Abstract
In the ongoing debate in the area of critical leadership studies, the nature of leader–follower relationships is a thorny issue. The nature of followership has been questioned, especially whether followers can display resistance behaviours while maintaining their follower position. Addressing this issue requires a dialectical approach in which followers and leaders alike are primary elements in leadership co-production. Followers who face destructive leaders are of special interest when leadership is studied as a co-creational process. This context favours the emergence of a full range of behavioural profiles in which passives and colluders will illustrate the destructive leadership co-production process, and those who resist demonstrate that followers may not follow the leader and still keep a followership purpose. A two-step data analysis procedure was conducted based on the behaviour descriptions of 123 followers having a destructive leader. A qualitative analysis (i.e. content analysis) showed a set of behaviours and their antecedents that suggest three main groups of followers: resisters, obedient and mixed behaviour. Treating these data quantitatively (i.e. latent class analysis), six followers’ profiles emerged: active resistance, passive resistance, passive obedience, conflict avoidance, support and mixed. Our findings provide evidence that followers who resist may do it for the sake of the organisation. We discuss our findings in light of followership theory, whereby joining role-based and constructionist approaches allows us to argue that followers may still be followers even when they do not invariably follow their leader.
Keywords
Introduction
‘In the context of over a century of studying leadership, we are still at the beginning of followership research. There is a great deal of work to be done’ (Riggio, 2020).
Recent research highlights conceptual issues pertaining to the studies of leadership calling for conceptual and empirical clarifications on the nature of the leadership–followership relationship (Blom and Lundgren, 2020; Collinson, 2017). This call arises in a context in which these and other authors (e.g. Learmonth and Morrell, 2017) advance different positions regarding the voluntary nature of followership.
The debate mirrors the demanding evolution of the followership approach, in which several concepts such as leader, leadership, follower and followership coexist with distinct meanings and assumptions attached to them. Accordingly, some reviews (Crossman and Crossman, 2011; Shamir, 2004; Uhl-Bien et al., 2014) demonstrate that this evolution has departed from the traditional leadership studies (where the focus was placed on leaders) to a real followership approach.
Followership involves the study of ‘the nature and impact of followers and following in the leadership process’ and whether this can be achieved by a role-based view (rank based) or by a constructionist approach (social influence process) (Uhl-Bien et al., 2014: 89). The constructionist approach conforms to the debate in critical leadership studies (CLS) on the nature of leadership–followership (e.g. Collinson and Tourish, 2015). Accordingly, and following a post-structuralist perspective, Collinson (2006, 2008) argues that leadership holds a power-based and knowledge nature that allows a dynamic relationship(s) amongst followers and leaders to be co-produced, and in which followers can develop and deploy different identities when playing their role in the leadership co-production: conformist selves may hide different motivations but perform according to the leader’s requests; resistance may emerge as an oppositional behaviour to control practises and can be designed as a way to build an alternative. However, resistance may entail harmful consequences (e.g. Latan et al., 2019) that compel followers to disguise their behaviours. Underlying this identity is a dramaturgical self (Collinson, 2006) in which a manipulative nature allows followers to alternate between conformist and oppositional forms.
From our standpoint, clarifying these conceptual issues is critical, as labelling is more than lettering. Langer and Benevento (1978) show some classic experiments in which participants change their performance levels by being labelled as workers or bosses. This shows how inducing power differences through labelling produces considerable behavioural and psychological effects (Hoption et al., 2012; Vanderslice, 1988). Conceptual clarification thus has a practical impact. Additionally, once we accept leadership is a co-produced process, we are holding both leader and followers accountable. A third motive to uphold the clarification concerns its possible conceptual redundancy because if (a) followership is ultimately the role followers play in the leadership process (e.g. Uhl-Bien et al., 2014), and (b) leadership co-production includes both downward and upward tactics (e.g. Collinson, 2006), then (c) followership is intrinsic to leadership (Rost, 1993, 2008). In line with this, followership is considered of such critical importance that it is not possible to conceive of leadership without it (Uhl-Bien and Carsten, 2018).
This study departs from previous research defining leadership as a dynamic and bidirectional influential process in which the leader and followers both play their role(s) based on asymmetrical power conditions (e.g. Collinson, 2006; Collinson and Tourish, 2015). In such a complex and dynamic process, a follower may decide not to follow the leader. This occurs, for example when the follower’s upwards influence is intended to display resistance (e.g. Tepper et al., 2001). Learmonth and Morrell (2017: 265) asked, ‘In what sense can a person intelligibly remain a follower while simultaneously displaying dissent and resistance’? – stressing the apparent contradiction in terms. In other words, does not following the leader imply not being a follower?
To answer this question, one should depart from ‘why people follow?’. Recent work (Bastardoz and Van Vugt, 2019) shows followership as an adaptive group process, where followers wilfully decide to follow an individual they acknowledge as leader. Conceiving the follower role as an individual decision stresses its voluntary nature (e.g. Alvesson and Blom, 2015). However, the organisational formal hierarchical structure may challenge this free will assumption (Blom and Lundgren, 2020). This is not to say that voluntariness is not important when studying followership in organisations, it means that it needs to be carefully analysed.
Accordingly, empirical research has been called to contribute to the debate on the followership nature (Blom and Lundgren, 2020; Collinson, 2017). We agree that when a leader displays destructive or unethical behaviours (e.g. Pelletier et al., 2019; Thoroughgood et al., 2018), it is especially important to answer this call. We conceive this as a critical context to analyse these issues because it threatens the basic assumption that leaders and followers work together for the organisational good. Knowing that followership comprehends behaviours displayed in interaction with a leader with the ultimate purpose of meeting organisational goals (Carsten et al., 2014), can someone decide not to follow a destructive leader so he or she can follow the common organisational goals? Departing from Uhl-Bien et al.’s (2014) differentiation between the followership role-based and the constructionist views, we will answer this question throughout this work. From a role-based view, subordinates displaying resistance cease their follower role and engage in a manager–subordinate relation. However, the constructionist view allows these subordinates to display leading behaviours (e.g. resistance) and remain followers of the organisational good.
Enabling followers to take part in the leadership process gives them both the opportunity and the responsibility to think critically (Kelley, 1988) and behave courageously (Chaleff, 1995) to check the leader. Although it is well accepted that destructive leadership is a dynamic process that depends not only on the leader’s behaviours but also on followers and context (e.g. Kellerman, 2016; Padilla et al., 2007), followership typologies do not provide profiles of resistant followers (Lipman-Blumen, 2005; Schyns et al., 2018; Thoroughgood et al., 2012). Starting with Collinson’s (2006, 2008) followership identities, and adopting an exploratory approach, this study contributes to the conceptual clarification involving leadership and followership by empirically identifying a full range of followers’ behavioural profiles when facing leaders that display destructive behaviours.
In the first section we present the literature review to explore the nature of the leader–follower relationship. Destructive leadership is discussed as a co-produced and dynamic process. With the premise that followers are active agents in any leadership process, our focus moves to followers themselves and how they have been classified in the literature. This section finishes with followership typologies that have been conceptualised for the specific context of destructive leadership.
In the method section, we provide information about the study’s participants, procedure and data analysis strategy, detailing the mixed nature (qualitative and quantitative) of this research. The results section thereafter describes the main findings of this study, which are then examined in the discussion section. Before presenting the main conclusions, we acknowledge some limitations and offer suggestions for future research.
The nature of the leader–follower relationship
Early leadership research overvalued the leader’s role (Uhl-Bien et al., 2014). This is evident in leader-centric approaches in which followers are considered mere recipients (Uhl-Bien and Carsten, 2018), and both success and failure are of the sole responsibility of leaders (Oc and Bashshur, 2013). Later, follower-centric perspectives on leadership brought followers into the equation. However, these approaches mainly focus on leaders, and how followers’ perceptions help to construct leaders and to identify effective or ineffective leaders (Crossman and Crossman, 2011). Other perspectives, such as the relational- and the role-based views, emphasise the wilful and active role of followers in the relationship with their leaders. The relational view emphasises that leaders and followers together achieve leadership goals (Uhl-Bien et al., 2014). The role-based perspective is well illustrated with behavioural typologies (e.g. Chaleff, 1995; Kelley, 1988). The evolution of these perspectives paved the way for the study of followership as ‘a relational role in which followers have the ability to influence leaders and contribute to the improvement and attainment of group and organisational objectives. It is primarily a hierarchically upwards influence’ (Crossman and Crossman, 2011: 484). However, it was not until the emergence of the constructionist followership perspective that leaders and followers came to be considered as co-producers of leadership (for an historical review on the topic see Uhl-Bien et al., 2014). From this approach, the object of study is the interaction between leaders’ behaviours and followers’ reactions in producing the leadership process.
The co-production of leadership exposes the critical role followers may play in a partnership (with the leader) aimed to benefit the organisation (e.g. Carsten et al., 2018). Although working together, leaders and followers take different positions in this power-imbalanced process (e.g. Collinson, 2008, 2018). Accordingly, to advance knowledge on followership/leadership, one should first understand why people follow, as evidenced by Lapierre and Carsten (2014) book’s title Followership: What is it and why do people follow?. This question is of primary interest since being a follower is a decision, that is a voluntary action (Lapierre, 2014), where there is no room neither for fear based nor for blind obedience (Blom and Lundgren, 2020).
Why people follow is rooted in its evolutionary and adaptive nature that aims to guarantee the group’s protection: ‘humans have a flexible followership psychology that enables them to select and follow the right kind of leaders under the right conditions, determine an appropriate engagement level and switch from being a follower to a leader whenever appropriate’ (Bastardoz and Van Vugt, 2019: 81). Therefore, followership is not confounded with a rigid hierarchy or formal position (Alvesson and Blom, 2015), as followers are those who decide who is the leader, based on the recognition of the ability to achieve the common good. Followers monitor the leader’s behaviours continuously. Together, these assertions highlight the attributional nature of leadership, where leaders are not just appointed but rather acknowledged as leaders (Hinrichs and Hinrichs, 2014).
All in all, the voluntary decision to follow the leader departs from the assumption that both the leader and the group follow an idea of a common good. Up to this point, we can compare the leadership process to a followership chain where all the actors (leader and followers) are aligned to follow organisational goals. In order to make the process efficient, different roles emerge: the followers follow the leader while the leader is following the common goals. In such a situation those ‘followers’ who decide to resist to a leaders’ legitimate influence are ‘non-followers’, and, consequently, may harm the collective good (Carsten et al., 2014).
It is then timely to ask what happens when the leader is not perceived as legitimate, for example because he or she no longer strives for common goals? When this happens, a fiduciary contract is breached, and previous assumptions are necessarily challenged. This is of interest within the organisational context where followers, although recognised as active and protective agents (Carsten and Uhl-Bien, 2013) cannot replace a bad leader (Bastardoz and Van Vugt, 2019). For all these reasons, followership needs to be explored when basic assumptions are disrupted, answering the call of Blom and Lundgren (2020) to extend knowledge on the voluntary nature of followership.
As mentioned, when followership is conceived as a chain, the ultimate purpose is to achieve the organisational good. This is in line with Carsten et al.’s (2014) claim that followership is all about how followers relate with leaders with the aim of meeting organisational goals. Thus, when perceiving that the leader is blocking that flow, followers can voluntarily decide to challenge the leader or to resist (Carsten et al., 2010; Carsten and Uhl-Bien, 2013). When they make such a decision, they keep their formal role (i.e. subordinate) but are no longer that leader’s followers. At this point, three things may happen: (a) the follower conforms (Thoroughgood et al., 2012), (b) the follower decides to leave (Schmid et al., 2018) and (c) the follower resists and challenges the leader’s attempts (Tepper et al., 2001). There is no voluntary will in following the leader in any of these situations, ceasing the subordinate’s follower role (Blom and Lundgren, 2020). However, minding the chain idea, those subordinates who voluntarily decide to challenge the leader to seek for an alternative (Collinson, 2006) will remain followers, that is organisational followers. Thus, studying followership within this specific context allows us to propose two layers of analysis: (1) a dyadic level (leader/manager and follower/subordinate relationship) related to a role-based approach; (2) a systemic level (organisational goals and follower/subordinate relationship) that recalls a constructive view. By dissecting these levels of analysis, we can conceive resistance as not following the leader (i.e. being a subordinate in a manager–subordinate relation) while following the purpose of a common good (i.e. being an organisational follower) and reiterating their role in leadership co-production (Carsten et al., 2010).
From what has been described, it becomes clear that the explanation of the conceptual nature of the leadership–followership nature is of critical importance not only for theory but also for practical purposes, for example attributing responsibility when a leader displays destructive or unethical behaviours (e.g. Pelletier et al., 2019; Thoroughgood et al., 2018). In power-imbalanced situations, implicitly assumed in the leader–follower relationship (Blom and Lundgren, 2020; Magee et al., 2005), those who are powerless may displace their responsibility and commit crimes of obedience – ‘I did what I have been told to do’ (Beu and Buckley, 2004; Carsten and Uhl-Bien, 2013); or, to gain an advantage, may use strategies such as ingratiation (De Clercq et al., 2019; May et al., 2014). Total responsibility for destructive leadership can be wrongfully placed on the leader from a leader-centric perspective. However, when one takes a co-production perspective, it becomes clear that leadership is not an exclusive leader-dependent-process (Collinson, 2005), but one in which followers take part and can resist and/or challenge the leader. Co-production of leadership implies a co-responsibility (Uhl-Bien et al., 2014).
Destructive leader(ship)
Echoing the co-production thesis, some researchers (e.g. Kellerman, 2016; Thoroughgood et al., 2018) approach destructive leadership as a dynamic process that depends not only on the leader’s behaviours but also on followers and context. Thus, the presence of a leader who displays destructive behaviours is not the same as destructive leadership. Destructive leadership emerges when susceptible followers enable those destructive behaviours (Padilla et al., 2007). As Kellerman (2016) clearly states, ‘Leadership – It is a system, Not a Person’ as it requires a leader, follower and a context in which to occur. Accordingly, she presents a typology for bad leadership and not for bad leaders (Kellerman, 2005).
Given the exploratory nature of this study, we decided to follow a well-established and comprehensive definition of destructive leaders’ behaviours. Moreover, as we place our focus on how followers deal with perceived leaders’ destructive behaviours, the definition we choose emphasises this behavioural approach: ‘The systematic and repeated behaviour by a leader, supervisor or manager that violates the legitimate interest of the organisation by undermining and/or sabotaging the organisation’s goals, tasks, resources and effectiveness and/or the motivation, well-being or job satisfaction of subordinates’ (Einarsen et al., 2007: 208).
This definition is compatible with the idea that a destructive leader is not the same as a destructive leadership. The latter holds a dialectical nature that is expressed as a ‘dynamic, co-creational process between leaders, followers and environments, the product of which contributes to group and organisational outcomes’ (Thoroughgood et al., 2018: 627). This dynamic approach draws our attention to the role that followers play within this process (e.g. Kellerman, 2004; Lipman-Blumen, 2005).
Followers as active agents in the leadership process
The CLS dialectical perspective on leadership–followership leverages followers’ potential on leadership co-production, arguing that they represent both knowledgeable and proactive agents (Collinson, 2005). An important consequence of this theoretical proposal comes from Collinson’s studies (e.g. 2006) from which three followers’ identities emerged: conformist (display disciplined behaviours), resistant (show their opposition and try to construct alternatives) or dramaturgical (manipulate their behaviour based on self-consciousness and knowledge).
Followers’ classifications.
Differently from the others, Collinson’s (2006) proposal is founded on a constructionist view. This is meaningful as from a post-structuralist perspective, context plays a critical role (Uhl-Bien et al., 2014). Accordingly, processes are not stagnant but fluid, and in permanent construction, depending on workplace dynamics (Collinson, 2008). For this reason, it is important to analyse how (and which) followers’ profiles emerge under the context of a leader who displays destructive behaviours.
To our knowledge, none of the presented followers’ classifications were developed through the specific lens of destructive leadership. However, some of their authors draw attention to the key role that followers may (should) play: ‘Instead of viewing followers as the “good soldiers” who carry out commands dutifully, we need to consider followers as the primary defenders against toxic leaders or dysfunctional organisations’ (Kelley, 2008: 14). In line, Chaleff (1995) argues that followers should show the courage to challenge the leader. Carsten et al. (2010) show that followers view themselves as being able to challenge the leader. This behaviour, in opposition to an obedience role, requires that followers possess a belief in co-production (Carsten and Uhl-Bien, 2013; Carsten et al., 2018). Collinson (2006) highlights the importance of followers’ resistance identity when leaders fail to perform in the best interest of the organisation, contrasting it with the conformity identity’s negative impact. Kellerman (2004, 2016) stresses this negative impact, arguing that passive followers play an active role in bad leadership by allowing it.
Destructive leader(ship) followers
Followers’ classifications within the destructive leadership context.
Thoroughgood et al. (2012) argue that followers can be conformers or colluders. Conformers tolerate destructive behaviours less actively and can be classified into three types: lost souls (emotionally dependent), bystanders (driven by pragmatism) and authoritarians (act under hierarchical expectations). Colluders have a more active role, and it is possible to find followers that are acolytes (share the leader’s values) and those who are opportunists and aim to be the next leader. In a similar vein, Schyns et al. (2018) propose five types of followers with different kinds of involvement in destructive leadership. Victims strengthen the leader’s destructiveness due to their fear. Bystanders allow destructiveness as they simply observe what is happening. Conformers follow the leader’s orders to avoid punishment. Colluders align with the leader for personal gains and they may share the leader’s values or be opportunistic. Lastly, puppet masters are those who push leaders to destructive behaviours.
Typologies partially overlap (anxious followers, lost souls and victims; pragmatic, bystanders and conformers; the leader’s entourage, acolytes and colluders; malevolent and puppet master) and comprise only followers that in one way or another enable destructive behaviours. As these typologies pertain to the identification of the destructive leadership process (Pelletier et al., 2019), they do not comprehend followers that could help to stop the destructive leadership phenomenon. However, it is known that even under oppressive conditions followers resist (e.g. Thomas, 2003). We contend that this is an important research gap on which we support the goal to provide an integrative and empirically based typology.
Even though followers’ classifications overlook a full-range behaviour model, some empirical studies have been analysing followers’ specific behaviours. Studies focussing on followers’ passivity towards destructive leaders mostly concern crimes of obedience (e.g. Beu and Buckley, 2004; Carsten and Uhl-Bien, 2013; Gibson et al., 2018) and silence (e.g. Kiewitz et al., 2016; Lam and Xu, 2019; Xu et al., 2015). Studies focussing on followers’ challenging behaviours include topics such as insubordination (e.g. Mackey et al., 2019), resistance (e.g. Carsten and Uhl-Bien, 2013; Greenbaum et al., 2013; Tepper et al., 2006) and whistle-blowing (e.g. Latan et al., 2019; Near and Miceli, 1985; Reckers-Sauciuc and Lowe, 2010). More broadly, some studies have drawn attention to followers’ coping strategies (May et al., 2014; Nandkeolyar et al., 2014; Yagil et al., 2011) within the context of destructive leaders’ behaviours. These behaviours range from passivity to active challenge. Although in different ways, all these behaviours express followers’ involvement in the leadership process (Kellerman, 2016). However, sometimes, when facing leaders with destructive behaviours, followers decide to withdraw and resign their formal and constructionist participation in the leadership process, by quitting the organisation or intending to do so (Tepper, 2000). When followers are not aligned with the leader and opt to stay, a possible reason is that they will do it for the sake of the organisation, or they still believe things can change. Accordingly, research shows that turnover intention and organisational commitment are strongly associated (Haque et al., 2019; Vandenberghe and Tremblay, 2008). Intention to stay in the organisation can be taken as an indication of organisational pride, but people can also stay due to a lack of alternatives (e.g. Hom et al., 2012). Reluctant stayers were recently found to show more dysfunctional behaviours such as workplace deviance (Singh, 2019). Overall, these findings suggest that followers who simultaneously display resistant behaviours and experience low levels of turnover intentions probably match ‘resistant identity’ and thus are striving to build an alternative (Collinson, 2006).
Summarising, we reason that followers’ resistance towards destructive leaders’ behaviours can be examined in light of a followership approach (Uhl-Bien et al., 2014). This can be done by joining both role-based and constructionist views. In a role-based approach, the relationship is founded in formal hierarchical positions, and subordinates may not follow the manager (Blom and Lundgren, 2020). A constructionist approach comprehends a dynamic nature where power position shifts are allowed and a follower may take the lead (Bastardoz and Van Vugt, 2019). Accordingly, opposing behaviours might be taken as a temporary leading position where a subordinate and a follower role can co-occur at different layers: when one decides not to follow a destructive leader they will remain a role-based follower (i.e. a subordinate) that, in a constructionist approach, is taking a leading position to pursue organisational goals (i.e. a follower). For this reason, we contend that yes, one is still a follower even when one does not follow the destructive leader.
Method
Participants and procedure
Participants were recruited through snowball sampling using Qualtrics online survey software distribution link. Snowball sampling is an approach that is especially useful for exploratory purposes and reaching hidden populations (Babbie, 2014) and has been chosen in empirical studies targeting destructive leadership (e.g. Schmid et al., 2018; Shaw et al., 2011). Given the specific requirements of the sample, we opted to use this technique.
With an extensive online survey and assuming an exploratory approach (open-ended questions), we provided an adaptation of Einarsen et al.’s (2007) definition of destructive leader(ship), and asked participants whether they have had or are currently dealing with such a leader. Of 145 respondents, 96 (66.2%) declared having had or currently having a destructive leader. Among those with such experience, 51 participants provided enough data to conduct content analysis. As we were interested in broad and full-range behaviours, we asked participants to describe not only their personal experience but also to offer the example of behaviours shown by up to two co-workers when dealing with a destructive leader. We also asked respondents to provide their behavioural antecedents (e.g. motivations) and some demographic variables.
Given that not all participants answered the three conditions, we were able to gather only 123 descriptions of how followers deal/dealt with the destructive leader, their behavioural antecedents and some demographic variables. We were also able to get descriptions of leaders’ behaviours. The mean age of the followers is 38.1 years old (SD = 11.4), with 70.7% of women (87 cases) and 29.3% of men (36 cases).
Data analysis strategy
To identify how followers behave when dealing with destructive leaders, we opted for a two-step data analysis: qualitative content analysis followed by a latent class analysis (LCA). Content analysis enabled us to categorise followers’ behaviours (and antecedents). Then, we defined a matrix where all behaviours (and antecedents) were classified as 2 (when present) and 1 (when absent). After analysing this matrix, we isolated three main groups (those who perform any type of obedient behaviour, those who try to challenge the leader and followers that perform both). At that point, we computed LCA in order to explore the presence of ‘hidden’ profiles within each group. Figure 1 summarises the overall process. Data analysis strategy.
Data analysis strategy: Qualitative content analysis
The data analysis process had a mixed nature combining both deductive and inductive approaches (Patton, 2015). We started our analysis classifying leaders’ destructive behaviours according to previous literature (e.g. abusive, exploitative and laissez-faire). For this purpose, we considered two main ideas: (1) Kellerman’s (2005) continuum of destructive leader’s behaviours ranging from ineffective to unethical behaviours and (2) Einarsen et al.’s (2007) model of displaying destructive behaviours against the followers and/or against the organisation. Crossing levels of severity (Kellerman, 2005) with two possible targets (Einarsen et al., 2007), we labelled categories according to well-established classifications that match each of the described behaviours. For example when someone described a leader displaying humiliating behaviours (severe behaviour) against followers (target), we classified it as abusive; a leader showing ineffective behaviours (less severe) due to lack of involvement (organisation directed) was labelled as laissez-faire.
Regarding followers’ behaviours and antecedents’ categories, we developed a conceptual matrix crossing existing followers’ classifications (e.g. Chaleff, 1995; Collinson, 2006; Kelley, 1988) looking for behavioural patterns. We extracted a priori categories such as challenge, support, critical thinking and hierarchical expectations. While developing the conceptual matrix, we were faced with equivalent concepts, that is conceptual overlapping. The nature of those concepts may have different roots, but in the end, they translate into similar behaviours. For example Zaleznik (1965) observes that some followers show dominance as an expression of the wish to control authority. In the specific context of facing a destructive leader, all these behavioural profiles express a challenging attitude. In a similar vein, when not aligned with the leader, Chaleff’s (1995) ‘partner followers’, Kellerman’s (2008) ‘activists’ and ‘diehards’ and Collinson’s (2006) ‘resistance selves’ will display challenging behaviours. Each time encountered, we considered these as overlapping dimensions and merged them into a single category.
To shed light on possible motivations for resistance behaviour, and to enrich the debate on why resistant followers opt to remain or withdraw, we defined a priori, a category pertaining to intention to quit. Whenever we identified content that did not fall into one of our a priori categories, we created a new one. We used MAXQDA 2018 (Kuckartz and Rädiker, 2019) to organise data and code sentence segments.
The content analysis procedure followed Bauer’s (2000) recommendations regarding the categorisation procedure (each a priori category has a fixed number of codes; codes are independent), coherence (a priori codes were derived from a conceptual matrix), transparency (presentation of the codes list, its frequency and illustration quote) and reliability measured through inter-rater agreement (a second independent coder coded 20% of the qualitative data, and a Fleiss’ Kappa of 0.936 was obtained). The coding units were words, sentences or paragraphs, each variable was coded only once, and our unit of analysis was each follower’s description (Boyatzis, 1998).
Data analysis strategy: LCA
After coding followers’ behaviours and antecedents, LCA was applied to find followers’ profiles according to behaviour patterns and their antecedents. LCA is a widely used technique in a number of research fields (e.g. Birmingham et al., 2020; Liu and Vivolo-Kantor, 2020; Parr et al., 2016) that enables multivariate categorical data testing and computes mixed models based on a probabilistic rationale (Hagenaars and McCutcheon, 2002), estimating the extent of each class in the population, and the probability of someone to belong to a specific class, which is useful to assign individuals in a categorisation system. LCA is an exploratory technique that is more flexible than other clustering procedures and enables the creation of latent variables (classes) form observed data (Oberski, 2016). Thus, this procedure does not use predetermined cut points and provides statistical fit indices that help to retain the best model. To increase data analysis sensitivity, we opted to analyse at the subcategory level, when existing. Moreover, only categories with a frequency of greater than three were considered for convergence’s sake.
We used the R software’s poLCA package and followed Linzer and Lewis’ (2011) recommendations to compute LCA. Firstly, we tested a basic latent class model using the behaviour variables. Then, we tested latent class regression modelling with covariates (behaviour antecedents or attributes) that we were expecting to operate as predictors of class membership. The number of classes (i.e. followers’ profiles) was judged based on Bayesian information criterion (BIC) and Akaike information criterion (AIC) model comparison fit indices, where the lowest values show the best model. Whenever contradictory information was found, we have considered two recommended decision criteria: BIC is recommended for larger samples (Dziak et al., 2020), and model selection should also be supported on its interpretability (Finch and French, 2015; Oberski, 2016; Parr et al., 2016).
Results
Following the frame presented in Figure 1, this section starts by showing content analysis findings comprising the leader’s destructive behaviours; followers’ three major behavioural groups; and their attributes. Then, LCA results are showed to identify the number of final classes and which attributes (behaviour’s antecedents) are linked to which profile.
Content analysis
Leaders’ behaviours content analysis results.
Followers’ behaviours content analysis results.
Note. Only frequencies identified with a star (*) were considered in the latent class analysis. These cases fulfil the requirement of having been mentioned three or more times and refer to the lowest categories’ level.
Obedience behaviours followed the same classification system and are presented as active (those who decide to conform or act in conformity) and passive (those who enable by doing nothing). From active obedience, the most frequent categories are support (behaviours that back the leader’s actions) and subservience (behaviours that manifest active obedience). The passive obedience dimension comprehends passivity (being passive to the leader’s actions) and to avoid conflicts (behaviours aimed to address consensus and avoid conflicts).
The other groups of behaviours comprise horizontal behaviours (to protect or to oppose co-workers); actions that show levels of inconsistency such as sharing disagreement when talking with others but behaving passively with the leader and pretending different behaviours with different people, and demonstrating the will to leave the organisation.
Although they are not very frequently cited, we were able to find nine types of attributes reported three or more times. In the resistance group (N = 48), the most frequent categories were strength (feeling confident and empowered; ‘(…) strong personality, confident and fearless’), high critical thinking (being able to question and think clearly; ‘(…) didn’t recognise leader’s competency to manage the tasks the leader was supposed to do’) and high competence (being perceived as competent; ‘(…) perform better than the leader’).
In the obedience group (N = 53), the most frequent categories were fragility (low self-esteem and self-confidence; ‘(…) more fragile/sensitive and with low self-esteem’), fear (displaying fear motivated reactions; ‘(…) because of feeling afraid’), ambition (ambition, greed and opportunism; ‘(…) most likely because of the willingness to stay within a company and not losing the opportunity for a career growth’) and low critical thinking. In the mixed behaviour group, the most prevailing categories were strength, ambition and pragmatism (being guided by behaviours’ consequences; ‘(…) the need for monthly remuneration and being the only person in his household who contributed for the family’s income’). So, explanations of the observed behaviours are mostly of a dispositional, attitudinal and emotional nature.
LCA
Fit indices for behaviour model decision.
Note. Boldface type shows the chosen model. BIC: Bayesian information criterion; AIC: Akaike information criterion
Fit indices for regression model decision.
Note. As these are regression models to identify predictors for different classes, we have not computed models comprising a single class. Boldface type shows the chosen model. BIC: Bayesian information criterion; AIC: Akaike information criterion.
Manifest variables probabilities for the two-class resistance group model.
Note. Boldface type shows the highest probabilities in each class.
Latent class regression results (resistance group).
Note. Boldface type shows significant covariates.
Manifest variables’ probabilities for the three-class obedience group model.
Note. Boldface type shows the highest probabilities in each class.
Latent class regression results (obedience group).
Note. Boldface type shows significant covariates.
Characteristics such as fragility, fear and low levels of critical thinking do not predict whether or not a follower belongs to conflict avoiders or passives groups. Fragility is least likely to be witnessed in the supporters group, which is also the least fearful. Ambitious followers are more likely to be found within supporters, closely followed by avoiders, and lastly by passives. The same is observed regarding hierarchical expectations, which are more likely to be observed in followers belonging to the passives, and least likely in those belonging to the conflict avoiders’ group. Low critical thinking is a feature found equally in passives and avoiders but not in the supporters’ group, which has the lowest likelihood of including such followers.
Correspondingly, the passive group most likely includes fragile, fearful and unambitious followers who have the lowest critical thinking and the highest levels of hierarchical expectations of all groups. Conflict avoiders stand out for being the group with the lowest hierarchical expectations, resembling passives in their fragility, fearfulness, low levels of critical thinking and moderate levels of ambition. Supporters have the highest number of ambitious followers, co-occurring with the lowest fragility, fear, low critical thinking and having intermediate levels of hierarchical expectations.
Discussion
The main purpose of this study is to contribute to the debate on the nature of followership. More specifically, we focus on the question ‘are followers who do not follow still followers?’ This issue emerges from a conceptual debate in which authors call for empirical contributions (Blom and Lundgren, 2020; Collinson, 2017; Learmonth and Morrell, 2017). Departing from a destructive leader-based context that challenges basic and evolutionary assumptions (Bastardoz and Van Vugt, 2019), we propose a two-layer approach based on Uhl-Bien et al.’s (2014) role-based and constructionist approaches. We argue that an integrative interpretation of these perspectives allows us to reason that, when attempting to resist to a destructive leader, subordinates (as defined by their position) may display leading position behaviours (from a transitory power shift) aimed to achieve the organisational good, and thus, remain as organisational followers while not following the leader.
Leadership studies have long treated followers as mere recipients in a leader–follower relationship, placing attention on leaders’ characteristics and styles (e.g. trait and behaviour leadership approaches) (Oc and Bashshur, 2013). This resulted in negative associations with the role of followers, who were considered passive agents (Agho, 2009; Chaleff, 2008). Overlooking followers’ agency may have led to a conceptual confusion between the terms of leader and leadership. This theoretical misperception gained orderliness through the emergence of the followership field of studies, which has been gradually growing and mobilising researchers’ interest in followers as important players in the leadership process (e.g. Carsten et al., 2010; Chaleff, 1995; Kellerman, 2008).
We believe that studying followers in the specific context of facing a dysfunctional leader will help to clarify the followership nature. This specific context allows us to analyse both destructive leadership co-production and resistance mechanisms, enabling the development of a full-range behavioural typology. Moreover, as we will detail throughout this section, the emergence of resistance profiles provides evidence that ‘not following’ may be compatible with followership.
Accordingly, we looked for participants who have experienced a destructive leader(ship) situation and asked them about their leader’s behaviours. Findings revealed two main categories in line with Kellerman’s (2005) unethical and ineffective dimensions. Within the unethical dimension, followers’ descriptions emphasised follower- and organisation-directed behaviours. The former includes abusive (Tepper, 2000, Tepper et al., 2017) and exploitative behaviours (Schmid et al., 2019). The latter was scarcely mentioned, which may reflect its less harmful effects on followers (Schmid et al., 2018) or possible gains for followers (Einarsen et al., 2007). Concerning ineffective behaviours, the emergence of both incompetence and laissez-faire categories provides evidence of its destructive nature (Aasland et al., 2010; Schilling, 2009; Skogstad et al., 2007). Ultimately, the main features of participants’ descriptions match those proposed in the literature, which offers encouragement that our findings pertain to these types of leader.
The literature review suggests that when facing a destructive leader, followers may display diverse behaviours (e.g. obedience, ingratiation and resistance). However, as mentioned, an integrative model is lacking. Our results mirror this complex behavioural portfolio showing individuals are complex entities where an all-or-nothing approach is illusory. Accordingly, we found different profiles within both the obedience and resistance groups. Within each of the subgroups (e.g. passive resisters, conflict avoiders, etc.), behavioural consistency is expected (Sherman et al., 2010), meaning that we can anticipate followers to behave consistently when interacting with a destructive leader. However, followers’ agency and contextual constraints interact and allow a shifting nature in followers’ profiles (Collinson, 2006) which is evidenced by the mixed group eclectic profile.
Essentially, findings regarding followers’ behavioural profiles allow a more comprehensive interpretation of leadership’s/followership’s dynamic nature, in which one needs first to distinguish and then to integrate the role-based and constructionist perspectives. In light of this dialectic, we discuss our findings for the three main behavioural groups: obedience, resistance and mixed.
Obedience: When followers co-produce destructiveness
Through the naked eye, the obedience group recalls a traditional conformity role, resulting from a corrective nature of power (Collinson, 2018). Milgram’s (1963) experiment is a classic illustration of this, in which obedience has a hierarchical nature. However, one must notice that compliance may disguise other motives. According to Fromm (1977), and his ‘Fear of Freedom’, obedience may derive from the security and comfort of simply leaving decisions to someone else. Shamir (2004) argues that conformity behaviours may camouflage calculative motivations. Regardless of the motivation, obedience behaviours reinforce the status quo (Kellerman, 2012).
Our findings shed light on how destructive leadership is co-produced in this group. The three-class model covers three different obedience profiles (passive followers, conflict avoiders and supporters) expressing distinct processes that enable destructive leadership. Passive followers’ behaviours are mainly characterised by showing total inaction regarding the leader’s behaviours. Their most distinctive characteristic is the presence of hierarchical expectations that seem to emerge as justification for the follower’s behaviour based on the idea of authority (e.g. Gibson et al., 2018; Milgram, 1963). These followers share some characteristics with individuals classified as conflict avoiders: fragility, fear and low levels of critical thinking. These features, especially fear and fragility, have been highlighted by Lipman-Blumen (2005) and Thoroughgood et al. (2012) in describing how followers enable leaders’ destructive behaviour. Low levels of critical thinking are also related to less effective followership styles (Kelley, 1988).
In line with theoretical predictions on the topic, findings show a group of followers that support the destructive leader’s behaviour (Lipman-Blumen, 2005; Schyns et al., 2018; Thoroughgood et al., 2012). Alongside the supportive behaviours, ingratiation and subservience behaviours are also observed in this class. Accordingly, Yagil et al. (2011) showed that ingratiation behaviours are a strategy that increases in the presence of a destructive leader. This group’s most distinctive characteristic is ambition, which has been discussed within followership literature as a characteristic of followers that enables destructive leadership, also related to greed and opportunism (Lipman-Blumen, 2005). The existence of different profiles of obedient followers suggests that obedience is a multi-meaning concept that ranges from pure conformism to calculative behaviours, as proposed by Collinson (2008) and Shamir (2004). Despite its diverse manifestations, we reason that obedience can be taken as consent in which individuals enable the destructive process. In other words, these actors are taking part and co-producing the situation, and as a consequence, are co-responsible for its destructiveness. Although all three obedience classes are co-responsible for the destructive leadership process, the motives underlying the profiles suggest different levels of alignment. This idea was proposed by Blom and Lundgren (2020) who discussed the involuntary nature of followership, extending previous research on the idea that followership (in opposition to the superior–subordinate relationship) implies a ‘voluntary reduction of influence’ (Alvesson and Blom, 2015: 272). Accordingly, it is reasonable to question whether the passive and conflict avoiders can be conceived as voluntary followers. Once their behaviour is rooted in motivations such as hierarchical expectation, and fear, it seems fair to consider them as subordinates, who fail the follower role at both the dyadic and the systemic layer. Conversely, those who support the leader’s actions are leader’s followers. However, they fail to meet the systemic follower role when we consider the organisational good. That being said, we also argue, in line with Collinson (2008), that regardless of being leader–follower or superior–subordinate, the critical nature of these relations resides in power dynamics. Moreover, from our standpoint, the pivotal issue regarding followers’ willingness to follow Blom and Lundgren (2020) is linked to followers’ motives, which deserve further attention and research. We reason that this more in-depth inspection of followers’ motivations to obey helps organisations to engage in practises and develop policies that supply or feed followers’ needs and/or concerns (e.g. fear of losing job). If an organisation adopts policies whereby employees’ participation is valued (e.g. providing feedback systems), we can expect that obedience elicited by hierarchical expectations will be reduced.
Regardless of the voluntariness of their behaviours, obedient followers display compliance behaviours and follow their leader. This willingness to follow shows how obedient followers and destructive leaders co-produce destructive leadership.
Resistance: When followers decide not to follow the leader
As mentioned, when facing destructive leaders, followers may display resistance behaviours (e.g. Tepper et al., 2001). When followers resist, they are by definition not following the leader. Not following the leader means not taking part in the destructive leadership process. With this idea in mind, it is appropriate to ask whether these followers are, indeed, followers. Recalling our two-layer approach in articulation with Uhl-Bien et al.’s (2014) views of followership – role-based and constructionist – we can interpret resistance as a manifestation of followership. Accordingly, these individuals are role-based subordinates in a dyadic level where they resist their manager’s attempts. At the same time, from the systemic layer, the constructionist perspective allows individuals to display resistance as a power-based behaviour where followers do follow the organisational good. Moreover, the idea of scrutinising the leader, and act to benefit the collective, restores the evolutionary and adaptive role of followership (Bastardoz and Van Vugt, 2019) and is in line with works that highlight the active and courageous nature of good followers (e.g. Chaleff, 1995; Kelley, 1988).
There are both conceptual arguments in the literature as well as empirical grounds reported in this study that support the contention that followers may remain followers when they decide not to follow a destructive leader. When followers are not aligned and do not want to comply, they can opt to resist or leave. When leaving the organisation is not possible (Hom et al., 2012; Sheridan et al., 2019), followers are known to display deviant behaviours (Singh, 2019). There are two main forms or workplace deviance: organisational and interpersonal (Bennett and Robinson, 2000). Interpersonal deviance has been reconceptualised and it is now well established that it involves behaviours targeted against co-workers and supervisors (Hershcovis et al., 2007; Mitchell and Ambrose, 2007). Our findings show vertical deviance through verbal aggressiveness (‘(…) talk back to the supervisor’) and horizontal deviance when followers describe behaviours against co-workers (‘(…) spreading negativity through other co-workers’). The existence of resisters who simultaneously show low intentions to leave and low levels of deviant behaviours may suggest their pro-organisational motivation to resist a destructive leader. Under these circumstances, resisters may not follow the destructive leader precisely in order to not co-produce destructive leadership and thus remain faithful to and follow their ideal of ‘good’ leadership. We agree that resistance may guide the goal of re-establishing the desired condition or building an alternative (Collinson, 2006), sending a message of disapproval (Tepper et al., 2001).
Our findings support this contention. Accordingly, it is informative that amongst the emerging categories there is a low frequency of intentions/behaviours of leaving the organisation. The probability of finding intentions to abandon the organisation was nil in the passive resistance class and rose to only 15% in those that fall into the active resistance class. Likewise, the probability of finding individuals who express verbal aggressiveness is only 5% in the passive resistance class and 7% in the active resistance class. Regarding behaviours against co-workers, the probability within passive resisters is only 14% and nil in the active resistance class. These probabilities make deviant behaviour very unlikely in the sample, which is also concomitant with a low intention to leave.
As mentioned, our findings showed two resistance profiles: those who play an active challenging role and those that opt to take a more covert resistance strategy. This partially goes in line with Tepper et al.’s (2001) proposal of constructive resistance and passive dysfunctional resistance behaviours. The active resistance group shows a defiant profile in which followers dare to challenge the leader (e.g. Chaleff, 1995). With respect to the passive resistance class, the two main behaviours are ignoring the leader and focussing on the task. Although it is possible to find followers who actively display resistance, this is not the core behavioural profile of this group. We take passive resistance as operating at an underground level, which is in line with previous proposals of camouflage dynamics and cover behaviours linked to resistance (Collinson, 2005; Rodrigues and Collinson, 1995). We contend that passive resistance may be an alternative path of defiance preferred by those who do not possess enough strength to engage in a straightforward confrontation. Findings from the content analysis showed that strength is related to feelings of confidence and empowerment, which is linked to a perceived sense of power (Anderson et al., 2012). This perception is known to relate to voicing (Kim et al., 2019) and silence (Morrison et al., 2015) behaviours. Moreover, the exhibition of behaviours against the co-workers, although with low occurrence, is aligned with the idea that followers may displace aggression when they cannot carry it out against the transgressor (Mitchell and Ambrose, 2012), suggesting the lack of power to oppose the leader.
The mixed group: When followers play both sides
This group is the most ambiguous and needs further research. However, it may echo Collinson’s (2006: 185) dramaturgical selves who can turn into ‘skilled manipulators of self and information’ when facing more authoritarian contexts. Thus, these followers will become tacticians and typically display both resistance and conformist behaviours. Our findings detect followers who exhibit this mixed profile, supported by antecedents such as ambition and pragmatism. The complex nature of these individuals may justify its relative scarcer occurrence in the sample. Most of the descriptions within this profile combine passive obedience behaviours (e.g. inaction and avoiding conflicts) with active resistance (e.g. to challenge) or passive resistance (e.g. task orientation and ignoring the leader). It is unclear if followers deliberately engage in different behaviours at different moments, in a manipulative and calculated manner, or if these behaviours simply co-occur. However, this mixed nature can be an expression of disguised resistance (Collinson, 2018).
Limitations and future research
This study presents some limitations that may guide future research. This work is based on the well-established assumption that followership assumes working towards organisational goals (Carsten et al., 2014). However, this may not match all possibilities. Accordingly, future research may challenge this assumption and ascertain its limits, namely, how strong is this motivation when contrasted with selfish motivations when voluntarily following or resisting.
A larger sample may have been preferable, but one should note that data were collected from open-ended questions, and previous or present experience with destructive leaders was required. This may explain the relatively low-frequency levels for attributes. Despite this, results show some influential features such as strength or ambition.
Working simultaneously with self-reports and observer ratings can be a limitation since the cognitive processes may bias data. However, there is an indication in the literature that observer ratings can be as precise as self-reports in such latent constructs as personality traits (Oh et al., 2011), and conducting our analyses with and without self-report data showed the same number of classes for the main behavioural groups.
This study reports specific types of destructive leaders’ behaviours. However, we did not segment the followers’ analysis accordingly, due to insufficient frequency. Considering Schmid et al.’s (2018) findings, future research may consider targeting each of the subsamples to test whether our results depend on a specific leader’s action(s). Moreover, as destructive leadership is a dynamic process (e.g. Kellerman, 2005; Padilla et al., 2007), it will be important to study how contextual factors influence followers’ behaviours. This issue seems particularly important in order to gain knowledge of followers belonging to the mixed behaviour group.
Our results show that resistant followers scarcely mention any intention to quit, displaying challenging strategies. It will be important for future research on this topic to collect data regarding followers’ organisational commitment (Allen and Meyer, 1990) to assess whether followers’ low rates of turnover intentions are related to organisational loyalty (normative or affective commitment) or lack of alternatives (continuance commitment). However, one should notice that although commitment is negatively related to turnover intentions (Meyer et al., 2002), in the presence of abusive supervision, normative commitment is expected to be linked to voicing behaviours, and continuance commitment to be related to low levels of voice (Gabler and Hill, 2015). It is, therefore, reasonable to expect that the resistance group shows low intentions to leave due to ethical and moral values, unlike passive followers (from the obedience group) who simply have no alternatives.
A more detailed and in-depth analysis of followership and the shifting nature of followers’ roles (resulting from the interaction between agency and contextual constraints) require an extension of our work, specifically targeting the mixed group. Such a group’s emergence provides evidence that followers are not static agents and may engage in different and sometimes contradictory behaviours. As mentioned, the nature of the study design precludes a deep understanding of this group and calls for qualitative and longitudinal research specifically designed to comprehend all its complexity. Although behavioural consistency is likely (e.g. Sherman et al., 2010) and favours group intra-homogeneity, we expect that longitudinal designs and non-structured qualitative methods (e.g. interview) will provide empirical robustness to this issue. To close this section, we must highlight that despite the need for further research, the mere emergence of a mixed group in our study allows us to gain evidence on the non-static nature of being a follower.
Conclusion
The nature of leadership–followership remains immersed in an intense debate (Blom and Lundgren, 2020), especially in the area of CLS (Collinson, 2014, 2017; Learmonth and Morrell, 2017; Tomkins et al., 2020). In line with a dialectical approach to leadership–followership dynamics, in which workplace conditions and power imbalances play an essential role (Collinson, 2008, 2018), we studied followers’ profiles when they face destructive leaders.
This study provides two major contributions. Firstly, our results contribute to a conceptual clarification regarding the nature of followership, by showing why and how resistance should be considered in a comprehensive followers’ behaviour model. Moreover, this study helps to consolidate the post-structuralist perspective, in which followers (although or because (Foucault, 1977, 1979) of being powerless participants) play an active and crucial role in leadership dynamics. Secondly, to achieve our main purpose, we developed an integrative empirical-based model of six behavioural profiles that emerge in the presence of a destructive leader: active resistant followers, passive resistant followers, passives, conflict avoiders, supporters and mixed behaviour followers. Although followers volition and agency have been widely considered (e.g. Chaleff, 1995; Kellerman, 2008; Kelley, 2008) as the most important factors in uncovering followers’ types, within the destructive leadership context, they are still theoretically based and favour followers that enable destructive leadership (Lipman-Blumen, 2005; Schyns et al., 2018; Thoroughgood et al., 2012). Our results largely confirm such profiles but highlight a group of individuals who opt not to follow the leader and display resistance seemingly for pro-social reasons. Accordingly, the low levels of intention to quit within the resistance group suggest that these followers are active producers of the leadership process.
Moreover, joining role-based and constructionist perspectives in a two-layer approach allows us to uphold the idea that followers may still be followers even if they do not follow the destructive leader. This claim is not inconsequential, as we have detailed in the introduction section, because labelling is not lettering. Accordingly, it is important to close this work with two critical assertions: (1) regardless of the voluntary nature of followers’ (subordinates’) obedience, following a destructive leader (supervisor) helps to develop destructive leadership (organisational destructiveness); (2) followers who decide not to follow the leader are, nevertheless, organisational followers and play a critical role in the non-destructive leadership process.
Footnotes
Acknowledgements
The authors would like to thank the editor, Dennis Tourish, and two anonymous reviewers for their constructive and insightful comments and suggestions during the revision process.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was supported by a Research Grant from the Portuguese Foundation for Science and Technology (FCT) awarded to Teresa Almeida (Ref. SFRH/BD/132041/2017).
