Content uploaded by Angel Gómez
Author content
All content in this area was uploaded by Angel Gómez
Content may be subject to copyright.
On the Nature of Identity Fusion:
Insights Into the Construct and a New Measure
A
´ngel Go´mez
Universidad Nacional de Educacio´n a Distancia
Matthew L. Brooks and Michael D. Buhrmester
University of Texas, Austin
Alexandra Va´zquez
Universidad Nacional de Educacio´n a Distancia
Jolanda Jetten
University of Queensland
William B. Swann Jr.
University of Texas, Austin
Previous research has documented the consequences of feeling fused with a group; here we examine the
nature of identity fusion. Specifically, we sought to determine what fusion is and the mediating
mechanisms that lead fused individuals to make extraordinary sacrifices for their group. Guided by the
assumption that fusion emphasizes the extent to which people develop relational ties to the group, we
developed a measure designed to capture feelings of connectedness and reciprocal strength with the
group. In 10 studies, the newly developed scale displayed predicted relationships with related measures,
including an earlier (pictorial) measure of fusion and a measure of group identification. Also as expected,
fusion scores were independent of several measures of personality and identity. Moreover, the scale
predicted endorsement of extreme progroup behaviors with greater fidelity than did an earlier pictorial
measure of identity fusion, which was, in turn, superior to a measure of group identification. Earlier
evidence that the personal and social selves of fused persons are functionally equivalent was replicated,
and it was shown that feelings of agency and invulnerability mediated the effects of fusion on extreme
behavior. Finally, Spanish- and English-language versions of the verbal fusion scale showed similar
factor structure as well as evidence of convergent, discriminant, and predictive validity in samples of
Spaniards and Americans, as well as immigrants from 22 different countries. This work advances a new
perspective on the interplay between social and personal identity.
Keywords: identity fusion, social identity, group identification, self-verification
Although members of extremist groups have been launching
suicide attacks for centuries (see e.g., Durkheim, 1897/1951; Ped-
ahzur, 2005), such attacks continue to inspire bewilderment. Why
are people willing to sacrifice themselves to injure others? Recent
research has suggested that the construct of identity fusion may
provide one answer to this question. That is, individuals whose
responses to a pictorial measure indicate that their personal iden-
tities are “fused with” their group are especially inclined to express
willingness to sacrifice themselves for their group (see e.g.,
Swann, Go´mez, Dovidio, Hart, & Jetten, 2010; Swann, Go´ mez,
Huici, Morales, & Hixon, 2010; Swann, Go´mez, Seyle, Morales, &
Huici, 2009). But if previous research has illuminated the conse-
quences of being fused with a group, less is known about (a) the
nature of identity fusion, (b) the processes that distinguish fusion
from related psychological constructs, and (c) the mechanisms that
mediate the effects of fusion on extreme progroup behavior. To
illuminate these issues, we begin by discussing the nature of
identity fusion and how it differs from a rival conceptualization of
alignment with groups, identification.
Identity Fusion and Identification as Distinct
Measures of Alignment With Groups
One perspective on the difference between these two forms of
alignment with groups is offered by Brewer and Gardner’s (1996)
distinction between relational and collective group ties. Ties to
relational groups are based on members’ personal connections and
This article was published Online First February 28, 2011.
A
´ngel Go´ mez and Alexandra Va´zquez, Departamento de Psicologı´a
Social y de las Organizaciones, Universidad Nacional de Educacio´n a
Distancia (UNED), Madrid, Spain; Matthew L. Brooks, Michael D.
Buhrmester, and William B. Swann Jr., Department of Psychology, Uni-
versity of Texas, Austin; Jolanda Jetten, School of Psychology, University
of Queensland, Brisbane St. Lucia, Queensland, Australia.
This research and the preparation of this article were supported by
Research Fund Grant PSI2009-07008 from the Spanish Ministry of Science
and Innovation to A
´ngel Go´ mez, Matthew L. Brooks, Michael D. Buhrm-
ester, and William B. Swann Jr. We thank Ana Lisbona and Greg Hixon for
advice on the statistical analyses.
Correspondence concerning this article should be addressed to A
´ngel
Go´ mez, Departamento de Psicologı´a Social y de las Organizaciones, Fac-
ultad de Psicologı´a, UNED, Dcho 1.58 C/Juan del Rosal 10, 28040 Madrid,
Spain, or to William B. Swann Jr., UT Department of Psychology, 1
University Station A8000, 108 East Dean Keeton, Austin, TX 78712-0187.
E-mail: agomez@psi.uned.es or swann@mail.utexas.edu
Journal of Personality and Social Psychology © 2011 American Psychological Association
2011, Vol. 100, No. 5, 918–933 0022-3514/11/$12.00 DOI: 10.1037/a0022642
918
relationships with other group members (see Aron, Aron, Tudor, &
Nelson, 1991; Markus & Kitayama, 1991). In contrast, ties to
collective groups are based on members’ perception of overlap
between their own characteristics and prototypical properties of
the ingroup (e.g., shared qualities or outcomes, commitment to a
common goal). Whereas members of relational groups tend to
perceive fellow members of the group as unique and hence irre-
placeable members of a larger “family” (Brewer & Gardner,
1996), members of collective groups perceive fellow members as
categorically undifferentiated and interchangeable (see e.g., Tajfel
& Turner, 1979; Turner, Hogg, Oakes, Reicher, & Wetherell,
1987).
Although members of all societies appear to form both relational
and collective ties to groups, cultures may vary in the prevalence
of these two forms of group alignment (see e.g., Brewer & Chen,
2007; Nisbett, Peng, Choi, & Norenzayan, 2001; Yuki, 2003).
Whereas members of East Asian cultures tend to perceive their
group memberships in personal, relational terms (a “relational
orientation”), members of Western individualistic cultures tend to
perceive their group memberships in categorical terms (a “collec-
tive orientation”). We suggest that, complementing these cultural
differences in predominant orientation to relationships, within a
given culture, people’s ties to their groups may reflect one or the
other orientation. We argue further that just as measures of identity
fusion emphasize degree of relational orientation to the group,
measures of identification emphasize degree of collective orienta-
tion to the group (see also Prentice, Miller & Lightdale’s, 1994,
distinction between common-identity and common-bond groups).
The tendency for highly fused people to have a relational
orientation to the group may have important implications. Rather
than focusing on the group as a relatively abstract social category,
fused persons perceive it as a “family” consisting of members who
all share a common bond. Such familial attachment may give rise
to two key sentiments. First, it may engender a powerful sense of
connectedness wherein fused persons come to believe that they
and other group members are functionally equivalent. Second, it
may foster the perception of reciprocal strength: Just as they
believe that they will do anything for other group members, so,
too, do they believe that other group members will do the same.
The result of these perceptions of connectedness and reciprocal
strength may be a powerful desire to act on behalf of the group,
even if extreme action is required (see e.g., Allport, 1962).
Identification with a group has very different implications. Most
important, highly identified people will align themselves with the
collective while perceiving their fellow group members as cate-
gorically undifferentiated and interchangeable (see e.g., Tajfel &
Turner, 1979; Turner et al., 1987). Indeed, numerous investiga-
tions using the minimal-group paradigm have shown that when
people are assigned to groups on the basis of trivial (even explic-
itly random) criteria, they become biased toward that group—
despite never having encountered a single group member (see e.g.,
Billig & Tajfel, 1973; Turner, Sachdev, & Hogg, 1983). As such,
although identified people may display solidarity with the collec-
tive, their positive sentiments will be directed to the category and
not necessarily to its members. Furthermore, because highly iden-
tified individuals perceive group members as categorically inter-
changeable, their loyalty to the collective will not necessarily
compel them to rush to the assistance of individual members of the
collective when such individuals are imperiled.
Support for the distinction between fusion and identification
comes from evidence that scores on measures of identity fusion are
associated with endorsement of extreme behavior while control-
ling for group identification. For example, in one set of studies,
identity-fused individuals were particularly likely to endorse fight-
ing and dying for their country and appeared to equate threats to
the group with threats to the self (see e.g., Swann et al., 2009).
Moreover, in several variations of the classic trolley dilemma,
fused persons endorsed saving fellow group members by plunging
themselves in front of a speeding trolley (Swann, Go´mez, Dovidio,
et al., 2010). Fused persons appear to be especially inclined to
endorse self-sacrifice when their personal or social identities have
been activated (Swann et al., 2009) or when their feelings of
agency are amplified by physiological arousal (Swann, Go´mez,
Huici, et al., 2010). In summary, researchers have recently pro-
vided a wealth of evidence that identity fusion consistently out-
predicts identification in predicting the tendency for people to
protect fellow group members and to endorse extreme progroup
behavior.
But if identity fusion serves to “cock the progroup activity
trigger,” it is unclear what variables cause people to actually pull
the trigger. Agency may be one such variable. That is, the high
levels of connectedness and reciprocal strength that fused people
feel toward their group may compel them to tether their feelings of
personal agency to the group. Such feelings of agency should, in
turn, predispose people to take action on behalf of the group by
protecting fellow group members. In fact, recent research has
demonstrated that feelings of agency mediated the relationship of
fusion to progroup behavior (Swann, Go´mez, Huici, et al., 2010,
Studies 3– 4).
Yet if agency explains why fused people translate their progroup
sentiments into corresponding actions, it fails to specify why fused
people are especially likely to endorse extreme behaviors, actions
that most would see as risky. The perceptions of reciprocal
strength engendered by fusion may be critical here. Insofar as
fused persons believe that they and other group members syner-
gistically strengthen each other, fused persons may conclude that
they are invulnerable relative to actors who do not enjoy complete
confidence that their fellow group members will act on their
behalf. Feelings of invulnerability have been linked to the propen-
sity to engage in dangerous behavior (see e.g., Greene, Krcmar,
Walters, Rubin, & Hale, 2000; Ravert et al., 2009). Hence, fusion
may foster perceptions of invulnerability, and such perceptions
may, in turn, motivate extreme progroup behavior.
In short, our goal here is to provide a richer, more complete
account of the causal impact of fusion by determining whether the
feelings of connectedness and reciprocal strength associated with
fusion give rise to feelings of agency and invulnerability that, in
turn, combine to foster endorsement of extreme behaviors. To test
our mediational hypotheses, we determined whether (a) fusion is
related to the two mediators (agency and invulnerability), (b) the
two mediators are related to endorsement of progroup action, and
(c) controlling for the effects of the mediators eliminates the
relationship between fusion and endorsement of progroup behav-
ior.
