You are on page 1of 30

Journal of Management Scientific Reports

Vol. 1 No. 1, February 2023 42-71


DOI: 10.1177/27550311231152901
© The Author(s) 2023
Article reuse guidelines:
sagepub.com/journals-permissions

Supervisor’s organizational embodiment is leader


group prototypicality: Addressing construct
redundancy through replication
Daan van Knippenberg
Rice University, Houston, TX, USA
Diana Lee
Drexel University, Philadelphia, PA, USA

The core concept in the social identity theory of leadership is leader group prototypicality (LGP),
the perception that the leader is representative of a shared group (team, organization) identity.
The theory was proposed in the 1990s and inspired a body of research that established the theory
as well-supported. We argue that the 2010 construct of supervisor’s organizational embodiment
(SOE) captures the same construct as LGP. The use of different labels for the same construct is
an impediment to the development of broad-ranging theory. It is therefore important to empiri-
cally establish whether SOE can indeed be seen as another name for LGP. We address this issue
through replication tests, establishing that core LGP findings and core SOE findings replicate
with both LGP and SOE measurement, as well as by showing that LGP and SOE items represent
the same underlying factor. We discuss how recognizing that SOE is another name for LGP
allows for the integration of two separate streams of research. We also propose referring to
the construct of interest exclusively as LGP and forgoing the SOE label that was proposed
after LGP was already established in the literature.

Keywords: Leadership; social identity; leader group prototypicality; supervisor’s


organizational embodiment

Corresponding Author: Daan van Knippenberg, Jones Graduate School of Business, Rice University, 6100 Main
Street, Houston, TX 77005, USA.
Email: dvanknippenberg@rice.edu

42
van Knippenberg and Lee / Supervisor’s organizational embodiment 43

In the 1990s and 2000s, the social identity theory of leadership established itself as a per-
spective in leadership research that is unique in putting follower perceptions of the leader
through the lens of shared social identity center stage (Hains et al., 1997; Hogg, 2001;
Platow & van Knippenberg, 2001; van Knippenberg, 2011; van Knippenberg & Hogg,
2003). Core to the theory is the concept of leader group prototypicality, the extent to
which the leader is perceived to be representative of shared group identity (Hogg, 2001;
i.e., is similar to the group prototype, the mental representation of the group; Rosch, 1978;
Turner et al., 1987). The group, in the definition of leader group prototypicality, is understood
to broadly refer to any social grouping that individuals can perceive themselves to be a
member of—organizations, teams, societal groups, etc.—and is not limited to the traditional
group dynamics understanding in which group exclusively refers to small interacting groups
(Turner et al., 1987). In a related vein, leader in the definition of leader group prototypicality
is used to broadly refer to any individual in a leadership position regardless of whether they
are called leader, manager, supervisor, or something else. The social identity theory of lead-
ership posits that leader group prototypicality is an influence on leadership effectiveness
because social identity (self-conception based on group membership) reflects shared social
reality, including shared values, norms, and aspirations (i.e., what “we” value, see as norma-
tive, and aspire to). Because shared values, norms, and aspirations guide behavior, group pro-
totypical leaders derive influence from the perception that they represent these values, norms,
and aspirations. The social identity theory of leadership has been well-supported in survey
and experimental research across a range of countries and study settings with both attitudinal
and behavioral indicators of leadership effectiveness, as also recently established in meta-
analysis (Steffens et al., 2021).
Drawing on a tradition of studying the employee–organization relationship from a social
exchange perspective (e.g., Rhoades & Eisenberger, 2002), in 2010, Eisenberger et al. pro-
posed the concept of supervisor’s organizational embodiment (SOE), which they defined as
members’ perception of their supervisor’s shared identity with the organization—the extent
to which the supervisor shares the organization’s characteristics. The overlap between the
concept of leader group prototypicality (LGP) and the concept of SOE is evident when
we compare their definitions as being representative of shared group/organizational identity,
especially when we consider that the social identity theory of leadership was explicitly
proposed to apply to all social groups, including organizations, and to all individuals in
leadership positions, including supervisors. SOE can thus be understood to be one specific
application of the LGP concept—leader/supervisor prototypicality of the organization—that
is fully covered by the conceptualization of LGP. This is something that Eisenberger et al.
may not have realized because the LGP naming convention does not refer to the organiza-
tion. Even when social identity theory of leadership publications do conceptually and empir-
ically use LGP to include LGP of the organization, this may have led Eisenberger et al. to
not recognize the concept redundancy. The reason why SOE researchers did not recognize
the concept redundancy with LGP is not our concern, however; our focus is on addressing
the construct redundancy itself.
The use of different labels for the same construct, as we see here for LGP and SOE, can
result in streams of research that develop in isolation and as a consequence forego possibilities
44 Journal of Management Scientific Reports / February 2023

for integration (Block, 1995). This is exactly what happened in the case of LGP and SOE (van
Knippenberg & Dwertmann, 2022). Integration of research streams where possible is essen-
tial to the development of a broader-ranging theory, regardless of whether it is to resolve
inconsistencies or to recognize the broader applicability and usefulness of a theoretical per-
spective (Cronin et al., 2021). Thus, proposing and studying SOE without recognizing that
it is another label for LGP (albeit limited to prototypicality of the organization) is problematic
for theory development, because it leads the field to forego opportunities for integration of
these research streams. Establishing that SOE is another label for LGP would thus be impor-
tant in bringing these streams of research together to develop a broader-ranging theory. It
would integrate theory and evidence from SOE research into the broader and more established
body of the social identity theory of leadership. Moreover, it would provide a bridge between
the social identity theory perspective that lies at the basis of the social identity theory of lead-
ership (Hogg, 2001) and the social exchange theory perspective that lies at the basis of
research on SOE (Eisenberger et al., 2010). Studying the construct redundancy of the SOE
concept with the LGP concept may thus also contribute to the development of integrative
theory bridging social identity and social exchange perspectives beyond the study of LGP
and SOE. Empirically establishing whether there is ground for such integration is the contri-
bution we aim to make with the current study. We do so through replication tests to determine
whether core LGP findings and core SOE findings replicate with both LGP and SOE measure-
ment and by investigating whether LGP and SOE measurement is better viewed as represent-
ing the same underlying factor in principal components analysis (PCA) and confirmatory
factor analysis (CFA).

Theoretical background and hypotheses


Considering the definitions of LGP and SOE in and of itself supports the conclusion that
LGP and SOE refer to the same construct. This was the conclusion also drawn in recent
reviews that treated SOE evidence as concerning LGP (Steffens et al., 2021; van
Knippenberg & Dwertmann, 2022). Putting that conclusion on firmer empirical ground
requires complementing these earlier considerations of the conceptual definitions of LGP
and SOE with empirical evidence. We are not aware of any study assessing both concepts
to determine their measurement overlap. Moreover, the relationships tested in LGP and
SOE research are different and as a result, we cannot base conclusions about concept redun-
dancy on the similarity of findings. Both types of evidence are important and complementary
in assessing construct redundancy empirically—which in turn is important in complementing
the conceptual assessment of construct redundancy.
Thus, the core aim of the current study is to empirically address the LGP–SOE construct redun-
dancy issue that can be identified on the basis of construct definitions. To do so, we focus on testing
whether LGP findings replicate with SOE measurement and SOE findings replicate with LGP mea-
surement, as well as whether principal components analysis (PCA) and confirmatory factor anal-
ysis (CFA) suggest that LGP and SOE items are more appropriately considered to reflect one
factor. We argue that triangulation by considering both findings from replication tests and from
PCA and CFA is important. A first reason for this is that any shortcoming in operationalization
(an issue we address in the following when we discuss measurement validity concerns with the
original SOE scale) could result in the conclusion that operationalizations do not fall on one
van Knippenberg and Lee / Supervisor’s organizational embodiment 45

factor, but not because of concept uniqueness but because of measurement validity problems. A
second reason is that PCA and CFA evidence only speaks by implication to the issue of
whether SOE findings replicate with LGP measurement and vice versa, which is another key con-
sideration in integrating evidence from both research streams.
In the current study, we, therefore, see PCA and CFA evidence only as one indicator in the
empirical consideration of construct overlap and put some emphasis on replication tests. The
focus in these replication tests is whether conclusions are the same regardless of the measure
used (i.e., LGP vs. SOE). If, as we propose, SOE is another label for LGP, we should replicate
key findings from LGP and SOE research with measures of both LGP and SOE. Conversely,
if LGP and SOE represent different constructs, findings may diverge, for instance, such that
an LGP measure is more likely to replicate key findings from LGP research and an SOE
measure is more likely to replicate key findings from SOE research. Measurement imperfec-
tions can also be an issue here, but complementing the PCA–CFA approach with the replica-
tion approach allows us to triangulate evidence for stronger conclusions. What we consider
next, therefore, are core theory and findings from LGP and SOE research that are well-suited
for our replication approach to construct redundancy.

