Studying collective leadership: The road ahead

Challenge 1: Collective leadership as an ambiguous leadership space

A primary challenge associated with CL is the ambiguity of the space in which it resides. Three factors influence this observation. First, there are many forms of CL, as our earlier mapping exercise and Rapid Appraisal make clear (see also Denis et al., 2012; Pearce et al., 2009). In this special issue, for example, Spiller et al.’s article (2020) on Māori ecosystems defines CL as relationships that are distributed across generations in living and non-living entities. The article by Clifton et al. (2020) reflects the negotiated sharing of leadership among team members in a hierarchical context; and Langley et al. (2020) study co-leadership dyads in a healthcare setting.

Second, the distinction between collective and hierarchical leadership is frequently equivocal and diffuse. Knowledge workers, dispersed sources of authority, unclear roles and responsibilities, beneficial practices with a dark side, and the fluidity of participatory decision-making can all contribute to rendering authority more uncertain and negotiable (Chreim, 2015; Denis et al., 2010; Gronn, 2009, 2015; Holm and Fairhurst, 2018; Quick, 2014). Hierarchical influences can also be both emergent and deliberate (Denis et al., 2012). For example, they may permit experiments or fluctuations (Lichtenstein and Plowman, 2009), boost it through collective visioning processes (Ensley et al., 2006; Houghton et al., 2003), or actively impede it with autocratic or aversive management styles (Pearce and Sims, 2002). In this special issue, Maupin et al. (2020) also caution analysts to consider multiple levels of analysis beyond the individual and the collective to consider the system dynamics that may further contribute to the ambiguity of the leadership space.

Finally, despite the prevalence of qualitative work in the extant CL research, there is a positivity bias in CL research, emphasizing the rational (e.g. depoliticized relations and goal convergence), short-term (e.g. devoid of historical influences), and separateness from hierarchy (e.g., by asking, ‘Which is better?’ versus ‘How do they interrelate?’). By excluding factors such as these, many researchers have sought to control the context to foster more generalizable findings; but the unfortunate side effect has been to underplay issues of disorder, power, competition, and conflict (Chreim, 2015; Denis et al., 2012; Gronn, 2015; Sergi et al., 2012; Smith et al., 2018).

To address the ambiguity of the leadership space, our first guideline to scholars is to decipher the configurations of CL and its power-laced foundations. In providing this guideline we are influenced by Peter Gronn (2011: 445–446), who argues that a configuration is a constellation of parts emerging as wholes, such that the latter are no longer reducible to their parts, and they exert a downward causal role in influencing members’ thinking and acting. In other words, configured members are consciously aware of acting as a decision-making unit, not merely as isolated or aggregated individuals (Gronn, 2015).

In a configurational analysis of an ambiguous leadership space, we should expect three features. First, expect hybrid relationships to blend formal and informal roles, cross levels of hierarchy or whole organizations, and/or shift back and forth over time (Davis and Eisenhardt, 2011; Gronn, 2011). This was certainly the conclusion of the article, which demonstrated that the ‘doing’ of leadership is not limited to the formal leader in a team meeting, as informal leadership emerged both in conjunction and in conflict with formal leadership.

Second, we should expect variety in the CL forms. Individual specializations may vary, and collectives may divide work differentially, thus complement each other in different ways (Chreim, 2015; Denis et al., 2010; Hodgson et al., 1965). Indeed, one of the key jobs of the analysist is to chart the contours of this variety, much as Langley et al. (2020) found with the six power-laced configurations of co-leadership across 20 dyads in four healthcare organizations: dyads of one (who does all of the work); consulting dyads (hierarchical, but involving of the other); boundary duos (equal due to different boundary spanning expertise); management duos (who work separately); management units (who work interchangeably); and mission units (who are mission driven to work interchangeably). This work is emblematic of a configurational analysis.

Finally, we should expect role ambiguity, tension, contradiction, paradox and, above all, the operations of power (Collinson, 2014; Denis et al., 2010; Endres and Weibler, 2019; Foldy, 2019; Ospina and Saz-Carranza, 2010; Schnurr and Schroeder, 2019). A configurational analysis is inescapably tied to issues of power and influence because it goes to the heart of the communication process. To wit, communication does not just convey content, but defines relationships of power and control in every utterance (Watzlawick et al., 1967) and in relatively fixed and reproducible patterns however conflictual or convoluted those patterns might be (Fairhurst and Uhl-Bien, 2012; Fairhurst et al., 1995). Because CL is inescapably embedded within a field of power relationships (Gronn, 2015), a configurational analysis is ideally suited to delineate the contours of that field. Indeed, one of the reasons that Langley et al.’s (2020) configurational analysis stands out is due to the subtleties of the power dynamics associated with the dominance of a managerialist institutional logic.

