizes the scores of different levels of commitment and their respective

compliance. Affective commitment seems to be the strongest sole determinant of compliance,

although the difference compared to normative commitment is not significant. Interestingly, a high

level of continuance commitment seems to temper compliance: groups with high continuance

commitment and high levels of other forms of commitment show lower compliance than groups with

only other high forms of commitment. Solely high continuance commitment results in mere

compliance (44). Interestingly, even no form of commitment results in compliance, although the

score is the bare minimum (41). This finding might be a presentation of the status quo bias as

described in the introduction.

This difference between compliance levels is important in successfully implementing innovations:

affective and normative commitment are required to exert extra effort, work cooperatively or to

truly champion a new initiative. Continuance commitment is enough to ensure compliance, which

means that employees do no more than formally required, and might do so reluctantly. For

managers, knowing how to invoke one type of commitment should benefit the implementation

success and ultimately the long run.

(20)

20

To the best of my knowledge, there is currently one research that has investigated the effect of

supervisor’s power bases on employees’ co-operative behavior regarding change, which is the work

of Munduate & Dorado (1998). However, the focus of the study also lies in the intercorrelations of

power bases besides co-operative behavior: it is argued that power bases based on position (reward,

coercive, position) have an effect on personal power bases, such as expert power and referent power.

Findings include a positive relationship between reward power and referent power, and one

between expert power and referent power.

Co-operative behavior, which conceptually shows much overlap with my measurement of

compliance (cooperation and championing, specifically), was found to have a positive relationship

with referent power. An interesting conclusion is the notion that personal power bases have a larger

effect than formal ones. This would be supported by Pierro et al. (2008), who indicates that people

are more willing to comply to soft power tactics compared to harsh power tactics. However, one

limitation of this study would be exactly the previous statement: reward power was measured using

the older classification, which did not make an distinction between personal reward and impersonal

reward, whereas the latter is expected to be more formal and based on hierarchy. Another limitation

would be the limited use of power bases: the older models of six power bases was used, instead of

eleven.

With this, I conclude the theory section. In the last paragraphs, I wrote about the model of

commitment to change and expanded on a study that tested a supervisor’s power usage in increasing

co-operative behavior towards an organizational change. In the next section, I draw my conceptual

model and formulate the hypotheses to be tested. First, I explain my reasoning for each hypothesis,

and at the end of the section, I summarize the hypotheses and show the model.

Hypotheses

The first hypothesis involves the distribution of power bases in four overarching factors, which are

(21)

21

collected data. The results of Raven et al. (1998) provide partial support of this hypothesis: in an

exploratory factor analysis 7 factors were determined and expert power and information power held

a common factor. Similarly, reciprocity power and equity power share a common factor and reward

and coercion impersonal. Lastly, personal reward and personal coercion load on the same factor.

Power bases Overarching factor Interpretation of factor Expert power, information

power

Soft1 Rationality

Referent power, dependence power

Soft2 Relationship oriented based on

social norms Position power, reciprocity

power, equity power

Harsh1 Relationship oriented based on economic norms

Reward impersonal power, coercive impersonal power, personal reward power

Harsh2 Carrot and stick model

Table 5: four overarching factors of individual power bases

The changes that I propose to those factors is that referent power and dependence power group

together, position power is added to reciprocity power and equity power and reward personal power

and coercion personal power groups together with reward impersonal power and coercive

impersonal power.

Grouping expert and information power together fits intuitively: both refer to rationality as a

persuasion tactic. The difference between expert power and information power is the degree to

which the information is shared and understood by a subordinate (Elias, 2008). Expert power relies

on a supervisor being perceived as more knowledgeable compared with the subordinate, whereas

information power relies on the presentation of persuasive logic or information (Elias, 2008). I argue

that expert power could potentially change a subordinate’s perception of the value of a change,

although to a lower degree, as presenting information could. Therefore, information power and

expert power would correlate with affective commitment to change: information regarding the

benefits of a change would alter employees’ evaluations of its value. Similarly, drawing on expert

(22)

22

However, Oreg (2006), found a positive correlation between information and cognitive resistance to

change, which can be described as the mental evaluation of the worth and benefit of a change. Two

explanations are that the content of the information can be of influence, and that the relationship

between information and cognitive resistance is non-linear; too little, or too much information

results in cognitive resistance. The amount of information lies outside of the scope of this research,

but informational power is assumed to concentrate on the potential benefits of a change (Elias,

2008). Similarly, expert power is conceptualized as a supervisor being more knowledgeable and using

this superior knowledge to the benefit of subordinates (Elias, 2008). In future studies concerning the

interaction model of interpersonal influence, negative expert power can be distinguished and direct

or indirect forms of information.