Core relationships in the social identity theory of leadership


Social identity refers to self-definition in terms of group membership, where group is under-
stood to generically refer to any social grouping one can perceive oneself to be a part of (Tajfel &
Turner, 1979). Accordingly, social identity may reflect demographic characteristics such as
nationality, race, or gender, social groupings such as sports fan affiliations or friendship
groups, or professional and organizational groupings such as a work team or employing orga-
nization (Hogg, 2003). For research in organizational behavior, these job-related aspects of
social identity are particularly relevant. Because social identifications guide behavior, organi-
zational identifications are helpful concepts in understanding organizational behavior
(Ashforth & Mael, 1989; van Knippenberg & Hogg, 2018). In this respect, it is important to
note that these notions of social identity and social identification may concern different
group memberships (e.g., team identification, organizational identification) but reflect the
same underlying construct (i.e., social identification; Turner et al., 1987). Accordingly, differ-
ent social identifications can be measured with the same scale items by adapting the group that is
being referred to, as van Knippenberg and van Schie (2000) for instance did to measure orga-
nizational and workgroup identification (cf. Riketta & Van Dick, 2005).
Social identities are mentally represented as group prototypes, knowledge structures that
capture one’s subjective understanding of what defines the group—what makes the group
a group, what group members have in common, and what distinguishes the group from
other groups (Turner et al., 1987). Such group prototypes capture what the group values, con-
sider normatively appropriate, and aspire too (Steffens et al., 2021). Social identification thus
in effect implies thinking of the self in terms of the group prototype and to internalizing group
values, norms, and aspirations. Through social identification, group prototypes thus become
sources of influence in that identification leads group members to treat group values, norms,
and aspirations as behavioral guides. A direct extension of this is the notion that group
members differ in how prototypical they are—in the extent to which they are seen to
reflect group prototypical characteristics—and that more group prototypical members can
46 Journal of Management Scientific Reports / February 2023

be more influential by virtue of being perceived to represent shared group values, norms, and
aspirations (Turner et al., 1987). Group prototypes in essence are mental representations of
social identity and thus can be applied to any membership group—team, organization, demo-
graphic group, etc.—without changing the meaning of the concept (Turner et al., 1987).
The social identity theory of leadership is a development of these notions of identity, group
prototypes, and group prototypicality to understand how shared group membership provides
leaders with a basis of influence to the extent that they are perceived to embody the shared iden-
tity based in that group membership—are perceived to be group prototypical. The foundational
proposition in the theory is that leaders are more influential the higher their leader group pro-
totypicality (LGP), because the perception of embodying shared group identity, and thus
shared values, norms, and aspirations, increases follower openness to the leader’s influence.
Because notions of social identity, social identification, and group prototypes are explicitly con-
ceptualized to apply to all membership groups, group prototypicality too is a term that applies to
all social groups. Accordingly, whereas the majority of empirical studies of LGP has focused on
LGP of small groups such as teams, the LGP notion applies much more broadly and equally
applies to organizations (Giessner et al., 2009; Sluss et al., 2012; Sluss & Ashforth, 2008;
Ullrich et al., 2009; van Knippenberg, 2016, 2020; van Knippenberg et al., 2008; van
Knippenberg & Hogg, 2003), and other larger groupings such as the students of a university
(Platow et al., 2006; Platow & van Knippenberg, 2001).
Notions of leader group prototypicality are closely tied to the concept of social identifica-
tion, and a first focus for replication is the moderating role of group identification that has
been a core element in the social identity theory of leadership from its conception (Hains
et al., 1997; Hogg, 2001; Platow & van Knippenberg, 2001). Because core to the social iden-
tity theory of leadership is the notion that LGP derives its influence from shared social iden-
tity, a key proposition is that only to the extent that followers identify with the group, team, or
organization (i.e., derive their identity from the group membership) is LGP influential. This
moderating effect of follower social identification has been established across experimental
and survey research (for a review, see van Knippenberg, 2011) as well as meta-analytical
(Steffens et al., 2021).
Because the moderating role of group identification lies at the heart of the social identity
theory of leadership, it is an important effect for a replication test with an SOE measure.
Social identity theory of leadership research has relied on a variety of attitudinal and behav-
ioral indicators of leadership effectiveness, with frequent reliance on leadership evaluations
such as perceived leadership effectiveness, which also seems an appropriate criterion for rep-
lication tests here (van Knippenberg, 2011). Because SOE is specifically defined in reference
to the organization, this test would focus on the organization as the reference point for iden-
tification and LGP.
Core to our argument is the proposition that SOE is another name for LGP and not another
construct. We, therefore, refer to LGP and SOE to differentiate their literatures and measure-
ment and not to suggest these labels represent different constructs. Our formal hypothesis
statements, however, should not be separate statements for LGP and SOE because formal
statements of hypotheses concerning relationship derived from theory are understood to
concern relationships between constructs rather than measures. Accordingly, in hypotheses
statements, we refer to “LGP–SOE” to reflect that we make identical predictions for both
van Knippenberg and Lee / Supervisor’s organizational embodiment 47

measures that we argue reflect the same construct. For the moderating role of organizational
identification in the relationship with perceived leadership effectiveness, we thus posit:

Hypothesis 1: The positive relationship of LGP–SOE with leadership effectiveness is stronger


with higher follower organizational identification.

The second foundational moderation effect identified in the social identity theory of leader-
ship concerns the interaction of LGP and instantiations of group-serving behavior, leader behav-
ior in the service of group interests (in which LGP is seen as the moderator; van Knippenberg &
Hogg, 2003). In the social identity theory of leadership, an important notion is that LGP imbues
trust in the leader’s group-serving motivation and that this trust makes followers more open to the
leader’s influence (Giessner & van Knippenberg, 2008; Platow & van Knippenberg, 2001; van
Knippenberg & van Knippenberg, 2005). Leaders may also build such trust by engaging in
group-serving behavior, such as making personal sacrifices in the service of the group (i.e.,
which is perceived to follow from group-serving motivation because it comes at personal
costs; van Knippenberg & Hogg, 2003). The social identity theory of leadership proposes that
LGP instills trust in the leader’s group-serving motivation and that there are diminishing
returns on such trustworthiness signals, such that when trust is established further signals of trust-
worthiness are less influential. LGP is therefore proposed to substitute for the influence of group-
serving behavior, rendering leadership effectiveness relatively high independent of leader group-
serving behavior. This moderation effect has been established in experimental as well as survey
research for a variety of operationalizations of group-serving behavior (van Knippenberg, 2011).
Recently, it has also been established in meta-analysis (Steffens et al., 2021). We focus here on
leader self-sacrifice to capture leader-group serving behavior, because the LGP × self-sacrifice
interaction was established across four samples using experiments and surveys and in that
sense probably is the empirically most strongly established instantiation of this interaction
effect (van Knippenberg & van Knippenberg, 2005). For this foundational moderation effect
too, a replication test with SOE measurement is valuable.

Hypothesis 2: LGP–SOE substitutes for the influence of leader-self-sacrifice such that the positive
relationship of leader self-sacrifice with leadership effectiveness is weaker with higher LGP–SOE.

