© Academy of Management Journal 2012, Vol. 55. No. 5, 1053-1078. http://dx.doi.org/10.5465/ainj.2011.0064
EMBEDDING ETHICAL LEADERSHIP WITHIN AND ACROSS ORGANIZATION LEVELS JOHN M. SCHAUBROECK Michigan State University SEAN T. HANNAH Wake Forest University BRUCE J. AVOLIO University of Washington STEVE W. J. KOZLOWSKI Michigan State University ROBERT G. LORD Akron University LINDA K. TREVIÑO Pennsylvania State University NIKOLAOS DIMOTAKIS Georgia State University ANN C. PENG Michigan State University We develop and test a model linking ethical leadership with unit ethical culture, both across and within organizational levels, examining how both leadership and culture relate to ethical cognitions and behaviors of lower-level followers. The data were collected from 2,572 U.S. Army soldiers representing three organizational levels deployed in combat. Findings provide limited support for simple trickle-down mechanisms of ethical leadership but broader support for a multilevel model that takes into account how leaders embed shared understandings through their influence on the ethical culture of units at various levels, which in turn influence followers' ethical cognitions and behavior. The influences of ethical leadership occur not only directly, among immediate followers within a unit, but also indirectly, across hierarchical levels, through the cascading of ethical culture and senior leaders' influences on subordinate leader behavior. We discuss scholarly and practical implications for understanding how leaders transmit ethical influence both down and across large organizations. (e.g.. Brown, Harrison, & Treviño, 2005), aspects of organizational context, such as low levels of "ethical culture" ("bad barrels" [e.g., Treviño & Youngblood, 1990]), and the interaction of individual predispositions and context (e.g., Treviño, 1986). In this study, we extend previous research and theory to examine how ethical cognitions and behaviors of organization members reflect their responses to a web of direct and indirect influences of ethical leadership and unit-level norms, standards, and sanctions pertaining to ethical behavior transmitted across multiple levels of an organizational hierarchy.
Unethical organizational behaviors have been attributed to the effects of individual "had apples" (e.g., Hannah, Avolio, & May, 2011), their leaders The authors acknowledge the United States Army soldiers who, despite serving their nation in an active combat zone, volunteered to participate in this research. We also thank the Army chaplains who traveled between combat outposts to collect this data, the leadership of Multinational Force Iraq, and the U.S. Army's Center for the Army Profession and Ethic (CAPE), which sponsored this research. We also want to express our appreciation to Megan Huth for her assistance with the data. 1053
Copyright of the Academy of Management, all rights reserved. Contents may not he copied, emailed, posted toa listserv, or otherwise transmitted without the copyright holder's express written permission. Users may print, download, or email articles for individual use only.
1054
Academy of Management Journal
Scholars have called for taking a multilevel approach to more fully understand how leadership at higher levels of organizations influences lower-level followers (e.g., Waldman & Yammarino, 1999). Yet prior empirical work on leadership has primarily focused on direct relationships between leader behaviors or traits and the responses of their immediate followers (Yammarino & Dansereau, 2008). Researchers have not comprehensively tested the various loci through which upper-level leaders influence the cognitions and behavior of lower-level followers. Thus, currently litfle is known about how leadership and contextual factors at higher organization levels influence outcomes at lower levels. From a metatheoretical perspecfive, how a coherent system of social influence is transmitted across multiple levels of an organization is fundamental to understanding organizational behavior (Kozlowski & Klein, 2000; Simon, 1973). We develop a multilevel model in which ethical leadership, which is seen as a leader's use of social influence to promote ethical conduct (Brown et al., 2005), exhibited at various levels of an organizational hierarchy, not only directly influences immediate followers' ethical conduct, but also indirectly influences the ethical beliefs and conduct of followers at lower levels owing to (a) the replication of ethical leader behaviors among subordinate leaders and (b) the embedding of shared understandings that represent observable elements of work unit ethical culture. Our proposed model is tied to a foundational framework of organizational culture developed by Schein (1985, 2010), who posited mechanisms through which leaders embed their assumptions into the "thinking, behavior, and feelings" of groups (1985: 223). We draw from Schein's "embedding mechanisms" (1985: 224) and "shared cultiural elements" (1985: 169) and extend his theoretical framework by developing and testing a multilevel model linking leadership, shared cultural elements, and their direct and indirect effects on follower ethical cognitions and behaviors. This approach provides a starting framework for iHiderstanding how different leaders influence ethical outcomes within a complex multilevel system. THEORETICAL FRAMEWORK
Our research model draws from Schein (1985, 2010) to explain how senior leaders influence the ethical conduct of followers at lower levels by embedding their expectations and assumptions into the observable manifestations of ethical culture. Schein defined culture as a system of shared assimaptions that can have a strong influence in directing followers' behaviors and beliefs. Organiza-
October
tional culture is represented as varying in layers, with the deepest layer being a broad system of assumptions and deeply held shared meanings, and the surface layer representing more tangible, observable factors that reflect those assumptions. Schein (1985, 2010) argued that leaders at all levels can influence the surface layer of an organization's culture, including visible artifacts such as behavioral norms, policies, and standards, particularly as these relate to specific domains of influence that Schein described as shared cultural elements. By influencing these surface elements, leaders can help others interpret their unit and organization's culture in terms of how it relates to their roles and expectations. Unit Ethical Culture Shared understandings concerning ethical conduct can be seen as a shared cultural subelement, or a "microcosm" of an overall organizational culture (Treviño, 1986, 1990). This aspect is the part of the culture that distinguishes what is "heroic" from what is "sinful" (Schein, 1985: 79) and is composed of patterns of shared understandings related to unethical and ethical conduct reflecting the norms, standards, sanctions, and rewards applied to behaviors deemed desirable and undesirable in an organization. This focus is consistent with how Treviño (1986, 1990) and Treviño, Butterfield, and McCabe (1998) defined ethical culture. Specifically, in their view ethical culture is "a subset of organizational culture, representing a multidimensional interplay among various 'formal' and 'informal' systems of behavioral control that are capable of promoting either ethical or unethical behavior" (Treviño et al., 1998: 451-452). This use of the term "culture" should not be equated with broader definitions of culture that relate to the shared assumptions that combine to form deep organizational culture. Rather, in the current study, ethical culture refers to surface-level cultural artifacts that are observable and were called shared cultural elements by Schein (1985). This conceptualization is similar to descriptions of climate in other research (e.g., Zohar & Luria, 2005). However, there exists a longestablished construct and measure of ethical climate (Victor & Cullen, 1988) in the organizational ethics literature that refers to criteria used in ethical judgment processes and the level of analysis used by parties in making ethical judgments. This construct has spawned a significant amount of research over the last 20 years (Arnaud & Schminke, in press). Thus, we use the label "ethical culture" rather than "ethical climate" to be consistent with Schein (1985, 2010), noting that we are focusing on
Schaubroeck et al.
2012
the surface level of ethical culture, following Treviño (1990; Treviño et al, 1998), and to avoid confusion with the established ethical climate construct and its different meaning. "Formal" ethical culture systems include policies (e.g., codes of ethics that are enforced), authority structures, reward systems, and ethics training programs. "Informal" systems include peer behavior, use of ethical language, myths, stories, and ethical norms. When the members of an organization or a unit in an organization have strong shared understandings about such matters, they are expected to be more cognizant of ethical issues, to avoid unethical conduct personally, and to discourage unethical conduct in others (Treviño, 1986). Such units penalize unethical behavior, reward virtuous behavior, and maintain strong ethical norms. This conceptualization contains botb global properties pertaining to organizational codes and shared social properties within the unit concerning agreed-upon norms and prescribed patterns of behavior. Following Schein, we refer to such domains in general as "shared cultural elements" and refer to Treviño and colleagues' (1998) construct as it applies to the domain of ethics at various hierarchical levels as "ethical culture." Whereas one may also use the term "ethical culttire" to refer to an organization as a whole, we expect that ethical culture varies across different units and levels of an organization despite common global influences, such as organizational codes of ethics and other norms, rules, and regulations at the organization or societal level. In addition, as Schein noted, "'shared'
1055
understanding means that vmit members recognize a particular feeling, experience or activity as common" (1985: 168; emphasis in original). From a multilevel theory perspective, such constructs are compositional in form and exhibit restricted within-unit variation (Kozlowski & Klein, 2000). Overview of Research Model In this study, we develop a multilevel model for understanding how leadership and culture influence ethics-related cognitions and behaviors, and we test it using survey data collected from a large sample of deployed soldiers at different hierarchical levels. Specifically, we examined leadership by assessing four separate potential mechanisms for leadership effects across organizational levels, as displayed in Figure 1. First, a senior leader's style may be mirrored by that leader's followers (Bandura, 1971). To the extent that subordinate leaders' behavior mirrors that of their senior leaders, a positive downward relationship exists between the styles of leaders at adjacent levels. Second, indirect effects of higher-level leaders that involve multiple downward linkages are possible; these effects may occur, for example, when subordinate leaders mirror senior leaders' styles and tbose subordinate leaders' styles in turn influence how their followers think and behave. The passage of direct and indirect effects down a hierarchy of authority has been variously labeled "the cascade of leadership," "the trickle-down model," and "the falling dominoes
FIGURE 1 Direct and Indirect EÉfects of Ethical Leadership and Ethical Culture across Hierarchical Levels^
Level 3
Ethical Leadership (L3)
Ethical Gulture (G3)
Level 2
Ethical Leadership (L2)
Ethical Gulture (G2)
Level 1 Ethical Leadership (LI)
Ethical Gulture (Gl)
Ethical and Unethical Gognitions and Behaviors (01)
° Dashed lines denote unexplained direct effects that have been conventionally labeled as "bypass" effects.
1056
Academy of Management Journal
effect" (Bass, Waldman, Avolio, & Bebb, 1987; Mayer, Kuenzi, Greenbaum, Bardes, & Salvador, 2009; Waldman & Yammarino, 1999). Third, an indirect leadership effect may occur through the horizontal and lateral linkages shown in Figure 1; for instance, a more senior leader may foster an ethical culture that then spreads to influence culture and behavior in lower organizational units. This effect of leaders on unit cultures is a more diffuse and complex process involving multiple coactors. It may be reflected in multiple effects showing mediation at different levels, such as through subordinate leader and follower behaviors and through leaders' effects on unit culture at lower levels (Bass & Bass, 2008). Finally, senior leaders' influence can skip levels and directly affect lowerlevel followers' bebavior as a result of their direct interactions (Detert & Treviño, 2010). This creates a fourth potential mechanism of leader influence referred to as a bypass processing model (Yammarino, 1994).
Outcomes of Ethical Leadership and Ethical Culture Prior research on ethical leadership has assessed a limited set of criteria. A recent review of the ethical leadership literature by Brown and Mitchell (2010) noted that besides research on organizational citizenship behavior (OCB), surprisingly little empirical research has examined how ethical leadership relates to follower ethical or unethical behavior. Theory and research suggest that the normative and informational influences of a unit with strong ethical leadership should support ethical conduct and curb organization members' serious ethical transgressions. Collective understandings about ethical conduct that become embedded in shared cultural elements should also encourage stronger beliefs among individual members about their moral agency, including (a) moral efficacy and (b) intentions to report others' ethical violations. Hannah et al. defined moral efficacy as "an individual's belief in his or her capabilities to organize and mobilize the motivation, cognitive resources, means, and courses of action needed to attain moral performance within a given moral domain, while persisting in the face of moral adversity" (2011: 24). With higher moral efficacy, individuals are more likely to convert moral judgments and intentions into ethical actions. Hannah and Avolio (2010) proposed that moral efficacy would be bolstered when followers are immersed in a context that provides the "means" (e.g., leader support and policies supportive of ethical actions) to act ethically.
