The Emergence of Collective Leadership During a Terrorist Attack: Dynamic Role Boundary Transgressions as Central in Aligning Efforts

Crisis Leadership in Extreme Events and Contexts

Crisis leadership has been researched for decades (Pearson & Mitroff, 1993), yet recent years have seen more focus on how differences in event and context characteristics influence the role of leaders (Hällgren et al., 2018; Hannah et al., 2009). Crisis events and contexts vary in extremeness, such as in terms of level of threat, location, magnitude and probability of consequences, and psycho-social proximity. Hällgren et al. (2018), for instance, distinguished between events related to core activities and/or unrelated contexts. In risky contexts, organizations manage operational risks that may trigger events such as a plane crash, oil rig explosion, or product tampering (Mitroff et al., 1987; Weick et al., 1999). In emergency contexts, critical events, such as in wildland firefighting units (Useem et al., 2005), emergency hospital wards (Klein et al., 2006), and speacial weapons and tactics team (SWAT) teams (Bechky & Okhuysen, 2011), in which risk responses become more planned for and routinized, are managed daily. In disruptive contexts, however, events range from genocide to natural disasters and malicious acts such as terrorism, which are unrelated to the organizational context and for which organizations are largely unprepared (Birkeland et al., 2017). Thus, crisis leadership demands and needs may be seen as varying on a continuum from more to less extreme.

In this study, we narrow our scope to study crisis leadership on the more extreme end of the continuum of being disruptive. While there may be plans and structures in place for different risk and crisis potentials (Selart et al., 2013), effective response is likely to require the mobilization of ad hoc structures, demanding structural flexibility and improvization. This diverges substantially from the static bureaucratic structures often found in existing plans and structures in risky and emergency contexts, such as in emergency wards (Klein et al., 2006) or firefighting units (Bigley & Roberts, 2001). Further, such extreme situations tend to evoke distress, shock, and horror, putting pressure on leaders not to break down psychologically (DeRond & Lok, 2016; Sommer et al., 2016). Leadership, in these situations, is to some extent already failing, and leaders risk exacerbating the crisis through their actions, inactions, and failed actions (Weick, 1993). Unsurprisingly, researchers have contended that the immediate response phase is the most challenging phase for leaders to manage (Sommer et al., 2016).

Leadership during this phase is likely to involve the overall purpose of containing the crisis, preventing loss and further escalation, normalizing the situation, and mitigating circumstantial effects (Hadley et al., 2011; Pearson & Clair, 1998; Williams et al., 2017). In this phase, prescriptive models tend to portray crisis leadership efforts as a series of consecutive steps such as making sense of the event, assessing potential consequences, making decisions, setting priorities, and implementing solutions (Hadley et al., 2011; Pearson & Clair, 1998). However, the in-situ leadership process is usually emergent, collective, and dynamic and continually subject to shifting situational needs and demands (Hannah et al., 2009; Pearson & Clair, 1998; Wu et al., 2021). Past empirical studies have suggested that sensemaking and giving (Baran & Scott, 2010; Weick et al., 1999) are key to developing a response strategy. Further, coordination is necessary to support such a strategy (Bigley & Roberts, 2001; DeChurch et al., 2011), and past research has suggested that fostering coping (DeRond & Lok, 2016; Kahn et al., 2013) and developing personnel (Bergman et al., 2019; Klein et al., 2006) may be important in this phase. The current study aims in particular to address leadership across structures during disruptive events, as this is an under-researched level of analysis (Carter et al., 2020; Hannah et al., 2009). Hence, in line with a call for a more grounded approach to leadership in these situations (Hadley et al., 2011; Wu et al., 2021), we move beyond general and individualistic leadership style theories as well as prescriptive models to examine the role of leadership in extreme events. A functional approach seems especially appropriate for examining leadership across structures in these extreme situations.

Leveraging a Functional Leadership Perspective

According to the functional leadership perspective, leadership involves complex social problem-solving through collective efforts (Carter et al., 2020; Fleishman et al., 1991; Mumford et al., 2000). A leader’s main objective is “to do, or get done, whatever is not being adequately handled for group needs” (McGrath, 1962, p. 5). Rather than specifying the behaviors that constitute leadership, such as transformational or servant leadership behaviors, the functional approach views leadership as a role. Effective leadership requires translating critical demands and needs into a pattern of functional behaviors, yet what constitutes functional behaviors may vary across contextual and situational factors as well as over time (Marks et al., 2001). The functional view focuses on the critical needs of the collective, as implied by the systems view of organizations (Fleishman et al., 1991; Katz & Kahn, 1978). Thus, the functional perspective seems suitable for analyzing leadership in temporary mobilized response structures, orchestrating efforts at the meso level of analysis in organizations that is higher than the team but lower than the organization (DeChurch et al., 2011). Further, this view acknowledges the emergent, collective, dynamic, and multilevel aspects of leadership (Burke et al., 2018).

Three aspects underlying this functional view seem particularly central for further developing our understanding of in extremis leadership across structures. First, it is deliberately inclusive when considering who leads, meaning that leadership can be preplanned and formal and/or informal or emergent. Anyone who fulfills the critical collective needs of the situation is considered a source of leadership. Furthermore, in any given collective, multiple sources may typically perform leadership tasks simultaneously, and leadership may take different forms and be collective. Morgeson and colleagues (2010) identified four sources of leadership based on the locus (i.e., internal or external) and formality (i.e., formal or informal) of leadership. The examination of the interplay between different formal and emergent leadership sources and forms in extreme events is ripe for further exploration.

Second, the functional view emphasizes the importance of identifying key leadership functions that meet the needs of the collective. While quite a few studies have examined this in teams (e.g. Burke et al., 2018; Morgeson et al., 2010), research is scarce when it comes to examining leadership across structures, such as in multiteam systems (DeChurch et al., 2011; DeChurch & Marks, 2006). Because these, by definition, are made up of “units of units,” leadership functions are likely to be homologous across levels of analysis and include performance and interpersonal dimensions (Klein & Kozlowski, 2000). However, leadership functions are not universal to all collectives (Marks et al., 2001). Thus, different settings are likely to render varying leadership functions that are more or less critical for achieving overall purpose(s) and proximal goals across structures. Accordingly, the most critical functions during in extremis leadership situations are yet to be revealed (Carter et al., 2020).