However, as the article by Maupin et al. (2020) make very clear, there are many ways to document configurations, such as through discourse analysis, relational event modeling, and dynamic network analysis, to which we would add ethnographic observation, interaction process analysis, and other related methodological approaches. The power-imbued question, ‘Who, or what unit, has established the right to say what to whom and when?’ must be answered, and Maupin et al.’s (2020) strong argument to combine methods to arrive at a more in-depth analysis that captures nuances and broad patterns to answer this question appears increasingly warranted in complex organizational environments.

Connecting this challenge with the theoretical map, note that these power-infused configurations happen at both the group and systemic levels (Dimension 1), so this guideline is broadly relevant. Moreover, it is valuable, though in different ways, for scholars viewing CL as type and as lens (Dimension 2). Type researchers have a fruitful history of documenting various configurations, which should certainly continue, but these scholars have been less successful in tracing the role of power in how those constellations emerge and are either sustained or undermined. Lens researchers can follow this guideline by asking how and why people construct the configurations that they do. How do particular meanings of leadership and attributions of leadership result in, and become the product of, particular configurations? From a Foucauldian perspective, the very act of construction represents a struggle for power.

Challenge 2: Addressing the question, ‘What is collective leadership?’

A second and related challenge concerns definitional issues inherited from traditional leadership studies, but made worse by the ambiguity of the CL leadership space. The definitional issues to which we refer involve leadership’s plasticity as a concept—that is, the numerous language games played in and around leadership by actors, but also by analysts for whom definitional consensus has been elusive (Fairhurst, 2007; Grint, 2011; Kelly, 2008, 2014). As traditional leadership scholars have wrestled with the multifarious meanings for leadership, the cleanest strategy from a methodological point of view has been (and still is) to settle on an a priori definition linked to organizational role, authority, or network position.

However, leadership can be especially difficult to define in ambiguous spaces because it may not be seen as relevant. For example, in their extensive review of the literature, Denis et al. (2012: 274) queried whether or not leadership disintegrates as a concept in collaborative settings: ‘The interactive processes whereby leadership is produced easily shade into decision-making, collaboration, or simply work. When “leadership” can no longer be attached to individuals at all, there is a danger that it may become a chimera.’

Taking their cue from traditional leadership research, CL analysts may sidestep this issue by setting the meaning of leadership as information brokerage in network studies (e.g. DeRue et al., 2015; Fransen et al., 2018; Marion et al., 2016) or organizational role in team interactions (for a review, see Denis et al., 2012). Yet, the same problems persist in CL as with traditional leadership, i.e., why those in formal managerial or supervisory roles may not be seen as leaders) (Bedeian and Hunt, 2006), or why others lacking such roles may well be (Quinn and Spreitzer, 2001). The same could be said for network centrality but, more than that, the ability of whole networks to be seen as acting in a leadership capacity (Gronn, 2015).

The expediency of analysts’ setting the meaning of CL a priori because of pragmatic research demands or methodological appeal is understandable. However, too often it ignores how assigned and emergent roles vis-a-vis individual or collective agency interact with attributions of leadership, or the lack thereof, based on advancing the task or other outcomes, influential acts of organizing, asymmetric influence, and the like. Given the specter of the disintegration of leadership as a concept in collective settings, cannot more be done to refute this argument?

Thus, we argue that, while configurational analyses are essential, they are not sufficient. We follow Iszatt-White (2010: 411; cited in Chreim, 2015) in this regard, who argues:

When we talk, for example, of ‘delegating’ or ‘participating’ leadership behaviours . . . we need to step outside the mere technical definitions of these terms to understand . . . what these kinds of leadership really look like in a given setting. And it takes situated knowledge of individual leaders, their settings and followers, to determine whether a specific behaviour will be received as participating leadership or something else entirely (interference, for example, or inability to delegate) . . . [The meaning of an act of leadership] is inseparable from its situated occurrence, rather than being a property of the act itself . . . [L]eadership is ‘brought off’ as leadership . . . as members of the setting orient to particular leadership practices in a manner which recognizes them as such.

