Symmetrical and asymmetrical outcomes of leader anger expression: A qualitative study of army personnel

Frameworks for examining emotions in organizations

To assist in interpreting a follower’s perception of leaders expressing anger, we draw upon two major frameworks: the DTM perspective already mentioned and Emotions as Social Information (EASI, see e.g. Van Kleef, 2014). While each provides an insight into expressions of anger at work and their consequences, they share some conceptual overlap in terms of display rules and norms. Diefendorff et al. (2011) note that display rules (i.e. organizationally prescribed rules for the expression of emotion) can emerge – when accepted by the group – as shared norms (i.e. display norms for the expression of emotion) within workgroups, thereby helping the group to determine what are appropriate emotional expressions at work. We argue that display norms may also vary from display rules if the work group accepts that these norms are consistent with the context they work within and consistent with the group history that contributes to that context. These emotional display norms are, therefore, also relevant in relation to applying the DTM (Geddes and Callister, 2007) to our data. In particular, while norms infer a standard of emotional expression, the DTM notes that there are expression and impropriety thresholds that these norms work within.

The final approach we draw on is the EASI model as it provides a link between display norms and how individuals (here more specifically followers) interpret those display norms. Van Kleef (2014) argues that emotional expressions influence the targets of those emotions (and others around them) by triggering a series of mental processes in which the individual tries to make sense of the emotions being expressed. Specifically, Van Kleef (2009, 2014) notes that both the observer’s affective reaction (e.g. emotional contagion, reduced liking or reciprocal anger) and the inferences (based upon cognitive appraisal of the situation) concerning the emotion directed toward the individual drives her/his eventual behavior. Key here is the level of motivation the observer develops to get a deeper understanding of the situation at hand (Van Kleef, 2009). This is particularly significant when considering the discrete emotion of anger. Whereas other negative emotions such as sadness and fear generally produce avoidance responses, anger is one of the few negatively valenced emotions to produce an approach response pattern (Marsh et al., 2005). Further to this, Van Kleef (2014) argues that the amount of effort the observer puts into understanding an emotion will be dependent on their motivation and the context in which the initial emotions were expressed. Clearly, a part of the context for employees is the display norms established by the organization. In our study, we explore what role context plays in the expression of anger and how followers responded to those anger expressions and their motivations for responding in specific ways.

Display norms in organizations and the military

As noted earlier, the focus of our research is on the observer’s and target’s perceptions of anger expressions in light of the display norms established by the organization. Seen in this light, norm content is concerned with where, why, how and to whom anger is expressed (Gibson and Callister, 2010). Norm strength, by contrast, determines the degree to which members agree with these norms, and also what the repercussions of breaking norms will be (Gibson and Callister, 2010). In light of these multiple considerations, Diefendorff and Greguras’s (2009) work on norms shows that display norms and rules in the workplace can be incredibly complex and nuanced, making them a challenging construct to study.

The cultivation of anger and aggression (as influenced by display norms and rules) has been discussed in wider organizational contexts. For instance, Rafaeli and Sutton found that bill collectors were expected to display negative emotions to fuel a certain amount of aggression toward individuals in an attempt to collect debts (Rafaeli and Sutton, 1987; Sutton, 1991). Still, Diefendorff and Greguras (2009) argue that many organizational display norms encourage no expressions of anger, or at least expressions of lesser intensity than it is really felt and that control of anger is important. Gibson and Callister (2010) note this paradox and argue that examining follower perceptions of display rules and norms at work with regard to anger, and how this impacts their reactions to certain emotional expressions, remains to be understood. We note that the DTM, which places emphasis upon the norms that shape the impropriety threshold, provides a framework to increase our understanding of this phenomenon.

Moving on to considering anger display norms in the military, there is a public perception of military personnel experiencing significant anger and aggression both in training and in their working day. Linkh and Sonnek (2003) assert that the army’s attitude to anger is somewhat paradoxical. On the one hand, it seeks to cultivate anger and aggression in soldiers in training situations to prepare them for combat (e.g. staff sergeant yelling at soldiers). On the other hand, the military also places a high value on self-control and conformity (both reflect that there are expectations around emotional display norms). We note, however, that anger and aggression are not the same (see Geddes and Stickney, 2011), albeit they can be used in the military context sometimes simultaneously to achieve desired outcomes. As we have discussed, anger is a reaction to perceptions of injustice, whereas aggression is an instrumental response where an individual seeks to dominate others (Pinto, 2014). On this basis, individuals can be angry and aggressive, angry and not aggressive, or not angry but aggressive. This distinction is not always clear.