October
Individuals' intentions to report the ethical violations of others are also related to moral agency. If unethical behavior is to be addressed in organizations, authority figures must know about it and therefore must set conditions to promote follower reporting. Followers tend to keep their knowledge of ethical problems to themselves for a number of reasons, including fear of retaliation, a sense that nothing will be done, or habituation to silence in authority situations (Detert & Edmondson, 2011; Kish-Gephart, Detert, Treviño, & Edmondson, 2010). Ethical leadership and a strong ethical culture can be expected to enhance followers' wiflingness to speak up because they are more likely to feel protected from retaliation and to believe that posifive acfions will be taken to address their concern. In past research, ethical leadership has been posifively related to "voice" and the likelihood that followers will report problems to authoriües (Brown et al., 2005; Walumbwa & Schaubroeck, 2009). Beyond being seen as promoting followers' moral efficacy and enabling them to report violations and to refrain from ethical transgressions, ethical leaders are also seen as moral role models who encourage followers to strive toward moral ideals and engage in ethical behavior that exceeds expectations (Brown & Treviño, 2006). We draw from scholarly work on virtuous ethical behavior (e.g.. Walker & Henning, 2004) in defining exemplary behavior as undertaking imusual personal risk or sacrifice to achieve a desirable collective purpose. It is important to note that these outcomes—moral efficacy, intentions to report violations, transgressions, and exemplary behavior—are conceptualized as "compilation constructs" (Kozlowski & Klein, 2000). Rather than regarding them as constructs that are emergent and shared at the unit level, as are ethical leadership and ethical culture, we conceptualize them as configurations of individual belief states, intentions, and rates tbat capture the descriptive averages among individuals in a unit. Transmission of Ethical Leadership through Ethical Culture Direct effects. The notion that there is a relationship between wbat leaders do and the norms and codes of their units—such as the direct effects of ethical leadership on ethical culture shown as L3 ^ C3, L2 -» C2, and Ll -^ Cl in Figure 1, which summarizes our model—is quite well accepted among organizational scholars (Schein, 2010; Yukl, 2002). Leaders establisb priorities for their units and expectations for what will be rewarded, punished, and tolerated. In addition to employing these more transactional means, leaders can also serve as
2012
Schaubroeck et al.
moral exemplars for followers to help them understand what is right and wrong and to be attracted to developing a more ethically centered self-image (Walker & Herming, 2004). However, previous leadership studies have not explicitly integrated unit shared cultural elements (Schein, 1985, 2010) into either their theoretical framing or their analyses of how these elements may mediate the relationships between leader behavior and follower outcomes occurring within or between organizational levels. Such cultural elements are likely to be important because, in addition to directly impacting followers' beliefs and behaviors, leaders create shared understandings that affect these outcomes (Depret & Fiske, 1993; Dragoni, 2005). With regard to ethics-related outcomes, follower cognitions and behavior should tend to align with these shared normative understandings, which are represented as ethical culture. Thus, in our study, ethical leadership is expected to be associated with ethical culture, which in tvirn promotes more positive beliefs among members about their personal capacity to act as moral agents, as well as reduced ethical transgressions and increased exemplary ethical behavior. Ethical leadership transmitted through embedding mechanisms. To influence shared cultural elements at a particular unit level and then transmit this influence to lower levels, Schein argued, leaders strive to embed their beliefs, values, and assumptions into members' shared understandings. Leaders do this by using a range of primary and secondary "embedding mechanisms" through which they influence the visible artifacts of shared cultural elements (Schein, 1985: 223). The primary mechanisms are seen as having the most direct and strongest effects on the beliefs, values, and assumptions of followers. Among these primary mechanisms, Schein described "what leaders pay attention to, measure, and control" as being the most powerful (2010: 237; cf. Feldman, 1984). Other primary embedding mechanisms include "deliberate role modeling, teaching, and coaching" and "allocating rewards and status" (Schein, 2010: 236). Followers' perceptions of a leader's ethical leadership relate directly to these primary embedding mechanisms. Specifically, Brown and colleagues defined ethical leadership 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" (2005: 120). Ethical leadership includes what has been labeled a "moral manager" aspect and a "moral person" aspect (Brown & Treviño, 2006: 597). As a moral manager, a leader
1057
specifies conduct that is proper and improper, explains the rewards and sanctions associated with behaving ethically and unethically, appropriately disciplines infractions, and rewards virtuous behavior. As a moral person, the leader models exemplary behavior and leads in a principled, just, and caring manner. Ethical leadership as conceptualized by Brown et al. (2005) incorporates elements of Schein's (1985, 2010) primary embedding mechanisms, including paying attention to, measuring, and controlling followers' behavior, as well as the primary mechanisms of role modeling and coaching followers to exhibit exemplary ethical behavior and to avoid unethical behavior. Other embedding mechanisms may also be represented in ethical leadership. Schein described how a leader's approach to handling "critical incidents" (cf. Feldman, 1984), allocating resources, and "selecting, promoting, and excommunicating" (Schein, 1985: 235) followers are additional primary embedding mechanisms. When a member's ethical infraction comes to a leader's attention, this event can be seen as a critical incident. How the leader responds is diagnostic and will influence followers' perceptions of the leader as a moral manager. Failing to discipline the follower and use this critical incident as a lesson to others may undermine her/his authority in ethical matters and persuade followers that ethical standards are not enforced in the unit. Alternatively, a significant loss of status or dismissal ("excommunication") of a follower for an ethical infraction may reinforce perceptions of ethical leadership and, in turn, increase confidence that sanctions are used to enforce ethical standards. One can also imagine that leaders who allocate resources to teaching proper ethical conduct, such as through ethical training, would more likely be seen as moral managers or ethical leaders. Ethical cultiure may also involve secondary embedding mechanisms in which leaders engage. Schein (1985, 2010) described these as being less powerful and potentially more ambiguous than the primary embedding mechanisms. These secondary mechanisms refer to such artifacts as stories, workflow designs, and other organization structures, procedures, and formal statements of leaders. For example, ethical leaders may tell stories of how an individual's exemplary ethical behavior was rewarded, and/or stories of how past organization members' unethical conduct was disciplined. Such stories may be very salient in members' memories, and they may pass them along to others to whom the story will serve as an incentive or as a deterrent. These secondary mechanisms are also consistent with ethical leadership, as stories and formal state-
1058
Academy of Management Journal
ments with implications for ethical conduct are a potential tool for managers seeking to lead ethically. Together, the primary and secondary embedding mechanisms are expected to shape the formation of shared, unit-level perceptions of ethical leadership as a compositional construct (Kozlowski & Klein, 2000). Ethical leadership in turn becomes embedded into shared understandings about ethical norms, standards, and sanctions in a unit—that is, ethical culture—which is also compositional. Studies by Brown and colleagues (2005) and other recent work collectively indicate that when individuals perceive their leader as a proponent and exemplar of ethical behavior, they report individual and work unit psychological states that are conducive to more ethical conduct (Walumbwa, Mayer, Wang, Wang, Workman, & Christensen, 2011; Walumbwa & Schaubroeck, 2009). When they perceive stronger ethical leadership, followers are more likely to engage in prosocial behaviors and less likely to engage in deviant or counterproductive behaviors (Mayer et al., 2009; Neubert, Carlson, Kacmar, Roberts, & Chonko, 2009). However, we are not aware of any research that has examined the linkage between ethical leadership and ethical cultvue, much less research concerning how ethical culture may aid in the transmission of ethical leadership in a manner that influences important ethics-related cognitions and behaviors of followers within and between organizational levels. Hypothesis 1. Ethical leadership is positively related to ethical culture within a hierarchical level. Hypothesis 2. Through ethical culture, ethical leadership is indirectly and positively related to unit members' moral agency (their belief in their capacity to act effectively as moral agents and their intentions to report unethical behavior) and exemplary ethical behavior and negatively related to their ethical transgressions. Hypotheses 1 and 2 recognize that differences likely exist in leadership and ethical culture among units at a particular organizational level. In the next section, we discuss how our theoretical framework is operationalized at different hierarchical levels and how higher-level ethical leadership and ethical culture are associated with outcomes at the lowest level. Cascading Effects of Ethical Culture and Ethical Leadership Unit-level differences in ethical culture may be determined in part by variation in unit ethical lead-
October
ership as described above. Importantly, however, these unit differences in ethical culture may also reflect the impact of higher-level leaders who can influence the ethical culture of lower-level units, either directly through their personal contact with members, or indirectly through their influence on the thoughts and behaviors of lower-level leaders. Below, we discuss these direct and indirect processes further in the context of leader behavior cascades and bypass processes. Differentiation and cascading of ethical culture. To date, the limited research on ethical culture has focused chiefly on followers' perceptions of ethical culture in their overall organization (e.g., Treviño et al., 1998). Given that culture has both global and local emergent properties, however, ethical culture may differ among organizational units (Glisson & James, 2002). For example, differences may emerge between units, both within and across management levels, in the extent to which norms for conduct are made clear to members, ethical standards are enforced, and norms emerge for tolerating behaviors that conflict with established codes. Factors other than leader behavior may also influence ethical culture at lower levels of an organization. These include a superordinate ethical culture of the organization that conveys standards of conduct, sanctions for their violation, norms for ethical conduct, and more for the entire organizational community (Schein, 1985). A mechanism that Schein does not explicitly examine is how cultural elements in a particular unit in an organization tend to mimic content from cultural elements established in higher-level units. Ethical cultures of lower-level units are nested within broader organizational units and thus may be affected by the cultural elements operating at higher levels. These effects may be somewhat independent of the proximal, within-level influences of leadership. For example, shared cultural elements can be explained in part by the past actions of higher-level leaders, through the creation of designs and structures, systems and procedures, creeds and codes, stories, rituals, and rites (Schein, 1985, 2010). Ethical cultures at one level, however, are not necessarily fully replicated at lower levels. In this regard, Schein (1985, 2010) described how cultural elements of different groups in an organization can differ, conflict, or align with one another. As Brown and Treviño (2006) explained, organizational leaders and groups develop tools and practices for reasoning and acting upon ethical matters that can become embedded over time. Unit ethical cultures that are operating at higher organizational levels that are seen as effective by lower-level leaders
2012
Schaubroeck et al.