As such, our second guideline is that, in some way, analysts must establish how leadership is made relevant in a collective setting. Certainly, the easiest way to do this is to collect attributional data to see how leadership, or the failure of leadership, enters into the vernacular of actors either in terms of individual or conjoint agency or both. However, researchers, depending on their theoretical bent, may disagree with actors about whether leadership is or is not present. Another approach, drawing from multiple measures research, is to consider more than one definition of leadership and let them creatively play off one another as the data tell their story (Fairhurst, 2017). Endres and Weibler (2019) recommend a version of this by suggesting that researchers capture the influence component at the collective level (read: configuration), but also the joint-motivational identity-based source of this influence component.

In their study of the implementation of shared leadership in a Danish municipality, Holm and Fairhurst (2018) assessed CL both in terms of role sharing, a commonly accepted practice in shared leadership research (Denis et al., 2012), but also leadership attributions. The latter proved instrumental in interpreting ambiguous role-sharing behaviors between the formal leader and his senior team. Had they studied role sharing alone, they would neither have established the relevance of leadership vis-a-vis what actors oriented to, nor fully understood how to interpret the CL–hierarchical configuration.

In this special issue, Sklaveniti (2020) establishes the relevance of leadership by relationally immersing herself in her ethnographic fieldwork, which allowed her to continuously (re)construct understandings of leadership moments together with research participants. She uses this strategy to good effect by zooming in on key moments and zooming out to discern turning points and other effects across meetings and workgroup interactions. Similarly, the study of Māori ecosystems in this special issue by Spiller et al. (2020) examines the enactment of leadership discourses derived from the culturally derived Māori worldview, mātauranga Māori, which defines the terms of individual and CL enactment in service to others and expressions of the collective will.

This guideline has relevance for the variety of scholarship captured in our theoretical map. Scholars who see the group or the system as the locus of leadership need to directly attend to how CL is different from other processes: for example, at the group level, CL must be analytically decoupled from teamwork and team processes, while system-level researchers should differentiate CL from other forms of coordination. Scholars defining CL as type can establish the relevance of leadership by documenting plural forms within formal structures, but must avoid the danger of confounding leadership with formal authority. Scholars defining CL as lens can benefit from this guideline in order to overcome obstacles associated with an interpretive stance. For example, interpretive researchers are reluctant to enter the field with pre-established definitions of leadership or of asking about leadership directly. However, this threatens establishing the relevance of leadership among actors. One way to address this is to develop a theoretical scaffolding (Thorne, 2016) that includes a tentative understanding of leadership (Sanchez and Ospina, forthcoming), for example, as an unequal influence relationship that produces direction (Crevani and Endrissat, 2016). Researchers could also explore questions such as: What do we gain or lose from using concepts such as leadership versus collaboration versus teamwork? Whose interests are served by one term versus another?

Challenge 3: Embracing process fully

A third challenge in conducting CL research is to move away from management studies’ long history of: (a) collecting data that produces static snapshots; (b) theorizing variance from those data; and (c) deriving weak process models, in which arrows connect to preexisting and stable boxes suggesting the temporal evolution of things or substances whose identity remains intact (Bakken and Hernes, 2006; Feldman, 2017; Langley and Tsoukas, 2017: 3).

There are legitimate questions as to whether such research practices sufficiently inform analysts of the relational, experiential, and temporal details of collective and hierarchical leadership’s co-enactment in complex organizations (Denis et al., 2012; Fairhurst, 2017; Gronn, 2002, 2011, 2015). For example, can we say CL exists when a formal leader invites participation, but then (consciously or unconsciously) dominates the discussion or ignores an emerging consensus? At what point does CL exist when team members new to sharing control ask for specifically timed control moves by formal leaders? What is the hybrid relationship that evolves between a collective and the hierarchy, especially when the latter gradually becomes less controlling? Finally, what about team members who opportunistically find ways to use the hierarchy in order to sidestep it?

These examples point to the necessity of strong process models and three features that are challenging to capture: experience, heterogeneity, and temporality (Langley and Tsoukas, 2017). The experience of action and its consequences affects actors, ‘changing qualitatively the phenomena they help co-create’ (Langley and Tsoukas, 2017: 4). CL study would greatly benefit from addressing this challenge because evolving interactions and their consequences are the very stuff of leadership attributions—or the lack thereof. Understanding that the lived experience of CL in relation to hierarchy is a dynamic and ongoing process represents a significant empirical challenge, but addressing it deepens our understanding of the ontologies of leadership, whether in individual or collective forms, and thus has relevance to collaborative settings.

Likewise, heterogeneity refers to the diversity of individual experience. Even repetitive experiences occur at a unique moment in time. They are entrenched in the sequential flow of time, circumstance, and whom we are as actors at any given moment (Guerlac, 2006: 82; Langley and Tsoukas, 2017). However, the challenge of heterogeneity is also to capture multiple voices and the competing interests, discourses, and interpretations that give rise to the complexity of organizational life (Putnam et al., 2016). Indeed, the challenge of heterogeneity extends to analysts who regard actors’ leadership attributions as incidental because a particular kind of analysis excludes it.

