INTERPERSONAL RELATIONS AND GROUP PROCESSES
The Psychological Bases of Ideology and Prejudice:
Testing a Dual Process Model
University of Auckland
Claire Wagner and Ilouize du Plessis
University of Pretoria
University of Kansas
The issue of personality and prejudice has been largely investigated in terms of authoritarianism and
social dominance orientation. However, these seem more appropriately conceptualized as ideological
attitudes than as personality dimensions. The authors describe a causal model linking dual dimensions of
personality, social world view, ideological attitudes, and intergroup attitudes. Structural equation mod-
eling with data from American and White Afrikaner students supported the model, suggesting that social
conformity and belief in a dangerous world influence authoritarian attitudes, whereas toughmindedness
and belief in a competitive jungle world influence social dominance attitudes, and these two ideological
attitude dimensions influence intergroup attitudes. The model implies that dual motivational and
cognitive processes, which may be activated by different kinds of situational and intergroup dynamics,
may underlie 2 distinct dimensions of prejudice.
The study of prejudice, ethnocentrism, and intergroup hostility
has been an important topic of social scientific inquiry for much of
the past century. During this time, two general approaches have
dominated research. One has viewed prejudice and ethnocentrism
as a group phenomenon to be explained in terms of intergroup
dynamics and processes (Pettigrew, 1958; Sherif, 1967; Tajfel &
Turner, 1979). A second, which has been relatively neglected for
the past half century, has viewed prejudice as an individual phe-
nomenon and set out to explain individual differences in people’s
propensity to hold ethnocentric intergroup attitudes (Adorno,
Frenkel-Brunswick, Levinson, & Sanford, 1950; Altemeyer, 1981;
Pratto, 1999; Sidanius & Pratto, 1999).
The importance of the individual-differences approach to prej-
udice has been empirically demonstrated by the generality of
prejudice. Persons who are unfavorably disposed to one particular
out-group or minority tend to be less favorable to others. This
effect is typically powerful, with effect sizes averaging around .50
(Adorno et al., 1950; Bierly, 1985; Duckitt, 1992), and pervasive,
holding across very different out-groups and even for totally
fictitious out-groups (Fink, 1971; Hartley, 1946).
The generality of prejudice suggests that some stable attribute of
individuals, such as personality or enduring beliefs, may cause a
generalized disposition to adopt prejudiced and ethnocentric atti-
tudes. Two such individual-differences dimensions have empirical
support. First, an authoritarian personality dimension was origi-
nally described by Adorno et al. (1950) and later refined by
Altemeyer (1981, 1988, 1998), who used his Right Wing Authori-
tarianism (RWA) Scale to measure the three covarying attributes
of conventionalism, authoritarian aggression, and authoritarian
submission. Second, Sidanius and Pratto (1993, 1999) conceptu-
alized and measured social dominance orientation (SDO) as a
“general attitudinal orientation toward intergroup relations, reflect-
ing whether one generally prefers such relations to be equal, versus
hierarchical” and the “extent to which one desires that one’s
ingroup dominate and be superior to outgroups” (Pratto, Sidanius,
Stallworth, & Malle, 1994, p. 742).
Both the RWA and the SDO Scales have been shown to pow-
erfully predict a wide range of political, ideological, and inter-
group phenomena and to be particularly powerful predictors of
chauvinistic ethnocentrism and generalized prejudice (Altemeyer,
1988, 1998; Sidanius & Pratto, 1999). Despite this, the two scales
seem relatively independent, often being nonsignificantly or only
weakly correlated with each other (Altemeyer, 1988, 1998; Mc-
Farland, 1998; McFarland & Adelson, 1996; Sidanius & Pratto,
1999). Altemeyer (1998) therefore recently suggested that the
John Duckitt, Department of Psychology, University of Auckland,
Auckland, New Zealand; Claire Wagner and Ilouize du Plessis, Depart-
ment of Psychology, University of Pretoria, Pretoria, South Africa; Ingrid
Birum, Department of Psychology and Research in Education, University
Portions of this article were included in a paper presented at the Annual
Scientific Meeting of the International Society of Political Psychology,
Seattle, Washington, July 2000, and some of the findings from Study 2
were also reported in Duckitt (2001).
Correspondence concerning this article should be addressed to John
Duckitt, Department of Psychology, University of Auckland, Private Bag
92019, Auckland, New Zealand. E-mail: email@example.com
Journal of Personality and Social Psychology, 2002, Vol. 83, No. 1, 75–93
Copyright 2002 by the American Psychological Association, Inc. 0022-3514/02/$5.00 DOI: 10.1037//0022-35220.127.116.11
RWA and SDO Scales measure two different kinds of authoritar-
ian personality, with both determining a basic susceptibility to
generalized prejudice and ethnocentrism.