would be expected to result in those lower-level leaders' embedding them into their own units. This is especially likely when lower-level leaders perceive that they will be rewarded for reinforcing these cultural elements in their own units. Another potential lever helping ethical culture to have cascading influences at lower levels derives from leader contact and peer-to-peer contact across units. Such interactions can reinforce key aspects of the culture of the higher-level unit to which they all belong and are nested through normative and informational influences (Cialdini & Trost, 1998). For example, leaders and members at a particular level collaborate and socialize with peers and leaders in the context of meetings held at a higher organizational level. In this way, workers develop a common point of reference for embedding ethical culture into their respective units. Thus, we expect that higher-level-unit ethical cultiure will be associated with ethical culture at adjacent levels, as illustrated by the vertical lines linking C3, C2, and Cl in Figure 1. Hypothesis 3. Ethical culture at higher hierarchical levels is positively related to ethical culture at lower hierarchical levels. Influences of ethical leadership on ethical culture at lovi^er levels. The literature further suggests that ethical leadership at higher levels may be associated with ethical culture at lower levels. In some cases, these effects may be direct influences of higher-level leaders on lower-level ethical culture through direct interactions with lower-level followers. Such direct effects have been called "bypass effects" because they bypass the effects of lower-level leaders (Yammarino, 1994) or lower-level cultural elements (L3 -^ C2 or L2 -^ Cl). Detert and Treviño (2010) reported qualitative evidence of such bypass effects of senior leader behavior on voice-related practices of lower-level followers. There is also some evidence of bypass effects from quantitative field studies (Dvir, Eden, Avolio, & Shamir, 2002; Yang, Zhang, & Tsui, 2010). According to Yammarino, this may occur because leaders often interact directly with members at lower levels. Schein (1985, 2010) observed that senior leaders' expectations are also conveyed by other means to the broader organization, through secondary embedding mechanisms such as formal communications, policies, and stories. Such mechanisms may reinforce the leaders' reputation and expectations among members, thereby influencing their cognitions and actions in a way that is distinct from the influence of their immediate leader. Any examination of such processes nevertheless presup-
1059
poses a relationship between higher-level ethical leadership and lower-level ethical culture. Thus, we propose the following: Hypothesis 4. Ethical leadership is positively related to ethical culture at lower hierarchical levels. However, what appears to be a bypass effect may actually be a product of embedding processes that are omitted from an analysis. Below, we consider two different indirect pathways through which senior leaders may embed their ethical influence at lower levels. Studies presenting evidence of bypass effects have not examined the extent to which higher-level leaders have embedded their assumptions and expectations in shared cultural elements at their own (higher) level. Civen that ethical leadership is expected to be positively related to ethical culture within each organization level (Hypothesis 1) and these shared culture elements are in tiirn expected to cascade to lower levels (Hypothesis 3), one indirect way that higher-level leaders can influence ethical conduct at a lower level is by influencing the ethical culture at their own level, which then cascades to lower levels (i.e., L3 ^ C3 -> C2 or L2 -» C2 -* Cl paths). Hypothesis 5. Higher-level ethical leadership indirectly influences ethical culture at lower hierarchical levels through ethical culture at the same (higher) hierarchical levels. Another indirect path linking ethical leadership to ethical culture at lower levels is the conventional leader behavior cascade. Leaders at higher levels indirectly affect the beliefs and behaviors of lowerlevel followers through their subordinate leaders' emulation of their behavior (as shown by the vertical lines connecting L3, L2, and Ll in Figure 1). Drawing on social learning theory (Bandura, 1971), Mayer and colleagues (2009) examined the indirect effects of ethical leadership on members' behavior transmitted through common ethical leadership practices across two organization levels. They asked followers in multiple organizations to each rate both their immediate supervisor's ethical leadership and the ethical leadership of "top management." These ratings were positively correlated, and the ethical leadership of both top management and immediate supervisor were positively related to OCB and negatively related to levels of deviant behavior followers observed in their units. Moreover, the ethical leadership of the immediate supervisor mediated the effects of top management ethical leadership on group deviance and group OCB. Additional research has examined how other leadership constructs, such as transformational
1060
Academy of Management Journal
leadership, correlate across levels (Bass et al., 1987; Chun, Yammarino, Dionne, Sosik, & Moon, 2009; Yang, Zhang, & Tsui, 2010). For example, to explain how the effects of leader behavior cascade across managerial levels, Bass and colleagues (Bass et al., 1987) concluded that junior supervisors mimic the behavior of their senior supervisors, as described by social learning theory (cf, Weiss, 1977), and then become models for their own followers, Chun et al, (2009) found that subordinate leaders' personal identiñcation with their leaders and internalization of their values explained indirect effects of the latter's charismatic and transformational leadership on follower outcomes of satisfaction, helping behavior, and job performance. Normative and informational influences provide another potential avenue for explaining the cascading replication of leaders' behaviors. People tend to conform their behavior to the expectations of others, either to be liked or respected (normative influence) or to be accurate or correct (informational influence) (Cialdini & Trost, 1989). Ethical leaders make it clear that they will not condone unethical behavior while also presenting persuasive arguments as to the beneflts of ethical practices to subordinate leaders (Brown & Treviño, 2006). Whether through behavior modeling, informational influence, or compliance with authority figures, ethical leadership is expected to be positively correlated across hierarchical levels. Given the positive relationship we have already proposed between ethical leadership and ethical culture at the same level (Hypothesis 1), this cascading effect of ethical leadership provides an additional mechanism through which higher-level leaders may indirectly influence the ethical culture in lower-level units (see, e.g., paths L3 ^ L2 ^ C2 and L2 ^ Ll ^ Cl in Figure 1). Hypothesis 6. Ethical leadership at higher organizational levels is indirectly related to ethical culture at lower hierarchical levels through ethical leadership at lower levels. Higher-Level Facilitation of Ethical Leadership To the extent that leader behavior can influence shared understandings about ethical conduct at lower levels, as we have argued above, influence from a higher-level leader may potentially reinforce or countervail the influence that the lower-level leader has on his or her unit. When the higher-level leader is not a proponent of ethical conduct, members of lower-level units may be less likely to develop strong norms and standards for their own ethical behavior (i.e,, strong ethical culture), even
October
though their immediate supervisor exhibits a high level of ethical leadership (Detert & Edmondson, 2011), When the leader of a lower-level unit seeks to develop a more ethical culture in her unit, her own leader may facilitate that by reinforcing the importance of such behaviors to her subordinates. A higher-level leader's influence in facilitating the behavior of the leader at the next lower level is a leadership enhancer (Howell, Dorfman, & Kerr, 1986), which suggests a cross-level interaction predicting ethical culture at the lower level (e.g., L2 X Ll -^ Cl in Figure 1). Hypothesis 7. Ethical leadership at higher hierarchical levels moderates the influence of subordinate ethical leadership: A high level of ethical leadership at a higher organization level is associated with a stronger influence of subordinate leaders' ethical leadership on ethical culture in their units. METHODS Sample and Procedures This study was commissioned by the U.S. Army, which asked us to evaluate the ethical conduct, cognitions, attitudes, and well-being of soldiers diuring their combat deployment in Iraq in May 2009, The surveys were designed by the authors and were administered through the Inspector General (IG; the Army's official investigative office) and chaplains. The aim was to maintain a "chain of custody" with a trusted agent to reassure soldiers that their leaders would not have access to their survey responses. To achieve broad representation, the IG randomly selected two brigades from each of the four Army divisions conducting combat operations in Iraq. Two battalions were randomly selected from each of those two brigades. Three companies were randomly selected from each of those battalions, as were three platoons from those companies and three squads from each platoon. We studied leaders and followers from the lowest three managerial levels in the U,S, Army: the typically nine-person squad (lowest level), the three- to foursquad platoon (middle level), and the three- to fourplatoon company (highest level). Each level had its own leader who reported to the leader at the next higher level. The questionnaires were administered to members of the randomly selected units in person in small assemblies by Army chaplains who traveled between combat outposts in regions with active combat operations. The chaplains received a standardized training course that covered the purpose of the research, protocol, and the survey items. The
Schaubroeck et al.
2012
chaplains returned the completed surveys to the IG, who delivered them to the research team. All respondents provided informed consent, and soldiers understood they were free to decline to participate in this study. The sampling frame included approximately 4 percent of the soldiers in the theater of operations at that time, or about 6,000 servicemen and servicewomen. A total of 2,572 surveys were returned, for an estimated 43 percent response rate. For squadand platoon-level aggregate variables, to ensure adequate representation at each nested organizational level, we designated that platoons must have at least 2 squads reporting, and each squad in turn must have a minimum of 4 members reporting. Of the 322 squads sampled, 280 (90%) met this criterion. Of 130 platoons sampled, 96 (78%) met this criterion. These platoons represented 42 companies. Those units meeting our criteria for the analyses required for this study totaled 2,048 soldiers. Missing data on certain variables (and combinations of variables across levels) limited our final analysis sample sizes to 172 squads, 78 platoons, and 39 companies for cross-level analyses, and 243 squads, 85 platoons, and 40 companies for withinlevel analyses. Mean unit sizes for the latter were 6.09 (s.d. = 2.24) for squads, 3.61 (s.d. = 1.50) for platoons, and 4.57 (s.d. = 1.45) for companies. Most respondents were male (88%), and their average age was 27.5 years. About half (54%) were married, and 48 percent reported having children. All had high school degrees or the equivalent, and 14 percent had college degrees. These statistics approximately mirror the demographic profile of the overall U.S. Army contingent in Iraq (MHAT-IV, 2006). Nearly 68 percent of the respondents were
1061
active duty U.S. Army soldiers. The remainder were activated members of the U.S. Army National Guard or the U.S. Army Reserve. Eight percent were commissioned officers. Measures All outcome variables were reported by squad members and aggregated to the squad level. Predictors and mediators were reported by multiple and separate rating sources at each of the squad, platoon, and company levels, and we aggregated ratings at each level. Squad-, platoon-, and companylevel variables were reported by squad members, squad leaders, and platoon leaders and sergeants, respectively. As shown on the diagonal in Table 1, coefficient alpha reliabilities for the measures were all above .70. We examined the outcomes only at the squad level. These frontline units had the greatest opportunities and risks for serious ethical misconduct arising from contact with combatants, noncombatants, and other operational events. In addition, our research questions concerned only the effects of higher-level leader behavior and ethical culture on outcomes at the lowest organizational level. Outcome variables. Ethical transgressions of different degrees of severity were measured with a set of items used in a previous study of the ethical behavior of U.S. soldiers during previous years of the Iraq war (MHAT IV, 2006). These items seek personal observations of unethical behavior exhibited by members of a respondent's squad. They were adapted for use in this study. The target behaviors presented after the stem "rate the number of times you are aware that members of your
TABLE 1 Variance Components" Variance Components'' Variables
Level 1: Squad
Platoon level Ethical culture Ethical leadership Squad level Ethical culture Ethical leadership Moral efficacy Peer exemplary behavior Transgressions vs. the Army Transgressions vs. noncombatants Report intentions
56.74% 99.82 78,78 60,69 87,12 87,01 72.54
Level 2: Platoon
Level 3: Company
67.85% 87.85
32.15% 12,15
25,52 0.08 21.18 0,03 12.80 2,65 0,01
17.74 0,10 0,04 39.28 0,08 10,35 27,44
° n = 243 for the squad level, n = 85 for the platoon level, and n = 40 for company level, ^ Percentage of variance between groups at different hierarchical levels.
1062
Academy of Management Journal
squad . . . " represented two categories of transgressions, which in this context refer to a total of seven significant and formally punishable organizational actions of two types: transgressions against noncombatants (i.e., "mistreated non-combatants," "unnecessarily damaged non-combatant's property," "killed or injured a non-combatant when it was not necessary") and transgressions against the Army (i.e., "defied the ROE [Rules of Engagement, i.e., the policies given by commanders as to the proper use of force in combat] to accomplish the mission," "lied or falsified reports," "stole," "defied legal orders"). Peer exemplary behavior was measured using items developed specifically for this study. They involve observations of peers undertaking personal risks and engaging in self-sacrifice in ways that are conventionally regarded in the military as virtuous and necessary to accomplish collective goals. Respondents were asked to "rate the number of times they had seen fellow squad members "put themselves in physical danger to protect a fellow Soldier/Marine," "put themselves in physical danger to protect a non-combatant," and "challenge leaders' orders in order to protect a non-combatant." Transgressions and exemplary behavior frequencies were reported on a 0-5 scale ("never" to "five times or more"). Moral efficacy is the belief that one is capable of acting effectively as a moral agent. It was measured with five items developed by Haimah and Avolio (2010) and rated on a scale ranging from 0, "not at all confident," to 10, "totally confident." The items specified an individual perception. Sample items include "I am confident that I can confront others who behave unethically to resolve the issue" and "I am confident that I can determine what needs to be done when I face a moral/ethical decision." We used five items to assess report intentions, which measure an individual's intention to report unethical conduct of other soldiers when it is observed. The items were tailored to the unique context of war in an environment with close proximity to a nonconjbatant population. As with moral efficacy, the items specified an individual perception. A sample item is "I would report a unit member if I saw him or her violating ROEs [rules of engagement]." The other items referred to "injuring or killing an innocent non-combatant," "unnecessarily destroying private property," "mistreating a non-combatant," and "stealing from a noncombatant." Report intentions, together with ethical leadership and ethical cultiu"e (see below), were rated from 1 ("strongly disagree") to 5 ("strongly agree"). Ethical leadership and ethical culture. We measured ethical culture as a reference shift composition
October
construct (i.e., items targeted the tmit level [Chan, 1998]) using the ten-item ethical culture scale developed by Treviño et al. (1998). Sample items include "Penalties for unethical behavior are stricfly enforced in this imit" and "The expressed ethical standards serve as "window dressing" only in this unit" (reverse-coded). Ethical leadership was also measured as a reference shift composition construct, via the ten-item scale developed by Brown et al. (2005). Sample items include "My leader discusses ethics or values with unit members" and "My leader sets an example of how to do things the right way in terms of ethics." One item ("disciplines unit members who violate ethical standards") was omitted from the scale in the analyses because it too closely overlapped with content in the ethical culture scale. Ethical culture and ethical leadership were rated independently at each level. Squad members rated squad ethical leadership and ethical culture; squad and section leaders rated platoon ethical leadership and ethical cultinre; and platoon leaders and platoon sergeants rated company ethical leadership and ethical culture. Combat exposure. All hypothesis tests controlled for combat exposure, which we measured using items adapted from an instrument developed for combat research by Hoge, Castro, Messer, McGurk, Cotting, and Koffinan (2004). Respondents were asked to rate the number of times they had the following experiences (0, "never," to 5, "five times or more"): "an improvised explosive device/booby trap has exploded near you," "received direct fire," "received indirect fire," "have been in a threatening situation where you were unable to respond due to ROE," "other members of your unit were killed or seriously wounded in theater," and "you have personally seen individuals killed or seriously wounded in theater." For control purposes we used the mean of these items for individuals' reports within each unit, at the lowest (squad) level of analysis. We controlled for combat exposure to reduce the likelihood that effects of ethical leadership on our dependent variables could be explained by psychological distress resulting from combat exposure. Analysis Procedure We used hierarchical linear modeling (HLM; Bryk & Raudenbush, 1992) to test the hypothesized relationships among our study variables. Several analyses required simultaneous estimation of regression models at three distinct levels of analysis (company, platoon, and squad) to predict the study outcomes at the first level of analysis. Because our hypotheses concern comparisons among groups both within and between hierarchical levels, we
Schaubroeck et al.