Core relationships in the study of SOE


The roots of SOE research are in the study of employee relationships from the perspective
of social exchange theory (Homans, 1961). In the social exchange analysis of organizational
behavior, relationships are understood to be guided by the norm of reciprocity (Gouldner,
1960). High-quality relationships are those that are characterized by reciprocal exchange of
valued material and immaterial goods, such that parties in the relationship perceive they
derive value from the relationship and that the give and take in the relationship is fair (i.e.,
balanced in its reciprocity). In organizational behavior, there is a long-standing tradition to
use this social exchange perspective to understand the relationship between leaders and fol-
lowers—primarily through the construct of leader–member exchange (LMX; Graen &
Scandura, 1987)—and employees’ relationship with their employing organization—probably
most prominently through the concept of perceived organizational support (Eisenberger et al.,
48 Journal of Management Scientific Reports / February 2023

1986). Whereas the focus is on different relationships (leader–follower, employee–organiza-


tion), the underlying theoretical notion is the same. The more individuals experience their
relationship as being of high quality (as understood in social exchange theory), the more indi-
viduals will display attitudes and behavior that are positive contributions to the relationship—
and positive outcomes from the perspective of organizational behavior (Graen & Uhl-Bien,
1995; Rhoades & Eisenberger, 2002).
One set of such outcomes links the leader–follower relationship to the employee–organi-
zation relationship and this has been the core focus of SOE research. In the social exchange
analysis, leaders are seen as representatives of the organization. Accordingly, one’s relation-
ship with one’s leader reflects on one’s psychological relationship with the organization. A
better relationship with the leader (higher LMX) is associated with a better relationship with
the organization, as captured through such concepts as organizational commitment, organi-
zational identification, and perceived organizational support. These notions were the starting
point for Eisenberger et al. (2010) to argue that LMX is positively related to employee orga-
nizational commitment and that this relationship should be stronger the more the leader is
perceived to represent the organization. That is, if the relationship between LMX and
organizational commitment is caused by the perception that the leader represents the
organization, the relationship should be stronger the more the leader is perceived to be rep-
resentative of the organization. To capture variations in the degree to which the leader
(supervisor in Eisenberger et al.’s 2010 terminology) is perceived to be representative of
the organization, they introduced the concept of supervisor’s organizational embodiment
(SOE).
From the notion that employees may take their relationship with their leader as indicative
of their relationship with the organization, Eisenberger et al. (2010) argued that SOE
strengthens the relationship between LMX and organizational commitment. The more the
leader is perceived to represent the organization, the more one’s relationship with the
leader is related to one’s psychological relationship with the organization. Eisenberger
et al. (2010) established this interactive influence of LMX and SOE on organizational com-
mitment. Later studies following the same logic established it for perceived organizational
support (Eisenberger et al., 2014) and organizational identification (Hussain & Shahzad,
2018). Thus, given how closely tied the LMX by SOE interaction on attitudes toward the
organization is to the study of SOE, replication tests with both LPG and SOE measures of
the interaction with LMX on organizational commitment, perceived organizational
support, and organizational identification is valuable in addressing the issue of concept
redundancy.

Hypothesis 3a: The positive relationship of LMX with organizational commitment is stronger
with higher LGP–SOE.

Hypothesis 3b: The positive relationship of LMX with perceived organizational support is stron-
ger with higher LGP–SOE.

Hypothesis 3c: The positive relationship of LMX with organizational identification is stronger
with higher LGP–SOE.
van Knippenberg and Lee / Supervisor’s organizational embodiment 49

Table 1
Principal components analysis with one vs two factors

Two factors
One
factor 1 2

Leader Group Prototypicality (van Knippenberg & van Knippenberg, 2005)


My supervisor resembles the members of my organization .80 .65 .21
My supervisor is very similar to the members of my organization .81 .85 −.01
My supervisor represents what is characteristics about my organization .82 .89 −.04
My supervisor has very much in common with the members of my organization .76 .27 .58
My supervisor is a good example of the kind of people that are members of my .84 .81 .06
organization
My supervisor embodies the norms of my organization .84 .75 .14
Supervisor’s Organizational Embodiment (Eisenberger et al., 2010)
When my supervisor encourages me, I believe that my organization is encouraging me .78 −.00 .90
When my supervisor is pleased with my work, I feel that my organization is pleased .76 −.09 .97
When my supervisor compliments me, it is the same as my organization complimenting .80 .07 .84
me
When my supervisor pays attention to my efforts, I believe that my organization is paying .80 .05 .87
attention to my efforts
My supervisor is characteristic of my organization .84 .88 −.02
My supervisor and my organization have a lot in common .77 .92 −.14
When I am evaluated by my supervisor, it is the same as being evaluated by my .74 .76 .01
organization
My supervisor is representative of my organization .81 .79 .05
My supervisor is typical of my organization .83 .81 .05

Entries are component loadings. Loadings for the two-factor solution are after OBLIMIN rotation.

Measurement considerations
The impression that SOE is another name for LGP is enhanced by the high degree of sim-
ilarity between the survey items used to measure LGP and SOE. Considering the van
Knippenberg and van Knippenberg (2005) LGP measure, which in its 6-item form (as per
their study 4) or in shortened forms is the most frequently used LGP measure (Steffens
et al., 2021) and Eisenberger et al.’s SOE measure, there is a striking similarity between
items (these LGP–SOE items can be found in Table 1). Compare for instance the van
Knippenberg and van Knippenberg items (here adapted to refer to supervisor and organiza-
tion, as per the social identity theory of leadership practice to adapt terminology to the
group of interest, including the organization; e.g., Giessner et al., 2009), “My supervisor rep-
resents what is characteristics about my organization” and “My supervisor has very much in
common with the members of my organization,” with the Eisenberger et al. items, “My super-
visor is characteristic of my organization” and “My supervisor and my organization have a lot
in common.”
The Eisenberger et al. (2010) SOE scale also contains items that raise validity concerns
from the perspective that they arguably concern attributed or hypothetical effects of SOE
more than SOE itself, such as “When my supervisor encourages me, I believe that my
50 Journal of Management Scientific Reports / February 2023

organization is encouraging me.” Such confounding of the leadership concept of interests


with its attributed or hypothetical effects constitutes a measurement validity problem (van
Knippenberg & Sitkin, 2013) that in and of itself does not reflect on the conceptual consid-
eration of the redundancy of the SOE label with the LGP concept. It does reflect on its empir-
ical consideration, however, in that any shortcoming in the SOE measure can reduce the
strength of the evidence in favor of construct redundancy for reasons of measurement validity
rather than construct uniqueness. We therefore also take these measurement issues into
account as we develop our empirical analysis.
Possibly in recognition of this measurement validity issue, some studies of SOE dropped the
items concerning attributed/hypothetical consequences (labeling these items the “shared experi-
ence” items; Eisenberger et al., 2014; Shoss et al., 2013) and used the items that have the greatest
similarity with the LGP items, arguing that these are more core to the SOE construct. We therefore
also explored the results of the SOE scale without the attributed/hypothetical consequences items.
Based on concept definitions and operationalizations, our expectation is that the LGP and SOE
measures are very highly correlated and that a one-factor solution for the items of LGP and SOE
scales better describes the data than a two-factor solution differentiating LGP and SOE scales.
Probability testing of correlations speaks to the existence of a relationship and not to whether
the correlation is high enough to constitute evidence of construct redundancy, however, and the
validity concern with the “hypothetical consequences items” of the SOE scale suggests that the
fit of the one-factor solution may be compromised for other reasons than construct uniqueness
(a reflection of the more general issue that PCA and CFA cannot differentiate construct uniqueness
from measurement validity problems). For these reasons, we refrained from stating a formal
hypothesis here and approached the correlational and PCA/CFA evidence from the perspective
that its interpretation is a judgment call more than a matter of probability testing.

Methods
Sample and procedure
Our focus was on replication of findings that for the LMX by SOE interaction have only been
established with single-source data (i.e., whereas the Eisenberger et al., 2010, 2014 studies
included multisource data, the tests of interest here were based on single-source percept-percept
data) and that for the LGP by identification and LGP by self-sacrifice interactions have also been
established with single source survey data (in addition to experimental data). Whereas single-
source data are not ideal and potentially problematic for some purposes, single-source effects
cannot explain the interactions in regression that are the focus of our replication tests (Evans,
1985). Moreover, the key consideration is whether the same results are found with LGP and
SOE measures and not whether we have the strongest possible evidence for any given relation-
ship. Therefore, single-source data sufficed for the current purposes.
In 2022, we used Prolific Academic’s online panel service to recruit a sample of full-time adult
working individuals fluent in English, drawn from a wide variety of countries across five conti-
nents. Following the payment guidelines of Prolific Academic, participants were paid $1.50 for
their participation. As a data quality control, we included two attention checks in the survey (e.g.,
“click ‘Strongly disagree’ for this question”). Sixteen respondents out of our initial sample of 405
did not pass either of the attention checks, and their data were removed from the analysis. From
van Knippenberg and Lee / Supervisor’s organizational embodiment 51

the remaining respondents, 195 identified their gender as male, 193 as female, and 1 as other. To
be able to include gender as a control variable, this forced us to drop the “other” category, leaving
a final sample of N = 388 respondents for the replication tests.

Measures
Because our focus was on testing whether findings from studies identifying the concept of
interest as LGP replicate with a measure associated with the SOE label and vice versa, we
adopted measurement from the studies we sought to replicate in order to ensure that a possible
failure to replicate could not be attributed to the use of different measures than the original
studies. Except for age and gender, all responses were assessed on 7-point scales ranging
from strongly disagree to strongly agree. The scales used here are included in Appendix A.