Referent power and dependence power might be less intuitive to group together, but both power

bases seem to rely on a social norm regarding relationships, which corresponds to the identification

process of influence as described by Kelman (1958): an individual adopts required behavior because

that is associated with the desired relationship. Referent power draws on the social norm of liking,

being a human tendency to be willing to do more for people that we like, compared to people that

we do not like (Cialdini, 2007). Dependence power relies on the norm of social responsibility to

persuade compliance (Raven et al. 1998). Therefore, I argue that this factor can be described as

relationship oriented based on social norms, and I expect it to positively affect compliance to change

via normative commitment. Since normative commitment refers to obligations, power tactics with

the purpose of increasing social duties would increase employees’ normative commitment.

Dependence power and referent power have been stably clustered in the soft spectrum, so I expect

this aggregated factor to be soft again (Pierro et al., 2008, Raven et al., 1998).

Equity power and reciprocity power have been grouped together in earlier research (Raven et al,

1998), since both refer to concepts such as compensation, fairness and reciprocity. Regan (1971)

(23)

23

insignificant. Instead, the small favor produced an obligation to comply to a request based on

reciprocity. Obedience to authority seems to be another well-established norm, based on Milgram’s

(1963) famous study. I argue that reciprocity, equity and authority reflect a factor oriented at

maintaining relationships based on economical norms. Based on prior research, I expect the power

bases of this factor to be defined as harsh (Pierro et al., 2008, Raven et al. 1998). Furthermore, I

argue that the power bases correlate to normative commitment, since a request based on reciprocity,

equity or authority would create an obligation or duty to comply.

Lastly, I expect reward power impersonal and personal and coercion power impersonal and personal

to load together via a factor that I call the carrot and stick model. This factor might be the most easily

interpretable, as rewards and coercion have been widely used by managers and experienced by

employees. Although personal reward can load as soft, in Western cultures it seems to load mainly as

harsh (Raven et al. 1998, Pierro et al., 2008). Therefore, I expect all these factors to load on the harsh

spectrum of power. This factor is expected to affect continuance commitment, since rewards and

coercive tactics explicitly refer to the costs or benefits of noncompliance (e.g. promotion for

compliance, being fired for noncompliance). The hypotheses are formulated as follows:

Hypothesis 1: the individual power bases can be loaded on four distinct factors, being rationality (soft1, SP1), social norm oriented relationship (soft2, SP2), economic norm oriented relationship (harsh1, HP1) and carrot and stick model (harsh2, HP2).

Hypothesis 2: power tactics related to rationality positively influence compliance to change, mediated by affective commitment to change.

Hypothesis 3: power tactics related to relationships based on social norms positively influence compliance to change, mediated by normative commitment.

Hypothesis 4: power tactics related to relationships based on economic norms positively influence compliance to change, mediated by normative commitment.

(24)

24

Hypothesis 5: power tactics related to the carrot and stick model positively influence compliance to change, mediated by continuance commitment.

In summary, the conceptual model is shown in image 7 below.

Image 7: conceptual model of power bases, commitment and compliance

This concludes the hypotheses section. I argued that the eleven power bases can be captured within

four factors (soft1, soft2, harsh1, harsh2), each with their own effect on affective, normative or

continuance commitment only. If these hypotheses are accepted, managers could gain insight in how

to use their power to gain the highest rates of compliance from employees. Ultimately, this would

benefit the bottom line, since the implementation success of innovation improves. In order to test

the hypotheses, data has been collected. The next section describes the collection method and

survey items, sample characteristics, and the preliminary steps required to prepare the data for

(25)

25

Methods

Data collection

The data was collected by means of a survey, which incorporated several control variables,

independent variables, mediator variables and the dependent variable. 105 responses have been

collected, of which nine have been deleted due to missing values: eight responses did not have any

values for power, commitment and compliance and one response did not have any values for

commitment and compliance. Table 6 below gives an overview of the scales used for the control

variables and table 7 presents the scales used for the independent, mediator and dependent variable.