2012
grand-mean-centered the independent variables at each level. For indexes of variance explained, we present a pseudo-/?^ [~R^) statistic (Snijders & Bosker, 1999). This statistic indicates the proportional reduction in the total variance of each variable across levels of analysis that is derived from introducing independent variables at higher levels.
1063
ership and ethical culture. The fit with the data was quite strong (x^[8] = 11.45, p = .18, SRMR = .05, CFI = .99, NNFI = .97). The AVEs for the two latent factors of ethical leadership and ethical culture are .85 and .64, which is substantially higher than the squared correlation between them (.42). Data Aggregation
RESULTS Measurement Models A number of confirmatory factor analyses (CFAs) were conducted to test the distinctiveness of the constructs in the study. Given our focus on squadlevel ethical cognitions and behaviors, we used the mean of squad members' ratings to conduct the first set of CFAs. Because of the relatively small sample sizes, we reduced the total number of indicators in each CFA model by using item parcels as indicators of the latent variables (Landis, Beal, & Tesluk, 2000). We formed item parcels by randomly combining two or three items into one parcel. This yielded a total of 24 indicators for the eight latent variables, which is suitable for a sample of 243 squads. The congeneric measurement model, which specifies that indicators load only onto their corresponding latent variables, fit the data well (A^[224] = 393.86, CFI = .97, NNFI = .97, SRMR = .05). The average variance extracted (AVE) for each latent factor ranges from .56 (transgressions against the Army) to .91 (ethical leadership) with a mean of .74. The AVE for each construct exceeds the conventional criterion of .50 and is larger than the square of any correlation between the constructs, thus supporting the convergent and discriminant validity of the construct measures (Fornell & Larcker, 1981). In addition, we tested alternative models by specifying the indicators of two or more of the constructs as loading onto the same latent variable. None of the possible alternative models fit the data as well as the hypothesized, congeneric model. We then tested a CFA at the platoon level [n = 96) that specified a congeneric model of ethical leadership and ethical culture only, as these are the only factors at the platoon level. The results indicate a good fit for this model (/[8] = 10.43, CFI = .99, NNFI = .99, SRMR = .04). The AVEs for the ethical leadership and ethical culture latent variables are .74 and .91, which is larger than the squared correlation between them (.31). The small sample of companies (n = 42) provides less confidence in the robustness of CFA findings. We nevertheless specified the CFA model at the company level using the same item parcels for ethical lead-
We conceptualized ethical leadership and ethical culture as composition constructs, assessing the extent to which members of a unit share perceptions of these constructs. Thus it was necessary to examine evidence for restricted within-unit variance as a justification for aggregating measures of these constructs to the relevant unit level (James, 1982; Kozlowski & Hattrup, 1992). Two indexes assess restricted within-unit variance: (a) r^^giß, an index of within-group consensus or agreement (James, Demaree, & Wolfe, 1984) and (b) the intraclass correlation coefficient, ICCl, an index of within-group consistency, or interrater reliability (Bliese, 2000). The r^g^ß index is known to be biased downwards when unit sizes are small (e.g., n = 5) (Kozlowski & Hattrup, 1992). Because our units were relatively small, we followed other leadership research (Den Hartog, House, Hanges, & Ruiz-Quintanilla, 1999) and used the maximum observed variance as the "expected value" (Lindell & Brandt, 1997) in the computation of r^g^j^ for each construct within each unit of interest. Kozlowski and Klein characterized this method of justifying aggregation as a "construct-by-group approach" (2000: 35). Median agreement for squad-level composition constructs was .89 for ethical leadership and .84 for ethical culture; platoon-level median agreement was .82 for ethical leadership and .81 for ethical culture; and company-level median agreement was .90 for ethical leadership and .78 for ethical culture. Thus, unit consensus exceeded conventional standards for data aggregation (Bliese, 2000). ICCl indexes the effect size of unit membership, and ICC2 indexes the reliability of an aggregated unit-level variable (Bliese, 2000). Kozlowski and Klein (2000) characterized this theoretical approach to aggregation as a "construct-by-sample approach" (2000: 35), as the indexes are based on the ratio of between-unit and total overall variance, rather than being unit specific. In the case of our sampling plan, however, we have assessments of three different target levels by different raters; thus, the "sample" is different at each level. The ICCl and ICC2 values for ethical leadership and ethical cultural elements were .12 and .36 (ethical leadership), and .14 and .41 (ethical culture) for the squad
1064
Academy of Management Journal
level; 09 and .19 (ethical leadership) and .32 and .52 (ethical culture) for the platoon level; and .17 and .29 (ethical leadership) and .22 and .36 (ethical culture) for the company level. The range of these ICC values is comparable to those reported in the literature and indicates that these data are suitable for aggregation (Bliese, 2000). Whereas ethical leadership and ethical cultural elements were conceptualized as composition constructs that necessitated empirical justification for aggregation, we conceptualized the outcome constructs as compilation variables that are descriptive of individual-level beliefs, intentions, and rates of observations in a given unit. These constructs are not necessarily shared phenomena and thus an examination of restricted within-unit variance is not needed or proper to meaningfully aggregate them to the squad level (Bliese, 2000; Kozlowski & Klein, 2000).
October
levels of analysis, as did squad-level ethical culture. Almost all squad-level variance in ethical leadership, however, was at the level of the squad. We nevertheless proceeded in testing Hypothesis 6, which pertains to platoon ethical leadership effects on squad ethical leadership, because there is substantial variance in the outcomes and predictors of squad ethical leadership, and thus multilevel analyses are required to correctly test the appropriate models (see LaHuis & Ferguson, 2007). Hypothesis Testing Table 2 shows the means, reliabilities, standard deviations, and correlations of the variables. Hypothesis 1 posits direct effects of leaders that are such that ethical leadership will be positively related to ethical culture within each hierarchical level. As Table 3 shows, ethical leadership was associated with ethical culture at the company, platoon, and squad levels [y = .69, p < .01, y = .56, p < .01, and y = .45, p < .01, respectively). Hypothesis 1 is therefore supported. Hypothesis 2 proposes that ethical leadership will be related to outcomes indirectly through ethical culture at the lowest level (e.g., Ll -» Cl ^ 01). Table 4 shows the results of our tests of Hypothesis 2. We foimd that squad ethical culture provided an indirect path from squad ethical leadership to squad ethical outcomes in terms of report intentions (7 = .06, p < .05), moral efficacy (7 = .09, p < .05), transgressions against the Army (7 = -.06, p < .05), and peer exemplary behavior (7 = .11, p < .01). There was no significant indirect effect on transgressions against noncombatants (7
Variance Components In a final preliminary step, we conducted a variance components analysis to examine whether variance at the lower hierarchical levels was adequate to justify multilevel analyses predicting outcomes at these levels. We examined two nested models: platoons nested in companies, and squads nested in platoons and companies (see Table 1). For the platoons nested in companies model, both variables included at this level (ethical leadership and ethical cultinre) showed adequate variance at the platoon level (67.9 and 87.9 percent, respectively). Similarly, all squad-level ethical outcomes displayed adequate variance at the higher and lower
TABLE 2 Means, Standard Deviations, and Correlations among Study Variables" Variable 1. Squad ethical leadership 2. Squad ethical culture 3. Moral efficacy 4. Peer exemplary behavior 5. Transgressions vs. the Army 6. Transgressions vs. noncombatants 7. Report intentions 8. Combat exposure 9. Platoon ethical leadership 10. Platoon ethical culture 11. Company ethical leadership 12. Company ethical culture
Mean s.d. 3.22 2.96 6.53 0.46 0.41 0.26 3.60 0.90 3.40 3.03 3.61 3.32
.96 0.58 0.36 .50* 1.26 .19** 0.41 -.01 0.38 -.28** 0.34 -.17** 0.61 .27** 0.62 -.18** 0.67 .17* .11 0.47 .13 0.62 0.49 .15**
10
11
12
.74
.23** .10
-.30** -.13 .15* -.08 .36** .20* .26** .35**
.95 .01
-.11 -.20** .37** -.07 .10 .06 .05
-.04
.72
.32** .39** -.08 .64** .22**
.11
.12 .00 .07
-.17** -.11 -.12
.82
.69** .35** .01
.80
-.37** .39**
.96
.42**
.03
.08
-.13 -.06
-.05 -.09 -.06
.01
.88
.16* .95 -.01 .57** .87 -.06 .14 .09 .05 .53** .21*
.94 .69
.88
" Rows 1-8 represent within-squad correlations. Rows 9-10 represent correlations between platoon leadership and culture variables and squad variables aggregated to the platoon level. Rows 11-12 represent correlations between company leadership and culture variables and squad and platoon variables aggregated to the company level. Values on the diagonal in italic are alpha reliability estímates. For the company-level correlations, n = 39; for the platoon-level correlations, n = 78; and for the squad-level correlations, n = 172. * p < .05 **p < .01
2012
Schaubroeck et al.
