Ethical leadership and deviant employee behaviours:
Moderated mediation effects on
unethical pro-organizational behaviour and pro-social rule breaking
Master Thesis
Kim den Uijl
6087957
Date of submission: June 29, 2015
Version: Final
Qualification: MSc Business Administration – Leadership and Management track
Institution: Amsterdam Business School, University of Amsterdam
First supervisor: Prof. Dr. Deanne den Hartog
This document is written by student Kim den Uijl who declares to take full responsibility for
the contents of this document.
I declare that the text and the work presented in this document is original and that no sources
other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of
Table of contents
1. Abstract ... 5
2. Introduction ... 6
3. Literature review ... 11
3.1 Ethical leadership ... 11
3.2 Ethical leadership and employee attachment ... 12
3.3 The mediating role of employee attachment ... 15
3.3.1 Mediation to unethical pro-organizational behaviour (UPB) ... 15
3.3.2 Mediation to pro-social rule breaking (PSRB) ... 19
3.4 The moderating role of ethical climate ... 21
3.4.1 Moderation between employee attachment and UPB ... 22
3.4.2 Moderation between employee attachment and PSRB ... 23
4. Research method ... 25
4.1 Research approach and design ... 25
4.2 Measures ... 25
4.3 Data analysis ... 27
4.4 Sample ... 27
5. Results ... 28
5.1 Data screening and preliminary analyses ... 28
5.1.1 Missing values and outliers ... 28
5.1.2 Reliability tests ... 29 5.1.3 Normality tests ... 30 5.1.4 Assumption tests ... 31 5.2 Descriptive statistics ... 33 5.3 Regression analyses ... 34 5.3.1 UPB ... 34 5.3.2 PSRB ... 36 6. Discussion ... 38 6.1 Discussion of results ... 38
6.1.1 Theoretical implications and future research suggestions ... 38
6.2 Strengths and limitations ... 44
6.3 Suggestions for future research ... 46
7. Conclusion ... 49
8. References ... 50
Appendices ... 62
Appendix A: Scale development ... 62
Appendix B: Histograms and Q-Q plots ... 68
Appendix C: Scatterplots of the residual terms ... 70
Appendix D: Histograms and P-P plots of the residuals ... 71
Appendix E: Scatterplots independent and dependent variables ... 74
List of figures
1 – Conceptual model ... 9List of tables
1 – Values, standard errors, and Z-scores for skewness and kurtosis ... 302 – Correlation matrix ... 34
3 – Results of the regression analysis for moderated mediation to UPB ... 35
1. Abstract
This study examines the relationships between ethical leadership and two forms of
employee pro-organizational, but deviant, behaviours: unethical pro-organizational behaviour
(UPB) and pro-social rule breaking (PSRB). Employee attachment is proposed as mediating,
and ethical climate as moderating these relationships. The data were collected by asking 204
employees to answer a questionnaire. Results showed that ethical leadership had no
significant effect on UPB through employee attachment, and ethical law and codes climate
did not moderate this relationship. For PSRB, a significant negative relationship was found
from ethical leadership through employee attachment under a low and moderate ethical rules
2. Introduction
Over the past decades, there has been a growing concern in society about
organizational ethics. Unethical practices that have taken place within companies such as
Enron and Tyco International have continuously made headlines in newspapers (Forbes,
2013). As ethics are among the most important matters in the long-term sustainability of any
business, questions of ethics have also become increasingly urgent for companies themselves
(Fombrun & Foss, 2004). The European commission showed that ethics in business can have
a positive impact on competitiveness (European Commission, 2009). On the other hand,
unethical acts in organizations can lead to a damaged reputation, and as a result, a loss of
market value (Karpoff, Lee, & Martin, 2008). Organizations now need more ethical leaders
who also encourage ethical behaviour in followers (Stouten, Van Dijke & De Cremer, 2012).
A response to this trend has been the emergence of the concept of ethical leadership in
the scientific literature. This type of leadership is defined as “the demonstration of
normatively appropriate conduct through personal actions and interpersonal relationships, and
the promotion of such conduct to followers through two-way communication, reinforcement,
and decision-making” (Brown, Treviño & Harrison, 2005, p. 120). Thus, ethical leaders
function as role models for ethical behaviour to their followers (Brown & Treviño, 2006).
Ethical leadership has been shown to reduce unethical behaviour at the workplace,
such as deviant workplace behaviour, misconduct, and bullying (Avey, Palanski &
Walumbwa, 2011; Mayer, Kuenzi & Greenbaum, 2010a; Stouten et al., 2010). It is also
associated with a whole range of other, positive, employee outcomes, such as organizational
citizenship behaviour (Avey et al., 2011; DeConinck, 2014; Kacmar, Bachrach, Harris &
Zivnuska, 2011; Ogunfowora, 2014), job satisfaction (Neubert, Carlson, Kacmar, Roberts &
Chonko, 2009; Ogunfowora, 2014), organizational commitment (Neubert et al., 2009; Zhu,
2011), work engagement (Cheng, Chang, Kuo & Cheung, 2014; Den Hartog & Belschak,
2012), work effort (Kacmar, Carlson & Harris, 2013; Toor & Ofori, 2009), helping (Kacmar
et al., 2013), self-efficacy (Walumbwa et al., 2011), voice (Cheng et al., 2014; Walumbwa &
Schaubroeck, 2009), creativity (Chughtai, 2014), and reduced turnover intentions (Elçi, Şener,
Aksoy & Alpkan, 2012).
Ethical leadership is a relatively new field (Brown et al., 2005) and most research has
focused on positive outcomes. However, recently, some researchers have begun to focus their
attention on possible boundary conditions or even negative outcomes of ethical leadership.
For example, while ethical leadership has a positive indirect effect on employee well-being
through job satisfaction, it has a negative direct effect on employee well-being (Yang, 2014).
The cause may be that when leaders show a higher ethical conduct than their employees, this
can make employees feel pressured, and therefore reduce their well-being. Another study has
shown that the relationship between ethical leadership and organizational citizenship
behaviours (OCBs) is positive but curvilinear (Stouten, Van Dijke, Mayer, De Cremer &
Euwema, 2013). This suggests that there is a positive relationship between ethical leadership
and employee OCB, but beyond an optimal level of ethical leadership, OCB starts to
diminish. Thus, the presumption in most of the previous research and literature, that ethical
leadership always has beneficial effects, is not necessarily true. Also looking at curvilinear
relationships, Miao, Newman, Yu and Xu (2013), argue that moderate levels of ethical
leadership can even lead to unethical pro-organizational behaviour (UPB) in a Chinese
context. These authors reason that ethical leadership will lead to a positive social exchange
relationship between employees and their supervisors, in which employees are willing to
engage in pro-organizational behaviour such as UPB, even if it is unethical. To conclude,
there are indications in the literature that ethical leadership does not yield only positive
possibly ‘darker’ side of ethical leadership by investigating the relationship between ethical leadership and employees’ unethical pro-organizational behaviours (UPB) and pro-social rule
breaking (PSRB). The research question is as follows:
“How are ethical leadership and employee unethical pro-organizational behaviours and pro-social rule breaking related?”
