Content uploaded by Thomas Fraser Pettigrew
Author content
All content in this area was uploaded by Thomas Fraser Pettigrew on Jan 24, 2019
Content may be subject to copyright.
European Journal of Social Psychology,
Vol.
25,57-75
(1995)
Subtle and blatant prejudice in western
Europe*
T.
F.
PETTIGREW
University
of
California, Santa Cruz,
U.S.A.
and
R.
W. MEERTENS
University
of
Amsterdam, The Netherlands
Abstract
This paper develops, measures, and tests two types of intergroup prejudice-blatant
and subtle. Blatant prejudice is the traditional, often studied form; it is hot, close
and direct. Subtle prejudice is the modern form; it is cool, distant and indirect. Using
data from seven independent national samples from western Europe, we constructed
10-item scales in four languages to measure each of these varieties of prejudice. We
report the properties, structure and correlates
of
both scales across the seven samples,
and make initial checks
on
their validity. The cross-nationally consistent results support
the value of the blatant-subtle distinction as two varieties of prejudice. While they
share many correlates, their distinctive differences suggest better specification of these
correlates of prejudice. And the blatant-subtle distinction also aids in more precise
specification
of the effects of prejudice
on
attitudes toward immigrants. The paper
closes with a normative interpretation of Subtle Prejudice.
INTRODUCTION
Primary attention to intergroup prejudice has focused upon the more blatant, tradi-
tional
forms
of
the phenomenon. But in recent decades indirect forms
of
prejudice
have come to preserve racial, ethnic and religious stratification. These more subtle
*The authors presented earlier versions of this paper at the First European Congress on Psychology
in Amsterdam, June 1989; the European Social Psychology Association in Budapest, June 1990; the
Racism Conference of
Passages
in Paris, June 1991; and the International Congress on Psychology in
Brussels, July 1992. A Rockefeller Foundation Fellowship to the first author to the Bellagio Study and
Conference Center at Bellagio, Italy facilitated the paper’s final preparation. The complete questionnaires
with instructions in Dutch, English, French and German can be obtained from the authors: Kerr Hall,
University
of
California, Santa Cruz, CA 95064 USA; or Department of Psychology, University
of
Amster-
dam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands.
CCC
00462772/95/010057-19
0
1995
by John Wiley
&
Sons, Ltd
Received
19
October
1992
Accepted
30
December
1993
58
T.
F.
Pettigrew and
R.
W.
Meertens
forms are similarly described though variously named. In France it is called
‘.
. .
a new under-the-skin racism’ (Freriks, 1990); in Germany, ‘latent’ prejudice (Berg-
mann
&
Erb, 1986); in Britain, ‘the new racism’ (Barker, 1984); in the Netherlands,
‘everyday racism’ (Essed, 1984); and in the
U.S.,
‘aversive’ (Kovel, 1970), ‘symbolic’
(Sears, 1988), or ‘modern racism’ (McConahay, 1983; Pettigrew, 1989).
The common theme across all these new forms is that they are
covert
means of
expressing prejudice that differ from the old-fashioned forms. Otherwise, quite diverse
conceptualizations are advanced. Thus, Essed
(1
984) emphasizes the commonplace
quality
of
the phenomenon; Kovel
(1970)
views it as a defence against recognizing
one’s own bigotry; McConahay (1983) sees it as an attempt to attribute outgroup
disadvantage to causes other than discrimination; and Sears (1988) traces it to early
socialization.
Given the varying conceptualizations, the domain’s measurement remains prob-
lematic. From a wide domain of possibilities, investigators have drawn on distinctive
subsets of elements aimed at their own particular emphasis. Thus, though various
measures have been proposed, no indicators of these newer prejudice forms have
won general acceptance in the literature. Indeed, conflicting measures have detracted
from the empirical evidence for the phenomenon (Pettigrew,
1989).
This paper
addresses these problems in two ways. First, it offers an explicit three-component
conceptualization of this newer form of prejudice. Second, it presents scales with
solid psychometric properties to measure the two prejudice types.
In
The Nature
of
Prejudice,
Allport (1958, p. 10) defined prejudice
as
‘an antipathy
based upon a faulty and inflexible generalization’. To this useful approach that
involves both affect and cognition, we would add that prejudiced attitudes tend
to
form ideological clusters of beliefs that justify discrimination. What we shall label
blutant
and
subtle
prejudice are two contrasting expressions
of
this central phenome-
non. Blatant prejudice is hot, close and direct. Subtle prejudice is cool, distant,
and indirect.
The components of these two varieties of prejudice uncovered in social psychologi-
cal research reflect these differences in style. Two components repeatedly emerge
in work on blatant prejudice (Allport, 1958). The first involves perceived threat
from and rejection of the outgroup. In its full racist form, blatant prejudice includes
belief in the genetic inferiority of the outgroup. Such a belief explains away any
outgroup disadvantage in the society, and thus denies that discrimination exists.
The second component of blatant prejudice involves opposition to intimate contact
with the outgroup. This anti-intimacy component focuses upon an emotional resis-
tance against any intergroup sexual contact or intermarriage. It also contains a power
element. Hence, working under the supervision
of
an outgroup member is rejected.
We propose that subtle prejudice is revealed by three more covert components,
each of which is expressed in ways deemed normative and acceptable in western
societies. The first
is
the defence
of
traditional values.
Victim blaming is often involved.
Outgroup members are seen
to
act in unacceptable ways, and not to perform in
ways
necessary to succeed. What is regarded as acceptable and necessary behaviour
is construed in terms of the ingroup’s traditional values. The second component
entails
the exaggeration
of
cultural differences.
Instead
of
invoking genetic inferiority,
subtle prejudice attributes outgroup disadvantage to cultural differences. These differ-
ences are often genuine; but subtle prejudice exaggerates them through gross stereo-
types. The third component denies positive emotional responses toward the outgroup.
Subtle and blatant prejudice
59
This feature does not admit negative feelings toward the outgrouponly the more
covert
denial
of
positive emotions.
Extensive research has focused on each of these components, though our concep-
tion of subtle prejudice is the first to bring these literatures together. Consider each
in detail.
The defence of traditional values
Early research consistently found prejudice greatest on the political right (Allport,
1958).
Work in the 1940s and 1950s on the prejudice-prone ‘authoritarian personality’
found strong relationships between the syndrome and conservative ideology-
labelled ‘conventionalism’ (Adorno, Frankel-Brunswik, Levinson and Sanford,
1950). To be sure, this research was biased toward finding this result (Christie and
Jahoda, 1953). Rokeach (1960) later showed that the phenomenon could be found
in Great Britain on the political left as well. But on balance recent work supports
the original formulation. With Canadian samples, Altemeyer (1 988) found his modern
measures of authoritarianism are highly intertwined with political conservatism. And
American research with subtle measures reveals links between racism and right-wing
voting, ideology and party identification (Sears, 1988; Weigel and Howes, 1985).