TABLE 3 Relationships between Ethical Leadership and Ethical Gulture within Hierarchical Levels (Hjqpothesis 1)" Dependent Variable: Ethical Culture'' Variables Company-level model Ethical leadership ~R^ Platoon-level model Ethical leadership -R^ Squad-level model Ethical leadership
ß .69**
5.74
.47
.56**
4.18
.24
.45**
6.29
.31
" All effects are standardized. In all the analyses, combat exposure was entered as a control variable (not included in this table). For the squad level, n = 243; for the platoon level, n = 85; and for the company level, Í7 = 40. ^ Ethical culture refers to the culture at the same level as the independent variable; e.g., for the company-level model, tbe dependent variable is company ethical culture. **p < .01
= - . 0 1 , p < .20). Hypothesis 2 was therefore supported for all outcomes except one. In addition to these indirect effects, ethical leadership also had significant direct effects on report intentions (7 = .14, p < .05) and transgressions against the Army (7 = -.26, p < ,05). Hypothesis 3 predicts a positive relationship between higher- and lower-level ethical culture (C3 -^ C2 and C2 —> Cl). To more rigorously test the
1065
cascading relationships of ethical culture as represented in Hypothesis 3 and also to test subsequent hypotheses involving multilevel effects, we simultaneously estimated the effects of the higher-level variables (platoon and company ethical culture) on corresponding variables measured at the lower levels. This relationship was significant for ethical culture in all three hierarchical paths: company to platoon (7 = ,26, p < ,05), company to squad (7 = .37, p < .01), and platoon to squad (7 = ,23, p < .01). Hypothesis 3, pertaining to the cascading effects of ethical culture, is therefore supported. Table 5 presents a simimary of these analyses. Hypothesis 4 predicts a positive relationship between higher-level ethical leadership and lowerlevel ethical culture (L3 -^ C2 or L2 -^ Cl). Table 6 displays the results of a series of models testing this hypothesis. Higher-level ethical leadership was related to ethical culture at lower levels (7 = ,21, p < .10, for the company to platoon model; 7 = .22, p < ,05, for the company to squad model; and 7 = ,41, p < .01, for the platoon to squad model). These findings together support Hypothesis 4. Hypotheses 5 and 6 elaborate Hypothesis 4 by specifying indirect relationships between ethical leadership and lower-level ethical culture transmitted through ethical culture at the same (higher) level (H5; L3 ^ C3 -» C2 or L2 ^ C2 -^ Cl) and through lower-level ethical leadership (H6; L3 -^ L2 -^ C2 or L2 ^ Ll ^ Cl). In tests of Hypothesis 5, only one out of the three possible indirect rela-
TABLE 4 Ethical Leadership Effects on Squad Member Outcomes tbrough Squad Etbical Culture (Hypotbesis 2)°
Variables Squad-level main effects Ethical leadership ~R^ Ethical culture ~R^ Mediated model Ethical leadership Ethical culture ~R^ Decomposition of ethical leadership effects Total Direct Indirect
ß .20**
ß
3.23
.14
.22**
3.06
0.75
.17**
2.24 1.90
.20 .14
1.84
-.05 .20*
ß -.14*
* -2.02
.17
2.29
.01
.15
.06*
,06
t
.00
.14
.14* ,14*
Transgressions vs. Noncombatants
Moral EfiBcacy
Report Intentions
-.10 -.12 -.03
ß
t
-.32**
-4.33
.22
-1.52
.16
-0.50 2.02
Transgressions vs. the Army
-.28** -.26* -.13*
ß .07
-4.25
.22**
-2.56 -1.78
.17 .24**
.17
.24
.26
.04
-.13 -.12 -.01
-.32 -.26 -.06*
.28 .17
1.95
0.43
1.73
.11**
All coefficients are standardized. Combat exposure was included as a control variable (not sbown in this table), n = 243. * p < .05 **p < .01
0.90 3.08
.26
.01
-.05 .09*
t
.22
.19
-1.38 -0.43
Peer Exemplary Behavior
-0.38 3.21
2.88
October
Academy of Management Journal
1066
TABLE 5 Standardized Relationships between Higher- and Lower-Level Ethical Culture (H5rpothesis 3)" Dependent Variable: Lower-Level Ethical Culture"'
TABLE 6 Standardized Effects of Higher-Level Ethical Leadership on Lower-Level Ethical Culture (Hypotheses 4 and 5)" Dependent Variable: Lower-Level Ethical Culture""
Variables Company to platoon model Company ethical culture Company to squad model Company ethical culture Platoon to squad model Platoon ethical culture
Variables .26* .04
2.31
,37** .18
5,30
,23* .10
3,12
ß
Company to platoon model—Main effects Company ethical leadership Company to squad model—Main effects Company ethical leadership
" All coefficients are standardized. Combat exposure is included as a control variable (not shown in this table), n = 7Ö for the platoon level and JI = 39 for the company level, ^ Lower-level ethical culture refers to the lower level of the model tested; e,g,, for the company to squad model, the dependent variable is squad-level ethical culture. * p < ,05 ** p < ,01
tionships was significant. Specifically, company ethical leadership was indirectly related to squadlevel ethical culture through company ethical culture. There was no significant indirect effect of higher-level ethical leadership on lower-level ethical culture for the company to platoon or the platoon to squad models. Thus, Hypothesis 5 is only partially supported. Regarding Hypothesis 6, the indirect path from company-level ethical leadership to platoon-level ethical culture through platoon-level ethical leadership was significant (y = .11, p < .05). The indirect path from platoonlevel ethical leadership to squad-level ethical culture through squad ethical leadership was also significant (y = .08, p < .01). Table 7 presents these findings, which overall support Hypothesis 6. In supplementary analyses, we examined whether ethical leadership has any direct influence on lower-level ethical culture after the indirect effects specified in Hypotheses 5 and 6 have been accounted for. Accounting for the cascading influence of company and platoon ethical culture (Hypothesis 5) fully explained the direct effect of company ethical leadership on squad ethical culture. However, the direct influence of platoon ethical leadership on squad ethical culture remained, even after we further accounted for the indirect effect through squad ethical leadership (Hypothesis 6). Insofar as neither lower-level ethical leadership nor same-level ethical culture explain this platoon to squad relationship, it can legitimately be labeled a "bypass effect."
1,78 .01
Platoon to squad model—Main effects Platoon ethical leadership
22* 07
2,55
41**
5.26
18
Company to platoon model—Indirect effects
Company ethical leadership Company ethical culture Decomposition of company ethical leadership effects Total Direct Indirect Company to squad model—Indirect
03 25 01
21 03 15
-0.09 1.18
1,17
effects
Company ethical leadership Company ethical culture
03 41**
-0.46 3,88
18
Decomposition of company ethical leadership effects Total Direct Indirect Platoon to squad effects—Indirect
22 03 28**
3,24
35**
3.72 1.13
effects
Platoon ethical leadership Platoon ethical culture Decomposition of platoon ethical leadership effects Total Direct Indirect
10 19
41 35 05
1,12
" Combat exposure was entered as a control variable (not shown in this table), n = 172 for the squad level, n = 78 for tbe platoon level, and n = 39 for the company level. ^ Lower-level ethical culture refers to the lower level of the model tested; e.g., for the company to squad model, the dependent variable is squad-level ethical culture, ^p < ,10 * p < ,05 ** p < ,01
Hypothesis 7 proposes that ethical leadership at higher hierarchical levels will moderate the relationship between lower-level ethical leadership and lower-level ethical culture. As shown in Table
2012
Schaubroeck et al.
1067
TABLE 7 Standardized Higber-Level Etbical Leadersbip Effects on Lower-Level Etbical Culture tbrougb Lower-Level Etbical Leadersbip (Hypotbesis 6)" Dependent Variable: Lower-Level Ethical Culture"" Variables Company to platoon model—Main effects Company ethical leadership -R^ Company to squad model—Main effects Company ethical leadership ~R^ Platoon to squad model—Main effects Platoon etbical leadership
ß .20^ .02
1.69
.15 .03
1.39
22* 07
2.55
.22*' .05
2.99
41** 18
5.26
08 54** 33
0.78 4.10
Company to platoon model—Indirect effects Company ethical leadership Platoon ethical leadership Decomposition of ethical leadership effects Total Direct Indirect Company to squad model—Indirect effects Company ethical leadership Squad ethical leadership Decomposition of ethical leadership effects Total Direct Indirect Platoon to squad model—Indirect effects Platoon ethical leadership Squad ethical leadership ~R^ Decomposition of ethical leadership effects Total Direct Indirect
1.78 .01
20 08 11* 19** 45** 18 45 45 03 32** 42** 20 41 32 08**
1.58 2.80 6.31
1.37 4.60 5.99
2.71
" Combat exposure was entered as a control variable (not shown in this table), n = 172 for the squad level, n = 76 for the platoon level, and /I = 39 for the company level. ^ Lower-level ethical culture refers to the lower level of the model tested; e.g., for the company to squad model, the dependent variable is squad-level ethical culture, ^ p < .10 * p < .05 **p < .01
8, a significant moderating effect was found for both the company to platoon and the platoon to squad model (interaction y = .22, p < .05, and y — .08, p < .05, respectively). Specifically, although the effects of lower-level ethical leadership are positive and significant in each case, the slopes are significantly stronger when higher-level ethical leadership is high rather than low, thus supporting Hypothesis 7. Figures 2 and 3 illustrate these relationships. To follow up on our analyses of Hypothesis 7, we conducted two supplementary analyses. First, we examined whether higher-level ethical culture moderates the relationship between lower-level
ethical leadership and ethical culture at the lower level (similar to the hypothesized moderating role of higher-level ethical leadership). Second, we assessed whether the moderating effects of higherlevel ethical leadership discussed in Hypothesis 7 are mediated through higher-level ethical culture. Higher-level ethical culture was found to moderate the effects of lower-level ethical leadership in a way that was similar to the moderating effect of higher-level ethical leadership. Platoon-level ethical leadership and squad-level ethical leadership exhibit a stronger relationship to platoon and squad ethical culture when company and platoon ethical culture levels are both high (interaction y = .34, p
October
Academy of Management Journal
1068
TABLE 8 Interactions of Higher- and Lower-Level Ethical Leadership Predicting Lower-Level Ethical Culture (Hypothesis 7)" Dependent Variable: Lower-Level Ethical Culture'' Variables
f
Y
Company to platoon model Company ethical leadership Platoon ethical leadership Cross-level interaction term
.05
.39** .22*
~R^
.36
Company to squad model Company ethical leadership Squad ethical leadership Cross-level interaction term
.11** .27**
Platoon to squad model Platoon ethical leadership Squad ethical leadership Cross-level interaction term
0.54 5.29 2.01
ical culture than for low ethical culture (p < .05); Figures 4 and 5 illustrate. Findings from this set of mediated moderation analyses indicate higherlevel ethical culture mediates the moderating effects of higher-level ethical leadership. This suggests that higher-level ethical culture, as a more proximal moderator of lower-level ethical leadership effects, is a mechanism that may more fully explain the effects of ethical leadership that were specified in Hypothesis 7. DISCUSSION
.04 .34
.17** .26** .08*
2.74 5.57 0.68 4.77 6.69 2.06
.37
" Combat exposure was entered as a control variable (not shown in this table), n = 172 for the squad level, n = 78 for the platoon level, and n = 39 for the company level. '' Lower-level ethical cultiu-e refers to the lower level of the model tested; e.g., for the company to squad model, the dependent variable is squad-level ethical culture. * p < .05 ** p < .01
< .05, platoon, and y = .16, p < .05, squad, respectively). The simple slopes, although significant in each case, were significantly stronger for high eth-
Ethical leadership represents a domain of leader behavior that accords with what Schein (1985, 2010) described as primary mechanisms through which leaders embed their expectations and assumptions into the fabric of an organization. Schein's model, however, does not precisely specify how these embedded understandings and expectations of leaders become suffused through multiple organizational levels and influence follower outcomes. Our study builds on Schein's theoretical framework in a way that begins to lay out the specific modes of transmission through which ethical leadership at different hierarchical levels may permeate an organization. In our model, the role of ethical culture is essential to understanding these relationships. We described and tested three basic categories of ethical influence transmission down and across an organizational hierarchy. First, we found that
FIGURE 2 Interaction of Company-Level Ethical Leadership and Platoon-Level Ethical Leadership Predicting Platoon-Level Ethical Culture" 3.5
3.3
Platoon Ethical Culture
3.1
2.9
2.7
2.5
Low Platoon Ethical Leadership High Platoon Ethical Leadership •Low company ethical leadership
High company ethical leadership
" For low company ethical leadership, simple slope = .26, f = 2.54, p < .05. For high company ethical leadership, simple slope = .53, Í = 4.29, p < .01.
2012
Schaubroeck et al.
1069
FIGURE 3 Interaction of Platoon-Level Ethical Leadership and Squad-Level Ethical Leadership Predicting Squad-Level Ethical Culture* 3.5
3.3
3.1
Squad Ethical Culture
2.9
2.7
2.5
Low Squad Ethical Leadership • Low platoon ethical leadership
High Squad Ethical Leadership • • •• • High platoon ethical leadership
" For low platoon ethical leadership, simple slope = .20, t = 3.81, p < .01. For high platoon ethical leadership, simple slope = .31. t = 8.07, p < .01.
