British Journal of Psychology (2018)
©2018 The British Psychological Society
I follow, therefore I lead: A longitudinal study of
leader and follower identity and leadership in the
Kim Peters* and S. Alexander Haslam
School of Psychology, The University of Queensland, Brisbane, Queensland, Australia
It is acknowledged that identity plays an important role in a person’s leadership
development. To date, however, there has been little consideration of the possibility –
suggested by the social identity perspective –that individuals who identify as followers
may be especially likely to emerge as leaders. We test this possibility in a longitudinal
sample of recruit commandos in the Royal Marines. Recruits rated their identiﬁcation
with leader and follower roles ﬁve times over the course of their 32-week training
programme. Recruits’ leadership and followership were evaluated by their commanders,
and their leadership was assessed by their peers. Analysis indicated that while recruits
who identiﬁed as leaders received higher leadership ratings from their commanders,
recruits who identiﬁed –and were perceived –as followers emerged as leaders for their
peers. These ﬁndings suggest that follower and leader identities underpin different aspects
of leadership and that these are differentially recognized by others.
For many scholars, practitioners and laypeople, leadership and followership are very
different things. Indeed, many see one as the direct opposite of the other. Thus
writing in Forbes, Bradberry (2015) asks ‘Are you a leader or a follower?’ and
Leadership and followership are mindsets. They’re completely different ways of looking at the
world. One is reactive, and the other is proactive. One is pessimistic; the other is optimistic.
Where one sees a to-do list, the other sees possibilities.
This distinction is preserved in most academic treatments too. So, while leadership is
deﬁned as a capacity to exert inﬂuence over others, followership is deﬁned as a
willingness to accept it (Bligh, Pillai, & Uhl-Bien, 2009; Hollander & Webb, 1955).
Accordingly, the aim of many individual difference approaches was to establish which of
these two categories a person belongs to (e.g., after Gibb, 1958; Stogdill, 1948). As
Torpman (2004) observes, in contemporary organizational contexts this means that
leadership typically has a ‘differentiating function’ in which a key managerial objective is
to identify would-be leaders and then set them apart from followers (physically,
If leaders and followers are indeed differentiated in this way, then people who see
themselves as leaders should be unlikely to see themselves as followers. More
*Correspondence should be addressed to Kim Peters, School of Psychology, University of Queensland, St. Lucia, Qld 4072,
Australia (email: firstname.lastname@example.org).
importantly, if leadership and followership are different mindsets, then it is the people
who see themselves as leaders (not followers) who should come to be recognized as
leaders by others. The existing empirical evidence is broadly consistent with this
possibility, as it has found that people who identify as leaders are more likely to attain
leadership positions (Chan & Drasgow, 2001; Kark & van Dijk, 2007; Lord & Hall, 2005).
However, there is very little (if any) evidence that people who identify as followers are less
likely to emerge as leaders. Indeed, there is reason to question the assumption that
leadership and followership operate in the ﬁxed antagonistic ways suggested above. This
is because group-based perspectives on leadership suggest that there may be contexts in
which a follower identity will increase a person’s capacity to inﬂuence other group
members (Haslam, Reicher, & Platow, 2011; Hogg, 2001; Reicher, Haslam, & Hopkins,
2005). This study aimed to directly test the antagonistic assumption by exploring the
unique contributions of people’s leader and follower identities to observers’ recognition
of their leadership.
Identifying as leader and follower
Leadership is generally deﬁned as the process of inﬂuencing other people in ways that
motivate them to contribute to the achievement of collective goals (e.g., Haslam, 2004;
Rost, 2008). This deﬁnition implies that, within a particular group at any given point in
time, people will occupy one of two complementary roles: that of leader, who exerts
inﬂuence, or that of follower, who responds to that inﬂuence. Although group members
do not need to see themselves in terms of these roles for leadership to occur, individuals
often internalize leader and follower roles as identity constructs that speak to their unique
worth within a social group (Ashforth, 2001; Chan & Drasgow, 2001; DeRue & Ashford,
2010; Epitropaki, Kark, Mainemelis, & Lord, 2017; Katz & Kahn, 1978; Sluss, Dick, &
Thompson, 2011). Importantly, as with any other self-categorization, when a group
member internalizes a role as leader or follower they are more likely to adopt and act in
terms of role-related goals, values and norms (Turner, Hogg, Oakes, Reicher, & Wetherell,
1987; Turner, Oakes, Haslam, & McGarty, 1994). So, where identifying as a leader should
increase a person’s tendencies to engage in behaviours that are prototypical of leaders
(e.g., by being dominant and conﬁdent; Epitropaki & Martin, 2004; Guill
en, Korotov, &
Mayo, 2012; Smith & Foti, 1998), identifying as a follower should increase their tendencies
to engage in behaviours that are prototypical of followers (e.g., by being industrious and a
good citizen; Sy, 2010).
While there is a growing body of work that has explored the impact of a leader identity
on a person’s leadership, very little (if any) research has directly examined the
implications of taking on a follower identity. One reason for this is the fact that, as
noted above, there is often an implicit assumption in the leadership literature that
follower and leader identities have ﬁxed and antagonistic effects on leadership
outcomes. For example, Chan and Drasgow’s (2001) Motivation to Lead scale –which
is seen as a stable predictor of leadership performance –captures a person’s self-concept
of the self as leader not follower. By implying that a leader identity comes at the cost of a
follower identity (and vice versa), this perspective suggests that effective leadership will
be facilitated by boosting a person’s identity as leader and suppressing that of follower.