Third, what constitutes effective leadership may vary over time and across levels (Klein & Kozlowski, 2000). For example, some researchers have suggested that leadership is likely to be cyclical by moving from transition/planning to action phases (Marks et al., 2001) and that leadership capacity often develops over time (Day et al., 2004). Although the process of leadership may start out as an input, it can become an outcome of prior input in later cycles. Furthermore, across levels, a key issue is how leaders align efforts over time. While traditional leadership research describes top management as representing long-term or strategic interests and lower levels of leadership as representing operational or short-term interests, combined actions across levels are likely to contribute to the leadership process. Thus, understanding leadership across structures hinges on grasping how strategic and operational leadership are aligned over time (DeChurch et al., 2011).

Using these building blocks from the functional leadership view, we address the following three questions using data collected at the corporate headquarters of a multinational energy corporation during a terrorist attack and siege: Who emerges as leaders? What are the critical leadership functions in the temporary mobilized response structure? And how does leadership develop over time and across levels to align efforts in the crisis response?

The Research Context

In 2013, the year of the terrorist attack, the affected corporation had six business areas (BAs) and approximately 30,000 employees. It was ranked one of the 40 most profitable companies in the world. The self-acclaimed high-reliability organization (Weick et al., 1999) depended on managing the risks associated with its operations and other activities in an error-free manner. This meant that there were multiple procedures in place to stay heedful to and manage risks (Weick & Roberts, 1993). By acknowledging that crises might still occur, the company had set up preparedness plans for several crisis scenarios, including for industrial accidents, malicious and criminal acts, and natural disasters. Furthermore, it provided regular crisis management training to a pool of personnel including leaders, experts, and support staff in preparation for such events.

The preparedness plans prescribed the mobilization of a corporate-level crisis management team (CMT) at the headquarters in case a triggering event exceeded the line organization’s ordinary capacity or threatened overall corporate interests. These plans outlined two lines of notification: one for the corporate organization (external) and one for the CMT (internal) structure. The plans typically described nine duty positions in the CMT (crisis leader, chief-of-staff, communications, human resources, medical doctor, insurance, international officer, rescue liaison and technical staff), although the scope of the crisis response structure was scalable and flexible and could be altered at any time during a crisis response. If more than one team were mobilized, this structure would be referred to as the crisis management organization (CMO).

The Crisis Response

Crisis response efforts began immediately. On-duty personnel and others who had been quickly mobilized began working toward the overall purpose of crisis response, namely to contain the crisis by saving lives, securing assets, and protecting the company’s reputation. Initial efforts to save lives focused on rescuing the hostages while ensuring that the situation did not escalate in terms of risking more people’s lives. This led to the evacuation of more than 2000 people across the affected region in case of surprise and concerted attacks on other facilities. The response also involved securing the production plant, where dangers such as hydrocarbon activity could lead to an exposition if exposed to hostile action. Efforts were also directed at communicating with hostages and hostage-takers and establishing dialog with partner companies and cooperating firms, national authorities, and military actors. Furthermore, it involved ensuring that the corporation’s international strategy concerning operations in politically unstable regions of the world would not become scrutinized until the immediate situation was contained, so as not to jeopardize strategic investments through a reputational blame-game.

What began as a single CMT evolved into a CMO with eight specialized crisis response units. By the fourth day, at the midpoint of the situation, the temporary mobilized response structure had increased to an astonishing number of around 125 people, and the (informal–emergent) leader–member ratio was close to 1:5. The CMO was manned on a 24–7 basis for eight consecutive days before being demobilized. Examples of the activities performed in the different units include the following: (a) the response unit coordinated activities and logistics, including evacuating escaped and rescued hostages; (b) the hostage incident response team (HIRT) communicated with hostages and (up to a certain point) the hostage takers; (c) the communications unit issued press releases and held press conferences; (d) the human resource unit provided psychosocial support to victims and next of kin; (e) the corporate communications unit communicated with the board of directors, joint-venture partners, national authorities, and military agencies; (f) the security unit assessed travel safety and plant damage and secured the plant; (g) the business continuity unit assessed the financial risks and opportunity for restarting plant operations; and (g) the recovery unit facilitated an investigation and transferred the remaining tasks and responsibilities to the corporate line organization. Figure 1 depicts a timeline of crisis and the emergence of different units in the CMO. Leadership across this temporary organizational structure at the corporate headquarters during those eight days provides the backdrop for the current analysis.

OPEN IN VIEWER

Figure 1.

Timeline of crisis developments as well as responses to situational needs and demands.

Data Sources and Data Collection

The primary data include 46 interviews and 80 h of observations (over 300 memos) collected over the course of eight days. Immediately upon arrival at the crisis management facilities, the chief of staff and unit leaders informed all other personnel about the author’s presence and status as a researcher. While building rapport with the personnel, the author provided the personnel with oral information about the study and informed that (a) interview participation was voluntary, (b) they had the right to withdraw from the study and have their interview records destroyed at any time, (c) interview information would be treated confidentially, and (d) the researcher could be contacted at any time to answer questions.

The participants agreed for interviews to be conducted during brief, high-intensity periods, with the interviews disrupted if necessary. Hence, we used brief snapshot interviews, as recommended by Wildman et al. (2012), who suggested that interviews in mission-critical situations should be conducted in a way that would fit the situational demands. Lengthier interviews were conducted during low-intensity periods, following McCracken’s (1988) recommendations. Due to the constraints, snapshot interviews lasting 5–25 min each were conducted during the daytime over the first four days. In the evenings and from day four onward, the typical interview length was 1–1.5 h. Of the 46 interviews conducted, 29 were brief (i.e., 5–25 min) and 17 were of longer duration (i.e., 1–1.5 h).

