1
Abstract
Modern organizations increasingly focus on self-managing teams, giving high levels of discretion and autonomy to its members. The current study investigates whether these groups are indeed hierarchy-free, or whether informal hierarchy inevitably rises up against pressures to maintain equality. In an experimental setting, the presence versus absence of a formal hierarchy was manipulated in student groups. Results confirmed the hypothesis that groups develop more linear and steep informal hierarchies when a formal hierarchy is absent. The experience of relationship conflict appeared to explain this development, suggesting that social dynamics play an important role in why groups form influence hierarchies. As the presence of a strong formal group hierarchy decreases these social dynamics, an informal hierarchy is no longer functional.
Keywords: formal hierarchy, informal hierarchy, influence, social dynamics, conflict,
linearity, steepness
2 When and Why Groups Develop Informal Influence Hierarchies.
Organizations have long enforced strong hierarchical structures on work groups, characterized by enduring formalized distributions of work units and work positions (James & Jones, 1976).
Formal hierarchical structures are generally based on clear-cut patterns of leadership, strict task assignments and centralized decision making processes (Jones, 2007). Research has demonstrated that formal hierarchical structures can carry decisive functional benefits for groups because they tend to facilitate group coordination, minimize intra-group conflict and increase group member motivation (Anderson & Brown, 2010; Magee & Galinsky, 2008).
Nevertheless, work groups that employ strict formal hierarchy also suffer from the fact that this hierarchy is difficult and costly to change, potentially creating group inertia and oppressing creative dissent (Magee & Galinsky, 2008). These negative effects contradict the aim of many organizations today to constitute of 'knowledge workers' and 'creative entrepreneurs' (Diefenbach & Sillince, 2011; De Dreu & West, 2001). For this reason, organizations are increasingly giving work groups full autonomy and discretion in how to perform their tasks (Diefenbach & Sillince, 2011). In self-managing teams, group members manage, assign and schedule their work themselves (Kirkman & Rosen, 1999).
Although the organizational trend to reduce formal group hierarchies has been well recognized, little empirical work exists on the effects of this trend on the self-management strategies that these groups employ (Kirkman & Rosen, 1999). The present research aims to investigate this question. I propose an intriguing paradox: In groups where formal hierarchy is weakened, informal hierarchical structure is likely to take its place, with clear-cut (albeit informal) influence differentials and role distributions between group members (cf.
Diefenbach & Sillince, 2011). This means that in an effort to reduce formal group hierarchies, organizations may actually enhance a group's informal hierarchy – promoting (informal) vertical differences between members despite an “explicit distaste for hierarchy” (Gruenfeld
& Tiedens, 2010, p. 1263).
This research contributes to prior research on the development of hierarchies in work
groups. So far, no studies have empirically tested the relationship between formal group
hierarchies and the development of informal group hierarchies. It is my aim to demonstrate
that the creation of a hierarchy-free organization may be impossible. Even with explicit
organizational goals of empowerment, participation, flexibility and equality, informal group
hierarchies may inevitably rise up against these pressures. Furthermore, responding to recent
3
calls in the literature (e.g. Magee & Galinsky, 2008), I will take a group-level approach. This enables me to incorporate the effects of social dynamics (i.e. interactive processes) as an explanation why hierarchy develops.
In the sections below, I will first discuss the benefits of a formally imposed group hierarchy for groups before introducing the idea that groups can also develop an informal group hierarchy through social dynamic processes. I subsequently formulate my central hypothesis for the two main features that define an informal group hierarchy: linearity and steepness. My hypotheses are tested in an experimental study where the presence and absence of a formal hierarchical structure was manipulated. Hierarchical linearity and steepness were measured by using dyadic dominance scores of the members in the experimental groups. The linearity and steepness of an informal hierarchy are expected to be lower when a formal hierarchy is present compared to when it is absent.
Theoretical Background Formal group hierarchies
Formal group hierarchies are installed to control and integrate group activities (Child, 1972). The formal structuring of groups typically consists of two core elements: horizontal and vertical divisions of labour (Bunderson & Boumgarden, 2010). The formal horizontal division of labour determines how tasks are allocated and divided among group members. For example, work groups can differ in their degree of task interdependence. Low task interdependence requires less coordination of work activities and less cooperation among group members than high task interdependence (Ronay, Greenaway, Anicich & Galinsky, 2012).