However, a number of commentators have pointed out that the
items of the RWA Scale and those of its predecessor, the F Scale
(Adorno et al., 1950), do not pertain to personality traits and
behavior but to social attitudes and beliefs of a broadly ideological
nature (Feldman & Stenner, 1997; Goertzel, 1987; Saucier, 2000;
Stone, Lederer, & Christie, 1993, p. 232; Verkuyten & Hagen-
doorn, 1998). The items of the SDO Scale also consist of state-
ments of social attitude and belief, and Pratto et al. (1994; see also
Sidanius & Pratto, 1999) described the SDO Scale as a measure of
enduring beliefs. Moreover, the RWA and F Scales have been
shown to be highly reactive to situational threat manipulations. For
example, Altemeyer (1988) showed that scenarios that depicted
threatening social changes markedly increased RWA scores (cf.
also Doty, Peterson, & Winter, 1991; Meloen, 1983; Sales, 1973;
Sales & Friend, 1973).
Both the RWA and the SDO Scales therefore seem more ap-
propriately viewed as measuring social attitude or ideological
belief dimensions rather than personality. This is supported by a
review of research showing that investigations of the structure of
sociopolitical attitudes and sociocultural values have typically
revealed two roughly orthogonal dimensions (Duckitt, 2001, see
Table 3 and pp. 46–49), with one corresponding broadly to RWA
and the other to SDO. The RWA-like dimension has typically been
labeled social conservatism, traditionalism, or collectivism at one
pole and personal freedom, openness, or individualism at the other
pole. The SDO-like dimension has typically been labeled eco-
nomic conservatism, belief in inequality, power, or power distance
at one pole and social welfare, social concern, egalitarianism,
universalism, or humanitarianism at the other pole (cf. Saucier,
2000; Schwartz, 1992).
Although studies of the structure of sociopolitical attitudes and
of social values have used a wide variety of measures of these two
broad value-attitude dimensions, the RWA and SDO Scales have
typically emerged as the most powerful and consistent predictors
of ideological and intergroup phenomena (McFarland, 1998; Mc-
Farland & Adelson, 1996), probably because both are psychomet-
rically highly reliable and unidimensional measures. In contrast,
most other measures of these two social attitude and social value
dimensions have not been particularly reliable and have often been
contaminated by response biases such as direction of wording
effects. However, these other measures have typically correlated
highly with either RWA or SDO. For example, in his recent,
large-scale investigation, Saucier (2000) obtained a correlation of
.77 between the RWA Scale and a more general attitudinal mea-
sure of social conservatism.
If RWA and SDO are more appropriately conceptualized as
measuring social or ideological attitudes and values, this suggests
that the question of what, if any, personality dimensions may
underlie prejudice has not yet been answered. Indeed, it suggests
that the question could be profitably broadened to what underlying
psychological dynamics in individuals might underlie both gener-
alized prejudice and the two ideological attitude dimensions of
authoritarianism and social dominance. Drawing on the develop-
ment of the theoretical construct of social world view by psycho-
logical anthropologists (D’Andrade, 1992; Ross, 1993; Strauss,
1992), Duckitt (2000, 2001) has recently proposed a model of how
personality and social world views might together determine ideo-
logical attitudes and generalized prejudice.
D’Andrade (1992) and Strauss (1992) have suggested that indi-
viduals’ social values and attitudes tend to express or represent
what they have termed motivational goal schemas. These motiva-
tional goal schemas consist of a motivational goal that has been
made salient for the individual by the activation of particular social
schemas, or ways of perceiving social reality. Motivational goals
that are important for individuals derive from social schemas that
are highly accessible for them. Such highly accessible social
schemas can be seen more broadly as forming coherent social
world views, or relatively stable interpretations or beliefs about
others and the social world (Ross, 1993), so making particular
motivational goals generally salient for individuals. Individuals’
world views reflect their social realities but should also be influ-
enced by their personalities, or generalized behavioral disposi-
tions, as personality influences the nature, outcome, and interpre-
tations of individuals’ interactions with their social realities. In
addition to influencing individuals’ motivational goals indirectly
by influencing their world views, individuals’ personalities also
seem likely to directly influence the kind of motivational goals that
might be generally salient for them. The motivational goals that are
important or chronically salient for individuals are then directly
expressed in their social attitudes and values. For example,
Schwartz (1992) has conceptualized sociocultural values as direct
expressions of the importance of motivational goals for individuals
and cultural groups.
This conceptual framework can therefore be readily applied to
the social attitude dimensions of RWA and SDO, with each di-
mension representing a pair of opposing motivational goal sche-
mas. Thus, high RWA expresses the motivational goal of social
control and security, activated by a view of the world as dangerous
and threatening. Low RWA expresses the opposing motivational
goal of personal freedom and autonomy, activated by a view of the
social world as safe, secure, and stable. The predisposing person-
ality dimension is that of social conformity versus autonomy.
