VU Research Portal
Sticking together or falling apart: Group identification as a psychological determinant
of group commitment versus individual mobility.
Ellemers, N.; Spears, R.; Doosje, B.J.
published in
Journal of Personality and Social Psychology
1997
DOI (link to publisher)
10.1037//0022-3514.72.3.617
document version
Publisher's PDF, also known as Version of record
Link to publication in VU Research Portal
citation for published version (APA)
Ellemers, N., Spears, R., & Doosje, B. J. (1997). Sticking together or falling apart: Group identification as a
psychological determinant of group commitment versus individual mobility. Journal of Personality and Social
Psychology, 72, 123-140. https://doi.org/10.1037//0022-3514.72.3.617
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal ?
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
E-mail address:
vuresearchportal.ub@vu.nl
Download date: 09. Aug. 2022
Journal of Personality and Social Psychology
1997,
Vol. 72, No. 3, 617-626
Copyright 1997 by the American Psychological Association, Inc.
0022-35J4/97/S3.00
Sticking Together or Falling Apart: In-Group Identification
as a Psychological Determinant of Group Commitment
Versus Individual Mobility
Naomi Ellemers
Free University
Russell Spears and Bertjan Doosje
University of Amsterdam
Two experiments investigated how in-group identification, manipulated with a bogus pipeline tech-
nique affects group members' desire for individual mobility lo another group. In die first experiment
(JV
= 88), the in-group had low status, and group boundaries were either permeable or impermeable.
Low identifiers perceived the group as less homogeneous, were less committed to their group, and
more strongly desired individual mobility to a higher status group than did high identifiers. The
structural possibility of mobility afforded by permeable group boundaries had no comparable effect.
The second experiment (JV = 51) investigated whether in-group identification can produce similar
effects when relative group status is unknown. Even in the absence of an identity threat, low identifiers
were less likely to see the groups as homogeneous, felt less committed to their group, and more
strongly desired individual mobility than did high identifiers. Results are discussed with reference
to social identity and self-categorization theories.
For some fans of sports teams, supporting "their" team at
home and away, not only rejoicing with them after success but
also sticking with them through failure and defeat, are integral
features of group life. Indeed, even though there are no formal
restrictions that prevent supporters from turning their backs on
"their" team, or even from switching loyalties to another, more
successful team, such a course of action would be unthinkable
for the true fan (see e.g., Wann & Branscombe, 1990). Thus,
whether fans stick with their team or not is determined not so
much by the question of whether alternatives for their adulation
are objectively available; rather, this is a matter of psychological
commitment stemming from the importance of that particular
team to the supporter's identity. In the present research we ex-
amined the role such psychological factors play in people's
inclination to stand by their group or to leave. More specifically,
we investigated how in-group identification is related to the
pursuit of individual mobility versus group loyalty and
commitment.
The role of in-group identification in social perception and
behavior is elaborated on by social identity theory (e.g., Tajfel,
1978;
Tajfel & Turner, 1979) and self-categorization theory
(Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Social iden-
Naomi Ellemers, Department of Social Psychology, Free University,
Amsterdam; Russell Spears and Bertjan Doosje, Department of Social
Psychology, University of Amsterdam, Amsterdam.
The names of the authors are listed in random order: all contributed
equally to this article. This research was supported by the Dutch Organi-
zation for Scientific Research (N.W.O., SGW Grant 575-70-053)
awarded to Russell Spears. We appreciate the help of Jolanda Jetten
with Experiment 2, and we thank John Turner for his most insightful
comments on a previous version of this article.
Correspondence concerning this article should be addressed to Naomi
Ellemers, Department of Social Psychology, Free University, Van der
Boechorststraat I, 1081 BT Amsterdam, The Netherlands. Electronic
mail may be sent via the Internet to n.ellemers@psy.vu.nl.
tity theory was developed to explain why, under certain circum-
stances, people may act in terms of group memberships (i.e.,
their social identity) rather than behave as distinct individuals.
