Download as pdf or txt
Download as pdf or txt
You are on page 1of 61



0%$7KHVLV

“The impact of ethical and


participative leadership on
innovative work behaviour
in tech firms:

The role of self-efficacy and creative process engagement

Anastasopoulou Kyriaki & Angelis Dimitrios

Supervisor
Philippe Rouchy

Karlskrona, Sweden
September 2020

ZZZEWKVHPED
 '(3$570(172),1'8675,$/(&2120,&6



This thesis is submitted to the Department of Industrial Economics at Blekinge Institute of Technology
in partial fulfilment of the requirements for the Degree of Master of Science in Industrial Economics
and Management. The thesis is awarded 15 ECTS credits.

The author(s) declare(s) that they have completed the thesis work independently. All external sources
are cited and listed under the References section. The thesis work has not been submitted in the same or
similar form to any other institution(s) as part of another examination or degree.

Author information:
Kyriaki Anastasopoulou
kirkianast@gmail.com

Dimitrios Angelis
dimitriosangelis@yahoo.com

Department of Industrial Economics


Blekinge Institute of Technology
SE-371 79 Karlskrona, Sweden

Website: www.bth.se
Telephone: +46 455 38 50 00
Fax: +46 455 38 50 57

 


$FNQRZOHGJHPHQWV

We would like to thank and express our sincere gratitude to our supervisor, Professor Philippe Rouchy
for his guidance and support throughout the time of conducting this research.
Further, we would also like to express our appreciation to Mr. Anders Wrenne for his coordination and
guidance for this Master Thesis course.
The same appreciation is shown to all our professors and tutors from BTH University who imparted
their knowledge to us.
We also thank our families, friends and colleagues for their support. Finally, special thanks must go to
all the professionals who participated in our survey whom without this study would not have been
possible.


 


$EVWUDFW

Background: Nowadays, companies from the technological sector confront extreme competition and it
is always a challenge for leadership teams to increase competitiveness. This study aims to investigate
innovation advancement within tech companies in an international context from the leadership and
management incentives point of view. Leadership plays a vital role in giving direction to the path an
organization should follow. It is of significant interest to examine how leadership can drive an
organization to innovative thinking. Different approaches and leadership styles can be adopted and
practiced by leaders to produce different outcomes on employees’ creative culture. Additionally,
individual characteristics of the employees such as self-efficacy and creativity may allow the innovative
behaviours to strive and create a workplace culture that is inducing innovative output. Innovative work
behaviour is becoming more popular or even mandatory within several firms in the technological sector
in contrast to previous decades.

Purpose: The purpose of this study is to explore how ethical leadership and participative leadership
style can affect innovative work behaviour and to examine if the creative process engagement and self-
efficacy lead to enhanced innovative work behaviour.

Methodology: For this thesis a quantitative approach for data collection, as well as data analysis is used.
This study is based on a SEM model which contains close-ended questions that were answered through
a self-administrated questionnaire. The survey was answered by employees working at companies from
the field of technology, covering different positions. A total of 177 respondents answered the
questionnaire, and the results were analysed both in quantitative and qualitative ways. IBM SPSS
software was used for the statistical analysis, and AMOS 26 for the Structural Equation Modeling tests.

Results: The results of the statistical analysis performed unveiled that the aspects of ethical and
participative leadership can positively affect creativity and innovation and that self-efficacy can
positively relate to creative process engagement.

Conclusion: This study contributes in showing that two positive ways of management, ethical and
participative can be introduced by leaders that are interested in increasing creativeness and innovation
at work; it also shows that for the sample tested, ethical and participative leadership does not necessarily
has a major effect on employee’s self-efficacy.

Delimitations: The geographical locations, the time and sample size, the choice of participating
organizations, and the framework designed for the evaluation of the theoretical problem are considered
as limitations for this study. This research is mainly limited to professionals working in the technological
sector and the study is restricted in time since the participants had to answer in a certain time frame. To
conclude, the sample size of the survey even though is satisfactory for its intended use, could be higher.

Keywords: Innovative Work Behaviour, Self-Efficacy, Creative Process Engagement, Creativity,


Ethical Leadership, Participative Leadership, Leadership Styles.

 


Table of contents

 ,QWURGXFWLRQ
 3UREOHPGLVFXVVLRQ
 3UREOHPIRUPXODWLRQDQGSXUSRVH
 'HOLPLWDWLRQV
 7KHVLV6WUXFWXUH

 7KHRU\
 /LWHUDWXUH5HYLHZ
 ,QQRYDWLRQDW:RUN
 ,QIOXHQFHRI/HDGHUVKLS0HWKRGV
 ,PSRUWDQFHRI(PSOR\HH&UHDWLYLW\
 7KHRUHWLFDO)UDPHZRUN
 (WKLFDO/HDGHUVKLS
 3DUWLFLSDWLYH/HDGHUVKLS
 &UHDWLYH3URFHVV(QJDJHPHQW
 6HOIHIILFDF\
 ,QQRYDWLYH:RUN%HKDYLRXU
 6XPPDU\

 5HVHDUFK0HWKRG
 2YHUYLHZ
 7KH6L[6WDJHVRI6(0
 6WDJH'HILQLQJ,QGLYLGXDO&RQVWUXFWV
 6WDJH'HYHORSLQJWKH2YHUDOO0HDVXUHPHQW0RGHO
 6WDJH'HVLJQLQJD6WXG\WR3URGXFH(PSLULFDO5HVXOWV
 6WDJH$VVHVVLQJ0HDVXUHPHQW0RGHO9DOLGLW\
 )LQDO6(06WDJHV 
 'DWD&ROOHFWLRQ
 6XUYH\V
 'HVFULSWLYH,QIRUPDWLRQRQWKH3RSXODWLRQ6XUYH\HG
 4XHVWLRQQDLUH
 0LVVLQJ'DWD

 6XUYH\'HVFULSWLYH5HVXOWV

Y


 2YHUYLHZ
 5HVSRQGHQW'HPRJUDSKLFV
 *HQGHU
 $JH
 /RFDWLRQRI(PSOR\PHQW
 (GXFDWLRQDO/HYHO
 3RVLWLRQ

 (PSLULFDO5HVXOWV6WDWLVWLFDO$QDO\VLV
 7KHRUHWLFDO6(00RGHO&RQVWUXFWLRQ
 6(07KHRUHWLFDO0RGHO$QDO\VLV
 $VVHVVPHQWRI1RUPDOLW\IRUWKH6(07KHRUHWLFDO0RGHO
 &RQILUPDWRU\)DFWRU$QDO\VLVRIWKH7KHRUHWLFDO6(00RGHO
 0RGHO)LW&KHFNV
 ([SORUDWRU\)DFWRU$QDO\VLV
 5HOLDELOLW\
 )DFWRU5HGXFWLRQXVLQJ5RWDWHG&RPSRQHQW0DWUL[
 7RWDO9DULDQFHRI&RPSOHWH'DWDVHW
 6XLWDEOH0RGHO6HOHFWLRQ8VLQJ)DFWRU5HGXFWLRQ
 7RWDO9DULDQFHRI6XLWDEOH0RGHO'DWDVHW
 &)$0RGHO$QDO\VLVDQG&RUUHFWLRQV
 (VWLPDWHV5HYLHZ
 )LQDO&)$0RGHO
 0RGHO)LW&KHFNV
 &RQVWUXFW9DOLGLW\DQG5HOLDELOLW\

 $QDO\VLVDQG'LVFXVVLRQ
 )LQDO6(00RGHO7HVWLQJDQG9DOLGLW\&KHFNV
 &RPSDULVRQRI7KHRUHWLFDO6(00RGHOZLWK)LQDO6(00RGHO
 7KHRUHWLFDO$QDO\VLV
 (WKLFDO/HDGHUVKLS
 3DUWLFLSDWLYH/HDGHUVKLS
 6HOI(IILFDF\
 &UHDWLYH3URFHVV(QJDJHPHQW
 ,QQRYDWLYH:RUN%HKDYLRXU

 &RQFOXVLRQ

YL


 &RQFOXVLRQV
 ,PSOLFDWLRQV
 /LPLWDWLRQV
 )XUWKHU:RUN

 %LEOLRJUDSK\

 $SSHQGLFHV
$SSHQGL[$

YLL


List of Tables
7DEOH&RQVWUXFWV'HVFULSWLRQ
7DEOH$VVHVVPHQWRIQRUPDOLW\7KHRUHWLFDO6(00RGHO
7DEOH9DOXHVWRFKHFN0RGHO)LW
7DEOH5HOLDELOLW\RI'DWDLQ6366
7DEOH6WDWLVWLFDO6LJQLILFDQFH
7DEOH)DFWRU5HGXFWLRQ5RWDWHG&RPSRQHQWLQ6366
7DEOH7RWDO9DULDQFH5RWDWLRQ6XPV
7DEOH)DFWRU5HGXFWLRQ²6XLWDEOH0RGHO
7DEOH7RWDO9DULDQFH5RWDWLRQ6XPV
7DEOH&)$0RGHO9DOXHV
7DEOH(VWLPDWHVDIWHUYDULDEOHVGHOHWHG
7DEOH)LQDO&)$0RGHO&KHFNV
7DEOH)LQDO&)$0RGHO6WDQGDUGL]HG(VWLPDWHV
7DEOH&RQVWUXFW9DOLGLW\$9(DQG&5
7DEOH&URQEDFK V$OSKDUHOLDELOLW\RIILQDO&)$
7DEOH&RQVWUXFWV&RUUHODWLRQV
7DEOH&RUUHODWLRQV9DULDQFH069DQG$69
7DEOH6XPPDU\RI5HVXOWVIURPWKH6(00RGHO
7DEOH([DPLQLQJ+\SRWKHVHVDVSHU&59DOXH&ULWHULRQ
7DEOH6WDWXVRIWKH5HODWLRQVKLSV+\SRWKHVHVLQWKH)LQDO6(00RGHO

YLLL


List of Figures
)LJXUH5HVHDUFK)UDPHZRUN
)LJXUH3DWK'LDJUDP
)LJXUH*HQGHU
)LJXUH$JH
)LJXUH/RFDWLRQRI(PSOR\PHQW
)LJXUH(GXFDWLRQDO/HYHO
)LJXUH3RVLWLRQ
)LJXUH7KHRUHWLFDO6(00RGHOEXLOGLQ,%0$PRV
)LJXUH7KHRUHWLFDO6(0PRGHOILUVWUXQ
)LJXUH&)$RI6(0PRGHO
)LJXUH&)$0RGHODIWHU)DFWRU5HGXFWLRQ
)LJXUH)LQDO&)$0RGHO
)LJXUH)LQDO6(00RGHOZLWK&RQVWUXFWV

L[


List of abbreviations

AGFI Adjusted Goodness-of-fit

AVE Average Variance Extracted

ASV Average Shared Variance

CE Creative Process Engagement

CFA Confirmatory Factor Analysis

CFI Comparative Fit Index

CMIN Chi-square Value

CMIN/DF Relative Chi-square

CR Critical Ratio

EFA Exploratory Factor Analysis

EL Ethical Leadership

GFI Goodness of Fit Index

GOF Goodness of-fit

IBM International Business Machines

IW Innovative Work Behaviour

MSV Maximum Shared Variance

NFI Normed Fit Index

PhD Doctor of Philosophy

PL Participative Leadership

RMSEA Root Mean Square Error of Approximation

R&D Research and Development

UK United Kingdom

SEM Structural Equation Modeling

SE Self-Efficacy

SPSS Statistical Package for the Social Sciences

Ȝ Lambda or Factor Loading

[


1. Introduction

Globalization, technological developments, and harsh market environments are some of the several
reasons that have driven businesses to give a great emphasis on innovation (Akram, et al., 2016). Due
to swift technological advancements, service sector organizations in particular are examining innovation
in order to survive the rivalry (Archibugi, et al., 2013).

The leadership of a firm must follow an optimistic method to accomplish innovation, as innovation
cannot be achieved in a negative environment (Donate, et al., 2015). Participative leadership style can
be categorized as a positive leadership style in which personnel are given the prospect of engagement
in decision making and problem-solving through inspiration, support and encouragement (Somech &
Wenderow, 2006). Participative leaders provide teams with accountability by involving them in decision
making activities (Sauer, 2011). A participative leader may improve employees’ efficiency,
organizational citizenship behaviour and many other positive behaviours (Miao, et al., 2014).
Participative leadership embraces the teamwork approach (DuBrin, 2013). Research has shown that low
performing teams are often governed by the team leader, although well performing teams are identified
to embrace shared leadership (DuBrin, 2013). Participative leadership provides the experience of team
spirit, and offers the warmth of a friendly professional environment.

Considering the developing devotion applied to corporate social responsibilities and business ethics,
leaders nowadays are needed to work ethically. Therefore, ethical leadership has been at the attention
of both academia and industry in the past years (Kalshoven, et al., 2011). In the process of making,
endorsing and realizing new concepts, methods or processes, many risks, complications, clashes, and
even ethical dilemmas may be encountered; this shows that ethical leadership that highlighted principles,
social responsibility, independence, and people direction (Brown & Trevin˜o, 2006) can be a possible
interpretation of innovative work behaviour. Ethics examine moral responsibilities or differentiate the
right from wrong (DuBrin, 2013). According to DuBrin, “An ethical leader is honest and trustworthy,
therefore has integrity” (DuBrin, 2013). It is vital to consider integrity as an important factor which
needs to be adopted by all professional cultures; it is believed that integrity can give positive results
with regards to employee performance. Integrity signifies adherence to logical morals; it contemplates
engaging in the right beliefs irrespective of psychological or public constraint (DuBrin, 2013).

By conducting this study, it is intended to examine how participative and ethical leadership impacts
employee’s innovative work behaviour and the primary mechanisms.

Problem discussion
The theoretical problem of the research is to investigate the impact of ethical and participative leadership
on employees’ self-efficacy and creative process engagement, and subsequently on their innovative
work behaviour.