However, there are perspectives that suggest that these identities need not be
antagonistically related. For instance, some researchers (e.g., Haslam et al., 2011) have
argued that it is possible for leader and follower identities to coexist within a person (albeit
in different contexts and at different times). Furthermore, in one notable study, Hollander
2Kim Peters and S. Alexander Haslam
and Webb (1955, see also Tanoff & Barlow, 2002) found that if naval aviation cadets
nominated a particular peer as someone that they would choose to lead, they were almost
certain to nominate the same person as someone they would choose to follow.
Importantly, this pattern of nominations was not accounted for by a preference for
working with friends. This pattern of ﬁndings led the researchers (see also Hollander &
Offermann, 1990) to conclude that, while leadership and followership are typically seen
as distinct, they are often aligned –not least because when people are looking for
leadership they often turn to those who have previously shown followership. Similar
patterns have also been observed in non-military contexts. In particular, in a large sample
of executives, Agho (2009) found widespread agreement for the possibility that effective
followership is a prerequisite for effective leadership. Further, Baker, Mathis, and Stites-
Doe (2011) found that health workers perceived that, in a context of organizational
change, followers possess both exemplary leader and follower characteristics. If these
perspectives are correct in suggesting that leader and follower identities (and behaviours)
are not opposite poles of a scale, then this raises the possibility that both identities may
contribute positively to leadership.
Identifying as leader and follower and leadership recognition
There is a growing body of work which shows that seeing oneself as a leader can be
expected to facilitate leadership development because it is likely to increase a person’s
motivation to engage in, and persist with, behaviours that will help him or her to succeed
as a leader and to acquire formal leadership positions. The importance of this motivational
function is highlighted by the observation that to be an effective leader a person may need
to work actively on acquiring a range of social, cognitive, and behavioural skills over an
extended period of time (Day & Halpin, 2004; Lord & Hall, 2005). Chan and Drasgow’s
(2001) large-scale study of male military recruits to the Singaporean military provides
evidence of this possibility. This study found that recruits who identiﬁed more highly as a
leader were evaluated as having greater leadership potential by assessment centre raters;
they also received higher ratings of leadership potential from their peers and supervisors
upon completion of their training programme. Along similar lines, Day and Sin (2011)
found that students who reported higher levels of leader identity were rated by their
coaches as exhibiting higher levels of leadership within their teams. Miscenko, Guenter,
and Day (2017) also observed a positive association between postgraduate students’
identiﬁcation as leaders and self-reported leadership skills over the course of a 7-week
leadership training module. Together, these strands of theory and evidence suggest that
people who identify as leaders are more likely to be recognized as such. This therefore
provides the basis for our ﬁrst hypothesis:
Hypothesis 1: Group members who identify more highly as leaders are more likely to be
perceived as leaders.
At the same time, though, when one considers the group that provides the context for
such leader development behaviours, it is less obvious that a leader identity will always
increase a person’s capacity to inﬂuence other group members. Indeed, the social identity
approach to leadership, which incorporates social identity theory (Tajfel & Turner, 1979)
and self-categorization theory (Turner, 1982; Turner et al., 1987, 1994), provides the
basis for expecting that a follower identity may be more helpful in this regard.
Leading by following 3
According to this approach, a person’s inﬂuence within a group is determined to a
large degree by group members’ perceptions that he or she represents, and is committed
to, the group (Haslam et al., 2011; Hogg, 2001; Reicher et al., 2005; Van Knippenberg &
Hogg, 2003; van Knippenberg, van Knippenberg, De Cremer, & Hogg, 2004; Turner &
Haslam, 2001). This claim is borne out by evidence that prototypical group members, who
are seen to embody ideal characteristics of the group, are more likely to be endorsed as
leaders than their less prototypical peers (e.g., Platow & van Knippenberg, 2001; Ullrich,
Christ, & van Dick, 2009). It is also supported by evidence that group members are more
likely to endorse a person’s leadership and to display followership if that person clearly
identiﬁes with the group and is committed to realizing its ambitions rather than their own
personal ones (e.g., Duck & Fielding, 2003; Giessner, van Knippenberg, van Ginkel, &
Sleebos, 2013; Graf, Schuh, Van Quaquebeke, & van Dick, 2012; Haslam & Platow, 2001;
Jetten, Duck, Terry, & O’Brien, 2002; van Dick, Hirst, Grojean, & Wieseke, 2007; van
Knippenberg & van Knippenberg, 2005). Illustrative of both these processes, archival
research by Steffens and Haslam (2013) showed that in all the Australian Federal elections
that have taken place since 1901, Prime Ministerial candidates were much more likely to
have been successful (winning 34 of the 43 elections) if they made more references to ‘we’
and ‘us’ in their campaign speeches than their opponent.
According to the above arguments, whether leader and follower identities can be
expected to facilitate social inﬂuence and leadership is a function of the extent to which
these identities encourage the behaviours that signal a person’s oneness with, and
commitment to, the group. All else being equal, it seems that a follower identity may be
more likely to promote this group-serving behaviour than a leader identity. Not least, this
is because a leader identity may increase a person’s motivation to differentiate themselves
from potential followers (Lord & Hall, 2005; Van Knippenberg & Hogg, 2003), where this
can be realized by engaging in prototypical leader behaviours of dominance and
conﬁdence (Epitropaki & Martin, 2004; Guill
en et al., 2012; Smith & Foti, 1998). In
contrast, prototypical follower behaviours include displays of industry and citizenship
(Smith & Foti, 1998; Sy, 2010), where these behaviours that more clearly situate a person
as striving within and for the group. Together, then, these strands of theory and evidence
provide the basis for a second hypothesis:
Hypothesis 2: Group members who identify more highly as followers are more likely to be
perceived as leaders.