The current research focuses on the second formal element; the formal vertical division of labour (i.e. hierarchy) within groups, typically installed through the presence (or absence) of a formal group leader with authority and power (Bunderson & Boumgarden, 2010; Kirkman &
Rosen, 1999; Shelly & Troyer, 2001). According to Anderson & Kilduff (2009), influence allocation to a formal leader is a primary source of hierarchical group differentiation within groups. A formal leader has the capacity to influence the behaviours of other group members and can impose certain structures on groups (Galinsky, Gruenfeld & Magee, 2003;
Thompson, 1961). Accordingly, he or she has the ability to change actions of others and, as a
result, has influence on group processes and group outcomes (Kirkman & Rosen, 1999
Anderson, Spataro & Flynn, 2008).
4
Organizations appoint formal leaders to work groups so that they can create a match between organizational goals and group activities, hereby maximizing group efficiency and group performance (Child, 1972; James & Jones, 1976). Empirical evidence indeed supports the idea that installing a strong formal hierarchy through the presence of a leader has functional benefits (Bunderson & Boumgarden, 2010). For example, by providing role- relevant information and resources to group members, a formal leader can decrease group conflict (House, 1971). Moreover, it is generally believed that organizations only grant formal leadership positions to those who are highly competent and motivated (Magee & Galinsky, 2008). As such, most group members tend to perceive a formal group hierarchy to be legitimate and accept the presence of a formal leader.
Informal group hierarchies
Hierarchies exist not only formally in groups; they can also exist informally (Anderson &
Kilduff, 2009; Schmid Mast, 2002; Van Vugt, 2006). An informal group hierarchy reflects the person-dependent social relationships of dominance and subordination that naturally emerge within a group and become persistent over time through repeated social interactions (Diefenbach & Sillince, 2011). The few group members who become highly ranked within an informal group hierarchy are allocated more influence over group processes and group outcomes compared to those lower ranked.
Previous research clearly demonstrates why informal rank differences tend to emerge between individuals in groups, even when group roles are undefined at initial stages of group interaction (Bales, 1950; Leavitt, 2005; Mazur, 1973; Van Vugt, 2006). The inevitable nature of this development suggests that group hierarchies may have an evolutionary function.
Ethological literature on animal groups confirms that informal group hierarchies increase chances of group survival. For instance, in animal groups, informal hierarchical differentiation can optimize the coordination of group activities (e.g. hunting; Chase, 1974) and has also been related to resource allocation (e.g. food; Hemelrijk, 2002). Functionalist theories on the existence of informal hierarchies in human groups likewise posit that informal hierarchies can be equally performance enhancing as formal group hierarchies (Magee &
Galinsky, 2008). In addition, it has been argued that an informal group hierarchy can also fulfil psychological needs, such as establishing social relationships, developing a social identity and acquiring status and recognition from others (Gruenfeld & Tiedens, 2010).
Two conceptual frameworks attempt to explain how informal hierarchies can develop
naturally within groups. The 'correlational' or 'prior attributes' framework (Chase, 1974),
5
which has long been the primary explanation for informal status differentiation in human groups (Chase, 1980; Magee & Galinsky, 2008), focuses on individual characteristics of group members as a critical determinant of the formation process. Traits like gender, age, race (Berger, Rosenholtz & Zelditch, 1980), dominance (Judge, Bono, Ilies & Gerhardt, 2002) and extraversion (Anderson, John, Keltner & Kring, 2001) correlate highly with an individual's rank in a group. The rationale behind personal traits as determinants of one’s group rank is based on the functional perspective on informal group hierarchies, stating that such traits carry expectations of competence in all kinds of group situations. For example, males are stereotypically believed to be better leaders, and will be granted more influence, compared to women. This way, personal traits can indeed shape group members' perceptions about each other (Berger et al., 1980).
Research, however, does not fully support the ‘prior attributes’ framework. Informal hierarchies have been established in groups where members are relatively homogeneous to one another. For example, there have been instances where two same aged male group members with equal amounts of trait dominance and extraversion still yielded different influence levels within their group, and this development could not be explained by chance alone (see Landau, 1951). This finding goes against the central assumption of the ‘individual attributes’ framework that groups consisting of members who are approximately similar to each other with respect to their individual characteristics should develop highly egalitarian informal structures.