LGP–SOE. As per Steffens et al. (2021), the most frequently used measure of LGP is van
Knippenberg and van Knippenberg (2005) 6-item measure or shortened versions thereof (e.g.,
Platow & van Knippenberg, 2001). We, therefore, used the 6-item measure for LGP van
Knippenberg and van Knippenberg (2005) used in their survey test of the LGP × leader self-
sacrifice interaction on perceived leadership effectiveness (note that the original items were in
Dutch and that there may be different but functionally equivalent English translations of these
items available; cf. Steffens et al., 2021). Consistent with SOE measurement terminology, as
well as with LGP theory and research practice that suggests adapting how the items refer to
the leader and the group to the study context, LGP items referred to “supervisor” and the
“organization.” A sample item is, “My supervisor represents what is characteristics about
my organization.” For SOE, we used the 9-item measure from the Eisenberger et al.
(2010) study we sought to replicate in the LMX × SOE interaction on organizational commit-
ment, which is the original source of this measure. A sample item is, “My supervisor is char-
acteristic of my organization.” Both scales had good reliability, with Cronbach’s α = .91 for
van Knippenberg and van Knippenberg’s LGP scale and Cronbach’s α = .93 for Eisenberger
et al.’s SOE scale.

LMX. We used the 12-item LMX measure from Liden and Maslyn (1998) deployed by
Eisenberger et al. (2010). A sample item is, “I like my supervisor very much as a person.”
This measure had good reliability, Cronbach’s α = .95.

Leader self-sacrifice. To measure leader self-sacrifice, we used the 5-item measure from
van Knippenberg and van Knippenberg (2005). A sample item is “My supervisor makes per-
sonal sacrifices in the team’s interest.” The scale had good reliability, Cronbach’s α = .92.

Organizational identification. We measured organizational identification with the 6-item


measure from Mael and Ashforth (1992), which has been used in survey research establishing
the moderating role of organizational identification in LGP effects (Cicero et al., 2008; Pierro
et al., 2007) and as the criterion for the LMX × SOE interaction in Hussain and Shahzad
(2018). A sample item is, “When I talk about my organization, I usually say ‘we’ rather
than ‘they’.” This measure had good reliability, Cronbach’s α = .92.
52 Journal of Management Scientific Reports / February 2023

Leadership effectiveness. We measured perceived leadership effectiveness with van


Knippenberg and van Knippenberg’s (2005) 5-item scale. A sample item is, “My supervisor
is a good leader.” Cronbach’s α = .96, indicating that reliability was good for this scale.

Organizational commitment. Following Eisenberger et al. (2010), we assessed organiza-


tional commitment with the 6-item affective organizational commitment scale from Meyer
and Allen (1997). A sample item is, “My organization has a great deal of personal
meaning for me.” Cronbach’s α = .88 for this scale, indicating good reliability.

Perceived organizational support. Following Eisenberger et al. (2014), we assessed per-


ceived organizational support with an 8-item measure from Eisenberger et al. (1986). A
sample item is, “Help is available from the organization when I have a problem.”
Reliability for this measure was good, Cronbach’s α = 80.

Controls. We added sex (0 = female, 1 = male) and age as frequently used controls in the
studies we build on. Typically, controls would be added based on conceptual consider-
ations. Here, we deviate from that practice because our concern is less with whether the con-
trols make conceptual sense and more with creating similarity with the studies we seek to
replicate.

Results
We analyzed evidence for construct redundancy of LGP and SOE in two ways. The first
was to compare in PCA and CFA a one-factor model combining LGP and SOE items and a
two-factor model with separate LGP and SOE factors. The second was a series of regression
analysis to replicate key findings from LGP and SOE studies separately for the van
Knippenberg and van Knippenberg (2005) LGP scale and the Eisenberger et al. (2010)
SOE scale. These two types of analyses are independent in the sense that the results of
the PCAs and CFAs were expected to lead to the conclusion that a one-factor solution is
better than a two-factor solution, but the replication tests should be conducted separately
for LGP and SOE measures regardless of the conclusions from the PCA and CFA. All anal-
yses were conducted using SPSS with the exception of the CFAs which were conducted
with LISREL.

Comparing one-factor and two-factor models


Because, as noted in the Introduction, a consideration of item content validity suggested
that the SOE measure consists of a set of items very similar to the items of the LGP scale
and a more problematic set of items confounding SOE and its attributed or hypothetical
effects, we preceded the CFA with a PCA to explore factor loadings for one-factor and
two-factor solutions. We expected a one-factor model to be better, but we reasoned that if
two factors would emerge, they would be correlated. Therefore, in PCA, we compared a one-
factor solution with a two-factor solution with OBLIMIN rotation, allowing the factors to be
correlated.
van Knippenberg and Lee / Supervisor’s organizational embodiment 53

The one-factor solution explained 64% of the variance and, as can be seen from Table 1, resulted
in high component scores for all items (all > .70), giving another indication that the LGP and SOE
scales represent the same underlying factor. The alternative two-factor solution is associated with
only an additional 8% of variance explained and does not separate LGP and SOE scales. Rather, as
per Table 1, it primarily separates the more problematic SOE items that confound SOE and its
attributed or hypothetical effects from the items that more purely assess LGP–SOE. Put differently,
to the extent that PCA would suggest a two-factor solution, it is not because LGP and SOE scales
assess separate constructs, but because the SOE scale could separate into two factors.
Next, we conducted CFA to compare the one-factor model with a two-factor model sep-
arating LGP and SOE scales. Results of this analysis show that the fit of the one-factor model,
χ2(90) = 1108.19, RMSEA = .17, CFI = .83, is moderate and only marginally better than that
of the two-factor model, χ2(89) = 1168.61, RMSEA = .18, CFI = .82. Importantly, however,
as per the PCA, it is clear that the fact that the fit of the one-factor model is suboptimal and
only modestly better than the two-factor model is primarily due to the duality of the SOE scale
and not to any separation of LGP and SOE scales.
Subsequently, we analyzed LGP and SOE items without the SOE attributed/hypothetical
consequence items (cf. Eisenberger et al., 2014; Shoss et al., 2013). PCA of the LGP items
and the remaining four SOE items yielded only one factor with an eigenvalue > 1, explaining
69% of the variance, and with component loadings all > .72. A forced two-factor solution
with OBLIMIN rotation still yielded a solution with all items loading higher on the first
factor. Testing the fit of the one-factor model in CFA showed a better fit than with the hypo-
thetical consequence items included, χ2(35) = 166.35, RMSEA = .098, CFI = .96. We can
thus conclude that not including the items that are associated with measurement validity
issues, following considerations of measurement validity as well as more recent SOE mea-
surement practice (Eisenberger et al., 2014; Shoss et al., 2013), supported the conclusion
that LGP and SOE items represent the same factor.
Thus, neither PCA nor CFA supported the conclusion that a two-factor solution separating
LGP and SOE scales would be better than a one-factor solution; PCA and CFA result in com-
bination supported the conclusion that LGP and SOE scales measure the same construct more
than that LGP and SOE scales would measure different constructs. These conclusions are
unequivocal when excluding the SOE items that are associated with measurement validity
concerns; they come with some caveats regarding these problematic items when they are
included. Indeed, complementing the item content validity concern raised earlier, results
also suggest that future studies should consider excluding the items from the SOE scale
that confound LGP–SOE with its effects (as per Eisenberger et al., 2014; Shoss et al.,
2013). Whereas the latter issue is beyond the scope of the current analysis, we nevertheless
need to put some emphasis on it here to be very explicit about the fact that the less-than-ideal
fit of the one-factor solution does in no way suggest that LGP and SOE are different con-
structs but rather that there is a measurement issue with the SOE scale. (In supplementary
analyses below, we explore the hypothetical consequences items a bit more.)

Correlations and descriptive statistics


Table 2 displays means, standard deviations, and intercorrelations for all study variables.
The most noteworthy correlation here is the one between the van Knippenberg and van
54
Table 2
Means, standard deviations, and intercorrelations of study variables

Mean SD 1 2 3 4 5 6 7 8 9

1. Sex .51 .50


2. Age 31.19 8.52 −.10
3. Leader group prototypicality 5.07 1.24 .05 −.13**
4. Supervisor’s organizational embodiment 5.11 1.18 .05 −.13** .92***
5. Leader self-sacrifice 4.56 1.42 .02 −.09 .32*** .36***
6. Organizational identification 4.29 1.51 .13** .07 .43*** .51*** .38***
7. Leader–member exchange 5.21 1.21 .07 −.12** .38*** .44*** .77*** .42***
8. Organizational commitment 4.01 1.37 .10* .04 .48*** .54*** .39*** .72*** .48***
9. Perceived organizational support 4.56 1.05 .01 −.10 .63*** .67*** .53*** .60*** .57*** .64***
10. Leadership effectiveness 5.26 1.47 .08 −.14** .39*** .46*** .77*** .42*** .90*** .51*** .61***

N = 388 listwise. Sex is coded 0 = female, 1 = male.