This half-century of research findings underlies the first component of the Subtle
Scale-the defence of traditional values. Conservatives defend these values more
than the left. They are also more likely to scapegoat and blame outgroups who
appear to threaten these values (Allport, 1958). Consequently, such issues as minori-
ties needing to try ‘hard enough’, work their ‘way up’, and teach their children
traditional values emerge as effective, yet ostensibly non-prejudicial, indicators of
subtle prejudice.
The exaggeration of cultural differences
During the 1960s, social psychology emphasized this second component of Subtle
Prejudice. Rokeach (1960) proposed that perceived dissimilarity of beliefs was the
foundation
of
all forms of prejudice.
This bold claim sparked a flurry of research. Soon a consensus modified Rokeach’s
sweeping contention. Perceived belief dissimilarity and race are
both
important
(Smith, Williams and Willis, 1967; Stein, Hardyck and Smith, 1965; Triandis and
Davis, 1965). Prejudice overwhelmed the belief dissimilarity factor among whites
from racist areas of the American South (Smith
et al.,
1967) and when norm-violating
intimate contact was involved (Triandis and Davis, 1965). Yet one study
of
black
children in the southern
US.
revealed belief dissimilarity more important than race
for even intimate contact (Moe, Nacoste and Insko, 1981). The critical feature is
the normative demands of the social context, not intimacy
per se.
Brown and Turner
(198 1) note that Rokeach confused the interpersonal and intergroup levels of analysis,
for belief dissimilarity is an interpersonal phenomenon, while prejudice is largely
an intergroup phenomenon. Moreover, belief differences with black Americans are
simply assumed and exaggerated by white Americans when no personal information
about the blacks was supplied (Stein
et al.,
1965). It is this triggering by race and
ethnicity of attributed difference, that constitutes the core process (Dienstbier, 1972;
Moe
et al.,
1981).
60 T.
F.
Pettigrew
and
R.
W.
Meertens
Phrased in the language of this earlier work, then, the subtly prejudiced exaggerate
actual group differences and thereby enhance their assumptions of belief dissimilarity.
This process sets the outgroup aside as ‘a people apart’-wholly unlike the ingroup.
Rokeach (1 960) showed that this process further contributes to prejudicial assump-
tions and actions. Note that exaggerated cultural differences are ostensibly non-
prejudicial-the covert key to subtle prejudice; for it hardly appears prejudiced to
report on actual and obvious intergroup differences.
The denial
of
positive emotions
This component of subtle prejudice derives from experimentation on both sides of
the Atlantic. Dijker (1987) showed that Dutch respondents could easily report on
emotions they experienced in encounters with minority group members. Abelson,
Kinder, Peters and Fiske (1982), using data from American surveys, found that
positive and negative feelings toward politicians were almost independent of each
other. Moreover, they found these affect reports to be less semantically filtered and
less subject to consistency pressures than trait judgements. This finding suggests
that reports on feelings could more easily act as covert indicators of subtle prejudice.
Similarly, laboratory studies (Gaertner and McLaughlin, 1983; Dovidio, Mann
and Gaertner, 1989), using white American college subjects, found no differences
in the ratings of whites and blacks on such negative attributes as ‘lazy’. But they
found systematic biases favouring whites on such positive attributes as ‘ambitious’.
Hence, for these students, ‘blacks are not worse, but whites are better’ (Dovidio
et
al.,
1989, p. 88). These results cannot be explained by self-presentation-that
these subjects consciously withheld their racial biases for the negative traits. The
Gaertner and McLaughlin (1983, Studies
1
and
2)
experiments used a highly indirect
measure of bias based on reaction times.
With these explicit conceptions of blatant and subtle prejudice, this paper tests
the following propositions:
[i]
Two forms of prejudice-blatant and subtle--can be usefully distinguished
and measured.
[ii] Blatant and subtle prejudice, as two forms of generalized intergroup preju-
dice, will be moderately intercorrelated.
[iii]
As
dependent variables, blatant and subtle prejudice will share similar corre-
lates. But the differences between them will specify more precisely these
relationships.
[iv] As independent variables, blatant and subtle prejudice will predict different
responses to outgroups and immigrant policy.
METHOD
Samples
Our
data are from seven independent probability surveys conducted as part of the
European Community’s Euro-Barometer Survey No.
30
during the fall of 1988 in
France, the Netherlands, Great Britain and then-West Germany’. Turkish immi-
I
Our
colleagues on this project are Professor James Jackson, of the University
of
Michigan, Professor
Gerard Lemaine of I’Ecoli des Hautes Etudes en Sciences Sociales, Professor Ulrich Wagner,
of
Marburg
Universitat, and Andreas Zick
of
the Bergische Universitat, Wuppertal.
Subtle and blatant prejudice
61
grants served
as
the target outgroup for the entire German sample. But the study
drew two separate samples each in the other countries
so
that two different minorities
could serve as target outgroups. In France, one sample focused on Asians, the other
on North Africans. In the Netherlands, one focused on Surinamers, the other on
Turks. In Great Britain, one focused on West Indians, the other on Asians.
All minority respondents were removed from the data. This left 455 French respon-
dents asked about North Africans; 475 French respondents asked about Asians;
462 Dutch respondents asked about Surinamers; 476 Dutch respondents asked about
Turks; 471 British respondents asked about West Indians; 482 British respondents
asked about Asians; and 989 West German respondents asked about Turks. Details
about the sampling procedures and the full schedule of the Euro-Barometer
30
survey
are available in Reif and Melich (1991).
Measures
Blatant and subtle prejudice
We developed 10-item Likert scales to measure each of the prejudice types. Table
1
shows the scales in their English form. Standard Likert-scale scoring was used.
Thus, item responses were scored
1,
2,
4
and 5 on a strongly disagree, somewhat
disagree, somewhat agree and strongly agree dimension, with higher scores indicating
greater prejudice. No answers were assigned the individual’s mean on those scale
questions answered. This procedure was used only for those answering at least four
of the 10 questions. Removing respondents with less than four answers to the Blatant
or Subtle Scales resulted in a sample loss of less than
3
per cent.
From a pool of over 50 separate items, we chose
10
items to measure each type
on the basis of our conceptualization of the two forms and factor analyses. These
exploratory factor analyses yield in four of the seven samples the two factors held
to underlie blatant prejudice: (1) threat and rejection and (2) anti-intimacy. In the
remaining three samples, a single factor emerged.
For the 10-item Subtle Prejudice Scale, all three components held to underlie
the phenomenon emerged in five samples: (1) the defence of traditional values,
(2)
the exaggeration of cultural differences, and
(3)
the denial of positive emotions.