FIGURE 4 Interaction of Company-Level Ethical Culture and Platoon-Level Ethical Leadership Predicting Platoon-Level Ethical Culture" 3.5
3.3
Platoon Ethical Culture
3.1
2.9
2.7
2.5
Low Platoon Ethical Leadership
High Platoon Ethical Leadership
»Low company ethical culture • • #t • High company ethical culture ° For low company ethical culture, simple slope = .21, í = 2.12, p < .05. For high company ethical culture, simple slope = .56, í = 4.12, p < .01.
Academy of Management Journal
1070
October
FIGURE 5 Interaction of Platoon-Level Ethical Culture and Squad-Level Ethical Culture Predicting Squad-Level Ethical Culture" 3.5
3.3
3.1 Squad Ethical Culture
2.9
2.7
2.5
Low Squad Ethical Leadership »Low platoon ethical culture
High Squad Ethical Leadership • • •• • High platoon ethical cultin-e
" For low platoon ethical cultiire, simple slope = .06, f = 1.02, p > .20. For high platoon ethical culture, simple slope = .23, f = 3.90, p < .01.
ethical leaders embed their expectations into the ethical culture that is shared and understood by members of their own units (Hypothesis 1). Through this mechanism, ethical leaders indirecfly influence their immediate followers' cognifions (e.g., moral efficacy) and behaviors (e.g., transgressions) through unit ethical culture (Hypothesis 2). Thus, ethical leadership appears to influence ethical culture within each hierarchical level, and much of the influence of leaders at each level is indirect, transmitted through ethical culture at the same hierarchical level. Second, ethical culture cascades across hierarchical levels to a substantial extent (Hypothesis 3), thus enabling a new and previously unstudied pathway for leadership influence to transmit across levels. Further, we predicted (Hypothesis 4) and found that ethical leadership is positively related to ethical culture at lower levels. Direct influences of leader behavior on the behavior of followers at lower levels have been examined previously (e,g,, Dvir et al,, 2002), Of greater interest in this study was to determine whether ethical leadership's influence is embedded through effects of ethical culture at the same level (Hypothesis 5) and/or through effects of lower-level ethical leadership (Hypothesis 6) in ways that explain these overall cross-level effects of ethical leadership on
the lower-level ethical culture. Both hypotheses were partially supported, and of three overall crosslevel direct effects stated in Hypothesis 4, only one—the connection between platoon ethical leadership and squad ethical culture—was not explained by indirect processes transmitted through lower-level leadership or higher-level ethical cu.lture. Third, we found that leaders who exhibit a high level of ethical leadership may facilitate the influence of subordinate leaders' ethical leadership on their followers (Hypothesis 7). Post hoc analyses showed that ethical culture at the higher level mediated the facilitating effect of higher-level ethical leadership. This highlights the importance of congruence in ethical leadership across multiple levels and how ethical leaders depend on support from their principals. Such cross-level interactions are thought to be critical for understanding complex systems (Page, 2007), but they have rarely been examined generally, nor specifically with respect to ethical leadership and culture. These interactions and the direct and indirect effects of ethical leadership transmitted through ethical culture indicate that a more holistic view of both proximal and distal leadership processes that impact follower behavior is required to more fully appreciate
2012
Schaubroeck et al.
the locus and transmission of leadership influence in complex organizations. Table 9 summarizes all the findings concerning the hypotheses. The most interesting of our findings was that local leadership, higher-level ethical culture, and higher-level ethical leadership influence local understandings of ethical culture. These findings suggest that much more attention should be paid to the role of shared understandings about ethical conduct (ethical culture) at each hierarchical level in an organization. A Multilevel Systems Perspective on Ethical Leadership and Ethical Culture From a meta-theoretical perspective, the notion of organizations as social entities of interacting and coacting elements operating at multiple levels of a system is one with a long history in the organizational sciences, one going back at least to Roethlisberger and Dickson (1939) and apparent in the writings of the many systems theorists of the 1950s and 1960s (e.g., Likert, 1961; Katz & Kahn, 1966). A persistent theme in the writings of these and other scholars is the central role of leaders in the enactment of a system of human relations across levels of
1071
organizations. Likert's concept of leaders as linking pins who transmit values and shape cultural integration goes to the core of "classic" organizational systems theory. Although leadership is widely viewed as central for linking organizational systems together, for the most part it has not been researcbed from a multilevel theory and integrated systems perspective (Yammarino & Dansereau, 2008). Such research is critical to evaluating system cohesion as a fundamental assumption of organizational behavior (Kozlowski & Klein, 2000). Our model included three levels of leadership in a complex organization which, combined with a large sample size, allowed us to establish how ethical leadership becomes embedded in a multilevel, multiperson system with both direct effects of leaders on their immediate followers and indirect, cross-level effects on lower-level organization members. Researcb and scbolarship about leadership has thus far examined the processes through which leaders suffuse their influence in an organization primarily by referencing the trickle-down relationships between leaders' behavior at different levels (Bass et al., 1987; Mayer et al., 2009) and/or direct effects of behavior across levels (e.g., Dvir et al., 2002). By integrating horizontal (within-level)
TABLE 9 Summary of Hypothesis Test Results" Hypothesis
Figure 1 Relationships
HI. Ethical leadership is positively related to ethical culture within each hierarchical level. H2. Through ethical culture, etbical leadership is indirectly and positively related to unit members' moral agency (moral efficacy and report intentions) and exemplary ethical behavior and negatively related to their ethical transgressions, H3, Ethical culture at higher hierarchical levels is positively related to ethical culture at lower hierarchical levels,
S S Ll^Cl Ll -^ Cl ^ 01 (moral efficacy) Ll —> Cl ^ 01 (report intentions) Ll —> Cl —>• 0 1 (transgressions vs. Army) Ll -» Cl -» 0 1 (transgressions v, NC) Ll —> Cl - • 0 1 (exemplary behavior) C3 ^ C 2 C3 ^ C l
H4. Ethical leadership is positively related to ethical culture at lower hierarchical levels. H5, Higher-level ethical leadership indirectly influences etbical culture at lower hierarchical levels through ethical culture at the same (higher) hierarchical levels, H6, Etbical leadership at higher organizational levels is indirectly related to ethical culture at lower hierarchical levels through ethical leadership at lower levels, H7, Ethical leadership at higher hierarchical levels moderates the influence of subordinate ethical leadership in such a way that high etbical leadership at a higher organization level is associated with a stronger influence of subordinate leaders' etbical leadership on ethical culture in their units.
Results
S S N S S S S S
s s s
L2->C1 L3 ^ C3 -> C2 L3 ^ C3 -^ Cl L2 ^ C2 -^ Cl L3 ^ L2 -* C2 L2 -• Ll ^ Cl
N S N S S
L3 X L2 •C2 L2 X Ll •Cl
S S
" See Figure 1 for path diagram. Abbreviations: C = "etbical culture"; L = "ethical leadership"; N = "not supported"; NC = "noncombatants"; O = "outcomes"; S = "supported." Numbers after abbreviations denote hierarchical level; e.g., L3 denotes ethical leadership at company (3rd) level.
1072
Academy of Management Journal
influence processes with vertical and diagonal cross-level linkages, we are proposing ways not only to study ethical leadership in organizations, but also to highlight the different transmission mechanisms for leadership constructs more generally. Horizontal (within-level) effects. We found strong horizontal, within-level effects of ethical leadership on ethical culture at each of the three levels. Such results are consistent with the widely held assiunption that the ethical leadership of an individual's immediate supervisor affects the proximal ethical culture of the individual's unit. Further, at the lowest level, squad-level ethical culture was significantly related to fovir out of five of the outcome measures, including report intentions, moral efficacy, transgressions against the Army, and exemplary ethical behavior. These simple horizontal effects indicate the value of selecting and developing ethical leaders at lower levels who proactively maintain favorable ethical norms and uphold high standards for ethical conduct in their own units. Vertical direct and indirect effects of embedded ethical culture and leadership. Our model specifying the indirect transmission of leadership influence has two core elements: First, leader behaviors at higher levels tend to be similar to leader behaviors at adjacent lower levels. Second, we observed that more senior leaders influence lowerlevel leaders' and followers' behaviors and cognitions via indirect pathways. One of these indirect pathways observed in the c\n"rent study was through the ethical culture at adjacent lower levels. Ethical culture at the highest hierarchical level (company, in this study) was positively related to ethical culture at both lower levels, as per Hypothesis 3. This served as a foundation for subsequent hypotheses concerning indirect effects of ethical leadership on lower-level outcomes. Cross-level indirect effects. In addition to the vertical and horizontal effects proposed in ovir model, the diagonal dashed lines in Figure 1 represent direct effects of higher-level ethical leadership on lower-level ethical culture. Company-level ethical leadership exhibited significant diagonal effects on ethical culture at both the platoon and the squad levels, and platoon-level ethical leadership also directly influenced squad ethical culture (Hypothesis 4). This relationship would conventionally be interpreted as a bypass of the lower-level leaders, but we suggest that is a rather liberal interpretation unless one also accounts for mechanisms that may explain such direct cross-level effects, such as mediation through lower-level leaders' behavior or through the cascading influence of shared culture. We found that the indirect relationship of
October
company ethical leadership through platoon ethical leadership fully accounted for the relationship between the former and platoon ethical culture. In addition, accounting for the cascading influences of ethical culture explained the aforementioned direct effect of company ethical leadership on squad ethical culture. The authors of one previous study observed bypass effects of leader behavior after accounting first for cascading leadership to the next level (Yang et al., 2010), but no previous study bas simultaneously accounted for indirect pathways through shared cultural elements in which leaders embed their influence. This suggests the possibility that bypass effects of leader influence across levels may be at best a weak phenomenon once all embedding processes (indirect relationships through lower-level leaders and same-level culture elements) are taken fully into consideration. One direct cross-level relationship, that between platoon ethical leadership and squad ethical cultvue, persisted after we accounted for the other hypothesized influences on squad ethical culture (i.e., the simultaneous effects of L2 ^ Ll —> Cl and L2 —> C2 —» Cl). Even using ovu- more conservative definition and test, this qualifies as a leadership bypass. This finding is similar to the results presented by Dvir et al. (2002) in that the platoon leaders' transformational leadership in their study had a significant effect on their indirect followers' behavior, but not on the behavior of their immediate followers (squad leaders). Explaining direct effects of this nature remains an interesting question for future research. One interpretation may be that senior leaders serve as more powerful role models when they are similar to and represent a potential possible self for subordinate leaders (Bandura, 1971). In the current study, the leaders at level 1 (squad leaders) were all noncommissioned officers (NCOs), whereas the formal leaders at level 2 (platoon leaders) were commissioned officers. The most salient role models to the squad leaders may have instead been the informal leaders who were the more senior NCOs in the organization (i.e., platoon sergeants and first [company] sergeants). This alternative senior leadership referent may have attenuated squad leaders' emulation of their platoon leaders, which in turn enabled platoon leaders to influence squad ethical cultme in distinct ways. As we noted above under the heading "Variance components," although platoon ethical leadership explained a significant portion of the between-squad variance at tbe platoon level, the total variance of squad ethical leadership at the platoon level was very small. Outcomes of ethical leadership and ethical cul-
2012
Schaubroeck et al.
ture. Our study examined the extent to which ethical leadership at different levels, both directly and through ethical culture, is associated with ethical transgressions, exemplary behaviors, and the moral agency cognitions of moral efficacy and reporting intentions. In measuring transgressions and peer exemplary behavior, we aggregated individuals' reports about frequencies of certain behaviors occvirring among members of their squads. We expected this approach to enhance reliability of reporting and encourage more honest reporting than might be expected if soldiers were reporting about themselves or specific peers. Transgressions included mistreatment of noncombatants as well as the more general rule breaking that typifies behavior in many other organizations, such as thievery, misreporting performance, and defying directives. Peer exemplary behavior and moral efficacy showed the strongest positive relationships with squad ethical culture, and transgressions against the organization (e.g., stealing) were negatively related to squadlevel ethical culture. Moral efficacy and report intentions are manifestations of moral agency, reflecting orientations to respond in purposeful ways to opportunities to confront unethical behavior (Hannah et al., 2011). Ethical context may influence these dispositions, as we observed significant, positive effects of both squad- and platoon-level ethical culture on squad mean levels of moral efficacy and report intentions. To ovu" knowledge, this is the first empirical study linking ethical leadership to such constructs of moral agency. To better test causality, future research might profitably examine baseline levels of moral efficacy and report intentions before followers are exposed to particular levels of ethical leadership and ethical culture that may encoiuage them to be more (or less) effective moral agents.