Studies have been performed to demonstrate relationships between ethical leadership or participative
leadership style and their impact on innovative work behaviour. Since leadership is made of knowledge
of personal, decision making and other aspects of labour activities that are difficult to measure, this
study investigates those relations in order to figure out what is the influence on innovative work
behaviour of both the moral aspects of ethical leadership and participative leadership style.

Problem formulation and purpose


This thesis examines how different types of leadership (ethical and participative) affect the employee’s
behaviour towards innovative work. In order to investigate those aspects, a SEM (Structural Equation
Modeling) model with five constructs is used. This allows to study the relationships between elements
of behaviour that other studies cannot perform. Firstly, it analyses two main inputs (i) ethical leadership
and (ii) participative leadership style, that reveal to be in significant relationship with secondly - two
important creative behaviour mediators which are (iii) self-efficacy and (iv) creative process




engagement, and finally, it examines if these inputs and mediators can positively influence the one
output which is the (v) innovative work behaviour.

Delimitations
To write this thesis, a sample of 177 professionals in the technological sector was collected. The study
was limited with samples from technological sectors, mainly Oil & Gas and Shipping. Some respondents
were from construction and advertising industry firms, but still within the technological sector. It was
intended to examine professionals’ answers from mainly Greece and the UK, but professionals from
some other countries have participated. The survey embraces selected employees of different level of
seniority and profession, from management to engineering, finance, administration, etc.

In addition, the sample size has been gathered during a strict one-week time period. The creation of the
questionnaire and its distribution to the population took 2 months. The participants had a limited time
to answer the questionnaires due to a strict schedule and therefore, the collection period was limited to
one week. It has resulted in some intensive activities to contact and insure following-up of respondents.
This means that more time would have helped this research in gathering a larger sampling. The time
spent to gather all the data was considered as the most possible and effective use of time in order to
deliver this study.

Thesis Structure
The thesis uses several chapters, including: (1) Introduction; (2) Theory; (3) Research Methodology; (4)
Survey Descriptive Results; (5) Empirical Findings / Statistical Analysis; (6) Analysis and Discussion;
and (7) Conclusion. The theory chapter consists of two main parts, the literature review and the
theoretical framework. The literature review presents the background work performed and explains the
pre-requisites for this research, categorizes the subject being investigated, highlights the specific area of
interest, and at the same, time emphasizes their theoretical and practical significance. The theoretical
framework explains the area of interest, presents a Structural Equation Modeling (SEM), and analyses
the hypotheses created for the modelled constructs. The research methodology section gives a detailed
description of the research design adopted in the study, and explains analytically how the participants
were sampled. The survey descriptive results section presents the data collected, and provides the
demographics data which contribute in the theoretical analysis. The empirical findings /statistical
analysis section uses statistical tools to analyse the collected data from the sample, elaborates on all
steps taken to derive a valid model, and the final analysis and discussion section concludes to the final
structural model, and explains what the results obtained from the analyses mean. The conclusion aims
to discuss the findings and the analysis results, and intends to answer the research question, to highlight
the outcomes of the research and give suggestions for future work.




2. Theory

This chapter underlines the literature review which defines the theoretical framework of the thesis. It
provides an overview of the ethical and participative leadership, and emphasizes their relation to
innovative work behaviour. A high-level overview of self-efficacy and creative process engagement is
also provided to explain their influence on innovative work behaviour. This chapter constitutes the
theoretical framework upon which the SEM (Structural Equation Modelling) is based. It contains the
constructs and hypotheses for measuring and testing how leadership components affect the innovative
work behaviour.

Literature Review
Studies on ethical or participative leadership exist but are not necessarily combined in scientific articles
to tackle the issue of innovative output in the workplace. Therefore, this study focuses on ethical
leadership or participative leadership style, and their impact specifically related to employees’
performance and innovative work behaviour.

The current study links ethical leadership and participative leadership style with the self-efficacy,
creative process engagement and innovative work behaviour. This combination has not been
investigated by prior studies. The major rationale behind ethical and participative leadership is to
appreciate the role positive leadership plays, with employee’s output seen as self-efficacy, creative
process engagement and innovative work behaviour. It is of great importance to examine these two
different leadership styles together, because each of them touches upon different areas of technological
development. The participative management is generally fitting in a high technology environment
whereby the workforce is high educated and cannot be managed with simple top-down principles. The
ethical leadership is covering industries that imply either ethical principle to deal with others (such as
medicine) or the environment (such as oil& gas and shipping).It can be argued that these two positive
styles of leadership can be applied almost everywhere as even in the firm environments of
manufacturing, labour could participate in a production line improvement, or focus in working ethically
by respecting others, the law and the environment.

Innovation at Work
Lately, innovative work behaviour has presented a compelling importance because of the continuous
competition and the fast-changing global market (Shanker, et al., 2017). Work innovation is related with
change or altering behaviour that anticipates to adjust the tactic that a business works (Madrid, et al.,
2014). Innovative work behaviour expresses intention of introducing new products, services, concepts,
processes or procedures, as well as development and implementation of new divisions or establishments
in an organisation (De Jong & Den Hartog, 2008). Innovative work behaviour requires changing, so
people’s readiness for change contributes to the outcome of innovative achievement (Kwahk & Lee,
2008). Therefore, since prior studies have emphasized the importance of the innovation at work, this
study will contribute further on the examination of how ethical and participative leadership styles can
endorse employee’s innovativeness at work.

Influence of Leadership Methods


Even if the leadership literature has been examined through several prior studies, there is not a common
understanding and agreement on how leadership works and what it means (Akram, et al., 2016).
Although there is a lot of research with respect to managers’ leadership styles, there is still a lack of
understanding which are the factors that lead to innovative work behaviour (Akram, et al., 2016), and
there is also limited knowledge about the outcomes of this novel way of employees’ behaviour
(Trivisonno & Barling, 2016). Exceptionally, another study presented how ethical leadership improved
employees’ job performance from enhancing motivation (Piccolo, et al., 2010). This study states that
ethical leadership strengthens employees’ elemental motivation by arranging the neutral and individual
job aspects. It provides a better viewpoint in the encouragement aspect that ethical leadership offers and




the possible impact on the employees’ intrinsic motivation. The current study outlines how employee
self-efficacy is affected from ethical leadership but also from participative leadership style.

Another paper examining if participative leadership promotes the innovative work behaviour proved
that when leaders practice this style of management, the personnel feel more dedicated to adjust and
subsequently their innovative behaviour increases (Fatima, et al., 2017). The relevant study intends to
test different mediating mechanisms than those used in the aforementioned study to further explain the
relationship between participative leadership style and innovative work behaviour. In addition, it is vital
to take into consideration that the diversity of personality will affect the outcomes.

This thesis links leadership methods to innovative work behaviour from an employees’ creativeness and
self-belief perspective. A model is proposed and in order to contribute to the existing participative and
ethical leadership literature in several ways: (1) highlighting the employees’ self-efficacy – shows the
positive correlation of the leaders practicing participative and ethical leadership with the subordinates
in order to innovate; (2) emphasizing on employees’ creative process engagement – reliance on the
process of how participative and ethical leadership influences employees’ innovative work behaviour
through the mediation of creativity engagement.

Importance of Employee Creativity


Considering the continuously changing environments, high competition, and technological changes,
managers tend to motivate their employees to be more creative (Shalley & Gilson, 2004). Prior studies
have presented that employee creative process engagement can positively affect organizational novelty,
survival, and effectiveness (Shalley, et al., 2004). This study examines the creative process engagement
by employees and its correlation with both participative and ethical leadership practices. In this thesis,
creative process engagement is considered as a mediator that can enhance innovative work behaviour.

According to a previous study, for creativity to exist, managers should support and promote it, since
they are those that know better than anyone else within an organization which employee could improve
creativity; managers can also affect the occurrence of creativity (Zhang & Bartol, 2010). Within the
same study, it is also stated that enhancing leadership includes empowering employees’ motivation and
therefore, this action may lead to a positive impact on their creativity. This paper by Zhang & Bartol
(2010), leaves the margin for further research on how the individual variable of self-efficacy interrelates
with creativity. Hence, the present study will try to examine if self-efficacy of employees has a positive
relationship with the creative process and promotes the generation of innovative and valuable ideas.

Theoretical Framework
To investigate the impact of ethical and participative leadership in innovative work behaviour, the
theoretical framework is intended to test if self-efficacy and creative process engagement positively
influence the innovative work behaviour. The research framework is illustrated in Figure 1.




Figure 1 Research Framework

Ethical Leadership
Ethical leadership is defined as an effort from the leader to promote an appropriate conduct through
individual actions and interpersonal relationships to his/her followers, through an existing
communication between them and decision making (Brown, et al., 2005). The ethical leader is therefore
implied to have honesty, honour, unselfishness, reliability, cooperative incentive, and righteousness as
a moral person (Brown & Trevin˜o, 2006). The moral manager influences his employees’ approaches
and actions via the practicing of his ethical leadership behaviour (Trevin˜o & Brown, 2004). Ethical
managers/leaders are keen on impacting positively the work of the others, including groups,
organizations, and even entire societies (Brown & Trevin˜o, 2006). Their supporters are more interested
in producing new concepts for the reason of achieving the expected goals. Since the ethical leaders are
characterized by positive traits and are committed to the organization (De Hoogh & Den Hartog, 2008),
their followers are feeling safe to express their new ideas and share their knowledge with their colleagues
(Janssen, 2003).
Ethical leadership is crucial for organizations because it helps them to decrease business costs by
engaging fairness and morality towards their employees and all other stakeholders (Thomas, et al.,
2004). Prior research has investigated ethical leadership’s positive role, since it decreases the destructive
behaviours of employees and discourages the possible immoral practices (Mayer, et al., 2009).

Hypothesis 1: Ethical leadership is positively correlated to creative process engagement.


Hypothesis 2: Ethical leadership is positively correlated to self-efficacy.

Participative Leadership
Participative leadership is perceived as a positive style of leadership, and it is defined as an approach in
which the manager allows his/her peers to participate in decision making (Kahai, et al., 1997). The
leader creates opportunities to subordinates for contributing in problem-solving by the use of
encouragement (Somech, 2006). A sense of responsibility is created on subordinates when participative
leaders allow them to have a role in decision making (Sauer, 2011). Participative leadership application
has a high probability in producing good performances and various positive behaviours. Another
empirical study has shown that the impact of participative leadership on work results in different
industrial and cultural fields (Kahai, et al., 1997). It has been shown that participative leadership
increases job performance (Yousef, 2000). Participative leaders do not enforce decisions on
subordinates. However, they welcome employees’ suggestions and they take decisions based on
consensus (Somech, 2006). Employees by this way feel honoured and privileged, and at the same time
motivated that their leaders respect them, and treat them equally with the rest of the organization’s



members. The observation of the leader’s positive behaviour in the work environment on a repeated
basis makes employees to adopt such a behaviour as well. These assumptions allow the creation of the
following hypotheses:

Hypothesis 3: Participative leadership is positively correlated to creative process engagement.


Hypothesis 4: Participative leadership is positively correlated to self-efficacy.

Creative Process Engagement


Creative process engagement is defined as employees’ involvement in methods and processes relevant
to creativity (Zhang & Bartol, 2010). A pure indication of creative process engagement is when
employees are involved or engaged in intellectual processes relevant to creativity. The Creative process
engagement has three critical dimensions (Reiter-Palmon & Illies, 2004). The first dimension of
creativity process is when the employee identifies the issue, objectives, processes, limitations, and
information for the problem to be solved. The time spent for this first stage is key for the quality and
validity of problem’s resolution (Reiter-Palmon & Illies, 2004). When the problem is identified, the
employee continues by collecting and processing relevant information (Zhang & Bartol, 2010). The
second stage associates with searching information relevant with the identified problem in order to be
better understood (Mumford, 2000). The time spent on searching information positively affects the
solution’s quality, and is likely for the creativity to be increased. Finally, the last dimension in the
creativity process is to contemplate and create concepts related to the obstacle, while incorporating the
relevant information (Zhang & Bartol, 2010). Combining and organizing all the gathered information
creates a new way of understanding, and the applications and implications of this new understanding
obviously conclude to a set of new ideas (Mumford, 2000). Creative process engagement helps the
employees to engage in creative activities and remain committed throughout the creative process, until
new and useful ideas are realized (Hennessey & Amabile, 2010). According to the above arguments, the
following hypothesis is presented:

Hypothesis 5: Creative process engagement is positively related to innovative work behaviour.

Self-efficacy
Self-efficacy has been formed as an idea in social learning theory, and firstly shaped in the social
psychology grounds (Bandura, 2000). Bandura suggested that self-efficacy is crucial in tasks’
performance because it influences humans’ decisions, achievements and stamina. Based on this theory,
individuals can self-control and self-regulate; this means that their emotions and behaviours can be
controlled and their destiny determined at their will. These mental processes relate significantly with
people’s behaviour. Continuing Bandura’s idea, being in the possession of knowledge, skills and prior
accomplishments are not appropriate predictors of the future performance. However, people’s belief in
their own capabilities to accomplish different tasks and jobs is effective on their performance quality
(Bandura, 1997). Humans learn the standards of behaviour via direct modelling or verbal
encouragement; helping others to believe in their capabilities is reinforcing their motivational and
behavioural standards (De Hoogh & Den Hartog, 2008). To conclude, the aforementioned lead to the
following hypothesis:

Hypothesis 6: Self-efficacy is positively correlated to innovative work behaviour.


Innovative Work Behaviour
Innovative work behaviour is recognizing the problems, initiating and incorporating (as an employee,
or a member of a group, or organization) new and practical ideas concerning products, services, and
work methods, as well as having the adequate behaviour to establish and carry out these ideas aiming to
enhance personally or in terms of business (De Jong & Den Hartog, 2007). The last years, the great
value of innovative work behaviour has been shown due to the continuously changing global market as
well as the expanded competition (Shanker, et al., 2017). According to current trends in recruitment,
most of organizations check applicants’ cognitive and innovative abilities in order to assure that their




workforce will act with innovative work behaviour (Delgadová, et al., 2017). Organizations proceed in
changing their leadership styles for the reason of the vital role leaders play in promoting employees’
efficiency (Strom, et al., 2014).