Discussions of anger within the military predominantly consider it a negatively valenced emotion, and associate it with violence (Linkh and Sonnek, 2003; Taft et al., 2007) and Post Traumatic Stress Disorder (or PTSD – Morland et al., 2012; Teten et al., 2010; Worthen, 2011; Wright and Russell, 2013). While this should not lead to the conclusion that anger in the military is always caused by PTSD, coupled with negative outcomes, it is a significant example of the inherent assumption and treatment of anger as an undesirable emotion. This notion is similarly reflected in the leadership literature (Madera and Smith, 2009; Waldman et al., 2011). Taken together, there are often competing perceptions on the usefulness and utility of anger expressions in the context of the military. It is exploring these competing perceptions that renders our study theoretically (in the context of the DTM) and practically relevant.

Leader expressions of anger and their perceived consequences

The examination of anger in leadership studies has generally supported the symmetrical assumption of a link between anger and negative outcomes. For example, Lewis (2000) notes that leader anger expressions tend to reduce observer perceptions of leader effectiveness. Similarly, Madera and Smith (2009) found that particularly in times of crisis, leaders displaying anger were evaluated less favorably than those expressing sadness in response to a failed product. Goleman (1998) also suggests that anger expressed by leaders is related to leader ineffectiveness, as it signifies a lack of emotional control. Perhaps as a result of this common perception, it has been reported that employees (including leaders) are inclined to hide or suppress feelings of anger owing to a concern for maintaining professionalism, as well as a fear of sanctions if the anger expression is perceived as violating display norms (Geddes and Stickney, 2011; Kramer and Hess, 2002; Stickney and Geddes, 2004).

Overall, these findings represent a symmetrical assumption between leader anger as a negatively valenced emotion, and negative consequences as perceived by others, even though ‘organizational life does not represent such a neat juxtaposition’ (Lindebaum and Jordan, 2012: 1027). While comparatively little research has examined conditions under which anger expressions can have positive consequences, there are some notable studies where leader anger expressions have led to positive consequences for the individual and/or the organization. For example, Tiedens (2001) found that people confer more status to those who express anger than those who expressed other ‘negative’ emotions such as sadness. Similarly, diverging from Lewis’s (2000) findings, Lindebaum and Fielden (2011) found that anger expressions in the construction industry were linked to perceived leader effectiveness. Moreover, Van Kleef and colleagues (2009) demonstrated that after a leader displayed anger, followers developed a deeper desire to understand the situation that precipitated the leader’s anger in order to improve the follower’s performance. In a similar vein, studies in the realm of negotiation show that expressions of anger typically aided in extracting concessions from negotiation counterparts, the exception being when anger expressions were perceived as inappropriate by the counterpart owing to a violation of an explicit display rule (Van Kleef and Côté, 2007). Similarly, when the counterpart had a powerful position, expressions of anger backfired by evoking competitive and retaliatory responses. Further to this, a recent empirical study shows that expressed anger predicts perceived improvement with problematic situations at work, while suppressed anger induced perceptions that the situation at work had deteriorated (Stickney and Geddes, 2014). Clearly, the literature presents contradictions regarding anger’s ability to elicit positive and negative outcomes, with findings being varied and, therefore, heavily dependent upon the context and the unique situation.

To make better sense of these contradictions, we developed this study examining anger expressions in the military. As mentioned earlier, we draw upon a combination of the DTM (Geddes and Callister, 2007) and Van Kleef’s EASI model (2014) to help interpret our data. Specifically, we employ these theoretical frameworks to assess the perceived appropriateness of leader anger displays as a key contingency to the observer’s processes of affective reactions or inferences. As Van Kleef highlights, affective reactions or inferences can lead to different outcomes in the case of anger expressions, and it is this difference that helps better understand the contradictions identified above in terms of anger expressions and their outcomes at work.