Importance of Context: Ethical Culture and Multilevel Leadership In keeping with the embedding of leadership as described by Schein (1985, 2010), much of the influence of ethical leadership on ethical outcomes that was observed in this study was mediated by unit-level ethical culture. Consequently, models of leadership and ethical behavior that omit the effects of ethical culture at different hierarchical levels may be underspecified. For example, one might conclude that senior leaders have a direct influence on outcomes at a lower level that results from direct interactions between these leaders and lower-level followers (Yammarino, 1994), whereas the influence of these leaders may in fact be indirect, trans-
1073
mitted through their influence on culture at their own levels, which then cascades to lower levels. This study also begins to address calls for examining leadership in context (e.g., Avolio, 2007); in this instance, shared cultural elements (Schein, 1985) relating to ethical conduct represents context. Importantly, the effects of lower-level ethical leadership on ethical culture at the same level were contingent on ethical leadership at the higher level. Ethical leadership at a lower level had a stronger positive influence on ethical culture at that level when the leader at the next higher level was reported to exhibit a high level of ethical leadership. Additional analyses indicated that this facilitating effect of higher-level ethical leadership may occur through the mediating role of ethical culture. Specifically, ethical leadership at higher levels promotes ethical culture at those levels, which in turn facilitates a more favorable impact of ethical leadership on ethical culture at the next lower level. Thus, when studying ethical leadership in the conventional way, focusing chiefly on the influences of the immediate leader, one can potentially be misled into drawing inferences about the magnitude of these effects, when in fact they may be to some extent contingent on qualities of ethical leadership at higher levels. This moderating effect of higherlevel ethical leadership suggests that senior leaders play a significant role in cultivating shared ethical understandings at lower levels, working not only through direct effects and indirect influences through ethical culture, but also by constraining or facilitating the influence of ethical leadership at lower levels. Practical Implications Leaders should be encouraged to embed their assumptions and expectations concerning ethical conduct among organizational members by engaging in ethical leadership and by seeking to create strong ethical cultures in their units. Through the mechanisms examined in the current study, this leadership strategy may influence ethical culture at lower levels. As conceptualized by Brown and colleagues (2005), ethical leadership involves many of the primary and secondary mechanisms that Schein (1985, 2010) noted are critical for a leader to embed influence. Specifically, ethical leaders pay attention to providing contingent consequences for desired eind undesired behavior. They serve as models and coaches to followers in ways that encourage them to exhibit exemplary ethical behavior and to avoid unethical behavior. Critical incidents concerning ethical conduct, such as severe employee infractions, can be highlighted as a key
1074
Academy of Management Journal
means to teach followers about ethical standards and sanctions and to reinforce the message that they can be effective moral managers and agents. Ethical leaders can also allocate resources to improve followers' ethical understandings, while also using levers such as selection, promotion, training, and performance appraisals in ways that reinforce these actions. An ethical leader can further reinforce such understandings by utilizing Schein's secondary mechanisms, such as telling engaging stories about ethics that are likely to be repeated by others, or using formal means, such as speeches and policies, to communicate expectations. Using the concept of ethical culture was helpful to our understanding and ability to explain the influences of ethical leaders within and across levels. Organizational leaders might pay more attention to the role that ethical culture plays in establishing the standards and boundaries for ethical conduct in their organizations. These shared elements can be tracked by surveying organization members to determine whether leaders are creating unit ethical cultures that support ethical behavior or whether they are creating obstacles to it that might reduce followers' moral agency and increase the frequency and severity of unethical conduct. Another important practical implication of this study is that to understand the influence of a direct leader, it may be necessary to look up one or more levels in an organization to obtain a more complete picture of the locus ofthat leader's style and effects. Our results suggest that building the full leadership capacity of an organization requires viewing leadership more as an integrated system of relationships that operate across hierarchical levels, driven substantially by both leader behaviors and ethical culture that are present within and across levels. It is also important for leaders to understand how their influence on their direct reports who are leaders themselves can be used to transmit and reinforce their indirect leadership to followers at lower levels of their organization. Such leadership may focus on aligning performance and ethical expectations or emphasize the importance of a particular "guiding" message that the more senior leader wants all lower-level leaders and organization members to embrace. To lead, therefore, requires that an individual think about both direct and indirect impacts as represented through his or her subordinate leaders and the assumptions and expectations the individual is embedding in the ethical culture. Furthermore, knowing how senior leaders can facilitate the effects of subordinate leaders' ethical leadership by exhibiting strong ethical leadership themselves reinforces how important it is for organizations to ensure senior leaders
October
are selected and developed to exhibit high levels of ethical leadership. Study Strengths and Limitations A number of distinct empirical strengths are associated with this study. First, unlike most prior studies of leadership transmission across levels (e.g., Chun et al., 2009; Mayer et al., 2009), we used independent ratings of leader behavior at each organizational level. This may at least partially explain why the leader behavior cascade effects in our study are somewhat weaker than those found in previous studies. We also encourage researchers to obtain independent measures of ethical culture at different levels, as we did. Relying upon the same lower-level participants to evaluate shared understandings about cultural elements (e.g., Zohar & Luria, 2005) or the behavior of leaders (e.g., Mayer et al., 2009) at multiple higher levels potentially inflates levels of agreement. This inflation of observed relationships may occur because of samesource bias but also because participants at lower levels may lack adequate information with which to gauge the behavior of leaders and the cultures of units at higher levels. By measuring ethical culture using distinct sources, we were also able to identify very satisfactory levels of discriminant validity of ethical culture at different levels. Whereas a general organizational culture may explain a portion of the variance in ethical culture shared across levels, these cross-level relationships were low to moderate in magnitude, indicating that global organizational culture could not explain the homogeneity within units. We also examined whether variance in company (platoon) ethical leadership and in ethical culture ratings was related to variance in platoon (squad) ethical leadership or platoon (squad) ethical culture ratings. The highest correlation, which concerned the relationship between the variances of squad and platoon ethical culture ratings, was quite low (r = .15). Second, unlike earlier multilevel leadership studies, our analyses related to ethical leadership also accounted for the effects of subunit ethical culture. Not only did the influences of ethical culture from higher levels explain additional variance in outcomes, but also, after controlling for these elements, we no longer observed some direct (bypass) effects of higher-level leadership on outcomes that have been the subject of previous research and theory (Yammarino, 1994; Yang et al., 2010). Third, although the effects of ethical leadership transmitting through lower-level leaders' behavior explained only a small portion of
2012
Schaubroeck et al.
the variance in outcomes among lower-level members, when also considering indirect ethical leadership effects that were transmitted through ethical culture at different levels, we found that higher-level ethical leadership was broadly related to squad members' ethical cognitions and behavior. A fourth advantage of our study involved our examination of the various links in the chain of influence from company-level leader behavior to squad-level beliefs and behavior. We took a conservative approach toward examining these paths, by taking only two-segment paths (e.g., variable A to variable B and variable B to variable C) that were found to represent statistically significant indirect effects when we calculated total indirect effects. We pursued this strategy because sampling variation around the indirect effects is too substantial in a multilevel study of this scope to cumulate indirect effects that were expanded across multiple bivariate relationships. As with any field research project, however, a number of limitations could impact the interpretation of our findings. One issue is the potential to generalize the findings to other organizational contexts. Over 89 million people serve in militaries worldwide (International Institute for Strategic Studies, 2010), Police officers and firefighters, who constitute a large segment of the global workforce, in some ways share the often extreme context of military work (Hannah, Uhl-Bien, Avolio, & Cavarretta, 2009). Nevertheless, the U.S. military, like other militaries around the world, is a unique organization. Additional research will need to be conducted in other types of organizations before one can be confident as to the generalizability of our findings. For example, militaries have hierarchical structures, and thus it would be useful to study more organically structured organizations to determine if the means by which higher-level ethical leadership is conveyed to lower-level followers is different in such organizations. Another example of how organizational context may influence the generalizability of our findings concerns leader selection. Yammarino (1994) noted that to the extent that leaders play a role in selecting their subordinate leaders, they may be inclined to choose individuals who are more likely to behave in a manner similar to how they themselves would behave. Subordinate leaders may also tend to have a closer relationship with leaders who have selected them, and this close interaction may promote behavioral similarity. In the U.S. Army, however, at lower levels of leadership the assignments of leaders are normally made at higher
1075
levels. Future studies examining samples in which such selection processes are more prevalent may be able to determine if leaders' selection of their immediate junior leaders enhances the cascading of leader behavior. A number of unmeasured variables may help to explain the pattern of results we reported. For example, it would be useful in future research to incorporate other potential influences on unit ethical culture, such as the presence or absence of formal ethical codes at each level. In addition, other potential mechanisms could be assessed in future research, such as leaders' influencing the identities of followers at lower levels (Chun et al,, 2009). Future studies may also examine whether disagreement among members about cultural elements, as reflected in within-unit variance on such measures (Zohar & Luria, 2005), is itself a moderator variable that influences the strength of cascades of such elements or relationships between leader behavior and the mean levels of cultural elements of a given unit or lowerlevel units. Given that our data collection was based on a cross-sectional design, we cannot make definitive inferences about causal processes. To address alternative models, we conducted supplementary analyses in which the order of ethical leadership and ethical culture was varied at each level. These analyses determined that the direct effects of ethical culttire on squad-level ethical outcomes were stronger than the effects of ethical leadership. Nevertheless, we cannot rule out alternative causal orders and mechanisms. In addition, the dependent variables examined in this study were obtained from self-reported data provided by squad members. Although there was reasonable agreement among members of a squad as to the frequency with which their members engaged in transgressions or exemplary behaviors, we cannot be certain as to the validity of their reports, and it is likely that some transgressions were not reported accurately or were underreported. Aggregating to the squad level, however, made it less likely that individuals' idiosyncratic attributions about the same observed behaviors would bias the study findings. Conclusions
We developed a multilevel model to guide a study of the effects of ethical leadership and shared understandings about ethical conduct (ethical culture) on lower-level followers' ethical cognitions and behavior. Our findings indicate that ethical cognitions and conduct reflect multiple normative and informational influences, including not only
Academy of Management Journal
1076
ethical leadership and ethical culture inherent in an immediate unit, but also reflecting the influence of leaders and ethical culture at higher levels. The understandings concerning norms, standards, and sanctions for ethical behavior (that is, ethical culture), which reflected the expectations and desires of leaders at various levels, were found to be the most potent and proximal influence on follower cognitions and behavior. REFERENCES Arnaud, A., & Schminke, M. In press. The ethical climate and context of organizations: A comprehensive model. Organization Science, orsc.1110.0698. Avolio, B. J. 2007. Promoting more integrative strategies for leadership theory building. American Psychologist, 62: 25-33. Bandura, A. 1971. Social learning theory. Morristown, NJ: General Learning Press. Bass, B., & Bass R. 2008. The Bass handhook of leadership: Theory, research, and managerial applications. New York: Free Press. Bass, B. M., Waldman, D. A., Avolio, B. I., & Bebb, M. 1987. Transformational leadership and the falling dominoes effect. Group & Organization Studies, 12: 73-87. Bliese, P. D. 2000. Within-group agreement, non-independence, and reliability: Implications for data aggregation and analysis. In K. J. Klein & S. W. Kozlowski (Eds.), Multilevel theory, research, and methods in organizations: 349-381. San Francisco: Jossey-Bass. Brown, M. E., & Mitchell, M. S. 2010. Ethical and unethical leadership: Exploring new avenues for future research. Business Ethics Quarterly, 20: 583-616.