In two samples, the second and third components folded into one factor. Further
details of the scales’ properties are provided below.
Independent variables
Earlier research has shown consistent relationships between various measures of
prejudice and nine independent variables assessed
in
this study (Allport, 1958). First,
to check on the generality of prejudice against many outgroups, we developed a
three-item measure
of
ETHNOCENTRISM. This scale sought favourability ratings
from 0 to
100
for three diverse groups not covered by our other measures: southern
Europeans, black Africans and Jews. Asking for favourability ratings resembles our
Subtle Scale items in that they are ostensibly non-prejudicial. Indeed, in every sample
the average rating for each group was either neutral
or
favourable
(50+).
These
62
T.
F.
Pettigrew and
R.
W.
Meertens
Table
1.
The blatant and subtle prejudice scales
Threat and rejection factor items: the Blatant Scale
1.
West Indians have jobs that the British should have. (strongly agree to strongly disagree)
2.
Most West Indians living here who receive support from welfare could get along without
it
if they tried. (strongly agree to strongly disagree)
3.
British people and West Indians can never be really comfortable with each other, even
if
they are close friends. (strongly agree to strongly disagree)
4.
Most politicians in Britain care too much about West Indians and not enough about
the average British person. (strongly agree to strongly disagree)
5.
West Indians come from less able races and this explains why they are not as well
off as most British people. (strongly agree to strongly disagree)
6.
How different
or
similar do you think West Indians living here are to other British
people like yourself-in how honest they are? (very different, somewhat different, some-
what similar, or very similar)
Intimacy factor items: the Blatant Scale
1.
Suppose that
a
child of yours had children with a person of very different colour and
physical characteristics than your own.
Do
you think you would be very bothered,
bothered, bothered a little, or not bothered at all, if your grandchildren did not physically
resemble the people on your side of the family?
2.
I
would be willing to have sexual relationships with a West Indian. (strongly agree
to strongly disagree)
3.
I
would not mind if a suitably qualified West Indian person was appointed as my boss.
(strongly agree to strongly disagree)
4.
I would not mind
if
a West Indian person who had a similar economic background
as mine joined my close family by marriage, (strongly agree to strongly disagree)
Traditional values factor items: Subtle Scale
1.
West Indians living here should not push themselves where they are not wanted. (strongly
agree to strongly disagree)
2.
Many other groups have come to Britain and overcome prejudice and worked their
way up. West Indians should do the same without special favour. (strongly agree to
strongly disagree)
3.
It is just a matter of some people not trying hard enough. If West Indians would only
try harder they could be as well off as British people. (strongly agree to strongly disagree)
4.
West Indians living here teach their children values and skills different from those required
to be successful in Britain. (strongly agree to strongly disagree)
Cultural differences factor items: Subtle Scale
How different or similar do you think West Indians living here are to other British people
like yourself..
.
(very different, somewhat different, somewhat similar, or very similar)
1.
In the values that they teach their children?
2.
In their religious beliefs and practices?
3.
In their sexual values
or
sexual practices?
4.
In the language that they speak?
.
.
.
Have you ever felt the following ways about West Indians and their families living
Positive emotions factor items: Subtle Scale
here..
.
(very often, fairly often, not too often, or never)
1. How often have you felt sympathy for West Indians living here?
2.
How often have you felt admiration for West Indians living here?
ratings
were
reversed
so
that high
scores
register high ETHNOCENTRISM. Alphas
for this
scale
ranged between
+0.73
and
+0.81.
An
APPROVAL
OF
RACIST
MOVEMENTS measure asked respondents how
Subtle and blatant prejudice
63
much they approved or disapproved of ‘movements in favour
of’ or
‘opposed to
racism’. High scorers approved racist and disapproved anti-racist movements. The
third variable, INTERGROUP FRIENDS, tapped the diversity of friendship
networks. Diversity is defined broadly-‘people of another nationality’, ‘religion’,
‘race’, ‘culture’, even ‘social class’. For each category, respondents reported if they
had ‘many’ (two points), ‘a few’ (one point) or
no
friends (no points). High scorers
report highly diverse friends.
The fourth predictor, POLITICAL CONSERVATISM, is measured by self-place-
ment on
a
10-point scale ranging from the political left to right. High scorers are
more conservative. The fifth predictor, GROUP RELATIVE DEPRIVATION is
measured by a single item drawn from Vanneman and Pettigrew
(1972):
‘Would
you say that over the last five years people like yourself in [France] have been economi-
cally a lot better
off,
better
off,
the same, worse
off,
or
a
lot worse
off
than most
[North Africans] living here’. High scorers feel relatively deprived in group terms.
POLITICAL INTEREST consists of two items that ask respondents how interested
they are in politics and in European Community politics. High scorers report keen
interest on both items. To measure NATIONAL PRIDE, respondents reported how
proud they were to be British, Dutch, French or German. Finally, EDUCATION
and AGE provide social context. Past research has shown strong relationships with
both variables (Allport, 1958). We measured education by years of schooling reported
and age from reported birth date.
Dependent variables
We asked three policy questions concerning immigrants to gauge the criterion validity
of the two scales. The first asks about the RIGHTS
OF
IMMIGRANTS who are
not citizens: Should their rights be extended, restricted, or left as they are? The
second focuses on IMMIGRATION POLICY. ‘There are a number
of
policy options
concerning the presence of [Turkish] immigrants living here. In your opinion which
is the one policy that the government should adopt in the long run?’ Though the
query asks for only one response from the card listing six policies, the interviewers
marked all alternatives mentioned. At the extremes, the preferred future policy choices
ranged from not sending ‘back to their own country’ any of the outgroup to sending
them all back. The four intermediate choices varied who should be sent back: those
not born here, those ‘not contributing’
to
the economy, those ‘who have committed
severe criminal offences’, or those ‘who have no immigration documents’.
The third dependent variable concerned PREFERRED MEANS ‘to improve rela-
tions between the [British] and [non-British] living here’. We asked respondents if
they thought it ‘a good idea’ or ‘a bad idea’ to adopt each of nine different actions:
‘prosecute in the courts people who say, write, or
do
things which are racist or
anti-foreigner’; ‘promote the teaching of tolerance and mutual respect in schools’;
‘encourage contact between [the British] and others both in the neighbourhood and
in associations’; ‘learn the language
of
others’; ‘insure that people in the public
services and teachers treat [British] and [non-British] equally’; ‘make naturalization
easier’; ‘expand international exchange programmes for young people’; and ‘draw
the attention of people in the press, radio and TV to the part they can play in
eliminating racial prejudices’.
64
T.
F.
Pettigrew and
R.
W.