Generally, it is hypothesized that all constructs positively affect Innovative Work Behaviour.

Summary
The presented literature review summarises the significant findings from previous researches which are
essential for this thesis. Several relative studies have been found for the variables examined (“Innovative
Work Behaviour”, “Ethical Leadership”, “Participative Leadership”, “Self-Efficacy” and “Creative
Process Engagement”). Previous studies have considered some of these variables combined together,
but no study has examined all the five derived constructs at the same time in a single model. As already
referred through the relevant thesis, both Participative and Ethical Leadership will be examined to
further explain their relationship with innovative work behaviour. It has to be noted, that in reality,
different types of personality affect the findings/results of this research. In addition, this research study
was also conducted to examine if employees high self-efficacy and creative process engagement can
produce innovation and whether the aforementioned management styles influence the mediators
positively.




3. Research Method

Overview
This study is based on a quantitative methodological approach called SEM, Structural Equation
Modeling. It is used because it evaluates and measures relationships between a set of variables and
indicators, and provides evaluation of which ones are reliable for the underlying theory tested. This
chapter summarizes the methodology used to pull together the SEM model and details the various
constructs.

Structural Equation Modeling (SEM) can explain the relationships among multiple variables, which are
expressed in equations that characterize all of the relationships among dependent and independent
variables (constructs) of in the analysis. Constructs are factors signified by multiple variables (Hair, et
al., 2014). SEM examines the relationships of dependent and independent variables in the same theory
(Hair, et al., 2014).

A SEM model expresses a theory as a systematic set of relationships which explain innovative work
behaviour reliably and specifically. A SEM model introduces a measurement model (how variables
represent constructs) and the structural model (constructs interrelationships); a structural model includes
structural relationships between latent constructs, these are: participative leadership style, ethical
leadership, creative work engagement, self-efficacy and innovative work behaviour.(Hair, et al., 2014).

According to Hair et al, “a latent construct cannot be measured directly but can be represented or
measured by one or more variables (indicators). In combination, the answers to these questions give a
reasonably accurate measure of the latent construct (attitude) for an individual." In our thesis, the
indicators are represented by the questions as explained later in this chapter, in accordance with SEM
theory.

SEM measurement theory consists of a “Series of relationships that suggest how measured variables
represent a construct not measured directly (latent). A measurement theory can be represented by a
series of regression-like equations mathematically relating a factor (construct) to the measured
variables" (Hair, et al., 2014).

The Six Stages of SEM


SEM is used to evaluate by data representation the realization feasibility of the theory. According to
Hair et al, “SEM can be defined by a six-stage decision process. This process reflects the unique
terminology and procedures of SEM.”

The six stages are listed below (Hair, et al., 2014):

¾ Stage 1: Defining individual constructs

¾ Stage 2: Developing the overall measurement model

¾ Stage 3: Designing a study to produce empirical results

¾ Stage 4: Assessing the measurement model validity

¾ Stage 5: Specifying the structural model

¾ Stage 6: Assessing structural model validity




Stage 1: Defining Individual Constructs


A decent measurement theory is essential to get useful outcomes from SEM (Hair, et al., 2014).
Considerable time must be devoted at an initial stage of the research process to guarantee valid decisions
are taken; defining the constructs is an important step in the research method.

The Operationalization of the constructs is an important follow up step, by selecting its measurement
scale and type. Constructs of the model can be defined and operationalized as similarly done in previous
studies. This study’s literature research regarding constructs, identified scales that have previously
performed well. Past research is an effective way to keep the standard of quality of new studies.
Regarding data collection, this thesis uses questionnaires; it is common for past questions from parallel
research to be incorporated in questionnaires (Hair Jnr., et al., 2010).

This study’s constructs are listed below:

¾ Ethical Leadership (EL)


¾ Participative Leadership (PL)
¾ Self-Efficacy (SE)
¾ Creative Process Engagement (CE)
¾ Innovative Work Behaviour (IW)

Stage 2: Developing the Overall Measurement Model


The model uses SEM and comprises of both a measurement model and a structural model. By using a
survey, the variables are measured. In SEM, the constructs are non-observable or latent factors that are
expressed by a variate that contains multiple variables. Using this methodology, several variables
incorporate mathematically a construct’s representation (Hair, et al., 2014).

After the constructs were specified, the measurement model was then created. All constructs used in the
model were defined together with the relationships between constructs, or else, the hypotheses. In this
model analysis, the constructs were considered unidimensional, which means that each indicator is
related to only one construct, thus cross-loadings are set to zero. The number of indicators per construct
is five for 4 constructs and 7 for one construct. This results in a total of twenty-seven indicators. It is
worth mentioning that more indicators do not automatically produce a better result. More indicators can
produce greater reliability results, but they need a larger sample (Hair, et al., 2014).

Path diagrams are advantageous in defining and communicating the theoretical model structure to the
program (Hair, et al., 2014). The model parameters specified (as shown in Figure 2) could be estimated
since the theoretical model structure was finalised. The path diagram created consists of 27 indicators
and is shown in Figure 2. The 27 questions created to represent the 27 indicators were part of the 2nd
stage of SEM.

Our thesis has concluded in using SEM as it is found to be the best method to assess management styles
and their influence to work behaviours. According to Hair, "Researchers are attracted to SEM because
it provides a conceptually appealing way to test theory. If a researcher can express a theory in terms of
relationships among measured variables and latent constructs (variates), then SEM will assess how well
the theory fits reality as represented by data."




Figure 2 Path Diagram

Stage 3: Designing a Study to Produce Empirical Results


After creating the theoretical model comprising of its constructs and indicators, the study’s design and
estimation issues must come into consideration. The analysis emphasises on three main points, “(1) the
covariances or correlations of data to be studied; (2) the consequence for missing data and its
provisions and (3) the effect of sample size” (Hair, et al., 2014).

The theory of communality is a suitable method for the purposes of reviewing the sample size issue;
communalities signify the average variation among the indicator variables described by the model (Hair,
et al., 2014). The selected model comprises of five constructs, and this study examines their
communalities.

This research aims to produce a minimum sample size of 150. According to Hair et al, a sample size of
150 is suitable for models with seven constructs or less and modest communalities (i.e. 0.5), and also
when there aren’t any under identified constructs.

Chapter 4 outlines the results from the 27 questions and forms part of Stage 3 of SEM. In addition,
Chapter 5, Empirical Results / Statistical Analysis, is also part of the 3rd stage of SEM and notably
includes Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA).

Stage 4: Assessing Measurement Model Validity


Validity checks defined the process of this research study as the study’s main statistical goal was to
provide a valid and reliable model. All the proposed models tested, with their validity checks to be a top
priority, and all necessary statistical enhancement methods were applied to achieve a valid model. The
validity of the SEM model was examined, by checking the fitness or goodness of-fit (GOF) results of
the measurement model which was expected to present acceptable levels. Additionally, more scrutinised
methods for determining convergent and discriminant validity as well as reliability of the model were
adopted. These steps are presented in chapter 5, the empirical findings / statistical analysis of this study.




Final SEM Stages: 5 & 6


These two stages are forming the final steps of this study, and include the creation of the final SEM
model and its validity checks. Chapter 6 elaborates on the creation of the final structural model, and
explains how the final CFA model derived in the empirical findings/statistical analysis (Section 5), is
transformed and explained in Structural Equation Modeling. It also provides a detailed analysis
discussion of the final proposed model of this study, and explains what the new formed model and its
constructs mean.

Data Collection
A survey was chosen in order for data to be collected for this study. After taking into consideration that
the expected number of participants was around 200 and the participants were coming from different
geographical locations and could be contacted through e-mail and social media platforms, the survey
method was found to be the most appropriate to use. It was anticipated to use answers from 150-200
questionnaires to provide a reliable result. This thesis intended to attain a minimum sample size of 150
responses as it used a model with 5 constructs. Larger samples could contribute better results, but in the
current study the recommended minimum sample of 150 is proposed for the number of SEM constructs
used (Hair, et al., 2014).

A brief explanation of survey as well as explanations of the structure of the questionnaire used in this
study will follow.

Surveys
Using surveys as a data collection method has a significant advantage of giving the possibility to collect
data from many respondents and compare the results in an easy way (Miller & Brewer, 2003). In
addition, as previously referred, the findings can be checked and duplicated by other researchers.
Conducting a survey has also some drawbacks. For example, the participant of the survey may
misunderstand some of the existing questions and this also may affect negatively the validity of the
study (Miller & Brewer, 2003). The potential of not responding to surveys is considered as another
problem of this data collection method. This concerns those participants that choose not to answer some
specific questions (Miller & Brewer, 2003).

The survey used for the relevant study is of self-administrated type. Self-administrated questionnaires,
most of the times are either on-line surveys or mail surveys (Bryman, 2012). These ways of spreading
the survey were both used here. These types of surveys used, were time-effective and gave the possibility
of reaching a greater number of respondents. Additionally, since it was difficult to reach the participants
in other ways, it was easier to approach them through internet. Moreover, specifically these surveys can
easily be modified and ease in answering them through text boxes (Bryman, 2012).

Descriptive Information on the Population Surveyed


The designed survey asked all the participants at the individual part, to indicate their gender, age group,
educational level, work position and location of employment. In the analysis part, all the information
gathered from the population that participated in the current survey will be further investigated.

Firstly, gender was measured as female or male. The respondents were called to choose among four age
groups covering all the ages existing in a working environment starting from 18-30 and continuing by
31-40, 41-50 and up to 50 years old. In continuation, the educational background of the professionals
was measured as High-school, 2-yr College, 4-yr College, Master Level and PhD Level. Regarding the
location of employment, each respondent was free to write his/her own country since the questionnaire
was spread to different countries (mainly European Countries).




Questionnaire
As previously indicated, this study used an electronic questionnaire to collect the data. This
questionnaire aimed to gather data related to the leadership styles, creative process engagement, self-
efficacy and innovative work behaviour. The data collected through the relevant questionnaire were also
used to conclude if the different leadership styles, creative process engagement and self-efficacy have
an impact on the reported innovative work behaviour.

The designed questionnaire has a closed question structure, and can be found at the end of this study in
Appendix A. It was divided into three parts. The first section was created to explain to the participants
the purpose and scope of this study, as well as ensuring them that all their personal information filled in
the questionnaire would remain confidential.

The second part consisted of the demographic questions of the respondents, which includes, gender,
age, education background, position and location of employment.

The third part contained questions for each one of the measured indicators, i.e. Ethical Leadership (EL),
Participative Leadership (PL), Creative Process Engagement (CP), Self-Efficacy (SE) and Innovative
Work Behaviour (IW), in which a five-stage scale is used to measure them. The scale presented the
following ranges “1: Strongly disagree”, “2: Disagree”, “3: Neutral”, “4: Agree” and “5: Strongly
agree”. This questionnaire was distributed online via email and other messaging services.

This survey concluded by urging all the respondents to forward the questionnaire to other candidates
adequate to answer the same questions, in order to further maximize the number of responses and
increase the validity of the results.

Missing Data
During the process of data collection, and when using a questionnaire as a data collection tool, the
problem of missing data may arise (Gyimah, 2001). It was possible that the respondents may have
skipped some of the questions in terms of minimizing the answering time or s/he may not observe some
of the questions need to be filled. Also, it is a fact that some respondents may feel unsecure of sharing
their personal opinions and details. Therefore, it was of great importance to try to minimize the missing
data that could possibly occur for the current study.

For this reason, the questionnaire was spread in two formats. The first one consisted of a word document
accompanied with a message emphasizing that the respondents should fill all the questions, since
uncompleted questionnaires will hot help the analysis. This would eliminate the possibility of missing
a question. The second one was an electronic questionnaire that limits the possibility of having missing
data, since all the questions were mandatory to be filled in order to submit it. Hence, it was assured that
all the submitted questionnaires were dully filled in. To conclude, both formats also contained a small
text assuring the respondents that all their personal details would remain confidential.




4. Survey Descriptive Results

Overview
This section summarizes the results of the data collection. Firstly, the demographic details of the
respondents are presented in order to understand the background of the population that has been
surveyed, and how the results relate to this population. Secondly, a data set was generated in order to
perform a SEM analysis.

Respondent Demographics
All the respondents were contacted via email and explained the importance of the study together with
the link leading them to a web-based questionnaire using “google forms” or an attached questionnaire
document in word format. The survey was sent out to a broad network of professionals mostly from the
United Kingdom, Greece and some other European countries (e.g. Sweden, Norway etc.). All the
respondents of this study were full-time employees. Some of the participants voluntarily spread the
survey to further assist in increasing the number of the respondents and enhancing the validity of this
research study. After receiving the first completed questionnaires within the first three days, a reminder
e-mail was sent to respondents in order to achieve as many collections as possible. The participating
invitation was sent to just over 200 respondents. The collection period was ran for one week; during this
time, 177 completed questionnaires were collected.

In this section, the demographic data are presented through graphs created with the help of Microsoft
Excel, as well as with tables extracted from IBM SPSS analysis software.

Gender
The following bar chart illustrates the number of men and women participants. In this study of the 177
respondents in total, 79 of them were female, while the male participants 98 in total, found to be more
in number than that of women workers (Figure 3).

'ĞŶĚĞƌ
ϭϮϬ

ϭϬϬ
ϵϴ
ϴϬ
ϳϵ
ϲϬ

ϰϬ

ϮϬ

Ϭ
&ĞŵĂůĞ DĂůĞ

Figure 3 Gender




Age
For this survey the respondents had to choose among four age groups, 18-30, 31-40, 41-50 and over 50
years old. The highest percentage of 34% showed that the respondents belong to the first age group of
18-30 years. The second higher of 29% corresponds to the second age group of 31-40. The percentages
of the two last age groups were almost similar, since workers of 41-50 years old represented the 19% of
the population and professionals up to 50 years old almost the 18% (Figure 4).