Method

Consistent with this study’s aim, we employ a phenomenological approach. This enables us to embrace ‘the complexity of the human individual and their actions’ (Ardley, 2011: 637), as well as prioritizing the individual’s own interpretation of a situation in relation to the research question of this study. More specifically, a phenomenological research approach focuses on the subjects’ own interpretation of a situation or experience, their lived experience (Sanders, 1982) and their construction of reality through narration (Groenewald, 2004). It is for this reason that, in describing this study, we follow the traditions of the method and exclude concerns for validity, reliability, objectivity, generalizability and verification (Amis and Silk, 2008). Instead, and consistent with writings of other qualitative researchers (Lincoln and Guba, 1985; Lindebaum, 2015), we focus upon several ‘quality’ criteria of qualitative research. First, we seek credibility instead of internal validity. This implies an authentic representation of the lived experiences of followers. We achieved this in our study by allowing individuals space to express themselves freely with regard to the phenomena under investigation, in addition to seeking respondent feedback from soldiers. Second, we focus on transferability instead of external validity. This involves drawing out the applicability of findings, for instance, in other contexts. As suggested later, there is a nexus between the findings of this study and other studies interested in how the nature of tasks can moderate the link between leaders expressing emotions and follower outcomes (Visser et al., 2013). Third, we consider dependability instead of reliability. Essentially, this factor asks researchers to minimize researcher idiosyncrasies. We have addressed this by way of ‘bracketing’, which involves noting down as many conscious presuppositions as the researcher can establish prior to analyzing the data in order to ensure these presuppositions do not cloud the interpretation of the data (for a detailed treatment on ‘bracketing’, see Holt and Sandberg, 2011; Hycner, 1985; Sandberg, 2000). Fourth, and finally, good research in this paradigm relies on confirmability instead of objectivity. Involved here is a sense of researcher self-criticism, such as changing the scope and labels of the themes presented later in response to discussions among two of the researchers of this study. In addition, this has been addressed by being rigorous in the analysis of data through a systematic reading and re-reading of the interview scripts – as opposed to the creation of researcher distance and non-involvement (Etherington, 2004).

As noted earlier, display norms in the workplace can be incredibly complex and nuanced, making them a challenging construct to study. Therefore, Gibson and Callister (2010) encourage researchers examining display norms in organizations to employ a range of qualitative methods that allow for ‘rich’ descriptions of perceptions of display norms. The phenomenological interview uses minimal structure, allowing both the individual to present their meaning structures as spontaneously as possible, and scholars to investigate the human experience from the inside of the individual using introspection (Osborne, 1994). This was crucial for our study as our research question involved examining the internal processes through which followers perceive leader anger. Readers wishing to immerse themselves further into phenomenology as a vehicle of analysis can consult several existing reviews (Holt and Sandberg, 2011; Sanders, 1982).

Findings

The findings represent the key themes determined from the clusters of relevant meaning following specific guidelines for phenomenological interview analysis, as detailed above. We emphasize that, although this inquiry was guided by an overarching research question, it is the nature of qualitative research that interviewees provide the structure of the data, rather than researchers imposing pre-defined structure as commonly occurs in quantitative research (Dey, 1993). Therefore, the data are broader than what we sought to elicit with the research question.

Themes extracted are labelled (i) paradoxical attitudes to anger expression and changing display norms, (ii) symmetrical outcomes of leader anger expression and (iii) asymmetrical outcomes of leader anger expression. Table 1 provides a summary of our findings.

OPEN IN VIEWER

Table 1.

Summary of findings.