* p < .05, ** p < .01, *** p < .001
van Knippenberg and Lee / Supervisor’s organizational embodiment 55

Knippenberg (2005) LGP scale and the Eisenberger et al. (2010) SOE scale (i.e., the original
nine-item scale; we used the full SOE scale here despite the measurement concerns to preempt
the question whether the problematic items may be what differentiates SOE from LGP). The r
= .92 here offers another clear indication that these are two measures of the same construct.
Not surprisingly given the very strong correlation, correlations of these two measures with
other constructs are highly similar, further pointing to the issue of construct redundancy.
We also note that some of the other correlations are potentially problematic from a measure-
ment perspective, but less of a concern here in that measurement inflation of correlations between
independent variable and dependent variable makes it more difficult to detect interactions
(Evans, 1985). That is, support for our hypotheses cannot be explained by inflated correlations.
Moreover, the highest correlation here is the r = .90 for LMX and leadership effectiveness,
which are constructs that are not involved in the same replication test. Thus, even though
such a high correlation suggests a measurement validity problem (likely due to the LMX
measure confounding relationship quality with leadership evaluations; Gottfredson et al.,
2020), it is not a problem that bears on our replication tests (i.e., our question is whether we
can replicate SOE findings for this LMX measure used by Eisenberger et al., 2010, with both
LGP and SOE scales and not whether this LMX measure is a good measure).

Replication tests
For the replication tests, the main question was whether using the SOE scale would result in
the same conclusions as using the LGP scale. All regression analyses for these tests used stan-
dardized scores for the interaction components and age and gender as controls. Table 3 displays
the result of these analyses, with the first set of regression statistics concerning results for the
LGP scale and the second set for the SOE scale (again, note that we used the full SOE scale here
despite the measurement concerns, to preempt the question whether the problematic items may
be what differentiates SOE from LGP). In order, Table 3 displays results for tests of LGP–SOE
interaction with organizational identification (Hypothesis 1) and with leader self-sacrifice
(Hypothesis 2) on leadership effectiveness, and LGP–SOE × LMX interactions on organiza-
tional commitment (Hypothesis 3a), perceived organizational support (Hypothesis 3b), and
organizational identification (Hypothesis 3c). All these interactions were significant with
highly similar regression weights for LGP and SOE scales.
As per recommendations by Aiken and West (1991), we computed simple slopes for one
SD above and below the mean of the moderating variable to further determine the nature of
these interactions. Following the studies we draw on for these replication tests, we treat orga-
nizational identification as the moderator for the LGP–SOE × organizational identification
interaction and LGP–SOE as the moderator for the interactions with leader self-sacrifice
and LMX. Table 4 provides the statistics for the simple slope tests and Figures 1a–5b graph-
ically represent the interactions with simple slopes for high and low values of the moderator.
In principle, it is a judgment call whether the similarity of findings is high enough to conclude
that they corroborate the conclusion that the LGP and SOE scales represent the same construct.
As can be seen in Tables 3 and 4 and Figures 1a–5b, however, findings were so similar that it is
safe to conclude that the use of LGP and SOE scales resulted in essentially the same findings.
This is much more in line with the conclusion that SOE is another name for LGP than with the
conclusion that SOE and LGP would be different constructs. All interactions were significant
56 Journal of Management Scientific Reports / February 2023

Table 3
Results of regression analysis

LGP SOE

b SE β p b SE β p

Leadership effectiveness
Sex .00 .13 .00 .98 .03 .13 .01 .84
Age −.02 .01 −.12 .01 −.02 .01 −.11 .006
LGP–SOE .39 .07 .26 <.001 .50 .08 .34 <.001
Identification .48 .07 .32 <.001 .39 .08 .27 <.001
LGP–SOE × identification .20 .06 .15 <.001 .21 .06 .16 <.001
Leadership effectiveness
Sex .16 .09 .05 .09 .14 .09 .05 .13
Age −.01 .01 −.04 .23 −.01 .01 −.03 .34
LGP–SOE .23 .05 .16 <.001 .31 .05 .21 <.001
Self-sacrifice 1.03 .05 .70 <.001 .98 .05 .66 <.001
LGP–SOE × self-sacrifice −.18 .04 −.13 <.001 −.21 .04 −.15 <.001
Organizational commitment
Sex .18 .11 .06 .11 .19 .11 .07 .08
Age .02 .01 .13 .002 .02 .01 .12 .002
LGP–SOE .46 .06 .34 <.001 .54 .06 .39 <.001
LMX .57 .06 .41 <.001 .52 .06 .37 <.001
LGP–SOE × LMX .25 .05 .20 <.001 .24 .05 .19 <.001
Perceived organizational support
Sex −.10 .07 −.05 .15 −.09 .07 −.05 .18
Age .00 .00 .01 .88 −.00 .00 .00 .99
LGP–SOE .49 .04 .47 <.001 .52 .04 .50 <.001
LMX .46 .04 .44 <.001 .43 .04 .40 <.001
LGP–SOE × LMX .20 .03 .21 <.001 .18 .03 .18 <.001
Organizational identification
Sex .30 .13 .10 .02 .31 .12 .10 .02
Age .03 .01 .16 <.001 .03 .01 .16 <.001
LGP–SOE .47 .07 .31 <.001 .60 .07 .40 <.001
LMX .54 .07 .35 <.001 .46 .07 .30 <.001
LGP–SOE × LMX .24 .06 .17 <.001 .20 .06 .15 <.001

N = 388 listwise. Sex is coded 0 = female, 1 = male.


Note. LGP: leader group prototypicality; SOE: supervisor’s organizational embodiment; LGP–SOE: column header
indicates which of the two scales is used; LMX: leader–member exchange.

replications of the effects we set out to replicate regardless of whether we used the LGP or the
SOE scale and findings are extremely similar for both scales. The LGP–SOE relationship with
leadership effectiveness is stronger with higher identification (Hypothesis 1), leader self-
sacrifice is more strongly related to leadership effectiveness for lower LGP–SOE because
LGP–SOE substitutes for leader self-sacrifice (i.e., leads to relatively high leadership effective-
ness regardless of self-sacrifice; Hypothesis 2), and LMX is more strongly related to organiza-
tional commitment (Hypothesis 3a), perceived organizational support (Hypothesis 3b), and
organizational identification (Hypothesis 3c) with higher LGP–SOE.
van Knippenberg and Lee / Supervisor’s organizational embodiment 57

Table 4
Simple slope tests

LGP SOE

b SE p b SE p

LGP–SOE × identification → effectiveness


high identification .59 .10 <.001 .71 .11 <.001
low identification .19 .09 .03 .29 .09 <.001
LGP–SOE × self-sacrifice → effectiveness
high LGP–SOE .85 .07 <.001 .78 .07 <.001
low LGP–SOE 1.21 .06 <.001 1.19 .06 <.001
LGP–SOE × LMX → commitment
high LGP–SOE .82 .09 <.001 .76 .09 <.001
low LGP–SOE .31 .07 <.001 .28 .07 <.001
LGP–SOE × LMX → perceived support
high LGP–SOE .67 .06 <.001 .60 .06 <.001
low LGP–SOE .26 .05 <.001 .25 .05 <.001
LGP–SOE × LMX → identification
high LGP–SOE .77 .10 <.001 .66 .11 <.001
low LGP–SOE .30 .08 <.001 .26 .08 .001

N = 388 listwise.
Note. LGP: leader group prototypicality; SOE: supervisor’s organizational embodiment; LGP–SOE: column header
indicates which of the two scales is used; LMX: leader–member exchange.