Because extreme situations may cause people to feel vulnerable, the researcher was careful not to interview personnel who appeared distressed (DeRond & Lok, 2016). Further, being present during a crisis may introduce an involvement paradox (Langley & Klag, 2019). This was resolved by the researcher, who is also a trained crisis psychologist, by providing brief psychosocial support when appropriate, while four other crisis psychologists working for the firm and not involved in the research were engaged to support such needs among the personnel. Since the CMO was located several thousand miles away from the terrorist attack, the interviewed personnel were in no immediate physical danger, as may be the case in other extreme situations (Hannah et al., 2009).

The researcher made personal contact with each interviewee at least twice during data collection, which was crucial for achieving data saturation, which is an important aspect of a grounded theory approach (Glaser & Strauss, 1967). Such member checks included presenting interviewees with current observations and interpretations and allowing them to correct information or provide further detail to gain a fuller picture of their experiences. This was also an important part of ensuring credibility and preventing confirmation bias (Guba et al., 1994), and it allowed interviewees to withdraw information if they desired.

In addition to interviews, observations were a primary source of data. In the CMO, the researcher was given security clearance and an access card and could move freely to solicit interviews and observe leaders and their units in action, including during informal and formal meetings.. The CMO consisted of eight rooms, including one in an adjacent building. Observations were carried out for 8–12 h per day, resulting in approximately 80 h of observations and 100 audio recorded observational memos.

Complementary data were collected before, during, and after the crisis as part of a larger research project to examine crisis preparedness and training at the corporation. These data include five pre- and five post-crisis interviews with regard to preparing for and learning from crises, 32 occasions of observations (including 24 crisis management training sessions, 2 workshops, and 4 meetings), and archival data including preparedness plans (6), printouts of electronic logs (2), official and internal crisis management investigation reports (3), and pictures (11).

The triangulation of data from multiple sources, particularly complementary data, was important from a verification standpoint to strengthen interpretive validity (Guba et al., 1994; Langley & Stensaker, 2012). The interview data were transcribed and then organized, coded, and analyzed using Atlas.ti7 software. All data were collected by the author, and their storage and use were approved by the Norwegian Social Science Data Services (NSD).

Data Analysis

The data were analyzed inductively in an iterative fashion, shifting between the data, interpretations, and relevant literatures. Thus, although depicted below in a stepwise manner, the findings were, as is common in inductive research, revised and refined iteratively (Klag & Langley, 2013). The data analysis examined three thematic questions separately: (a) who leads (sources and forms), (b) what leaders do (functions/behaviors), and (c) how leadership develops across levels and over time (e.g., role boundary transgression).

For each topic, we approached analysis in three coding steps, namely first- and second-order coding and overarching conceptual categorizing, as is common in grounded theory (Glaser & Strauss, 1967). During first-order coding, the authors used labels as close to the words used in vivo by the interviewees as possible, based on a minimum level of meaning. Second, the codes were refined, subsuming the first-level codes under second-order codes. Third, the second-order codes were organized into three overarching conceptual categories (Strauss & Corbin, 1990). A functional leadership view (Morgeson et al., 2010) was found useful for consultation throughout this coding process to identify leadership sources and behaviors across the CMO.

The researcher continued analyzing the data in a top-down manner, looking for emergent and dynamic patterns using visual mapping (see Figure 1) and temporal bracketing strategies (Langley, 1999). During this process, the primary data were compared with the complimentary data to include situational and contextual aspects. Matching concepts with facts helped for instance to visualize the emergence of collective forms of leadership across four quadrants (Morgeson et al., 2010). Furthermore, visual mapping helped identify leadership functions and subtasks that were repetitive and cyclical. By depicting how leaders emerged and interacted over time, leadership capacity and alignment could be identified as phenomena that changed dynamically.

To maintain the “checks and balances” of a realist interpretivist stance, keeping close to the data while acknowledging that the top-down analysis included “conceptual leaping” was key (Klag & Langley, 2013). Three interviewees and another researcher were brought in during the final phase of analysis to review the process of analysis and findings, leading to additional refinement (DeRond & Lok, 2016). Evidence is presented in detail in the findings section.

The Emergent Sources of in Extremis Leadership

Upon arriving at the CMO, the researcher began a qualitative examination of crisis leadership by asking interviewees the following question: “Who are the leaders [of the CMO]?” Rather than being straightforward, though, the answer to this question seemed quite complex. First, the interviewees described the CEO and the director of the business area (BA) affected by the crisis as the formal leaders of the corporation in general, as well as during this crisis. However, in accordance with the crisis preparedness plans, the chief of staff was the formal leader of the CMO. Other formal leaders within the CMO included the unit leaders (e.g., the head of the communications unit) that had been appointed. Thus, formal leadership was both external and internal to the CMO.

Furthermore, almost all interviewees referred to several other informal and emergent leaders as key contributors to leadership in the CMO. As an interviewee said during our first snapshot interview:

There are more leaders here than you could have possibly imagined [before the crisis]. For instance, we’re flooded with corporate staff and line leaders, and I am a leader as well, although no plans describe this role. (II8)

As the interviewee implied, several ad hoc leaders emerged in the CMO. They were either experts who had become responsible for a specific task domain, such as “hostage communication,” or personnel who had stepped into a role to relieve a formal leader. Furthermore, corporate staff and line leaders external to the CMO took on leadership roles in the same manner. These leaders were typically responsible for a corresponding domain in the corporate organization, such as human resources, and could offer useful advice and resources during the crisis response, although their roles were not mentioned in the preparedness plans. Overall, interviewees consistently referred to four sources of leadership during the crisis. While two of these sources were formally defined in preparedness plans before the crisis, two were informal and emergent.