Theme	Sense-making	Implications for understanding link between display norms, affective reactions/inference and outcomes	Examples
Paradoxical attitudes to anger expression	Military establishment discourages expressions of anger in recent years; does not want to be seen as autocratic, but rather encourage transformational leadership styles or ‘lovey-dovey’ approaches.Conversely, in practice soldiers still express anger at work as it can be useful in getting points of view across and bringing attention to issues that need to be addressed.	Display norms in the military are changing and thus somewhat paradoxical; expressions of anger still commonly seen at work, however much the military at large tries to combat it.Expressions of anger could be construed as bullying and involve formal punishments, in line with the organization’s established protocol for dealing with bullying.	‘I think there is definitely a perception that the army is trying to combat . . . the army is investing a huge amount of time with its instructors at all levels to move away from that sort of transactional leadership . . . it’s moving to a far more transformational style . . .’ (Follower 8, Lieutenant)
			‘If someone was in their office shouting and screaming at somebody . . . no one would really take any notice of it . . . No one really bothers what it’s about, because they know it’s probably warranted’. (Follower 18, WO Class 2)
Symmetrical outcomes to leader anger	Followers had negative emotional reactions to leaders when it was considered unwarranted.Followers observing leader anger toward a subordinate often think the leader has poor leadership qualities.	Both affective reactions and inferences can result in negative outcomes to leader anger. This occurs when leaders transgress display norms. Particularly where observers perceive leader anger as an injustice, they make negative inferences about them, resulting in avoidance behaviors from the target.	‘People will walk away just calling you names behind your back, thinking you’re an absolute nob . . . People won’t approach you’. (Follower 12, Captain)‘He treated [an individual] with utter contempt . . . it was a real angry outburst . . . I thought it was horrendous and utterly inappropriate’. (Follower 13, Captain)
Asymmetrical outcomes to anger	Followers respond to leader anger by assessing their performance/behavior and made efforts to rectify the issue and address the anger-eliciting event.When in intense training practices, or out in combat situations, followers respond to leader anger immediately, by carrying out the task required, or mirroring the leader emotion toward the threat.	Where leader anger is perceived as justified, soldiers accept it and make inferences about their performance and/or behavior. Subsequently, they improve their performance behavior.In high-pressure situations, followers respond to leader anger with immediate affective reactions. This involves mirroring the leader’s anger in order to produce the kind of controlled aggression to deal with arduous and stressful situations. Also, their immediate behavioral reactions seem to take place without any time for high levels of information processing or inferences.	‘He got really angry on the radio. He said “get that fucking cigarette out of your mouth now”. [Laughs] . . . I did it because I realized I had forgotten where I was and forgotten what I was doing. He was right . . . I sorted it immediately and nothing more was said’. (Follower 19, Sergeant)‘If you’re on exercise and you’re doing a platoon attack . . . That can result in a lot of . . . shouting; you don’t want to be quietly tapping people on the shoulder and saying, come on, you could possibly do this a bit better. You need to get the point across immediately’. (Follower 7, Captain)

Discussion

This research project was guided by the question: what conditions contribute to the expression of anger that crosses the impropriety threshold being seen as productive or counterproductive by observers or targets of that anger? The data that emerged from our study led us to offer several important theoretical contributions. The first thing to note is that, in examining dynamic processes at work, simple models do not always capture the complexities that emerge at work. This seems particularly relevant in the context of display norms, as mentioned before. While we have outlined both evidence of negative outcomes of anger expressions (e.g. Lewis, 2000) and previous research regarding the positive outcomes of anger (e.g. Stickney and Geddes, 2014), our data revealed that both positive and negative outcomes emerged simultaneously. One factor influencing this is the paradoxical attitude toward expressions of anger, which reveal that, on the one hand, certain display rules are enforced to combat perceptions of a bullying culture (see account of Follower 12, Captain). On the other hand, however, there are certain characteristics of the situation and task at hand that seem to require the presence of anger to resolve these (see account of Follower 7, Captain). This tension, we suggest, exerts a constant pull and push on the location of the impropriety threshold. It is not static or fixed, but subject to continuous negotiation among those who express anger and the observers or targets of that expression. As our analysis suggests, one relevant factor or condition in that negotiation is the nature of task, especially when urgency had to be instilled into the situation. We note that, in the context of the DTM, the nature of the task as such is not explicitly mentioned as a factor prompting the crossing of the impropriety threshold.

A further issue arising from the data was the negative connotations followers associated with anger when it was either a consistent response or the only response from leaders. Several of the followers identified this as an issue inferring that a limited range of emotional expressions (i.e. only expressing anger) or expressing anger at a level that was inappropriate to the situation at hand resulted in the follower losing respect or liking for that leader (i.e. an effect beyond the impropriety threshold). This is partly consistent with prior research showing that the effects of anger diminish if it is constantly displayed (Lindebaum and Fielden, 2011). Certainly, there were examples in the data that supported this conclusion. Based on our data, we note that the impropriety threshold is not just about intensity of expression, but also about the consistency of the anger response. It appears that the impropriety threshold becomes reduced for observers when they perceive constant expressions of anger from leaders.