To test this mediational model, we first developed a measure of
fusion that directly assessed the feelings of connectedness and
reciprocal strength theoretically associated with fusion. Past re-
search relied exclusively on a pictorial measure that was derived
919
ON THE NATURE OF IDENTITY FUSION
from a measurement device designed to assess attachment in close
relationships (see e.g., Aron, Aron, & Smollan, 1992; see also
Cialdini, Brown, Lewis, Luce, & Neuberg, 1997; Maner et al.,
2002). Composed of a series of pictures that represent different
degrees of overlap between the self and other, the Inclusion of
Other in Self (IOS) scale (Aron et al., 1992) was conceptualized as
a measure of the degree to which people possess a “sense of being
interconnected with another” that entails a tendency to view the
self as “including resources, perspectives, and characteristics of
the other” (p. 598). Several group researchers (Coats, Smith,
Claypool, & Banner, 2000; Smith & Henry, 1996; Tropp &
Wright, 2001) adapted the IOS to capture alignment of respondents
with groups. Building on this work, Schubert and Otten (2002)
added an option in which the self and group were completely
overlapping. Swann et al. (2009) further modified this measure by
creating a scale in which participants selected from among five
pictures the one that best represented their relationship with the
group. Scores on the scale were distributed bimodally, with
“fused” persons selecting the most extreme option in which the
circle representing the “self” was completely immersed in the
larger circle representing the “group” and nonfused persons select-
ing the other four options (for a discussion of the psychometric
properties of the fusion scale, see Swann et al., 2009).
Yet if the strong track record of the pictorial measure of fusion
seems to support the adage that “a picture is worth a thousand
words,” it remains unclear precisely which “words” participants
have in mind when they endorse the fused option. To be sure,
participants’ informal accounts (Swann et al., 2009) generally
supported the notion that the state of fusion truly reflects the
connectedness and reciprocal strength constructs that are thought
to underlie fusion. That said, the validity of retrospective reports
have been challenged (see e.g., Nisbett & Wilson, 1977). A more
compelling way to establish that connectedness and reciprocal
strength actually do underlie the state of identity fusion would be
to devise an instrument that specifically focuses on these con-
structs. To this end, we developed and validated a verbal measure
of identity fusion.
In the process of validating our new measure of identity fusion,
we addressed three shortcomings of earlier research on this topic.
First, all of the research has been conducted with Spanish nationals
residing in Spain, raising the possibility that fusion effects might
be restricted to members of this group. Second, the pictorial
measure that has been employed in earlier fusion research consists
of a single item only, and single-item scales are notoriously unre-
liable in many applications (see e.g., Churchill, 1979; Guilford,
1954; Nunnally, 1978). Third, because scores on the pictorial
measure have been distributed bimodally when completed by
Spanish nationals, it was necessary to treat it as a dichotomous
index. In general, dichotomous indices are less powerful than
continuous ones (see e.g., Nunnally, 1978), and in the all-too-
common instances in which endorsement of the fused option is low
(e.g., fusion with the United States is typically in the vicinity of
20%), recruitment of sufficient numbers of fused participants is
challenging (in this instance and hereafter, fused participants re-
fers to endorsers of the “E” option on the pictorial measure of
fusion). More generally, the bimodal distribution could be an
artifact of how participants construed the layout of the pictorial
options; if so, then scores on a verbal measure of identity fusion,
or even the scores of non-Spaniards on the pictorial measure,
might be distributed normally. The development of a verbal mea-
sure of fusion may therefore not only add a new assessment device
to the methodological toolbox of future fusion researchers but it
may inform our conceptualization of fusion.
Overview of Studies
To deepen our understanding of the antecedents of extreme
progroup behavior, we sought to learn more about the nature of
identity fusion and simultaneously address the methodological and
psychometric shortcomings associated with the pictorial measure
of fusion. To these ends, we developed and validated a verbal
measure of fusion. The measure was designed to tap both of the
core features of identity fusion: (a) perception of connectedness
with the group and (b) reciprocal strength. Upon creating an initial
pool of items for the scale, we asked both American and Spanish
participants to complete it and identify items that they perceived to
be confusing (Studies 1a and 1b). We then had another sample of
participants complete the scale to evaluate its psychometric prop-
erties (Study 1c). Factor structure and internal consistency were
assessed, and items that diminished coherence of the scale were
deleted. To assess temporal stability, we had another sample of
participants complete the scale once and then again 6 months later
(Study 2).
Two studies (Studies 3a and 3b) were conducted to assess
discriminant validity. Most important, we tested the hypothesis
that fusion is related to, but distinct from, group identification, the
standard measure of alignment with groups. To further assess
discriminant validity and examine convergent validity, in Study 4
we covaried the verbal fusion scale with the earlier pictorial
measure as well as several other scales that we did, or did not,
expect to be related to the newly developed verbal fusion measure.
We then conducted a series of investigations of predictive
validity. In Study 5, we determined whether fusion assessed at one
point in time predicted endorsement of fighting and dying 6
months later. The next two studies in this series (6a and 6b)
investigated the responses of Spaniards to interpersonal variations
of the trolley dilemma. Two additional studies (7a and 7b) focused
on the willingness of immigrants to Spain to fight and die for their
country of origin. Finally, the last study in this series (8) examined
the willingness of American participants to fight and die for their
country.
The last two studies focused on the nomological validity of the
verbal fusion scale. In Study 9, we asked whether the relationship
between fusion and extreme actions for the group might be medi-
ated by feelings of agency and invulnerability. In Study 10, we
sought to replicate evidence for the functional equivalence as-
sumption of the identity fusion formulation (Swann et al., 2009,
Experiment 1 and 2): Does challenging a personal self-view am-
plify endorsement of extreme behavior for the group among
strongly fused individuals but not among weakly fused persons?
Scale Construction and Psychometric Analysis:
Studies 1–2
In the first series of studies, independent samples of individuals
responded in parallel to a series of 13 items that we believed might
tap identity fusion. For each item, respondents indicated the extent
to which they felt that the statement reflected their relationship
920 GO
´MEZ ET AL.
with their country on a scale ranging from 0 (strongly disagree) to
6(strongly agree). Participants in Studies 1a, 1c, and 2 were
Spanish undergraduates at the Universidad Nacional de Educacio´n
a Distancia (UNED) in Madrid, Spain, who completed the study on
the web. Participants in Study 1b were Americans (predominantly
nonstudents) who completed the study through the website Me-
chanical Turk (for a discussion of properties of this participant
population, see Buhrmester, Kwang, & Gosling, in press).
Studies 1a–1c were designed to identify items to be included in
the scale. Studies 1a and 1b were specifically designed to deter-
mine whether all items made sense to participants. To this end, we
had participants complete the scale and indicate which, if any, of
the original 13 items struck them as ambiguous or confusing.
Cognizant that cultural and linguistic differences might influence
perceptions, after creating and back-translating Spanish- and
English-language versions of the scale, we had separate samples of
433 Spaniards (72% women; mean age !32.45 years, SD !
10.28) and 357 Americans (67% women; mean age !34.79, SD !
12.23) complete the scale in their native language. Each sample
identified one item as confusing, leading us to designate both items
for deletion from all subsequent analyses.
Study 1c was designed to prune from the scale any items that
failed to capture a unitary fusion construct. To this end, 1,981
Spanish undergraduates (72% women; mean age !31.64, SD !
9.51) completed the 13 original items. After deletion of the two
items that were identified as confusing in Studies 1a and 1b, we
conducted a principle components factor analysis on the remaining
11 items. Three factors emerged, with factor one accounting for
more variance than did the other factors (Factor 1 !40.12%,
Factor 2 !15.96%, and Factor 3 !9.73%). We accordingly
assumed that Factor 1 best captured the fusion construct and
subsequently deleted four items because they failed to load above
.50 on Factor 1 or loaded ".50 on one of the other two factors.
To test whether the remaining seven items captured a unitary
construct, we conducted two confirmatory factor analyses (CFAs)
using Analysis of Moment Structures (AMOS; Arbuckle, 1997). A
one-factor model revealed a good fit, with the fit indices exceeding
the .930 benchmark (comparative fit index [CFI] !.989, normed
fit index [NFI] !.987, goodness-of-fit index [GFI] !.991) and
the residual index falling below the .08 benchmark (root-mean-
square error of approximation [RMSEA] !.053). Nevertheless,
mindful that the items we designed to capture feelings of connect-
edness (Items 1– 4) and reciprocal strength (Items 4 –7) might load
on different factors, we compared the fit of a one-factor model
with the fit of a two-factor (connectedness and reciprocal strength)
model. The fit for the one-factor model was superior to the one for
a two-factor model (CFI !.971, NFI !.969, GFI !.977) and the
residual index (RMSEA !.083). Note that the results of a
Shapiro–Wilk test (W!.976, p#.001) indicated that scores on
the measure of verbal fusion were normally distributed (this was
also true of all 10 studies in the article).
In short, the final seven-item scale contained a single factor that
accounted for 51.60% of the variance and had a coefficient alpha
of .84. The final items and factor loadings are displayed in Table
1. Note that the correlation between the verbal and pictorial mea-
sure in Study 1c was substantial, r(1979) !.58, p#.001.
Nevertheless, despite this sizeable correlation, in all regression
analyses the variance inflation factor was always lower than 10
(i.e., !2.35), ruling out concerns regarding multicollinearity, and
in none of the studies reported in this article did the correlation
between the verbal and pictorial measures of fusion exceed .68.
These correlations, as well as the vital statistics for all 10 studies,
can be found in Table 2.
Study 2 focused on the temporal stability of the verbal fusion
scale. Spanish undergraduates (N!652) completed indices of
identity fusion measured verbally and pictorially and a measure of
identification (Mael & Ashforth, 1992; items are included in Table
3). Six months later, 620 participants from the original sample
(73% women; mean age !32.64, SD !9.14) completed all
measures again. The test–retest correlation for the verbal fusion
scale was respectable, r(618) !.71, p#.001. This stability index
exceeded the stability of the pictorial measure of fusion, r(618) !
.56, p#.001, z!4.16, p#.001, which in turn exceeded the
stability of the identification scale, r(618) !.44, z!2.82, p#.01.
Moreover, alphas for the fusion scale were .82 and .87 at Time 1
and Time 2, respectively. The alphas for the identification scale
were slightly lower: .78 and .76 at Time 1 and Time 2, respec-
tively. It thus appears that the verbal fusion scale is equivalent to
the identification scale in internal consistency but has higher
temporal stability than both the identification and pictorial mea-
sure of fusion.