In Table 1, the framework developed by Morgeson et al. (2010) has been adapted to this situation to show that leadership could stem from four sources that were either formal or informal and either external (a permanent corporate organization) or internal (a temporary CMO). According to these designations, the sources of leadership could be classified as follows: (a) the CEO and BA director: formal–external, (b) the chief of staff and unit leaders: formal–internal, (c) ad hoc task leaders: internal–informal, and (d) corporate line and staff leaders: external–informal.

OPEN IN VIEWER

Table 1.

Four Leadership Sources During the Corporate Crisis Response to the Terrorist Attack.

		Formality of leadership
		Formal	Informal
Locus of leadership	External	CEO/BA director (a)	Corporate staff/line leaders (c)
	Internal	Chief of staff /unit leaders (b)	Ad hoc leaders (d)

Note. Adapted from Morgeson et al. (2010). Leadership sources in quadrant (a,b,c, d).

Emergent Leadership as a Collective

The interviewees described that leadership during the crisis event was, in addition to being provided by many sources, predominantly shared by two or more people in different ways, thereby taking multiple forms. Collective forms of leadership occurred within all the sources of leadership in the quadrants depicted in Table 1. For example, when the crisis struck, the CEO was abroad on business travel, so the BA director acted as his deputy. However, the BA director and the CEO continued to share their role as external–formal leaders after the CEO’s return. As one interviewee stated:

He [the CEO] and [the BA director] really work together in concert. (IF3)

This dyadic leadership was evidenced by the two external–formal leaders ensuring that one or both was always available to leaders in the CMO either over the phone or in person.

The corporate staff and line leaders, classified as external and informal leaders in Table 1, also described sharing leadership in different ways. For example, some corporate leaders found it necessary to share leadership during the day and rotate it during the night. As one leader said:

We made sure that Friday through Saturday [during the night], it was one of us, and then there was another one on Saturday, and on Sunday, there was a third. We shared the tasks. During the daytime, we were all present, more or less. (EII2)

Furthermore, collective leadership involved leaders from different quadrants leading together. For example, in the response unit, leadership was mostly shared by three people: one formal and two informal leaders. One of the informal leaders described the triad as follows:

I am part of the triangle that leads this. During the daytime, it’s me and [name of formal–internal leader] and [name of internal–informal leader]. During the nighttime, it’s [name of internal–informal leader] and the [name of internal–informal leader]. (II3)

Although the interviewee had never been formally acknowledged as a leader, he was described by many others as part of the leadership triangle. Similar leadership constellations were described by the interviewees throughout the response structure.

Finally, collective leadership instances could involve leaders from all four quadrants and include different people at different times, depending on which problem was the most critical. On this topic, one of the leaders said:

You know, I’ve told you that we have micro-meetings with kind of all of us every now and again? It’s typically the CEO, some corporate guy, me or [the response unit leader], and one of the [task force] guys. Different people at different times, I guess. The critical people. The people who know best right there and then. I have to tell you, I think this is where leadership happens. We’re doing it together, the four of us, this leadership thing, when it comes to the bigger purpose. (IF1)

Leadership was carried out by multiple leaders working together, while constellations varied depending on the criticality of pressing problems and situations. This underscores the centrality of leaders engaging in efforts beyond the boundaries of their roles, whether they were formal or informal and external or internal to the CMO, to achieve the CMO’s overall purpose. Thus, perhaps the most unique characteristic of the emergent and collective leadership was that leaders engaged in what could be referred to as role boundary transgressions.

These role boundary transgressions involved leaders stepping out of their roles, sharing roles, and taking on roles that did not exist in the preparedness plans and procedures established before the crisis. For example, the CEO described himself as a “tool” for the leaders in the CMO, suggesting that he had transgressed the boundaries of his role as an external leader of the CMO. The external corporate line and staff leaders also performed functions within the CMO, similarly transgressing the boundaries of their other roles, as they had no formal roles in the CMO. The ad hoc leaders who emerged internally within the CMO are examples of the same.

Thus, leaders came together to not only lead within a specific role, function, or level but also to engage in role boundary transgressing leadership. Among the explanations for engaging in these transgressions, the leaders mentioned the high stakes, urgency, ambiguity, and novelty of various situations. While performing under immense pressure, both the formal and emergent leaders reported feeling a sense of responsibility and need for control and emphasized having a “fear of glitches,” which explains why individuals from different quadrants led at the same time.

Overall, the emergence of different sources and collective forms of leadership hints at the nature of the leadership that unfolded during the response to the terrorist attack. Most importantly, it involved role boundary transgressions. Next, the critical functions that the leaders enacted throughout the crisis response are examined.

The Functions of Collective Leadership

The interviewees described the role of leadership in the CMO as purpose-driven and directed toward orchestrating collective efforts. However, there were several challenges to collective performance. Most critically, this was a psychologically surprising, frightening, and nearly intolerable situation to respond to, for which pre-existing strategic frameworks and plans did not really exist. Further, the existing routines and structures for problem solving were immediately deemed inadequate, and the CMT was altered and extended with multiple emergent units. In addition, while under tremendous pressure to perform, the personnel had to cope with both cognitive overload and severe emotional stress while interacting in new ways, potentially creating tensions and conflict. At the same time, new personnel unfamiliar with working together instantly had to develop ways to solve problems together. Confronted with these challenges, leaders had to engage in critical functions to facilitate collective performance.

The analysis reveals that along the performance dimension, leaders engaged in strategic framing and ad hoc structuring, and, along an interpersonal dimension, they engaged in relational coping and instant developing. Table 2 depicts the leadership dimensions, functions, and sub-tasks. Each is examined more closely in the following section.

OPEN IN VIEWER

Table 2.

Summary of Leadership Dimensions, Functions, and Subtasks.

