Blackboxing leadership: Methodological practices leading to manager-centrism

The scholarly literature on leadership has long been characterized by leader-centrism, in the sense of a focus on individual leaders, their characteristics and actions. This tendency has been strongly criticized, not least by scholars with a critical perspective. However, we still see a strong emphasis on leaders and managers in empirical studies of leadership. In this article, we suggest that this tendency is at least in part a consequence of common methodological blackboxing practices within leadership studies. We identify two such blackboxing practices: delegation, where identification of the core phenomenon is left to informants, and proxying, when more easily defined phenomena are taken to stand for leadership. We suggest that a consequence of such practices is an unintended focus on managers, and attempts to avoid leader-centrism that rely on these blackboxing practices therefore paradoxically might result in manager-centrism.


Introduction
During the last few decades, leadership studies in general and critical leadership studies in particular have struggled to move away from leader-centrism and leader romantization. The seminal studies of romance of leadership by Meindl and his colleagues (Meindl et al., 1985) have been a forceful inspiration. Critical leadership studies have contributed with a critical gaze on how labels such as leader is used (Collinson, 2014;Learmonth and Morrell, 2017) and on tendencies to exaggerate and romanticize for instance authenticity in leaders (Tourish, 2019). Moreover, several contemporary developments in leadership theory are grounded in an ambition to de-center the leader. Theories include leadership as practice (Raelin, 2016), leadership as an interactional accomplishment (Clifton et al., 2020), leadership in the plural (Denis et al., 2012), and varieties of collective leadership (Ospina et al., 2020). Instead of focusing on individual leaders, leadership is here approached as a decentered phenomenon (Crevani et al., 2010;Ospina et al., 2020).
Yet, despite these efforts, when it comes to empirical research, managers still take center stage. In this journal, more than three quarters of the empirical leadership articles during 2019-21 had managers or other formal leaders as the focal object of study. Further, conflation of formal positions as manager and leadership processes have been a common practice not least in survey based studies, where managerial position is commonly taken as a proxy for leadership (Ashford and Sitkin, 2019;Bedian and Hunt, 2006). Together with a strong focus on individual characteristics, experiences, and actions of people in managerial positions, research has contributed to an exaggerated and romanticized view on individuals in power (Collinson et al., 2018).
Even when scholars draw attention to leadership as distributed amongst several actors or when non-human actors are drawn into analysis, it is managers and managerial activities that provide the prime focus for empirical efforts in for example shadowing and observations. While it is very reasonable to study managers within the field of organization and management, it is problematic when the fact that it is managers acting that makes the researchers label the phenomenon under study "leadership". We are not proposing a distinction between managers and leaders as different persons or roles, but we are drawing on an understanding of leadership as an interpersonal process, broadly conceived to be a relational phenomenon (Uhl-Bien, 2006) involving interpersonal influence (Rost, 1991;Yukl, 2013) and the accomplishment of direction, commitment and alignment (Alvehus and Crevani, 2022;Crevani, 2018;Drath et al., 2008). That is, we treat leadership as a social phenomenon that is not reducible to any individual or formal position. This means that researchers need to ask whether and why a certain phenomenon becomes regarded as leadership. For example, is listening "leadership" when it is done by managers, but not when it is done by their subordinates (Alvesson and Sveningsson, 2003)? In practical terms, this means that we would expect a substantiated claim for why a certain something is considered to be an aspect of leadership, other than the observation that it is being done by a person in a formally superordinate position. Moreover, we would expect to find this phenomenon also in other contexts. It makes sense to sometimes study managersbut it does not make sense to predominantly study managers.
In essence, despite various attempts to escape leader-centrism, we still see a firm focus on managers. A central question concerns the reasons for this. One argument could be that the focus on managers is simply a matter of a practical choice and that researchers could just as well have chosen, say, basketball players, participants in social movements, or high school studentsbut for some reason they did not. This might give an impression that the obsession with managers is a consequence of leader romanticism among researchers. However, we suggest that the tendency is too strong to attribute only to romanticism. Instead, we argue that this manager-centrism is partly produced by a particular combination of conceptual and methodological practices across the field of leadership studies, namely specific ways of blackboxing the core phenomenon of study. There is an extensive and deep lack of clarity about the phenomenon of leadership that is shared between traditions as diverse as quantitative social-psychological leadership studies, interpretive leadership studies, and the as-practice/collective/plural leadership movement that is covered up by a set of methodological practices. At loss when it comes to identifying the core phenomenon, researchers turn to studying managers. In brief, our argument is that current blackboxing practices in leadership studies result in manager-centrism.
We will develop our argument through a close and critical reading of a small number of illustrative and well-conducted studies from a quantitative social-psychologically oriented tradition, studies from a qualitative interpretative tradition, and from studies explicitly aiming to approach leadership as a decentered phenomenon. The few studies we have the space to address can of course not represent any of these traditions in their entirety. Moreover, we wish to emphasize that we are not intending to criticize these studies as being faulty; on the contrary, we have chosen them as they are well-conducted studies that we hold in high regard. And being of high quality, they illustrate strong and problematic tendencies within the field as suchthey are not illustrations of failures.
Things-in-the-world, ideas, and blackboxing For our argument, we draw on a very simple distinction offered by Hacking (1999), between thingsin-the-world 1 and ideas. Things-in-the-world are, according to Hacking, precisely that: what we might see or perceive or feel, including very, very small things that we cannot see with the naked eye (such as quarks) and more nebulous things such as emotions (directly observable only to the person having the emotion). The point of the term thing-in-the-world is to distinguish it from how we label and think about this thing-in-the-world. The latter is what Hacking calls ideas. Ideas are not to be understood as only scientific concepts, but any idea anyone might have about something. He takes the example of the refugee woman. The refugee woman as a human being is a thing-in-the-world, but the label "refugee woman" is an idea attributed to that thing-in-the-world by humans that has consequences for how, for example, other people relate to that particular person. Yet, the person as a thing-in-the-world exists whether we use labels such as "woman" and "refugee" or not. Hacking goes on to use this and similar distinctions for a rather sophisticated discussion; however, we wish to simply borrow this distinction, and to keep it simple so as to not get lost in an intricate discussion of concepts and reality.
Moreover, we would wish to suggest that there are such things-in-the-world as interpersonal influence processes going on in organizational contexts (formal organizations or not). These include actions, interactions, and emotions but of course also assessments and judgements, and of construction of context, agents and agency, and so on. But it is something happening that can be experienced by people. This may later be labeled "leadership" by either the participating humans or the researchers, or both (Alvehus, 2019). The label leadership, on the other hand, and the theories of it, are ideas.
There might be reason to elaborate briefly the epistemological position we take. First of all, we do not wish to position ourselves within a particular school of thought, but rather to present a quite simple conceptual framework that enables us to problematize existing research practices across a range of different epistemological traditions. Put simply, and building on Hacking, we work from the idea of a reality of social actions in the sense that such actions (but not thoughts and individually ascribed meaning) is in principle observable for a third part. We thus afford the realm between people, the realm of social action and interaction, an ontological reality in and of itself (Fairhurst, 2011), not reducible to (but obviously influenced by) individual or collective beliefs and cognitions or to various kinds of social structures. We also work from the assumption that people make sense of and ascribe meaning to social actions and subsequently act on such ascribed meanings. Social action is in that way predicated on the social, cultural, and situated context within which it occurs. Social action is performed by intelligent and reflexive actors, not by inanimate physical objects (Hacking, 1999). This sets our position apart from arguments that leadership only exists as ideas, for example that leadership is an essentially empty category. Such a position is of course also possible, and makes for interesting discussions on the implications of for example leadership discourse on people's identity, or for "studying the ideological character of leadership in language" (Kelly, 2013: 907). However, it also means that it never aims to study leadership as a thing-in-the-world.
On, then, to our second conceptual resource: the notion of blackboxing. When we refer to the blackboxing of leadership, we broadly refer to the way in which researchers avoid specifying the phenomenon under study, even provisionally, whilst at the same time treating this avoidance as unproblematic. More generally, Latour (1987: 2-3) describes a "black box" as a word "used by cyberneticians whenever a piece of machinery or a set of commands is too complex. In its place they draw a little box about which they need to know nothing but its input and output." For our purposes, the central feature of blackboxing is that "many elements [are] made to act as one" (Etzkowitz, 1987: 695), the outcome appearing as a taken for granted fact. A simple example is when a respondent fills in a questionnaire, being asked about an organization's "culture". Difficult questions such as the potential discrepancy in what the respondent and the researcher mean by "culture", what information the respondent has access to, the particular circumstances when filling in the questionnaire, and so on, are unacknowledged, and the respondent's answer is taken to represent "culture" as a reportable fact about the characteristics of an organization. Potentially, any such black box can be opened up, and its taken-for-granted argument unpacked. This would lead to new black boxes being unveiled, as arguments can always burrow furtherthe "expansion process is theoretically infinite" (Jackson, 2008: 438). Yet, for practical purposes, such as research, an explicit or implicit agreement is normally established at some point on which parts to blackbox, in order to avoid infinite regress (Klein, 1999). A black box, once established, is generally accepted within a scientific field (Latour, 1987). Thus, blackboxing "solves" both theoretical and methodological problems by reducing the explanatory burden on the researcher. All research relies on blackboxing; the interesting questions concern what it is that is blackboxed, and what the consequences of blackboxing are. Thus, we do not suggest that blackboxing could be avoided in leadership research. What we instead highlight is how blackboxing leadership tends to solve major problems for leadership researchers, and the consequences of the blackboxing practices in terms of reproducing manager-centrism in leadership research.
In this paper we will highlight two blackboxing practices: blackboxing by delegation and blackboxing by proxying. These labels have originated organically over time as we have read a plethora of empirical leadership studies and started to ask questions around what they actually do study. Of key concern for understanding how these blackboxing practices operate is how authors actually identify leadership in empirical data. In order to show how they work, we will for each practice give three examplesone from a quantitatively oriented study, one from an interpretive qualitative study, and one from a study based on notions of as-practice/collective/plural leadership. The third category is of course of particular interest, as such approaches often explicitly aim at challenging leader-centeredness. Whilst the examples are to some extent arbitrarily chosen, we have aimed at finding well-designed and well-argued paperswe are in no way aiming to criticize the selected examples per se. On the contrary, by their high quality they testify to the acceptance of the practices we identify. Moreover, we can of course not see them as representatives of broad research streams; the diversity of leadership research has been repeatedly shown (see e.g. Gardner et al., 2020;Rost, 1991; for as-practice/collective/plural leadership research in particular, see Denis et al., 2012;Ospina et al., 2020). This diversity is in itself not a problem: Despite different ontological and epistemological positions, the very same methodological practices prevail.

Blackboxing by delegation
Blackboxing by delegation is when researchers leave the definition of central concepts (such as leadership) to informants. The researchers may well formally define the concept themselves but when approaching the empirical material, it is left to the informants to interpret what the concept should mean.
In questionnaire-based studies, this is something of a standard approach. Theories in this tradition are often built on the basis of the idea of someone performing particular behaviors, and the data is collected through questionnaires where respondents are asked to answer a range of questions. Of interest here is that more often than not, those questions include far more than reporting on observations of behaviors. Such questions typically ask respondents to assess and judge the focal person's inner states, intentions, beliefs and feelings (Banks et al., 2021;Larsson et al., 2021). Moreover, and more to the point here, the behaviors in focus are typically described on a relatively abstract level, and the respondent asked to assess whether this happens more or less often (rather than, for instance, reporting precisely how many times during a time interval). Essentially, a wide range of assessments and judgements are left to the respondents.
Such of delegation can for instance be found among studies taking a social network orientation to leadership (Carter et al., 2015). This is the case in the highly influential and innovative study of shared leadership by Carson et al. (2007). Instead of taking formal positions as team managers as a proxy for leadership, the researchers used questionnaires to ask team members to indicate to whom they turned for leadership, then processed the responses to see the distribution of leadership within the teams. In this way, they were able to see how different patterns of leadership configurations developed in different teams. However, what is meant by leadership is left to the respondents to interpret. By explicitly asking the team members who they turned to for leadership (rather than for instance asking for specific behaviors or perceptions or experiences), it was up to the respondents to interpret what leadership might mean to them in this context. This interpretation thus potentially includes a range of unspecified processes, such as perceptions, prototyping, attribution (Martinko et al., 2018), and so on. In other words, these processes are blackboxed and neither explained nor explored. The definition of the central concept is delegated to the survey respondents by simply using the concept in the questions.
However, asking someone else to identify leadership is also found among interview-based and ethnographic studies. In ethnography and ethnography-inspired methods, the view of "the Other" is a key element. A fundamental idea with an ethnographic endeavor is to understand how others make sense activities and situations, and how meaning-making constructs the world they inhabit (Geertz, 1973). Leadership researchers have often taken up such approaches, thus studying leadership through the experiences and vocabularies of the other.
A recent example is Einola and Alvesson (2021). In their study the identification of leadership as a phenomenon depends on whether leadership is a recurring theme in the relationship. In that sense they rely on the informants' own labelling of the phenomenon. They acknowledge that the qualities of the relationship may fluctuate and that the understandings of the relationship may not be shared by all informants. Identification of leadership is thus delegated to the actors. However, the situated and idiosyncratic meanings of these labels are not explored. Hence, the process of labelling as well as the precise meaning of those labels are blackboxed. More specifically, the authors identify a "clear case of leadership" (p. 852) where a team formally appoints a person to a leader role. However, it is not explored what that role entails, that is, what the team members themselves mean by someone being a leader. The study relies on the team members defining leadership by choosing someone and calling this role a leader. The link between the concept of leadership (an idea) and some interpersonal processes (a thing-in-the-world) is blackboxed and left to the informants to handle.
This practice is also found among studies of relational and collective leadership. In an influential article by Cunliffe and Eriksen (2011), the authors describe how they through an abductive process came to a change their understanding of leadership: What we began to notice in our conversations with [Federal Security Directors] was that they did not talk about individualistic heroic action, but their sense of the importance of mundane small details, actions and conversations [...] and importantly, that these were usually seen in relation to others. (Cunliffe andEriksen, 2011: 1428) From this, a key element in their approach to leadership emerged, as they argue for a view of leadership that takes the experience of "being-in-the-world" (p. 1433) as a starting point. They want to draw "attention to the 'mundane' yet revealing intentions, values and judgments that leaders see as crucial to leading in complex situations" (p. 1443). Thus, what they study as leadership is the vocabularies used by a certain type of manager, that is, their ideas as expressed through language. Leadership in this study becomes what these managers "see as crucial", and thus the substance of leadership is actually decided on by these managers, not by the researchers themselves. Again, the precise meanings attached to these descriptions, the precise processes through which they are chosen, as well as the situated functions of choosing them in the conversation at hand (the interview), are left unexplored. In this way, the processes and mechanisms through which the label leadership is connected to a thing-in-the-world is essentially blackboxed.
In summary, blackboxing by delegation encourages studies not of social processes, practices, or relationships, but about respondents' ideas about such processes, practices and relationships.
The object of study thus becomes these ideas, and maybe how those ideas are utilized by the respondents to (discursively) make sense of their world. We are thus in the realm of studying ideas and as such close to studying culture or institutionalized beliefs and logics. This might be very interesting, but our main point here is that it is distinct from trying to explore and observe leadership as a thing-in-the-world. One might argue that social practices, as things-in-the-world, are shaped by the ideas people hold about such practices. However, to explore that relationship, one would need to explore both the ideas and the practicesconflating them defeats the purpose. Exploration of how some thing-in-the-world came to be labeled leadership by someone would reasonably involve a nuanced and detailed analysis of the practice of labelling something as leadership, assuming that the labelling is meaningful in the context where it is produced, and of the thing-in-the-world that is being labelled.

Blackboxing by proxying
Blackboxing by proxying means to study other things, such as results, effects, or signs of leadership, and take them to in some implicit way stand for leadership. This can be seen in both socialpsychological studies and in more qualitatively oriented studies of leadership.
A particularly interesting case is a study of leadership in the gig economy (Fest et al., 2021). In the experiment-based study, the researchers hired online workers to perform text transcription tasksthey de facto enter into a managerial relationship with these tempworkers. They then varied payment systems, motivational messages, and "charismatic leadership communication techniques" (p. 11) to test whether quantity and quality of worker output varied, also testing against demographic variables. (We will not delve into the results here, but we do note that their study suggests monetary rewards increase output whilst praise reduces it.) When studying charismatic leadership communication techniques, they argue: Charismatic leaders use communication tactics, which can be organized in three major categories that can be reliably coded. The first category is "frame and vision", by which the leader tries to draw attention to the key issues of the job. The second category is "substance", which is used to justify the mission and announce strategic goals. "Frame and vision" can be provided by (i) metaphors, (ii) rhetorical questions, (iii) stories and anecdotes, (vi) contrasts, and (v) three-part lists. "Substance" can be induced by (vi) expressing moral conviction, (vii) expressing sentiments of the collective, (viii) setting high and ambitious goals, and (ix) creating the confidence that workers will be able to reach these goals. (Fest et al., 2021: 11) What the researchers focus on, however, is leadership just as a potentially mobilizing force. The task and performance criteria are predetermined by the set-up of the study. No collaboration is possible (hence leadership cannot organize anything). We argue that this is a blackboxing technique where one aspect of leadership (motivating through praise) is taken to stand for the full leadership phenomenon of charismatic leadership, which elsewhere is seen to empower and mobilize to new and challenging tasks. Moreover, leadership is proxied in the sense that short textual, one-way messages are taken to stand for charismatic leadership ("charismatic leadership tactics"). Moreover, the messages are assumed to be perceived and experienced by the subjects in the same way that the researcher classifies them, that is, as giving praise. This substitutes a presumed fixed meaning for the complex phenomenon of social interaction. The blackboxing thus consists of proxying leadership by studying text messages and their effects, leaving unexplored the processes and mechanisms that link the concept of leadership to a thing-in-the-world. This leads the authors to conclude that their study can "advance and inform leadership theory" and that "virtual charismatic leadership is possible in certain environments" (p. 18). The point here is not to discuss the validity and reliability of these results. The point is that the methodological techniques employed enable the authors to discuss their findings in terms of leadership.
A similar case in this regard is Kempster et al.'s (2019) paper on authentic leadership and emotional labor. In the paper, the authors criticize authentic leadership theory for its focus on "relational transparency", arguing that it is "problematic at the level of managers attempting to lead on a daily basis" and even that it is "misplaced and potentially harmful" (Kempster et al., 2019: 332, 333). Instead, they argue for a focus on emotional labor where role identification and fidelity to purpose are central components of authentic leadership. In the paper they approach leadership empirically by one autoethnographic study and one more traditional ethnographic study. Through this, they assert that the experience of authenticity is better captured by acknowledging the way managers achieve role identification and fidelity to purpose. Undoubtedly, they problematize naïve notions of authenticity in the authentic leadership literature. The question is, however: In what sense is this leadership?
While we fully acknowledge that trying to influence people in an emotionally charged situation (as in their study) demands emotional labour, our point is that the link between the emotional labour and the leadership process is left unexplained and unexplored. The mechanisms linking emotional labour to the complex interpersonal phenomenon of leadership is essentially blackboxed. Seemingly, this is authentic leadership rather than authentic something else, because it is a manager doing and experiencing it. Only by proxy, we assert, as they assume that what these managers are doing is leadership by performing the kind of emotional labor suggested in the article. In this sense, it is an example of the well-known tendency in leadership studies to use managers and managerial work as proxies for leadership (Ashford and Sitkin, 2019;Bedian and Hunt, 2006).
In studies focusing on leadership as practice, leadership is often defined as a process and without any predetermined differentiation between leaders and followers. Simpson et al. (2018) make a contribution to this literature as develop the understanding of how leadership emerges in interaction, and also to contribute to the field methodologically. A key element is that they want to understand leadership as a processual phenomenon and as emerging from interpersonal interaction, primarily talk. They analyze a massive amount of data, over 220,000 words of transcript, from a series of meetings with a senior management team. In order to more closely focus on leadership, they look for conversational turns and identify "instances where remembered pasts and anticipated futures were immediately adjacent in the same speech act" (p. 651). Thus, the operational definition of leadership in this study is a specific kind of speech act and its "performative effects" (p. 651). The different performative effects are then mapped over time, resulting in identification of different phases in the management team's process. In the final analysis, these movements are seen to "signal the presence of leadership" (p. 656).
Leadership is thus studied in terms of signals, identified as conversational turns. These turns are clearly found in the interaction, and as such true to the aim of studying leadership as a processual phenomenon, rather than individual characteristics. What is less clear, though, is if these conversational turns in themselves are outcomes of leadership processes (since they are signals) or if they are a constitutive part of those processes. This relationship between the conversational turns (identified through an impressive analysis) and the complex phenomenon of leadership is left implicit. Moreover, the analysis stops at the identification of these turns, leaving the situated production of them unexplored (beyond identification of turns in accordance with the definition). Conversational turns, as signals of leadership, are seemingly taken as proxies for leadership as a complex interpersonal processa process which thereby is black-boxed.
In summary and returning to Hacking's (1999) distinction, what we find here are studies of a variety of things-in-the-world. These include motivational texts, emotions, and shifts in conversations. These observable things-in-the-world are then taken to stand for the idea of leadership, that is, for a complex phenomenon that at its core includes social interaction, experiences, assessments, and ascription of meaning. We argue that practices such as these abound and many of them are likely taken for granted. We acknowledge that of course things-in-the-world can be labeled differently and that theoretical concepts may be overlapping. Yet, that is not quite the point in this case. What is interesting is how concepts such as motivation, emotions and conversational turns, turn into leadership. These have been studied for a long time but predominantly not related leadership. How do they become identified as leadership in leadership studies? Seemingly, because they appear in studies of managers. We suggest that one important consequence of these blackboxing practices is a prevailing and perhaps increasing manager-centrism.

The prevailing focus on managers
Through the two practices of delegation and proxying, studies accomplish researching leadership while failing to specify how what is studied is understood to be leadership, that is, the link between what is studied and the complex phenomenon of leadership (as a thing-in-the-world) is left implicit. In our first three cases, it was left to respondents and informants to use the label of leadership to select and identify the focal phenomenon, and the precise ways this label of category was used, was unexplored. In the last three cases, specified phenomena were explored (effects of reading short texts, experiences of emotional labour, and conversational turns) and used as proxies for leadership, leaving the link between these and the complex phenomenon of leadership only implied.
How then, is the claim that the research is about leadership made credible and convincing? Studies using delegation relies on recruiting the research subjects as fellow ethnographers (Mol, 2002). The research subjects are expected to be able to use the label of leadership in a way that makes sense to them and that resonates with what the researchers have intended. A basic problem here, however, is that there is no reason to believe that the research subjects use the label "leadership" in the same sense as it is used in the researcher's theoretical framework. Indeed, the distinction between lay concepts and theoretical concepts has a long history within sociology (Denzin, 2017). Moreover, there might be a range of reasons for a research subject to use (or not use) the label leadership that is distinct from the research questions. For instance, the label leadership has been shown to hold very strong identity implications (Ford and Harding, 2007) and that it even can take on the quality of a fantasy, as such out of contact with observable reality (Sveningsson and Larsson, 2006). More basically, the label is embedded in the research subject's discursive and cultural context and is given meaning within this, rather than in the researcher's theoretical framework. The studies we discussed above were all conducted in contexts where this label was already utilized, that is, they studied managers and their colleagues in formal organizations. We would expect that individuals in such contexts tend to engage with the popular leadership discourse, for example communicated through leadership development programs. Using the label then reflects the meaning it is given within this discourse (without making that meaning explicit) but there is no reason to assume this is identical to the meaning of the label within the theoretical framework.
If we want the research participants to be well versed in leadership discourse, it is a very good idea to study managers or their colleagues, who also get exposed to such discourse. It makes less sense to study a context where the label "leadership" is more foreign to peoplefor instance, to ask choir singers about whom they turn to for leadership, or ask children on a playground about who their leader is. As a consequence, utilizing delegation as a blackboxing strategy would reasonably lead to a preference for studying precisely managers in formal organizations.
At times, particularly among researchers adhering to some variant of social constructionism, the argument is made that we have to rely on what people themselves call leadership, otherwise we as researchers impose our worldview on the studied context. However, Silverman (1989: 218) warned researchers: "Avoid treating the actor's point of view as an explanation" (for the case of leadership, see also Alvehus and Crevani, 2022). The problem can be illustrated by another theoretical term: power. Imagine that we would only study power through the vocabularies of respondents. If so, subtle forms of power (such as ideology, disciplinary power, subjectification) would likely seldom appear in studies as such forms of power operate just by not being recognized by actors. Thus, definition by delegation inhibits theoretical development and, even worse, it leaves definitions implicit.
Also studies using proxying mainly rely on studying managers or manager-subordinate relationships. In the studies we have discussed text messages, conversational turns, and experiences of emotional labor seem to be understood as (aspects of) leadership since they are found in managers or in manager-follower relationships. We can understand this as a metonymic relationship, where a part stands for the whole, or a halo-effect, where the assumed leadership of the manager so to speak latches on to, or skews the perception of, everything around it. One can wonder how the chosen phenomena would be labelled, had they occurred on other context (the choir, the playground, in an emerging social movement, and so on). Moreover, is there a difference between at appreciative text message sent by a HR department, sent by a colleague, and by a supervising manager? Following Latour (1987), we would assume that the assemblage of, say, colleague + text message constitutes something different from the assemblage supervising manager + text message. If the latter is taken to be "leadership", then we have precisely defined leadership as something supervising managers do, that is, we are tied to a strong manager-centrism.
In summary, we see a strong dominance of manager-centrism in the empirical studies in this journal as well as in other journals devoted to leadership studies. We suggest that this is in part a consequence of the methodological blackboxing of the phenomenon of leadership, accomplished through delegation and proxying. Leadership is treated as something that is not empirically specified and that seemingly need not be specified by the researcher. We are witnessing a "missing what" (Rawls, 2008: 709): We are missing the attention to the central phenomenon, however this is defined. Yet more troublesome is the fact that blackboxing makes it very difficult to locate the central phenomenon, leadershipa problem that is "solved" by turning to studying managers. Hence we perhaps, in particular in cases where leadership is sought to be decentered, will paradoxically find increased manager-centrism.

Conclusion
Thirty years ago, Rost (1991) famously argued that a century of leadership studies essentially had focused on phenomena other than the central interpersonal influence relationship, even though most definitions agree on this being at the heart of the concept "leadership". We argue that this still remains true: that leadership studies are plagued by a focus on "peripheral aspects" and that "the essential nature of what leadership is" continues to evade leadership scholars (Rost, 1991: 4). Despite the seeming consensus that at the center of leadership lies a social and interpersonal process of influence, the vast majority of research engages with one of the individuals engaged in such processes. Managers become a focal point for studying leadership, also in studies trying to avoid just this. This paper is driven by the thought that there is indeed some things-in-the-world that can be fruitfully understood in terms of leadership. We could of course argue that the concept of leadership is an empty container (Kelly, 2013) and come to the conclusion that it should not be studied at all, or that it can only be studied as an idealeadership as ideology, discourse, meaning, or similar. This is all in order, but it is not the underlying gist of our argument. Rather, our point is that if leadership scholars want to study leadership as a thing-in-the-world, they need to develop new epistemological and methodological practices. Such practices would need to open up the black boxes we have discussed above. This means engaging with critical discussion of how and why a certain thing-inthe-world is leadership, and not taking this for granted. It might involve a reflexive discussion about how a label such as leadership comes to be employed, what it is associated with, and exploring in more detail the role of elements otherwise taken as proxies in the complex assemblage that we tend to theorize as leadership. This may appear controversial, and so may our claims; but as we aim to provoke discussion, we also note that controversies are what opens up black boxes (Latour, 1987).
Exactly how new leadership research practices would look like, we are not sure; we also acknowledge that they would of course create new black boxes that would need further discussion. The key idea with this paper is to open up such conversations. To move leadership research forward, we need perhaps not develop more intricate ontologies on the one hand, or search for more distinct and operationalizable definitions on the other. Instead, leadership scholars need to be aware of the methodological practices they engage with, develop new such practices, and avoid routinely blackboxing leadership. Perhaps then can leadership scholars move from studying managers to actually studying leadership.