A standard 7 point Likert scale was used, with the addition of an eight option ‘undecided’, which was

treated as a missing value. I decided that one response was unfit for analysis, since 77% of its

answers were undecided. ‘Organizational change’ was consciously chosen over ‘innovation

implementation’ since I expected the former to be more intuitive for non-business students and the

latter to be associated with new, high-tech products. In the Appendix, the items are presented as

used in the survey.

Control Item Description Open

Age Continuum in years

Gender Binary; male or female Nationality Open

Tenure Continuum in years or months

Employment status Binary; part time or full time (cutoff point 30 hours per week)

Position Open

Department size Continuum in full time employees

Change significance 8 point Likert scale (Herscovitch & Meyer, 2002)

Change impact 3 item, 8 point Likert scale (Herscovitch & Meyer, 2002)

Table 6: control variables

Independent Item Mediator Item Dependent Item Reward impersonal

3 item, 8 point Likert scale Raven et al.

(1998) Affective 6 item, 8 point Likert scale Herscovitch & Meyer (2002)

Compliance 101 point continuum

Coercive impersonal Continuance

Expert Normative

Referent Information Position Reciprocity

(26)

26 Dependent

Equity

Personal reward

Table 7: independent, mediator and dependent variables

Compliance was measured on a 101 point continuum, with anchor points active resistance (0-20),

passive resistance (21-40), compliance (41-60), cooperation (61-80), and championing (81-100).

Herscovitch & Meyer (2002) measured cooperation, championing and compliance with an eight item,

six item and three item scale besides the continuum. Due to the compliance scale having low

reliability (alpha of .49) and the current length of the survey, I have not included these items.

Frequencies were run using SPSS to check for missing values and for all items, missing values were

<10%. EP1,2 and 3 had the most missing values for an aggregated construct, respectively 7,4 and 4.

PP1 had the most missing values for a single item (8 missing).

Primarily, I attempted to collect data within a single organization via personal contact with one of its

employees. This proved to be unsuccessful, and I changed the data collection process to include

multiple organizations. Unfortunately, this proved unsuccessful as well, after which I contacted

anyone who had recently experienced an organizational change or is experiencing one. Distribution

of the survey was done via Facebook, which resulted in very low response rates as estimated in table

8. One organization was willing to share the survey on its Intranet, being Drostenburg: a school for

children with mental and/ or physical disabilities.

Source Members – theoretical reach Estimated response rate

University of Amsterdam – MSc Business Administration

500 <2%

Nederlandse studenten alpen club, Amsterdamse studenten alpen club (Dutch student mountaineering club)

1400 combined <2%

Dutch Climbing 750 <1%

Free University – BSc International Business Administration

440 <1%

Korea University Buddy Assistants (KUBA) 7100 <.01% ‘Respondenten gezocht’ (respondents wanted) 2000 1%

Drostenburg 70 <1%

(27)

27

This strategy proved ineffective. Eventually, I contacted over 150 people by sending individual emails,

Facebook messages and replying to other survey’s posted, drawing on reciprocity power (earlier

favors done by me, or answering surveys), dependence power (amplifying the need of others to

answer the survey) and referent power (contacting friends).

Sample characteristics

In the upcoming paragraphs I introduce the sample and discuss its characteristics. Table 9 below

shows an overview of the demographics. As shown, the sample mainly reflects Dutch students and

young professionals: 79% is between 19 and 30, 87% is Dutch, 75% has 3 years or less working

experience and 58% is employed part time. The sample was reasonably balanced concerning gender:

55% of respondents identified as male and 45% as female.

Age Nationality Tenure Employment status 19 - 30 79% The Netherlands 87% 0-1 year 35% Part time 58% 30 - 40 12% Germany 2% 1-2 years 22% Full time 42% 40 - 62 10% United States of America 2% 2-3 years 18%

Other 9% 3-4 years 8%

4-5 years 2% 5-6 years 3% >6 years 11%

Table 9: demographics of sample

These findings reflect my experience of encountering follow-up emails to the survey. Many

respondents were willing to answer but had difficulty in doing so, since they had not experienced

many organizational changes and asked my advice. By giving advice, providing examples and giving

interpretations of organization and change, I could have subconsciously influenced the research. At

times, I made recommendations based on personal knowledge.