Dimension	Function	Description of Functions and Subtasks
Performance dimension	Strategic framing	This function involves framing the extreme situation by making sense of it and devising a course of action in response to evolving, ill-defined, and ambiguous problems rather than executing a ready-made plan, as well as proving meaning in relation to choices made. This includes: Making sense of inadequate information (sensegiving); Anticipating potential crisis developments (proactive forecasting); Making changes in accordance with these anticipations and setting priorities for what to address (prioritizing); Explaining the rationale behind strategic choices (meaning-making).
	Ad hoc structuring	This function concerns ad hoc mobilizing and altering response structures in a flexible manner to fit strategic choices and developments in an evolving extreme situation. This includes: Notifying personnel, assigning them roles and responsibilities, and altering structures according to changing needs (mobilizing); Timing dependent activities and ensuring interaction between defined response domains (coordinating); Ensuring access to resources within and across the system (boundary spanning).
Inter-personal dimension	Relational coping	This function concerns facilitating coping though social support to personnel who are ill-prepared and to manage the relational tensions that may occur in a pressing, extreme, and psychologically challenging situation. It includes: Conveying that crisis reactions may occur and how to deal with them (fostering coping); Mitigating the reactions that occur (regulating emotions); Addressing potential relational tensions and conflict (facilitating cooperation).
	Instant developing	This function concerns the instant developing of personnel who are ill-prepared for managing the sudden and extreme situation. This includes: Displaying relevant behaviors (serving as a role model); Providing positive and negative feedback (reinforcing); Providing brief instructions and encouraging participation (instructive encouragement).

Leadership Dynamics Over Time and Across Levels

While the overall purpose of containing the situation remained the same throughout the eight days that the CMO was mobilized, the proximal and distal goals changed depending on the developments in the crisis situation (Figure 1). Further, the leaders engaged in the same functions in a repetitive and cyclical manner and in different ways, depending on what the situation warranted. However, while the emergence of additional leaders provided much needed leadership capacity and enabled efforts to become more specialized, this could have led to misalignment and fragmentation across the CMO, as the enhanced leadership capacity increased the need for alignment in the corporate crisis responders efforts. Thus, the importance of role boundary transgressions become evident. Next, we examine how the emergence of leadership capacity influenced the occurrence of role boundary transgressions to foster alignment of efforts occurring over time.

Driving, Enabling, and Boundary Conditions

The leadership that took place was described by the interviewees as “exceptional” amid the unfolding situation. Perhaps therein lies some cues to how and why such leadership could occur. It was driven by the extreme nature of the crisis event, in which urgency, criticality, and threat reached a threshold beyond the capacity of the day-to-day organization, requiring leadership that differed substantially from that in other situations. Nevertheless, the extensive emergence of informal leadership as well as the boundary role transgressions would probably not emerge in all organizations or in all situations. Indeed, in many organizations, people would be likely to feel that they would have to defer to formal leaders, “stay in their lane,” and hold to standardized processes and systems. For leadership of this kind to occur, people would have to both want to and feel that they could adapt and shift their roles and take charge, and at the same time this would have to be allowed by other people. According to the data, three enablers seem to have facilitated this kind of leadership: (a) flexible crisis preparedness plans and procedures; (b) a pool of procedurally trained crisis management personnel; and (c) a crisis management culture.

Who Leads Across Structures During Such an Extreme Situation?

Through the first set of findings, this study sought to explore who leads across ad hoc and temporary mobilized structures during in extremis leadership situations. Intriguingly, leadership was enacted by multiple formal and emergent informal leaders leading together. While notions of shared and collective leadership have been illustrated within teams across a number of contexts (Day et al., 2004; Denis et al., 2012; Pearce & Conger, 2003), extreme context research has tended to focus on the effectiveness of individual formal leadership (Avolio et al., 2022; DeChurch & Marks, 2006; Dixon et al., 2017). However, this study indicates that addressing only one source of leadership may not provide an adequate account of the leadership involved in these situations. Instead, leadership during the terrorist attack was both formal and emergent and both external and internal to the CMO, in line with Morgeson et al.’s (2010) leadership source quadrant. In disruptive extreme events and contexts, there is an immense demand for immediate and effective leadership (Hällgren et al., 2018; Hannah et al., 2009). The findings underscore that, in particular, the emergence of informal leadership seems to alleviate pressure to immediately perform and enable a response to an array of threatening, complex, and interdependent problems at the same time.

Proposition 1: During extreme situations, utilizing emergent (informal) sources of leadership to engage in leadership functions across temporary mobilized response structures is more likely to lead to the achievement of overall purpose(s) and goals than utilizing only formal leaders to engage in such efforts due to the immense pressure to perform effectively to solve an array of complex problems.

Further, the interviewees explained that leadership during the crisis event was, with limited exceptions, carried out by multiple leaders from all four sources of leadership engaging in these efforts across structures in concert. Thus, contrary to past scholars’ propositions that leadership under pressure is demonstrably filled by one person at a time, whether by a formal or an emergent leader (e.g., Klein et al., 2006), the current study found the opposite to be the case. Indeed, the power dynamics were not as straightforward as in the case of dynamic delegation (Klein et al., 2006) or in incident command systems (Bigley & Roberts, 2001) where formal leaders can retract powers delegated to others due to their hierarchical position. Instead, the leadership during this situation appeared to share similarities with “pooling leadership at the top” in top management teams (TMTs) (Denis et al., 2012) and the notion of team heterarchy (Aime et al., 2014), in which leaders are mutually dependent on each other and power relations may change over time. It appears that the complexity of the situation, together with the risk of exacerbating it, for instance by making the wrong choices, motivated leaders to work together. This way, they could expand their cognitive–emotional problem-solving capacity, ensuring timeliness albeit without sacrificing quality for speed (Sommer & Pearson, 2007).

Proposition 2. Multiple sources of leadership engaging in collective leadership efforts in temporary mobilized response structures are more likely to achieve their overall purpose and goals than in structures in which only single sources of leadership engage in such efforts due to the expanded cognitive–emotional problem-solving capacity that working together can provide to ensure timeliness without sacrificing quality for speed.

What do Leaders Do in Such an Extreme Situation?