Meertens
Scales
Blatant Subtle
Blatant Scale
87-90?
Threat and rejection 90-94 41-68
Traditional values 3&68
80-88
Subtle Scale 48-70
7342
Intimacy 85-91 37-60
Cultural differences
0848
63-77
Positive emotions 35-56 55-68
RESULTS
Subscales*
T
and
R
Intim.
T.
Val.
Cul.
Dif.
+Emot.
7341
40-70 26-54
53-67
59-71
70-93
0-5 0940 2143
57-72
29-50 35-52 30-44 12-31
61-73
Properties
of
the prejudice scales
Table
2
provides the range of correlations across the seven samples between the
two principal scales and their five sub-scales.
In
the diagonal cells, Table
2
also
provides the Cronbach alphas for all seven scales. Though four different response
formats are employed, the alphas for the Blatant and Subtle Scales reveal adequate
reliabilities across all samples. The alphas, however, are consistently higher for the
Blatant Scale (0.87 to 0.90) than the factorially more complex Subtle Scale (0.73
to
0.82)’.
Surprisingly, the alphas tend to be lower for the Dutch samples
on
which
the scales were originally devised. The sub-scales’ alphas are lower but often adequate,
even for the two-item Positive Emotion Scale.
The correlations between the Blatant and Subtle Scales range between
+0.70
in
France to
+0.48
in the Netherlands. This sharing on average of
40
per cent common
variance, given measurement error, meets the expectation that the two scales would
be moderately interrelated.
Take note of the positive emotions sub-scale. We asked the respondents how often
they had ever felt two positive (admiration and sympathy) and two negative (irritation
and fear) emotions about the minority group in their midst. These survey findings
strongly confirm the laboratory results of Dovidio
et
al.
(1989). The negative emo-
tions, fear and irritation, do
not
scale
in
either the blatant or subtle measures. But
the two positive emotions relate well to the other items of the Subtle Scale. Thus,
respondents scoring high on the other Subtle Scale items do not report more negative
feelings about minorities than others. But they do report
less
often feeling admiration
and sympathy for minority groups.
’The scale mean substitution for missing items inflates these alphas. When recalculated for
just
those
respondents who answered all items of each scale, the effect was minimal. The Blatant Scale alphas
now ranged from
0.84
to
0.89,
the Subtle from 0.70 to 0.81.
Subtle and blatant prejudice
65
The structure
of
the prejudice scales
Is there empirical justification for conceptualizing blatant and subtle as two types
of prejudice? We offer two types of evidence. Here we present the results of the
confirmatory factor analyses of the two scales. Later we shall note that the two
scales have distinctive relations with issues regarding immigrants.
Table
3
presents the results of confirmatory factor analyses using
EQS
(Bentler,
1989) on all seven independent samples. Given the tentativeness with which structural
models must be evaluated (Bollen, 1989, Chapter
7),
replication across all seven
samples is necessary to compare the models. Four competing models are tested.
(A) The
one-factor model
tests if the correlational matrix
of
all
20
items is best
represented as simply measuring one latent factor-prejudice. (B) The
uncorrelated
two-factor model
tests
if
the data form two separate, uncorrelated latent factors-
blatant and subtle prejudice.
(C)
The
correlated two-factor model
tests if the item
matrix is best described by two separate and correlated latent factors. (D) The
second-
order hierarchicalmodel
tests if the matrix can be best accounted for by two first-order
latent factors (blatant and subtle) that load equally and are subordinate to a second-
order latent factor (prejudice)’.
Table
3’s
key finding is the support it offers for a two-factor model of prejudice.
In six samples, the one-factor and uncorrelated two-factor models provide signifi-
cantly inferior fits than the correlated and hierarchical models. The exception is
the German sample, for which the hierarchical model remains superior.
Also in Table
3,
the hierarchical model offers the significantly best fit in the German
and Dutch samples. In a fourth sample, the British asked about West Indians, the
hierarchical model is also significantly the best; but this findings is limited by a
conditional constraint. For this one sample, the disturbance (error) parameter of
the subtle prejudice latent factor is constrained at its lower bound. For the remaining
samples from Britain and France, the correlated two-factor model provides a fit
equal to that of the hierarchical model. Thus, in all seven tests, the hierarchical
model is equal or superior to the competing models. Moreover, the hierarchical
model attains an acceptable median of 0.926 across the seven samples for the compara-
tive index.
Yet these differences between the correlated two-factor and hierarchical models
are not
so
large and consistent as to render a definitive choice between them. We
shall later explore a Guttman-type alternative that views subtle prejudice as an inter-
mediate category between blatant prejudice and tolerant egalitarianism.
The correlates
of
the prejudice scales
Table
4
provides the mean effect sizes from nine-variable regressions for the major
predictors of the two scales. By averaging the effect sizes across the samples, we
see the relative strength of each predictor of the two scales holding constant the
other predictors. While all these independent effect sizes are highly significant
(p
<
0.000
l), only three
8s
are relatively large-both the ETHNOCENTRISM relation-
Transformations normalized all item distributions and equated variances. Two conditions applied to
all models: error terms were equated
for
three sets of items that shared identical response formats; and
no blatant items were predicted by the subtle factor or
vice
versa.
For the hierarchical models, an equality
constraint was entered
for
the blatant prejudice
and
subtle prejudice paths (Bentler, 1989, pp.
3941).
The
estimation method used
for
all models was maximum likelihood.
66
T.
F.
Pettigrew
and
R.
W.
Meertens
Table
3.
Goodness
of
fit indices
for
four
confirmatorv
factor
analvtic models*
Model Chi-sq. df Chi-sq.ldf Bentler’s
(1989)
fit indices
Normed Non-normed Compara-
t
ive
The British about West Indians
(N
=
471)
[A] One-factor
500.0 136
[B] Uncorrelated two-factor
744.9 137
[C] Correlated two-factor
456.7 136
[D] Second-order
hierarchical?