That said, however, we also found that there were times when expressing anger beyond the impropriety threshold was accepted as reasonable. For instance, Follower 19’s (Sergeant) recounting of one of his commanding officers going ‘apeshit’ and swearing at him. This would be considered in most workplaces (and in the military) as exceeding the impropriety threshold. In this case, however, the Follower found this response to be justified and accepted the anger expression without developing negative connotations, despite the breach of the impropriety threshold.

Our data also support research showing that expressions of anger result in observers or targets seeking information to resolve situations. In line with the EASI model (Van Kleef, 2014), we found that anger expressions resulted in followers reflecting on their own behavior to assist them in making sense of an anger expression (see also Van Kleef et al., 2009: for an example of enhanced follower performance as a result of leader anger expressions). We further note that affective reactions also produced asymmetrical outcomes by virtue of the nature of task at hand (i.e. instilling urgency in the situation), as already discussed above in the context of the DTM. In these instances, affective reactions took the form of emotional contagion, which transmitted leader anger and aggression to the followers.

Practical implications

There are several broader practical implications we wish to draw attention to as well. First, the army seems to be explicit in terms of establishing expectations of behavior and associated display rules to address past and current accusations of leader bullying (Townsend, 2014). Our data, however, reveal a very real difference between display rules and display norms. Although there are explicit standards of acceptable behavior, our data reveal many implicit ‘norms’ that, together, dictate when a leader’s expression of anger is appropriate or not as perceived by observers and/or targets. Often, the accepted norm (freedom to express anger that exceeds the impropriety threshold at time) is not in line with organizational display rules. Some of this explains the apparent paradoxical attitude toward anger held by the military, as this and prior studies suggest (Linkh and Sonnek, 2003). We suggest that this is related to the special context within which all military organizations operate. However, these paradoxical tendencies can inhibit the very learning process at the heart of display norms. As Johnson et al. (2000) argue, display norms in organizations tend to be understood through a gradual learning process. Applied to our study, that gradual learning process is likely to be hampered vis-a-vis the paradoxical explicit and implicit information that both leaders and followers face. As our data have shown, anger can be quintessentially important vis-a-vis the task context of military (i.e. high-pressure combat situations), albeit some soldiers believed that the army has moved toward a much softer approach to training soldiers to avoid accusations of bullying. This would seem to suggest that the army places higher emphasis on emotional self-regulation by soldiers, and display rules are the conduit through which this expectation is communicated.

While there is still an inherent assumption and treatment of anger as an undesirable emotion in the military, the nature of some tasks to be performed in the context of the military nevertheless seems to necessitate it or in some cases produce it. The implications of this paradoxical situation, therefore, can be profound for military personnel, especially in terms of how they are socialized into, and trained for, the job. Our data indicate that there is a difference between display rules established by the organization in this context, the display norms required to achieve desired performance outcomes, the standards of behavior expected, and the preparation of soldiers for engaging in operational circumstances where extreme emotions are expected to be experienced. Soldiers need to be assisted in negotiating this dynamic process and a clear distinction provided for them between occasional anger utilized to enhance their training and performance and persistent angry responses, which may be interpreted as bullying.

Second, while remaining cautious not to generalize excessively from our data, results of this study may have relevance for employees in other organizational contexts, such as those who work in crisis situations (e.g. in the emergency services or police) or in organizations where significant and urgent change or action is required. Our study showcases anger as a particularly useful leadership tool in high-pressure situations where the safety or well-being of individuals or groups is paramount, owing to its potential to inject urgency and elicit immediate affective reactions and subsequent desirable behavioral responses in followers to meet the requirements of the situation.

Lastly, as observers of leader anger expression may react differently to those who are the target of the anger, leaders should be aware of the need to manage reactions to their anger in both observers and targets. Therefore, we suggest that leaders should be able to explain the reasons for their anger (at least post-hoc the event) not only to the target in order for them to make behavioral changes, but also to observers, whose reactions can be equally important to outcomes, even though they may not be directly involved. Being able to provide such explanations under conditions of ambiguity (e.g. the paradoxical situation in the military that anger expressions are both discouraged and encouraged) can help prevent attributional errors on the part of targets and observers (Geddes and Callister, 2007), and positively influence future leader–follower interactions (Dasborough and Ashkanasy, 2002). Indeed, this should be the other side of the EASI equation; that is, an angry person being able to explain their anger as well as the target seeking information about the anger.