The second set of findings shows that the four leadership functions seem to fulfill collective needs along a performance and interpersonal dimension across the response structure in this extreme situation.

Starting out with the performance dimension, strategic framing was a key function, resembling strategy development (DeChurch et al., 2011) and including four context-specific subtasks. First, through sensegiving, the leaders attempted to make sense of what was happening (Weick, 1993) in order to enable a collective cognitive shift (e.g., perceiving the crisis or significant changes) and communicating this to others (Baran & Scott, 2010; Carrington et al., 2019; Foldy et al., 2008). Second, proactive forecasting included anticipating an unknown future while the escalation potential was uncertain to develop a prescriptive shared mental model of how to respond proactively to the situation (Pearson & Mitroff, 1993). Third, by prioritizing, leaders needed to choose what to do and when, based on limited information and access to resources, as action was needed before it would be too late (Hadley et al., 2011). Finally, through meaning-making (Boin et al., 2005), leaders attempted to influence how their choices would be perceived by key stakeholders, for instance by relating them to crisis management values. This points to the salience of “politization” of in-situ crisis leadership (Boin et al., 2005). Taken together, the subtasks within this function illustrate the need to link operational and long-term strategic leadership (Bundy et al., 2017; DeChurch et al., 2011) in dynamically developing extreme situations. Thus, leaders are pivotal in engaging in strategically framing subtasks to ensure that all responses are in line with the overall purpose(s) of their efforts while proximal goals may shift as the situation changes.

Another key function along the performance dimension was ad hoc structuring, which included three subfunctions. The interviews revealed that in-situ crisis leadership involved not only coordination within planned or existing structures (Bigley & Roberts, 2001; DeChurch et al., 2011) but also the ad hoc structuring of the response itself. First, mobilizing involved notifying more people and structures than the plans could account for and scaling and restructuring roles, units, and activities as unexpected developments occurred (Beckhy & Okhuysen, 2011). Second, coordinating included integrating a collective set of interdependent tasks in a timely manner (DeChurch et al., 2011), such as through synchronized schedules, routines, and status meetings. This placed a heavy toll on the workload of leaders, compared to, for example, those in emergency ward units, where coordination patterns are usually more stable (Hannah et al., 2009). Third, boundary spanning, a collaborative external–internal linking activity (Tushman, 1977), included defining domains and ensuring access to resources across the improvizational system. Importantly, placing the ad hoc mobilized structure in such a way that it reported directly to the CEO aimed to resolve the paradox of needing to expend more resources to resolve the situation while needing to centralize authority to align efforts (Dutton, 1986). While being “kept close,” however, more than through authority-based control, this enabled fast structural adaptation to shifting demands, supporting the implementation of strategic choices in a dynamically evolving situation. Taken together, leaders are pivotal for engaging in the ad hoc structuring subtask to ensure that all responses are in line with the overall purpose(s) of their efforts while proximal goals may shift as the situation changes.

Proposition 3a. Engaging in strategic framing leadership behaviors is positively related to achieving the overall purpose(s) and proximal goals across temporary mobilized response structures, compared with not engaging in such functional behaviors.

The first function identified within the interpersonal dimension is relational coping. The interviewees not only described the horror of the terrorist attack and siege but also performing under pressure to potentially trigger psychological reactions and conflicts. While managing emotional and relational aspects tends to be overlooked in the crisis literature (Kahn et al., 2013; Wu et al., 2021), such aspects appear important to manage. First, through fostering coping, leaders provided information about how crises can impact people psychologically to prevent and “take the edge off” some of these reactions and promote cognitive coping (Brown et al., 2005; Lazarus & Folkman, 1984). Second, in case stress emotions threatened to become overwhelming, leaders would engage in regulating emotions, for instance by seeking support from peers or an on-site psychologist to influence the valence and intensity of such emotions (Nesse & Stensaker, 2021). Third, in a situation in which conflicts, tensions, and power struggles could have easily caused response inertia (Boin et al., 2005), leaders attempted to prevent this by facilitating cooperation. This included involving responders in discussions to create commitment and resolve tensions before they would turn into conflicts across the structure. This appears to have been particularly important, given that achieving the overall purpose may require a change in priorities in multiple units, potentially causing tension and conflict (Carter et al., 2020). Taken together, leaders must engage in the relational coping subtasks to ensure that all responses are in line with the overall purpose(s) of their efforts while proximal goals may shift as the situation changes.

Second, along an interpersonal dimension, leaders engaged in the function of instant developing. While performing under pressure during the terrorist attack, any thorough attempt to develop personnel was made difficult; however, this seems to have prompted learning in other ways. First, leaders would engage in role modeling. By relying on learning through observations, they would display behaviors for others to adopt (Bandura, 1977), which was an effective way of teaching others with minimal effort. Second, the leaders would engage in reinforcing desired behaviors. Past research has shown that while providing positive feedback is more effective in self-managing teams, negative feedback may be warranted in crisis situations (Morgeson, 2005). In this situation, however, the leaders would reinforce positive behaviors, as they found that negative feedback could cause stress reactions in a pressing situation with limited time to follow up. Third, leaders would engage in instructive encouragement, for instance by declaring what to prioritize, yet not how to do it, and stating their confidence in others to figure it out. Indeed, they would not always know how to solve a problem themselves but had to trust that others would. This contrasts with past research, for instance in medical emergency ward settings (Klein et al., 2006), where more senior personnel are expected to retract delegated authority in urgent and critical situations. Typically, developing personnel is perceived as a within-team function (Morgeson, 2005). Thus, while it may be surprising that leaders from all quadrants engaged in this, it illustrates that in extremis leadership requires having to “learn on the job,” irrespective of, for example, seniority and experience. Taken together, it appears pivotal for leaders to engage in instantly developing subtasks to ensure that all responses are in line with the overall purpose(s) of their efforts while proximal goals may shift as the situation changes.

Proposition 4a. Engaging in relational coping leadership behaviors is positively related to achieving overall purpose(s) and proximal goals across temporary mobilized response structures, compared with not engaging in such functional behaviors.