Discriminant and Convergent Validity: Studies 3– 4
To date, the “gold standard” measure of alignment with groups
has been group identification. For this reason, it is crucial to
demonstrate that the new verbal fusion scale is distinct from extant
measures of identification. In selecting a measure of identification
as a standard of comparison, we searched for the identification
scale that seemed most likely to compete successfully with our
measure of fusion. In an initial effort to identify this scale, Swann
et al. (2009) correlated the pictorial fusion measure with three
measures of identification from these sources: Jetten, Branscombe,
Schmitt, and Spears (2001); Mael and Ashforth (1992); and Tropp
and Wright (2001). The results indicated that Mael and Ashforth’s
scale was more strongly associated with fusion, r(198) !.56, p#
.01, than was Jetten et al.’s scale, r(112) !.26, or Tropp and
Table 1
Factor Loadings for Items in the Spanish- and English-
Language Versions of the Verbal Fusion Scale Used in Studies
1c and 8
Item
Study 1c
(Spanish sample)
Study 8
(U.S. sample)
I am one with my country. .787 .722
I feel immersed in my country. .696 .616
I have a deep emotional bond
with my country.
.741 .794
My country is me. .655 .778
I$ll do for my country more
than any of the other group
members would do.
.707 .851
I am strong because of my
country.
.744 .757
I make my country strong. .690 .682
Note. These factor loadings are based on a principle-axis factor analysis
with a single-factor solution specified.
921
ON THE NATURE OF IDENTITY FUSION
Wright’s scale, r(248) !.23. More recently, Swann, Go´mez,
Huici, et al. (2010) showed that when it comes to predicting
endorsement of the extreme progroup behaviors that have been the
focus of fusion research, no measure of identification outper-
formed the fusion measure, but Mael and Ashforth’s (1992) mea-
sure of group identification was a stronger predictor of progroup
behavior than was a recently developed scale created by Leach et
al. (2008). Considered together, these findings suggest that if there
is one identification scale that is likely to compete successfully
with our fusion measure in predicting progroup behavior, it is Mael
and Ashforth’s. We accordingly used Mael and Ashforth’s scale as
the representative identification scale.
Studies 3a and 3b were designed to determine whether identity
fusion is distinct from identification. In Study 3a, 1,000 Spanish
undergraduates (68% women; mean age !35.33, SD !9.73)
completed the verbal fusion (%!.83) and identification (%!.88)
scales. We then submitted these responses to an exploratory factor
analysis. As shown in Table 3, two factors emerged, with the first
factor including the seven items from the verbal fusion scale and
a second factor including the six items from the identification
scale.
To verify the factor structure and test our assumption that fusion
and identification were distinct constructs, in Study 3b we asked
889 Spanish undergraduate students (75% women; mean age !
Table 3
Summary of Exploratory Factor Analysis for Identity Fusion and Identification for Study 3a Using Maximum Likelihood Estimation
(N !1,000)
Item
Factor loadings
Identity fusion Identification
a
I am one with my country. .78 .22
I feel immersed in my country. .59 .23
I have a deep emotional bond with my country. .65 .10
My country is me. .72 .04
I$ll do for my country more than any of the other group members would do. .64 .38
I am strong because of my country. .62 .37
I make my country strong. .60 .29
When someone criticizes my country, it feels like a personal insult. .27 .78
I am very interested in what citizens of others countries think about my country. .28 .59
When I talk about my country, I usually say “we” rather than “they.” .38 .65
Successes of my country are my successes. .25 .80
When someone praises my country, it feels like a personal compliment. .32 .83
If a story in the media criticized my country, I would feel embarrassed. .33 .65
Eigenvalues 3.97 3.43
% of variance 30.54 26.37
Note. Factors loading over .40 appear in bold, and items in italics are from Mael and Ashforth’s (1992) scale.
a
Mael & Ashforth (1992).
Table 2
Summary of the Studies
Study N
% Fused
(pictorial scale) Participants
Fusion pictorial and
fusion verbal correlation
Identity fusion
(verbal) Identification
a
tM SD M SD
1a 433 35 Spaniards .54
!!!
2.13 1.23 3.27 1.28 13.35
!!!
1b 357 27 Americans .43
!!!
2.12 1.32 3.65 1.15 15.63
!!!
1c 1,981 27 Spaniards .58
!!!
2.03 1.13 3.08 1.22 16.75
!!!
2 32 Spaniards 27.52
!!!
Time 1 652 .61
!!!
2.01 1.07 3.25 1.21
Time 2 620 .62
!!!
1.97 1.22 3.10 1.23
3a 1,000 34 Spaniards .55
!!!
2.02 1.07 3.37 .97 35.92
!!!
3b 889 33 Spaniards .57
!!!
2.08 1.34 3.57 1.25 35.81
!!!
6a 92 38 Spaniards .61
!!!
2.70 1.52 3.56 1.34 6.06
!!!
6b 93 33.3 Spaniards .59
!!!
2.63 1.67 3.48 1.37 7.06
!!!
7a 79 19 Immigrants .46
!!!
2.06 1.28 3.47 1.30 10.13
!!!
7b 37 19 Immigrants .63
!!!
2.10 1.21 3.50 1.24 8.11
!!!
10 137 29 Spaniards .68
!!!
2.52 1.55 3.37 1.44 8.36
!!!
Note. Studies 4, 5, 8 and 9 do not appear in the table because the data would be identical to those of the other studies that used the same sample.
Specifically, Study 4 used the Study 1c sample, Study 5 used the Study 2 sample, Study 8 used the Study 1b sample, and Study 9 used the Study 1c sample.
a
Mael & Ashforth (1992).
!!!
p#.001.
922 GO
´MEZ ET AL.
27.91, SD !7.66) to complete the verbal fusion scale (%!.84)
and the identification index (%!.81). We conducted a CFA using
AMOS to test whether the two-factor model (seven-item fusion
factor and six-item identification factor) generated by the explor-
atory analysis could be confirmed in a new data set. Items were
permitted to load on only the components they were expected to
load on, and no item errors were permitted to correlate. The
two-factor model displayed in Figure 1 revealed a good fit, with
the fit indices exceeding the .930 benchmark (CFI !.962, NFI !
.950, GFI !.962) and the residual index falling below the .08
benchmark (RMSEA !.056). The two-factor model presents a
better fit than an alternative model in which all the items load on
a single latent variable (CFI !.911, NFI !.899, GFI !.919) and
the residual index falling over the .08 benchmark (RMSEA !
.088).
Having shown that verbal fusion is distinct from identification,
in Study 4 we proceeded to examine the relationship of the fusion
scale to potentially related constructs. We expected that, on the
basis of the assumption that feelings of connectedness motivate
fused people to tether their feelings of agency to progroup behav-
ior, identity fusion would be associated with agency for the group.
In addition, we expected that, on the basis of the assumption that
fused people believe that the self and group synergistically
strengthen one another, identity fusion would be associated with a
perception of invulnerability. At the same time, we hoped to rule
out the possibility that fusion merely tapped a personality trait,
such as self-efficacy, empathy, aggressiveness, self-concept clar-
ity, or essentialism.
To address these issues, we asked 1,981 Spaniards (whose
responses to the fusion scale were also used for the factor analysis
conducted in Study 1c) to complete measures of each of the
potentially related constructs discussed in the foregoing paragraph.
With the one exception described in the next paragraph, partici-
pants indicated their responses on 7-point scales ranging from 0
(totally disagree)to6(totally agree).
For the measure of agency, we used five items based on
Haggard and Tsakiris’s (2009) discussion of the agency con-
struct (see also Swann, Go´ mez, Huici, et al., 2010). With
reference to Spain, participants rated their agreement with the
following: “I am able to control what my group does,” “I am
able to control what my group does in the same way that I
control what I do,” “I usually feel responsible for what my
group does,” “I am responsible for my group’s actions,” and “I
feel as responsible for what my group does as for what I do.”
These items formed a cohesive scale (%!.79).
For the measure of invulnerability, we developed a five-item
scale. Participants indicated their degree of agreement with the
following items: “In the face of danger, I am convinced that my
group and I will survive,” “Nothing bad can happen to me or my
group,” “Anything could damage me or my group” (reverse-
scored), “My group is less vulnerable than most other groups,” and
“My group will be able to cope with any sort of threat.” The scale
was internally consistent (%!.73).
We assessed self-concept clarity because it seemed possible
that identity fusion might be especially high among people who
enjoy high levels of self-concept clarity. To assess self-concept
clarity, we had participants complete the seven-item scale de-
veloped by Campbell et al. (1996). Like self-concept clarity,
individual differences in self-efficacy might also encourage
people to fuse with their group. To test this possibility, we used
Jerusalem and Schwarzer’s (1992) six-item scale. We specu-
lated that empathy might foster the strong feelings of group
alignment inherent in fusion. We accordingly assessed empathy
using a shortened 12-item version of Davis’s (1980, 1983)
measure.
Theoretically, individuals with a more essentialized view of
people (e.g., one that holds that differences are rooted in biological
and genetic differences) should be more sensitive to social cate-
gorization and group differences, and this could contribute to
fusion. We accordingly assessed essentialism using a four-item
scale developed by Bastian and Haslam (2006). Individual differ-
ences in aggressiveness could also be responsible for the enhanced
agency and strength that characterizes fusion. For the measure of
aggressiveness, we used the nine-item scale originally developed
by Buss and Perry (1992) and later adapted to the Spanish popu-
lation by Andreu, Pen˜a, and Gran˜ a (2002).
The results in Table 4 largely supported our predictions. That
is, verbal fusion was independent of individual differences in
self-concept clarity, self-efficacy, empathy, aggressiveness, and
essentialism. This finding is important because it suggests that
the temporal stability of scores on the verbal fusion measure
does not reflect a tendency for one of the foregoing traits to
masquerade as fusion. In contrast, the measure of verbal fusion
appears to be especially effective in capturing feelings of
agency and invulnerability, feelings that we believed would
mediate the relationship of fusion to endorsement of extreme
behavior. Before testing this possibility (see Study 9), we
examine the predictive validity of the verbal fusion scale.
Figure 1. Confirmatory factor analysis for Study 3b, including the seven-
item scale of identity fusion and the six-item scale from Mael and Ashforth
(1992). The disattenuated correlation between fusion and identification is
significantly lower than 1 ( p#.001).
923
ON THE NATURE OF IDENTITY FUSION
Predictive Validity: Studies 5– 8
In this series of studies, we compared the predictive validity of
the verbal fusion scale with that of the pictorial fusion and iden-
tification scales. The criterion variables were the ones featured in
most past research on identity fusion: endorsement of extreme
progroup actions (i.e., fighting and dying for the group; Swann,
Go´mez, Huici, et al., 2010; Swann et al., 2009) and willingness to
sacrifice one’s life in interpersonal variations of the trolley di-
lemma (see e.g., Swann, Go´mez, Dovidio, et al., 2010). We began
by determining whether the scale was predictive of endorsement of
extreme progroup action over a 6-month period (Study 5). We then
explored its capacity to predict responses to the trolley dilemma in
Spain (Studies 6a and 6b). To determine whether fusion effects
generalized to members of other cultures, we asked whether the
scale could predict the responses of immigrants to Spain from 22
different countries both contemporaneously (Study 7a) and after a
6-month delay (Study 7b). Finally, to generalize our findings to
speakers of English as well as Spanish, we assessed the factor
structure, internal consistency, and convergent and discriminant
validity of an English-language version of the scale and then
assessed its predictive validity with American participants
(Study 8).