In sum, our theoretical analysis provides reason to expect that the impact of leader and
follower identities on a person’s ability to lead may be more complementary than has
previously been assumed and that identifying with either role may confer beneﬁts. In what
follows, we present the results of a longitudinal ﬁeld study that was designed to shed light
on this possibility. This study examined the unique impact of marine recruits’
identiﬁcation as leader and follower (measured repeatedly over the course of their
training) on commander and peer perceptions of their leadership on completion of their
Although not part of our a priori reasoning, as will become evident from our results,
different audiences appeared to respond differently to recruits’ identiﬁcation as leader or
follower. Speciﬁcally, our ﬁndings are consistent with the possibility that commanders,
who observe recruits from outside their group, are particularly sensitive to recruits’
tendencies to behave in line with the generic leadership prototype (Epitropaki et al.,
2017; Lord & Hall, 2005), while ingroup peers are more sensitive to recruits’ tendencies to
4Kim Peters and S. Alexander Haslam
represent and advance their group (Platow & van Knippenberg, 2001). These observa-
tions, if robust, support a more nuanced understanding of the implications of leadership
role identities for social inﬂuence and leadership recognition, and the conditions under
which they are likely to translate into the attainment of leadership positions. We expand
on this (and alternative) theorizing in the Discussion.
Participants were 218 Royal Marine recruits belonging to one of four troops that
commenced their training in September or October of 2008. (Our sample size was
determined by troop commanders’ willingness to only grant us access to four troops for
the duration of their training). As the ban on women in ground close combat roles in the
United Kingdom was only lifted in 2016, all were male. Surveys were completed during
scheduled classroom time, and all recruits consented to participate. The ﬁnal sample for
analytic purposes consisted of the 68 recruits who (1) completed at least one of the ﬁve
surveys (63 recruits completed four or more), (2) received commander ratings of their
leadership and followership midway through their training and again on its completion,
and (3) were rated by their peers as part of the Commando Medal vote. The ﬁnal sample
were on average 20.90 years old (SD =2.76, range 16–28 years).
Sample attrition was primarily due to the arduous nature of marine commando
training. The Royal Marines is a specialized infantry force with particular expertise in
warfare in extreme environments (including arctic and mountainous terrain). Every year,
approximately 1,200 recruits embark on the 32-week training programme, which is
considered to be one of the most physically demand ing specialist infantry training regimes
in the world. According to Royal Marines Commando Training Centre statistics, on
average, just over half of recruits successfully complete their training, and only around
40% complete it as a member of their original troop (injured recruits undergo
rehabilitation before continuing with a new troop and this may delay their completion
by many months). Importantly, because recruits are not permitted to discharge from
training before the end of Week 4, we were able to examine whether there were any
differences in leader and follower identity between the ﬁnal and initial samples on the ﬁrst
four surveys. This analysis did not reveal any evidence that recruits who completed the
training within the time frame captured by our study differed in their endorsement of a
leader or follower identity or in their interpersonal dominance from the recruits who did
not, all t(214–227) <|1.23|,p>.222.
As part of a broader project examining recruit belonging and career motivation,
participants were asked to complete three scales relating to leader and follower identity.
We measured leader identity with three items (as=.56–.80): ‘I think that I am a natural
leader’, ‘I have the skills and abilities to lead others’, and ‘I prefer to keep with my peers,
rather than standing out as a leader’ [reversed]. We measured follower identity with two
items (r=.28, p=.026, to r=.68, p<.001): ‘I think it is more important to get the job
done than to get my way’, and ‘If another person has a good idea, I am prepared to work
hard to make it happen’. Finally, to control for the possibility that identity endorsement
could be confounded with recruits’ interpersonal dominance (see Nevicka, De Hoogh,
Leading by following 5
Van Vianen, Beersma, & McIlwain, 2011, for evidence that such narcissis tic characteristics
predict leader categorization), we also measured dominance with two items (r=.49,
p<.001, to r=.75, p<.001): ‘I think that my ideas are usually better than other peoples’
ideas’, and ‘I enjoy getting other people to do things my way’. Responses to these various
items were made on 7-point Likert scales (1 =strongly disagree, 7 =strongly agree).
Conﬁrmatory factor analysis with recruits who completed the identity scales on Day 1
(N=216) indicated that the three-factor model provided an adequate ﬁt for the data: LR
(11) =29.31, p=.002, RMSEA =.088 (95% CI: 0.050–0.127), CFI =.95,
SRMR =.050. These scales were administered during scheduled classes on Day 1, Day
5, Day 15, Day 25, and Week 25 of training.
Troop commanders, who each took responsibility for up to 15 recruits, were asked to
rate recruit leadership and followership on two occasions: midway through the training
(Week 16) and again on its completion (Week 31). Leadership was measured with the
following two items (r=.80–.88, p<.001): ‘This recruit has displayed leadership
qualities’ and ‘This recruit has leadership potential’. Followership was measured with the
following two items (r=.67–.85, p<.001): ‘This recruit has responded well to the
leadership of others’ and ‘This recruit has the poten tial to respond well to the leadership of
others’. Responses were made on identical 7-point Likert scales (1 =strongly disagree,
To measure peer evaluations of leadership, as part of the end-of-training voting for the
Commando Medal, each recruit was asked to rate each member of their troop according to
whether he ‘embodied the commando spirit’ (5-point Likert scales, 1 =not at all; 5 =very
much). The commando spirit encompasses the attributes of leadership, courage,
unselﬁshness, cheerfulness, and determination, and the Commando Medal is explicitly
awarded in recognition of these qualities (King, 2004; see Figure 1). As suggested by the
ordering of the attributes on the medal –further borne out in our conversations with
training commanders –this award is primarily understood as a recognition of leadership.