The ‘ethologistic’ framework provides an alternative mechanism through which informal
hierarchies can emerge in groups; through within-group social dynamics (Chase, 1980; Chase,
Tovey, Spangler-Martin & Manfredonia, 2002). This framework proposes that informal
hierarchies tend to develop on the basis of interactions between group members rather than on
the traits of individual group members. The presence of social dynamics has been clearly
illustrated by Chase et al.'s (2002) study of hierarchy development in groups of fish. The fish
(Metriaclima zebra) were placed together in tanks repeatedly, where they interacted to
establish an informal hierarchy. In between these episodes, the fish were put in isolation long
enough to forget one another. It was found that groups consisting of the exact same fish
formed different hierarchies the first and the second time they interacted (i.e. the fish
occupied different positions in the rank order). When hierarchy development would be
dependent on prior attributes only, the same hierarchy would have developed after every
6
interaction round. The study thus clearly shows the importance of social dynamics that take place during interaction.
In the ethological literature on animal groups, where the social dynamics model has mainly been developed, social dynamics are described as conflicts between animals of the same group expressed through agonistic encounters and physical fights (Flack & de Waal, 2004). Conflict characteristics, such as conflict intensity and conflict frequency, have been used as important explanatory mechanisms for why informal group hierarchies arise (Flack &
de Waal, 2004; Thierry, 1985). Although in human groups conflict is more subtle and difficult to observe, it has been suggested to play an equally important role in establishing influence rank orders, and thus, informal group hierarchies (Bendersky & Hays, 2012; Chase, 1980;
Magee & Galinsky, 2008).
Conflict in human groups arises from perceived incompatibilities or differences of opinion between group members (Greer, Jehn & Mannix, 2008). It is commonly divided into three basic types: relationship, task and process conflict. Relationship conflict refers to the incompatibilities among group members that are not task-related, such as tension with respect to personal norms, values, preferences & interpersonal style (Jehn & Mannix, 2001). Task conflict is the disagreement among group members about task related issues, such as how tasks should be performed or what kind of strategies should be applied (Jehn, Greer, Levine &
Szulanski, 2008). Lastly, process conflict is the controversy about logistical issues, thus dividing roles and responsibilities for accomplishing tasks (Jehn et al., 2008).
Hypothesis Building
When organizations impose a formal hierarchy upon groups, the formal leader is responsible for indicating which members perform which tasks, when they need to perform them and with whom (Bunderson & Boumgarden, 2010; Janicik & Bartell, 2003). Through these actions, the formal leader imposes the hierarchy on the groups, minimizing all three types of conflict.
First, influence contests are often experienced as relationship conflict by the actors
involved (Gould, 2003). A formal group hierarchy with established work relations often
makes influence and respect issues official and unambiguous (Thompson, 1961). Therefore,
group members do not need to resolve ambiguity by engaging in such conflict. Second, as a
formal group hierarchy is also designed to enhance group effectiveness and efficiency (Jones,
2007), it makes it unlikely for group members to engage in process conflict; work activities
are already designed and scheduled. Finally, in formal group hierarchies, those with highest
7
rank generally receive more support from the other members, preventing task conflict to really develop prominently. Formally lower ranked individuals will simply evaluate standpoints from the formally higher ranked more positively than vice versa (Bales, 1950; Berger et al., 1980).
Notably, when one of the three types of conflicts does develop within groups with a formal leader, the leader can mitigate such conflict using his/her authority, ultimately deciding how the issues are resolved (Thompson, 1961). Moreover, he or she can officially rectify aggressive behaviour of subordinates (Strijbos, Jochems & Broers, 2004). As a result, conflicts are likely to remain small and non-disruptive.
When organizations do not impose a formal hierarchy upon groups but create self- managing teams, group members are expected to solve relationship, process and task issues themselves. I therefore argue that all three types of conflict are more likely to occur in this situation.