445.4 134
The British about Asians
(N
=
482)
[A] One-factor
419.9 136
[B] Uncorrelated two-factor
631.8 137
[C] Correlated two-factor
361.6 136
[D] Second-order
hierarchical
360.0 134
[A] One-factor
368.5 136
[B] Uncorrelated two-factor
442.6 137
[C] Correlated two-factor
296.1 136
[D] Second-order
hierarchical
279.1 134
[A] One-factor
494.6 136
[B] Uncorrelated two-factor
569.8 137
[C] Correlated two-factor
470.1 136
[D] Second-order
hierarchical
309.4 134
[A] One-factor
412.9 136
[B] Uncorrelated two-factor
740.4 137
[C] Correlated two-factor
385.0 136
[D] Second-order
hierarchical
383.6 134
[A] One-factor
465.1 136
[B] Uncorrelated two-factor
712.7 137
[C] Correlated two-factor
423.5 136
[D] Second-order
hierarchical
420.8 134
[A] One-factor
980.3 136
[B] Uncorrelated two-factor
1637.9 137
[C] Correlated two-factor
1120.7 136
(D] Second-order
hierarchical
698.5 134
The Dutch about Turks
(N
=
476)
The Dutch about Surinamers
(N
=
462)
The French about North Africans
(N
=
455)
The French about Asians
(N
=
475)
The Germans about Turks
(N
=
989)
3.68
5.44
3.36
3.32
3.06
4.61
2.66
2.69
2.71
3.23
2.18
2.08
3.64
4.16
3.46
2.31
3.04
5.40
2.83
2.86
3.42
5.20
3.11
3.14
7.21
13.96
8.24
5.21
0.866
0.801
0.878
0.881
0.871
0.806
0.889
0.889
0.878
0.853
0.902
0.907
0.843
0.819
0.851
0.902
0.890
0.803
0.898
0.898
0.868
0.798
0.880
0.881
0.869
0.782
0.851
0.907
0.859
0.762
0.874
0.875
0.870
0.716
0.897
0.895
0.885
0.850
0.921
0.927
0.830
0.797
0.842
0.916
0.892
0.766
0.903
0.901
0.862
0.761
0.880
0.878
0.839
0.715
0.812
0.891
0.897
0.829
0.910
0.912
0.907
0.838
0.926
0.926
0.918
0.892
0.943
0.947
0.879
0.854
0.887
0.941
0.923
0.831
0.930
0.930
0.901
0.828
0.914
0.914
0.885
0.795
0.865
0.923
~ ~ ~ ~~~
*All
28
chi-squares are significant
(p
<
0
001)
Using the Likelihood Ratio
(or
chi-square difference) Test (Bollen,
1989, p 292), the chi-square of the hierarchical model is significantly different from those of the other models in each
sample
(p
<
0
001)
except with the correlated two-factor model
[C]
for
British opinions of Asians and the French
opinions of both North Africans and Asians
+For
this model only, the disturbance parameter for the subtle latent factor was constrained at the lower bound
Subtle and blatant prejudice
67
ships and RACIST MOVEMENT APPROVAL with blatant prejudice. However,
many of the average effects are reasonably robust, with
ds
ranging from 0.21 to
0.35
(Cohen, 1977). All but three variables predict both scales equally well.
Table 4. Mean effect sizes of major predictors for blatant and subtle prejudice across samples*
Variables Blatant racism scale Subtle racism scale
Cohen’s
Cohen’s
d
r
Fail safe
d
I
Fail
safe
no.
no.
Ethnocentrism +0.72 +0.34 1168 +0.54 +0.26
Racist movement approval +0.557 +0.27 682 +0.21t
+0.10
Political conservatism +0.28 +0.14 173 +0.25t +0.12
National pride +0.17
+0.08
64 f0.177
+0.08
Intergroup friends
-0.31
-0.15 222 -0.35 -0.17
Group relative deprivation +0.29t
+O.
14 188 +0.12 +0.06
Political interest -0.217
-0.11
113 -0.19 -0.09
Education -0.26 -0.13 161 -0.23 -0.11
Age +0.13t +0.07 34 +0.16t
+0.08
Age (without U.K. samples) +0.24t +0.12
58
+0.34t +0.17
*All
effects are significant at least at the
level
ofp
<
0.0001.
?Indicates
that
the effect sizes are significantly heterogeneous atp
<
0.05.
666
87
287
130
26
59
79
124
48
123
For
both scales and all samples, the three-item measure
of
ETHNOCENTRISM
proves the major predictor (median standardized beta weights-SB
=
+0.29 for
blatant and +0.27 for subtle).
A
more direct measure of generalized hostility to
outgroups, APPROVAL
OF
RACIST MOVEMENTS, relates primarily to the Bla-
tant Scale (median
SB
=
+0.20). This predictor is especially strong in two samples:
the British on Asians and the French on North Africans (median
SB
=
+0.29).
These results may reflect the mobilizing effects
of
such movements as the National
Front in France. Such mobilization could increase the correlation with blatant preju-
dice by linking the variables ideologically and furthering the legitimization of blatant
prejudice in the nation’s political discourse.
Those who report having more and diverse INTERGROUP FRIENDS score signi-
ficantly lower on blatant (median
SB
=
-0.12) and especially subtle prejudice in
all samples (median
SB
=
-0.16). POLITICAL CONSERVATISM is another signi-
ficant predictor of both the Blatant and Subtle Scales (median
SBs
=
+0.13
and
+0.11).
GROUP RELATIVE DEPRIVATION is important for blatant prejudice (median
SB
=
+0.10).
As in American results (Vanneman and Pettigrew, 1972), those who
feel that people like themselves are slipping behind the minority group are more
blatantly prejudiced across all samples. We also tested
a
measure of Individual Rela-
tive Deprivation: ‘Would you say that over the last five years
you
have been economi-
cally a lot better
off,
the same, worse off, or a lot worse off than other [Germans]
like yourself?’ Also as in previous work, this variable did not relate to either type
of prejudice in the multivariate tests.
POLITICAL INTEREST relates consistently negatively to both prejudice mea-
sures, reaching statistical significance in 10 of 14 comparisons (median
SB
=
-0.09).
68
T.
F.
Pettigrew and
R.
W.
Meertens
NATIONAL PRIDE explains further variance in half the samples (median
SB
=
+0.06).
The British samples yield the three largest relationships (median
SB
=
+0.15).
EDUCATION proves to be an important negative correlate, with the less educated
consistently higher on both prejudice scales (median
SB
=
-0.09). AGE, however,
yields contradictory results. In the samples from France and the Netherlands, older
respondents express significantly more blatant prejudice (median
SB
=
+O.
15)-the
typical result in past research (Allport, 1958). But among the British, especially
those asked about West Indians, it is the young who express the most blatant prejudice
(median
SB
=
+0.24).
Internal checks and inquiries with the data collection agency
indicate this rare finding is not an artifact of sampling or non-response biases. If
replicated, this result suggests ominous difficulties for future British race relations.
We also calculated average effect sizes for each of the five subscales, and obtained
more specified results. Thus, POLITICAL CONSERVATISM had its greatest impact
on the TRADITIONAL VALUES subscale; AGE on the INTIMATE subscale;
and INTERGROUP FRIENDS on the INTIMATE and POSITIVE EMOTION
subscales.
Table
4
shows that some of the average effects, especially those of RACIST MOVE-
MENT APPROVAL and AGE, derive from heterogeneous effect size samples. This
lack of homogeneity reveals large effect size differences among the samples that
indicate that these particular effects must be interpreted cautiously. Table
4
also
supplies Rosenthal’s (1980) ‘fail-safe’ index for each effect. This index specifies the
number of additional independent studies showing no effect that must exist to bring
the overallp above
0.05.