Construct means, standard deviations, and correlations are summarized in Table 1. From
this, it can be seen that recruits were, on average, more likely to identify as a follower than
Figure 1. The Commando Medal.
6Kim Peters and S. Alexander Haslam
they were to identify as a leader, t(68) =10.71, p<.001. Troop commanders were
likewise more likely to report that their recruits were effective followers than that they
were effective leaders, both at training midpoint, t(68) =6.42, p<.001, and on training
completion, t(68) =6.80, p<.001. There was very little evidence that a recruit’s
identiﬁcation as a leader came at the expense of his identiﬁcation as a follower, as the
negative correlation between these measures was small and non-signiﬁcant. In fact,
commanders perceived that the recruits who displayed more leadership were also those
who displayed more followership. Importantly, the more recruits endorsed leader or
follower identities, the more they were perceived to be better leaders at the end of their
Together, these ﬁndings suggest that the general assumption that leader and follower
identities (and behaviours) are antagonistic may be unwarranted and that both identities
may make positive contributions to a person’s leadership. In the analyses below, we ﬁrst
examine the stability of recruits’ leader and follower identities before conducting formal
tests of our hypotheses.
Leader and follower identity stability
Role identities are considered to be at least somewhat stable within people over time,
thereby providing them with the ongoing motivation that is required to acquire a role and
succeed in it. The intraclass correlation coefﬁcients showed that a sizeable amount of
variance in identity ratings (especially leader identity ratings) was due to differences
between (rather than within) recruits: leader identity ICC =.64; follower identity
ICC =.23. It is therefore plausible that these identity constructs could contribute to the
recruits’ leadership development over the course of their training.
Table 1. Means and correlations for recruit, commander, and peer ratings of leadership and
Construct Mean SD 1234567
Recruit identity measures
1. Leader identity
4.43 0.94 –
2. Follower identity
5.91 0.59 .04 –
3.68 0.96 .32** .20 –
4.24 1.58 .10 .15 .16 –
4.35 1.55 .32** .12 .13 .36** –
5.59 1.21 .02 .03 .07 .24* .16 –
5.68 1.02 .14 .03 .06 .02 .27* .05 –
8. Commando spirit
3.27 0.60 .06 .24* .05 .04 .34** .17 .51***
Identity ratings aggregated across surveys.
Commando Spirit rated on a 5-point Likert scale (all others rated on 7-point scales).
*p<.05; **p<.01; ***p<.001.
Leading by following 7
To understand whether participating in the training affected recruits’ identiﬁcation as
leader or follower, we regressed the identity ratings onto the measurement time point
using ordinary least squares method (OLS). We controlled for the repeated nature of these
ratings by clustering standard errors at the level of the recruit (Wooldridge, 2003). Table 2
summarizes recruits’ mean levels of identity endorsement at each of the ﬁve measurement
points. This analysis did not provide any evidence that ratings of leader identity varied over
=.00, F(1, 67) =2.29, p=.135. However, there was evidence that recruits’
endorsement of a follower identity did vary, at least to a small extent, R
67) =4.60, p=.036. In particular, recruits were more likely to endorse a follower
identity earlier in their training, b=0.07, t(67) =2.15, p=.036. When we repeated
this analysis for recruits’ dominance, we also found evidence of temporal variation,
=.04, F(1, 67) =15.10, p<.001, with recruits’ ratings of interpersonal dominance
increasing over time, b=0.14, t(67) =3.89, p<.001. Therefore, while progressing
through commando training appeared to have little impact on recruits’ identiﬁcation with
a leader role, it may weaken their identiﬁcation as a follower and boost their dominance
(albeit to a small degree).
Commander leadership and followership ratings
As a ﬁrst test of our hypotheses, we assessed the association between recruits’ leader and
follower identities and commanders’ end-of-training perceptions of their leadership, using
hierarchical OLS regression and clustering standard errors at the level of the participant to
account for the nested structure of the data (Wooldridge, 2003). All recruit self-report
measures were centred at the scale mean. To enable a consideration of temporal effects,
we calculated the two-way interactions between the three centred self-report variables
and the measurement time point (coded: Day 1 =2, Day 5 =1, Day 15 =0, Day
25 =1, Week 25 =2). Table 3 contains the unstandardized regression coefﬁcients from
In Step 1, we regressed leadership perceptions onto the identity and dominance
ratings (see Figure 2). This step accounted for a marginal amount of variance in leadership
ratings, F(3, 67) =2.66, p=.055. In line with H1, recruits who identiﬁed more with the
leader identity were ultimately perceived to be better leaders by their commanders. In
contrast, and contra H2, there was no evidence that identifying as a follower conferred any
In Step 2, we included the leadership and followership ratings that commanders
provided halfway through the training programme. This allowed us to ascertain whether
identity accounted for changes in leadership recognition over the second half of the
Table 2. Means and 95% conﬁdence intervals for recruit identity endorsement over time
Construct Leader identity Follower identity Dominance
Day 1 4.56 (4.24–4.88)
Day 5 4.49 (4.20–4.79)
Day 15 4.33 (4.01–4.64)
Day 25 4.34 (4.06–4.62)
Week 25 4.34 (4.08–4.60)
Note. Column means with superscripts containing no common letters differ at p<.05.
8Kim Peters and S. Alexander Haslam
training course. This step resulted in a signiﬁcant improvement in model ﬁt, Wald F(2,
67) =9.00, p<.001, and the overall model was signiﬁcant, F(5, 67) =7.07, p<.001.
Importantly, leader identity was a signiﬁcant positive predictor over-and-above midpoint
leadership perceptions, suggesting that leader identity continues to contribute positively
to recruit leadership recognition over time. Interestingly, recruits who were initially
perceived by their commanders to be good followers were perceived to be marginally
poorer leaders at the end of training.