Study 5. Does the Verbal Fusion Scale Predict
Endorsement of Fighting and Dying for
the Group a Half-Year Later?
To test this possibility, we asked 620 Spanish undergraduates
(whose responses were also examined in Study 2) to complete
measures of our three predictors. Six months later, we assessed
their willingness to fight and die for their group. For the measure
of willingness to fight for the group, on 7-point scales ranging from
–3 (totally disagree)to3(totally agree), participants rated their
agreement with five items (e.g., “I would fight someone physically
threatening another Spaniard”). For the measure of willingness to
die for the group, participants indicated their agreement with two
items (e.g., “I would sacrifice my life if it saved another group
member’s life”). Because past research has shown that the mea-
sures of willingness to fight and die are conceptually overlapping
and highly correlated (Swann et al., 2009), we combined them into
a single index dubbed endorsement of extreme progroup behaviors
(%!.81).
To determine how effectively each of the three predictor vari-
ables predicted endorsement of extreme behavior, we ran a mul-
tiple regression in which the predictors were pictorial fusion (ef-
fect coding: –1, 1) and verbal fusion and identification (both
centered). As can be seen in the top row of Table 5, the verbal
fusion measure and the pictorial fusion measure were stronger
predictors than was identification (zs"1.84, ps#.05), with more
fusion leading to more endorsement of extreme behavior. No
difference was produced between the pictorial and the verbal
measures of fusion ( p".22).
One could argue that the verbal fusion scale was such a strong
predictor of the outcome measure (endorsement of extreme pro-
group behavior) due to overlap in the content of the items in the
predictor and outcome measure. Although this seems unlikely
given that the fusion measure focuses on feelings about the group
and the outcome measure focuses on behavioral intentions, we
nevertheless put this possibility to empirical test. First, we con-
ducted an exploratory factor analysis (N!889) that included all
items from the measures of verbal fusion and endorsement of
extreme behavior. The results indicated that the predictor and
outcome measures tapped into two different factors, the first factor
including the seven items from the verbal measure (34.96% of
variance), and the second factor including the seven items in the
endorsement of extreme behavior index (16.09% of variance).
Second, we conducted a CFA using AMOS (N!1,000) to
determine whether the two-factor model (seven-item fusion factor
and seven-item endorsement of extreme behavior factor) generated
by the exploratory analysis could be confirmed in a new data set.
Items were permitted to load on only the components they were
expected to load on, and no item errors were permitted to correlate.
The results confirmed the two-factor model, revealing a good fit,
with the fit indices exceeding the .930 benchmark (CFI !.955,
NFI !.947, GFI !.936) and the residual index falling below
the .08 benchmark (RMSEA !.074). The two-factor model
presented a better fit than did an alternative model in which all
the items loaded on a single latent variable (CFI !.857, NFI !
.846, GFI !.870) and the residual index fell over the .08
benchmark (RMSEA !.108).
Studies 6a and 6b: Does the Verbal Fusion Scale
Predict Endorsement of Self-Sacrifice Among
Spaniards in the Trolley Dilemma?
Studies 6a and 6b are based on two distinct interpersonal vari-
ations of the “footbridge dilemma” developed by Swann, Go´mez,
Dovidio, et al. (2010, p. 1178). In Study 6a, Spanish undergradu-
ates (N!92; 57.6% women; mean age !33.88, SD !6.88)
participated voluntarily in a study that involved responses to moral
dilemmas. In counterbalanced order, participants completed the
measures of pictorial fusion, verbal fusion (%!.89), and identi-
fication (%!.80), all with reference to the group Spain. Each
participant then received a variation of the footbridge dilemma in
which a runaway trolley would crush and kill five ingroup mem-
bers (e.g., Spaniards) unless the participant jumped from the bridge
into the trolley’s path. Each participant then chose between letting
the trolley crush the five ingroup members and sparing them by
sacrificing her or his own life.
Table 4
Convergent and Discriminant Validity in Study 4
Construct
Fusion
Identification
a
Verbal Pictorial
Related
Feelings of agency .413
!!!
.236
!!!
.098
!!!
Invulnerability .393
!!!
.265
!!!
.102
!!!
Unrelated
Self-concept clarity &.045
!
.040 &.110
!!!
Self-efficacy .170
!!!
.118
!!!
.091
!!!
Empathy .058
!
.043 .062
!!
Aggressiveness .014 .064
!!
&.023
Essentialism .130
!!
.050
!
.120
!!
a
Mael & Ashforth (1992).
!
p#.05.
!!
p#.01.
!!!
p#.001.
924 GO
´MEZ ET AL.
To determine whether our predictor variables were associated
with responses to the trolley dilemma, we used a binary logistic
regression with three predictors: pictorial fusion (effect coding: –1,
1) and verbal fusion and identification (both centered). The out-
come measure was whether participants endorsed jumping to their
death versus allowing the trolley to crush five ingroup members.
As can be seen in Table 5, the verbal fusion measure was a
stronger predictor than was either pictorial fusion or identification,
with more fusion leading to more willingness to sacrifice oneself.
In addition, the second row of the right column reveals that
pictorial fusion was a stronger predictor than was identification. As
in previous investigations of this dilemma (Swann, Go´mez, Dovi-
dio, et al., 2010), most fused persons (74%) chose to sacrifice
themselves and most nonfused persons (79%) chose to let the
trolley kill their fellow Spaniards.
The results therefore confirmed our hypothesis that the verbal
measure of fusion would predict participants’ responses to the
trolley dilemma in much the same manner that the pictorial mea-
sure did. As in the earlier research, the measure of identification
failed to predict participants’ responses. To determine whether the
effect of the verbal fusion measure would generalize to another
variation of the trolley dilemma, we conducted a follow-up inves-
tigation.
In Study 6b, Spanish undergraduates (N!93; 50.5% women;
mean age !34.09, SD !6.83) participated voluntarily in a study
that involved responses to moral dilemmas. In counterbalanced
order, participants completed the measures of pictorial fusion,
verbal fusion (%!.92), and identification (%!.84), all with
reference to the group Spain.
Participants were asked to imagine that it was March 11, 2004,
the day when terrorists detonated several bombs in the Atocha
trolley station in Madrid. They were standing on a footbridge over
several tracks to the station just after the bombs exploded. Sud-
denly they saw the terrorists running on one set of tracks below.
Beside them, another Spaniard was preparing to jump down into
the path of an approaching trolley. The Spaniard knew that he
would die but also knew that the approaching trolley would avoid
him by veering onto the track where the terrorists were running,
killing them. The participant was then given the option of either
allowing the other Spaniard to jump or pushing him aside and
jumping to his or her own death, causing the trolley to divert and
kill the terrorists.
To determine whether our predictor variables were associated
with responses to the trolley dilemma, we used a binary logistic
regression to test whether pictorial fusion (effect coding: –1, 1) and
verbal fusion and identification (both centered) predicted the ex-
tent to which participants endorsed allowing the other Spaniard to
jump or pushing him aside and jumping. As can be seen in the third
row of the left and middle columns of Table 5, the verbal fusion
measure was a stronger predictor than was either pictorial fusion or
identification, with more fusion leading to more willingness to
sacrifice oneself. In addition, pictorial fusion was a stronger pre-
dictor than was identification. As in previous investigations of this
dilemma (Swann, Go´mez, Dovidio, et al., 2010), most fused per-
sons (69%) chose to sacrifice themselves and most nonfused
persons (89%) chose to let the fellow Spaniard jump.
The results therefore confirmed our hypothesis that the verbal
measure of fusion would predict participants’ responses to the
trolley dilemma in much the same manner that the pictorial mea-
sure did. As in the earlier research, the measure of identification
failed to predict participants’ responses.
Study 7a: Does the Verbal Fusion Scale Display
Criterion Validity Among Immigrants to Spain?
To address this question, we tested 79 students at UNED (84.8%
women; mean age !31.05, SD !7.74) who had immigrated from
22 different countries (approximately 50% were from South Amer-
ica, 30% were from Eastern Europe, and 20% were from Western
Europe). The students had been residing in Spain for 3–5 years.
When nationality or time in Spain were entered into the analyses,
no effects of either variable emerged.
All completed the three predictors (i.e., %s!.83 for the verbal
measure of fusion and .67 for identification). The referent for the
measures of fusion, identification, and endorsement of extreme
progroup behavior (%!.77) was always their country of origin.
To assess how strongly each of the three predictor variables was
associated with endorsement of extreme behavior, we ran a mul-
Table 5
Summary of Regressions, With Standardized Betas and Raw Correlations for Predictors of Endorsement of Extreme Behavior
Study
Verbal measure of fusion Pictorial measure of fusion Identification
a
B b sr B b sr B b sr
5 0.24
!!!
.25
!!!
.16
!!!
0.20
!!!
.19
!!!
.13
!!!
0.10
!!
.11
!!
.09
!!
7a 0.38
!!!
.46
!!!
.39
!!!
0.22
†
.20
†
.18
†
&0.06 &.07 &.06
7b 0.33
!!
.45
!!
.35
!!
0.41
!
.39
!
.37
!
0.01 .01 .02
8 0.41
!!!
.48
!!!
.33
!!!
0.14
!
.11
!
.10
!
0.18
!
.13
!
.09
!
9 0.40
!!!
.42
!!!
.34
!!!
0.04 .04 .03 &0.04 &.04 &.04
10 0.66
!!!
.51
!!
.34
!!
0.22
!
.15
!
.13 0.13 .10 .07
Wald OR Wald OR Wald OR
6a 1.23
!!
8.08
!!
3.42
!!
0.64
!
4.87
!
1.90
!
0.01 0.01 1.01
6b 1.47
!
4.66
!
4.36
!
1.39
!
3.99
!
4.03
!
&0.10 0.05 0.90
Note. B !raw regression coefficient; b!standardized regression coefficient; sr !semipartial correlation; OR !odds ratio.
a
Mael & Ashforth (1992).
†
p#.06.
!
p#.05.
!!
p#.01.
!!!
p#.001.
925
ON THE NATURE OF IDENTITY FUSION
tiple regression in which the three predictors were entered simul-
taneously. As can be seen in Table 5, the verbal fusion measure
was a stronger predictor than was either pictorial fusion or iden-
tification (zs"1.81, ps#.05), with more fusion leading to more
endorsement of extreme behavior. In addition, pictorial fusion was
a stronger predictor than was identification (z!1.68, p#.05).
Study 7b: Does the Verbal Fusion Scale Display
Predictive Validity Among Immigrants to Spain a
Half-Year Later?