Figure 1a
Interaction leader group prototypicality (LGP) × organizational identification (OID) on
leadership effectiveness
58 Journal of Management Scientific Reports / February 2023

Figure 1b
Interaction supervisor’s organizational embodiment (SOE) × organizational
identification (OID) on leadership effectiveness

Figure 2a
Interaction leader group prototypicality (LGP) × leader self-sacrifice on leadership
effectiveness
van Knippenberg and Lee / Supervisor’s organizational embodiment 59

Figure 2b
Interaction supervisor’s organizational embodiment (SOE) × leader self-sacrifice on
leadership effectiveness

Figure 3a
Interaction leader group prototypicality (LGP) × leader–member exchange (LMX) on
organizational commitment
60 Journal of Management Scientific Reports / February 2023

Figure 3b
Interaction supervisor’s organizational embodiment (SOE) × leader–member exchange
(LMX) on organizational commitment

Figure 4a
Interaction leader group prototypicality (LGP) × leader–member exchange (LMX) on
perceived organizational support
van Knippenberg and Lee / Supervisor’s organizational embodiment 61

Figure 4b
Interaction supervisor’s organizational embodiment (soe) × leader–member exchange
(lmx) on perceived organizational support

Figure 5a
Interaction leader group prototypicality (LGP) × leader–member exchange (LMX) on
organizational identification
62 Journal of Management Scientific Reports / February 2023

Figure 5b
Interaction supervisor’s organizational embodiment (SOE) × leader–member exchange
(LMX) on organizational identification

We repeated the SOE regression analysis with a shortened, four-item scale that did not
include the hypothetical consequences items. Not surprisingly, results were very similar to
those reported in Table 3: interaction with identification, b = .21; interaction with leader self-
sacrifice, b = −.19; interaction with LMX on commitment, b = .29, on perceived organiza-
tional support, b = .18, on identification, b = .26. Thus, use of the shortened SOE scale also
supported the conclusion that the LGP and SOE scales are functionally equivalent in replicat-
ing findings from both LGP and SOE research.

Supplementary analyses
Our concern in the current study is with construct redundancy in the conceptualization of
SOE vis-à-vis the already established concept of LGP. It is not with the more problematic
aspect of the SOE scale in the items that invoke attributed or hypothetical consequences
and thus confound measurement of the construct with its effects (cf. van Knippenberg &
Sitkin, 2013). Even so, for a fuller assessment of the overlap between LGP and SOE research,
there is value added in considering how these problematic items may impact replication tests.
This is the issue we addressed in supplementary analyses.
First, we conducted a PCA with the SOE items only, resulting in two factors with eigen-
values greater than 1. An OBLIMIN rotation showed that these factors represented the uncon-
founded SOE items versus the hypothetical consequences items, with all component loadings
van Knippenberg and Lee / Supervisor’s organizational embodiment 63

on the primary factor higher than .80 and all cross-loadings smaller than |.07|. Thus, there is an
empirical as well as an item content case to consider treating the SOE scale as separating into
two factors, an LGP–SOE factor and a hypothetical consequences factor.
Because the latter factor represents a measurement validity problem, there is less con-
ceptual value in exploring its behavior. For the current purposes, however, there may be
empirical value in considering how the replication tests work out for a “hypothetical con-
sequences scale” that is formed by these problematic items only. Using the same regres-
sion approach as for the hypothesis tests, all replication tests with the hypothetical
consequences scale are significant and show highly similar results as the tests with the
LGP, full SOE, and shortened SOE scales. For the hypothetical consequences, scale
regression weights for the interactions are as follows (see Table 3 for comparison): inter-
action with identification, b = .18; interaction with leader self-sacrifice, b = −.15; interac-
tion with LMX on commitment, b = .22, on perceived organizational support, b = .19, on
identification, b = .19. If anything, coefficients for the hypothetical consequences scale
tend to be smaller, but the differences are trivial. When for the moment we put aside
the measurement validity issues with the SOE scale, the main conclusion here is that rep-
lication tests with a scale based on items that are most different from LGP measurement
lead to the same conclusions as analysis based on LGP measurement. This is a conclusion
that is not too surprising given the evidence in support of including the hypothetical con-
sequences items in the same factor as the other LGP–SOE items, but it is worth noting
explicitly for anyone wondering whether the hypothetical consequences items introduce
some uniqueness to SOE measurement—they do not as far as the current analysis is
1
concerned.

Discussion
From the perspective of developing a broader-ranging theory, it is an important question
whether SOE is LGP by another name rather than a separate construct (and given that the first
SOE paper was submitted for publication when the LGP concept was already well-established
in the published literature, this is the appropriate way to frame the question). When the con-
clusion is justified that SOE findings can be interpreted as concerning LGP, this can both be
seen as an instant broadening of the scope of the social identity theory of leadership (the more
established and broader theoretical perspective on LGP) as well as a bridge between the social
identity perspective on organizational behavior and the social exchange perspective on orga-
nizational behavior (i.e., recognizing that LGP and SOE studies both have a broader-ranging
theoretical backdrop). We argued that the most relevant in this respect were replication tests
of core LGP and SOE findings with both LGP and SOE measurement. Based on the very high
intercorrelation of the LGP and SOE scales, their highly similar pattern of correlations with
other study variables, the high component loadings for the PCA with one factor, the good fit
of the one-factor solution when the more problematic hypothetical consequence items were
dropped (cf. Eisenberger et al., 2014; Shoss et al., 2013), and the strong similarity in the
results of the replication tests, the conclusion that LGP and SOE measures capture the
same construct is more strongly supported than any suggestion that they would capture dis-
tinct constructs.
64 Journal of Management Scientific Reports / February 2023

Theoretical implications
The conclusion that SOE is another name for LGP is important because it means that the
separate streams of research on LGP and SOE can be integrated as concerning the same con-
struct. Such conclusions are essential in moving our field forward in creating broader-ranging
theory and consolidating rather than proliferating constructs (Cronin et al., 2021).
The first conclusion in this respect is about the naming convention, which is more impor-
tant than it may sound because it concerns addressing the issue of construct redundancy.
Moving forward, it stands to reason to uniformly refer to LGP rather than (also) to SOE.
The leader group prototypicality construct made its entry into the literature 13 years before
the supervisor’s organizational embodiment construct and at the time it was introduced
drew on a tradition of 20 years of research on group prototypes and group member prototy-
picality (Rosch, 1978; Turner et al., 1987). Whereas many LGP studies focus on LGP of
smaller groups such as teams and work groups, theory and empirical work have been clear
that the LGP construct is not limited to small interactive groups but equally applies to
broader social groupings such as organizations. There also is a much more voluminous
body of research on LGP than on SOE (Steffens et al., 2021). The conclusion, therefore, is
that to remedy the isolation of findings created by separate naming conventions for the
same construct, research should henceforth refer to only one construct—leader group proto-
typicality (LGP)—as the concept and naming convention that was introduced earlier and has
been more widely studied.
The recognition that SOE research concerns LGP allows for an important bridging of lit-
erature. Narrowly interpreted, this means that SOE findings reported in the literature can be
interpreted as concerning LGP. This is an important extension of the social identity theory of
leadership in linking LGP more firmly to employees’ psychological relationship with the
organization (cf. Sluss et al., 2012). Interpreted more broadly, recognizing that SOE evidence
pertains to LGP also invites further development of the social identity theory of leadership.
This is illustrated by a recent review by van Knippenberg and Dwertmann (2022) that
included evidence concerning the moderating effects of LGP–SOE.
Van Knippenberg and Dwertmann concluded that when SOE evidence is interpreted as
concerning LGP, this strengthens the case to consider the different mechanisms involved
in different interaction patterns for LGP. Van Knippenberg and Dwertmann observed that
findings for the interaction of LGP with leader group-serving behaviors consistently show
that LGP substitutes for leader group-serving behavior (as also replicated in the current
study). They contrasted this with findings for LGP by leader fairness interactions. For
these, the argument has been advanced that the trust in the leader that LGP instills also sub-
stitutes for leader fairness, which has also been linked to trust in the leader. There are studies
that support this perspective across samples and methods. At the same time, the argument has
also been advanced that LGP enhances the effects of leader fairness in that the fairness of
treatment one receives conveys one’s standing with the group (team, organization), Such
effects would be stronger the more the leader is perceived to be representative of the group
—an argument akin to the rationale for the LMX × LGP–SOE interactions. This perspective
too has been supported across samples and methods. Van Knippenberg and Dwertmann con-
cluded that when SOE findings were interpreted as concerning LGP, this added to the evi-
dence that LGP may enhance the effects of other elements of leadership to the extent that
van Knippenberg and Lee / Supervisor’s organizational embodiment 65