Data analysis: preliminary steps

Based on the description, I coded three control variables, from the research of Damanpour (1991):

administrative/ technical, profit/nonprofit and product/ process. Most description were adequate to

make a distinction between administrative (non-core work activities) and technical (core activities).

(28)

28

been treated as non-core work activities. The profit/nonprofit distinction proved more difficult:

hospitals and climbing (instructor or trainer) related activities were treated as undecided if the

description was insufficient to make a decision. This is due to some climbing instructors or trainers

simultaneously working for profit (e.g. in a gym) and not for profit (e.g. for a student association).

Words such as ‘association, club, member and committee’ were associated with nonprofit, whereas

expressions such as ‘management, intern, department, company, organization and profit’ were

classified as profit. Professional job descriptions (e.g. customer replenishment specialist, financial

controller) were also associated with for-profit organizations. The product/process distinction was

made based on the description, although in some cases this was inadequate to make an informed

decision and was treated as missing. Finally, no missing values were recorded for administrative –

technical, 12.6% (12) missing values for profit-nonprofit, and 14.7% (14) missing values for product –

process. The distribution is summarized in the table below. Most of the changes were classified as

administrative, for profit and process oriented.

Administrative Technical Profit Nonprofit Product Process

61% 39% 64% 23% 17% 69%

While processing the data, I discovered that the Likert scale of power and commitment to change

was reversed: strongly agree was 1 and strongly disagree was 7. Therefore, all indicative items were

recoded and the counter indicative items kept.

Three limitations require mentioning in this section. Firstly, coercive personal has unconsciously not

been included in the survey and therefore, no scores are available for this variable. Strangely, I

discovered this exclusion only when it was too late and the data collection had already started.

Therefore, in all subsequent analyses, ten power bases are utilized. Secondly, gender has not been

used as a control variable, since SPSS kept encountering an ‘unexpected value’ and stopped any

(29)

29

Recoding was done manually to 0 and 1, respectively, yet still no analyses could be conducted.

Examination of the scores did not help to detect any flaw.

This concludes the basic preliminary steps. The scores have been checked for missing data, items

have been recoded and counter indicative (in my case indicative) items are reverse scored. The

sample reflects mostly Dutch students and young professionals. Next, I assess the reliability and

validity of the items. Afterwards, scale means are computed, which are checked for normality.

Reliability, validity, scale means, normality

Reliability of power, commitment to change, and change impact was computed via Cronbach’s alpha

and found highly satisfactory. Cronbach’s alpha for affective, normative and continuance

were .96,.795 and .883, respectively. .96 for affective commitment could be expected, as Herscovitch

& Meyer (2002) recorded an alpha of .94. The power bases also had satisfactory alpha’s: reward

impersonal (.88), coercive impersonal (.88), expert (.914), referent (.829), information (.921), position

(.827), reciprocity (.790), dependence (.844), personal reward (.887) and equity (.941). Although

some scales could be improved significantly (>.10) by deletion of items, no items were deleted since

each power base is measured with the minimum of three items. A control variable change impact

consisted of three items and had an alpha of .826.

Exploratory factor analysis is conducted on the items of commitment, with three expected factors.

This corresponds to the found Eigenvalues and interpretation of the scree plot: the three factors

explain 70.21% of variance. KMO’s test is satisfactory with a value of .847 and Bartlett’s test is

significant at p<.05 (.000).

The pattern matrix shows that the items measuring affective commitment and continuance

commitment both load solely on one factor, without significant loadings on other factors. However,

normative commitment does not load as expected: NC1 has the highest loading on factor 2,

associated with continuance commitment, NC2 has a high secondary loading on factor 2 (.310 with

(30)

30

and NC6 shows highest loading on factor 2 with a secondary high loading on factor 1, associated with

affective commitment (.684 and .412 respectively). The items of normative commitment are shown

below. Pattern Matrixa Component 1 2 3 AC1 ,948 -,049 -,004 AC2 ,914 ,012 ,015 rAC3 ,939 -,034 ,017 AC4 ,791 -,053 ,057 rAC5 ,936 -,114 ,002 rAC6 ,911 -,057 ,078 CC1 -,215 ,723 ,004 CC2 -,113 ,803 -,127 CC3 -,100 ,794 ,120 CC4 -,068 ,778 ,152 CC5 -,097 ,780 ,037 CC6 -,088 ,772 ,025 NC1 ,330 ,708 -,002 NC2 ,293 ,310 ,441 rNC3 ,163 ,147 ,727 NC4 ,151 ,410 ,461 NC5 -,092 -,162 ,937 rNC6 ,412 ,684 -,113

Extraction Method: Principal Component Analysis. Rotation Method: Oblimin with Kaiser Normalization. a. Rotation converged in 5 iterations.