To test this possibility, we asked 37 immigrants (86.5% women;
mean age !30.86, SD !8.39) from 16 different countries to
complete measures of our three predictors.
In counterbalanced order, participants completed the measures
of pictorial fusion, verbal fusion (%!.82), and identification (%!
.69). Six months later, participants completed the measure of
endorsement of extreme behavior (%!.84).
To assess how strongly each of the three predictor variables was
associated with endorsement of extreme behavior, we ran a mul-
tiple regression in which each of our predictors was entered
simultaneously. As can be seen in Table 5, the verbal and the
pictorial fusion measures were stronger predictors than was iden-
tification (zs"1.66, ps#.05), with more fusion leading to more
endorsement of extreme behavior.
Study 8: Does the Verbal Fusion Scale Display Sound
Psychometric Properties and Predictive Validity
Among American Participants?
To cross-validate our measure of verbal fusion among English
speakers, we had one sample of Americans complete the English
language version of the fusion scale developed in Study 1c. Par-
ticipants volunteered to participate via Amazon’s Mechanical Turk
data collection platform for 10 cents (see Buhrmester et al., in
press, for evidence of the platform’s validity at low payment
levels). In Study 8, we asked 357 participants (67% women; mean
age !34.79, SD !12.23) to first complete the pictorial fusion
item, the verbal fusion scale (%!.87), and the measure of group
identification (%!.79) in counterbalanced order and in reference
to America.
Table 1 presents the items in the verbal fusion scale and the
factor loadings of each item. The verbal fusion factor accounted
for 55.70% of the variance. We conducted a CFA using AMOS as
we did in Study 1c to test whether the one-factor model could be
confirmed. The one-factor model revealed a good fit, with the fit
indices exceeding the .930 benchmark (CFI !.982, NFI !.969,
GFI !.975) and the residual index falling below the .08 bench-
mark (RMSEA !.070). The one-factor model presented a better
fit than did an alternative two-factor model (CFI !.969, NFI !
.956, GFI !.964) and the residual index (RMSEA !.092). We
also examined correlations between the verbal fusion scale, the
pictorial fusion measure, and the measure of identification. The
verbal fusion scale was more strongly related to the pictorial fusion
measure than was identification, rs(355) !.43 and .29, respec-
tively (z!2.15, p#.01).
The foregoing data indicate that the English version of the
verbal fusion scale is similar to the Spanish version in factor
structure and in some key indices of convergent and discriminant
validity. Encouraged, we proceeded to examine the predictive
validity of the English-language verbal fusion scale. To this end,
we tested participants’ endorsement of extreme progroup behavior
using an abbreviated version of the Swann et al.’s (2009) measure
that included two items assessing willingness to fight for, and die
for, one’s country (%!.89).
To assess how effectively each of the three predictor variables
predicted endorsement of extreme behavior, we ran a multiple
regression analysis in which each of our predictors was entered
simultaneously. As can be seen in Table 5, the verbal fusion
measure was a stronger predictor than was either pictorial fusion or
identification (zs"3.38, ps#.001). No difference appeared
between the pictorial measure of fusion and identification (z!.55,
p!.29). Apparently when the pictorial and verbal measures of
fusion were entered into the regression simultaneously, the verbal
measure accounted for much of the variance that would otherwise
have been associated with the pictorial measure.
Nomological Validity: Studies 9 –10
Nomological validity is a form of construct validity that reflects
the degree to which the construct behaves as it should within a
system of related constructs. To assess the nomological validity of
the verbal fusion scale, we conducted two studies. In the first, we
asked whether the predicted relationship between fusion and en-
dorsement of extreme behavior would be mediated by two vari-
ables: feelings of agency and invulnerability. In the second, we
sought to replicate crucial evidence for the fusion construct that
was first reported by Swann et al. (2009, Experiments 1 and 2).
Specifically, in support of the notion that personal and social
self-views are functionally equivalent among fused persons, they
discovered that activating a personal self-view amplified endorse-
ment of extreme progroup behavior among fused, but not non-
fused, individuals.
Study 9. Would the Relationship Between Fusion and
Endorsement of Extreme Progroup Action be
Mediated by Feelings of Agency and Invulnerability?
To address this question, we asked 1,981 Spaniards (whose
responses were also used in Studies 1c and 4) to complete mea-
sures of our three predictors (i.e., identity fusion measured verbally
[%!.84] and pictorially, and identification [%!.87]), two
potential mediator variables (i.e., feelings of agency and invulner-
ability), and Swann et al.’s (2009) measure of endorsement of
extreme action (i.e., fighting and dying [%!.81]) for the group.
We began by regressing the three predictors simultaneously on
extreme behavior for the group. As shown in Table 5, only verbal
measure predicted endorsement of extreme behaviors for the
group.
After this, we correlated the two potential mediators with our
outcome measure. Endorsement of extreme behaviors for the
group was significantly and positively correlated with feelings of
agency, r(1979) !.30, and invulnerability, r(1979) !.33 (all ps#
.001).
To test whether the relationship between identity fusion and
endorsement of extreme behavior for the group were meditated by
two potential mediators of this effect while controlling for identi-
fication, we conducted a pair of mediational analyses. In the first
926 GO
´MEZ ET AL.
analysis, pictorial fusion was the predictor; in the second, verbal
fusion was the predictor. Following Preacher and Hayes (2008),
we controlled for the main effect of identification by including it
as a covariate. Using the SPSS macro provided by Preacher and
Hayes, we conducted a bootstrapping test (nboots !5,000) for the
model.
Figure 2 displays the mediational analysis when the pictorial
measure of fusion was considered while controlling for identifica-
tion. The results of the analysis indicate that feelings of agency and
invulnerability partially mediated the effect of pictorial fusion on
endorsement of extreme behavior for the group (none of the
confidence intervals of the bootstrapping at the 95% confidence
interval for each of the potential mediators included zero, but the
effect of pictorial fusion remained significant when the mediators
were included in the equation). Figure 3 shows a similar pattern
when verbal fusion was the predictor. In this instance, however,
there was full mediation: When feelings of agency and invulner-
ability were entered into the equation, the effects of verbal fusion
were no longer significant. With this evidence in hand, we pro-
ceeded to ask whether, using the verbal fusion scale, we could
replicate crucial evidence for the functional equivalence of the
personal and social self.
Study 10: Does Activating a Personal Self-View
Amplify Endorsement of Extreme Behavior for the
Group Among Strongly Fused Individuals but Not
Among Weakly Fused Individuals?
Crucial support for the identity fusion formulation came from
Swann et al.’s (2009, Experiments 1 and 2) discovery of compen-
satory self-verification processes (see e.g., Swann & Hill, 1982)
among fused persons. To test the compensatory self-verification
hypothesis, the investigators challenged the personal identities of
fused persons by providing them with evaluations that contained
feedback that was notably more positive than the actual self-views
held by the participants (note that challenging their positive self-
views with negative feedback would have diminished theoretical
precision because such a challenge could trigger either self-
verification of self-enhancement strivings). The results indicated
that the challenge manipulation increased endorsement of extreme
progroup behaviors by fused persons but not nonfused persons. By
showing that challenging the personal self-views of fused persons
had effects similar to challenging their social self-views, this study
demonstrated that personal and social identities were functionally
equivalent among fused persons. The key goal of Study 10 was to
replicate this important finding using the verbal fusion measure.
Method. Spanish high school students volunteered to partic-
ipate in this research. The experiment was conducted in two waves
separated by 7 months. There was relatively little attrition between
the two waves, with 160 students completing Wave 1 and 137
participants (50% female; mean age !15.72, SD !1.26) com-
pleting Wave 2.
We altered the procedure used by Swann et al. (2009) in three
important ways. First, we had participants complete the verbal as
well pictorial measure of fusion. Second, we added a baseline
control condition in which participants received no feedback what-
soever. Third, whereas in the earlier studies participants learned
that the source of the feedback was an ingroup or outgroup
member, in this study the source was a neutral expert. The design
therefore included three factors: challenge of personal identities
(verified, challenged, control), fusion, and identification.
During Wave 1, participants completed several personality
scales and indicated how they behave when they are at school, with
their friends, with their families, with unknown people, etc. During
Wave 2, participants responded in counterbalanced order to the
verbal measure of fusion (%!.90), the pictorial measure of fusion,
and the identification scale (%!.86). Participants then learned
that the questionnaires they had completed during Wave 1 had
been evaluated by a group of psychologists who had prepared an
individual report about each participant. Participants who were
randomly assigned to the challenge condition learned that the
psychologists had formed impressions that were more positive
than the participants’ self-views on four of the five dimensions that
were evaluated (the five representative self-views were shy, inse-
cure, stubborn, nervous, and distrustful). Participants in the veri-
fying condition learned that the psychologists had formed impres-
sions that confirmed participants’ self-views on four of the five
dimensions that were evaluated. Participants in the control condi-
tion received no feedback. After the experimental manipulation,
participants responded to the measure of endorsement of extreme
FUSION
PICTORIAL
FIGHT DIE
INVULNERABILITY
AGENCY
IDENTIFICATION
B
= 0.03, p > .11, ns
B
= 0.42, p < .001
B
= 0.28, p < .001
B
= 0.20, p < .001
B
= 0.14, p < .01
B
= 0.28, p < .001
(B = 0.13, p < .01)
Total: 95% CI = .0986 to .1754 boots = 5000
Invulnerability: .0230 to .0533; Agency: .0069 to .0565
Figure 2. Study 9 reveals that agency, invulnerability, and loyalty partially mediate the effect of pictorial
fusion on endorsement of extreme behavior for the group. Numbers in parentheses refer to the beta after the
mediators were added to the regression equation. CI !confidence interval.
927
ON THE NATURE OF IDENTITY FUSION
behavior for the group on a scale ranging from 0 (totally disagree)
to 6 (totally agree;%!.84).
To determine whether the challenge manipulation influenced
participants’ perception that the psychologists saw them as they
saw themselves, we asked participants to rate the following three
items on a 7-point scale ranging from 0 (totally disagree) to 6
(totally agree): “The evaluators of my questionnaire have treated
me in such a way that they have made me feel understood,” “The
evaluators of my questionnaire understand me,” and “The evalu-
ators of my questionnaire see me as I see myself.” The resulting
scale was internally consistent (%!.92).