How Does Leadership Develop Over Time and Across Levels to Align Collective Efforts?

The third set of findings concerns how leadership dynamics unfolded over time and across levels. First, the observations and interviews revealed that while actions changed as warranted by the situation, the core leadership functions remained the same throughout the crisis. Particularly, in line with the Marks et al.’s (2001) typology, to achieve the overall purpose, the leaders appeared to engage in strategic framing during transition/planning phases (when engaging in strategic framing) and in the three other functions during action phases in a cyclical and repetitive manner until the situation had been resolved. This indicates that the taxonomy developed for teams by Marks and colleagues (2001) is homologous across levels. Furthermore, the emergence of (informal) leaders in the CMO reflects the finding that in teams, leadership first serving as an input may later become an output that can increase leadership capacity (Day et al., 2004). Here, this occurred on a much larger scale, as interviewees described it occurring at the system level, where emergent leaders engaged in the four leadership functions leading new and emergent units and teams. Over the course of the crisis, enacting the four functions repeatedly in multiple affected domains provided not only leadership capacity but also an increasingly specialized response. This created what past research has referred to as “bricolage,” or the structuring and restructuring of adaptive, temporary, and ad hoc mobilized cross-functional expert teams (Beckhy & Okhuysen 2011) exerting “pockets of control” (Wolbers et al., 2018) and increasing in number over time. Being able to span an increasing number of domains where control is regained is key to containing a crisis (Boin et al., 2005; Pearson & Clair, 1998) and contributes to the overall purpose of response.

Proposition 5. Engaging in functional leadership behaviors in a cyclical and repetitive manner in multiple emergent specialized units in different affected domains until the situation is resolved is positively related to achieving the overall purpose(s) and proximal goals across temporary mobilized response structures, compared with not engaging in functional behaviors in a cyclical and repetitive manner in specialized emergent units over time.

While achieving much-needed leadership capacity in multiple units and domains made it possible to specialize efforts, this could have led to fragmentation, thereby challenging the alignment of efforts over time. In particular, representational gaps and inconsistencies regarding how shared problems are conceptualized and resolved (Cronin & Weingart, 2007; Firth et al., 2015) could occur. The leaders, mindful of the need to bridge these gaps, engaged in role boundary transgressions, extending and stepping out of their roles to engage in leadership functions across levels and units over time. Through the role boundary transgressions, the leaders were able to create a shared mental model of where, when, and how alignment was needed across the system (Murase et al., 2014). It is worth noting that while synchronization of actions across teams is critical for success (Davison et al., 2012; Marks et al., 2001), temporary fragmentation is not necessarily counterproductive (Wolbers et al., 2018). When specialization was at a high level, some degree of fragmentation was inevitable and seemed to have been effective. This is because multiple important proximal goals could be pursued in different specialized units at the same time, thereby contributing to achieving the overall purpose of containing the crisis (Wolbers et al., 2018). This was observed by the leaders, who would only intervene if convergence, such as a system-level change in priorities, did not occur in due time. The leaders, dependent on one another to succeed, experienced this interdependence changing over time, and it was strongly related to who could achieve the overall purpose at what time. Specifically, at the outset of the crisis, the external leaders’ role boundary transgressions were aimed towards aiding the internal leaders to manage the initial situation, whereas toward the end, the internal leaders’ role boundary transgressions were aimed towards aiding external leaders to normalize the situation.

Proposition 6. Engaging in dynamic role boundary transgressions to achieve alignment over time is positively related to achieving the overall purpose(s) and proximal goals across temporary mobilized response structures at the outset, specifically when external (formal and informal) leaders support internal leaders, and towards the end (demobilizing) when internal leaders support (formal and informal) external leaders.

This research has important implications for the understanding of the emergence and dynamics of leadership across ad hoc mobilized structures in extreme crisis situations (Hällgren et al., 2018; Hannah et al., 2009). First, examining the more disruptive end of the extreme events continuum, leadership is illustrated as planned and emergent, dynamically evolving, and “pooled” with representatives from across the system. Drawing on pluralistic leadership perspectives (Denis et al., 2012), the study complements past research that has predominantly focused on individual leadership styles. This kind of leadership ensures time-limited and purpose-driven leadership claimed by and granted to numerous leaders from different leadership sources (Morgeson et al., 2010), supported and limited by specific enabling and boundary conditions. The interplay of different sources of leadership (Morgeson et al., 2010), claims, and grants of leadership (DeRue & Ashford, 2010), as well as the conditions that enable such leadership, should be taken into account and theorized further in advancing research on leadership in disruptive contexts in the future. Further, as organizations are likely to experience extreme situations for which resilience needs to be developed in advance (Hällgren et al., 2018; Williams et al., 2017), they need to prepare to combine planned and impromptu efforts. In high reliability organizations (HROs) such as the one studied, there is a tendency to rely on formality and preparedness plans (Selart et al., 2013). Yet, as this study shows, during extreme events, this appears not to be enough. Instead, organizations could benefit from developing plans and procedures that allow for flexibility in how leadership and response structures are organized based on situational needs and demands. This would include granting both formal and emergent crisis leaders the authority to act on behalf of the organization in critical and urgent situations to achieve the overall purpose of resolving a crisis.