Finally, there is the issue of temporality in which order is consequential and performative. Langley et al. (2013) characterize this aspect of process as becoming (Chia, 1996)—an orientation marked by fluidity and activities-over-product. For example, consider the autocratic leader who strategically sequences the activities in team meetings to appear highly participative. The manager’s talk time may be minimal, but she exerts authority through agenda setting and creating the official record of the outcomes. No one can accuse her of not letting her people have their say, but neither does the sequencing of these actions look completely participatory.

One of the analytical challenges of temporality is to understand the limitations of one’s bracketing of the continuous flow of actors’ experiences. For example, a survey approach asks for a retrospective summary of behavior, eliding the sequential details of how the manager above operates. An interaction process analysis of team meetings, by contrast, targets the sequential nature of the manager’s strategy to invite participation while controlling it at the front and back ends of the meetings, while deemphasizing broader contextual concerns. An in-depth ethnographic or interview study of the implementation of a shared leadership intervention might gloss over accounts or observations of meeting patterns to focus on the ways in which this manager loosened (or failed to loosen) her control orientation over time as her team became more self-sufficient. The point here is that, if one’s CL research is raising more questions than it answers, including more than one temporal lens, albeit labor intensive, it is necessary to provide the richest picture possible.

This point reinforces our third guideline, which is for CL research to adopt strong process models (Langley and Tsoukas, 2017; Maupin et al., 2020; Uhl-Bien and Arena, 2017). This is because: (a) the experience of action and its consequences is important to interpret evolving CL configurations, especially with hierarchy; (b) the heterogeneity and multiplicity of experience is necessary to understand the organizing effects of multiple and varied attributions of leadership in a collective setting; and (c) understanding the consequentiality and performativity of temporality broadens or truncates our views of CL and the discernment of its overall workings and effects. To the extent possible, future CL research needs to move in the direction of incorporating more of these elements to provide the requisite variety necessary to explain the ambiguity of the CL leadership space.

All of the empirical articles in this special issue give some attention to process and to the three features. Spiller et al. (2020), for example, see temporality in a radically different way than most process models, as time stretches into the ancient past. Moreover, future and past are relative terms, as time has more of a cyclical flow. Sklaveniti (2020) pays particular attention to the experience of participants as she identifies turning points in the streaming deluge of action. Langley et al. (2020) document the heterogeneity of consequences of actors engaging in a theoretically similar structure.

Collinson (2014: 47) argues that one way to do this is to first be wary of over-interpreting the distinctions that we draw as analysts (e.g., leader–follower, leader–manager, and so on) because ‘they can oversimplify complex relationships and reduce them to either/or polarities that downplay or neglect interrelations, tensions, asymmetries, and contradictions’. He then suggests reframing leadership binaries such as control–resistance, dissent–consent, male–female—to which we could add individual–collective, informal–formal, and many more—as multiple, intersecting dialectics (Collinson, 2005, 2014; Denis et al., 2010; Zoller and Fairhurst, 2007).

Indeed, a dialectical view can capture the dilemmatic nature of CL given the constant presence of multiple levels of hierarchy in complex organizations, the intended and unintended consequences produced by virtuous cycles of interconnecting practices that mesh and reinforce, and/or vicious cycles of those that conflict and contradict across these levels, and the ways in which multiple, competing logics and deep-seated tensions emerge in the dynamics and practices of organizing (Collinson, 2014: 47; Denis et al., 2010). A dialectical view has the potential to deliver on a strong process model of CL, especially as regards (multiple levels of) hierarchy and levels of analysis.

Again, this guideline is broadly applicable to the range of CL scholarship. Process models can be fertile ways of understanding leadership, its emergence and its changes over time, in both groups and systems. Process models are also useful for both ‘type’ and ‘lens’ scholars, though they are likely to look quite different. Type researchers often use boxes-and-arrows with clearly differentiated elements that can be put into a relatively clear temporal order, thus suggesting temporal causality. They tend to be less successful at rich illustrations of experience and heterogeneity. Lens scholars often see boxes-and-arrows as reductionist, and their models can be stronger at capturing experiences of co-construction grounded in on-going, moment-to-moment interactions without having to reduce them to something that could fit in a box. However, it would be beneficial for these scholars to foreground temporality, rather than just stay with an ontological commitment to a ‘becoming’ view of process.