To assess the effectiveness of the challenge manipulation, we
conducted a pair of multiple regressions. We first regressed the
challenge manipulation (two dummy-coded variables were created
with the control condition as the reference group: Dummy Code 1
compared the verify condition with the control condition, whereas
Dummy Code 2 compared the challenge condition with the control
condition), pictorial fusion (effect coding: –1, 1), identification
(centered), and all two-way and three-way interactions on percep-
tions of the evaluation; we then repeated this analysis substituting
verbal fusion for pictorial fusion. The effect of Dummy Code 2
was significant for the first regression, including pictorial fusion
among the predictors, B!–1.17, t(125) !–2.36, p#.05, as well
as for the second regression, including verbal fusion among the
predictors, B!– 0.99, t(125) !–2.14, p#.05. This indicates that
participants in the challenge condition (M!2.75, SD !1.65) felt
less verified than did participants in both the control condition
(M!3.57, SD !1.73) and the verified condition (M!3.66,
SD !1.49; both ps#.05). No other main or interaction effects
were significant ( ps".11).
Results and discussion. To determine whether fusion and
challenge interactively predicted endorsement of extreme behavior
for the group and also whether the verbal measure of fusion had
the same effect as did the pictorial measure of fusion, we con-
ducted a pair of regressions, first for the pictorial measure of fusion
and then for the verbal measure of fusion.
We tested our hypothesis using multiple regression analyses.
Endorsement of extreme behavior for the group was regressed onto
identification (centered), fusion (effect coding: –1, 1), Dummy
Code 1, Dummy Code 2, and the two- and three-way interactions.
There was an interaction between fusion and the Dummy Code 2
(B!0.85), t(125) !2.78, p#.01. As shown in Figure 4A, fused
participants expressed more endorsement of extreme behavior for
the group in the challenge condition (M!4.06, SD !1.11) than
did participants in the control condition (M!2.65, SD !1.27),
t(25) !3.02, p#.01. In contrast, among nonfused participants,
those in the challenge condition (M!1.71, SD !1.13) were no
more inclined to endorse extreme actions than were those in the
control condition (M!1.45, SD !1.67), t(60) !0.95, p!.34.
There was no significant interaction between fusion and the
Dummy Code 1, however (B!– 0.11), t(125) !– 0.47, p".63,
indicating that as shown in Figure 4A, no difference emerged
between the control condition and the verification condition for
either fused (M!2.18, SD !1.35), t(24) !3.83, p".34, or
nonfused (M!1.67, SD !0.81), t(61) !0.10, p".91, partic-
ipants. Therefore, the key prediction was confirmed: Challenging
the personal self-views of fused participants amplified their en-
dorsement of progroup behavior, but the challenge manipulation
had no impact on nonfused participants.
In addition, a main effect of fusion emerged (B!0.38),
t(125) !2.39, p#.01, such that fused participants were more
inclined to endorse extreme behavior for the group than were
nonfused participants (M!2.90, SD !1.45, and M!1.61, SD !
1.01, respectively). A main effect of identification also emerged
(B!0.47), t(125) !2.71, p#.01, with greater identification
being associated with more endorsement of extreme behavior for
the group. Finally, there was a main effect of Dummy Code 2 (B!
0.88), t(125) !2.79, p#.01, such that participants in the chal-
lenge condition displayed more endorsement of extreme behavior
for the group (M!2.40, SD !1.55) than did participants in either
the control condition (M!1.82, SD !1.25) or the verifying
condition (M!1.82, SD !1.00). No other effects were signifi-
cant ( ps".35).
We tested the effects of verbal fusion by regressing the two
dummy codes, verbal fusion and identification (both centered), and
all two-way and three-way interactions on endorsement of extreme
behavior for the group. As expected, there was an interaction
between fusion and Dummy Code 2 (B!0.50), t(125) !2.21,
p#.05. As shown in Figure 4B, strongly fused participants in the
challenge condition expressed more endorsement of extreme be-
FUSION
VERBAL
FIGHT DIE
INVULNERABILITY
AGENCY
IDENTIFICATION
B
= 0.31, p < .001
B
= 0.18, p < .001
B
= 0.32, p < .001
B
= 041, p < .001
B
= 0.13, p < .01
(B = 0.09, p > .10)
B
= 0.05, p > .10, ns
Total: 95% CI = .0753 to .1260 boots = 5000
Invulnerability: .0420 to .0758; Agency: .0233 to .0620
Figure 3. Study 9 reveals that agency, invulnerability, and loyalty partially mediate the effect of verbal fusion
on endorsement of extreme behavior for the group. Numbers in parentheses refer to the beta after the mediators
were added to the regression equation. CI !confidence interval.
928 GO
´MEZ ET AL.
havior for the group than did strongly fused participants in either
the control condition (B!1.58), t(125) !8.16, p#.001, or the
verification condition (B!– 0.56), t(125) !–2.99, p#.01. In
contrast, among weakly fused participants, no between-condition
differences emerged ( ps".26). There was also a main effect of
fusion (B!0.69), t(125) !3.85, p#.001, indicating that the
more the participants were fused, the more they endorsed extreme
actions for the group. No other main or interaction effects were
significant ( ps".20).
A rival explanation of the results of this experiment is that
challenging fused people with positive feedback emboldened them
to endorse extreme behavior for the group. Examination of the
specific positive feedback in the challenge manipulation (e.g.,
“secure,” “calm,” “flexible,” and “trustful”), however, suggests
that the manipulation would have encouraged less rather than more
extreme behavior. Moreover, nothing in this rival hypothesis in-
dicates why the challenging feedback failed to encourage more
endorsement of extreme behaviors among nonfused participants.
In summary, the results of Study 10 replicate earlier evidence
that challenging the personal self-views of fused persons amplifies
their endorsement of progroup behavior. This evidence that chal-
lenging the personal self-views of fused persons (but not nonfused
35
4
4.5
vior
2.5
3
.
Extreme Group Behav
Nonfused
Fused
1
1.5
2
Endorsement of E
Fused
0
0.5
Challenge Control Verify
A
B
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Challenge Control Verify
Endorsement of E xtreme Group Behavior
Weakly Fus ed
Strongly Fused
Figure 4. Endorsement of extreme behavior for the group as a function of pictorial fusion and experimental
manipulation (Panel 4A) and as a function of verbal fusion and experimental manipulation (Panel 4B) in Study
10. For expositional purposes, values for weakly and strongly fused were '1 standard deviation from the mean
(M!2.52, SD !1.55).
929
ON THE NATURE OF IDENTITY FUSION
persons) is functionally equivalent to challenging their social self-
views provides crucial support for the identity fusion formulation.
Moreover, the results show that the same results emerge whether
fusion is measured pictorially or verbally. This evidence that
respondents who score highly on the verbal fusion scale respond to
fusion-relevant manipulations much as do those who endorse the
fused option on the pictorial measure bolsters the nomological as
well as predictive validity of the verbal fusion scale.
General Discussion
Past research has shown that people who are fused with their
group are more inclined to endorse, and actually enact, progroup
behavior. Yet, the precise nature of fusion has remained unclear. In
this report, we suggest that identity fusion consists of a sense of
connectedness and reciprocal strength that is commonly experi-
enced by people who develop relational ties to their group. To test
this conceptualization of identity fusion, we developed a verbal
measure of the construct to complement an earlier pictorial mea-
sure.
Our findings indicated that the new fusion measure was closely
associated with the pictorial measure and equaled or exceeded the
earlier pictorial measure on all indices of construct validity. In
addition, the verbal measure of fusion was associated with two key
variables: feelings of agency and invulnerability. In one study
(Study 9), these variables fully mediated the relationship of verbal
fusion to endorsement of extreme behavior for the group (the
pictorial measure only partially mediated this relationship). A
further study offered evidence for the functional equivalence of the
personal and social self among fused persons: Although challeng-
ing the personal selves of highly fused persons amplified their
progroup behavior, it had no such impact on low scorers on the
verbal fusion measure. Here, as in earlier research on identity
fusion (Swann et al., 2009), people compensated in the wake of
challenges to the veracity of their personal selves in precisely the
same way that they would compensate in the wake of attacks on
their social selves— by amplifying their progroup behavior.
Whereas previous research has provided evidence that identity
fusion predicts various outcome variables while controlling for
group identification (see e.g., Swann, Go´mez, Dovidio, et al.,
2010; Swann, Go´mez, Huici, et al., 2010; Swann et al., 2009), in
this research we tested the relationship of fusion and identification
more directly. Exploratory and confirmatory factor analyses, for
example, revealed that fusion is not merely “identification plus”
but is instead a unique construct that emphasizes synergistic,
self– other influence processes: Verbal fusion items loaded on a
single factor (Studies 1c and 8), whereas items designed to capture
group identification loaded on a different factor (Studies 3a and
3b). These factor analytic results strongly suggest that the identity
fusion scale and the group identification scale tap different con-
structs.
Further support for the independence of the verbal fusion and
identification scales was provided by several studies of predictive
validity. Over a 6-month period, the fusion scale predicted en-
dorsement of extreme progroup behaviors (e.g., fighting and dying
for one’s group) with greater fidelity than did an earlier (pictorial)
measure of fusion as well as a measure of identification (Studies 5
and 7b). In addition, the verbal fusion scale outstripped its rivals in
predicting the likelihood that participants would endorse jumping
to their deaths in front of a speeding trolley to save a fellow group
member (Study 6a) or kill terrorists who threatened the group
(Study 6b). Finally, whereas the foregoing studies were conducted
with Spaniards, the final series of studies validated the verbal
fusion scale with two samples of immigrants from 22 different
nations (Studies 7a and 7b) as well as a sample of Americans who
completed an English-language version of the scale (Study 8). All
of the foregoing effects emerged while controlling for identifica-
tion.
Together, these studies highlight the influential role of a highly
agentic, proactive personal self in group-related activity. This
emphasis on the impact of the personal self on progroup behavior
represents a key difference between the fusion formulation and
some versions of the dominant model of group relations, social
identity theory.
Links to Social Identity Theory
Most contemporary analyses of group processes have been
guided by either social identity theory (see e.g., Tajfel & Turner,
1979) or self-categorization theory (see e.g., Turner, Oakes, Has-
lam, & McGarty, 1994). Both formulations assume that group
identification is an influential determinant of the tendency to band
together with other group members in derogating members of
outgroups (see e.g., Branscombe, Ellemers, Spears, & Doosje,
1999; Brewer, 1999) and viewing fellow ingroup members through
rose-colored glasses (see e.g., Hewstone, Rubin, & Willis, 2002;
Klar & Giladi, 1997; Voci, 2006). Nevertheless, in our research
and in several previous investigations of identity fusion (Swann,
Go´mez, Dovidio, et al., 2010; Swann, Go´ mez, Huici, et al., 2010;
Swann et al., 2009), identification was a weak predictor of en-
dorsement of extreme progroup activity. The predictive frailty of
the identification measure in past work does not seem to reflect the
use of any particular scale, such as the widely used Mael and
Ashforth (1992) scale. Indeed, recent evidence (Swann, Go´mez,
Huici, et al., 2010) suggests that Leach et al.’s (2008) recently
developed measure of identification is an even weaker predictor of
endorsement of extreme behavior than is Mael and Ashforth’s
scale.