Second, while some of the leadership functions and subtasks reflect prior work on what leaders do across structures, others differ in significant ways. As Carter et al. (2020) noted, most leadership research across structures (e.g., in multiteam systems) places an overwhelming emphasis on performance-related behaviors such as strategizing and coordinating (see, e.g., DeChurch et al., 2011; DeChurch & Marks, 2006). The lack of attention toward interpersonal aspects of leadership in these situations appears to be an oversight. Indeed, all the leadership quadrants appear to engage in behaviors related to regulating emotions and developing personnel. However, it is also worth noting, like Klein et al. (2006), we did not find “motivating” to be a key leadership function, despite it being important in most other settings. The demands of the situation, such as saving fellow employees from being killed by terrorists, appeared to be intrinsically self-motivating. This underscores that leveraging context-specific perspectives (Hällgren et al., 2018; Morgeson et al., 2010) may be necessary to further advance our understanding of what constitutes effective leadership in disruptive contexts. In practice, organizations could develop a pool of potential crisis leaders who have received training to be ready to emerge as leaders when such situations unfold, providing immediate leadership capacity. Specifically, the training could include developing leaders to execute the key leadership functions outlined in this study, using training needs analyzes to identify who would need what kind of training (Salas et al., 2012), as well as using crisis scenario simulations to provide training transfer to a real setting (Waller et al., 2014).

Third, contributing to understanding leadership dynamics, the study showed that while emergent leadership may provide much-needed capacity, it poses a threat to alignment, which is managed through role boundary transgressions across roles and levels. This contrasts with the ideals of command-and-control leadership, as seen in incident command (Bigley & Roberts, 2001) with emergency ward dynamic delegation systems (Klein et al., 2006), where control is achieved in critical moments through the top-down, hierarchical formality of leadership. According to this study, a disruptive context requires different and more dynamic leadership arrangements. Traditional means of control are shown to be relaxed, and control is regained without stifling initiatives and efforts that emerge. This reflects the work by Wolbers et al. (2018), who showed that control is gained not only through alignment but also fragmentation. This study, however, underscores the centrality of leadership in bridging the representational gaps caused by fragmented response, where role transgressions, occurring in the shadow of the (ordinary organization’s) hierarchy, allow fragmentational aspects to shift and be aligned over time and thereby support reaching overall goals. Despite the focus on alignment, such as in multiteam systems (MTS) research, this suggests that examining leadership dynamics that foster goal achievement may be more important than studying alignment in itself. Given that leadership arrangements of a hybrid and dynamic nature appear to be important,for instance when averting mass casualties and promoting survival is the purpose, role transgressions should be further theorized in such contexts. In line with this, organizations could benefit from developing a crisis management culture that would promote emergent and formal leaders to engage in role boundary transgressions to foster alignment of efforts over time and across levels. This “super function” is mutual and collective in nature, requiring leaders to accept shared responsibility, in contrast to notions and the strong-lived yet unlikely ideal of leadership as heroical, individual, and hierarchical in these situations.

Finally, a methodological implication of the study is the demonstration of the value of using a real-time explorative qualitative design in an under-researched leadership context. The study reveals aspects of leadership that are impossible to reach through quantitative, variance-based analysis, especially when it involves a single leadership source and measures leadership at one point in time. Observing a single leader’s effectiveness would, for instance, not fully capture factors such as the sources, forms, functions, and dynamics of leadership. Compared with other inductive methods, such as historiometric analyzes, our research design allowed us to avoid methodological issues such as relying on archival or recalled data, thereby reducing retrospective biases. For instance, real-time observations appear more sensitive to capturing dynamic and interpersonal dimensions of leadership than archival data. In particular, the centrality of role boundary transgressions in the alignment of efforts probably would not have been revealed through the most common research designs.

Limitations and Future Directions

This study is unique in that we were able to gather data during an extreme situation due to access that is rarely granted to researchers (Sommer et al., 2016). Such real-time access is necessary to advance this field of research, particularly pertaining to capturing the dynamics among key variables as they occur in the moment. This is crucial for theoretical development but also for the practical purpose of contributing to discovering how to navigate such situations in the future. However, the current study has some potential limitations, and we highlight the three most important ones.

First, while the inductive design in this study serves to contribute to generating theory, affording external validity in terms of identifying meaningful aspects of leadership in this context comes at the expense of being able to make inferences related to internal validity. Second, the study is based on a single case and reflects leadership in response to a specific situation. Therefore, the sources of leadership, functions, and developments over time and across levels might not apply during all kinds of similar situations. This poses a call for more similar studies to validate the taxonomy developed in this study. Third, the study assesses leadership occurring in physically distant headquarters, in a unit that is not itself directly in danger. Thus, it is possible that leadership may be different, either when in closer proximity or when the leader and the team that they lead is personally in danger or impacted by the event in some way.

Future research could benefit from using explorative and real-time research designs, as the understanding of this phenomenon is still nascent, rather than foregoing such aspects by using more mature research designs and risking not capturing important parts of the picture (Edmondson & McManus, 2007). Several avenues for future research are proposed. First, to deepen our understanding of the emergence of leadership, future studies could examine how emergent leadership is claimed, granted (DeRue & Ashford, 2010), rejected, and retracted in these situations. This would be especially useful for examination when it comes to collective leadership constellations and how these change over time to effectively manage extreme situations. In addition, examining the emergence of ethical and/or abusive leadership and which mechanisms foster one or the other, as well as if and how different mechanisms (such as shared leadership configurations) could counteract more destructive forms of leadership from occurring, would be valuable.

Another research avenue concerns examining the leadership functions along the interpersonal dimension across structures more closely. This has tended to be neglected in extreme situation research to date (Carter et al., 2020; Wu et al., 2021), albeit with some exceptions (Kahn et al., 2013; Nesse & Stensaker, 2021), yet it appears to be crucial for effective leadership in these situations. Future studies should be extended to examine functions along this dimension, their relative importance, and their interplay with other functions more thoroughly.

A third research avenue concerns leadership dynamics over time, especially in relation to power and influence, in emergent structural arrangements. Examining different organizing principles for how power and influence is distributed and reconfigured dynamically to avoid response inertia (Boin et al., 2005) and gain control is crucial and involves exploring when to foster alignment and when not to, in order to resolve such situations effectively.

Finally, when it comes to leadership in an extreme context, a research avenue would be to examine how, when leadership is most needed, leaders dare to take the lead beyond roles and expectations. This is evident in the centrality of the very brave relational activity of transgressing one’s role’s boundaries for the greater good, and so risking becoming more vulnerable than if not doing so.