Observe that all but the smaller effects are well protected.
Initial indications
of
validity
There remain questions concerning the validity of the two scales. In particular, is
subtle prejudice
reaZZy
prejudice? In addition to the confirmatory models just pre-
sented, we address this question by checking on the validity
of
both the Blatant
and Subtle Scales.
Table
4
shows the strength and consistency of both prejudice scales’ relationships
with the ETHNOCENTRISM measure. In addition, differences uncovered among
the samples have been previously documented with different measures. These results,
then, lend further evidence of criterion validity for the new scales. Hence, French
respondents on all scales-blatant, subtle, and each of the subscales, voice signifi-
cantly more prejudice against North Africans than Asians. Similarly, Dutch respon-
dents evince more prejudice against Turks than Surinamers (Hagendoorn and Hraba,
1987). Moreover, much-studied negative stereotypes are uncovered once again: per-
ceptions
of
‘lazy’ blacks in Great Britain and the Netherlands and ‘dishonest’ North
Africans in France.
Another means of checking on the validity of the new scales makes use of Figure
1’s
2
x
2
typology. Note the dividing lines for this topology are made at the logical
mid-point-between 30 and 31 on a possible range from
10
to
50.
Since the means
for the Subtle Scale are consistently higher than those of the Blatant Scale, few
respondents fall in the error cell (scoring high on the Blatant but low on the Subtle
Scale). We call those low on both scales
equalitarians;
those high
on
both
bigots;
and the critical group, those low on the Blatant but high on the Subtle Scale,
suhtles.
Subtle and blatant prejudice
69
+
blatant
prejudice
-
subtle prejudice
+
type
0
type
3
error
(<2%)
bigots
type
1
type
2
equalitarians subtles
Figure
1.
Typology
of
prejudice
Note this scheme implies a Guttman-like cumulative scale relationship between bla-
tant and subtle prejudice.
For each sample, Figure
2
relates these three types to the RIGHTS
OF
IMMI-
GRANTS question. Highly significant and consistent relationships are found in all
samples. The equalitarians are disproportionately among those who wish to extend
the rights of immigrants; bigots disproportionately among those who wish to restrict
further the rights of immigrants. Subtles assume a straddling, ostensibly non-preju-
diced position. More than either equalitarians
or
bigots, they simply wish to leave
the issue as it is.
Figure
2.
Prejudice types: attitudes toward changing immigrants’ rights.
70
T.
F.
Pettigrew
and
R.
W.
Meertens
Figure
3.
Prejudice types: send home those who commit crimes or have
no
papers.
Further criterion validity is provided by the question on IMMIGRATION
POLICY.
In all samples, bigots are disproportionately among the most extreme
in their negative views: significant majorities of them wish to send back either all
the outgroup or those not born in the country or not contributing to the economy.
By contrast, only equalitarians in any numbers favour not deporting any of the
outgroup. Figure
3
shows that subtles require justification for their exclusionary
policy preferences. Overwhelmingly, they favour sending back immigrants when there
is an ostensibly non-prejudicial reason for doing
so.
From 73 to
87
per cent
of
the subtles wish to deport the outgroup if they have committed serious crimes or
do
not have their immigration papers. These percentages are larger than either the
equalitarians or bigots in all samples.
This phenomenon
is
replicated for PREFERRED REMEDIES for improving rela-
tions with foreigners ‘living here’. In all
63
tests (nine items across seven samples),
the bigots are the most likely to regard such remedial efforts as ‘a bad idea’. The
equalitarians provide by far the most support in every sample for two decisive actions:
prosecuting racists in the courts and making naturalization easier. Such forceful
remedies are opposed by most subtles. They favour actions that are less committing
and direct, more vague and gradual, and place the responsibility on others. Hence,
subtles are almost as likely as equalitarians to support such ‘soft’ remedies as ‘promot-
ing tolerance
. . .
in schools’, ensuring that public servants treat nationals and non-
nationals equally, and expanding ‘international exchange programmes for young
people’.
Consider these data in terms of the normative acceptance of the nine remedies.
If we use the percentage of bigots who think a particular action to be a ‘good idea’
as afloor
index
of that action’s normative acceptance, then a revealing result emerges.
In all samples, there is an inverse relationship between a given remedy’s normative
acceptance and the difference between subtles and equalitarians in its support
Subtle
and
blatant
prejudice
7
1
(p
<
0.05
in each of five samples; Stouffer’s combined test for seven samples:
p
<
0.0001). Thus, the more generally acceptable a given remedy (even among bigots),
the less the difference between equalitarians and subtles in how much they think
it ‘a good idea’. This normative finding introduces
our
larger interpretation of these
results.
DISCUSSION
Two
types
of
prejudice
These data from 38 10 European respondents across four nations provide evidence
for the cross-national importance of the distinction between blatant and subtle preju-
dice. There are four types of evidence: the scalability
of
the two types; the confirma-
tory factor analytic results; the specificity
of
the correlates of the two types; and
the specificity of the effects of the two types.
The two
1
0-item Likert measures are moderately correlated, have adequate reliabili-
ties, and yield an exploratory factorial structure that mirrors earlier findings about
intergroup prejudice. The two subscales of blatant prejudice-threat and rejection
and anti-intimacy-are well-studied phenomena. The Subtle Scale’s three subscales-
the defence of traditional values, the exaggeration of cultural differences, and the
denial
of
positive emotions-also relate to substantial research literatures. The
socially acceptable rejection of minorities for ostensibly non-prejudicial reasons is
the key ingredient that links these components.
The confirmatory factor analytic results further support the blatant-subtle distinc-
tion. The two-factor models across the seven samples uniformly produce superior
fits to the correlational matrix than a one-factor model. But more experience with
the scales is necessary before a choice can be made between three possible structures.
The confirmatory analyses favour an hierarchical model in which blatant and subtle
prejudice are first-order latent factors that relate equally to a second-order latent
factor
of
generalized prejudice. But the correlated two-factor model does almost
as well. Moreover, the relations between the two scales and preferences regarding
immigration policy suggest a third, Guttman-like model in which subtle prejudice
constitutes an intermediate position between blatant prejudice and egalitarian toler-
ance.
Specified correlates
of
prejudice
Several theories of intergroup relations garner support in these cross-national data.