In Step 3, we included measurement time point and the three-two-way interactions
between time point and the recruit self-report measures. This did not signiﬁcantly
Figure 2. Plot of best ﬁt relationship between recruit follower or leader identity endorsement and
leadership evaluations by commanders (right) and peers (left).
Notes. Best ﬁt lines plot the impact of variations in focal identity at mean values of non-focal identity;
Commander and peer leadership ratings are standardized to allow comparability of scales.
Table 3. OLS regression unstandardized coefﬁcients predicting commander leadership and follower-
ship ratings at completion of training
Step 1 Step 2 Step 3 Step 1 Step 2 Step 3
Recruit identity measures
Leader identity .39* .37** .36** .12 .12 .13
Follower identity .19 .14 .08 .05 .04 .03
Dominance .04 .00 .01 .06 .06 .07
Leadership (training midpoint) .40*** .40*** .01 .01
Followership (training midpoint) .35
Leader identity 9time .01 .01
Follower identity 9time .07 .02
Dominance 9time .02 .05*
Constant 4.38*** 4.62*** 4.59*** 5.69*** 5.56*** 5.52***
.27*** .28*** .02 .02 .04
Notes.N=68, observations =304; Standard errors clustered at the participant level.
Time coded on 5-point scale (Day 1 =2, Day 5 =1, Day 15 =0, Day 25 =1, Week 25 =2).
p<.10; *p<.05; **p<.01; ***p<.001.
Leading by following 9
improve model ﬁt, Wald F(4, 67) =0.67, p=.616, which suggests that the relationship
between identity and leadership is stable over time.
In order to ascertain whether recruits’ leader and follower identities also have
implications for commanders’ ultimate perceptions of their followership, we repeated the
above analysis for followership evaluations. This time, none of the hierarchical regression
steps was able to account for a signiﬁcant amount of variance in followership ratings: Step
1F(3, 67) =0.62, p=.608, Step 2 F(5, 67) =0.41, p=.843, and Step 3 F(9, 67) =1.10,
p=.374. An examination of Table 3 reveals that, if anything, recruits who initially
reported higher levels of interpersonal dominance were ultimately perceived to be less
effective as followers.
Peer leadership ratings
As a second test of our hypotheses, we assessed the association between recruits’ leader
and follower identities and peer leadership evaluations in the form of ratings of
commando spirit. We followed the hierarchical regression approach described above,
with one exception: In Step 2, we also included commander’s end-of-course leadership
and followership evaluations. By doing this, we were able to assess how recruits’
leadership and followership behaviours (evaluated at two time points by commanders)
related to peer evaluations of their leadership. The unstandardized regression coefﬁcients
from this analysis are provided in Table 4.
Step 1, which included the three self-report measures, did not account for a signiﬁcant
amount of variance in leadership ratings, F(3, 67) =1.09, p=.358. While there was no
evidence that a leader identity positively affected peer leadership perceptions (contra
H1), there was weak evidence in support of H2, as recrui ts who endorsed higher levels of a
Table 4. OLS unstandardized regression coefﬁcients predicting peer leadership ratings
Step 1 Step 2 Step 3
Recruit identity measures
Leader identity .01 .05 .05
Follower identity .08
Dominance .01 .01 .01
Leadership (training midpoint) .01 .01
Leadership (training completion) .10
Followership (training midpoint) .09* .09*
Followership (training completion) .23** .23**
Leader identity 9time .00
Follower identity 9time .00
Dominance 9time .01
Constant 3.29*** 2.10*** 2.10***
.02 .34*** .34***
Notes.N=68, observations = 304; Standard errors clustered at the participant level.
Time coded on 5-point scale (Day 1 =2, Day 5 =1, Day 15 =0, Day 25 =1, Week 25 =2).
p<.10; *p<.05; **p<.01; ***p<.001.
10 Kim Peters and S. Alexander Haslam
follower identity received marginally higher ratings from their peers. Figure 2 graphically
represents the impact of each identity on leadership ratings.
Step 2, which included the commander leadership and followership perceptions, led
to a signiﬁcant improvement in model ﬁt, Wald F(4, 67) =6.16, p<.001. While there was
again no support for H1, there was further support for H2. Speciﬁcally, recruits who
endorsed higher levels of a follower identity received signiﬁcantly higher leadership
ratings. Interestingly, commander’s ultimate evaluations of recruits’ followership (but not
their earlier evaluations) also signiﬁcantly predicted peer leadership evaluations, and
recruits who were seen as effective followers were perceived by their peers to be more
effective leaders. Weaker evidence also emerged of a similar association between
commanders’ ultimate evaluations of recruit leadership and those of their peers.
Finally, Step 3, which accounted for temporal variations in the relationship with
identity, did not signiﬁcantly improve model ﬁt, Wald F(4, 67) =0.23, p=.918,
suggesting that the observed relationships were stable over time.
This study was designed to explore the relationship between recruits’ leader and follower
identities and their leadership emergence, as assessed both by their commanders and their
peers. Together, the pattern of results that we describe in this study provides partial
support for our hypotheses. First, in line with H1, there was evidence that recruits who
saw themselves as leaders were more likely to be seen as such by the commanders who
oversaw their training. This pattern did not change signiﬁcantly over time. Second, in line
with H2, there was evidence that recruits who saw themselves as followers (and were
seen as followers by their commanders) were more likely to be seen as leaders by their
peers. Again, this pattern did not change signiﬁcantly over time.