October
Cialdini, R., & Trost, M. 1998. Social influence: Social norms, conformity, and compliance. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handhook of social psychology, vol. 2: 151-192. Boston: McGraw-Hill. Den Hartog, D. N., House, R. )., Hanges, P. J., & RuizQuintanilla, S. A. 1999. Culture specific and crossculturally generalizable implicit leadership theories: Are attributes of charismatic/ transformational leadership universally endorsed? Leadership Quarterly, 10: 219-256. Depret, E. F., & Fiske, S. T. 1993. Social cognition and power: Some cognitive consequences of social structures as a source of control deprivation. In G. Weaiy, F. Gleicher & R. Marsh (Eds.), Control motivation and social cognition: 176-202. New York: Springer Verlag. Detert, J. R., & Edmondson, A. C. 2011. Implicit voice theories: Taken-for-granted rules of self-censorship at work. Academy of Management Journal, 54: 461-488. Detert, J. R., & Treviño, L. K. 2010. Speaking up to higher ups: How supervisors and skip-level leaders influence employee voice. Organization Science, 21: 249-270. Dragoni, L. 2005. Understanding the emergence of state goal orientation in organizational work groups: The role of leadership and multilevel climate perceptions. Journal of Applied Psychology, 90: 10841095. Dvir, T., Eden, D., Avolio, B. J., & Shamir, B. 2002. Impact of transformational leadership on follower development and performance: A field experiment. Academy of Management Journal, 45: 735-744. Feldman, D. G. 1984. The development and enforcement of group norms. Academy of Management Review, 9: 47-53.
Brown, M. E., & Treviño, L. K. 2006. Ethical leadership: A review and future directions. Leadership Quarterly, 17: 595-616.
Fornell, C, & Larcker, D. 1981. Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18: 39-50.
Brown, M., Harrison, D., & Treviño, L, 2005. Ethical leadership: A social learning perspective for construct development and testing. Organizational Behavior and Human Decision Processes, 97: 117-134.
Glisson, G., & James, L. R. 2002. The cross-level effects of culture and climate in human service teams. Journal of Organizational Behavior, 23: 767-794.
Bryk, A. S., & Raudenbush, S. W. 1992. Hierarchical linear models. Newbury Park, CA: Sage.
Hannah, S. T., & Avolio, B. J. 2010. Moral potency: Building the capacity for character-based leadership. Consulting Psychology Journal, 62: 291-310.
Chan, D. 1998. Functional relations among constructs in the same content domain at different levels of analysis: A typology of composition models. Journal of Applied Psychology, 83: 234-246.
Hannah, S. T., Avolio, B. J., & May, D. R. 2011. Moral maturation and moral conation: A capacity approach to explaining moral thought and action. Academy of Management Review, 36: 663-685.
Chun, J. U., Yammarino, F. J., Dionne, S. D., Sosik, J. J., & Moon, H. K. 2009. Leadership across hierarchical levels: Multiple levels of management and multiple levels of analysis. Leadership Quarterly, 20: 689707.
Hannah, S. T., Uhl-Bien, M., Avolio, B. J., & Gavarretta, F. 2009. A framework for examining leadership in extreme contexts. Leadership Quarterly, 20: 897-919. Hoge, G. W., Gastro, G. A., Messer, S. G., McGurk, D.,
2012
Schaubroeck et al.
Cotting, D. L, & Koffman, R. L. 2004. Comhat duty in Iraq and Afghanistan, mental health problems, and barriers to care. New England Journal of Medicine, 351: 13-22. Howell, J. P., Dorfman, P. W., & Kerr, S. 1986. Moderator variables in leadership research. Academy of Man-
agement Review, 11: 88-102. International Institute for Strategic Studies. 2010. The military balance. London: Routledge. James, L. R. 1982. Aggregation bias in estimates of perceptual agreement. Journal of Applied Psychology, 67: 219-229. James, L. R., Demaree, R. G., & Wolfe, G. 1984. Estimating within-group inter-rater reliability with and without response bias. Journal of Applied Psychology, 69: 85-98. Katz, D., & Kahn, R. L. 1966. The social psychology of organizations. New York: Wiley. Kish-Gephart, J., Detert, J., Treviño, L. K., & Edmondson, A. 2010. Silenced by fear: The nature, sources, and consequences of fear at work. In A. Brief & B. Staw (Eds.), Research in organizational behavior, vol. 29: 163-193. Bingley, U.K.: Emerald. Kozlowski, S. W. J., & Hattrup, K. 1992. A disagreement about within-group agreement: Disentangling issues of consistency versus consensus. Journal of Applied Psychology, 77: 161-167. Kozlowski, S. W. J., & Klein, K. J. 2000. A multilevel approach to theory and research in orgeinizations: Contextual, temporal, and emergent processes. In K. J. Klein & S. W. J. Kozlowski (Eds.), Multilevel theory, research and methods in organizations: Foundations, extensions, and new directions: 3-90. San Francisco: Jossey-Bass. LaHuis, D. M., & Ferguson, M. W. 2007. The accuracy of significance tests for slope variance components in multilevel random coefficient models. Organizational Research Methods, 12: 418-435. Landis, R. S., Beal, D. J., & Tesluk, P. E. 2000. A comparison of approaches to forming composite measures in structural equation models. Organizational Research Methods, 3: 186-207. Likert, R. 1961. New patterns of management. New York: McGraw-Hill. Lindell, M. K., & Brandt, C. J. 1997. Measuring interrater agreement for ratings of a single target. Applied Psychological Measurement, 21: 271-278. Mental Health Advisory Team (MHAT-IV). 2006. Operation Iraq Freedom 05-07. Office of the Surgeon General, U.S. Army Medical Command. http://www. armymedicine.army.mil. Mayer, D. M., Kuenzi, M., Greenbaum, R., Bardes, M., & Salvador, R. 2009. How low does etbical leadership
1077
flow? Test of a trickle-down model. Organizational Behavior and Human Decision Processes, 108: 1-13. Neubert, M., Carlson, D. S., Kacmar, K. M., Roberts, J., & Chonko, L. 2009. Tbe virtuous influence of etbical leadership behavior: Evidence from tbe field. Journal of Business Ethics, 90: 157-170. Page, S. E. 2007. The difference: How the power of diversity creates better groups, firms, schools, and societies. Princeton, NJ: Princeton University Press. Roethhsberger, F. J., & Dickson, W. J. 1939. Himian relations. In Management and the worker: 551-568. Cambridge, MA: Harvard College. Schein, E. H. 1985. Organizational culture and leadership (1st ed.). San Francisco: Jossey-Bass. Schein, E. H. 2010. Organizational culture and leadership (4tb ed.). Hoboken, NJ: Jossey-Bass. Simon, H. A. 1973. Tbe organization of complex systems. In H. H. Pattee (Ed.), Hierarchy theory: 1-27. New York: George Braziller. Snijders, T. A. B., & Bosker, R. J. 1999. Multilevel analysis: An introduction to basic and advanced multilevel modeling. London: Sage. Treviño, L. K. 1986. Ethical decision making in organizations: A person-situation interactionist model. Academy of Management Review, 11: 601-617. Treviño, L. K. 1990. A cultural perspective on changing and developing organizational ethics. Research In Organizational Change and Development, 4: 195230. Treviño, L. K., Butterfield, K. D., & McCabe, D. M. 1998. The ethical context in organizations: Influences on employee attitudes and bebaviors. Business Ethics Quarterly, 8: 447-476. Treviño, L. K., & Youngblood, S. A. 1990. Bad apples in bad barrels: A causal analysis of etbical decision-making behavior. Journal of Applied Psychology, 75: 378385. Victor, B., & Cullen, J. B. 1988. Tbe organizational bases of ethical work climates. Administrative Science Quarterly, 33: 101-125. Waldman, D. A., & Yammarino, F. J. 1999. CEO charismatic leadership: Levels-of-management and levelsof-analysis effects. Academy of Management Review, 24: 266-285. Walker, L. J., & Hennig, K. H. 2004. Differing conceptions of moral exemplarity: Just, brave, and caring. Journal of Personality and Social Psychology, 86: 629647. Walumbwa, F. O., Mayer, D. M., Wang, P., Wang, H., Workman, K., & Gbristensen, A. L. 2011. Linking ethical leadership to employee performance: The
1078
Academy of Management Journal
October
roles of leader-member exchange, self-efficacy, and organizational identification. Organizational Behavior and Human Decision Processes, 115: 204215.
from the University of Nebraska-Lincoln. His research focuses on leadership development; character development; leader identity, self-efficacy, and courage; and leadership in extreme and complex contexts.
Walumbwa, F. O., & Schaubroeck, J. 2009. Leader personality traits and employee voice behavior; Mediating roles of ethical leadership and work group psychological safety. Journal of Applied Psychology, 94; 1275-1286.
Bruce J. Avolio (
[email protected]) received his Ph.D. from the University of Akron. He is the Marion B. Ingersoll Professor and the director of the Center for Leadership & Strategic Thinking in the Foster School of Business, University of Washington. His current research focuses on examining how to accelerate leadership development across levels and determining the value of leadership.
Weiss, H. M. 1977. Subordinate imitation of supervisory behavior; The role of modeling in organizational socialization. Organizational Behavior and Human Performance, 19; 89-105. Yammarino, F. J. 1994. Indirect leadership: Transformational leadership at a distance. In B. M. Bass & B. J. Avolio (Eds.), Improving organizational effectiveness through transformational leadership: 26-47. Thousand Oaks, CA: Sage. Yammarino, F. J., & Dansereau, F. 2008. Multi-level nature of and multi-level approaches to leadership. Leadership Quarterly, 19: 135-141. Yang, J., Zhang, Z. X., & Tsui, A. S. 2010. Middle manager leadership and frontline employee performance: Bypass, cascading, and moderating effects. Journal of Managerial Studies, 47; 654-678. Yukl, G. 2002. Leadership in organizations (5th ed.). Upper Saddle River, NJ: Prentice Hall. Zohar, D., & Luria, G. 2005. A multilevel model of safety climate; Gross-level relationships between organization and group-level climates. Journal of Applied Psychology, 90; 616-628.
John M. Schaubroeck (
[email protected]) is the John A. Hannah Distinguished Professor of Psychology and Management at Michigan State University. He received his Ph.D. from Purdue University. His research interests relate to primarily to leadership processes, employee well-being, and ethical behavior. Sean T. Hannah (
[email protected]) is a colonel in the U.S. Army and the director of the Center for the Army Profession and Ethic, at West Point. He received a Ph.D. in organizational behavior with a focus on leadership
Steve W. J. Kozlowski (
[email protected]) is a professor of psychology in the Organizational Psychology Program at Michigan State University. He received his Ph.D. from The Permsylvania State University. His meta-theoretical work focuses on multilevel phenomena and the dynamics of emergence. Substantive research centers on individual and team learning, team effectiveness, leadership, and use of simulation to create synthetic experience. Robert G. Lord (
[email protected]) is a distinguished professor of psychology at the University of Akron. He received his Ph.D. from Carnegie-Mellon University. His research interests include leadership, motivation, information, and ethical behavior. Linda K. Treviño (
[email protected]) is Distinguished Professor of Organizational Behavior and Ethics in the Smeal College of Business, The Pennsylvania State University. She received a Ph.D. from Texas A&M University. Her current research focuses on behavioral ethics in organizational context with an emphasis on values, ethical leadership, ethical culture, internal reporting and whistle-blowing. Nikolaos Dimotakis (
[email protected]) is an assistant professor at Georgia State University. He received his Ph.D. from Michigan State University. His research focuses on affective processes, well-being and motivation. Ann C. Peng (
[email protected]) is a doctoral student in management at Michigan State University. She received her M.Phil, in management at Lingnan University of Hong Kong. Her research interests include leadership, employee well-being, and proactive work behavior.
Copyright of Academy of Management Journal is the property of Academy of Management and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