'

ϭϵй ϭϴй
хϱϬ
ϭϴͲϯϬ
ϯϭͲϰϬ
Ϯϵй ϯϰй ϰϭͲϱϬ

Figure 4 Age

Location of Employment
Geographical locations of the working environment are identified in this questionnaire. Focusing on
European respondents, the majority of them found to be from Greece by 58%, followed by United
Kingdom with a percentage of 29% and of 6% from other countries. The high percentage of Greece and
the United Kingdom was expected, as persons within those two countries were active in their support in
the process of data collection. Participants from Sweden acquire the 4% of the surveyed population,
while professionals from Norway obtained the smallest percent of 3% (Figure 5).

>ŽĐĂƚŝŽŶŽĨŵƉůŽLJŵĞŶƚ

'ƌĞĞĐĞ

Ϯϵй EŽƌǁĂLJ
KƚŚĞƌ

ϰй ϱϴй ^ǁĞĚĞŶ
ϲй hŶŝƚĞĚ<ŝŶŐĚŽŵ
ϯй

Figure 5 Location of Employment




Educational Level
Besides the location of employment, it is important to understand the level of education of the
participants, hence this questionnaire is also designed to capture this data. The respondents came from
a variety of industries in the technological sector. In terms of their educational level, almost more than
the half answered that they obtain a Master with a percentage of 51%. The second larger percentage of
40% of the population answered that they have accomplished their studies in a 3/4-year College. A small
amount of 4% of the participants obtain the higher university degree (PhD level). The 3% of the
respondents answered that they have a high-school degree and only a small minority of 2% found to
have a 2-year college degree (Figure 6).

ĚƵĐĂƚŝŽŶĂů>ĞǀĞů

ϮͲLJƌŽůůĞŐĞ
WŚ>ĞǀĞů Ϯй
ϰй
ϯͬϰͲLJƌŽůůĞŐĞ ϮͲLJƌŽůůĞŐĞ
ϰϬй ϯͬϰͲLJƌŽůůĞŐĞ
DĂƐƚĞƌůĞǀĞů
ϱϭй ,ŝŐŚͲƐĐŚŽŽů
,ŝŐŚͲƐĐŚŽŽů
DĂƐƚĞƌůĞǀĞů
ϯй
WŚ>ĞǀĞů

Figure 6 Educational Level

Position
Another important characteristic to be examined is the position of the respondents within their
companies. The relevant survey tries to cover a range of positions existing in technological firms, for
which the surveyed professionals are employed. The pie chart of Figure 5 illustrates the different
positions of the participants. Belonging to Management accounted for 24 per cent. Unsurprisingly, a
large amount of people of 36%, has Engineering roles, considering that the surveyed population works
mostly in technology sector. The 16% works in other positions different than those described in the
questionnaire. A smaller amount of 11% works in Finance/Accounting while, the 8 per cent corresponds
to Administration and the minor percent of 3% represents those working in R&D (Figure 7).




WŽƐŝƚŝŽŶ

ϯй ϴй
ϭϴй ĚŵŝŶŝƐƚƌĂƚŝŽŶ
ŶŐŝŶĞĞƌŝŶŐ
ϯϲй &ŝŶĂŶĐĞͬĐĐŽƵŶƚŝŶŐ
Ϯϰй
DĂŶĂŐĞŵĞŶƚ
ϭϭй
KƚŚĞƌ
ZΘ

Figure 7 Position




5. Empirical Results / Statistical Analysis

This section outlines the statistical analysis performed to evaluate the theoretical model. The analysis
examines the model’s validity and reliability, and re-structures the model as required in a step-by-step
fashion in order verify this study’s theory. The statistical analysis consists of the following steps:

1. Theoretical SEM Model Construction


2. SEM theoretical model Analysis
3. Confirmatory Factor Analysis of Theoretical SEM Model
4. Exploratory Factor Analysis
5. CFA Model Analysis and Corrections
6. Final CFA Model
7. Construct Validity and Reliability

Theoretical SEM Model Construction


For the purpose of this study, a model of five constructs was created, where each construct correlates to
specific questions in the questionnaire. The twenty-seven closed-ended questions in total are directly
related to the constructs provided through the questionnaire. All data collected were analysed with IBM
SPSS software for obtaining statistical results. A further analysis was conducted with AMOS software
to gain insight of how the constructs interacted among them.

The data was adjusted for input to SPSS for performing analysis of the SEM model. A unique code was
created for each question used from the five constructs (also mentioned in the Theoretical framework
part of this study). The demographic questions were used as “non-numerical” in the SPSS program. The
codes used for the five constructs of the SEM Model as presented below:

¾ Ethical Leadership : Question 1 = EL1, Question 2 = EL2, Question 3 = EL3, Question 4 = EL4,
Question 5 = EL5

¾ Participative Leadership : Question 1 = PL1, Question 2 = PL2, Question 3 = PL3, Question 4 =


PL4, Question 5 = PL5, Question 6 = PL6, Question 7 = PL7

¾ Self-Efficacy : Question 1 = SE1, Question 2 = SE2, Question 3 = SE3, Question 4 = SE4, Question
5 = SE5

¾ Creative Process Engagement : Question 1 = CP1, Question 2 = CP2, Question 3 = CP3, Question
4 = CP4, Question 5 = CP5

¾ Innovative Work Behaviour : Question 1 = IW1, Question 2 = IW 2, Question 3 = IW 3, Question


4 = IW 4, Question 5 = IW5

Note: All questions can be found more detailed in Appendix A.

The SEM model was created by categorizing the constructs to exogenous and endogenous. Ethical
Leadership and Participative Leadership considered to be exogenous since they are not dependent on
one another. This means that they have correlational relationships with other constructs and act as
independent variables in structural relationships (Hair, et al., 2014). The mediator constructs (Creative
Process Engagement and Self-Efficacy) were considered to be endogenous as they are indicating a
dependence relationship. Innovative work behaviour is classified as an endogenous parameter as well.




Table 1: Constructs Description


Exogenous Constructs Endogenous Constructs

Creative Process Engagement


Ethical Leadership

Self-Efficacy

Participative Leadership
Innovative Work Behaviour

After defining the relationships and path diagram of the model, the data were incorporated into the
program in a format suitable for analysis. Firstly, this analysis aimed to estimate the strength of the
aforementioned relationships. Secondly, it targeted to show how the constructs of the designed model
were associated with each other for testing and proving the theoretical hypotheses. The theoretical model
is shown in Figure 8.

Figure 8 Theoretical SEM Model build in IBM Amos




SEM Theoretical Model Analysis


After inputting the questionnaire data in Amos, the first step was to run the theoretical model as Shown
in Figure 8. In order to examine its normality. Figure 9 shows the SEM model and all the relationships
(loadings) as appear after calculating estimates. Before proceeding to the actual analysis techniques
presented in next section, a check of normality of data was performed at this stage for the purpose of
evaluating the model at this early stage.

Figure 9 Theoretical SEM model first run


Assessment of Normality for the SEM Theoretical Model
Normality of data check determines whether the examined dataset is well demonstrated by a normal
distribution. If the dataset is modelled correctly, the researcher will be enabled to evaluate and make
suggestion to the random samples tested from the overall population (Bryman & Bell, 2007).

The results of assessment of normality are presented in Table 2 below. The first column represents the
variables and corresponds to all the questions in the questionnaire, whereas the second and third columns




denote the maximum and minimum recorded values of the replies gained. The Skew and Kurtosis values
also shown in columns 4 and 6 respectively.

When a Kurtosis value is 0, it means that the dataset has perfect normality. Generally, Kurtosis values
between -3.0 and +3.0 can be deemed acceptable (Kline, 2010). Though, values between -2.0 and +2.0
are satisfactory when showing normal univariate distribution (George & Malley, 2010). All of the
Kurtosis values presented below are within the acceptable limits. Thus, it can be assumed that the
collected dataset has acceptable normality.

Column 5 (C.R.) represents the T-Values which will be examined during the CFA analysis of this thesis.
Three values are above 5 and are highlighted in amber colour; as according to Bentler & Byrne (Bentler,
2005) (Byrne, 2009) the C.R values > 5.0 indicate non normality. This indication showing that the 3
variables need to be deleted from the model.

Table 2: Assessment of normality Theoretical SEM Model


Variable min Max skew c.r. kurtosis c.r.
PL1 1,000 5,000 -,642 -3,485 ,509 1,381
PL2 1,000 5,000 -,376 -2,045 ,157 ,425
PL3 2,000 5,000 -,143 -,777 -1,057 -2,872
PL4 2,000 5,000 -,380 -2,061 ,166 ,451
PL5 2,000 5,000 -,185 -1,005 -,501 -1,360
PL6 2,000 5,000 -1,253 -6,808 1,558 4,232
PL7 2,000 5,000 -1,234 -6,701 ,743 2,017
IW5 1,000 5,000 -,177 -,963 -,561 -1,524
IW4 3,000 5,000 -,152 -,827 -,570 -1,549
IW3 2,000 5,000 -,522 -2,836 ,383 1,039
IW2 2,000 5,000 -,111 -,603 -,689 -1,870
IW1 2,000 5,000 -,344 -1,867 ,020 ,055
CP1 3,000 5,000 -,264 -1,434 -,764 -2,074
CP2 2,000 5,000 -,253 -1,376 -,553 -1,503
CP3 2,000 5,000 -,352 -1,913 -,509 -1,383
CP4 2,000 5,000 -,406 -2,207 -,220 -,597
CP5 1,000 5,000 -,239 -1,300 -,312 -,848
SE5 1,000 5,000 -,622 -3,377 ,265 ,720
SE4 1,000 5,000 -,603 -3,273 ,076 ,207
SE3 1,000 5,000 -,888 -4,825 1,464 3,976
SE2 2,000 5,000 -,507 -2,753 ,387 1,051
SE1 2,000 5,000 -,657 -3,567 ,388 1,054
EL5 2,000 5,000 -,851 -4,624 ,051 ,138
EL4 2,000 5,000 -,481 -2,614 -,504 -1,370
EL3 2,000 5,000 -,367 -1,994 ,779 2,117
EL2 3,000 5,000 -,122 -,665 -,480 -1,305
EL1 3,000 5,000 -1,136 -6,173 -,256 -,696
Multivariate 62,644 10,530




Confirmatory Factor Analysis of the Theoretical SEM Model


Confirmatory Factor Analysis (CFA) is used to test the questions and their linked constructs, and
evaluate whether the data fit the theoretical measurement model. The theoretical model created is related
on the background theory and literature review as presented in previous sections, thus, it is envisaged
that the SEM model has theoretical support. The arrows in a CFA model are bi-directional as seen in
Figure 10, in comparison to a dependence relationship presented in the theoretical SEM model. In CFA
all combinations of connected constructs (covariances) are used.

The estimates were calculated and are shown in Figure 10. The results obtained from the CFA model
can now be studied.

Figure 10 CFA of SEM model




Model Fit Checks


Next step is to investigate model of fit for the current CFA model, and obtain the most significant results
for the analysis. Model stipulates how well the factor structure projected model fits for the correlations
between variables in the dataset.

P value should be <.05 (Bagozzi & Yi, 1988) to have a significant Chi-square. Chi-square is positive.
Comparative Fit Index (CFI) should be larger than 0.90 (Bagozzi & Yi, 1988), although here it is 0.748.
The CFI is an incremental fit index otherwise an improved normed fit index (NFI). The CFI values have
a range between 0 and 1. Higher values prove better fit (Hair, et al., 2014).

The root mean square error of approximation (RMSEA) should be less than 0.095 (Bagozzi & Yi, 1988),
however here it is 0.063. RMSEA represents the level of error in a sample, and examines how well the
model fits with the population. Lower RMSEA values show better model fit (Hair, et al., 2014). More
specifically, all the values examined to prove a strong model fit of the CFA, as well as their required
limits are presented in Table 3.

Table 3: Values to check Model Fit


Value Required
Probability Level P 0.000 < 0.05
Chi-Square CMIN 531.152 No Limit
Comparative Fit Index CFI 0.748 > 0.9
The root mean square
RMSEA 0.063 < 0.095
error of approximation
*According to ( (Hair, et al., 2007); (C., & Baumgartner, 1995); (T.J. & W.,
& Schindler , 2001))

The results obtained after running the theoretical model in CFA were unsatisfactory. It was an obvious
indication that the theoretical framework needed to be looked deeper and use the statistical techniques
to create a model that will provide validity, reliability, and at the same time relate to the theoretical
background research of this thesis. These steps were necessary in this type of research as when a model
is valid, the theories created can be elaborated and explained in the analysis and discussion section.

Exploratory Factor Analysis


This Section uses SPSS Software to analyse the dataset obtained by the questionnaire and to examine
how reliable the model is prior to running a Confirmatory Factor Analysis of the theoretical model in
AMOS. Factor analysis is useful for achieving data reduction and suitable variable selection.

Factor analysis was used to understand the interrelationships between the model’s variables, as it is
suggested to be an essential step in order to define the structure of the data used (Hair, et al., 2014). This
thesis analysis started by using SPSS to evaluate which variables may have an impact in the model.

Reliability
The reliability of the model was examined with Cronbach's Alpha. The SPSS dataset shows values above
0.8 (Hair, et al., 2014). According to (George & Malley, 2003) the rule of thumb for the understanding,
the values for the Cronbach Alpha can be used, such as that:

ƒ values > 0.9:- “Excellent” reliability


ƒ values > 0.8:- “Good” reliability
ƒ values > 0.7:- “Acceptable” reliability
ƒ values > 0.6:- “Questionable” reliability



ƒ values > 0.5:- “Poor” reliability


ƒ values < 0.5:- “unacceptable” reliability

The SPSS results obtained are presented in table Table 4 below.