These ﬁndings support previous claims that a person’s identity plays an important role
in their emergence as a leader over time; they also point to the import ance of considering a
broader array of identities beyond a person’s tendency to see him or herself as a leader. In
particular, our ﬁnding that a leader identity predicts subsequent leadership is in line with
the possibility that this identity motivates people to behave in ways that are consistent
with the leader prototype and to seek out leadership opportunities and positions (e.g.,
Day & Sin, 2011). At the same time, our ﬁnding that a follower identity is also predictive of
subsequent leadership is consistent with the possibility that this identity motivates people
to act as one with and on behalf of the group, creating the sense of shared identity that is
required for social inﬂuence (e.g., Haslam et al., 2011).
However, while our ﬁndings support the contention that leader and follower identities
do not come at the expense of one another (in this sample, they appear to be largely
orthogonal constructs) and therefore can both exert positive effects on a person’s
leadership, we did not ﬁnd evidence of general complementarity. That is, the leadership of
recruits who adopted a leader identity was most clearly recognized by those outside of
their group (the commanders who oversaw their training), while the leadership of recruits
who adopted a follower identity tended to be appreciated only by their fellow ingroup
members. This raises the possibility of a more nuanced relationship between leadership
role identities and leadership emergence, where the important question is less whether
leader and follower identities matter and more when and to whom they matter.
Our ﬁndings suggest that one important moderator of the relationship between
leadership role identities and leader emergence is the structural position of the person
Leading by following 11
doing the evaluating. The responses of evaluators who were external, and superior, to the
group that provides the context for leadership (i.e., commanders) were different to the
responses of evaluators who were located within the group (i.e., peers). There are at least
three explanations of why structural position may affect leadership evaluations in the way
that we observed: (1) involvement in the leadership process, (2) sensitivity to impression
management, and (3) notions of leadership potential.
The ﬁrst of these concerns the personal involvement of the evaluator in the leadership
process that they are evaluating. Where peers are likely to have personal experience of a
recruit’s capacity to inﬂuence them in ways that facilitate the achievement of their group’s
goals, commanders are not. This means that consistent with the predictions derived from
a social identity approach, peers (especially those who highly identify with their group)
will recognize and respond to the leadership of group members who are commited to and
represent the group (Haslam et al., 2011; Platow & van Knippenberg, 2001; van Dijke &
De Cremer, 2010; Van Knippenberg & Hogg, 2003). In contrast, commanders, who lack
the experience of being led by a given recruit, may attend to the extent to which a recruit
behaves in ways they believe a leader should (e.g., Epitropaki & Martin, 2004; Epitropaki
et al., 2017; Lord & Hall, 2005; Smith & Foti, 1998).
However, other explanations are possible too. The second explanation concerns the
involvement of the evaluator in a recruit’s impression management attempts. This time it
is more likely that a recruit who is seeking material advantage and advancement will direct
their impression management behaviours upwards, towards the evaluator with greater
structural power, rather than horizontally, towards ingroup members. Peers, for their
part, may be sensitive to, and suspicious of, a recruit’s impression management efforts
(e.g., in the form of ingratiation; Varma, Min Toh, & Pichler, 2006). If a leader identity
promotes such impression management behaviour, then our ﬁndings may reﬂect the fact
that superiors are more susceptible to them than equal-ranking ingroup members. The
ﬁnal explanation concerns the possibility that people in different structural positions have
different notions of leadership potential, and that while commanders are not recognizing
those recruits who are effective leaders of their peers, they may be attendin g to behaviours
that are predictive of leadership emergence in non-training contexts and across a career.
Gaining evidence of the validity of these alternative explanations is an important agenda
for future investigation.
While our ﬁndings point towards a more nuanced understanding of the impact of role
identities on leadership emergence, they also point to important structural barriers which
limit the capacity for individuals who are recognized as effective leaders by their peers to
attain formal leadership positions. In particular, our results suggest that commanders not
only rewarded those recruits who identiﬁed (and behaved) as leaders by recognizing their
leadership, they also (marginally) denied recognition to recruits who they perceived as
having the potential to be good followers. On the face of it, this is a recipe for establishing
ineffective leadership structures and increasing team dysfunction. Organizations that
utilize democratic processes for the selection of formal leaders (i.e., processes that tap
into team member experiences of leadership within their team) may well beneﬁt from
While there are a number of strengths to this study (including the multisource
longitudinal nature of the data set and relevance of the training context to leadership),
there are also a number of limitations, not least the relatively small size of the sample. Also,
although there was no evidence that recruits who completed their training in the time
frame that we were able to capture differed in their endorsement of leader and follower
identities from those who did not (either because they opted out of training, or were
12 Kim Peters and S. Alexander Haslam
injured and entered rehabilitation), we cannot eliminate the possibility that our ﬁnal
sample differed from our original one in their baseline leadership and followership
capabilities. Relatedly, though longitudinal, the correlational nature of this study means
that it is not possible to make any claims about the causal impact of a leader or follower
identity on leadership emergence. In future work, it is therefore important to address
these limitations (perhaps using experimental methodologies) so as to provide stronger
tests of our claim that the seeds of a person’s leadership may lie not only in their taking on
of the identity of a leader but also in their willingness to assume the identity of a follower
(Hollander & Webb, 1955).
When asked how they approached the task of developing the leadership of their
students, staff at the prestigious West Point Military Academy said: ‘We begin by
teaching them to be followers’ (Litzinger & Schaefer, 1982). From the perspective of
the existing literature, this would seem to be a rather paradoxical approach, because
in encouraging students to see themselves as followers the academy is failing to
promote the leader identity that motivates the acquisition of leadership skills and
positions. However, considered in light of our current ﬁndings, West Point trainers
are not wasting their time. In particular, while effective leadership does to some
extent rest on having the right skills and a willingness to assume positions of
responsibility, it cannot be reduced to them. And consistent with the possibility that
a follower identity motivates group-serving behaviours, we found that recruits who
identiﬁed as followers and who engaged in followership were more likely to be seen
as leaders by their peers. Accordingly, organizations that wish to develop leadership
capacity may be well served by promoting their members’ identities as followers, not
just as leaders.