Departing from this general approach, social identity theory
focuses on the different ways in which group members may
respond to unfavorable social status, and it specifies how differ-
ent beliefs about the properties of the social structure may lead
people to engage in either individualist or collective coping strat-
egies.
Self-categorization theory elaborates in more detail the
role of group identification, by specifying how salience of either
one's personal or social identity may guide various social per-
ceptions and behaviors. An interesting theoretical tension arises
when we try to predict the responses of members of lower
status groups from these two perspectives. Specifically,
self-
categorization theory predicts that people are more inclined to
behave in terms of their group membership because their com-
mon identity as group members is more salient. Social identity
theory, however, suggests that people generally tend to avoid
the association with a group that does not contribute to a favor-
able social identity (i.e., a lower status group), especially where
sociostructural conditions allow this (i.e., when group bound-
aries are permeable). An important question, then, is under
what circumstances members of lower status groups will feel
committed to their common identity and stick together instead
of opting for membership in a more attractive social group.
In previous research (Doosje, Ellemers, & Spears, 1995;
Spears, Doosje, & Ellemers, in press) we have demonstrated
that responses to group identity threat differed depending on
the person's prior level of in-group identification. In these earlier
studies, we measured perceptions of intragroup homogeneity
and heterogeneity in response to differences in group status
(Doosje, Ellemers, & Spears, 1995) and self-stereotyping as a
consequence of threats to either group status or group distinc-
tiveness (Spears et al., in press). Results from these studies
revealed that, compared to high identifiers, low identifiers were
more likely to accentuate intragroup heterogeneity and were less
617
618
ELLEMERS, SPEARS, AND DOOSJE
inclined to self-stereotype as a group member, when their
group's identity was threatened. We explained these findings by
arguing that low identifiers are more disposed to individual-
level responses, dissociating themselves from the in-group,
whereas high identifiers are more likely to display a group-level
reaction, "sticking together" when their group is threatened
(see also Lee & Ottati, 1995; Simon, 1992). So far, however,
our analysis in terms of individual- versus group-level responses
has remained somewhat speculative, because these earlier stud-
ies did not include a direct measure of group members' inclina-
tion to leave the threatened group. The aim of
the
present investi-
gation was to address this issue more explicitly and to assess
how strength of in-group identification affects group members'
desire for individual social mobility, when social identity is
threatened (Experiment 1), or even when it is not (Experiment
2).
The role of group identification as a determining factor has
been neglected by earlier experimental research in this area,
presumably because it has been methodologically more conve-
nient to regard it as an outcome or a correlate of behavioral
responses rather than as a cause of them. Rirthermore, measur-
ing group identification as a predictor does not rule out the
influence of additional causal factors correlated with this vari-
able.
In the present studies, we used a bogus pipeline procedure
developed specially to manipulate identification in order to sur-
mount these problems (cf. Doosje, Ellemers, & Spears, 1995).
This technique allowed us to experimentally induce different
levels of identification in group members and to study the ensu-
ing responses to the intergroup situation.
Experiment 1
Theoretical analyses of the desire for mobility to another
group as an individual strategy to cope with low group status
have mainly focused on people's beliefs about the properties of
the social structure as determinants of strategy preference (Taj-
fel,
1978; Tajfel & Turner. 1979). Empirical research in which
different kinds of social structures were experimentally manipu-
lated has confirmed the important role of these sociostructural
variables that determine the feasibility of different behavioral
options (for an overview, see Ellemers, 1993). Different studies
that have compared people's responses when group member-
ships were either flexible (permeable group boundaries) or fixed
(impermeable group boundaries) have yielded consistent results
(Ellemers, van Knippenberg, de Vries, & Wilke, 1988; Ellemers,
van Knippenberg, & Wilke, 1990). These studies indicate that,
when group boundaries are permeable, members of low-status
groups generally respond with decreased satisfaction with group
membership and decreased in-group identification. In other
words, when presented with the opportunity to gain membership
in another group, because group boundaries are permeable, peo-
ple seem to opt for an individual mobility strategy in response
to their group's low status rather than dealing collectively with
the group threat (cf. Lalonde & Silverman, 1994; Taylor &
McKirnan, 1984; Wright, Taylor, & Moghaddam, 1990).
At first sight these results seem to be symptomatic of a rather
opportunistic stance toward group membership. Insofar as the
group may contribute to positive social identity (as is the case
with a high-status group), people show strong in-group identi-
fication and express satisfaction with their membership in this
group. When, on the other hand, the in-group does not compare
positively to other groups, people seem quite prepared to take
advantage of the fact that group boundaries are permeable and
to leave their own group to gain membership in a group with
higher status. Social identity and self-categorization theorists
have consistently argued, however, that the group is more than
just a vehicle for serving personal advancement and egoistic
self-presentational goals (Tajfel, 1978; Tajfel & Turner, 1979;
Turner, 1975, 1991). From this theoretical point of view, we
therefore predicted that this individualistic response pattern may
not be as general as it may seem, and indeed, that the degree
to which people identify as members of their group can play a
crucial role in determining whether they are inclined to show
an individual-level or group-level response to group threat (cf.
Doosje, Ellemers, & Spears, 1995; Spears et al., in press).
When we take a closer look at the studies in which people
expressed the desire to leave their lower status group when
presented with an opportunity for individual mobility (Ellemers
et al., 1988, 1990; Lalonde & Silverman, 1994; Wright et al.,
1990),
it turns out that these studies were all conducted with
laboratory groups, to which people were assigned on a random
basis.
However, in some further experiments, in which group
membership allegedly was based either on some common trait
(Ellemers, Doosje, van Knippenberg, & Wilke, 1992), or was
reinforced because group members suffered common unfair
treatment (Ellemers, Wilke, & van Knippenberg, 1993), at-
tempts to achieve upward individual mobility in response to
permeable group boundaries were less prominent. Furthermore,
research investigating behavior in real-group settings reveals
that group members may refrain altogether from displaying stra-
tegic behavior serving their own personal self-interest when
their common identity as in-group members is sufficiently im-
portant (see de Gilder, 1993; Mlicki & Ellemers, 1996). These
results are consistent with our hypothesis that, when presented
with objective opportunities for individual mobility, the impor-
tance of the group to a person's identity affects that person's
psychological readiness to display individualistic behavior and
determines whether or not he or she will lake advantage of these
opportunities to improve his or her personal standing in the
social structure.
In sum, theoretical assumptions, together with our investiga-
tions of perceived intragroup homogeneity and self-stereotyping
as well as previous research examining individual mobility as
an identity management strategy, led us to hypothesize that the
tendency to use such individualistic strategies is determined at
least in part by the extent to which people feel involved with
or committed to their group (cf. Turner, Hogg, Turner, & Smith,
1984).
As indicated earlier, it is important to note that the studies
investigating individual mobility attempts have hitherto mea-
sured in-group identification as a dependent variable that is
indicative of the psychological readiness to leave one's group.
However, at a theoretical level (Tajfel, 1978; Turner, 1975,
1987),
strength of in-group identification can be seen as an
important cause of people's inclination to engage in individual-
istic or intergroup behavior. Group identification should there-
fore be investigated as an independent variable that is likely to
determine whether people opt primarily for individual mobility
or for social change attempts in response to identity threats.
Previous research in which effects of differential identity sa-
lience were studied has actually induced differential social con-
texts (e.g., Haslam, Turner, Oakes, McGarty, & Hayes, 1992),
GROUP IDENTIFICATION AND INDIVIDUAL MOBILITY
619
differential intragroup similarity (Kawakami & Dion, 1993), or
differential relevance of natural group memberships (Lalonde &
Silverman, 1994). In contrast, as noted earlier, degree of group
identification as an independent variable has been surprisingly
neglected, in part because, unlike salience, this is a feature of the
person as well as something that is influenced by the situation. In
recent work we have therefore devised and validated a bogus
pipeline procedure designed to directly manipulate, by experi-
mental means, identification with artificially created laboratory
groups to address the causally determinant role of identification,
unconfounded by other variables (see Doosje, Ellemers, &
Spears, 1995). This technique is theoretically important in the
present context given the problems associated with using identi-
fication in natural groups as a classification variable in an analy-
sis of variance (ANO\A; e.g., by means of a median split;
Roccas & Schwartz, 1993) or as a dependent variable in a
correlational analysis (e.g., Simon, Kulla, & Zobel, 1995). Pre-
cisely because identification develops and builds over a history
of association with a group, it is likely to be naturally con-
founded with a range of other factors that could also explain
commitment to the group (such as actual or perceived interde-
pendence, familiarity, interpersonal loyalties, or even force of
habit).
Experimental manipulation of identification in the pres-
ent study allowed us to discount these explanatory contenders.
In sum, despite evidence of a preference for individual mobil-
ity in the face of low group status, we proposed that group
members who feel involved with their group are more likely to
show commitment to their group, even when mobility is possi-
ble.
Thus, although permeable group boundaries may make peo-
ple aware of themselves as movable agents in the social struc-
ture,
whether they will take advantage of the opportunity to
move is likely to depend on how they experience the properties
of the social structure, particularly their identity investment in
the group (cf. Tajfel, 1978). Members of a low-status group
may therefore respond differently to objective opportunities for
individual mobility, depending on their level of in-group identi-
fication, such that low identifiers report a greater desire for
individual mobility than do high identifiers.
In the present experiment we induced low group status and
orthogonally manipulated permeability of group boundaries and
in-group identification. In this way, we could independently as-
sess effects of the structural possibility of achieving member-
ship in a higher status group (because group boundaries are
permeable) and of the psychological readiness to leave one's
group (because of low in-group identification) on the desire for
individual mobility. In other words, the joint manipulation of
structural and psychological factors within the same research
design enabled us to directly compare whether an individual-
level response to group threat is elicited by permeable group
boundaries, low in-group identification, or both. Our main pre-
diction was that the extent to which people identify with the in-
group would moderate or even override effects of group bound-
ary permeability.
Method
Overview
At each experimental session, participants were randomly divided into
two groups, allegedly on the basis of a problem-solving task that mea-
sured whether they were inductive or deductive thinkers. False feedback
on a group problem-solving task was used to induce relative group
status.
In this experiment, the in-group always had lower status than the
out-group. Furthermore, a bogus pipeline procedure was used to lead
participants to believe that they were strongly or weakly involved with
their group. Finally, further instructions informed group members either
that the composition of the groups would remain the same throughout
the experiment, or that some participants might change groups. In this
way, we manipulated in-group identification (low or high) and perme-
ability of group boundaries (permeable or impermeable) in a 2 X 2
between-subjects factorial design. The main dependent variables con-
sisted of measures of perceived group homogeneity, commitment to the
group, personal identification, and desire for individual mobility.
Participants
In the first experiment, participants were students of the University
of Amsterdam (N = 88). They were randomly assigned to one of the
experimental conditions, although men (n = 52) and women (n = 36)
were assigned in equal proportions to each cell. Their mean age was 21
years (range: 16-34). Students were approached in the university can-
teen and asked to participate in the study on a voluntary basis. Each
session lasted approximately 1.5 hr; participants received 15 Dutch
guilders (approximately $10.00 U.S.) as remuneration. At the end of
each session, participants were fully debriefed and were asked to not
discuss the experiment with fellow students.
Procedure
On arrival at the laboratory, participants were seated in a computer
room. About 10 participants were present for each session; they were
partially separated from each other by screens. The experimenter ex-
plained that three electrodes would be placed on one of their hands, to
measure their galvanic skin response while they performed the experi-
mental tasks. Next the experimenter put some electrode gel on the elec-
trodes and instructed participants how to put the three electrodes on one
of their hands (see Doosje, Ellemers, & Spears, 1995, for further details
about this procedure). After all participants had completed this, they
were told that the computers were connected with each other in a net-
work, that further instructions would be displayed on the computer
screen, and that they could answer any questions that appeared with the
keyboard or the mouse.
Categorization. Instructions on the computer screen explained that
the experiment was designed to study how people collaborate on a group
task. Tb be able to do this, participants first had to be divided into two
groups: "inductive" and "deductive" thinkers, on the basis of a test that
ostensibly measured their style of thinking (Doosje, Spears, & Koomen,
1995).
This test consisted of a word association task and a number
association task. Each item of this test presented participants with a key
word (e.g., house) or a key number (e.g., 4), after which they had to
indicate which of four alternatives they associated most strongly with
the key word (e.g., number, street, flat, or room) or number (e.g., 2,
16,
40, or 44). After completion of this test, participants were led to
believe that the main computer could determine their style of thinking.
In reality, all participants were assigned to the group of inductive thinkers
(a pilot study had not revealed evaluative differences or different expec-
tations on the basis of the two group labels). Furthermore, they were
informed that four of the other participants present were allocated to
the same group; the exact size of the other group was not revealed.
Further instructions told participants that, in line with previous research,
the two groups most likely would be of equal size and that men and
women would probably be equally represented in both groups.
Group
task.
Next, participants performed a group task that com-
prised eight items. Each item consisted of a brief description of a ' 'per-
sonnel problem;" for example,
620 ELLEMERS, SPEARS, AND DOOSJE
A flower shop is not doing well. It seems that, due to the recession,
flowers are the first thing that people economise on. It is only
possible to ensure the future of the company if the costs of personnel
are reduced.
Then, two possible solutions were suggested (e.g., "This can be
achieved in two ways: (a) by discharging a small number of employees,
or (b) by reducing the working hours of all employees"). Participants
were first asked to indicate which of these solutions they would person-
ally prefer. Then they received false feedback, ostensibly indicating the
preferences of their fellow in-group members, on the basis of which
each group member had to give his or her final decision. It was made
clear that the group could earn more points with this decision task, the
more group members made the correct final decision. Furthermore, it
was emphasized that final score on the group task would be corrected
for the number of group members, so that the scores of the two groups
could be compared.
Manipulation of in-group identification. After completion of this
group task, but before group members received feedback about tiieir
group's performance, we manipulated the level of in-group identifica-
tion. Participants were first asked to indicate on a 9-point scale (1 =
not at all, 9 = very much) to what extent they agreed with a number
of general statements that indirectly referred to group membership or to
contact with other people in general (e.g., "Relationships with other
people are very important to me"). Further instructions explained that
this was part of a measure tapping the extent to which participants felt
involved with their group. Participants were led to believe that the
computer could calculate this from several indices, namely their answers
to this questionnaire, the way they had collaborated with their fellow
group members during the group task, and the tension level in their
bodies while they worked on the group task and the questionnaire, as
measured with the electrode. We deliberately left unspecified how this
was calculated exactly, or what the range of possible scores would be.
We then manipulated the level of in-group identification by informing
participants in the low-identification condition that their group involve-
ment score (27 points) was lower than the average score for these
kinds of groups (40 points), whereas the group involvement score of
participants in the high-identification condition (53 points) lay above
this average score. Participants were asked to write their involvement
score on a form they had been provided, on which the average score
was preprinted.
Induction of low group status. After the manipulation of in-group
identification, participants received feedback about the two groups' per-
formance on the problem-solving task. These scores always indicated
that the in-group's performance was inferior to that of the out-group
(54 points for the in-group, 67 points for the out-group). Here too, we
reinforced the manipulation by indicating that the average score for this
group task was 61 points. Hence, the in-group performance also fell
below this average score. Again, participants were asked to write on the
form, on which the average score was preprinted, the score their group
had earned on the problem-solving task.
Manipulation of permeability. After receiving the feedback on the
group task, participants were told not only that the quality of the collabo-
ration in the group would depend on the style of thinking of the group
members, but also that it is important that people can adapt to their
group, and that their personal performance is compatible with the perfor-
mance of their group. Participants in the impermeable condition were
subsequently informed that the composition of the groups would never-
theless remain the same throughout the experiment and that they would
perform all experimental tasks with the same group. In the permeable
condition, however, participants were led to believe that the composition
of the groups might change during the course of the experiment. De-
pending on how well group members had been able to adapt to their
group, and how well their individual performance matched their group's
performance during the first experimental task, some participants would
be able to change groups for the remaining group tasks (cf. Ellemers et
al.,
1988, 1990, 1993). We deliberately chose to refer to various different
criteria in the permeability manipulation, to avoid the possibility that
participants could infer that they would be able to influence their chances
of being reassigned to the other group.
Dependent measures. After the experimental manipulations were in-
duced, the dependent measures were assessed. The first three questions
were intended to check these manipulations. Participants were asked to
what extent it would still be possible to change groups (1 = certainly
not, 9 = certainly), what the different measures had revealed about
their level of involvement with the in-group (1 = much below average,
9 = much above average), and what the relative performance of the in-
group on the group task had been (1 = much worse than the other
group, 9 = much better than the other group). Then the dependent
measures asked about the perceived homogeneity of the two groups (' 'To
what extent are inductive /deductive thinkers similar to each other?": 1
= not at all, 9 = very much). Furthermore, four questions asked how
strongly participants actually identified with or felt committed to the
group of inductive thinkers ("I identify with the inductive thinkers/I
see myself as an inductive thinker/I am glad to belong to the group of
inductive thinkers/I feel strong ties with other inductive thinkers:" 1 =
not at all, 9 = very much; a = .85). This scale is hereafter referred to
as group commitment, to distinguish it from the identification manipula-
tion (i.e., the independent variable). One question was asked about
participants' personal identification
(*
l
I am different from other people:''
I = not at all, 9 ~ very much). Then we tapped participants' pursuit
of individual mobility (cf, Worchel, Coutant-Sassic, & Wong, 1993),
by asking them two questions indicating with which group they would
rather perform another group task, and with which group they would
rather collaborate (a = .79). These two questions were answered on 9-
point bipolar scales, with lower scores indicating greater preference for
the group of deductive thinkers (the in-group) and higher scores indicat-
ing greater preference for the group of deductive thinkers (the out-
group),
the latter denoting stronger preference for individual mobility.
Results
Checks on the Manipulations
We omitted the data from 2 participants from the analyses
because they misunderstood the manipulations. One of" them
indicated that (s)he thought the in-group had performed better
than the out-group (whereas all participants were informed that
their group had performed worse than the other group), and
the other one failed to respond correctly on the check on the
identification manipulation. Results from the remaining 86 parti-
cipants show that they correctly indicated that their group had
performed worse than the other group {M = 2.29, which sig-
nificantly deviates from the scale midpoint, 5), F{\, 82) —
821.58,
p <
.001.
Furthermore, participants in the impermeable
condition considered it less likely (M = 2.00) than participants
in the permeable condition (M - 7.47) that some people might
change groups, F(l, 82) = 127.95, p < .001. Finally, partici-
pants in the Iow-in-group-identification condition reported less
involvement with the group of inductive thinkers (M = 2.09)
than those in the high-identification condition (M = 7.49), F(\,
82) - 965.50, p < .001.
Principal-Components Analysis
To check the discriminant construct validity of our main de-
pendent variables (i.e., commitment, group homogeneity, and
individual mobility), we conducted a principal-components
analysis. This analysis resulted in a three-factor solution, which