Table 4: Reliability of Data in SPSS


Item-Total Statistics
Cronbach's
Survey Alpha if
Construct Reliability
Question Item
Deleted
EL1 0.828 Good
EL2 0.828 Good
Ethical
EL3 0.828 Good
Leadership
EL4 0.826 Good
EL5 0.830 Good
PL1 0.826 Good
PL2 0.828 Good
PL3 0.834 Good
Participative
PL4 0.827 Good
Leadership
PL5 0.829 Good
PL6 0.826 Good
PL7 0.830 Good
CP1 0.826 Good
Creative CP2 0.826 Good
Process CP3 0.821 Good
Engagement CP4 0.825 Good
CP5 0.825 Good
SE1 0.829 Good
SE2 0.826 Good
Self-Efficacy SE3 0.826 Good
SE4 0.831 Good
SE5 0.825 Good
IW1 0.822 Good
Innovative IW2 0.821 Good
Work IW3 0.821 Good
Behaviour IW4 0.824 Good
IW5 0.826 Good

Factor Reduction using Rotated Component Matrix


A factor method was initially selected, and the total variance was looked at so the factors were extracted
with component analysis. The factor matrix determined the number of factors to be recalled. The
selected rotational method was vital to the way the analysis was performed; an orthogonal method is
normally selected when the factors are uncorrelated. In this analysis, VARIMAX is used.

After the rotated factor matrix was run, the interpretation of the results and the checking of the loadings
significance followed. Typically, component factor analysis is most suitable when data reduction is a




primary focus, estimating on getting the minimum amount of factors that can explain the maximum
percentage of the total variance represented in the original set of variables (Hair, et al., 2014).

In this analysis, factor reduction was performed in order to evaluate the significance of the loadings.
Practical significance was the preliminary check of the factor matrix regarding the factor loadings. A
factor loading is the correlation of the variable and the factor, where the squared loading is the measure
of the variable’s total variance explained by the factor (Hair, et al., 2014).

According to Hair (Hair, et al., 2014), when using practical significance, factor loadings are categorized
in:

1. Loadings in the range of ±0.30 to ±0.40 meet the minimal level for interpretation of structure
2. Loading greater than ±.50 are practically significant
3. Loadings exceeding 0.70 are indicate a well-defined structure and are the goal of any factor
analysis

Note: Exceptionally high loadings (greater than 0.80) are not typical. These recommendations are used
for examining the practical significance of the loadings, and are related for analyses with sample size of
100 or greater.

It should be noted that statistical significance guidelines differ to practical significance. In this analysis,
the stricter practical significance was used initially, but in order to maintain a good model which would
not vary from the theoretical model, the concept of statistical significance to be used for different sample
sizes was also considered for the purpose of concluding to a well-defined model. Table 5 contains the
sample sizes needed for each factor loading value to be considered significant (Hair, et al., 2014).

Table 5: Statistical Significance


Guidelines for Identifying Significant Factor
Loadings Based on Sample Size
Sample Size Needed
Factor Loading
for Significance

0.30 300

0.35 250

0.40 200

0.45 150

0.50 120

0.55 100

.060 85

0.65 70

0.70 60

0.75 50

The factor reduction in SPSS was achieved by using the default method of principal components,
extracting eigenvalues over 1 and suppressing absolute values to less than 0.4. The initial factor solution
was chosen to be rotated using Varimax rotation. Results are shown in Table 6. Following the statistical



significance guidelines presented above, for a sample greater than 150, the factor loadings are significant
from 0.45 and above, and thus, the model was set not to consider values below 0.4 for this factor
reduction exercise. The rotation was initially set for all 27 variables and run in order to obtain the results.

Table 6: Factor Reduction - Rotated Component in SPSS


Rotated Component Matrixa
Component
1 2 3 4 5 6 7 8 9
IW3 0.733
IW1 0.716
IW5 0.536
IW2 0.509 0.422
CP5
CP3 0.745
CP2 0.690
CP4 0.580
PL2 0.758
SE4 0.698
PL1 0.585
PL5 0.677
PL7 0.637
PL4 0.634
PL3 0.527
EL4 0.429
EL2 0.633
EL5 0.578
EL3 0.401 0.566
EL1 0.557
SE1 0.821
SE2 0.782
CP1 0.653
IW4 0.436
SE3 0.418 0.421
PL6 0.725
SE5 0.764
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.
a. Rotation converged in 14 iterations.

Nine factor loadings had a value of 0.70 significance as per practical significance guidance and fourteen
loadings of significance above 0.45 as per statistical significance recommendation. The rest of the factor
loadings could not be used for the structural model.




Total Variance of Complete Dataset


After checking the factor loadings, the total variance was examined. As presented in Table 7, nine
specified components explain the 61.3% of the complete data variance. In the social sciences, a model
that explains for 60 percent of the total variance (or even less) it is not unusual to be acknowledged as
satisfactory (Hair, et al., 2014). Nevertheless, this thesis focused in achieving a more significant
cumulative variance and many combinations of the model were tried out in order to select a suitable
model with higher variance percentage as described in section 5.4.4.

Table 7: Total Variance - Rotation Sums


Total Variance Explained

Component Rotation Sums of Squared Loadings


% of
Total Variance Cumulative %
1 2.462 9.118 9.118
2 2.321 8.596 17.715
3 2.015 7.461 25.176
4 1.973 7.308 32.484
5 1.899 7.032 39.515
6 1.715 6.351 45.867
7 1.426 5.282 51.149
8 1.418 5.251 56.400
9 1.309 4.850 61.250
Extraction Method: Principal Component Analysis.

Suitable Model Selection Using Factor Reduction


SPSS factor reduction analysis involved several tries in order to select a suitable model that can be
transferred to Confirmatory Factor Analysis (CFA) and explain the questionnaire and the related
questions derived. In Table 8, the proposed model is presented. The model consists of seventeen factor
loadings corresponding to seventeen variables and seven components. There are ten factor loadings
above 0.7 and eight above 0.45.
Table 8: Factor Reduction – Suitable Model
Rotated Component Matrixa
Component
1 2 3 4 5 6 7
IW1 ,743
IW3 ,705
IW5 ,678
IW2 ,630
CP2 ,720
CP3 ,711
CP4 ,633
CP1 ,503
SE1 ,819
SE2 ,811
PL5 ,718



Rotated Component Matrixa


Component
1 2 3 4 5 6 7
PL4 ,680
PL3 ,679
EL5 ,766
EL3 ,697
SE3 ,763
PL1 ,842
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.
a. Rotation converged in 8 iterations.

Total Variance of Suitable Model Dataset


The factor loadings check is completed for the selected model in section 5.4.4, and the total variance
was examined. As presented in Table 7, seven specified components explain the 66.71% of the complete
data variance, and thus it can be considered as a significant model. Therefore, the model can now be
examined under CFA analysis. Even though in this analysis exercise the ideal target of cumulative
percentage was 70%, it was decided to analyse in CFA the current model, since it would offer the
possibility to relate the original theoretical model with the final model and prove the literature review
of this study. It was decided to keep the 5 constructs with the remaining variables and run them in Amos.

Table 9: Total Variance - Rotation Sums


Total Variance Explained
Rotation Sums of Squared Loadings
% of Cumulative
Component Total Variance %
1 2,224 13,083 13,083
2 2,174 12,787 25,869
3 1,571 9,240 35,109
4 1,523 8,957 44,066
5 1,465 8,618 52,684
6 1,238 7,285 59,969
7 1,146 6,744 66,713
Extraction Method: Principal Component Analysis.

CFA Model Analysis and Corrections


From the original theoretical model of 27 variables, only 17 variables remained to be transferred for
CFA analysis. The aim of the next steps is to fine tune the CFA model and to examine its reliability and
validity before re-structuring a final SEM as a result of this research study.

After deleting 10 questions (IW4, CP5, EL1, EL2, EL4, SE3, SE4, PL2, PL6 & PL7) as a result of factor
reduction analysis, an assessment of normality was performed again and showed that all values
displayed were normal. It has to be noted that despite the fact EL1, PL6 and PL7 were significant in the
SPSS model, it proved to have non normal values in Amos from the theoretical CFA first run shown in



Figure 10. Thus, these variables were excluded from the SPSS model. In Figure 11 the first run after
factor reduction is shown.

Model Fit Summary was checked for the CFA model and the results were observed proved not to meet
the required values. With a CFI at 0.860 as shown in Table 10, the model needed to be looked at for
further correction. The results comparing to the theoretical model CFA have improved, and it can be
seen as the Goodness of fit of GFI 0.899 and AGFI of 0.858 presented in Table 10.

The same improvements can be observed by looking at where the CMIN value and the reduced Chi-
Square is given as well as in Table 10. Finally, the RMSEA values are shown and represent the level of
error in the model.

Table 10: CFA Model Values


Value Required
Relative Chi-square CMIN/DF 1,651 0> CMIN/DF < 5
Probability Level P 0.000 < 0.05
Chi-Square CMIN 179.910 No Limit
Goodness-of-Fit GFI 0.899 > 0.9
Adjusted Goodness-Of-Fit AGFI 0.858 > 0.8
Comparative Fit Index CFI 0.860 > 0.9
The root mean square
RMSEA 0.061 < 0.095
error of approximation
*According to ( (Hair, et al., 2007); (C., & Baumgartner, 1995); (T.J. & W.,
& Schindler , 2001))




Figure 11 CFA Model after Factor Reduction

Estimates Review
At this stage, the model was improved but still not to the validity level required for defining a well-
structured argument for the research. The standardized regression estimates were checked for values
below 0.5. By looking at the estimates as Table 11 shows, it can be observed that PL3, EL5, SE3, CP1
and PL1 were below the 0.5 threshold. The very low values were removed one by one so that the model



improves its reliability and validity results. After many runs, it was decided to delete PL1, PL3, SE3,
CP1, and IW5 in order to keep a model equilibrium to the theory. The final CFA model was formed
with 12 variables as described in section5.6.

Table 11: Estimates after 10 variables deleted


Estimate
EL5 <--- Ethical Leadership ,463
PL5 <--- Participative Leadership ,549
PL4 <--- Participative Leadership ,576
PL3 <--- Participative Leadership ,352
SE1 <--- Self-Efficacy ,622
SE2 <--- Self-Efficacy ,751
SE3 <--- Self-Efficacy ,297
CP1 <--- Creative Process Engagement ,493
CP2 <--- Creative Process Engagement ,590
CP3 <--- Creative Process Engagement ,729
CP4 <--- Creative Process Engagement ,618
IW1 <--- Innovative Work Behaviour ,655
IW2 <--- Innovative Work Behaviour ,651
IW3 <--- Innovative Work Behaviour ,652
IW5 <--- Innovative Work Behaviour ,584
PL1 <--- Participative Leadership ,225
EL3 <--- Ethical Leadership ,611

Final CFA Model


The final CFA model as shown in Figure 12 achieved the required values in all Goodness-of-fit
checks.




Figure 12 Final CFA Model




Model Fit Checks


After deleting further five questions highlighted in section 5.5.1, from the Standardized Regression
Weights estimates, the 12 remaining variables formed the CFA model. The estimates were calculated
and the results were examined.

The new Chi-Squared of 61.12 is supported since the probability factor is 0.036 (<0.05) with degrees of
freedom to be 43 as indicated in Table 12. The relative Chi Square, Goodness-of-fit, adjusted Goodness
of fit and Comparative Fit Index are all within acceptable limits.

Table 12: Final CFA Model Checks


Value Required
Relative Chi-square CMIN/DF 1.421 0> CMIN/DF < 5
Probability Level P 0.036 < 0.05
Degrees of Freedom DF 43 N/A
Chi-Square CMIN 61.122 No Limit
Goodness-of-Fit GFI 0.946 > 0.9
Adjusted Goodness-Of-Fit AGFI 0.903 > 0.8
Comparative Fit Index CFI 0.949 > 0.9
The root mean square
RMSEA 0.049 < 0.095
error of approximation

*According to ( (Hair, et al., 2007); (C., & Baumgartner, 1995); (T.J. & W.,
& Schindler , 2001))

Construct Validity and Reliability


In previous sections, the final CFA model was derived and produced valid results. Reliability of the
theoretical model was checked though Cronbach’s Alpha and proved to give good reliability. Validity
and reliability were examined further to provide extra scrutiny of the final model. This chapter intends
to precisely present the strong and weak points of the CFA model, and allow for further research to be
continued by indicating the room for improvement as well as the achievements of the study.

Construct validity was used to provide a check on the correctness of the model’s measurement. This
check expressed to what degree the set of variables measured the theoretical latent construct that they
were planned to. The better the construct validity, the more assurance can be provided that item
measurements of the sample exemplify the true score of the sample population (Hair, et al., 2014).
Convergent Validity means that items that are variables/indicators (questionnaire items) of a construct
should converge or share a high proportion of variance in common. A good rule of thumb indicates that
standardized loading estimates should be measured 0.5 or higher, and ideally 0.7 or higher. (Hair, et al.,
2014).

It can be concluded that the factor loadings are higher than 0.5, in some cases higher than 0.7. The only
factor loading which is below 0.5 is EL5, having a value of 0.361 in the model. It has to be noted that
in this model, trial and error practice was conducted, and EL5 was decided to remain in the model in
order to maintain 5 constructs in the CFA model, and explain the theory of ethical management in
relation to creative and innovative behaviours. In the later stage of this study, when the SEM model is
produced, EL5 is the only indicator remained that explains an ethical behaviour of management and
how it affects the employee innovative work behaviour as per literature review. In Table 13 the Estimates
of Standardized Regression Weights are identified.