In particular, one aspect of authoritarian personality theory (Adorno
et
al.,
1950;
Allport, 1958; Altemeyer, 1988) receives confirmation. The strength of the ETHNO-
CENTRISM predictor reveals the generalized nature of prejudice against out-
groups-a central tenet of authoritarian theory apart from its psychoanalytic
orientation. Moreover, the predictive value of the measures of POLITICAL CON-
SERVATISM, APPROVAL
OF
RACIST MOVEMENTS, NATIONAL PRIDE
and the lack of POLITICAL INTEREST also lends evidence for the characterization
of the authoritarian personality.
These findings also support contact theory. Even with the other predictors con-
12
T.
F.
Pettigrew
and
R.
W.
Meertens
trolled, the breadth of one’s friendship network is a significant contributor to explain-
ing scores from both scales-particularly the Subtle Scale. Two factors underlie
this impressive consistency. The specification of a ‘friend’, not just an ‘acquaintance’,
makes it more likely that the theory’s critical conditions are met (equal status in
the situation, common goals, interdependence, and authority sanction; Allport,
1958).
Moreover, the causal sequence is two-way: more tolerant people seek diverse friend-
ships as well as such contact leads to less prejudice. These cross-sectional data do
not permit an evaluation of the two paths.
These results also specify the effects for prejudice of relative deprivation. Group
relative deprivation contributed significantly to the prediction of blatant prejudice
in
six
samples. But
it
attained statistical significance in only two samples as a predictor
of subtle prejudice. Individual relative deprivation
did
not relate with either type
of racism. This latter finding casts doubt on ‘riff-raff and mass society theories
of prejudice-namely, that prejudice flourishes largely among the deprived and alien-
ated pockets of industrial societies. Moreover, a five-item scale of political efficacy
failed to correlate with either the Blatant or Subtle Scales (analyses not shown).
To be sure, the less educated and less politically interested are more prone to prejudice;
but these relationships are not dominant in Table
3.
For
subtle prejudice especially,
there is no indication that the prejudiced are marginal segments of their societies.
The specification of the correlates of prejudice made possible by this differentiated
approach is enhanced when analyses are done with the five subscales. Conservatives
score higher on the Subtle Prejudice Scale, because they are defending traditional
values against the new ways of the outgroup. Older people score higher on the
Blatant Prejudice Scale, because of their fear of intimacy with the outgroup. And
those respondents with many types of friends are unusually open to intimacy with
the outgroup, and more readily report having felt admiration and sympathy for
outgroup members.
Specified effects
of
prejudice
The blatant-subtle distinction also helps to specify how prejudice shapes attitudes
toward the treatment of immigrants. On all three dependent variables, the blatant,
subtle and equalitarian types show distinctive patterns. Bigots, who score high on
both scales, want to restrict immigrants’ rights further, send most or all of them
back to their ‘home country’, and do little to improve immigrant-native relations.
Equalitarians, who score low on both scales, present a sharp contrast. They wish
to enlarge immigrants’ rights, allow most or
all
of the immigrants to remain, and
act forcefully to improve immigrant-native relations.
Social psychology has long studied the bigots and equalitarians. What is important
about this approach are the subtles-those who score low
on
the Blatant but high
on the Subtle Scale. They adopt intermediate positions that reject minorities
in
socially acceptable ways. Thus, they would neither restrict further nor enlarge immi-
grants’ rights. They would not send all immigrants ‘home’, but would send ‘home’
those for whom there is an ostensibly non-prejudicial reason to do
so.
They reject
forceful means of improving immigrant-native relations; yet support ‘soft’ measures
that place the burden on others. It is subtle prejudice that deserves more research
attention in future work.
So
we close with a tentative interpretation of the phenome-
non.
Subtle and blatant prejudice
13
A
normative interpretation
of
subtle prejudice
We recognize the critical importance for prejudice of structural and cultural differ-
ences among the four nations surveyed. Such recognition makes the cross-national
generality of these psychological dimensions
of
prejudice even more impressive. Yet
a robust theory of subtle prejudice requires a wide social context, one that considers
the normative structure within which it is embedded.
We believe these results are best explained by
a
normative perspective (Pettigrew,
1991).
Three decades ago, Kelman
(1961)
made critical distinctions between three
processes of social influence: compliance, identification, and internalization. Com-
pliance involves the instrumental expression of opinion for gaining rewards and
avoiding punishment. Identification occurs when one adopts opinions based on a
self-defining relationship with another person. Internalization occurs when an indivi-
dual accepts influence because it is congruent with one’s value system.
We posit that western European countries have been developing a norm against
blatant prejudice. Consistent with
our
data, this norm is stronger and more deeply
established in some nations, such
as
the Netherlands, than in others. Equalitarians
internalize this norm, bigots ignore
or
reject it. Subtles comply with the norm, and
express their negative intergroup views only in ostensibly non-prejudiced ways that
‘slip in under the norm’.
Devine
(1989)
shows that this tendency of subtles is not only possible but routine.
She first showed that negative cultural stereotypes of the outgroup were equally
well known to her low- and high-prejudice subjects. Both types employed stereotype-
congruent evaluations
of
ambiguous behaviours when controlled inhibition of the
stereotype was precluded. Then she employed a consciously directed thought-listing
task and found only low-prejudice subjects tend to inhibit the automatically activated
stereotypes and replace them with negations of the stereotype and thoughts about
equality.
Applied to
our
conjectures, we propose that equalitarians (low on both scales)
perform like Devine’s low-prejudice subjects. The subtle respondents (low on the
Blatant, high on the Subtle Scale), like the equalitarians, are also inhibiting automati-
cally activated prejudiced responses-but only when blatantly prejudiced stimuli
are presented. Our conjecture is that the subtles have a higher threshold than equali-
tarians (but lower than bigots) before their inhibition of prejudiced responses is
activated. Thus, subtles tend to accept items of the Subtle Scale that equalitarians
reject.
This is not a conscious, ‘self-presentation’ technique to hide prejudice from others
(Gaertner and McLaughlin,
1983).
Indeed, the careful behaviour of subtles appears
to serve principally the need
to
protect oneselffrom the realization that one harbours
prejudiced attitudes. To this extent, the anti-blatant prejudice norm has been partially
accepted and internalized by subtles-witness their low scores on blatant prejudice.
But they have not as yet internalized the norm deeply enough to prevent the indirect
‘leaking’
of
anti-minority attitudes captured by the items on the Subtle Scale.
Note that this analysis does
not
posit an anti-subtle prejudice norm. The full
internalization of equalitarian values by equalitarians leads them to be the only
ones who routinely reject even the socially acceptable items
of
the Subtle Prejudice
Scale. In this sense, their responses, like those of the bigots, are less governed by
the society’s norms on intergroup relations. These are the groups that have long
14
T.
F.
Pettigrew
and
R.
W.
Meertens
been the objects
of
personality-oriented explanations
of
prejudice.
A
rounded under-
standing of the subtles, however, requires more attention to the relevant intergroup
norms, which we believe have been shifting in recent years.