NC1 I feel a sense of duty to work toward this change

NC2 I do not think it would be right of me to oppose this change rNC3 I would not feel badly about opposing this change

NC4 It would be irresponsible of me to resist this change NC5 I would feel guilty about opposing this change rNC6 I do not feel any obligation to support this change

Combined with the factor analysis, one interpretation of the third factor is that it reflects feelings of

(31)

31

second strongest score is NC3, which asks if respondents would feel bad about opposing the change.

NC2 and NC4 score moderately high on the third factor, whereas NC2 asks if it would be right to

oppose the change, NC4 asks if it would be irresponsible, which could be a question of morality.

Although morality could reasonably be expected to correlate with obligation, or sense of duty, NC6

and NC1 load negatively on the third factor.

Another interpretation is that NC1 and NC6 ask about supporting, or working for a change, contrary

to NC2 – NC5 which ask about resisting or opposing the change. This could explain why NC1 and NC6

have negative scores on factor 3. The items of continuance commitment seem to support this

interpretation: CC1 and CC2 have low, or negative, respective scores on factor 3, contrary to CC3-6

which score positive, higher scores. CC1 and CC2 indicate going along with the change, and CC3-6

indicate resistance to change. AC does not have items that clearly indicate resistance or going along

with the change, and most items score close to zero on the third factor. The items of CC1-6 are

shown below as well.

CC1 I have no choice but to go along with this change CC2 I feel pressure to go along with this change CC3 I have too much at stake to resist this change CC4 It would be too costly for me to resist this change CC5 It would be risky to speak out against this change CC6 Resisting this change is not a viable option for me

My decision is to use the items for normative commitment as done in the work of Herscovitch &

Meyer (2002), meaning that NC1-6 are used to calculate the mean scale of normative commitment.

Firstly, in their research NC1 and NC6 were the highest contributors of normative commitment,

which makes sense intuitively since normative commitment refers to an obligation (Herscovitch &

Meyer, 2002). NC1 and NC6 explicitly mention a sense of duty, or obligation, so it seems unfit to

exclude these items from the measurement. Secondly, the interpretation that factor 3 describes

resistance to change seems convincing. I acknowledge that this decision would potentially be a

(32)

32

Factor analysis was also conducted on the power bases, which yielded satisfactory results: KMO’s

test value was .756 and Bartlett’s test is significant at p<.05 (.000). 7 factors were found, which

corresponds to earlier results by Raven et al. (1998). The 7 factors explained in total 78.569% of

variance. Items relating to a power base loaded on the same factor, with no significantly high

loadings on other factors. The distribution of aggregated power bases did differ to some degree

compared with earlier research. Specifically, (1) information power and position power loaded

together, (2) reciprocity power did not share the same factor as equity power, (3) referent power had

the same factor as personal reward power. Although interesting to elaborate on, the purpose of this

analysis is to establish the validity of the measurements, which is satisfactory.

Another factor analysis was done with 4 expected factors as hypothesized: rationality (soft1),

relationship oriented based on social norms (soft2), relationship oriented based on economic norms

(harsh1) and stick and carrot model (harsh2). KMO’s test was satisfactory at .756 and Bartlett’s test

was significant at p<.05 (.000). 4 factors were found, which explained in total 62.76% of variance.

Three more factors with Eigenvalues higher than 1 were found, which fits the earlier factor analysis.

This pattern matrix is included in the Appendix for reference.

The loadings are not completely as expected. Factor 1 had high loadings of referent power, position

power, reciprocity power, equity power and personal reward power, factor 2 had high loadings of

expert power and information power, factor 3 had high loadings of reward power impersonal and

coercion power impersonal and factor 4 had high loadings of dependence power. Most items loaded

well within one factor, but several items had strong secondary loadings as well. These include;

 Referent power (1,2,3) having high secondary loadings on factor 2. RFP2 even has the highest loading on factor 2.