Table 13: Final CFA Model Standardized Estimates


Indicator Construct Estimate
EL5 <--- Ethical Leadership 0.361
EL3 <--- Ethical Leadership 0.784
PL4 <--- Participative Leadership 0.639
PL5 <--- Participative Leadership 0.511
SE1 <--- Self-Efficacy 0.658
SE2 <--- Self-Efficacy 0.732
CP3 <--- Creative Process Engagement 0.749
CP4 <--- Creative Process Engagement 0.633
CP2 <--- Creative Process Engagement 0.564
IW1 <--- Innovative Work Behaviour 0.602
IW2 <--- Innovative Work Behaviour 0.753
IW3 <--- Innovative Work Behaviour 0.811

Another method utilised in this study to evaluate convergent validity was by Average Variance
Extracted (AVE); as recommended by Hair et al (2014, p. 632), is a precise construct validity
assessment. AVE value is the sum of the squared factor loadings divided by the number of loadings
(lambdas or Ȝ). AVE values of each construct were calculated as presented in Table 14. Hair et al (2014,
p. 619) indicate that when using this method of calculation, the desired result should use the same as
factor loadings, i.e. an AVE of 0.5 or higher is a good rule of thumb signifying satisfactory convergence.
However, if AVE of less than 0.5, it means that more errors remain in the items than variance explained
by the latent factor structure (Hair, et al., 2014).

The AVE values for Innovative Work Behaviour (0.53) and Self-Efficacy (0.48) when rounding the
number meet the 0.5 threshold and thus can be said that AVE calculation classifies them with significant
validity. However, for Ethical Leadership (0.37), Participative Leadership (0.33) and slightly less for
Creative Process Engagement (0.43), the constructs are below 0.5 threshold. For Creative Process
Engagement, because the Composite Reliability is higher than 0.6, the validity value of 0.43 can be
accepted as it is above 0.4 (Fornell & Larcher, 1981).

The issues, appear to be with convergent validity on Ethical and Participative leadership, could be
computed and enhanced in future work, and is a result of the subjective choices of the sample’s
population. The issues of validity are elaborated further in the conclusions and suggested further work
sections. Nonetheless, for this thesis analysis, the basis of using estimates with values of 0.5 or higher
significance is used for evaluating an acceptable validity.

Construct reliability was also used to measure convergent validity. According to Hair et al (2014, p.
619) Coefficient alpha is still a commonly used reliability approximation despite the fact it may under
measure reliability. Composite Reliability (C.R.) can be calculated from the squared sum of factor
loadings (Ȝ) for each construct and the sum of the error variance for a construct (1-Ȝ2). The rule of thumb
for C.R. is normally 0.7 or higher for good reliability and 0.6 or higher for acceptable reliability. A high
construct reliability specifies existence of internal regularity, meaning that the variables consistently
represent the same latent construct (Hair, et al., 2014).

As shown in Table 14, the C.R. values for the 5 constructs are presented. Creative Process Engagement
(0.69) and Innovative Work Behaviour (0.77) have a high CR value (0.7 or higher) and represent a good
reliability of the measures. A Satisfactory reliability is also applied to Self-Efficacy (0.65) with Ethical
Leadership (0.51) and Participative Leadership (0.5) to have a slightly lower reliability from the
threshold suggested by Hair (Hair, et al., 2014).




Table 14: Construct Validity - AVE and C.R.


Variance Error
SUM SUM
Construct Qs Ȝ SUM2 Explained AVE Variance C.R.
(Ȝ 2) (1- Ȝ 2)
(Ȝ 2) (1- Ȝ 2)
Ethical EL5 0.361 0.13 0.87
1.31 0.74 0.37 1.26 0.51
Leadership EL3 0.784 0.61 0.39
Participative PL4 0.639 0.41 0.59
1.32 0.67 0.33 1.33 0.50
Leadership PL5 0.511 0.26 0.74
SE1 0.658 0.43 0.57
Self-Efficacy 1.93 0.97 0.48 1.03 0.65
SE2 0.732 0.54 0.46
Creative CP3 0.749 0.56 0.44
Process CP4 0.633 3.79 0.40 1.28 0.43 0.60 1.72 0.69
Engagement CP2 0.564 0.32 0.68
Innovative IW1 0.602 0.36 0.64
Work IW2 0.753 4.69 0.57 1.59 0.53 0.43 1.41 0.77
Behaviour IW3 0.811 0.66 0.34

The reliability result of the selected model with the 12 variables, if measured in SPSS using Cronbach’s
Alpha based on standardized items, gives a value of 0.739, which is considered acceptable reliability as
indicated previously in section 5.4.1.

Table 15: Cronbach's Alpha reliability of final CFA


Reliability Statistics
Cronbach's Alpha Based
Cronbach's Alpha N of Items
on Standardized Items
0.740 0.739 12

Discriminant Validity denotes how much a construct is different from other constructs. A high
discriminant validity indicates that a construct is distinctive and explains measures that other constructs
do not. Discriminant validity was measured by comparing the average variance-extracted values for any
two constructs with the square estimate correlation between these two constructs. The correct criterion
is that the variance extracted estimates is greater than the squared correlation estimate, thus the latent
construct explains more of the variance in its variable measures (Hair, et al., 2014). Table 16 lists the
CFA correlations and the squared values to be used for discriminant validity checks.

The model’s discriminant validity is checked by looking at the Maximum Shared Variance (MSV) and
the Average Shared Variance (ASV) of each construct and ensuring is not greater than the AVE of each
construct. In Table 17, the MSV and ASV values of each construct are shown. All values are smaller to
the AVE values of the constructs; thus, this test has passed, providing good evidence of discriminant
validity.

Table 16: Constructs Correlations.


Correlation Variance
Description
(Estimate) (Squared)
Ethical Participative
<--> 0.23 0.0529
Leadership Leadership
Ethical
<--> Self-Efficacy 0.269 0.072361
Leadership



Correlation Variance
Description
(Estimate) (Squared)
Creative
Ethical
<--> Process 0.301 0.090601
Leadership
Engagement
Innovative
Ethical
<--> Work 0.505 0.255025
Leadership
Behaviour
Participative
<--> Self-Efficacy 0.004 0.000016
Leadership
Creative
Participative
<--> Process 0.304 0.092416
Leadership
Engagement
Innovative
Participative
<--> Work 0.334 0.111556
Leadership
Behaviour
Creative
Self-
<--> Process 0.436 0.190096
Efficacy
Engagement
Innovative
Self-
<--> Work 0.304 0.092416
Efficacy
Behaviour

Table 17: Correlations Variance - MSV and ASV


Construct MSV ASV
Ethical Leadership 0.255025 0.117722
Participative
Leadership 0.111556 0.064222
Self-Efficacy 0.190096 0.088722
Creative Process
Engagement 0.322624 0.173934
Innovative Work
Behaviour 0.322624 0.195405

Typically, when convergent validity issues exist, it means that the used variables do not correlate well
with each other within their construct; it can be stated that in such a case their construct is not well
explained by its observed variables. In the model examined in this section, the validity is acceptable by
using high loading estimate values. When calculating AVE, it shows validity issues for Ethical and
Participative Leadership, and that denotes that the selected variables of Ethical and Participative
Leadership do not explain very well their parent factor for the population sample surveyed.

When discriminant validity issues exist, it means that the variables correlate better with variables outside
their construct than with the variables within their construct; in such case their construct is better
explained by variables from another construct. There are no discriminant validity issues in the CFA
model examined, and it can be said that the selected variables explain better their parent factor rather
than another latent factor.




6. Analysis and Discussion

The previous section presented the EFA and CFA techniques used to derive the final CFA model, and
investigated the relationships of the variables and the constructs, as well as analysed the validity and
reliability of the final model. The achievement of a valid model initiated the process of the final stages
of SEM.

This section discusses the new model formed and checks for its validity. In addition, the new meanings
of the constructs are discussed. The new model has still kept the same five constructs with the theoretical
model; however, the constructs are different in their essence since some variables have been deleted.
All of the constructs are represented by fewer variables in comparison to the theoretical model.
Notwithstanding the modification of constructs in terms of the variables forming them, the naming of
the constructs has not changed. The reasoning for not renaming the constructs was firstly that the model
shows no discriminant validity issues, which suggests that the variables checked in the final CFA model
are related to their parent construct more than any other construct. In theory, all remaining variables
belong to their theoretical part (i.e. either Ethical Leadership, Participative Leadership, Self-Efficacy,
Creative Process Engagement or Innovative Work Behaviour). Secondly, the continuity of the model
will ease the understanding of the reader and since the logical meaning falls under the same sector, it
was decided to keep the naming of new constructs identical and provide explanations of their new
meaning in the theoretical analysis section.

All the hypotheses created for the new SEM model are therefore compared to the hypotheses from the
theoretical model and represent parts of the theory from the five constructs used in the theoretical model.

The hypotheses previously formed from the theoretical model as presented in the literature review of
this thesis have now been changed. Hence, the hypotheses are redefined and represent the new structural
model derived from the final CFA model. The new hypotheses are listed below and their explanations
are elaborated in the following subsections:

Hypothesis 1: Participative leadership is positively correlated to creative process engagement.


Hypothesis 2: Ethical leadership is positively correlated to creative process engagement.
Hypothesis 3: Self-efficacy is positively correlated to creative process engagement.
Hypothesis 4: Creative process engagement is positively related to innovative work behaviour.

Final SEM Model Testing and Validity Checks


The final SEM model testing was performed to prove that the structural relationship of the remaining
variables and constructs can provide a valid model. Therefore, the next step was a Structure Equation
Modeling test of the final SEM model as presented in Figure 13. The process of deriving the final SEM
model was a result of the entire project work, and the detailed examination of the remaining variables
and relationships. It was carefully chosen in accordance with the evidence gained from the CFA analysis,
taking into account the variables correlations and all other statistical indications. After experimenting
on the structural modeling results, the final model was checked in detail.

The relationships that have been hypothesized among the constructs with regards to the remaining
variables were modelled for further estimation. In this respect, the final SEM model comprises of five
constructs, eleven variables and only four hypotheses as listed previously. The only variable that has
been deleted in comparison to the final CFA model is EL3, as it seemed to have high error variance
which affected the model’s fitness. A final enhancement of the SEM model goodness of fit was achieved
by incorporating a covariance between the error terms 23 and 25.




Figure 13 Final SEM Model with 5 Constructs

The overall model Fit was checked to ensure that the validity was achieved and can be shown in Table
18. It is clear that all the values examined are within the recommended limits. While testing and creating
the model, the process of selecting the most correlated directions led to obtaining the most significant
results for the majority of the examined values. More specifically, the required limit of each value, as
well as the results achieved from this SEM model are presented in Table 18.




Table 18: Summary of Results from the SEM Model


Test Values Result Limits
There is no actual limit, it
depends on model.
Chi-square 62.554
According to theory is supported
if p < 0.05
Degrees of Freedom 41 -
Probability Level 0.017 < 0.05
CMIN/DF 1.526 0-5
Goodness of Fit Index (GFI) 0.940 > 0.90
Comparative Fit Index (CFI) 0.932 > 0.90
Adjusted Goodness-Of-Fit (AGFI) 0.904 > 0.80
Root Mean Square Error of
0.055 < 0.095
Approximation (RMSEA)
*According to ( (Bagozzi & Yi, 1988); (Bentler, 2005); (Byrne, 2009); (Marsh & Hocevar,
1985); (Hair, et al., 2014))

According to the above table, the overall model fit is achieved, since it presents high comparative fit
index, high goodness of fit index, low probability level and low root mean square error of approximation.

To check the new hypotheses that occurred, the Critical Ration (CR), Path Coefficients and P values
were investigated and are presented in Table 19. The CR was calculated by dividing the regression weight
estimates by the estimates of its standard error. This value needed to be higher than 1.96 or lower than
-1.96 proving a two-sided significance at p < 0.05 (Hox & Bechger, 1998). In addition, the Path
Coefficients should not exceed 1 (Hair, et al., 2014).

According to the CR criterion, all the four new Hypotheses can be supported since they are within the
acceptable limits for Critical Ratio being higher than the 1.96 and probability level less than 0.05. Also,
all the path coefficients are positive, and they do not exceed the limit of 1.

Table 19: Examining Hypotheses as per CR Value Criterion


Path CR P
Hypothesis
No. Path Coefficients value > value
acceptance
do not exceed 1 1.96 < 0.05
Ethical Leadership ĺ Creative Process
H1 0.811 2.715 0.007 Supported
Engagement
Participative Leadership ĺ Creative
H2 0.391 2.420 0.016 Supported
Process Engagement
Self-Efficacy ĺ Creative Process
H3 0.434 3.249 0.001 Supported
Engagement
Creative Process Engagement ĺ Innovative
H4 1.000 4.233 >0.001 Supported
Work Behaviour

Comparison of Theoretical SEM Model with Final SEM Model


Some of the questions defined, presented and distributed to the population through the questionnaire
used for this study have been excluded from the final SEM model. Deleting variables when conducting
an analysis does not mean that the entire study is invalid, or that the determined questions were not
related to the constructs that constituted the theoretical model. When conducting SEM analysis, it is
expected that some of the questions may be eliminated as the process of the analysis carries on. The
results are highly related to the surveyed population. Different working environments, existing



management styles and the individual ways of thinking and acting of both the managers and the
subordinates, affect the strength of each question asked. The questions excluded through the construct
of the final SEM model did not present strong estimates and relationships in contrary to those remained,
whereas some of them included a high error variance, which means that they carried more error than
explanation to the theory.

When the Theoretical model was created, the relationships among the constructs were proposed which
led to the theoretical hypotheses’ definitions. After proceeding to the final SEM model definition, some
relationships between the constructs were changed, where some were eliminated. More specifically, the
differences of the relationships/hypotheses between the theoretical model and the final SEM model are
described in Table 20.

Table 20: Status of the Relationships/Hypotheses in the Final SEM Model


Status in the Final SEM
Relationships / Hypotheses
Model
Participative Leadership ĺ Creative Process Engagement Remained
Participative Leadership ĺ Self-Efficacy Removed
Ethical Leadership ĺ Creative Process Engagement Remained
Ethical Leadership ĺ Self-Efficacy Removed
Creative Process Engagement ĺ Innovative Work Behaviour Remained
Self-Efficacy ĺ Innovative Work Behaviour Removed
Self-Efficacy ĺ Creative Process Engagement New relationship added

From the initial model, three relationships/hypotheses were excluded. These relationships were changed
after running EFA and CFA analyses. An indicative examination of the relationships among the
constructs is done through correlations’ estimates occurred from the CFA analysis part (Table 16).

Firstly, Participative Leadership and Self-Efficacy have a significantly low correlation of 0.004.
Therefore, it is decided to exclude such a relationship to the SEM model. This low correlation between
these two constructs also indicates that from the population surveyed the participative leadership style
does not enhance the Self-Efficacy of the employees.

Secondly, another relationship found to present a low significance of 0.23 is that between Ethical
Leadership and Self-Efficacy. The low correlation between ethical leadership and self-efficacy led to
adjusting the new SEM model, and converting the initially defined endogenous construct to an
exogenous latent factor. Self-Efficacy is a virtue which can enhance innovation, but is less correlated or
affected by the examined leadership styles according to the population surveyed. Hence, such a
relationship is also excluded from the Final SEM model.

Finally, even though Self-Efficacy and Innovative Work Behaviour are more significant with a
correlation estimate of 0.304, it was decided to exclude this relationship and define a valid overall final
SEM model. Self-Efficacy proved to be important as an exogenous construct which can affect the ability
of the employee in engaging creativity, as the tests results showed from the population participated. It
can be observed that the relationship between Self-Efficacy and Creative Process Engagement is
stronger with an estimate of 0.436 and this also explained by the final SEM model.

Theoretical Analysis
In this subsection, the modified constructs are analysed and their new meaning is explained. The final
SEM model as shown in Figure 13 produced valid results and positive values in the proposed path
directions. The constructs “Innovative Work Behaviour” and “Creative Process Engagement” remained
as endogenous constructs similarly to the original model. The constructs “Participative Leadership” and
“Ethical Leadership” remained as exogenous constructs in the final model as also shown in the derived
theory. “Self-Efficacy” was converted to an exogenous construct instead of endogenous as firstly shown



in the theoretical model. In order to further understand if the exogenous constructs have a positive
influence on the endogenous, there is a need to examine their loadings. The stronger the loading, the
greater the impact. The meanings are described in the following subsections.

Ethical Leadership
Ethical Leadership positively contributes the Creative Process Engagement as per Hypothesis H1. This
construct has one remaining question “EL5”, from the initial five included in the questionnaire. “EL5”
corresponds to “I do not undermine publicly a subordinate by indicating his/her mistakes. I suggest
solution to that person in private”. Ethical Leadership explained significantly the creative behaviour of
the employees (0.81). By ethical management, the study results show that it is related to the respect of
management towards employees and the self-worth of the employees in the workplace (not undermining
the person in public). The ethical managers or employees should support their colleagues without
damaging their morale and professional image by trying to give them advices privately. For an employee
to get involved in creative methods and thinking, it is important to feel secure within the working
environment. Having a supportive dealing when mistakes are made, enhances the creative process; on
the other hand, when undermined publicly, the morale may produce a reverse outcome on innovative
performance. As initially defined in in Section 2.1.2, the current study intended to also examine if the
Ethical Leadership affects the Self-Efficacy, as stated by the study of Piccolo et al. However, the specific
population that answered the relevant questionnaires proved that there is not a high effect between these
two constructs.

Participative Leadership
Participative Leadership presents a positive effect on Creative Process Engagement as per Hypothesis
H2. This construct remained with two questions “PL4” and “PL5” from the seven included in the created
questionnaire. The first question concerning the Participative Leadership Style, is “PL4”- “I request
team members for alternative solutions when a problem occurs”. The second is “PL5”- “I consult group
input before making a decision”. The participative leadership style has an effect that is well defined. It
includes basically two main characteristics, leading by counting on the employees to deliver alternative
solutions (PL4) and consultation of the employees before decision making (PL5). Essentially,
participative leadership in its basic form, relates to a collaborative decision process between the
employees and their managers in the technological sector. This management style did not provide the
most significant result in terms of enhancing the creative behaviour, but it contributed in providing an
acceptable relationship to creativity (0.39). PL4 relates to collaborative management which may
increase the creativity of employees when searching for alternative solutions in order to achieve their
goal. PL5 relates to the involvement of the team in the decisions taken, which again is a method that
makes the professionals engage in creativity to some extent, prior to providing their consultation for
overcoming a given task. Respondents surveyed showed that they may rely on team decisions to proceed
to further actions. As already stated in Section 2.1.2, the Participative Leadership style used a different
mediating mechanism than prior studies, in order to study the innovative work behaviour, which in this
case was the Creative Process Engagement.

Self-Efficacy
Self-Efficacy positively affects the Creative Process Engagement according to Hypothesis number 3.
From the five questions of the survey, “SE1” and “SE2” are those kept in the final Model. “SE1”- “I
believe I can succeed at almost any endeavour to which I set my mind” clearly shows that for Creative
process engagement to be promoted, a professional needs to strongly believe in his/her abilities.
According to “SE2”- “I am confident that I can perform effectively on many different tasks” the workers
involved with creativity need to be confident that can be multi-tasking. Work efficacy is shown to be
important (0.43) but less significant to self-worth as defined in Ethical Leadership section 6.3.1, proving
that internal relations within a working environment are sensitive matters, especially when they are
linked to innovation. Hence, they need to be handled with extra attention and in a careful manner. The
new formed construct comprising of the two remaining questions proved the theory of this study, which
stated that an employee with high self-efficacy has a positive relationship with the creative process, and
promotes the generation of innovative ideas as referred in Section 2.1.3. The difference from the



theoretical model is that Self-Efficacy is acting as an exogenous construct and is not much related to the
style of management applied. From both perspectives, managers and subordinates, self-efficacy proved
to be a parameter that is important for creativity and innovation, but is less likely to be affected by
participative and ethical leadership.

Creative Process Engagement


Creative Process Engagement contributes in a positive way the Innovative Work Behaviour as per H4.
Additionally, this hypothesis corresponds to H5 created in the theoretical model of theory part. This
construct is now embodied by three questions “CP2”, “CP3” and “CP4” instead of five as firstly defined.
The first question is “CP2” - “I compile large amounts of detailed information in my area of expertise
for future use”. The second question is “CP3” - “I consider diverse sources of information for generating
new ideas” and the third question is “CP4” - “I consult a significant number of alternatives of the same
problem before I choose the final solution”. All three questions indicate that it is of vital importance for
an employee/manager to compile large amounts of alternative information from different sources that
can generate new ideas used to provide solutions and increase innovativeness. The significant positive
influence between Creative Process Engagement and Innovative Work Behaviour can also be justified
through its high loading (1.00). This high loading strongly reveals that an employee shall advance
creativity relevant processes which can result in achieving innovation. Innovation is always achieved as
a result of previously engaged creativity, where creativity doesn’t always results in achieved innovation.
In other words, creativity must exist for innovation to be accomplished. Creative process engagement is
highly correlated with innovative work behaviour in theory, and the model results validate the
background theory. The model could produce the same goodness of fit results by merging these two
meanings in one. Even though creative engagement is a step before innovation, it could be concluded
that creativity engagement is the basis of innovation before the implementation of innovation.

Innovative Work Behaviour


Following the analysis results, the new construct is composed by three remained questions “IW1” - “I
come up with new and practical ideas to improve performance”, “IW2” - “I search out new
technologies, processes, techniques, and/or product ideas” and “IW3” - “I suggest new ways to increase
quality”. The population sample from the technological sector resulted in being active with innovation
and engaging creativity at work. As explained in the previous paragraphs, self-worth, employee
competency, collaborative environment, and creativity explained by analytical approach towards
providing solutions (either by overlooking different methods of increasing performance or by
investigating more possible mechanisms for resolving a known issue), are factors that affect the
innovative work behaviour of employees and managers. The population examined involved 24%
management from the technological sector and 39% engineering and design, where innovation is a more
related topic to the day-to-day challenges than to other types of labour force (such as a nurse or a
postman for example). Therefore, it can be concluded that the population participated in this study seem
to embrace an Innovative Work Behaviour culture. In addition, as already hypothesized when defining
the theoretical model, since the construct “Creative Process Engagement” acts as a mediator between
the exogenous constructs and the innovative work behaviour, it can be considered that all the constructs
positively affect the Innovative Work Behaviour.




7. Conclusion

This chapter acknowledges the research problem expressed for this study, and elaborates on the diverse
findings and final outcomes presented. It also describes the implications of the study, its limitations and
suggests future studies that can follow.

Conclusions
This research study examined whether ethical and participative leadership can have a positive impact
on innovative work behaviour. Ethical leadership combined with the participative leadership style are
both classified as positive leadership, and combined together created a new unique aspect. There was
no research found that empirically proves the concept studied and presented in this thesis. Most
technological companies focus on innovation as it can be a paramount element for growth, for
competiveness, for meeting customer needs and even for attracting the best talent. This study focused
on people’s psychological empowerment, and examined how self-efficacy and creative process
engagement could act as mediator paths to innovative behaviour as a result of the combined ethical and
participative leadership.

In order to reach this study’s objective, a theoretical framework was developed based on the relevant
literature. The literature studied led to the creation of the theoretical framework. Based on the derived
theoretical framework, a representative model was created, that initiated the process of determining the
data required via a developed questionnaire. When the data was collected, the model was finally tested
and an empirical analysis was conducted. The thesis aimed to investigate the relationship between
leadership and innovation, and was specifically concentrated to answer the research problem: “to
examine how ethical and participative leadership affect the self-efficacy and creative process
engagement of the employees in order to provide effective innovative work behaviour.”

The findings of this research study, prompt to conclude that ethical and participative leadership will
indeed enhance creativity and innovation to some extent. However, the findings also demonstrate that
ethical and participative leadership do not affect employee’s self-efficacy significantly. Self-efficacy
however proved to be linked positively to work creativity. The outcome of the model is a result of an
empirical setting represented by the theoretical idea of this study. The final SEM model was defined
after the examination of findings and by selecting meaningful correlations between constructs. It was of
great importance to check how each construct explains the next construct, and give an explanation to
the overall model. The definition of the model also took into consideration the literature review and
previous related studies which contributed to the definition of the theoretical model. It was not intended
to only create a statistically very strong new model and destroy the theory behind this analysis. Thus,
there is room for some statistical improvements as presented in the limitations / future work sections.

Management methods and leadership styles can be proved to be complex to examine and subjective.
The industries that the survey was distributed as well as the locations play another role for the
findings/results examined. The population examined, predominantly belonged to the technological
sectors, in particular from the Oil and Gas and Shipping industries. One quarter of the population were
managers, and 40% were from engineering, design and research and development. Innovation is more
relative to technology-related environments when managing technological services or designs and
manufacturing of technological products. Leadership in combination with the psychological aspect of
professionals such as self-worth, self-belief and team spirit feeling, proved to positively influence
creativity and innovation in accordance with the sample tested.

Ethical Leadership produced weaker results than expected in terms of validity, and the questionnaire
did not explain the construct as expected. It has to be noted that these results may differ when examining
a different sample population from a different industry. In the Engineering sector, ethics is usually not
the issue, so by not achieving the expected statistical results does not mean that people are not ethical.
However, there was still an ethical management aspect in the overall model that showed a positive
impact on the creativity engagement of the employees.



Participative leadership in correlation to self-efficacy proved to be insignificant for the sample


population examined in this research. A reason for this result could be that the majority of the employees
surveyed do not experience a participative leadership style, and they experience an authoritarian
leadership style; for example, they are used to either give an order (managers) or follow orders
(subordinates), then self-efficacy may not be supported among the population, as the freedom of choice
at work is not a given. It does not mean necessarily that the participative leadership theory cannot
influence positively the self-efficacy in any sample population, but it is likely not to influence self-
efficacy in a rigid environment with less flexibility.

Implications
The study was able to empirically and quantitatively prove the relationship of ethical and participative
leadership on work innovation. The findings may offer practical implications for leaders aiming to
improve innovation at work. Leadership teams should vigorously apply ethical and participative
leadership style to be able to engage employees into creative thinking, which effectively leads them to
innovative work behaviour. Managers should look into enhancing creativity if they aim to improve their
organizations’ innovation performance. Moreover, leadership should lead by example and not
undermine subordinates in public; this research showed that not undermining subordinates enhances
creativity. Another strong aspect demonstrated was the inclusion of team’s opinion in decision making,
and the existence of a collaborative environment can support the creative process engagement which
can promote innovative ideas.

Employees that feel confident in achieving challenging tasks and also that can multi-task have a positive
relationship to creative thinking. As explained earlier, creative process engagement is a step before
innovation, as innovation deals with the practical implementation of new ideas and products. Self-
efficacy of employees is a positive factor for enhancing innovation as creativity behaviour is highly
correlated to innovativeness.

To increase innovation, leadership should be respectful, collaborative, consult ideas for solutions from
the team members, and work towards improving teams’ self-belief; by succeeding so, the teams could
believe in achieving perplexing goals, improve creativeness and increase the possibilities in achieving
innovation.

Limitations
Undoubtedly, a clear limitation of the relevant study is the population participated. The respondents are
mostly professionals coming from companies from the technological sector. Thus, this sample is not
representative of the larger population coming from a broad network of different sectors. In addition,
the framework created in order to evaluate the Innovative Work Behaviour might not be the most
illustrative comparatively with the population that the relevant questionnaire is spread. Another
limitation that should be considered is the size of the sample. Even if it is acceptable for such a study, it
could be higher. Finally, the way that the questions are defined through the questionnaire and the Likert
scale used to gather the answers can hide the risk that the respondents have different perceptions on the
scales, and this may had an effect on the results of the this study.

Further Work
According to the limitations presented above, several suggestions for future research are given in this
section. Firstly, the relevant study has selected to gather respondents from the technological sector in
order to examine innovative behaviour, since innovation is highly connected with this business area.
However, professionals from different sectors could be sampled and checked on innovativeness in
relations to the leadership styles examined in this study.

Additionally, a future study may modify some dimensions of the theoretical framework to better argue
the research question in relation with the population sampled, resulting in the creation of a new
theoretical model.




For further work, a balance from the number of the participants coming from different countries is highly
recommended. Different cultures and mentalities or perceptions from employees coming from other
geographical locations from the same sector could give an interesting perspective following this study.




8. Bibliography

Akram, T., Lei, S. & Haider, M., 2016. "The impact of relational leadership on employee innovative
work behavior in IT industry of China". Arab Economic and Business Journal, 11(2), pp. 153-161.

Archibugi, D., Filippetti, A. & Frenz, M., 2013. "Economic crisis and innovation: Is destruction
prevailing over accumulation?". Research Policy, 42(2), pp. 303-314.

Bagozzi, R. & Yi, Y., 1988. On the Evaluation of Structural Evaluation Models. Journal of the
Academy of Marketing Science, 16(1), pp. 74-94.

Bandura, A., 1997. Self – efficacy: The Exercise of Control. New York: W. H. Freeman.

Bandura, A., 2000. "Exercise of Human Agency Through Collective Efficacy". Current Directions in
Psychological Service, 9(3), pp. 75-78.

Bentler, P., 2005. EQS 6 Structural equation program manual. Encino, CA: Multivariate.

Brown, M. E., Trevin˜o, L. K. & Harrison, D. A., 2005. "Ethical leadership: A social learning
perspective for construct development and testing". Organizational Behavior and Human Decision
Processes, 97(2), p. 117–134.

Brown, M. & Trevin˜o, L., 2006. "Ethical leadership: A review and future directions". Leadership
Quarterly, 17(6), pp. 595-616.

Bryman, A., 2012. Social Research Methods. 4 ed. Oxford: Oxford University Press.

Bryman, A. & Bell, E., 2007. Business Research Methods. USA: Oxford University Press.

Byrne, B., 2009. Structural Equation Modeling With AMOS: Basic Concepts, Applications, and
Programming (Multivariate Applications). (New York: Routledge).

C., & Baumgartner, H., 1995. Beurteilung von Kausalmodelen. Marketing, 17(3), pp. 162-176.

De Hoogh, A. H. B. & Den Hartog, D. N., 2008. "Ethical and despotic leadership, relationships with
leader’s social responsibility, top management team effectiveness and subordinates’ optimism: A
multi-method study". Leadership Quarterly, 19(3), pp. 297-311.

De Jong, J. P. & Den Hartog, D. N., 2007. "How leaders influence employees’ innovative behaviour".
European Journal of Innovation Management, 10(1), pp. 41-46.

De Jong, J. P. & Den Hartog, D. N., 2008. "Innovative work behavior: Measurement and validation".
EIM Business and Policy Research, pp. 1-27.

Delgadová, E., Gullerová, M. & Ivanová, E., 2017. "Recruitment and selection processes in Slovak
enterprises and multinational corporations". International Journal of Organizational Leadership, 6(2),
p. 211.

Donate, M. J., Sánchez de Pablo & Jesús D, 2015. "The role of knowledge-oriented leadership in
knowledge management practices and innovation". Journal of Business Research, 68(2), pp. 360-370.

DuBrin, A., 2013. Principles of Leadership. 7th ed. Boston: South-Western Cengage Learning.

Fatima, T., Majeed, M. & Saeed, I., 2017. "Does Participative Leadership Promote Innovative Work
Behavior: The Moderated Mediation Model". Business & Economic Review, 9(4), pp. 139-156.




Fornell, C. & Larcher, D., 1981. Evaluating structural equation models with unobservable variables
and measurement error. Journal of Marketing Research, 18, pp. 39-50.

George , D. & Malley, M., 2010. SPSS for Windows Step by Step: A Simple Guide and Reference,17.0
update. 10 ed. Boston: Pearson.

George, D. & Malley, P., 2003. SPSS for Windows step by step: A simple guide and reference.11.0
update. 4 ed. Boston: Allyn & Bacon.

Ghauri, P. & Grönhaug, K., 2010. Research Methods in Busines Studies. 4 ed. Harlow: Pearson
Education.

Gyimah, S., 2001. Missing data in quantitative social research.. PSC Discussion Papers, 15(14), 1.

Hair, J. et al., 2007. Multivariate data analysis. Delhi: Dorling Kindersley (India) Pvt. Ltd Homburg.

Hair, J. F., Black, W. C., Babin, B. J. & Anderson, R. E., 2014. Multivariate Data Analysis. 7 ed.
Harlow: Pearson Education Limited.

Hennessey, B. A. & Amabile, T. M., 2010. Creativity. Annual Review of Psychology, Volume 61, p.
569–598.

Hox, J. & Bechger, T., 1998. An introduction to structural equation modeling. Family Science Review,
11, pp. 354-373.

Janssen, O., 2003. "Innovative behaviour and job involvement at the price of conflict and less
satisfaction with co-workers". Journal of occupational and organizational psychology, 76(3), p. 347–
364.

Kahai, S. S., Sosik, J. J. & Avolio, B. J., 1997. "Effects of Leadership Style and Problem Structure on
Work Group Process and Outcomes in an Electronic Meeting System Environment". Personnel
Psychology, Volume 50, pp. 121-146.

Kalshoven, K., Den Hartog, D. & De Hoogh, A., 2011. Kalshoven, K., Den Hartog, D. N., "Ethical
leadership at work (ELW): Development and validation of a multidimensional measure". The
Leadership Quarterly, 22(1), pp. 51-69.

Kline, B. R., 2010. Principles And Practice Of Structural Equation Modeling. 3 ed. New York: The
Guilford Press.

Kwahk, K. Y. & Lee, J. N., 2008. Kwahk, K. "The role of readiness for change in ERP
implementation: Theoretical bases and empirical validation". Information & Management, 45(7), pp.
474-481.

Madrid, H. P. et al., 2014. Madrid, H. P., Patterson, M. G., Birdi, K. S"The role of weekly highǦ
activated positive mood, context, and personality in innovative work behavior: A multilevel and
interactional model". Journal of Organizational Behavior, 35(2), pp. 234-256.

Marsh, H. & Hocevar, D., 1985. Application of confirmatory factor analysis to the study of self-
concept: First- and higher- order factor models and their invariance across groups. Psychological
Bulletin. 97, pp. 562-582.

Mayer, D. M. et al., 2009. "How low does ethical leadership flow? Test of a trickle-down model".
Organizational Behavior and Human Decision Processes, 108(1), pp. 1-13.




Miao, Q., Newman, A. & Huang, X., 2014. "The impact of participative leadership on job
performance and organizational citizenship behavior: distinguishing between the mediating effects of
affective and cognitive trust". The International Journal of Human Resource Management, vol. 25(no.
20), pp. pp. 2796-2810.

Miller , R. & Brewer, J., 2003. The A-Z of Social Research: A Dictionary Of Key Social Science
Concepts. u.o.:Sage. s.l.:s.n.

Mumford, M. D., 2000. "Managing creative people: Strategies and tactics for innovation". Human
Resource Management Review, Volume 10, pp. 313-351.

Piccolo, R., Greenbaum, R., Den Hartog, D. & Folger, R., 2010. "The relationship between ethical
leadership and core job characteristics". Journal of Organizational Behavior, 31(2–3), p. 259–278.

Reiter-Palmon, R. & Illies, J. J., 2004. "Leadership and creativity: Understanding leadership from a
creative problem solving perspective". Leadership Quarterly, Volume 15, pp. 55-77.

Sauer, S., 2011. "Taking the reins: the effects of new leader status and leadership style on team
performance". The Journal of applied psychology, 96(3), pp. 574-587.

Shalley, C. E. & Gilson, L. L., 2004. "What leaders need to know: A review of social and contextual
factors that can foster or hinder creativity". Leadership Quarterly, Volume 15, pp. 33-53.

Shalley, C., Zhou, J. & Oldman, G. R., 2004. "The effects of personal and contextual characteristics
on creativity: Where should we go from here?". Journal of Management, Volume 30, pp. 933-958.

Shanker, R., Bhanugopan, R., Van der Heijen, B. & Farrell, M., 2017. "Organizational climate for
innovation and organizational performance: The mediating effect of innovative work behavior".
Journal of Behavior, Volume 100, pp. 67-77.

Somech, A., 2006. "The effects of leadership style and team process on performance and innovation in
functionally heterogenous teams". Journal of management, 32(1), pp. 132-157.

Somech, A. & Wenderow, M., 2006. "The Impact of Participative and Directive Leadership on
Teachers' Performance: The Intervening Effects of Job Structuring, Decision Domain, and Leader-
Member Exchange". Educational Administration Quarterly, 42(5), pp. 746-772.

Strom, D. L., Sears, K. L. & Kelly, K. M., 2014. ”Work engagement: The roles of organizational
justice and leadership style in predicting engagement among employees". Journal of Leadership &
Organizational Studies, 21(1), pp. 71-82.

T.J., R. & W., & Schindler , A., 2001. Customer retention, loyalty and satisfaction in the german
cellular telecommunications market. Telecommunications Policy, 25(4), pp. 249-269.

Thomas, T., Schermerhorn, J. & Dienhart, J., 2004. "Strategic leadership of ethical behavior in
business". The Academy of Management Executive, 18(2), pp. 56-66.

Trevin˜o, L. K. & Brown, M. E., 2004. "Managing to be ethical: Debunking five business ethics
myths". Academy of Management Executive, 18(2), pp. 69-81.

Trivisonno, M. & Barling, J., 2016. "Organizational leadership and employee commitment".
Handbook of Employee Commitment, p. 305.

Yousef, D. A., 2000. "Organizational Commitment: A Mediator of the Relationships of Leadership


Behavior with Job Satisfaction and Performance in a Non-Western Country". Journal of Managerial
Psychology, Volume 15, pp. 6-28.




Zhang, X. & Bartol, K. M., 2010. "Linking empowering leadership and employee creativity: The
influence of phsycological empowerment, intrinsic motivation, and creative process engagement". The
Academy of Management Journal, 53(1), pp. 107-128.




9. Appendices

Appendix A

MBA-THESIS questionnaire
This questionnaire intends to examine the impact of ethical and participative leadership on innovative work
behaviour, as well as role of self-efficacy and creative process engagement, for the scope of this MBA thesis. All
the collected data concerning personal information will remain confidential.

DEMOGRAPHIC QUESTIONS

Female Male
Gender
տ տ

18-30 31-40 41-50 >50


Age Group
տ տ տ տ
3/4-yr Master
High-school 2-yr College PhD Level
Educational College level
Background
տ տ տ տ տ
Finance /
Management Administration Engineering R&D Other
Position Accounting
տ տ տ տ տ տ
Location of
employment

ETHICAL LEADERSHIP
Q Strongly Strongly
Statement Agree Neutral Disagree
No. Agree Disagree
EL1 I am honest and can be trusted. տ տ տ տ տ
I set an example of dedication (too strong) for the
EL2 տ տ տ տ տ
organization.
I provide fair and objective evaluation by
EL3 assessing people performance based on clear and տ տ տ տ տ
publicly known objectives.
I credit subordinate for an innovative idea publicly
EL4 տ տ տ տ տ
and emphatically.
I do not undermine publicly a subordinate by
EL5 indicating his/her mistakes. I suggest solution to տ տ տ տ տ
that person in private.

PARTICIPATIVE LEADERSHIP
Q Strongly Strongly
Statement Agree Neutral Disagree
No. Agree Disagree

PL1 I am consulted regarding important changes. տ տ տ տ տ

I have considerable independence and freedom to


PL2 տ տ տ տ տ
grab opportunities.




PARTICIPATIVE LEADERSHIP
Q Strongly Strongly
Statement Agree Neutral Disagree
No. Agree Disagree

I believe that most creative ideas are likely to


PL3 come from the group members rather than from տ տ տ տ տ
the manager.

I request team members for alternative solutions


PL4 տ տ տ տ տ
when a problem occurs.

PL5 I consult group input before making a decision. տ տ տ տ տ

I inform frequently the group of any information


PL6 տ տ տ տ տ
that could affect their work.

I believe that much more can be accomplished by


PL7 a team than by the same number of people տ տ տ տ տ
working alone.

CREATIVE PROCESS ENGAGEMENT


Q Strongly Strongly
Statement Agree Neutral Disagree
No. Agree Disagree
I think about the problem from multiple
CP1
perspectives. տ տ տ տ տ

I compile large amounts of detailed information


CP2 տ տ տ տ տ
in my area of expertise for future use.
I consider diverse sources of information for
CP3 տ տ տ տ տ
generating new ideas.
I consult a significant number of alternatives of
CP4 the same problem before I choose the final տ տ տ տ տ
solution.
I try to come up with solutions that move away
CP5 տ տ տ տ տ
from established ways of doing things.

SELF-EFFICACY
Q Strongly Strongly
Statement Agree Neutral Disagree
No. Agree Disagree
I believe I can succeed at almost any endeavour to
SE1 տ տ տ տ տ
which I set my mind.
I am confident that I can perform effectively on
SE2 տ տ տ տ տ
many different tasks.
SE3 The work I do is meaningful to me. տ տ տ տ տ
I have significant autonomy in determining how I
SE4 տ տ տ տ տ
do my job.

SE5 I have mastered the skills necessary for my job. տ տ տ տ տ

INNOVATIVE WORK BEHAVIOR


Q Strongly Strongly
Statement Agree Neutral Disagree
No. Agree Disagree
I come up with new and practical ideas to
IW1 տ տ տ տ տ
improve performance.




INNOVATIVE WORK BEHAVIOR


Q Strongly Strongly
Statement Agree Neutral Disagree
No. Agree Disagree
I search out new technologies, processes,
IW2 տ տ տ տ տ
techniques, and/or product ideas.

IW3 I suggest new ways to increase quality. տ տ տ տ տ

I look systematically for solutions when problems


IW4 տ տ տ տ տ
occur

I import solutions from other industries than the


IW5 տ տ տ տ տ
one I am working in currently.

Additional Comments:
(please feel free to add any comment that you consider important for the accomplishment of this study)

To maximize the number of the gathered responses and assure the validity of the results for the current study, it will be of high
importance to forward the survey to other suitable candidates.

Thank you for your time & attention!

Kyriaki Anastasopoulou & Dimitrios Angelis



You might also like