These two behaviours are chosen as possible ‘darker’ outcomes, because they are both
pro-organizational (or pro-social), in the sense that the organization is believed to benefit from
them, and are thus different from other unethical behaviours which are done out of
self-interest, such as stealing from the organization. Although it is clear that ethical leadership can
reduce self-interested forms of deviance (Avey et al., 2011; Mayer et al., 2010a; Stouten et al.,
2010), it can nevertheless relate positively to these two pro-organizational forms of deviance,
because these are conducted with the intention to benefit the organization. Ethical leadership
enhances employee attachment, which leads to engagement in these pro-organizational
behaviours because of a strengthened identification and social exchange relationship with the
organization (Kalshoven, Den Hartog & De Hoogh, 2011; Kalshoven, Van Dijk & Boon,
2013; O’Reilly & Chatman, 1986). The study presented in this thesis thus proposes a positive
relationship between ethical leadership and UPB and PSRB that is mediated by employee
attachment. Employee attachment is believed to mediate this relationship because ethical
leadership has a positive effect on attachment, as will be discussed in the theoretical
framework, and because attachment has positive relationships with pro-social behaviours in
general (O’Reilly & Chatman, 1986). In addition, because an ethical climate is believed to
have a reducing effect on deviant behaviours, it is proposed that ethical climate acts as a
moderator, in the sense that the relationship between employee attachment and UPB and
PSRB will be less strong for higher levels of ethical climate and strongest at low levels of
–
–
Figure 1. Conceptual model. Ethical leadership influences employee unethical pro-organizational behaviour and
pro-social rule breaking through employee attachment, moderated by ethical climate. It is further theorized, but not tested, that ethical leadership has a positive relationship with ethical climate.
This study will add to the current knowledge about ethical leadership in multiple
ways: first, it will examine the relationship between ethical leadership and unethical
pro-organizational behaviour (UPB) in a Western society, as a complement to the Chinese context
of Miao et al. (2013). Second, this study will look at the relationship between ethical
leadership and pro-social rule breaking (PSRB), as this is in many ways a similar concept to
UPB. The difference is that in PSRB explicit organizational rules are broken, while in UPB
the norms of society are broken (Ilie, 2012). Thus, the aim is to extend previous work by
examining whether ethical leadership has a similar or a different effect on these two employee
behaviours. Third, this study answers Brown et al.’s (2005) and Brown and Mitchell’s (2010) call for more research about ethical leadership’s influence on unethical behaviours, and the role of employee identification in this process.
It is important to investigate ethical leadership’s effect on these pro-social deviant behaviours because, although they are carried out with good intentions, they can have
negative effects for organizations. UPB can cause silence or a misrepresentation of the truth
about failures, errors, and misconduct (Knoll & Van Dick, 2013; Umphress & Bingham,
+ + – – + + +
Ethical leadership Employee attachment
Ethical climate
law and codes
Ethical climate
rules
Unethical pro-organizational behaviour (UPB)
Pro-social rule breaking (PSRB)
2011; Umphress, Bingham & Mitchell, 2010). This could lead to financial and legal problems
for the organization, such as loss of customers or corporate scandals (Ilie, 2012). When the
organization in question is a societal institution such as the police or health care service,
silence or commission can even damage the larger society (Knoll & Van Dick, 2013;
Umphress et al. 2010). PSRB can also have negative effects, such as unnecessary costs for the
3. Literature Review
3.1 Ethical Leadership
Ethical leadership can be viewed from a normative or prescriptive standpoint, i.e.
looking at how leaders should ideally behave, or from a social scientific or descriptive
standpoint, which entails describing the perception, antecedents and outcomes of ethical
leadership (Brown & Mitchell, 2010; Brown & Treviño, 2006). The social scientific literature
on ethical leadership is still fragmented, but growing (Avey et al., 2011). This paper will thus
focus on expanding the social scientific knowledge of ethical leadership.
Treviño, Hartman and Brown (2000) were among the first researchers to examine
from a descriptive standpoint what ethical leadership entails. They found that ethical leaders
are perceived to score high on two different dimensions: being a moral person and being a
moral manager. The former refers to the ethicality of the person, which shows itself in traits,
behaviours, and decision-making. The latter refers to leaders’ proactive efforts to influence
followers’ ethicality, by role modelling, using rewards and discipline, and communicating
about ethics and values. Brown et al. (2005) consequently developed a measure of ethical
leadership which incorporated both of these dimensions. Their widely used definition of ethical leadership is “the demonstration of normatively appropriate conduct through personal actions and interpersonal relationships, and the promotion of such conduct to followers
through two-way communication, reinforcement, and decision-making” (Brown et al., 2005,
p. 120). Thus, ethical leadership does not concern the intent or motivation of the leader, but
only the behaviour that is shown (Stouten et al., 2012).
Most studies on ethical leadership make use of the one-dimensional scale of Brown et
al. (2005). However, some authors have found that ethical leadership consists of multiple
components: morality and fairness, role clarification, and power sharing, and use a
Kalshoven et al., 2011). While these are different components of ethical leadership, the
correlations among them are still high. However, in line with most of the previous research, this study will make use of Brown et al.’s (2005) scale.
3.2 Ethical Leadership and Employee Attachment
The moral person component of ethical leadership (Treviño et al., 2000) entails being
perceived as honest and trustworthy. Indeed, empirical research has shown that ethical
leadership is positively related to consideration, honesty and trust (Brown et al., 2005).
Trusting behaviours from the leader have a ‘signalling’ effect towards employees, showing
them that they are respected and valued (Tyler, 1997). This generally evokes employee
attachment, known as “the [psychological] bond between the employees and their
organizations” (Riketta & Van Dick, 2005, p. 491). Employee attachment manifests in two
forms, namely identification with and affective commitment towards the organization (Meyer,
Becker & Van Dick, 2006; Riketta & Van Dick, 2005).
The first form of employee attachment, organizational identification, is a form of
social or group identification (Ashforth & Mael, 1989). This means that a person perceives
him- or herself as being one with a group, or as belonging to a particular group. Thus,
organizational identification is the perception of psychological oneness with the organization
to which a person belongs. It shows the extent to which a person derives his or her identity
from the organization (Mael & Ashforth, 1992). An individual with a high level of
organizational identification will want to help his or her organization move forward, because successes and failures from the organization become internalized. The organization’s failures are felt as personal failures, and the organization’s successes are felt as personal successes (Mael & Ashforth, 1992).