 IP (1,2,3) having high secondary loadings on factor 4.

(33)

33

Yet, I continued the analysis and calculated the mean scores of the variables as presented. The three

forms of commitment were calculated by taking the mean of the six items representing each form

(e.g. ACtot = mean AC1-6). Soft1 consists of the mean of XP1-3 and IP1-3, soft2 of the mean of DP1-3,

harsh1 incorporates the mean of RFP1-3, PP1-3, RP1-3, EP1-3 and PR1-3, and harsh2 summarizes the

mean of RIP1-3 and CIP1-3. Change impact was a control variable, aggregated from Ci1-3.

Descriptive statistics of all major variables of interest are presented in table 10 below. Most variables

seem reasonably normally distributed: Skewness and Kurtosis values are between -1 and 1, except

for affective commitment. I included the range of z-scores as a test for outliers: values below -3 or

above 3 were treated as such. Although no such values are found, compliance to change did have an

extreme minimum value, with a z-score of -2.72. Normative commitment and harsh1 similarly had

extreme minimum values with z-scores of -2.62 and -2.55, respectively. The Kolmogorov-Smirnov

rejected the null hypothesis of normal distribution for Cc, Ci, AC and SP2 and Shapiro-Wilk rejected

the null hypothesis of normal distribution for Cc, Ci, AC, CC, SP2 and HP2. The results of these tests

are shown in table 11 below.

Descriptive Statistics

N Mean

Std.

Deviation Skewness Kurtosis

Range z-score

Statistic Statistic Statistic Statistic Std. Error Statistic Std. Error Minimum Maximum Citot 95 3,8544 1,39151 ,248 ,247 -,792 ,490 -1.81 2.26 ACtot 95 4,2851 1,80083 -,256 ,247 -1,111 ,490 -1.82 1.51 CCtot 95 4,6088 1,53877 -,265 ,247 -,512 ,490 -2.35 1.55 NCtot 94 4,4163 1,30536 -,451 ,249 -,101 ,493 -2.62 1.98 SP1 95 4,1502 1,43308 -,148 ,247 -,837 ,490 -2.08 1.99 SP2 95 4,8386 1,51421 -,543 ,247 -,393 ,490 -2.31 1.43 HP1 95 3,9600 1,13678 -,269 ,247 ,145 ,490 -2.55 2.21 HP2 94 4,6321 1,55045 -,624 ,249 -,200 ,493 -2.34 1.53 Cc 91 59,20 19,941 -,470 ,253 -,585 ,500 -2.72 1.54 Valid N (listwise) 90

(34)

34

Tests of Normality

Kolmogorov-Smirnova Shapiro-Wilk

Statistic df Sig. Statistic df Sig.

Cc ,110 90 ,009 ,960 90 ,007 Citot ,119 90 ,003 ,964 90 ,015 ACtot ,105 90 ,015 ,940 90 ,000 CCtot ,065 90 ,200* ,966 90 ,019 NCtot ,076 90 ,200* ,973 90 ,055 SP1 ,088 90 ,083 ,973 90 ,057 SP2 ,129 90 ,001 ,945 90 ,001 HP1 ,074 90 ,200* ,984 90 ,337 HP2 ,086 90 ,094 ,944 90 ,001

Table 11: normality test. *This is a lower bound of the true significance.

Histograms of the variables were examined to check the assumption of normal distribution.

Combined with the rule of thumb regarding Skewness and Kurtosis, all variables except

affective commitment pass the test. For affective commitment, I have included the histogram,

which shows that the values are not normally distributed around the mean.

(35)

35

With the normality check, the preliminary steps of the data analysis is finished. The data does not

fully satisfy findings of earlier research, and affective commitment does not fully satisfy the

assumption of normality. Otherwise, the data seems fit for analysis. Affective commitment and

normative commitment is included, so the results should be interpreted with care. Next, the

correlation matrix is provided and the results of the regression analysis are presented. Each

hypothesis is evaluated and the tested conceptual model presented at the end of the section.

Results

Table 12 below shows the correlation matrix with independent, mediator and the dependent

variable.