First, with relations and positions of all individuals in the group being undefined, group members will engage in conflict to 'divide' their levels of influence. For example, in order to earn a high ranking position and gain more influence than others, group members tend to show off their competence (Burke, 1968). It has been found that as long as rank attainments continue, group functioning is impaired because 'too many cooks in the kitchen', or too many individuals trying to assert high influence, increases relationship conflict. (Kilduff &
Anderson, 2009; Loch, Huberman & Stout, 2000). Moreover, ambiguity about who does what leaves room for group members to argue about how to solve logistical issues, creating process conflict. Finally, when different opinions exist on task content, lack of deference makes it disputable whose opinion is most important (Berger et al., 1980), triggering task conflict. A final reason why conflict will occur in groups without a formal hierarchy is that a lack of behavioural regulation makes it simply more likely that group members attempt to cross boundaries of their role and are less easily corrected for it (McLean Parks, Ma & Gallagher, 2010). To conclude, research suggests that without any guidance of a formal hierarchy, groups tend to 'fight' until they have figured out their own informal hierarchy. My first hypothesis is therefore;
Hypothesis 1: In groups without a formal hierarchy, all three types of conflict will be higher compared to groups with a formal hierarchy.
Constructs describing informal hierarchy
8
An informal group hierarchy is generally captured by two basic constructs that reflect the main types of social influence structures that can develop within a group; linearity and steepness.
In a perfectly linear group hierarchy, member A influences all other individuals in the group; member B influences all other individuals except for person A; person C influences all other individuals except for person A and B; and so on. Perfect hierarchical linearity exists when all triads in a group are transitive. This is the case when in each triad, consisting of three individuals, individual A influences B, B influences C and when A also influences C.
There is overall deviation from hierarchical linearity within the group when there are intransitive triads. The more intransitive triads there are, the more likely it is that one can speak of a non-linear, group hierarchy. Such an non-linear hierarchy is characterized by equal influence or undecided relationships among group members. Note however that studies on linearity in animal groups (Chase, 1980; Chase et al., 2002; Flack & de Waal, 2004) and human groups (Berdahl & Anderson, 2005; Strayer & Strayer 1976; Strayer & Trudel, 1984;
Schmid-Mast, 2002) have found informal hierarchical structures with remarkably high degrees of linearity.
Figure 1: Transitivity in informal hierarchy
Conflicts (either relationship, process or task) between group members offer the chance to
discover a winner and a loser of that conflict, the winner being the person who is able to push
his/her opinion or convince the other; the loser being the person who is unable to do this. The
more conflict exists within a group, the more chances there are to discover 'winners' and
'losers'. Because the winners of conflict are granted more influence in groups, they end up
higher in the rank order compared to the losers of conflict. Accordingly, the more conflict
there initially exists within a group, the more chance there is that group members will
eventually differentiate from each other in influence rank. Positions within the hierarchy will
9
be negotiated just as long as status disagreement between respective team members exist (e.g.
no clear winner or loser is present; Kilduff & Anderson, 2009; Loch et al., 2000). Therefore, my second hypothesis is;
Hypothesis 2: I propose that in groups where formal hierarchy is absent, increased conflict (all three types) leads to higher linearity compared to groups where formal hierarchy is present and conflict (all three types) is low.
The second construct that informs us about the shape of the informal group hierarchy is hierarchical steepness. Hierarchy steepness refers to the extent to which relative rank within the group is dispersed (Christie & Barling, 2010). Specifically, it is measured by the size of absolute differences between adjacently ranked individuals in their influence levels (e.g. won dominance encounters, influence ratings; De Vries, Stevens & Vervaecke, 2006). When influence differences between adjacently ranked group members are large, the informal hierarchy is identified as relatively steep; but when only small influence differences exist, the informal hierarchy is identified as relatively shallow.
Figure 2: Steepness of informal hierarchy
I expect conflict to increase steepness of informal hierarchy, because group members over
time will get to know each other, and will remember each others' past behaviour. Thus, the
increased number of conflicts (and thus conflict outcomes) will add up towards an overall
judgment of someone's influence. Consider scenario 1, where one conflict has emerged
between member A and B, leading to one conflict outcome for that dyad in the group. When
A wins the conflict, member A will be perceived by him-/herself and the other team members
as being slightly more influential compared to member B. However, in scenario 2, there have
been ten conflicts between member A and B. Group member A is identified as the winner of
all these conflicts. Now, the perceived influence difference between member A and B will be
higher. The added wins and losses for person A and B respectively will cause influence
10
differences to increase, thus leading to higher steepness of the informal hierarchy. Therefore, my third hypothesis is;
Hypothesis 3: In groups where formal hierarchy is absent, increased conflict (all thee types) will lead to increased steepness of the hierarchy compared to groups where formal hierarchy is present and conflict (all three types) is low.