Limitations and future research

The method used in this study has inherent limitations, but the richness of the data analysis helps counterbalanced this. Qualitative data normally do not permit the testing of theories (Creswell, 1994). While it is crucial to recognize this limitation, it is worth-highlighting that the strength of qualitative research lies in the potential to refine our understanding (as is the case in this study) of existing theories (Pratt, 2009). Another potential limitation of the in-depth interview is that of interviewer bias (Antonakis et al., 2004). In order to limit researcher bias as an influencing factor in the interpretation of findings, we have followed clear guidelines for phenomenological research (Hycner, 1985) and have explained how this study complies with the ‘quality’ criteria for qualitative research as outlined by Lincoln and Guba (1985). Further to this, in line with all qualitative research (Kincheloe and Tobin, 2009), the phenomenological interview accepts the inability to generalize, and makes no such claims of generalizability or context-free results in this study (Ardley, 2011). With regards to the sample used, the study only included followers from one infantry regiment based in England and we acknowledge that no female respondents were interviewed. As such, findings may not be representative of the military at large, albeit one needs to bear in mind that – in phenomenological studies – larger samples do not necessarily yield more data (Sanders, 1982). Rather, we were concerned with providing ‘an account which allows for [a] depth of understanding, rather than one which unnecessarily generalizes the outcomes’ (Quinn, 2009: 258).

While this study has provided some theoretically and practically useful findings, there remain important areas for future research. First, we see the potential for future research to examine this phenomenon in other organizations and to expand the sample to include females. Future research would also benefit from considering the short- versus long-term outcomes of leader anger expressions. Our data revealed that constant displays of anger were related to an apparent lowering of the impropriety threshold. That is, in the long term, observers and targets dislike the angry individual, and will tolerate their anger less and will try to avoid future interactions (consistent with muted anger – see Geddes and Stickney, 2011). We can see that the exposure to constant anger might result in negative health consequences as a result of being exposed to leader anger over time (see Lawrence et al., 2011). This is one potential source for future research. There is also the opportunity to investigate the performance outcomes of experiencing constant anger. We can see that exposure to this kind of anger and the avoidance it generates could result in lower performance for observers and targets.

Although this has not been central to our original research question, we also note that display rules and norms in the military are perceived to be changing. As such, soldiers are left with conflicting ideas – owing to tensions between display rules and norms – of what is appropriate and what is not. While we know that these perceptions change over time (Stearns and Stearns, 1985), more research is required to understand how and why display norms change over time and how these norms are transferred within the workforce. For example, which factors (e.g. social, economic, environmental) induce display norms changes in organizations? How does this impact on how employees view display norms in their organization, both individually and collectively? How can we systematically study these changes over time, using which methods?

Finally, as already hinted at, exploring potential moderating effects of task-type in leadership studies is an important area for future research. Since the nature of tasks can vary significantly across industry and occupational contexts (compare, for example, the care of a nurse with the risk-taking of stock traders), it is plausible to suggest that this area offers significant potential for future research. As Johns (2006: 389) cautions, ‘context is likely responsible for one of the most vexing problems in the field: study-to-study’, adding that it often plays a larger role than individuals differences across research settings. More specifically, close appreciation of context (as it dictates the nature of tasks) can also aid in better understanding the missing linkages that can explain how activities of organizational members translate into organization-level outcomes (Goodman, 2000). Taken together, future research has the potential to progress both theoretically and empirically how leaders and followers interact emotionally to produce a variety of individual and organizational outcomes given a particular situation at hand within a certain moment in time.

Conclusion

In sum, our study reveals interesting paradoxical attitudes toward expressions of anger in the context of leader–follower interactions in the military. While the military seeks to foster certain display rules to combat perceptions of a bullying culture, there are certain characteristics of the situation and task at hand that seem to require the presence of anger to resolve these. This helps explain why sometimes anger expression can yield symmetrical outcomes, whereas in other instances (either below or above the impropriety threshold) it leads to asymmetrical outcomes. Our study contributes to this ongoing debate by demonstrating that the impropriety threshold is not static or fixed, but subject to continuous negotiations among those who express anger and the observers or targets of that expression.