Why are fusion measures more powerful predictors of endorse-
ment of extreme behavior than are measures of identification?
When people endorse items on identification scales such as “My
group’s successes are my successes,” they acknowledge shared
fate with a category but do not necessarily communicate a deep
connection with the members of that category. Indeed, according
to social identity theory, identification encourages people to view
various ingroup members as categorically interchangeable with
other members of the category. As a result, although identified
individuals may know what they can do for the group, they hesitate
to make extreme sacrifices for other group members. In contrast, in
the tradition of self-verification theory’s emphasis on a highly
agentic personal self (see e.g., Swann, 1983, in press), the fusion
construct is based on the assumption that when fused people enter
groups, they do not lose sight of their personal self but instead add
group-related action as a potential mode of personal self-
expression. The fusion measure is accordingly designed to tap
familial sentiments toward the group— connectedness and recip-
rocal strength—that emphasize the synergistic relationship of the
person to the group. For this reason, fused persons do not merely
930 GO
´MEZ ET AL.
know what they can do for other group members, they are highly
motivated to do it (Swann, Go´mez, Huici, et al., 2010, Studies 3
and 4, respectively). The powerful motivating force of identity
fusion is especially evident when devotion to the group undergoes
an acid test: that is, when progroup activity involves serious
sacrifice (for a related discussion in the context of leadership, see
Reicher, Hopkins, Levine, & Rath, 2005).
Issues of measurement and predictive validity aside, the fusion
formulation’s conceptualization of what happens to the personal
self when people become closely aligned with groups is funda-
mentally different from the one advanced in the original statements
of social identity and self-categorization theory. For example,
Turner et al.’s (1987) principle of functional antagonism posits a
zero-sum relationship between personal and social identities in
which the salience of one diminishes the salience of the other. Our
evidence that activating the personal identities of fused people
increased their endorsement of extreme group-related behavior
suggests that personal and social identities can be activated simul-
taneously and synergistically reinforce one another (see e.g.,
Simon, 2004). Similarly, earlier evidence that activating the social
identities of fused people increased the certainty of their personal
identities (Swann et al., 2009, Experiment 3) provides additional
evidence that personal and social identities may cooperate rather
than compete with one another.
Our findings therefore provide further evidence for the emerg-
ing conviction that functional antagonism should not be a default
assumption in social identity approaches (see e.g., Abrams, 1994;
Baray, Postmes, & Jetten, 2009; Pickett, Silver, & Brewer, 2002;
Postmes & Jetten, 2006; Reid & Deaux, 1996; Stephenson, 1981;
Turner, Reynolds, Haslam, & Veenstra, 2006). With this assump-
tion invalidated, the way is clear for developing a new, broader
understanding of the interplay of personal and social identities. We
believe that the fusion construct is of particular interest here,
because it highlights the ways in which personal and social iden-
tities may combine to create a whole that is greater than the sum
of its parts.
From this perspective, the main contribution of this report is not
a methodological one centered on the development of a verbal
measure of identity fusion. Instead, our primary contribution is
conceptual: Whereas previous work on identity fusion has focused
on the capacity of fusion measures to predict extreme progroup
behavior, we have focused on explicating the nature and mediators
of the identity fusion construct. Importantly, we have clearly
distinguished the fusion construct from group identification, the
dominant measure of alignment with the group. Whereas fusion
emphasizes the tendency for people to develop feelings of con-
nectedness and reciprocal strength with other group members,
identification emphasizes the tendency for group members to ally
themselves with a common group identity. Similarly, we have
distinguished our account from past treatments of the interplay of
personal and social identities based on social identity approaches
(Tajfel & Turner, 1979; Turner et al., 1987). Whereas these earlier
accounts have historically argued that personal and group identi-
ties are functionally antagonistic, the fusion construct assumes that
personal and social identities may combine synergistically to com-
plement and reinforce one another. Finally, we have explicated the
psychological mechanisms that mediate the relationship between
fusion and endorsement of extreme group behavior. Specifically,
previous evidence that feelings of agency mediate the relationship
of fusion to extreme progroup behavior (Swann, Go´mez, Huici, et
al., 2010) could not fully account for why fused persons endorse
extreme behaviors. Our evidence of the mediational role of invul-
nerability, together with independent evidence that invulnerability
is associated with diminished perception of risk (see e.g., Greene
et al., 2000; Ravert et al., 2009), suggests that fused persons
endorse extreme behaviors because they systematically underesti-
mate the risks associated with such behaviors.
Limitations and Future Questions
Skeptics might ask how we can be certain that connectedness
and reciprocal strength are the crucial elements involved in iden-
tity fusion as measured by the pictorial scale. In answering, we
begin by noting that when both the verbal scale and pictorial scale
were entered into regressions predicting endorsement of extreme
action, the verbal scale tended to eliminate the effects of the pic-
torial scale. This suggests that the verbal scale does indeed capture
the construct measured by the pictorial scale. More generally,
when it came to predicting endorsement of extreme behavior, the
verbal measure performed extremely well— consistently better
than did rival predictors. Hence, even if there is more to fusion
than connectedness and reciprocal strength, from a practical stand-
point it matters that the scale we have developed was strongly
predictive of our outcome variables.
Critics could point out that, whereas earlier articles on identity
fusion proposed that it is a dichotomous state, the findings we
report here indicate that it is distributed continuously. Apparently,
the earlier evidence that fusion was dichotomous was an artifact of
the pictorial scale used in that research. For example, it may be that
had we used a pictorial scale with a larger number of gradations,
scores would have been distributed normally. In any event, it is
fortuitous that the fusion construct is continuous because we have
found that, for many groups (e.g., political, religious, and even
ethnic groups), the percentage of people who endorse the fused
option on our pictorial measure is quite low. The development of
a continuous measure of fusion opens the door for examining the
effects of fusion within domains that have heretofore been imprac-
tical to investigate due to a paucity of fused participants.
Although our findings provide solid evidence of the conse-
quences of fusion as well as insight into the mechanisms under-
lying the effects of fusion, there is still much more to be learned
about the construct. For example, little is known about the origins
of fusion, specifically why some people become fused with a
particular group whereas others do not, as well as what factors
determine why some people are fused and others are merely
strongly identified with a group. Another intriguing question is the
impact of context on fusion; although our evidence of test–retest
stability indicates that fusion has a temporally stable component,
context surely influences people’s feelings of fusion at any given
moment. Furthermore, our evidence of the mediational role of
agency and invulnerability in fusion effects represents a first step
toward the construction of more elaborate causal models of the
antecedents and consequences of identity fusion. A logical next
step will be to test whether manipulating the mediators produces
corresponding changes in the outcome variables that fusion has
been shown to predict.
Still another issue is the influence of culture on fusion and its
manifestations. To be sure, the verbal fusion scale predicted im-
931
ON THE NATURE OF IDENTITY FUSION
portant outcomes among Spaniards, among immigrants to Spain,
and among Americans. Nevertheless, it is likely that culture plays
an important role in the levels of fusion present in any given
society. We suspect, for example, that relationally oriented soci-
eties such as those found in Asia cultivate higher levels of fusion
than do those in the West (see e.g., Brewer & Chen, 2007; Nisbett
et al., 2001; Yuki, 2003). Similarly, it will be useful to systemat-
ically test our assumption that, within cultures, fusion is associated
with relational orientation and identification is associated with
collective orientation. Finally, although we believe that one of the
important features of the studies we report here is illuminating the
psychological antecedents of extreme behaviors such as suicide
killing, future research might extend this work to positive behav-
iors that presumably emerge when people become fused with
groups that are designed to pursue prosocial goals.
Clearly, a wide array of important and exciting questions re-
garding the antecedents, nature, and consequences of identity
fusion await investigation. It is hoped that our theorizing regarding
the nature of identity fusion, in conjunction with our new measure,
will facilitate future efforts to answer these questions.
References
Abrams, D. (1994). Social self-regulation. Personality and Social Psychol-
ogy Bulletin, 20, 473– 483. doi:10.1177/0146167294205004
Allport, F. H. (1962). A structuronomic conception of behavior: Individual
and collective: I. Structural theory and the master problem of social
psychology. Journal of Abnormal and Social Psychology, 64, 3–30.
doi:10.1037/h0043563
Andreu, J. M., Pen˜ a, M. E., & Gran˜ a, J. L. (2002). Adaptacio´ n psicome´trica
de la versio´ n espan˜ ola del Cuestionario de Agresio´n [Psychometric
adaptation of the Spanish version of the Aggression Questionnaire].
Psicothema, 14, 476 – 482.
Arbuckle, J. L. (1997). Amos users’ guide version 4.0. Chicago, IL:
Smallwaters.
Aron, A., Aron, E., & Smollan, D. (1992). Inclusion of other in the Self
Scale and the structure of interpersonal closeness. Journal of Personality
and Social Psychology, 63, 596 – 612. doi:10.1037/0022-3514.63.4.596
Aron, A., Aron, E., Tudor, M., & Nelson, G. (1991). Close relationships as
including other in the self. Journal of Personality and Social Psychol-
ogy, 60, 241–253. doi:10.1037/0022-3514.60.2.241
Baray, G., Postmes, T., & Jetten, J. (2009). When I equals we: Exploring
the relation between social and personal identity of extreme right-wing
political party members. British Journal of Social Psychology, 48, 625–
647. doi:10.1348/014466608X389582
Bastian, B., & Haslam, N. (2006). Psychological essentialism and stereo-
type endorsement. Journal of Experimental Social Psychology, 42, 228 –
235. doi:10.1016/j.jesp.2005.03.003
Billig, M., & Tajfel, H. (1973). Social categorization and similarity in
intergroup behavior. European Journal of Social Psychology, 3, 27–52.
doi:10.1002/ejsp.2420030103
Branscombe, N. R., Ellemers, N., Spears, R., & Doosje, B. (1999). The
context and content of social identity threat. In N. Ellemers, R. Spears,
& B. Doosje (Eds.), Social identity: Context, commitment, content (pp.
35–58). Oxford, England: Blackwell.
Brewer, M. B. (1999). The psychology of prejudice: Ingroup love or
outgroup hate? Journal of Social Issues, 55, 429 – 444. doi:10.1111/
0022-4537.00126
Brewer, M. B., & Chen, Y. (2007). Where (and who) are collectives in
collectivism: Toward conceptual clarification of individualism and col-
lectivism. Psychological Review, 114, 133–151. doi:10.1037/0033-
295X.114.1.133
Brewer, M. B., & Gardner, W. L. (1996). Who is this “we”? Levels of
collective identity and self representations. Journal of Personality and
Social Psychology, 71, 83–93. doi:10.1037/0022-3514.71.1.83
Buhrmester, M. D., Kwang, T., & Gosling, S. D. (in press). Amazon’s
Mechanical Turk: A new source of inexpensive, yet high-quality, data?