REFERENCES
Abelson, R. P., Kinder, D. R., Peters, M. D. and Fiske,
S.
T. (1982). ‘Affective and semantic
components in political person perception’,
Journal of Personality and Social Psychology,
Adorno, T. W., Frankel-Brunswik, E., Levinson, D. J. and Sanford, R. N. (1950).
The Authori-
Allport, G. W. (1958).
The Nature
of
Prejudice,
Doubleday-Anchor, Garden City, NY, U.S.A.
Altemeyer, B. (1988).
Enemies of Freedom: Understanding Right- Wing Authoritarianism,
Jossey-
Bass, San Francisco, CA, U.S.A.
Barker,
M.
(1984).
The New Racism: Conservatives and the Ideology of the Tribe,
Aletheia
Press, Frederick, MD, U.S.A.
Bentler, P. M. (1989).
EQS: Structural Equations Program Manual,
BMDP Statistical Software,
Los Angeles, CA, USA.
Bergmann, W. and Erb, R. (1986). ‘Kommunikationslatenz, moral und offentliche meining.
[Latent communication, moral and public meaning]’,
Kolner Zeitschrift fur Soziologie
und
Sozialpsychologie,
38:
223-246.
Bollen, K. A. (1989).
Structural Equations with Latent Variables,
Wiley, New York, NY,
U.S.A.
Brown, R. J., and Turner, J. C. (1981). ‘Interpersonal and intergroup behavior’. In: Turner,
J. C. and Giles,
H.
(Eds)
Intergroup Behavior,
Blackwell, Oxford, England.
Christie, R. and Jahoda,
M.
(Eds) (1953).
Studies in the Scope andMethodof ‘The Authoritarian
Personality
’,
Free Press, Glencoe, IL, U.S.A.
Cohen, J. (1977).
Statistical Power Analysis for the Behavioral Sciences,
rev. edn. Academic
Press, New York,
NY,
U.S.A.
Devine, P. G. (1989). ‘Stereotypes and prejudice: Their automatic and controlled components’,
Journal of Personality and Social Psychology,
56:
5-18.
Dienstbier, R. A. (1972).
‘A
modified belief theory of prejudice emphasizing the mutual caus-
ality of racial prejudice and anticipated belief difference’,
Psychological Review,
79:
146160.
Dijker, A. J. M. (1987). ‘Emotional reactions
to
ethnic minorities’,
European Journal of Social
Psychology,
17:
305-325.
Dovidio, J. F., Mann, J. and Gaertner,
S.
L. (1989). ‘Resistance to affirmative action: The
implications of aversive racism’. In: Blanchard, F. and Crosby,
F.
(Eds)
Afirmutive Action
in Perspective,
Springer-Verlag, New York, NY, U.S.A.
42:
619-630.
tarian Personality,
Harper
&
Row, New York, NY, U.S.A.
Essed, P. (1984).
Alledaags Racisme
[Everyday racism], Sara, Amsterdam, the Netherlands.
Freriks, P. (1990). ‘Franse anti-racisten willen handen vuil maken. [French anti-racists want
Gaertner,
S.
L.
and McLaughlin, J. P. (1983). ‘Racial stereotypes: Associations and ascriptions
Hagendoorn, L. and Hraba,
J.
(1987). ‘Social distance toward Holland’s minorities: Discrimi-
Hewstone,
M.
and Brown, R. (Eds) (1986).
Contact and Conflict
in
Intergroup Encounters,
Kelman, H. C. (1961). ‘Processes of opinion change’,
Public Opinion Quarterly,
35:
57-78.
Kovel, J. (1970).
White Racism: A Psychohistory,
Pantheon, New York, NY, U.S.A.
McConahay,
J.
B. (1983). ‘Modern racism and modern discrimination: The effects of race,
racial attitudes, and context on simulated hiring decisions’,
Personality and Social Psychology
Bulletin,
9:
551-558.
Moe, J. L., Nacoste, R. W. and Insko, C. A. (1981). ‘Belief versus race as determinants
of
discrimination: A study of southern adolescents in 1966 and 1979’,
Journal of Personality,
41:
1031-1050.
to get their hands dirty-i.e., take action)’,
De Volkskrant,
May
1,
1990,2.
of positive and negative characteristics’,
Social Psychology Quarterly,
46:
23-30.
nation against and among ethnic outgroups’,
Ethnic and Racial Studies,
10:
12G
133.
Blackwell, Oxford, England, U.K.
Subtle
and
blatant prejudice
75
Pettigrew, T.
F.
(1971).
Racially Separate or Together?
McGraw-Hill, New York, NY, U.S.A.
Pettigrew, T.
F.
(1989). ‘The nature ofmodern racismin the United States’,
Revue Internationale
de Psychologie Sociale,
2: 291-303.
Pettigrew,
T. F.
(1991). ‘Normative theory in intergroup relations: Explaining both harmony
and conflict’,
Psychology and Developing Societies,
3
(1):
3-16.
Reif, K. and Melich, A. (1991).
Euro-Barometer
30:
Immigrants and Out-Groups in Western
Europe, October-November
1988,
Inter-university Consortium for Political and Social
Research, Ann Arbor, MI,
U.S.A.
Rokeach,
M.
(Ed.) (1960).
The Open and Closed Mind,
Basic Books, New
York,
NY,
U.S.A.
Rosenthaf,
R.
(1
980). ‘Summarizing significance levels’,
New
Directions for Methodology
of
Sears, D. D. (1988). ‘Symbolic racism’. In: Katz, P.
A.
and Taylor, D.
A.
(Eds)
Eliminating
Smith, C.
R.,
Williams, L. and Willis, R. H. (1967). ‘Race, sex, and belief as determinants
Stein, D.
D.,
Hardyck,
J.
A.
and Smith, M.
B.
(1965). ‘Race and belief: An open and shut
Triandis, H. C. and Davis, E. E. (1965). ‘Race and belief as determinants
of
behavioral inten-
Vanneman, R. D. and Pettigrew, T.
F.
(1972). ‘Race and relative deprivation in the urban
Weigel, R.
H.
and Howes, P. W. (1985). ‘Conceptions
of
racial prejudice’,
Journal
of
Social
Social and Behavioral Science,
5:
3346.
Racism: Profiles in Controversy,
Plenum, New York, NY,
U.S.A.,
pp. 53-84.
of friendship acceptance’,
Journal
of
Personality and Social Psychology,
5:
127-1 37.
case’,
Journal
of
Personality and Social Psychology,
1:
281-289.
tions’,
Journal
of
Personality and Social Psychology,
2:
71 5-725.
United States’,
Race,
13
(4): 461-486.
Issues,
41
(3): 117-138.