Hypotheses 1, 2 and 3 are summarized in Figure 3.
Figure 3: Conceptual model
Supplementary Prediction: Perceptions of influence
Theory suggests that simply abolishing formal hierarchical structure may not be sufficient to achieve anticipated group benefits, like individual autonomy, creativity and dissent. I therefore proposed that groups without formal hierarchy will inevitably develop informal hierarchy, to overcome problems of coordination and conflict. Why then, do so many organizations still believe in their empowerment initiatives? Is it possible that informal group hierarchies fulfil the same needs as formal hierarchies but lack their negative consequences?
Among other things, formal hierarchy has been criticized widely for its encouragement of conformity and group think, decreasing creativity (James & Jones, 1986). However, it is yet unknown whether informal group hierarchies have the same restrictive nature as formal group hierarchies, or whether they allow group members to speak up and disagree with each other (i.e., grant higher levels of influence to all members).
On the one hand, there is research among groups of children (Strayer & Trudel, 1984) and
adolescents (Savin-Williams, 1979) suggesting that behaviour of group members is
significantly limited once an informal hierarchy has developed. Individuals behave in line
with their informal rank, suggesting that informal hierarchy limits actors to share opinion and
11
dissent. In research on adult groups, it has been found that highly ranked individuals have more speaking time (Brown & Levinson, 1987) and task input (Fisek & Ofsche, 1970) compared to low ranked individuals (Brown & Levinson, 1987) and they more easily express true attitudes (Anderson & Berdahl, 2002). These findings suggest informal hierarchies to be restrictive especially to those who end up lower in the influence rankings.
Yet on the other hand, ethological research has found that animal groups that have been through episodes of severe conflict tend to develop effective mechanisms to cope with such conflict (De Waal, 1986). In human groups, after a conflict, group members tend to engage in social activities to repair damaged relationships and increase group cohesiveness (Seyfarth, 1980; Strayer & Trudel, 1984). Based on this idea, I argue that the whole process of becoming a functionally differentiated group is vital for developing an informal hierarchy. Specifically, I expect that the establishment of methods to deal with conflict removes group members' perceived boundaries to assert influence. In groups with formal hierarchy, dissent is perceived as a serious challenge, while in groups with informal hierarchy, dissent is perceived as constructive criticism to be resolved by conflict handling strategies. Because these groups have developed informal hierarchy themselves, they have gone through phases of conflict and conflict resolution and have learned to manage differences of opinion. Sharing dissent has became a natural part of the group process, and group members have learned to resolve such conflict. I therefore tentatively suggest that:
Hypothesis 4: Groups without formal hierarchy (and instead an informal hierarchy), will have higher perceptions of overall influence compared to groups with a formal structure.
The hypothesis is illustrated in Figure 4.
12 Figure 4: Group members' influence perceptions in formal and informal group hierarchies
Method Design and Participants
The experiment involved a between-subjects design with two conditions. In the first condition, the presence of a formal hierarchy was manipulated. The groups in this condition consisted of five group members; one formally appointed leader and four formally appointed subordinates. In the second condition, groups consisted of four group members. These group members were all appointed to the role of team member, which gave them no formally differentiated position. The sample consisted of 189 students of the University of Groningen (96 male, 93 female), who participated either for course credit or for money. The students were assigned randomly to one of the 42 mixed-gender groups. 21 groups were in the first condition, 21 groups were in the second condition.
Procedure
Upon arrival in the lab, participants signed an informed consent form about their
participation in the experiment. After that, participants were seated in separate cubicles where
they were informed about the experimental procedure, filled in a pre-task questionnaire, and
received materials to prepair for the group task. Group members individually solved the
NASA's 'Lost on the Moon' task, in which they imagine their group is a team of astronauts
who have crash-landed on the moon and have to return to the mother ship (Shelly & Troyer,
2001). They are told to have salvaged 15 items from the wreckage, which they have to rank in
order of importance. After all participants worked on the task individually, they were seated
in a room together and asked to develop a collective ranking of the 15 items. Having finished
13
the group task, participants were again seated into the individual cubicles to finish a post-task questionnaire. At the end of the experiment, they were debriefed about the purpose of the study, thanked and reimbursed for their participation.