Variable M

SD

Cc

AC

CC

NC

SP1

SP2

HP1

HP2

Cc

59.20 19.94 .

AC

4.29

1.80 .683** (.96)

CC

4.61

1.54 -.001

-.087

(.883)

NC

4.42

1.31 .523 .482 .567** (.795)

SP1

4.15

1.43 .354 .570 .016

.359** (.855)

SP2

4.84

1.51 .274 .360 .223*

.374** .167

(.844)

HP1

3.96

1.14 .265*

.102

.304 .403 .116

.161

(.902)

HP2

4.63

1.55 .169

.085

.529 .424 .021

.093

.429** (.901)

Table 12: correlation matrix of dependent, independent and mediator variables

The table shows that there are significant relations among the variables, which means that it makes

sense to further analyze the data. To test hypotheses 2,3,4 and 5, regression analyses are conducted.

Hypothesis 2: power tactics related to rationality positively influence compliance to change,

mediated by affective commitment.

The process tool within SPSS was used to test this hypothesis, via model 4. Soft1 was entered as

independent variable, compliance to change as dependent variable, and affective, normative and

continuance commitment as proposed mediators. Control variables included age, tenure,

(36)

36

type (product-process), change significance, change impact, soft2, harsh1, and harsh2. To test the

hypothesis, four steps are performed: the model is summarized for the effect of the independent

variable and control variable on each mediator alone, after which the full model is tested for the

dependent variable. In table 13 these results are shown. In tables 14,15,16 and 17, the direct and

indirect effects of SP1, SP2, HP1 and HP2 are given. Now, the hypotheses are evaluated based on the

results.

From table 13, it can be seen that affective commitment is significantly affected by (1) rationality, (2)

change impact, (3) relationship oriented based on social norms and (4) innovation type –

administrative versus technical.

Consequent Antecedent Affective commitment

R2=.7559, p<.01. Continuance commitment R2=.5290, p<.01. Normative commitment R2=.5222, p<.01. Compliance to change R2=.6762, p<.01.

Effect SE p Effect SE p Effect SE p Effect SE p

Constant .7538 1.1446 .5132 2.1128 1.3836 .1330 1.2890 1.1416 .2642 26.8048 2.3253 .1005 Soft power1 (X) .3099 .0966 .0023** .1438 .1168 .2241 .3534 .0964 .0006** -.9344 1.5293 .5441 AC - - - 5.8318 2.0651 .0069** CC - - - -4.5652 1.8288 .0161* NC - - - 4.5904 2.3253 .0543 Age -.0240 .0197 .2302 .0520 .0239 .0340* .0178 .0197 .3700 .1879 .2833 .5104 Te .0074 .0043 .0927 -.0053 .0052 .3215 -.0012 .0043 .7895 -.0655 .0611 .2890 Es -.0815 .3576 .8206 .0498 .4322 .9088 .0187 .3566 .9584 14.2265 4.8465 .0051** Si .0021 .0038 .5869 .0078 .0046 .0973 .0007 .0038 .8481 .0720 .0537 .1863 Cs .1423 .1065 .1874 .0822 .1287 .5259 .0126 .1062 .9064 .2305 1.4831 .8772 Citot .6538 .1084 .0000** -.4180 .1311 .0025** .0229 .1081 .8329 .3488 2.0223 .8638 SP2 .3714 .0938 .0002** .1796 .1134 .1194 .2184 .0935 .0236* .8370 1.4923 .5775 HP1 -.2275 .1455 .1242 .3430 .1758 .0567 .2416 .1451 .1021 4.9785 2.1271 .0236* AdTe -.9235 .4175 .0316* -.6240 .5047 .2221 -.3215 .4164 .4437 -1.0004 6.0464 .8693 PrNPr .1603 .3497 .6486 -.4909 .4227 .2511 -.7397 .3488 .0389* -13.3962 5.0063 .0102* PdPc .0259 .4062 .9494 -1.2017 .4910 .0179* -.9531 .4051 .0226* -7.4370 5.9337 .2163 HP2 .0750 .1137 .5123 .15240 .1374 .2675 .0393 .1133 .7305 -3.4502 1.5701 .0330*

Social power and innovation implementation : motivated compliance : a mediated model of power and implementation