Perspectives on Psychological Science.
Buss, A. H., & Perry, M. (1992). The Aggression Questionnaire. Journal
of Personality and Social Psychology, 63, 452– 459. doi:10.1037/0022-
3514.63.3.452
Campbell, J. D., Trapnell, P., Heine, S. J., Katz, I. M., Lavallee, L. F., &
Lehman, D. R. (1996). Self-concept clarity: Measurement, personality
correlates, and cultural boundaries. Journal of Personality and Social
Psychology, 70, 141–156. doi:10.1037/0022-3514.70.1.141
Churchill, G. A. (1979). A paradigm for developing better measures of
marketing constructs. Journal of Marketing Research, 16, 64 –73. doi:
10.2307/3150876
Cialdini, R. B., Brown, S. L., Lewis, B. P., Luce, C., & Neuberg, S. L.
(1997). Reinterpreting the empathy-altruism relationship: When one into
one equals oneness. Journal of Personality and Social Psychology, 73,
481– 494. doi:10.1037/0022-3514.73.3.481
Coats, S., Smith, E. R., Claypool, H. M., & Banner, M. J. (2000). Over-
lapping mental representations of self and in-group: Reaction time
evidence and its relationship with explicit measures of group identifica-
tion. Journal of Experimental Social Psychology, 36, 304 –315. doi:
10.1006/jesp.1999.1416
Davis, M. H. (1980). A multidimensional approach to individual differ-
ences in empathy. JSAS Catalog of Selected Documents in Psychology,
10, 85.
Davis, M. H. (1983). Measuring individual differences in empathy: Evi-
dence for a multidimensional approach. Journal of Personality and
Social Psychology, 44, 113–126. doi:10.1037/0022-3514.44.1.113
Durkheim, E. (1951). Suicide (J. A. Spaulding & G. Simpson, Trans.). New
York, NY: Free Press. (Original work published 1897)
Greene, K., Krcmar, M., Walters, L., Rubin, D., & Hale, J. (2000).
Targeting adolescent risk-taking behaviors: The contribution of egocen-
trism and sensation-seeking. Journal of Adolescence, 23, 439 – 461.
doi:10.1006/jado.2000.0330
Guilford, J. P. (1954). Psychometric methods. New York, NY: McGraw–
Hill.
Haggard, P., & Tsakiris, M. (2009). The experience of agency: Feeling,
judgment and responsibility. Current Directions in Psychological Sci-
ence, 18, 242–246. doi:10.1111/j.1467-8721.2009.01644.x
Hewstone, M., Rubin, M., & Willis, H. (2002). Intergroup bias. Annual
Review of Psychology, 53, 575– 604. doi:10.1146/annurev
.psych.53.100901.135109
Jerusalem, M., & Schwarzer, R. (1992). Self-efficacy as a resource factor
in stress appraisal processes. In R. Schwarzer (Ed.), Self-efficacy:
Thought control of action (pp. 195–213). Washington, DC: Hemisphere.
Jetten, J., Branscombe, N. R., Schmitt, M. T., & Spears, R. (2001). Rebels
with a cause: Group identification as a response to perceived discrimi-
nation from the mainstream. Personality and Social Psychology Bulletin,
27, 1204 –1213. doi:10.1177/0146167201279012
Klar, Y., & Giladi, E. E. (1997). No one in my group can be below the
group’s average: A robust positivity bias in favor of anonymous peers.
Journal of Personality and Social Psychology, 73, 885–901. doi:
10.1037/0022-3514.73.5.885
Leach, C. W., Van Zomeren, M., Zebel, S., Vliek, M., Pennekamp, S. F.,
Doosje, B., . . . Spears, R. (2008). Collective self-definition and self-
investment: A two-dimensional framework of group identification. Jour-
nal of Personality and Social Psychology, 95, 144 –165. doi:10.1037/
0022-3514.95.1.144
Mael, F. A., & Ashforth, B. E. (1992). Alumni and their alma mater: A
partial test of the reformulated model of organizational identification.
932 GO
´MEZ ET AL.
Journal of Organizational Behavior, 13, 103–123. doi:10.1002/
job.4030130202
Maner, J. K., Luce, C. L., Neuberg, S. L., Cialdini, R. B., Brown, S.,
Sagarin, B. J., & Rice, W. E. (2002). The effects of perspective taking
on motivations for helping: Still no evidence for altruism. Personality
and Social Psychology Bulletin, 28, 1601–1610. doi:10.1177/
014616702237586
Markus, H., & Kitayama, S. (1991). Culture and the self: Implications for
cognition, emotion, and motivation. Psychological Review, 98, 224 –
253. doi:10.1037/0033-295X.98.2.224
Nisbett, R. E., Peng, K., Choi, I., & Norenzayan, A. (2001). Culture and
systems of thought: Holistic vs. analytic cognition. Psychological Re-
view, 108, 291–310. doi:10.1037/0033-295X.108.2.291
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know:
Verbal reports on mental processes. Psychological Review, 84, 231–259.
doi:10.1037/0033-295X.84.3.231
Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York, NY:
McGraw–Hill.
Pedahzur, A. (2005). Suicide terrorism. Cambridge, England: Polity Press.
Pickett, C. L., Silver, M. D., & Brewer, M. B. (2002). The impact of
assimilation and differentiation needs on perceived group importance
and judgments of group size. Personality and Social Psychology Bulle-
tin, 28, 546 –558. doi:10.1177/0146167202287011
Postmes, T., & Jetten, J. (2006). Individuality and the group: Advances in
social identity. London, England: Sage.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling
strategies for assessing and comparing indirect effects in multiple me-
diator models. Behavior Research Methods, 40, 879 – 891. doi:10.3758/
BRM.40.3.879
Prentice, D. A., Miller, D. T., & Lightdale, J. R. (1994). Asymmetries in
attachments to groups and to their members: Distinguishing between
common-interest and common-bond groups. Personality and Social
Psychology Bulletin, 20, 484 – 493. doi:10.1177/0146167294205005
Ravert, R., Schwartz, S., Zamboanga, B., Kim, S., Weisskirch, R., &
Bersamin, M. (2009). Sensation seeking and danger invulnerability:
Paths to college student risk-taking. Personality and Individual Differ-
ences, 47, 763–768. doi:10.1016/j.paid.2009.06.017
Reicher, S., Hopkins, N., Levine, M., & Rath, R. (2005). Entrepreneurs of
hate and entrepreneurs of solidarity: Social identity as a basis for mass
communications. International Review of the Red Cross, 87, 621– 637.
doi:10.1017/S1816383100184462
Reid, A., & Deaux, K. (1996). Relationship between social and personal
identities: Segregation or integration. Journal of Personality and Social
Psychology, 71, 1084 –1091. doi:10.1037/0022-3514.71.6.1084
Schubert, T. W., & Otten, S. (2002). Overlap of self, in-group and out-
group: Pictorial measures of self-categorization. Self and Identity, 1,
353–376. doi:10.1080/152988602760328012
Simon, B. (2004). Identity in modern society: A social psychological
perspective. Oxford, England: Blackwell.
Smith, E. R., & Henry, S. (1996). An in-group becomes part of the self:
Response time evidence. Personality and Social Psychology Bulletin,
22, 635– 642. doi:10.1177/0146167296226008
Stephenson, G. M. (1981). Intergroup bargaining and negotiation. In J. C.
Turner & H. Giles (Eds.), Intergroup behaviour (pp. 168 –198). Oxford,
England: Blackwell.
Swann, W. B., Jr. (1983). Self-verification: Bringing social reality into
harmony with the self. In J. Suls & A. G. Greenwald (Eds.), Social
psychological perspectives on the self (Vol. 2, pp. 33– 66). Hillsdale, NJ:
Erlbaum.
Swann, W. B., Jr. (in press). Self-verification theory. In P. Van Lang, A.
Kruglanski, & E. T. Higgins (Eds.), Handbook of theories of social
psychology. Sage: London.
Swann, W. B., Jr., Go´ mez, A., Dovidio, J. F., Hart, S., & Jetten, J. (2010).
Dying and killing for one’s group: Identity fusion moderates responses
to intergroup versions of the trolley problem. Psychological Science, 21,
1176 –1183. doi:10.1177/0956797610376656
Swann, W. B., Jr., Go´ mez, A., Huici, C., Morales, F., & Hixon, J. G.
(2010). Identity fusion and self-sacrifice: Arousal as a catalyst of pro-
group fighting, dying, and helping behavior. Journal of Personality and
Social Psychology, 99, 824 – 841. doi:10.1037/a0020014
Swann, W. B., Jr., Go´ mez, A., Seyle, C. D., Morales, J. F., & Huici, C.
(2009). Identity fusion: The interplay of personal and social identities in
extreme group behavior. Journal of Personality and Social Psychology,
96, 995–1011. doi:10.1037/a0013668
Swann, W. B., Jr., & Hill, C. A. (1982). When our identities are mistaken:
Reaffirming self-conceptions through social interaction. Journal of Per-
sonality and Social Psychology, 43, 59 – 66. doi:10.1037/0022-
3514.43.1.59
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). Monterey, CA: Brooks–Cole.
Tropp, L. R., & Wright, S. C. (2001). Ingroup identification as inclusion of
ingroup in the self. Personality and Social Psychology Bulletin, 27,
585– 600. doi:10.1177/0146167201275007
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, England: Blackwell.
Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and
collective: Cognition and social context. Personality and Social Psy-
chology Bulletin, 20, 454 – 463. doi:10.1177/0146167294205002
Turner, J. C., Reynolds, K. J., Haslam, S. A., & Veenstra, K. (2006).
Reconceptualizing personality: Producing individuality by defining the
personal self. In T. Postmes & J. Jetten (Eds.), Individuality and the
group: Advances in social identity (pp. 11–36). London, England: Sage.
Turner, J. C., Sachdev, I., & Hogg, M. A. (1983). Social categorization,
interpersonal attraction and group formation. British Journal of Social
Psychology, 22, 227–239.
Voci, A. (2006). Relevance of social categories, depersonalization and
group processes: Two field tests of self-categorization theory. European
Journal of Social Psychology, 36, 73–90. doi:10.1002/ejsp.259
Yuki, M. (2003). Intergroup comparison versus intragroup relationships: A
cross-cultural examination of social identity theory in North American
and East Asian cultural contexts. Social Psychology Quarterly, 66,
166 –183. doi:10.2307/1519846
Received July 16, 2010
Revision received November 16, 2010
Accepted December 21, 2010 !
933
ON THE NATURE OF IDENTITY FUSION
A preview of this full-text is provided by American Psychological Association.
Content available from Journal of Personality and Social Psychology
This content is subject to copyright. Terms and conditions apply.