followers are concerned with their standing with the group (i.e., as per the relationships of
LMX with commitment, identification, and organizational supports).
The current evidence that LMX × SOE interactions can be seen as LMX × LGP interac-
tions puts empirical weight behind van Knippenberg and Dwertmann’s (2022) call to deter-
mine when LGP has substituting effects that are presumably trust-based and enhancing effects
that are presumably based on the social evaluation that is conveyed about one’s standing with
the group. This is a question that is also relevant for LMX specifically (and thus for the LMX
× LGP interaction). As per the SOE studies, there is an argument that LMX is informative
about one’s standing with the leader and the group the leader represents. There is also an argu-
ment that LMX is associated with trust in the leader (Rockstuhl et al., 2012). The rationale to
predict an enhancing effect of LGP in the relationship of LMX with commitment, identifica-
tion, and perceived organizational support clearly links these interactions to the social eval-
uation conveyed by LMX. Given that both LGP and LMX are also associated with trust in the
leader and the rationale to expect a substituting interaction pattern for such indicators of trust-
worthiness, the broader evidence base in the social identity analysis of leadership thus raises
the question of when LGP has a substituting rather than enhancing moderation effect in
LMX’s relationship with outcomes (e.g., in moderating the effects of leader group-serving
behavior).
Addressing this issue of enhancing versus substituting effects of LGP would help bridge
social identity (LGP, e.g., van Knippenberg & Hogg, 2003), social exchange (LMX; e.g.,
Eisenberger et al., 2010), and organizational justice (i.e., leader fairness; e.g., van
Knippenberg et al., 2007) perspectives on leadership and thus help build toward broader-
ranging leadership theory. Social identity theory, social exchange theory, and organizational
justice theory do not only concern leadership; however; all three perspectives more broadly
concern the employee–organization relationship (Shore et al., 2012). Therefore, developing
an integrative theory to bridge these perspectives specifically for the moderating role of
LGP in the effects of LMX and leader fairness may set the stage for the development of
further integrative theory bridging these perspectives beyond the role of LGP (e.g., can the
theory that captures when to expect enhancing versus substituting effects be extended to
LMX × leader fairness interactions?), and potentially beyond the role of leadership (e.g.,
organizational justice also has a systemic aspect directly associated with the organization;
cf. perceived organizational support). Part of the value of the current study thus lies in pro-
viding a stronger basis to motivate such integrative research efforts.

Implications for researchers


In addition to the more content and theory-focused research implications captured in our
consideration of theoretical implications, we may note two methods’ implications for
researchers in addressing the issues highlighted in the previous as well as other analyses of
LGP.
A measurement conclusion from the current evidence is that LGP can be measured with
items originally associated with LGP as well as with SOE. Several of these items are so
similar that it hardly makes a difference whether researchers draw on LGP or SOE measure-
ment and the replication evidence underscores that they can be expected to be functionally
equivalent. A caveat to this conclusion is that a subset of SOE items invokes effects attributed
66 Journal of Management Scientific Reports / February 2023

to SOE. This is a measurement validity concern (van Knippenberg & Sitkin, 2013) that also
calls into question how suited the original SOE scale is as operationalization of the unidimen-
sional construct is it intended to capture. Moving forward, it thus is advisable to avoid the use
of these items, as had already become the practice in some SOE studies (Eisenberger et al.,
2014; Shoss et al., 2013).
A second issue to note is that deliberately, and we contend with good reason, we limited
ourselves in the current study to single-source survey data: the proximal outcomes of the
LMX × SOE interactions concerned single-source survey data and the LGP effects we
sought to replicate have also been established with single-source survey data (as well as
experimentally). In a related vein, we concluded that online panel data would be appropriate
for the current purposes and that limitations of online panel data (such as lack of information
about response rate and the inability to correct for the possibility that there might be obser-
vations that are nested within the same leader) are no concern for the current purposes of com-
paring results for two different measures of the same construct (i.e., replication of earlier
findings that were not based on online panel data cannot realistically be attributed to the
current use of online panel data). Moving forward in developing the integrative work that
the current study would motivate (as per the previous section), however, it would be impor-
tant to rely on higher-quality data, for example by also conducting experimental studies and
by also assessing behavioral indicators of leadership effectiveness (e.g., performance; van
Knippenberg, 2011). Ultimately, any theory is only as strong as its supporting evidence
and what sufficed for the current purposes will not suffice moving forward in developing
this integration further.

Conclusion
The approach we adopted in the current study holds that replication tests are a powerful
way to address construct redundancy in complementing more traditional PCA and CFA evi-
dence and conceptual considerations because they directly speak to the extent to which find-
ings from the one research tradition can be integrated with findings from another research
tradition. We believe the current evidence for the replication of LGP and SOE effects is a
case in point. From the perspective of the value of construct consolidation rather than prolif-
eration, we would encourage researchers not only to follow up on the potential for integrative
research we outlined but also to pursue the investigation of other instances of construct redun-
dancy through similar replication approaches.

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) received no financial support for the research, authorship, and/or publication of this article.
van Knippenberg and Lee / Supervisor’s organizational embodiment 67

Orcid id
Daan van Knippenberg https://orcid.org/0000-0002-0269-8102

Note
1 We conducted additional replication tests (i) with a scale that combines all LGP and SOE items and (ii) with a
scale that combines LGP and the shortened SOE scale items. These tests showed very similar results too that we do
not report here but that are available upon request.

References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Sage.
Ashforth, B. E., & Mael, F. (1989). Social identity theory and the organization. Acad Manage Rev 14, 20–39.
Block, J. (1995). A contrarian view of the five-factor approach to personality description. Psychol Bull 117, 187–215.
Cicero, L., Bonaiuto, M., Pierro, A., & van Knippenberg, D. (2008). Employees work effort as a function of leader
group prototypicality: The moderating role of team identification. Eur Rev Appl Psychol 58, 117–124.
Cronin, M. A., Stouten, J., & van Knippenberg, D. (2021). The theory crisis in management research: Solving the
right problem. Acad Manage Rev 46, 667–683.
Eisenberger, R., Huntington, R., Hutchison, S., & Sowa, D. (1986). Perceived organizational support. J Appl Psychol
71, 500–507.
Eisenberger, R., Karagonlar, G., Stinglhamber, F., Neves, P., Becker, T. E., Gonzalez-Morales, M. G., &
Steiger-Mueller, M. (2010). Leader-member exchange and affective organizational commitment: The contribu-
tion of supervisor’s organizational embodiment. J Appl Psychol 95, 1085–1103.
Eisenberger, R., Shoss, M. K., Karagonlar, G., Gonzalez-Morales, M. G., Wickham, R. E., & Buffardi, L. C. (2014).
The supervisor POS–LMX—subordinate POS chain: Moderation by reciprocation wariness supervisor’s orga-
nizational embodiment. J Organ Behav 35, 635–656.
Evans, M. G. (1985). A Monte Carlo study of the effects of correlated method variance in moderated regression anal-
ysis. Organ Behav Hum Decis Process 36, 305–323.
Giessner, S. R., & van Knippenberg, D. (2008). “License to fail”: Goal definition, leader group prototypicality, and
perceptions of leadership effectiveness after leader failure. Organ Behav Hum Decis Process 105, 14–35.
Giessner, S. R., van Knippenberg, D., & Sleebos, E. (2009). License to fail?: How leader group prototypicality
moderates the effects of leader performance on perceptions of leadership effectiveness. Leadersh Q 20,
434–451.
Gottfredson, R. K., Wright, S. L., & Heaphy, E. D. (2020). A critique of the leader-member exchange construct: Back
to square one. Leadersh Q 31, 101385.
Gouldner, A. W. (1960). The norm of reciprocity: A preliminary statement. Am Sociol Rev 25, 161–178.
Graen, G. B., & Scandura, T. (1987). Toward a psychology of dyadic organizing. Res Organ Behav 9, 175–208.
Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to leadership: Development of leader-member
exchange (LMX) theory of leadership over 25 years: Applying a multi-level multi-domain perspective.
Leadersh Q 6, 219–247.
Hains, S. C., Hogg, M. A., & Duck, J. M. (1997). Self-categorization and leadership: Effects of group prototypicality
and leader stereotypicality. Pers Soc Psychol Bull 23, 1087–1099.
Hogg, M. A. (2001). A social identity theory of leadership. Pers Soc Psychol Rev 5, 184–200.
Hogg, M. A. (2003). Social identity. In M. R. Leary & J. P. Tangney (Eds.), Handbook of self and identity (pp. 462–
479). Guilford.
Homans, G. (1961). Social behavior. Harcourt Brace Jovanovich.
Hussain, S., & Shahzad, K. (2018). Role of supervisor’s organizational embodiment and organizational identification
on LMX and job performance relationship: A test of moderated-mediation model. J Manage Sci 5, 18–37.
Liden, R. C., & Maslyn, J. M. (1998). Multidimensionality of leader-member exchange: An empirical assessment
through scale development. J Manag 24, 43–72.
68 Journal of Management Scientific Reports / February 2023

Mael, F., & Ashforth, B. E. (1992). Alumni and their alma mater: A partial test of the reformulated model of orga-
nizational identification. J Organ Behav 13, 103–123.
Meyer, J. P., & Allen, N. J. (1997). Commitment in the workplace. Sage.
Pierro, A., Cicero, L., Bonaiuto, M., van Knippenberg, D., & Kruglanski, A. W. (2007). Leader group prototypicality
and resistance to organizational change: The moderating role of need for closure and team identification. Test
Psychom Methodol Appl Psychol 14, 27–40.
Platow, M. J., & van Knippenberg, D. (2001). A social identity analysis of leadership endorsement: The effects
of leader ingroup prototypicality and distributive intergroup fairness. Pers Soc Psychol Bull 27,
1508–1519.
Platow, M. J., van Knippenberg, D., Haslam, S. A., van Knippenberg, B., & Spears, R. (2006). A special gift we
bestow on you for being representative of us: Considering leader charisma from a self-categorization perspec-
tive. Br J Soc Psychol 45, 303–320.
Rhoades, L., & Eisenberger, R. (2002). Perceived organizational support: A review of the literature. J Appl Psychol
87, 698–714.
Riketta, M., & Van Dick, R. (2005). Foci of attachment in organizations: a meta-analytic comparison of the strength
and correlates of workgroup versus organizational identification and commitment. J Vocat Behav 67, 490–510.
Rockstuhl, T., Dulebohn, J. H., Ang, S., & Shore, L. M. (2012). Leader-member exchange (LMX) and culture: A
meta-analysis of correlates of LMX across 23 countries. J Appl Psychol 97, 1097–1130.
Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categorization
(pp. 27–48). Erlbaum.
Shore, L. M., Coyle-Shapiro, J. A.-M., & Tetrick, L. E. (2012). The employee-organization relationship:
Applications for the 21st century. Routledge.
Shoss, M. K., Eisenberger, R., Restubog, S. L. D., & Zagencyck, T. J. (2013). Blaming the organization for abusive
supervision: The roles of perceived organizational support and supervisor’s organizational embodiment. J Appl
Psychol 98, 158–168.
Sluss, D. M., & Ashforth, B. E. (2008). How relational and organizational identification converge: Processes and con-
ditions. Organ Sci 19, 807–823.
Sluss, D. M., Ployhart, R. E., Cobb, M. G., & Ashforth, B. E. (2012). Generalizing newcomers; relational and orga-
nizational identifications: Processes and prototypicality. Acad Manage J 55, 949–975.
Steffens, N. K., Munt, K. A., van Knippenberg, D., Platow, M. J., & Haslam, S. A. (2021). Advancing the social iden-
tity theory of leadership: A meta-analytic review of leader group prototypicality. Organ Psychol Rev 11, 35–72.
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. 33–47). Brooks/Cole.
Turner, J. C., Hogg, M., Oakes, P. J., Reicher, S., & Wetherell, M. (1987). Rediscovering the social group: A self-
categorization theory. Blackwell.
Ullrich, J., Christ, O., & van Dick, R. (2009). Substitutes for procedural fairness: Prototypical leaders are endorsed
whether they are fair or not. J Appl Psychol 94, 235–244.
van Knippenberg, B., & van Knippenberg, D. (2005). Leader self-sacrifice and leadership effectiveness: The mod-
erating role of leader prototypicality. J Appl Psychol 90, 25–37.
van Knippenberg, D. (2011). Embodying who we are: Leader group prototypicality and leadership effectiveness.
Leadersh Q 22, 1078–1091.
van Knippenberg, D. (2016). Making sense of who we are: Leadership and organizational identity. In M. G. Pratt,
M. Schultz, B. E. Ashforth, & D. Ravasi (Eds.), The Oxford handbook of organizational identity (pp. 335–
349). Oxford University Press.
van Knippenberg, D. (2020). Meaning-based leadership. Organ Psychol Rev 10, 6–28.
van Knippenberg, D., De Cremer, D., & van Knippenberg, B. (2007). Leadership and fairness: The state of the art.
Eur J Work Organ Psychol 16, 113–140.
van Knippenberg, D., & Dwertmann, D. J. G. (2022). Interacting elements of leadership: Key to integration but
looking for integrative theory. J Manag 48, 1695–1723.
van Knippenberg, D., & Hogg, M. A. (2003). A social identity model of leadership effectiveness in organizations. Res
Organ Behav 25, 243–295.
van Knippenberg, D., & Hogg, M. A. (2018). Social identifications in organizational behavior. In D. L. Ferris, R.
E. Johnson, & C. Sedikides (Eds.), The self at work: Fundamental theory and research (pp. 72–90). Routledge.
van Knippenberg and Lee / Supervisor’s organizational embodiment 69

van Knippenberg, D., & Sitkin, S. B. (2013). A critical assessment of charismatic-transformational leadership
research: Back to the drawing board? Acad Manag Ann 7, 1–60.
van Knippenberg, D., van Knippenberg, B., & Bobbio, A. (2008). Leaders as agents of continuity: Self continuity and
resistance to collective change. In F. Sani (Ed.), Self-continuity: Individual and collective perspectives (pp. 175–
186). Psychology Press.
van Knippenberg, D., & van Schie, E. C. M. (2000). Foci and correlates of organizational identification. J Occup
Organ Psychol 73, 137–147.

Appendix A: Measurement Scales


Leader Group Prototypicality (van Knippenberg & van Knippenberg, 2005)
My supervisor resembles the members of my organization
My supervisor is very similar to the members of my organization
My supervisor represents what is characteristics about my organization
My supervisor has very much in common with the members of my organization
My supervisor is a good example of the kind of people that are members of my
organization
My supervisor embodies the norms of my organization

Supervisor’s Organizational Embodiment (Eisenberger et al., 2010)


When my supervisor encourages me, I believe that my organization is encouraging me
When my supervisor is pleased with my work, I feel that my organization is pleased
When my supervisor compliments me, it is the same as my organization complimenting
me
When my supervisor pays attention to my efforts, I believe that my organization is paying
attention to my efforts
My supervisor is characteristic of my organization
My supervisor and my organization have a lot in common
When I am evaluated by my supervisor, it is the same as being evaluated by my
organization
My supervisor is representative of my organization
My supervisor is typical of my organization

Leader–member exchange (Liden & Maslyn, 1998)


I like my supervisor very much as a person
70 Journal of Management Scientific Reports / February 2023

I respect my supervisor’s knowledge of and competence on the job


My supervisor would defend me to others in the organization if I made an honest mistake
I am impressed with my supervisor’s knowledge of their job
My supervisor is the kind of person one would like to have as a friend
My supervisor is a lot of fun to work with
I do not mind working my hardest for my supervisor
I admire my supervisor’s professional skills
My supervisor would come to my defense if I were “attacked” by others
I am willing to apply extra efforts, beyond those normally required to meet my supervi-
sor’s work goals
My supervisor defends my work actions to a superior, even without complete knowledge
of the issue in question
I do work for my supervisor that goes beyond what is specified in my job description

Leader self-sacrifice (van Knippenberg & van Knippenberg, 2005)

My supervisor makes personal sacrifices in the team’s interest


My supervisor stands up for the team members’ interest, even when it is at the expense of
my supervisor’s own interest
My supervisor risks their position when they believe the goals of the team can be reached
that way
My supervisor is always among the first to sacrifice free time, privileges, or comfort if that
is important for the team’s mission
I can always count on my supervisor to help me in times of trouble, even if it is at costs to
them

Organizational identification (Mael & Ashforth, 1992)

When someone criticizes my organization, it feels like a personal insult


I am very interested in what others think about my organization
When I talk about my organization, I usually say ‘we’ rather than ‘they’
My organization’s successes are my successes
When someone praises my organization, it feels like a personal compliment
If a story in the media criticized my organization, I would feel embarrassed
van Knippenberg and Lee / Supervisor’s organizational embodiment 71

Leadership effectiveness (van Knippenberg & van Knippenberg, 2005)


My supervisor is a good leader
My supervisor is very effective as a leader
My supervisor leads in a way that motivated people
I enjoy working with my supervisor
My supervisor motivates me to exert myself for my organization

Organizational commitment (Meyer & Allen, 1997)


My organization has a great deal of personal meaning for me
I do not feel emotionally attached to my organization (R)
I would be very happy to spend the rest of my career working for my organization
I do not feel like part of the family at my organization (R)
I really feel as if my organization’s problems are my own
I do not feel a strong sense of belonging to my organization

Perceived organizational support (Eisenberger et al., 1986)


Help is available from the organization when I have a problem
The organization really cares about my well-being
The organization is willing to extend itself in order to help me perform my job to the best
of my ability
Even if I did the best job possible, the organization would fail to notice (R)
The organization cares about my general satisfaction at work
The organization shows very little concern for me (R)
The organization cares about my opinions
The organization takes pride in my accomplishments at work

You might also like