Agho, A. O. (2009). Perspectives of senior-level executives on effective followership and leadership.
Journal of Leadership & Organizational Studies,16, 159–166. https://doi.org/10.1177/
Ashforth, B. E. (2001). Role transitions in organizational life: An identity-based perspective.
Mahwah, NJ: Lawrence Erlbaum Associates.
Baker, S. D., Mathis, C. J., & Stites-Doe, S. (2011). An exploratory study investigating leader and
follower characteristics at US healthcare organizations. Journal of Managerial Issues,8, 341–
Bligh, M. C., Pillai, R., & Uhl-Bien, M. (2009). The social construction of a legacy: Summarizing and
extending follower-centered perspectives on leadership. In B. Shamir, M. Uhl-Bien, R. Pillai, & M.
C. Bligh (Eds.), Follower-centered perspectives on leadership: A tribute to the memory of James
R. Meindl. Greenwich, CT: Information Age Publishing.
Bradberry, T. (2015). Are you a leader or a follower? Forbes.com. Retrieved from https://www.
Chan, K. Y., & Drasgow, F. (2001). Toward a theory of individual differences and leadership:
Understanding the motivation to lead. Journal of Applied Psychology,86, 481–498. https://doi.
Day, D. V., & Halpin, S. M. (2004). Growing leaders for tomorrow: An introduction. In D. V. Day, S. J.
Zaccaro, & S. M. Halpin (Eds.), Leader development for transforming organizations: Growing
leaders for tomorrow (pp. 3–22). New York, NY: Routledge.
Day, D. V., & Sin, H. P. (2011). Longitudinal tests of an integrative model of leader development:
Charting and understanding developmental trajectories. The Leadership Quarterly,22, 545–
Leading by following 13
DeRue, D. S., & Ashford, S. J. (2010). Who will lead and who will follow? A social process of
leadership identity construction in organizations. Academy of Management Review,35, 627–
Duck, J. M., & Fielding, K. S. (2003). Leaders and their treatment of subgroups: Implications for
evaluations of the leader and the superordinate group. European Journal of Social Psychology,
33, 387–401. https://doi.org/10.1002/ejsp.153
Epitropaki, O., Kark, R., Mainemelis, C., & Lord, R. G. (2017). Leadership and followership identity
processes: A multilevel review. The Leadership Quarterly,28(1), 104–129. https://doi.org/10.
Epitropaki, O., & Martin, R. (2004). Implicit leadership theories in applied settings: Factor structure,
generalizability and stability over time. Journal of Applied Psychology,89, 293–310. https://doi.
Gibb, C. A. (1958). An interactional view of the emergence of leadership. Australian Journal of
Psychology,10(1), 101–110. https://doi.org/10.1080/00049535808255958
Giessner, S. R., van Knippenberg, D., van Ginkel, W., & Sleebos, E. (2013). Team-oriented
leadership: The interactive effects of leader group prototypicality, accountability, and
team identiﬁcation. Journal of Applied Psychology,98, 658. https://doi.org/10.1037/
Graf, M. M., Schuh, S. C., Van Quaquebeke, N., & van Dick, R. (2012). The relationship between
leaders’ group-oriented values and follower identiﬁcation with and endorsement of leaders: The
moderating role of leaders’ group membership. Journal of Business Ethics,106, 301–311.
en, L., Korotov, K., & Mayo, M. (2012). Is leadership a part of me. An identity approach to
understanding the motivation to lead. The Leadership Quarterly,26, 802–820. https://doi.org/
Haslam, S. A. (2004). Leadership. In A. Kuper & J. Kuper (Eds.), The social science encyclopedia (3rd
ed., pp. 566–568). New York, NY: Routledge.
Haslam, S. A., & Platow, M. J. (2001). The link between leadership and followership: How afﬁrming
social identity translates vision into action. Personality and Social Psychology Bulletin,27,
Haslam, S. A., Reicher, S. D., & Platow, M. J. (2011). The new psychology of leadership: Identity,
inﬂuence and power. London, UK & New York, NY: Psychology Press.
Hogg, M. A. (2001). A social identity theory of leadership. Personality and Social Psychology
Review,5, 184–200. https://doi.org/10.1207/S15327957PSPR0503_1
Hollander, E. P., & Offermann, L. R. (1990). Power and leadership in organizations: Relationships in
transition. American Psychologist,45, 179. https://doi.org/10.1037/0003-066X.45.2.179
Hollander, E. P., & Webb, W. B. (1955). Leadership, followership, and friendship: An analysis of peer
nominations. The Journal of Abnormal and Social Psychology,50, 163. https://doi.org/10.
Jetten, J., Duck, J., Terry, D. J., & O’Brien, A. (2002). Being attuned to intergroup differences in
mergers: The role of aligned leaders for low-status groups. Personality and Social Psychology
Bulletin,28, 1194–1201. https://doi.org/10.1177/01461672022812005
Kark, R., & van Dijk, D. (2007). Motivation to lead, motivation to follow: The role of the self-
regulatory focus in leadership processes. Academy of Management Review,32, 500–528.