Formal Hierarchy Manipulation
In the condition where a formal hierarchy was present, one of the group members was appointed as formal group leader (see Galinsky et al.; 2003). The power manipulation gave the leader control over resources (by letting him distribute a financial bonus) and gave the leader the ability to evaluate the subordinates. To legitimize the appointment of a leader, the experimenter told participants to complete a personality questionnaire that would determine their role in the team. However, the leader and subordinate roles were in fact randomly assigned among the group members. The group member with the leader role received the following instruction:
You will be the leader of the group. Together with your team, you will try to find a group solution the survival task. As the leader, you are in charge of structuring this process. You will decide how to structure the discussion, and you will set the standards by which your group members are evaluated. Because you are the group leader, you receive two raffle tickets that each represent a chance of winning a 10 euro extra reward. This raffle will be conducted at the end of this experiment. One of the raffle tickets is for you. At the end of the experiment, you will give the second raffle ticket to the subordinate that you think performed best in the group.
The remaining members of his or her group were placed in the subordinate role and received the following instruction;
Your role in the group is the role of subordinate. Together with the group, you will try to find a group solution to the survival task. One of the other members is appointed as leader of the group. He/she will be in charge. He or she directs the discussion process and sets standards to evaluate you and the other subordinates. In addition, the leader has received two raffle tickets that each represent a chance of winning a 10 euro extra reward in a raffle that will be conducted after this experiment. One of the tickets is for the leader, the other ticket will be distributed by the leader to the subordinate that performed best in the group.
In the condition where a formal hierarchy was absent, no information was provided about
formal role distributions. The members of all groups received the instruction to conform to the
role of team member, based on the following instruction:
14
During this experiment, you and the other people in your team are all group members. Together with the other members, you are responsible for finding a group solution to the survival task.
Therefore, you all carry responsibility for a good outcome of the task. At the end of the experiment, a financial bonus of 10 euros will be distributed to the best performing team member. Your performance will be based on a comparison between your individual task solutaions and the collective team solution.
General Measures
Manipulation check. Two questions were used to check the manipulation. The first question asked participants to indicate their formal role in the group as assigned by the experimenter (i.e. leader, subordinate or team member). The second question asked participants whether the experimenter had assigned to them a formal power role in the group.
Conflict. Conflict was measured using the scale of Jehn (1995). Participants indicated the frequency of experienced conflict using a 7-point response scale from 1 (never) to 7 (always).
Example items are: "There was much relationship tension in our team." (representing relationship conflict); "There were often disagreements about who should do what in our team." (representing process conflict); and "People in our team had conflicting ideas about the task we were working on." (representing task conflict). Although the scale measures all types of conflict seperately, it can also be used as a generic conflict scale (Jehn, 1995). Cronbach's alpha of the averaged generic conflict scale was .72.
Dyadic Influence Measure.
The strength of the informal hierarchy (i.e. linearity and steepness) is generally calculated based on the dyadic relationships between individuals in the group (Langbein & Puppe, 2004), as in animal groups, round robin analysis is not available.
For this reason, we used dyadic measurement to calculate linearity and steepness. The level of influence of each group member was measured by letting participants judge every dyad in the team in terms of who was most influential during the task (e.g. member A relative to member B). For every dyad, the answering options were: A was more influential than B; they were equally influential; B was more influential than A. Every actor in a four person group rates a total of six dyads: member A - B, Member A - C, Member A - D, Member B - C, Member B - D and Member C - D. The question order of paired comparisons was based on Ross' ordering method (1934) to eliminate space and time errors, to avoid regular repetitions which might influence judgment, and to maintain the greatest possible spacing between pairs in the group.
Linearity. On the basis of the dyadic influence scores, I calculated hierarchical linearity by
using a formula developed by Singh, Singh, Sharma & Krishna (2003). The formula is
15
appropriate for the measurement of linearity, because it incorporates proportions of wins and losses for individuals instead of absolute values. Thus, when some group members disagree about who is mostly influential, this information is not treated as error variance but as important calculation input. The linearity formula is based on the proportion of wins in pair- wise encounters (P
a) for each person in the group. The sum of these proportions is used to calculate the d
avalue. The formula is the following:
