Sie sind auf Seite 1von 12

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 239

Group Processes &


Intergroup Relations
2003 Vol 6(3) 239250

Subgroup Identification,
Superordinate
Identification and
Intergroup Bias Between
the Subgroups
Jukka Lipponen, Klaus Helkama and Milla Juslin
Department of Social Psychology, University of Helsinki

The primary aim of this study was to investigate the effects of subgroup and superordinate
identification on intergroup differentiation between the subgroups. Hypotheses based on the
social identity approach were formed and tested in two samples gathered from the same Finnish
shipyard: Sample 1 consisted of the employees of 20 different subcontractors working at the
shipyard, and Sample 2 consisted of the shipyards own workforce. The results from Sample 1
supported the idea that subgroup identification (identification with the subcontractor) is
positively related to ingroup bias, and superordinate identification (identification with the
shipyard) is negatively related to ingroup bias toward other subgroups under the same
superordinate category (shipyard). Among the shipyards own workers (Sample 2), in turn,
identification with the shipyard was significantly related to increased levels of ingroup bias
toward the subcontractors working there. The results, thus, indicate that the positive effects of
superordinate identification on subgroup relations may be limited to only some of the
subgroups.

keywords

identification, intergroup differentiation, social identity

S O C I A L identity theory (SIT; Tajfel & Turner,


1979) aims to explain peoples readiness to
favor ingroups in claiming that social categorization arouses self-evaluative social comparison
processes. Positive social identity is based to a
large extent on favorable comparisons that can
be made between the ingroup and some
relevant outgroup (Tajfel & Turner, 1979,
p. 16). According to Hinkle and Brown (1990)
and Kelly (1988), this means that SIT postulates
an explicit link between an individuals social

identity and the amount of intergroup differentiation displayed. It is argued that an individuals membership of various social groups

Authors note
Address correspondence to: Jukka Lipponen,
Department of Social Psychology, P.O. Box 9,
00014 University of Helsinki, Finland
[email: jukka.lipponen@helsinki.fi]

Copyright 2003 SAGE Publications


(London, Thousand Oaks, CA and New Delhi)
[1368-4302(200307)6:3; 239250; 033834]

G
P
I
R

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 240

Group Processes & Intergroup Relations 6(3)

makes an important contribution to his or her


self-concept by providing a source of social
identity, and that group members are engaged
in a search for positive ingroup distinctiveness.
Individuals strive to differentiate their ingroups
from other groups along the dimensions that
make favorable comparisons possible. The
more the individual identifies with a certain
group, the more his or her self-esteem depends
on a positive social identity deriving from that
group membership and, therefore, the greater
is the need to establish positive ingroup distinctiveness by means of intergroup differentiation. This means that we might expect a clear,
positive correlation between the level of
identification with a certain group, and intergroup differentiation (Hinkle & Brown, 1990;
Kelly, 1988).
This idea has been tested in many studies
which have examined the correlation between
the strength of group identification and the
amount of ingroup bias shown in intergroup
judgments. Some of these studies have been
carried out in organizational contexts: in a
bakery (Brown & Williams, 1984), a paper mill
(Brown, Condor, Mathews, Wade, & Williams,
1986), and a hospital (Oaker & Brown, 1986).
The ingroups of interest were the respondents
own work or departmental group, and the outgroups were normally other workgroups or
departments inside the same organization.
However, the strength of the correlation
between workgroup identification and ingroup
bias was ver y variable, ranging in different
groups from positive, through non-existent, to
negative. The rather unstable correlation
between group identification and ingroup bias
has been confirmed in many studies in various
settings (see Hinkle & Brown, 1990 for a
review). Although the identificationbias correlations were usually positive, they were often
relatively low. According to Hinkle and Brown
(1990), these results raise an important
question about the central assumption of SIT
that social identities are derived from and maintained by intergroup comparison (Hinkle &
Brown, 1990, p. 64).
In his recent state-of-the-art overview of social
identity research, Turner (1999) replies to this

240

criticism by claiming that, in fact, SIT has never


suggested a direct causal connection between
ingroup identification and bias. It has always
been assumed that this causal relationship is
moderated by a number of complicating
factors, and that ingroup bias is only one of the
several strategies which group members use to
achieve positive distinctiveness. He also points
out that the instrument used to measure social
identification in many previous studies (see
Brown et al., 1986) focuses more on personal
identity than on social identity.
Previous research methodology has also been
criticized by Perreault & Bourhis (1999).
According to them, many previous studies have
leaned on the original Tajfel (1978) definition
of identification, which does not distinguish
between the degree and quality of identification. The degree of identification refers to
the strength of an individuals identification
with the ingroup, independently of whether he
or she feels positive or negative about this
identification. The quality of social identification can be defined as the affective evaluation, positive or negative, of an individuals
group membership (Perreault & Bourhis,
1999). Previous research has tended to confuse
the degree of ingroup identification with the
quality of social identity within combined
measures of ingroup identification (e.g. Brown
et al., 1986). Recent studies that have used this
conceptual distinction have found significant
links between the degree of ingroup identification and ingroup bias (e.g. Branscombe,
Wann, Noel, & Coleman, 1993).
For the present study, the most interesting
explanation of the findings that identification
and bias were not always positively related has
been offered by Brown & Williams (1984). They
suggested that the individuals in their study
may have considered their own subgroup
(departmental) identity synonymous with their
superordinate group (corporate) identity:
when responding to the subgroup-identification scale, they may have actually been
identifying with a superordinate category or
professional groups, rather than with their own
particular subgroup. Thus, studies of
intergroup differentiation in organizations,

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 241

Lipponen et al.

especially concerning the identificationbias


link, should take into consideration the possibility of identification with two different foci at
the same time: the subgroup (workgroup,
department, etc.) and the superordinate group
(the whole organization).
According to common ingroup identity
model (Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993), if members of different
groups focus attention on a superordinate
category that encompasses both ingroup and
outgroup in a single social group, attitudes
toward former outgroup members will become
more positive. In other words, if an individual
identifies with the superordinate group, and
the relevant outgroup (other subgroup) is seen
as part of this group, the level of superordinate
identification should be related to lowered
levels of bias.
In the context of subgroups and superordinate groups, it is necessary to make a distinction between superordinate groups with nested
subgroups and superordinate groups with crosscutting subgroups (e.g. Hornsey & Hogg,
2000). Sometimes subgroups are nested
entirely within the superordinate category. In
this study, for example, the shipyards own
employees as a subgroup are nested entirely
within the category, shipyard, especially when
compared to the crosscutting subcategor y,
shipyard subcontractor, which extends beyond the
superordinate category, shipyard, since most of
the subcontractors also have functions and
employees outside the shipyard. According to
Hornsey and Hogg (2000), however, the extent
to which a persons own subcategory and other
subcategories are seen as crosscutting with or
nested within a superordinate category may be
more dependent on situational variables and
perception than on structural reality.
Superordinate identification has proved to
be negatively related to bias in certain previous
studies (e.g. Gaertner, Rust, Dovidio, Bachman,
& Anastasio, 1994; Terr y, Carey, & Callan,
2001). For example, Terry et al. (2001) found
that identification with the new organization
after a merger was negatively related to ingroup
bias toward the merger partner. There is,
however, reason to believe that superordinate

subgroup identification

identification does not necessarily work in all


circumstances or for all subgroups. According
to Mummendey and Wenzel (1999), for
example, a successful superordinate identity
must be inclusive and it should not reflect too
strongly the characteristics of a dominant
majority group. The optimal distinctiveness
theor y (e.g. Brewer, 1993) holds that there
exists a dynamic tension between competing
drives for distinctiveness and inclusiveness, and
it has been found that individuals in large,
overly inclusive groups may be motivated to
achieve greater distinctiveness (Hornsey &
Hogg, 1999). Nevertheless, it is hypothesized in
this study that subgroup identification (identification with the subcontractor) is positively
related, and superordinate identification
(identification with the shipyard) negatively
related, to bias in intergroup evaluations
between subgroups under the same superordinate group.

Method
The research site
In order to increase cost effectiveness and flexibility, many companies in the shipbuilding
industry rely more and more on the use of
suppliers and subcontractors. Shipyards try to
concentrate on their core competencies, and
they are widely outsourcing functions that are
outside of their main business. Consequently,
they often subcontract bigger or more complete parts of the building process, and this
trend seems to be international (Malinen,
1998).
In Finland, the breakthrough of shipyard
subcontracting took place in the late 1980s and
in the 1990s, and outsourcing has gained a
more strategic position in Finnish and
European shipbuilding since 1993. For
example, subcontractors were responsible for
about half of the work in three of the worlds
largest cruise-liners, which were built in two
Finnish shipyards between 1997 and 2000.
According to previous studies on the Finnish
shipbuilding industr y (e.g. Malinen, 1998;
Niemel, 1999; Pusila, 1995), there is also
evidence that the new organization of

241

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 242

Group Processes & Intergroup Relations 6(3)

production has increased productivity at


Finnish shipyards. According to the same
studies, however, the functioning of the shipyards does not promote trust between shipyard
and subcontractor, and cooperation between
subcontractors has also remained relatively
infrequent. On the one hand, it is true that
shipyards do need the capacity and know-how
of subcontractors, for example in many longterm research and design projects, in which
mutual trust of a high degree is also required.
On the other hand, shipyards aim at reducing
their dependence on subcontractors and, as a
result of tightened competition, price and the
desire for short-term advantage, have often outweighed long-term cooperation in practice
(Pusila, 1995). Similar findings about networks
and inter-firm relations have also been
reported in studies of industries other than
shipbuilding (Niemel, 1999).
The internal organization of the target
shipyard is rather complex: it consists of the
shipyards own departments and numerous
subcontractors side by side. Although the subcontractors may also have functions and other
customers outside the shipyards gates, it is the
shipyard that is usually the most important
customer of a single subcontractor. As a result
of this, the subcontractors and their employees
are highly dependent on the economic success
of the shipyard and on the future job opportunities it can provide. In addition, the simultaneous use of subcontracted labor and layoffs
or the threat of layoffs among the shipyards
own workforce has often caused conflicts and
distrust between management and the union.
There have also been strikes against the
expanded use of subcontracted labor (Pusila,
1995).
This study is based on two samples gathered
in 1996 in one of the four Finnish shipyards.
Access to the shipyard was obtained initially
through the Personnel Manager, and subsequently through the representatives of each
subcontractor. All subcontractors with employees at the shipyard in the beginning of
March 1996 were informed about the study,
and finally 23 of them agreed to participate in
it. Questionnaires for this Sample 1 were

242

delivered to all the employees who worked at


the shipyard under the participating subcontractors. The respondents mailed the completed surveys directly back to the researchers.
Of the 201 questionnaires delivered, 113 usable
ones were returned and the effective response
rate was 57%. No questionnaires at all were
returned by employees of three subcontractors.
The Sample 1 respondents thus represented 20
different subcontractors. Because of practical
limitations there was no way of comparing
respondents with non-respondents in Sample 1.
In order to preserve the anonymity of the
shipyard, its representatives were assured that
no figures concerning the total numbers of
employees or subcontractors will be presented.
Sample 2 consisted of the shipyards own
workforce. Questionnaires were randomly
distributed to employees in four different
sections, and completed surveys were returned
through the company mail system. The four
sections (two production and two planning
sections) were chosen in collaboration with the
Personnel Manager in order to match the two
samples in terms of the similarity in nature of
the job. Of the 266 questionnaires delivered,
168 usable ones were returned and the effective
response rate in Sample 2 was 63%.

Measures
Identification Based on Ashforth and Mael
(1989; Mael & Ashforth, 1992), organizational
identification is defined in this study as follows:
Organizational identification is the perception
and sense of oneness with or belongingness to
an organization, where the individual defines
him or herself in terms of the organization(s)
in which he or she is a member. In both
samples identification was measured by using
the same slightly modified 5-item version of
Mael and Ashforths (1992) Organizational
Identification Questionnaire. The original 6item version of the scale also contained the
item This organizations successes are my
successes, which was omitted from the scale in
this study. In this research context, the subcontractors were very dependent on the economic
success of the shipyard and there was a danger

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 243

Lipponen et al.

subgroup identification

Table 1. Confirmatory factor analysis for subgroup identification and shipyard identification.
Sample 1: Subcontracted employees (N = 113)
Factor
loadings

F1
F2

Items
Subgroup identification
1. When someone criticizes (name of the subcontractor), it feels like a
personal insult.
2. I am very interested in what others think about (name of the
subcontractor).
3. When I talk about (name of the subcontractor), I usually say we rather
than they.
4. When someone praises (name of the subcontractor), it feels like a personal
compliment.a
5. If a story in the media criticized (name of the subcontractor), I would feel
embarrassed.
Shipyard identification
1. When someone criticizes (name of the shipyard), it feels like a personal
insult.
2. I am very interested in what others think about (name of the shipyard).a
3. When I talk about (name of the shipyard), I usually say we rather than
they.
4. When someone praises (name of the shipyard), it feels like a personal
compliment.
5. If a story in the media criticized (name of the shipyard), I would feel
embarrassed.
a

.93
.74
.65

.72

.67

.36
.90
.63

Items excluded from the final solution.

that this item would not tap identification with


two different foci.
One distinctive feature of this scale compared to many previous identification scales is
that most items operationalize identification as
some kind of personal sensitivity to outsiders
views of the group. If outsiders views, positive
or negative, are taken ver y personally, it is
considered an indicator of a strong sense of
belongingness. As previously noted, Turner
(1999) has criticized some previous identification measures because they merely measure
personal identification and not social identification. It could be argued that Mael and
Ashforths (1992) identification scale, although
far from perfect, taps something essential of the
social aspect of social identification, perhaps
even better than many previous scales. This

conceptualization and operationalization also


distinguishes identification from Tajfels (1978)
original formulation and Brown et al.s (1986)
identification scale. For instance, evaluative
dimensions, such as a personal sense of pride in
the group membership, are not included in our
conceptualization of identity. The identification items in Sample 1 (see Table 1) were
tailored to each of the 20 subgroups.
In order to investigate identification with the
subgroup and the superordinate group in
Sample 1, confirmatory factor analysis was performed using the AMOS program (Arbuckle &
Wothke, 1999), allowing the factors to be correlated. When the five items that were supposed
to measure subgroup identification and the five
items for shipyard identification were included
in a two-factor solution, they showed a better fit

243

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 244

Group Processes & Intergroup Relations 6(3)

(GFI (goodness of fit index) = 0.867, RMSEA


(root mean square error of approximation) =
0.124, AIC (Akaike information criteria) =
134.5) than the one-factor solution (GFI =
0.783, RMSEA = 0.176, AIC = 196.6). However,
in the two-factor solution one item from the
subgroup identification scale and one from the
shipyard identification scale had high modification indices. After these items were excluded,
the two-factor solution considerably improved
(GFI = 0.952, RMSEA = 0.046, AIC = 57.5) (2
(19, N = 113) = 23.5, p > .10), compared to the
one-factor solution (GFI = 0.815, RMSEA =
0.183, AIC = 127.2) (2 (20, N = 113) = 95.2, p
< .001). The final scales in Sample 1 consisted
of four items for subgroup identification ( =
.84) and four items for shipyard identification
( = .70).
To facilitate comparisons between the two
samples, shipyard identification was measured
with the same four items in Sample 2 ( = .79).
Because of practical limitations, subgroup
identification (identification with the shipyards own employees) was not measured in
Sample 2.
Intergroup bias Intergroup bias was operationalized in terms of the difference between
the evaluative ratings of the ingroup and of the
outgroup. In both samples, scales of attitudes
toward the work motivation and qualifications
of the members of each subgroup (six items)
were developed for this study (e.g. (name of
the subcontractor) workers are usually ver y
skillful and competent Employees of (name of
the subcontractor) are not usually ready to
expend extra effort unless rewarded (R)). It
was anticipated that intergroup comparison
would be displayed on these dimensions in this
context, although there was no evidence or
reason to believe that there would be objective
differences between the groups. In Sample 1 the
same six questions were asked about the
ingroup (employees of their own subcontractor
firm,  = .77), and outgroup 1 (the shipyards
own workers,  = .76) and outgroup 2 (the
employees of other subcontractors at the
shipyard,  = .65). In Sample 2 we used the same
questions as in Sample 1. The same six questions

244

were asked about the ingroup (shipyards own


workers,  = .79) and the outgroup (employees
of subcontractors at the shipyard,  = .76).

Results
Before going into the main analysis a brief
description of the two samples is given. In
Sample 1, the mean tenure with the current
subcontractor was 4.5 years, ranging from one
month to 23 years, and 99% of the respondents
were males. The total number of employees of
the subcontractor firms ranged from 3 to 192
(M = 78.62), and the number of employees of
each subcontractor working at the shipyard
ranged from 1 to 37 (M = 19.88). In Sample 1
each respondents own ingroup was clearly
smaller than outgroup 1 (the shipyards own
employees) and outgroup 2 (employees of
other subcontractors working at the shipyard).
Among the shipyards own employees (Sample
2), the mean tenure was 14.7 years, ranging
from two months to 36 years, and of the respondents, 97% were males. In Sample 1 respondents were less identified with the shipyard (M
= 11.06 ) than in Sample 2 (M = 12.15, twotailed t-test: t = 2.43, p < .05).
In order to answer the question whether the
respondents in the two samples favored their
ingroup, we compared the mean evaluations of
the ingroup and outgroup(s) in the two
samples. The employees of different subcontractors (Sample 1) significantly favored their
ingroup (M = 20.84) over both the shipyards
own workforce (M = 18.53) (two-tailed t-test: t =
5.70, p < .001) and employees of other subcontractors (M = 20.19) (two-tailed t-test: t = 2.13,
p < . 05). They also rated other subcontractors
employees significantly higher than the shipyards own workforce (two-tailed t-test: t = 4.16,
p < .001). In Sample 2, the shipyards own
workers rated their ingroup (M = 19.51) higher
than the outgroup (M = 18.62), although the
difference was not significant (two-tailed t-test:
t = 1.90, p > .05).

Group identifications and ingroup bias in the


two samples
In Sample 1 the AMOS 4.0 structural equation

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 245

Lipponen et al.

subgroup identification

Figure 1. Regression with three unobserved variables: subgroup identification, shipyard identification and
ingroup bias toward other subgroups. Sample 1: Subcontractors employees (N = 113).

modeling approach (Arbuckle & Wothke,


1999) was used to examine the regression
model with three unobserved variables: subgroup identification and shipyard identification, both with four indicators, and
intergroup bias toward other subgroups, with
two indicators (bias toward Outgroup 1 and
bias toward Outgroup 2). The results with standardized estimates are presented in Figure 1.
The fit indexes indicated a very good fit (GFI =
0.939, RMSEA = 0.046, AIC = 85.7) (2 (32, N =
113) = 39.7, p > .10). As hypothesized, subgroup
identification was positively related ( = .84, p <
.001) and shipyard identification was negatively
related to ingroup bias ( = .35, p < .05). The
standardized estimate for subgroup identification seems to be extraordinarily high
compared to previous studies, in which the
correlation between identification and bias has
usually been around .20 or .30. However, we
cannot directly compare the standardized estimates between the latent variables (where the
measurement error is taken into account) to
direct correlations between observed variables.
For example, in this case, the direct correlation
between subgroup identification and ingroup
bias toward other subgroups is only .49 (p <

.001). Based on previous work of Huo, Smith,


Tyler, and Lind (1996) and Mummendey and
Wenzel (1999) it was reasonable to anticipate
possible interaction effects between subgroup
identification and shipyard identification.
Interaction was tested and it turned out to be
nonsignificant.1
Among the shipyards own employees
(Sample 2), identification with the shipyard was
related to more positive evaluations of the
ingroup (r = .52, p < .001), but it was totally
unrelated to outgroup evaluations (r = .01, ns).
The AMOS 4.0 structural equation modeling
approach (Arbuckle & Wothke, 1999) was used
to examine the regression model with two
unobserved variables: shipyard identification,
with four indicators, and intergroup bias
toward other subgroups, with one indicator
(bias toward subcontractors employees). The
results with standardized estimates are presented in Figure 2. The fit indexes indicated a
very good fit (GFI = 0.994, RMSEA = 0, AIC =
22.8) (2 (5, N = 168) = 2.8, p > .50). Contrary
to the findings obtained from Sample 1, among
the shipyards own employees identification
with the shipyard was positively related to
ingroup bias ( = .39, p < .001).

245

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 246

Group Processes & Intergroup Relations 6(3)

Figure 2. Regression with two unobserved variables: shipyard identification and ingroup bias toward other
subgroups. Sample 2: Shipyards own employees (N = 168).

Summary and discussion


In general, the results based on the sample of
employees of shipyard subcontractors (Sample
1) clearly supported the main hypotheses
derived from SIT that the amount of subgroup
identification is positively related to ingroup
bias toward other subgroups at the shipyard.
Although in line with the theory, the results
differ somewhat from those of many previous
studies carried out in different work organizations, in which a positive relationship was not
found (e.g. Brown & Williams, 1984; Brown et
al., 1986; Oaker & Brown, 1986). Among other
things, these previous results have been attributed to methodological problems: when
responding to the subgroup identification
scale, the respondents actually identified with
the superordinate categor y or professional
groups rather than with their own particular
subgroup (e.g. Brown & Williams, 1984). This
problem was taken into account in the present
study by measuring identification with these
two different foci separately at the same time.
Confirmatory factor analysis suggested that this
could be done empirically, and the results of
the present study clearly point to the usefulness
of the approach adopted. It seems necessary to
make the distinction between subgroup and
superordinate identification, and to take it into
consideration, particularly when the focus is on
intergroup relations in rather complex organizational contexts.
It should also be noted that there are certain
differences in research context between this
study and the ones previously referred to. The
target organizations in earlier studies (e.g.
Brown & Williams, 1984; Brown et al., 1986)

246

were more or less traditional line organizations,


and the ingroups of interest included different
departments and sections in the same organization, which (supposedly) had clear external
boundaries. The present study, in turn, was
carried out in a very different context. The
shipyard was a combination of a traditional line
organization and a complex network of different subcontractors. Subcontractors are both
juridically and economically separate from the
shipyards own line organization, and from
each other. Consequently, the internal division
into different subgroups is perhaps sharper in
the shipyard than it is in traditional line
organizations with different departments. In
addition, the internal division is clearly more
visible, since the employees of each subcontractor have workclothes of different colors
with the firms name on the back. In this kind
of environment it may be easier for the employees of different subcontractors to make a clear
distinction between ingroups and outgroups,
and to handle their separate subgroup and
superordinate identities. This may also partly
explain why subgroup identification turned out
to be a stronger predictor of ingroup bias than
it was in previous studies carried out in work
contexts.
There are also certain differences in the
operationalization of both identification and
ingroup bias between this study and previous
studies. Most of the studies carried out in work
contexts have used the identification scale
devised by Brown et al. (1986). In both samples
of this study, identification was measured on
scales based on Mael and Ashforths (1992)
Organizational Identification Questionnaire
(OID), which sets it apart from previous

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 247

Lipponen et al.

studies. It is possible, therefore, that some of


the exceptionally high correlations between
identification and bias found here are a result
of the identification scale used. This is pure
speculation for the moment, however, because,
at least to our knowledge, there are no other
studies on identificationbias relationship in
which Mael and Ashforths (1992) identification scale might have been used.
The original aim with Sample 1 was to
measure subgroup and shipyard identification
on identical scales. Based on confirmator y
factor analyses, different items were then
omitted from the two scales. It could be argued
that, as a result, the two measures were then no
longer strictly speaking comparable, or at least
that they could not be considered absolutely
equivalent measures of the same thing at
different levels of abstraction. However, the
major objective of the study was that the
measures used would distinguish identification
with the two separate groups as reliably as
possible, which could be considered a valid
reason to delete different items from the scales.
This would, of course, eliminate the possibility
of precisely comparing the identification levels
with each other in Sample 1, which in itself
would also have been interesting.
The measures of ingroup bias used in this
study clearly differ from those used in some
previous studies (e.g. Brown et al., 1986), in
which bias was operationalized and measured
in terms of the respondents subjective estimation of the contribution to the company of
the ingroup and each outgroup, as opposed to
the estimated work motivation and qualifications of the group members. It is possible that
the dimensions of intergroup comparison used
here are more relevant in this context than the
dimensions used in previous studies in other
contexts. It is also possible that values and
norms in this context do not prohibit discrimination between the subgroups (see Turner,
1999). Consequently, the stronger positive
association between subgroup identification
and bias found in this study may result partly
from the methodological and contextual differences.
The most important finding from Sample 1

subgroup identification

was, however, that a significant negative


relationship between shipyard identification
and evaluative bias toward the other subgroups
was found. It seems that for the employees of
the subcontractors the shipyard clearly served
as a true superordinate categor y, which
includes both the subcontractors employees
and the shipyards own workers. This result is
noteworthy for at least two reasons. First, there
are only a few published studies in which the
relationship between the level of superordinate
identification and ingroup bias has been tested
(e.g. Gaertner et al., 1994; Terry et al., 2001).
Second, although laborator y studies have
usually shown that bias is virtually eliminated
against those who share at least one category
membership of the two equally important
categories (Vanbeselaere, 1991), field studies
have often revealed that, in real life, categorizations tend to be of unequal significance (e.g.
Brewer, Ho, Lee, & Miller, 1987). Although it
has been proposed (e.g. Hewstone, Islam, &
Judd, 1993) that it is the personal importance
of the categorization that may explain this
significance, the effects of the individuals level
of identification with multiple social groups has
not usually been measured in studies on
crossed categorization. Consequently, the
present study offers one possible explanation
for this unequal significance of social
categories. It seems that it is the individuals
level of simultaneous identification with
different social groups that explains the bias.
Previous research based on the common
ingroup identity model (Gaertner et al., 1993)
has supported the idea that if subgroup
members see themselves as part of a common
superordinate categor y, intergroup relations
would be improved (Gaertner et al., 1994;
Terry et al., 2001). Mummendey and Wenzel
(1999), in turn, have proposed that ingroup
bias between the subgroups would be greatest
when both subgroup and superordinate
identification are high. In view of the findings
from Sample 1, the results seem to give clear
support to the idea that focusing on a superordinate category may, indeed, be a successful
strategy for improving subgroup relations, and
in this regard our results are in contradiction to

247

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 248

Group Processes & Intergroup Relations 6(3)

Mummendey and Wenzels (1999) predictions.


As the results based on Sample 2 indicate,
however, the positive effects of superordinate
identification on subgroup relations may be
limited to only some of the subgroups within a
certain superordinate group.
Among the shipyards own workers (Sample
2), identification with the shipyard was clearly
related to increased levels of ingroup bias
against the subcontractors employees. There
was also a strong positive correlation between
identification with the shipyard and ingroup
evaluations, but virtually zero correlation
between shipyard identification and evaluations
of subcontractors employees. These results can
be interpreted in the light of the recent work of
Mummendey and Wenzel (1999) and Hornsey
and Hogg (2000).
People may differ in how they construe and
categorize the social world, and there may also
be disagreement over adequate social categorizations between different groups (Hornsey &
Hogg, 2000; Mummendey & Wenzel, 1999). For
example, some subgroups may actually project
their subgroup identities onto the shared superordinate category, and it has been proposed
that this sense of ownership over the superordinate category may be especially strong for
subgroups with high status or power. Different
subgroups may therefore hold slightly different
conceptions about the nature and the extent of
a certain superordinate category, and about
what particular subcategories that superordinate category consists of. It is also possible that
some subgroups may actually project their
subgroup identity onto superordinate identity
and see the superordinate group identity as
synonymous with their own subgroup identity
(Hornsey & Hogg, 2000; Mummendey &
Wenzel, 1999).
If we think about the shipyard studied, which
consists of both the shipyards own workforce
and employees of numerous subcontractors
working there, it is not at all clear whether the
shipyards own workers consider the subcontracted workers an integral part of the
shipyard (although objectively other subgroups under the ver y same superordinate
group). For them, as a permanent and a

248

higher-status subgroup in this context, the


category shipyard does not necessarily include
the rather heterogeneous and constantly
changing subcategor y of subcontracted
workers. Instead, shipyard may actually be
interpreted as synonymous with their own subgroup category the shipyards own workers.
This may also explain why shipyard identification was positively related to bias and ingroup
evaluations, and unrelated to outgroup evaluations among the shipyards own workers.
Unfortunately we did not directly measure the
degree to which the respondents in Sample 2
categorized subcontractors employees as part
of the shipyard. In future studies on superordinate identification, it would be important to
take this into consideration.
Another potential weakness of this study is
that subgroup identification was not measured
in Sample 2, and, as a result, it was not possible
to perform parallel analyses for the two
samples. This means that the estimated paths
between shipyard identification and ingroup
bias in the two models (Figures 1 and 2) are not
strictly speaking comparable. Although this
does not jeopardize the main results here,
because the shipyard identification scales and
the evaluative rating scales in the two samples
were identical, we must be careful not to draw
too strong conclusions from our results. In
addition, we cannot make reliable judgments
about the causal direction of the significant
relationships found between the variables based
on correlational studies such as this one. It is
obvious that we need both controlled laboratory experiments and longitudinal field studies
in order to get a clearer picture of the causal
dynamics between subgroup identification,
superordinate identification and ingroup bias.
According to Mummendey and Wenzel
(1999) superordinate identity fails to improve
subgroup relations if a subgroup regards itself
as it alone defines the whole. Our findings also
point to this direction and imply that the same
social category may simultaneously serve as a
superordinate category for certain subgroups,
but it may also be synonymous with the
subgroup category for some other subgroups.
Consequently, we should be very careful and

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 249

Lipponen et al.

sensitive to the given social context when


deciding whether to use superordinate identification as a strategy for improving subgroup
relations.
Although the results of this study are based
on rather small and unique samples, the
findings may still have broader implications.
For example, the boundaries and the extension
of many real-world superordinate categories
(e.g. nationality) are quite often rather unclear,
and which subgroups (e.g. ethnic minorities)
are included and which are not may often be
dependent on perceptions and on the perceivers subgroup memberships (Mummendey
& Wenzel, 1999). In order to get a clearer
picture of the complicated relationship
between superordinate identification and intersubgroup bias, we need much more research
on the various strategies people use when
defining the boundaries and extension of realworld superordinate categories in different
everyday contexts.

Note
1. Interaction was tested with LISREL 8.50 following
the principles suggested by Jreskog & Yang
(1996) and Yang Jonsson (1998). Applying
parceling, two variables were created for both
subgroup identification and shipyard
identification. To indicate the interaction one
product variable was made. The fit of the model
was excellent (df = 22, 2 = 20.97, RMSEA = 0, CFI
= 1.0 and GFI = 0.96). The critical gamma
coefficient for interaction, however, was 0.125
(standard error = 0.104 and z value = 1.20).
Several other analyses were also made (e.g. with
four product variables and without parceling and
also using CFA factor scores) with a similar result.

Acknowledgments
This article is based on the first authors doctoral
thesis (Lipponen, 2001) conducted under the
second authors supervision. We would like to thank
the examiners of the thesis, Naomi Ellemers and
Rupert Brown, and the anonymous reviewers of this
manuscript, for their beneficial comments, and
Pertti Keskivaara for his help on statistical analysis.

subgroup identification

References
Arbuckle, J. L., & Wothke, W. (1999). Amos users
guide version 4.0. Chicago, IL: SmallWaters
Corporation.
Ashforth, B. E., & Mael, F. (1989). Social identity
theory and the organization. Academy of
Management Review, 14, 2039.
Branscombe, N. R., Wann, D. L., Noel, J. G., &
Coleman, J. (1993). In-group and out-group
extremity: Importance of threatened social
identity. Personality and Social Psychology Bulletin, 17,
381388.
Brewer, M. B. (1993). Social identity, distinctiveness,
and in-group homogeneity. Social Cognition, 11,
150164.
Brewer, M. B., Ho, H. K., Lee, J. Y., & Miller, N.
(1987). Social identity and social distance among
Hong Kong schoolchildren. Personality and Social
Psychology Bulletin, 13, 156165.
Brown, R., Condor, S., Mathews, A., Wade, G., &
Williams, J. (1986). Explaining intergroup
differentiation in an industrial organization.
Journal of Occupational Psychology, 59, 273286.
Brown, R., & Williams, J. (1984). Group
identification: The same thing to all people?
Human Relations, 37, 547564.
Gaertner, S. L., Dovidio, J. F., Anastasio, P. A.,
Bachman, B. A., & Rust, M. C. (1993). The
common ingroup identity model: Recategorization
and the reduction of intergroup bias. European
Review of Social Psychology, 4, 226.
Gaertner, S. L., Rust, M., Dovidio, J. F., Bachman, B.
A., & Anastasio, P. A. (1994). The contact
hypothesis: The role of a common ingroup identity
on reducing intergroup bias. Small Group Research,
25, 224249.
Hewstone, M., Islam, M. R., & Judd, C. M. (1993).
Models of crossed categorization and intergroup
relations. Journal of Personality and Social Psychology,
64, 779793.
Hinkle, S., & Brown, R. (1990). Intergroup
comparisons and social identity: Some links and
lacunae. In D. Abrams & M. Hogg (Eds.), Social
identity theory: Constructive and critical advances
(pp. 4470). Hemel Hempsted: Harvester
Wheatsheaf.
Hornsey, M. J., & Hogg, M. A. (1999). Subgroup
differentiation as a response to an overly-inclusive
group: A test of optimal distinctiveness theory.
European Journal of Social Psychology, 29, 543550.
Hornsey, M. J., & Hogg, M. A. (2000). Assimilation
and diversity: An integrative model of subgroup

249

02 GPI 6-3 Lipponen (dm/s)

6/9/03

9:24 am

Page 250

Group Processes & Intergroup Relations 6(3)

relations. Personality and Social Psychology Review, 4,


143156.
Huo, Y. J., Smith, H. J.,Tyler, T. R., & Lind, E. A.
(1996). Superordinate identification, subgroup
identification, and justice concerns. Psychological
Science, 7, 4045.
Jreskog, K.G, & Yang, F. (1996). Nonlinear
structural equation models: The KennyJudd
model with interaction effects. In R.E.
Schumacher & G.A. Marcoulides (Eds.), Advanced
structural equation modeling: Issues and techniques
(pp. 5778). NJ: Erlbaum.
Kelly, C. (1988). Intergroup differentiation in a
political context. British Journal of Social Psychology,
27, 319332.
Lipponen, J. (2001). Organizational identifications
Antecedents and consequences of identification in a
shipyard context. Doctoral dissertation, Department
of Social Psychology, University of Helsinki.
Mael, F. A., & Ashforth, B. E. (1992). Alumni and
their alma mater: A partial test of the reformulated
model of organizational identification. Journal of
Organizational Behavior, 13, 103123.
Malinen, P. (1998). To buy, sell, exchange and
produceA case study in the shipbuilding industry in
shipyard-subcontractor exchange . [In Finnish with an
English summary]. Turku, Finland: Publications of
the Turku School of Economics and Business
Administration.
Mummendey, A., & Wenzel, M. (1999). Social
discrimination and tolerance in intergroup
relations: Reactions to intergroup difference.
Personality and Social Psychology Review, 3, 158174.
Niemel, J. (1999). Crisis, industrial relations and
organizational learning at the Finnish shipyards. In
S. Lhteenmki, L. Holden, & I. Roberts (Eds.),
HRM and the learning organization. Turku, Finland:
Publications of the Turku School of Economics
and Business Administration.
Oaker, G., & Brown, R. (1986). Intergroup relations
in a hospital setting: A further test of social identity
theory. Human Relations, 39, 767778.
Perreault, S., & Bourhis, R. Y. (1999).
Ethnocentrism, social identification, and
discrimination. Personality and Social Psychology
Bulletin, 25, 92103.
Pusila, J. (1995). The shipyards in local labor market.
In H. Leimu & J. Pusila (Eds.), A change of
production paradigm in Finnish shipbuilding [In
Finnish with English summaries]. Turku: Finland:

250

Publications of the Turku School of Economics


and Business Administration.
Tajfel, H. (1978). Social categorization, social
identity and social comparison. In H. Tajfel (Ed.),
Differentiation between social groups: Studies in the
social psychology of intergroup relations (pp. 6176).
London: Academic Press.
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. 3347). Monterey, CA: Brooks/Cole.
Terry, D. J., Carey, C. J., & Callan, V. J. (2001).
Employee adjustment to an organizational merger:
An intergroup perspective. Personality and Social
Psychology Bulletin, 27, 267280.
Turner, J. C. (1999). Some current issues in research
on social identity and self-categorization theories.
In N. Ellemers, R. Spears, & B. Doosje (Eds.),
Social identity: Context, commitment, content
(pp. 634). Oxford: Blackwell.
Vanbeselaere, N. (1991). The different effects of
simple and crossed categorizations: A result of the
category differentiation process or of differential
category salience? European Review of Social
Psychology, 2, 274278.
Yang Jonsson, F. (1998). Modeling interaction and
nonlinear effects: A step-by-step LISREL example.
In R.E. Schumacher & G.A. Marcoulides (Eds.),
Interaction and nonlinear effects in structural equation
modeling (pp. 1742). NJ: Erlbaum.
Paper received 16 October 2001; revised version accepted
25 November 2002.

Biographical notes
is a lecturer at the department of
social psychology at the University of Helsinki. His
research interests include organizational
identification, intergroup relations and
organizational justice and values.

JUKKA LIPPONEN

is a professor at the department of


social psychology at the University of Helsinki. His
research interests include moral development,
values and national identity.

KLAUS HELKAMA

was a student of social psychology at


the University of Helsinki. She is currently
employed outside academia.

MILLA JUSLIN

Das könnte Ihnen auch gefallen