Katz, D., & Kahn, R. L. (1978). The social psychology of organizations (Vol. 2). New York, NY:
King, A. (2004). The ethos of the Royal Marines: The precise application of will. Exeter, UK:
University of Exeter. Retrieved from https://ore.exeter.ac.uk/repository/handle/10036/58653
Litzinger, W., & Schaefer, T. (1982). Leadership through followership. Business Horizons,25(5),
Lord, R. G., & Hall, R. J. (2005). Identity, deep structure and the development of leadership skill. The
Leadership Quarterly,16, 591–615. https://doi.org/10.1016/j.leaqua.2005.06.003
14 Kim Peters and S. Alexander Haslam
Miscenko, D., Guenter, H., & Day, D. V. (2017). Am I a leader? Examining leader identity
development over time. The Leadership Quarterly,28(5), 605–620.
Nevicka, B., De Hoogh, A. H., Van Vianen, A. E., Beersma, B., & McIlwain, D. (2011). All I need is a
stage to shine: Narcissists’ leader emergence and performance. The Leadership Quarterly,22,
Platow, M. J., & van Knippenberg, D. (2001). A social identity analysis of leadership endorsement:
The effects of leader ingroup prototypicality and distributive intergroup fairness. Personality
and Social Psychology Bulletin,27,1508–1519. https://doi.org/10.1177/01461672012711011
Reicher, S., Haslam, S. A., & Hopkins, N. (2005). Social identity and the dynamics of leadership:
Leaders and followers as collaborative agents in the transformation of social reality. The
Leadership Quarterly,16, 547–568. https://doi.org/10.1016/j.leaqua.2005.06.007
Rost, J. C. (2008). Leadership deﬁnition. In A. Marturano & J. Gosling (Eds.), Leadership: The key
concepts (pp. 94–99). New York, NY: Routledge.
Sluss, D. M., vanDick, R., & Thompson, B. S. (2011). Role theory in organizations: A relational
perspective. In S. Zedeck (Ed.), APA handbook of industrial and organizational psychology,
Vol. 1: Building and helping the organization (pp. 505–534). Washington, DC: American
Psychological Association. https://doi.org/10.1037/12169-016
Smith, J. A., & Foti, R. J. (1998). A pattern approach to the study of leader emergence. The Leadership
Quarterly,9, 147–160. https://doi.org/10.1016/S1048-9843(98)90002-9
Steffens, N. K., & Haslam, S. A. (2013). Power through ‘us‘: Leaders’ use of we-referencing language
predicts election victory. PLoS One,8(10), e77952. https://doi.org/10.1371/journal.pone.
Stogdill, R. M. (1948). Personality factors associated with leadership: A survey of the literature.
Journal of Psychology,25,35–71. https://doi.org/10.1080/00223980.1948.9917362
Sy, T. (2010). What do you think of followers? Examining the content, structure, and consequences
of implicit followership theories. Organizational Behavior and Human Decision Processes,
113(2), 73–84. https://doi.org/10.1016/j.obhdp.2010.06.001
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conﬂict. In W. G. Austin & S.
Worchel (Eds.), The social psychology of intergroup relations (pp. 33–37). Monterey, CA:
Tanoff, G. F., & Barlow, C. B. (2002). Leadership and followership: Same animal, different spots?
Consulting Psychology Journal: Practice and Research,54(3), 157. https://doi.org/10.1037/
Torpman, J. (2004). The differentiating function of modern forms of leadership. Management
Decision,42, 892–906. https://doi.org/10.1108/00251740410550952
Turner, J. C. (1982). Towards a cognitive redeﬁnition of the social group. In H. Tajfel (Ed.), Social
identity and intergroup relations (pp. 15–40). Cambridge, UK: Cambridge University Press.
Turner, J. C., & Haslam, S. A. (2001). Social identity, organizations, and leadership. In M. E. Turner
(Ed.), Applied social research. Groups at work: Theory and research (pp. 25–65). Mahwah, NJ:
Lawrence Erlbaum Associates.
Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the
social group: A self-categorization theory. Oxford, UK: Basil Blackwell.
Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and collective: Cognition and
social context. Personality and Social Psychology Bulletin,20, 454–463. https://doi.org/10.
Ullrich, J., Christ, O., & van Dick, R. (2009). Substitutes for procedural fairness: Prototypical leaders
are endorsed whether they are fair or not. Journal of Applied Psychology,94(1), 235. https://
van Dick, R., Hirst, G., Grojean, M. W., & Wieseke, J. (2007). Relationships between leader and
follower organizational identiﬁcation and implications for follower attitudes and behaviour.
Journal of Occupational and Organizational Psychology,80(1), 133–150. https://doi.org/10.
Leading by following 15
Van Dijke, M., & De Cremer, D. (2010). Procedural fairness and endorsement of prototypical leaders:
Leader benevolence or follower control? Journal of Experimental Social Psychology,46(1),
Van Knippenberg, D., & Hogg, M. A. (2003). A social identity model of leadership effectiveness in
organizations. Research in Organizational Behavior,25, 243–295. https://doi.org/10.1016/
van Knippenberg, B., & van Knippenberg, D. (2005). Leader self-sacriﬁce and leadership
effectiveness: The moderating role of leader prototypicality. Journal of Applied Psychology,
90(1), 25. https://doi.org/10.1037/0021-9010.90.1.25
van Knippenberg, D., van Knippenberg, B., De Cremer, D., & Hogg, M. A. (2004). Leadership, self,
and identity: A review and research agenda. The Leadership Quarterly,15, 825–856. https://doi.
Varma, A., Min Toh, S., & Pichler, S. (2006). Ingratiation in job applications: Impact on selection
decisions. Journal of Managerial Psychology,21,200–210. https://doi.org/10.1108/02683940
Wooldridge, J. M. (2003). Cluster-sample methods in applied econometrics. American Economic
Review,93, 133–138. https://doi.org/10.1257/000282803321946930
Received 18 October 2017; revised version received 29 March 2018
16 Kim Peters and S. Alexander Haslam