The second form of employee attachment, closely related to the first one, is
organizational commitment (Meyer et al., 2006; Riketta & Van Dick, 2005). This is defined as “the relative strength of an individual’s identification with and involvement in a particular organization” (Mowday, Porter & Steers, 1982, p. 226). Allen and Meyer (1990) proposed a three-component view of organizational commitment, consisting of affective, continuance,
and normative commitment. Continuance commitment is commitment as a result of perceived
costs of leaving the organization, while normative commitment results from the feeling of
obligation to stay with an organization. Affective commitment involves the emotional attachment to the organization: it is about a person’s identification with, involvement in, and enjoyment of membership in the organization. Out of the three commitment types, this study
will focus on affective commitment, because this component has the strongest association
with organization-relevant employee behaviour (Meyer, Stanley, Herscovitch & Topolnytsky,
2002), which is the focus of this study.
It is clear that there is a conceptual overlap between organizational identification and
affective commitment. However, it should be noted that some researchers argue that these are
two separate and empirically distinct constructs, with different antecedents and outcomes, and
should be treated as such (Gautam, Van Dick & Wagner, 2004; Riketta, 2005; Van
Knippenberg & Sleebos, 2006). For this study, we follow Riketta and Van Dick’s (2005)
approach, who argue that employee attachment can be used as an ‘umbrella term’ for both
concepts. In their meta-analysis, these authors found no difference whether commitment or
identification was used as a conceptualization of attachment. Thus, as noted, we
operationalize attachment as affective commitment and organizational identification.
The relationship between the leader and the follower influences not only attachment to
the leader, but also the broader attachment to the organization, because the relationship
follower (Sluss & Ashforth, 2008). Therefore, leader behaviours should have a large impact
on how the organization as a whole is perceived.
As stated before, ethical leaders enhance employee attachment because they are a
moral person: fair, honest and trustworthy. Therefore, they treat employees in a fair and
caring way (Brown & Treviño, 2006). The resulting feelings for employees of being valued
and respected (Tyler, 1997) have a positive impact on employee attachment because they lead
to positive attitudes in employees (Brown & Treviño, 2006; Zhu et al., 2004). Moreover, the
fairness of ethical leaders in particular, for example in decision making, enhances procedural
justice perceptions among followers (Folger & Konovsky, 1989). Procedural justice has to do
with the procedures in decision making, i.e. whether a fair, accurate, and unbiased decision is
made (Kernan & Hanges, 2002). This procedural justice in turn is a predictor of job
satisfaction, reduced turnover intentions, increased trust in management, but also
organizational commitment (Kernan & Hanges, 2002). Others studies confirm the idea that
procedural justice affects organizational commitment in a positive way (Aryee, Budhwar &
Chen, 2002). Moreover, another study has shown that in a similar way, organizational justice
mediates the relationship between authentic leadership and organizational identification
(García-Guiu, Molero & Moriano, 2015).
Another one of the three dimensions of ethical leadership is the sharing of power (De
Hoogh & Den Hartog, 2008). Because employee voice and input have positive effects on
procedural and interpersonal justice perceptions, which in turn have an impact on affective
commitment (Kernan & Hanges, 2002), this dimension of ethical leadership is likely to
strengthen employee attachment even more. Furthermore, ethical leaders and their employees
likely have a positive social exchange relationship (Brown & Treviño, 2006). This means that
organization (Bishop, Scott & Burroughs, 2000; Eisenberger, Armeli, Rexwinkel, Lynch &
Rhoades, 2001).
The moral manager component of ethical leadership (Treviño et al., 2000) can also
play a role in increasing employee attachment because it fosters an ethical climate (Neubert et
al., 2009). Because this ethical environment evokes commitment to the organization (Brown
& Treviño, 2006), ethical leadership will likely also indirectly influence employee
commitment. Thus, we expect a positive relationship between ethical leadership and ethical
climate. However, this relationship is displayed in the conceptual model as a gray dotted line
(figure 1, p. 9), because it is not the main focus of the study and will therefore not be tested.
Empirical research seems to support these ideas. Positive relationships have been
identified between ethical leadership and organizational identification (Riggio, Zhu, Reina &
Maroosis, 2010; Walumbwa et al., 2011), as well as between ethical leadership and
employees’ (affective) commitment (Den Hartog & De Hoogh, 2009; Kalshoven et al., 2011; Kalshoven et al., 2013; Kim & Brymer, 2011; Neubert et al, 2009). In line with theory and
previous findings, we expect that:
H1: Ethical leadership is positively related to employee attachment.
3.3 The Mediating Role of Employee Attachment
3.3.1 Mediation to unethical pro-organizational behaviour (UPB). Employee
attachment is known to have several positive effects on employee social or
pro-organizational behaviour (O’Reilly & Chatman, 1986). These effects can be explained using
social identity theory (Ashforth & Mael, 1989) and social exchange theory (Blau, 1964).
According to social identity theory, employees who identify with the organization have internalized the organization’s successes and failures as being their own successes and failures (Mael & Ashforth, 1992), and have thus tied their own identity to the organization.
Therefore we expect that highly identified employees will do whatever they can to help the
organization move forward, considering that people generally want to belong to a prestigious
group (Ashforth & Mael, 1989). Conversely, when identification is low, employees are not as
involved in coming up with solutions, because they do not perceive issues as their problem
(McLean Parks, Ma & Gallagher, 2010). Indeed, organizational identification has been shown
to relate to both in-role and extra-role performance (Riketta, 2005).
In addition, according to social exchange theory (Blau, 1964), employee attachment in
the form of commitment can be seen as reciprocation by the employee for favourable
treatment from the leader (Bishop et al., 2000; Eisenberger et al., 2001). Thus, employee
attachment can be seen as an indication of the presence of a positive social exchange
relationship (Konovsky & Pugh, 1994). Because ethical leaders are seen as representative of
the organization (Kalshoven & Den Hartog, 2009), the employee will likely reciprocate
towards the organization, rather than just the leader (Kalshoven et al., 2013). This
reciprocation can take the form of pro-organizational behaviours if employees believe that
pro-organizational behaviour is valued (Kalshoven et al., 2011). Empirical research shows
support for this notion: employee commitment has a positive relationship with pro-social
behaviour in general (Bishop et al., 2000; O’Reilly & Chatman, 1986).
However, a strong attachment to the organization can mean that employees lose sight
of ethical issues. High identification has been shown to be potentially destructive of
organization’s ethical learning, because highly identified employees start to defend unethical organizational behaviour (Ploeger & Bisel, 2013). Furthermore, over-identification, in which
case employees see almost no distinction between the self and the organization, can lead to
pro-organizational workplace crime (Vadera & Pratt, 2013).
A form of workplace deviance that is unethical and pro-organizational at the same
et al. (2010). Pro-organizational refers to the fact that the behaviour is carried out with good
intentions, to help the organization. It is also not a part of the formal job description or asked for by a leader (Umphress et al., 2010). Unethical means that the act is “either illegal or morally unacceptable to the larger community” (Jones, 1991, p. 367). Thus, UPB is behaviour
which is done to advance the organization, against the norms of society. The formal definition of UPB is: “actions that are intended to promote the effective functioning of the organization or its members (e.g., leaders) and violate core societal values, mores, laws, or standards of proper conduct” (Umphress & Bingham, 2011, p. 622). This can be done by commission, such as misrepresenting the truth, or omission, for example being silent on information which
could harm the organization (Umphress & Bingham, 2011).
Considering that a strong attachment to the organization can lead to increased
engagement in pro-organizational behaviour, as well as lead to a diminished focus on
ethicality, we propose that employee attachment will relate positively to UPB. Ethical leaders,
despite their roles as moral managers, could thus actually increase employees’ UPB through
an increased employee attachment to the organization.
Previous research findings seem to support this idea. Umphress et al. (2010) did not
find a significant relationship between organizational identification and UPB. They did find,
however, a positive relationship when positive reciprocity beliefs were present. Considering
that ethical leadership should invoke organizational identification as well as invoke
reciprocity (Brown & Treviño, 2006), we expect that organizational identification will have a
positive relationship with UPB. Other empirical research has also supported the notion that
both forms of employee attachment, organizational identification and commitment, could lead
to engagement in UPB (Effelsberg & Solga, 2013; Effelsberg, Solga & Gurt, 2014; Kalshoven
Still, some critical notes must be placed with regards to an alternative theoretical
perspective and previous empirical research. The alternative theory is that of social learning,
which poses that leaders can have an influence on followers through role modelling (Bandura,
1986). Because ethical leaders role model ethical behaviour, they should then reduce, not
increase, the amount of UPB by followers. However, considering that a strong attachment to
the organization can lead to a diminished focus on ethical concerns (Ploeger & Bisel, 2013),
we believe the effects of social identity and social exchange will outweigh the positive role
modelling effects and thus hypothesize a positive link between ethical leadership or employee
attachment and UPB.
As to the empirical research, Cullinan, Bline, Farrar and Lowe (2008) found a negative
relationship between employee commitment and ethically questionable behaviour that is
pro-organizational in nature. They used a different measure than Umphress et al.’s (2010) UPB scale, which could explain the differential effects. Cullinan et al.’s (2008) measure is highly specified because detailed scenarios are presented to the respondents. It could be that the
respondents had difficulty imagining such a specific situation which may not apply to their
own jobs. Another explanation is that Cullinan et al. (2008) asked the respondents what their
peers would do in the described situations, and used that as a measure of the tendency to
engage in ethically questionable behaviours, to reduce possible social desirable answers. In Umphress et al.’s (2010) study and in the study presented in this thesis, the focus is on what the respondent says he or she themselves would do. We believe this is a more accurate
predictor of individual UPB. Further, in the current study, the individual level of employee
attachment is used as a mediator, so the behaviour that follows should come from this same
individual. Concluding, the following hypothesis is proposed:
H2: Employee attachment mediates the relationship between ethical leadership and UPB.
3.3.2 Mediation to pro-social rule breaking (PSRB). Another pro-organizational
behaviour that employees can show is social rule breaking (PSRB). The concept of pro-social rule breaking was developed by Morrison (2006). It is defined as “any instance where an employee intentionally violates a formal organizational policy, regulation, or prohibition
with the primary intention of promoting the welfare of the organization or one of its
stakeholders” (p. 6). In her study, Morrison found that employees engage in PSRB for three reasons: to perform organizational tasks and duties in a more efficient way, to help a
co-worker with job tasks, or to provide better customer service.
Thus, commonalities between UPB and PSRB are that they are both
pro-organizational (or pro-social), in that the underlying motive is to help the organization. They
are also both intentional, meaning that they are not due to unintentional mistakes or errors.
The difference between these concepts is that UPB violates hypernorms, or societal standards
(Umphress & Bingham, 2011), while PSRB violates specific organizational rules. Also, with
UPB, the rules or norms which are broken can be implicit, while the rules that are broken in
PSRB are formal and made explicit by the organization. Another difference is that next to
wanting to help the organization, UPB can also be carried out to help the organizations’ members, while PSRB can also be carried out to help (one of) the organizations’ stakeholders.
To our knowledge, no research has yet investigated the relationship between employee
attachment and PSRB. However, PSRB can be characterized as a type of personal initiative
behaviour, which is positively influenced by affective commitment (Den Hartog & Belschak,
2007). Moreover, some authors argue that when ethical decision making is emphasized by
the organization (or leader), employees engage in PSRB to maintain the welfare of the
organization while following ethical criteria (Koo Moon & Kwon Choi, 2014). Thus, because
the source of employee attachment is ethical leadership, which emphasizes ethical decision
organizational misbehaviour intended to benefit the organization (OMB type O; Vardi &
Wiener, 1996). The predisposition of this type of organizational misbehaviour is proposedly
higher under person-organization identification (Vardi & Wiener, 1996), a form of employee
attachment. Furthermore, we expect that the same mechanism as with UPB applies:
employees will do their best to help the organization because they are highly identified and
want the organization to progress, and because they want to reciprocate towards their leader. Indeed, employees’ personal engagement (i.e., their emotional commitment towards the organization) has a positive relationship with PSRB (Luth & May, 2012).
However, there is also a conflicting theory, which argues that group authorities draw
their legitimacy from the social relationships with members of the group. This legitimacy in turn causes members to “internalize their feelings of obligation to obey group rules and the decisions of group authorities” (Tyler, 1997, p.323). The breaking of rules would result in negative implications for employees’ self-concept. Thus, the enhanced relationship between
the organization and its members, which comes about because of ethical leadership, should
cause employees to adhere to organizational rules, not break them. Notwithstanding, PSRB is a concept different from ‘normal’ rule breaking, for example for selfish reasons, because it is pro-social. It is specifically targeted to advance the organization. Whether attachment has a
positive or negative effect on PSRB is thus a matter of whether the rule-following
consequence of attachment outweighs the pro-social consequence of attachment. We believe
that the latter is the case. Thus, the following is hypothesized:
H3: Employee attachment mediates the relationship between ethical leadership and PSRB.
3.4 The Moderating Role of Ethical Climate
In order not to engage in unethical behaviour, employees need a clear and consistent
message about the ethical standard in an organization (Miao et al., 2013). Ethical climates
send such a message to employees because they convey leaders’ and the organization’s
expectations concerning ethics and rules to employees (Cullen, Victor and Bronson, 1993;
Victor & Cullen, 1988). According to social information processing theory (Salancik &
Pfeffer, 1978), employees search for cues in their environment to understand what the correct
and expected behaviour is. Mayer et al. (2010a) argue that an ethical climate can act as such a
cue.
An ethical climate is a type of work climate. A work climate can be characterized as a system’s perceived practices and procedures. When there is an ethical content to these
procedures, it can be described as an ethical work climate (Victor & Cullen, 1988). An ethical
climate is not just influenced by ethical leadership, but knows other antecedents and correlates
as well. For example, the organizations’ formal ethics codes, ethics committees, ethics
communication systems, ethics training programs, disciplinary processes, and top management’s commitment to ethics all help to create and preserve an ethical climate (Weaver, Treviño & Cochran, 1999). Victor and Cullen (1988) and Cullen et al. (1993)
identified several ethical climate types. The different types are classified based on how ethical
decisions are made (ethical criterion), and on the locus of analysis. This study will focus on
two ethical climate types, namely ethical law and codes climate, and ethical rules climate.
These types are chosen because they are about not breaking ethical codes, which is proposed
to affect UPB, and not breaking rules, which is proposed to affect PSRB, respectively. Both use ‘principle’ as an ethical criterion, which means that in ethical decision making, rules and laws serve as guiding mechanisms. Concerning the locus of analysis, a law and codes ethical
to the organization (e.g., society). In contrast, a rules-focused ethical climate concentrates on
the local level, or the social system itself (e.g., the organization).
3.4.1 Moderation between employee attachment and UPB. When there is a strong
ethical climate that focuses on law and professional codes, employees will perceive ethical
behaviour as the expected behaviour and will act accordingly. The normative pressures that
flow from a law and codes ethical climate should decrease unethical behaviour, even if it is
pro-organizational. Empirical research findings have shown support for these ideas: principled
ethical climates (under which law and codes ethical climate) are negatively associated with
unethical choice (Kish-Gephart, Harrison & Treviño, 2010). This leads us to the following
hypothesis:
H4: Ethical law and codes climate is negatively related to UPB.
Furthermore, it is expected that an ethical law and codes climate acts as a moderator of
the relationship between employee attachment and UPB. It is proposed that at high levels of
ethical law and codes climate, the presence of this clear ethical climate will weaken the
positive relationship between employee attachment and UPB. Matherne and Litchfield (2012)
have shown that moral identity weakens the relationship between affective commitment and
UPB. It is theorized here that ethical climate can act in a similar fashion to moral identity, and
can have a similar effect on UPB. A high ethical climate likely leads employees to think about
the possible negative consequences of unethical pro-organizational acts. When employees are
in a positive exchange relationship with the organization, they will therefore show less
unethical behaviour (Ilie, 2012). Thus, under a high level of ethical climate, employee
attachment will, on average, have a less strong effect on UPB.
Further, it is proposed that at low levels of ethical climate, the positive relationship
between employee attachment and UPB will be stronger. This is because in this situation, the
3.3.1), but the low ethical climate also ‘permits’ them to show behaviours that are not in line with the law and professional codes.
Finally, at moderate levels of ethical climate, there is a sufficient amount of
attachment to evoke pro-organizational behaviours from the employee, but the ethical climate
is not very clear and consistent. As a result, employees may get the impression that the
organization finds ethical issues important, but that the other interests of the organization are
even more important (Miao et al., 2013). This results in an ambiguous situation where ethical
standards and the interests of the organization are in conflict. Employees are then likely to
engage in a process of neutralization, where unethical behaviours are justified by other
reasons (Miao et al., 2013; Umphress & Bingham, 2011). Thus, under a moderate level of
ethical climate, the relationship between employee attachment and UPB stays the same.
Concluding, the following hypothesis can be stated:
H5: The positive relationship between employee attachment and UPB is moderated by ethical law and codes climate, so that this relationship is less strong for higher values of ethical law and codes climate.
3.4.2 Moderation between employee attachment and PSRB. When there is a high
ethical climate that emphasizes the following of rules, employees will know that following
rules is the expected behaviour and act accordingly. A rules ethical climate should make rule
breaking unimaginable, even if it is pro-social (Vardaman, Gondo & Allen, 2014). Indeed,
one empirical study shows that a rules and standard operating procedures ethical climate has a
negative association with rule bending (Borry, 2013). However, as described before, some
authors argue that when ethical decision making is emphasized by the organization or leader
(e.g., an ethical climate is present), employees are likely to engage in PSRB to maintain the
welfare of the organization (Koo Moon & Kwon Choi, 2014). This would seem to contradict
the ethical criteria (Koo Moon & Kwon Choi, 2014), and the ethical criteria in an ethical rules
climate are specifically targeted to the adherence to rules, engaging in PSRB would not be in
line with these criteria. Thus, employees will not be as likely to engage in PSRB in the
presence of an ethical rules climate. Thus, we state the following hypothesis:
H6: Ethical rules climate is negatively related with PSRB.
As to the moderation effect between employee attachment and PSRB, the same
principle applies here as with an ethical law and codes climate and UPB: it is expected that at
high levels of ethical rules climate, the relationship between employee attachment and PSRB
will be less strong, because it is very clear to employees which behaviours are expected of
them. At low levels of ethical rules climate, the employee has incentives to show pro-social behaviours, and the ethical climate ‘allows’ rule breaking to occur, which will result in an even stronger relationship between employee attachment and PSRB. Lastly, at moderate
levels of ethical climate, the ambiguity of following rules versus obtaining results is ensued
by a process of neutralization. Thus, the relationship between employee attachment and PSRB
is not changed.
H7: The positive relationship between employee attachment and PSRB is moderated by ethical rules climate, so that this relationship is less strong for higher values of ethical rules climate.
4. Research Methods
4.1 Research Approach and Design
This quantitative study was carried out using a cross-sectional survey design. A group
of five master students worked together to find respondents and administer the questionnaire
to them. For the broader research two different questionnaires, one for a supervisor and one
for a subordinate, were administered. For this study, however, only the responses from the subordinates’ questionnaires were used.
The population for this study consisted of employees in the Dutch workforce. The
respondents were selected through contacting a company or by contacting individual workers.
Because the population was quite large and diverse, the aim was for purposive heterogeneous
sampling. However, convenience sampling was also used because of difficulty in gaining
access to companies. Because of the sampling technique and use of multiple researchers, the
sample was quite diverse. However, only respondents who worked more than a certain number of hours per week (no ‘side jobs’) were used in the analysis.
This study strived for as many respondents as possible; however, a minimum of 105
responses is an acceptable sample size for the analyses conducted here (Green, 1991).
4.2 Measures
The questionnaire asked respondents for their demographics such as their gender
(nominal variable), age, and tenure (ratio variables). For the other constructs, existing and
validated Likert scales on a 7-point scale (completely disagree to completely agree) at interval
level were used. These were directly translated into Dutch to ensure better understanding in
the respondents. This was then checked by two supervisors, who made alterations when
necessary. For ethical leadership and UPB, a translation into Dutch was readily available from
the questionnaire too long, with the risk of a lower response or completion rate. All
translations and shortened versions can be found in appendix A.
Ethical leadership. Ethical leadership was measured by asking the employee about
his or her leader, using 10 items. These items are based on the measure of Brown et al.
(2005). An example of an item is “[My supervisor] disciplines employees who violate ethical standards”.
UPB. To assess unethical pro-organizational behaviour, Umphress et al.’s (2010)
6-item employee measure was used. An example of this measure is “If it would help my
organization, I would misrepresent the truth to make my organization look good”.
PSRB. Pro-social rule breaking was measured using 5 items from Dahling, Chau,
Mayer and Gregory’s (2012) 13-item employee scale. An example of a statement is “I violate
organizational policies to save the company time and money”.
Employee attachment. Employee attachment was measured using existing scales for
organizational identification and affective commitment. Mael and Ashforth’s (1992) 6-item
organizational identification measure was shortened to 4 items. An example of an item is “This organization’s successes are my successes”. Allen and Meyer’s (1990) affective commitment measure, which originally consists of 8 items, was shortened to 4 items. One of
these was “This organization has a great deal of personal meaning for me”.
Ethical climate. Lastly, two ethical climate types, namely law and codes climate, and
rules climate, were assessed using items from Victor and Cullen’s Ethical Climate
Questionnaire (1988; Cullen et al., 1993). An example of an item for law and codes climate is “People are expected to comply with the law and professional standards over and above other considerations”. An example for rules climate is “Successful people in this company go by the book”.
Control variables. To control for other possible effects, we decided to include age
and gender as control variables. Previous research suggests that women hold higher ethical
values than men, and thus engage in less unethical behaviour. Furthermore, older employees
also engage in more ethical behaviour (Appelbaum, Deguire & Lay, 2005). Thus, age and
gender were used as control variables.
4.3 Data Analysis
This study used regression analyses to test the hypotheses and model. To find out if
there are relationships between ethical leadership and UPB and PSRB, and mediating and
moderating effects of employee attachment and ethical climate, respectively, model 14 of Hayes’ (2013) PROCESS macro for SPSS was used. All analyses were conducted using IBM SPSS Statistics.
4.4 Sample
The data were collected using the online software of Qualtrics. The responses of
employees could be downloaded into an SPSS data file. In total, 289 employees were
approached to take part in the study, of which 204 responded by answering at least one
question of the questionnaire. Thus, the response rate is 70.6 percent. From these respondents,
180 people completed the entire questionnaire. The completion rate is thus 88.2 percent. The
5. Results
5.1 Data Screening and Preliminary Analyses
5.1.1 Missing values and outliers. Before the analyses were conducted, the data were
checked for missing values and outliers. When looking at the individual missing data by
excluding the cases pairwise, for ethical leadership N = 173, with 3,9 percent of the original
cases missing; for employee attachment N = 176, with 2,2 percent of cases missing; for
ethical law and code climate N = 175, with 2,8 percent of cases missing; for ethical rules
climate N = 175, with 2,8 percent of cases missing; for UPB, N = 179, with 0,6 percent of
cases missing, and for PSRB, N = 179, with 0,6 percent of cases missing.
For regression analysis, however, it is best to exclude cases listwise in case of a
missing value (Field, 2009). Especially for mediation analysis, where a ‘path’ is tested, a
missing value in one of the variables is not deemed appropriate (Hayes, 2013). The total
amount of respondents for ethical leadership, employee attachment, ethical climate, unethical
pro-organizational behaviour and pro-social rule breaking was 180. Listwise exclusion of
missing cases reduced this number to 162 (N = 162). This means there were 10 percent of
missing cases from the original data.
To check for outliers, the Z-scores of ethical leadership, employee attachment, ethical
law and codes climate, ethical rules climate, UPB, and PSRB were computed in SPSS. Scores
outside the range of -3.29 and 3.29 are significant outliers, because the probability of finding
cases outside this range is less than 0.1 percent (Field, 2009). Ethical leadership, employee
attachment, and ethical law and codes climate all had outliers on the low end, three in total.
Since these outliers could have resulted in a non-normal distribution, the Kolmogorov-Smirov
and Shapiro-Wilk tests for normality were run for both the dataset with outliers and for a
dataset with the outliers removed. The finding was that excluding the outliers would result in
kurtosis of the variables, the dataset with the outliers removed appeared to have less skewness
and kurtosis. However, removing outliers from the dataset is only a valid option if there is a
reason to believe the outlier is the result of incorrectly entered or measured data, or that the
case does not come from the intended population (Field, 2009). There were no mistakes found
in the data, and there was no reason to believe the cases came from a different population. The
outliers also did not affect the assumptions of regression analysis (see section 5.1.4), or create
a significant association were there would not have been one otherwise. Thus, the outliers
were retained in the dataset.
5.1.2 Reliability tests. A reliability test using Cronbach’s α was conducted for all
constructs in the model. A score higher than .7 indicates a reliable scale, with each item
measuring the same underlying construct (Field, 2009).
Concept 1, ‘ethical leadership’, had a reliability of α = .91. Removing any items would
not have increased the reliability. Concept 2, ‘employee attachment’, had a reliability of α =
.84. If the reliability for this variable would have been lower than α = .7, this could have
indicated that organizational identification and affective commitment should be treated as
separate constructs (Field, 2009). However, the reliability was high enough to treat it as a
coherent scale. Removing items from the scale would not have increased the reliability,
except for item 6, “I would be very happy to spend the rest of my career with this
organization”, which, when removed, would have increased the reliability with .005. This is
only a small increase in the reliability of the scale, which is already high. Thus, to prevent
unnecessary data loss, the item was retained in the final scale and analyses. Concept 3 and 4, ‘unethical pro-organizational behaviour’ and ‘pro-social rule breaking’ had a reliability of α = .77 and α = .83, respectively. Removing any items would not have increased the reliability. Lastly, concept 5, ‘ethical law and code climate’, had a reliability of α = .81 and concept 6, ‘ethical rules climate’ had a reliability of α = .78. Removing any items would also not have
increased the reliability. These reliabilities can also be found in the correlation matrix (table 2,
section 5.2).
5.1.3 Normality tests. Subsequent analyses of normality were performed for the same
concepts. The skewness and kurtosis and their standard errors were produced in SPSS, and
can be found in table 1. Skewness reflects the lack of symmetry in a normal distribution, while kurtosis reflects the ‘pointiness’ (Field, 2009). The Z-scores for the skewness and kurtosis were calculated by dividing the scores by their standard errors. In a Z-score the
average equals zero and the standard deviation equals one. Consequently, the Z-score of the
skewness and kurtosis can be compared to those of a normal distribution. In a large sample, a
Z-score between –2.58 and 2.58 means the data are normally distributed (Field, 2009). Ethical
leadership and employee attachment appear to be heavily negatively skewed and to have
positive kurtosis (Z-scores of -7.70, -4.37, 5.70, and 2.65, respectively).
Table 1. Values, standard errors, and Z-scores for skewness and kurtosis
Variable Skewness SE skewness Z-score skewness Kurtosis SE kurtosis Z-score kurtosis Ethical leadership -1.425 .185 -7.70 2.091 .367 5.70 Attachment -.800 .183 -4.37 .966 .364 2.65 UPB .026 .182 .14 -.586 .361 -1.62 PSRB .290 .182 1.59 -.444 .361 -1.23
Ethical L&C climate -.435 .184 2.36 .317 .356 .87
Ethical rules climate -.094 .184 -.51 .006 .365 .02
Additionally, the Kolmogorov-Smirnov and Shapiro-Wilk tests for normality were
run. According to the first test, ethical leadership, D(173) = 0.16, p < .001, employee
attachment, D(176) = 0.11, p < .001, unethical pro-organizational behaviour, D(179) = 0.09,
p < .01, pro-social rule breaking, D(179) = 0.08, p < .01, ethical law and codes climate
significantly non-normal. According to the second test, which has more power to detect
differences from normality, UPB was normally distributed. However, ethical leadership,
W(173) = 0.88, p < .001, employee attachment, W(176) = 0.96, p < .001, pro-social rule
breaking, W(179) = 0.98, p < .05, ethical law and code climate, W(175) = 0.98, p < .01, and
ethical rules climate, W(175) = 0.98, p < .05, were all significantly non-normal.
Normality was also checked for visually using histograms and Q-Q plots (see
appendix B). In the histograms, ethical leadership and employee attachment appear to be
leptokurtic. In the probability plots, ethical leadership and employee attachment indeed seem
to be non-normal. Ethical law and code climate also appears to be slightly non-normal in the
Q-Q plot.
The data for the non-normal variables could have been transformed to correct for the
skewness and kurtosis, however, the other variables would have to be transformed as well,
with the possibility of them becoming non-normal in return. Furthermore, transformations don’t always result in a more valid probability statement, and essentially change the hypotheses tested and constructs used, which makes interpretation difficult (Field, 2009).
Additionally, in larger samples, normality tests can turn out significant when in reality scores
only deviate marginally from a normal distribution (Field, 2009). Because normality of
variables is not an assumption of regression analysis (but instead whether the errors are
normally distributed, see section 5.1.4), and because a bootstrapping method will be used for
the testing of the moderation effect, we chose not to transform the data.
5.1.4 Assumption tests.Regression analysis has certain assumptions that have to be met in order to be able to draw conclusions about a population (Field, 2009). These
assumptions were tested and met.
Firstly, the predictor variable, ethical leadership, is quantitative in nature, and the
interval level. The variables are also unbounded, which means that there is enough range in
the variability of the responses.
Secondly, the variances of ethical leadership, employee attachment, ethical law and
codes climate, and ethical rules climate, are higher than zero. The standard deviations are
0.96; 1.0; 1.1; and 1.1, respectively.
Thirdly, there is no perfect multicollinearity between predictor variables. Correlations
between predicting variables are lower than .8 (see section 5.2). In addition, this was
examined by producing variance inflation factors (VIFs) in SPSS. The VIFs are only
somewhat greater than 1 (for UPB: ethical leadership 1.14, employee attachment 1.13, and
ethical law and codes climate 1.00; for PSRB: ethical leadership 1.15, employee attachment
1.14, and ethical rules climate 1.01) and the tolerances lie above 0.2 (for the predictors of
UPB: 0.88, 0.88, and 1.00; for the predictors of PSRB: 0.87, 0.88, and 1.00). This indicates
that no multicollinearity is present (Field, 2009).
Fourthly, relevant external variables (such as age and gender) that could influence the
outcomes are included in the model.
Fifthly, the variance of the residual terms is constant. The scatterplots of the residual terms show a ‘cloud’ of dots (see appendix C). Thus, the assumption of homoscedasticity has been met.
Sixthly, the Durbin-Watson test shows that errors are independent: a value between 1
and 3 is mostly acceptable, with a value of 2 indicating that the residuals are uncorrelated
(Field, 2009). The scores for variables are 2.25 (ethical leadership and attachment); 1.88
(ethical leadership and UPB); 2.00 (ethical leadership and PSRB); 1.82 (attachment and
UPB); 1.94 (attachment and PSRB); 1.89 (ethical law and code climate and UPB); and 1.89
Seventhly, the errors are normally distributed. The histograms of the residuals show
quite a normal distribution, as do the P-P plots of the residuals (see appendix D).
Eighthly, the values of the outcome variable are independent. Different respondents
completed the questionnaire individually.
Ninthly and finally, the relationships between predictors and outcome variables can be
assumed to be linear: the scatterplots do not show clear curves. Moreover, it is clear from the
scatterplots that some variables do not have strong relationships, because the typical ‘cigar’
shape of a linear relationship is missing. The scatterplots can be found in appendix E.
5.2 Descriptive Statistics
The means, standard deviations, correlations, and reliabilities of the variables are
provided in table 2. Because not all variables were normally distributed, Kendall’s т, a
nonparametric test, was chosen to compute the correlations instead of Pearson’s r. Kendall’s т was preferred as a nonparametric test instead of Spearman’s rho, because even though
Spearman’s rho usually shows higher correlations, Kendall’s т is argued to be a better estimation of the population’s correlation (Field, 2009).
Results show there is a significant relationship between gender and unethical
proorganizational behaviour (UPB), т = .14, p < .05, and prosocial rule breaking (PSRB), т =
-.13, p < .05. Women are significantly less likely to engage in these behaviours. The age of the
respondent is significantly negatively correlated with unethical pro-organizational behaviour
(UPB), т = -.18, p < .01, and positively correlated with perceptions of an ethical law and
codes climate, т = .15, p < .01. Ethical leadership is significantly correlated with employee attachment, т = .22, p (one-tailed) < .01, and, contrary to expectations, negatively correlated with unethical organizational behaviour (UPB), т = -.12, p < .01. Unethical
ethical law and codes climate, т = -.15, p (one-tailed) < .01. There is a significant negative
relationship between pro-social rule breaking (PSRB) and the presence of an ethical rules climate, т = .10, p (one-tailed) < .05. Lastly, ethical law and codes climate and ethical rules climate are significantly correlated, т = .38, p < .01.
Table 2. Correlation matrix
Variables M SD 1 2 3 4 5 6 7 8 1. Gender 1.52 .50 - 2. Age 33.54 11.60 .04 - 3. Ethical leadership 5.45 .96 .10 .10 (.91) 4. Attachment 4.91 1.00 -.05 .10 .22** (.84) 5. UPB 3.30 1.05 -.14* -.18** -.12* -.02 (.77) 6. PSRB 3.80 1.15 -.13* .04 .03 -.08 .10 (.83)
7. Ethical law and codes climate
5.00 1.05 .03 .15** .01 -.01 -.15** -.05 (.81)
8. Ethical rules climate 4.50 1.05 -.05 .03 -.06 .05 .03 -.10* .38** (.78)
* p < .05, ** p < .01. Hypothesised effects are one-tailed (in italics), other effects two-tailed. Cronbach’s α on the diagonal.
5.3 Regression Analyses
To test the hypotheses, Hayes’ (2013) PROCESS macro for SPSS was used to conduct
two moderated mediation analyses: one for UPB as the dependent variable and one for PSRB
as the dependent variable. Gender and age were entered as control variables. A 95%
confidence interval was used for the moderation bootstrapping. If this range does not include
zero, there is a significant relationship. Furthermore, the moderating effects were reported for
different values of the moderator, ethical climate: low (one standard deviation below the
mean), moderate (the mean), and high (one standard deviation above the mean).
5.3.1 UPB. Results of the analysis for UPB can be found in table 3. The output shows
p < .001. Thus, hypothesis 1 is supported. Furthermore, gender and age are significantly
negatively related to unethical pro-organizational behaviour (UPB), b = -0.33, p < .05, and b =
-0.02, p < .01, respectively, such that female and older employees engage less in UPB than
male and younger employees.
Other proposed effects turned out non-significant: ethical leadership and employee
attachment are not significantly related to UPB. Thus, no evidence for a mediating effect is
found, and hypothesis 2 is not supported. There is no indication of a significant relationship
between ethical law and codes climate and UPB, and thus hypothesis 4 is not supported.
Furthermore, ethical law and codes climate has no significant moderating effect on the
relationship between employee attachment and UPB, providing no support for hypothesis 5.
Table 3. Results of the regression analysis for moderated mediation to UPB
Mediator variable model: employee attachment
Predictor B SE t p
Constant 2.860 .498 5.749 .000
Ethical leadership .356 .079 4.493 .000
Gender -.047 .149 -.316 .752
Age .006 .007 .899 .370
Dependent variable model: UPB
Predictor B SE t p
Constant 6.541 2.013 3.250 .001
Employee attachment -.168 .416 -.403 .688
Ethical leadership -.128 .090 -1.419 .158
Ethical L&C climate -.294 .396 -.741 .460
Employee attachment x ethical L&C climate
.041 .082 .502 .616
Gender -.331 .159 -2.086 .039
Age -.022 .007 -3.118 .002
Direct effect of ethical leadership on UPB
Predictor Effect SE t p
Ethical leadership -.127 .090 -1.42 .158
Conditional indirect effects of ethical leadership on UPB at values of ethical climate
Moderator V Effect Boot SE Boot LLCI Boot UCLI
Ethical L&C climate, low 4.015 -.001 .054 -.093 .129
Ethical L&C climate, mod. 5.032 .014 .036 -.045 .107
Ethical L&C climate, high 6.049 .029 .047 -.062 .127
5.3.2 PSRB. To assess the hypotheses concerning PSRB, the same procedure was
followed. The results are shown in table 4. The results again show that ethical leadership is
significantly positively related to employee attachment, b = 0.36, p < .001. Age and gender
have no significant effects on PSRB. As to the hypothesized effects, ethical leadership and
employee attachment are not significantly related to PSRB. Thus, no evidence for a mediating
effect is found, and hypothesis 3 is not supported. Ethical rules climate has no significant
direct effect on PSRB, providing no support for hypothesis 6. There is also no overall
moderating effect to employee attachment and PSRB. However, certain levels of ethical rules
climate do appear to have a significant moderating effect on the mediated relationship
between ethical leadership and PSRB. For lower and moderate levels of ethical rules climate,
the proposed mediation from ethical climate to PSRB through employee attachment is
significant (Effect = -.104 and -.065, respectively), because the confidence intervals do not
include zero. However, contrary to expectations, this relationship is negative, and thus
Table 4. Results of the regression analysis for moderated mediation to PSRB
Mediator variable model: employee attachment
Predictor B SE t p
Constant 2.877 .498 5.773 .000
Ethical leadership .361 .079 4.548 .000
Gender -.073 .149 -.486 .628
Age .005 .007 .696 .487
Dependent variable model: PSRB
Predictor B SE t p
Constant 6.92 2.101 3.294 .001
Employee attachment -.639 .394 -1.620 .107
Ethical leadership .095 .103 .928 .355
Ethical rules climate -.589 .432 -1.364 .175
Employee attachment x ethical rules climate
.102 .085 1.192 .235
Gender -.330 .180 -1.827 .070
Age -.013 .008 1.580 .116
Direct effect of ethical leadership on PSRB
Predictor Effect SE t p
Ethical leadership .095 .103 .928 .355
Conditional indirect effects of ethical leadership on PSRB at values of ethical climate
Moderator V Effect Boot SE Boot LLCI Boot UCLI
Ethical rules climate, low 3.451 -.104 .064 -.295 -.017
Ethical rules climate, mod. 4.511 -.065 .045 -.190 -.004
6. Discussion
6.1 Discussion of Results
The hypothesis testing indicated that ethical leadership has a positive relationship with
employee attachment. While the correlations show that ethical leadership has a negative
relationship with UPB, in the regression analyses (that corrected for the effects of age and
gender) no relationship was found between ethical leadership and/or employee attachment and
UPB. Furthermore, ethical leadership appears to have an effect on PSRB through employee
attachment for some values of ethical rules climate. However, the effects were different in
nature than hypothesized: when ethical rules climate was low or moderate, ethical leadership
had a negative, not positive, effect on PSRB through employee attachment.
6.1.1 Theoretical implications and future research suggestions. The implications
for theory of this study are fivefold. Firstly, it confirmed existing research that argued for a
positive effect of ethical leadership on employee attachment. The fairness, consideration,
honesty and trust of ethical leaders likely lead to this attachment through their signalling
effect on employees. Future research could focus on these hypotheses by explicitly testing
them as mediators of the relationship between ethical leadership and employee attachment, for
example, it could be tested if justice perceptions act as a mediator of this relationship.
Secondly, one of the surprising findings from this research is that there was no
indication of a positive relationship between ethical leadership and UPB when correcting for
age and gender. This is not in line with existing theories and previous research that showed
positive or curvilinear relationships (Kalshoven et al, 2013; Miao et al., 2013). It is not
exactly clear why a different relationship was found, however, one explanation is that there
could have been curvilinear effects, as was the case in some of the previous research (Miao et
al., 2013). In the current study, the assumption of linearity was met, but a graph shows that a
the entire model with a curvilinear relationship, because of limitations in terms of statistical
programs (Hayes, 2013).
Another possible explanation is that previous research used other moderators in their
research that could explain the difference in the findings. For example, Kalshoven et al.
(2013) found that job autonomy acts as a moderator of the relationship between ethical
leadership and UPB through organizational commitment. Only at high levels of job autonomy
a mediation effect was found. These authors argued that at low levels of job autonomy,
employees do not always have the opportunity to engage in UPB. The same could apply to
PSRB, because Morrison (2006) found that autonomy predicted the occurrence of these
behaviours. A reason for this link between autonomy and deviant behaviour could be the lack
of close monitoring from the leader or organization (McLean Parks et al., 2010). Because job autonomy was not included in this study’s questionnaire, we were unable to test this
hypothesis. However, it provides a possible explanation for the rareness of significant
relationships in this study compared to previous studies. Another moderator that was included
in previous research was positive reciprocity beliefs from the employee. Positive reciprocity
beliefs reflect “the degree to which individuals endorse reciprocity in exchange relationships”
(Umphress et al., 2010, p. 769). Umphress et al. (2010) found that organizational
identification was not significantly positively related to UPB, unless this was combined with
positive reciprocity beliefs. This is connected to the social exchange theory’s proposition that
employees want to reciprocate towards their leader by showing pro-organizational behaviour.
It appears that this reciprocation only takes place when employees have these positive
reciprocity beliefs. This variable was not included in this study, so it could have accounted for
the absence of some significant relationships.
An alternative explanation for the lack of significant findings could be that leaders