Being higher in dispositional social conformity should create a
greater readiness to perceive threats to the existing social order and
see the social world as dangerous and threatening. High social
conformity should also have a direct impact on authoritarian
attitudes by making the motivational goal of social control, secu-
rity, and stability salient for the individual. There is empirical
support for the core proposition here from numerous studies that
have found an association between social threat and authoritarian-
ism (e.g., Altemeyer, 1988; Doty et al., 1991; Peterson, Doty, &
Winter, 1993; Sales, 1973; Sales & Friend, 1973). Altemeyer
(1988) has also reported powerful correlations between his RWA
Scale and a perception of the social world as dangerous and
threatening (measured by his Belief in a Dangerous World Scale;
On the other hand, SDO has been empirically linked with the
personality dimension of toughmindedness versus tenderminded-
ness, a lack of empathy, and power motivation (Eysenck, 1954;
Goertzel, 1987; McFarland, 1998; Pratto et al., 1994; Sidanius &
Pratto, 1999). Powerful correlations have also been found between
the SDO Scale and Altemeyer’s (1998) recently constructed Per-
sonal Power, Meanness and Dominance (PP-MAD) and Exploitive
Manipulative Amoral Dishonesty (E-MAD) Scales, which contain
items that were drawn largely from the older Machiavellianism
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
Scale (Christie and Geis, 1970) and that seem to express a view of
the social world as a competitive jungle characterized by a ruth-
lessly amoral, Darwinian struggle for resources and power (e.g.,
“It’s a dog-eat-dog world where you have to be ruthless at times”).
This suggests that the personality dimension ultimately underlying
SDO might be toughmindedness versus tendermindedness, char-
acterized by the contrasting traits of being hard, tough, ruthless,
and unfeeling to others, as opposed to compassionate, generous,
caring, and altruistic. Toughminded personalities tend to adopt a
view of the world as a ruthlessly competitive jungle in which the
strong win and the weak lose, and this view tends to activate the
motivational goals of power, dominance, and superiority over
others, which in turn are expressed in high SDO. Tenderminded
personalities, on the other hand, should tend to adopt the directly
opposing view of the social world as a place of cooperative and
altruistic harmony in which people care for, help, and share with
each other. This should make salient the motivational goal of
altruistic social concern, which is expressed in low SDO.
The causal model of personality, world view, ideological atti-
tudes, and ethnocentrism that is suggested by this reasoning is
summarized in Figure 1 (see also Duckitt, 2000, 2001). The model
sees the two ideological attitude dimensions, RWA and SDO, as
causally determined by their corresponding personality and social
world view dimensions. In addition, causal links are also proposed
among the two personality dimensions, the two world view dimen-
sions, and the two ideological attitude dimensions. First, because
toughmindedness is socially deviant and undesirable, more so-
cially conforming persons are less likely to report being high on
toughmindedness. Modeling a causal path from social conformity
to toughmindedness would therefore statistically control the im-
pacts of more socially desirable responding by persons higher in
social conformity on toughmindedness as well as on a competitive-
jungle world view, both of which tend to be seen as socially
undesirable and unconventional in most sociocultural contexts.
Second, if people see the social world as a ruthlessly competi-
tive jungle, this should inevitably cause them to see it as more
dangerous and threatening. On the other hand, if people see the
social world as more dangerous and threatening, this does not
necessarily also cause them to see it as a competitive jungle. This
suggests a causal path from a competitive world view to a dan-
gerous one but not the reverse.
Third, pressures for cognitive consistency should result in RWA
and SDO having positive, reciprocal, causal impacts on each other,
at least in more ideologized societies in which politics is organized
along a single left (low RWA and low SDO) versus right (high
RWA and high SDO) dimension. Thus, the correlation between
RWA and SDO over a number of samples was shown to be
significantly higher in Western European countries and New Zea-
land, possibly because powerful labor movements produced a
greater ideological organization of politics in those countries than
in the United States (Duckitt, 2001). The correlation between
RWA and SDO was also stronger for older than for younger
persons, seemingly reflecting the effects of individuals being po-
litically socialized in late adolescence and early adulthood in
societies in which politics tends to be ideologically organized
The strength of the effects of RWA and SDO on each other also
seem likely to depend on which ideology is more dominant in a
particular sociopolitical context. For example, in societies in which
authoritarian–conservatism is the dominant ideology into which
people are socialized earlier and more intensively, pressures for
ideological consistency between RWA and SDO should produce a
stronger effect of RWA on SDO than the reverse. Thus, the
direction and existence of effects between RWA and SDO should
be context dependent, being weak or nonexistent in societies in
which politics is not ideologically organized, stronger in societies
in which politics is ideologically organized, and with directional
differences possible where socialization into one ideological di-
mension is clearly dominant relative to the other.
Finally, on the basis of a good deal of previous research (Alte-
meyer, 1998; McFarland, 1998; McFarland & Adelson, 1996;
Pratto et al., 1994; Sidanius & Pratto, 1999), the model predicts
that both RWA and SDO will directly increase both pro-in-group
and anti-out-group attitudes. Personality and world view, however,
are not seen as having general, direct impacts on prejudiced or
ethnocentric attitudes, only indirect effects mediated through these
The model has been supported by recent research. Three studies
showed that the personality dimensions of social conformity and
toughmindedness can be reliably measured and related as expected
to ideological attitudes (Duckitt, 2001, Study 1). Thus, social
conformity was strongly correlated with RWA but not with SDO,
whereas toughmindedness was correlated with SDO but not with
RWA. Structural equation modeling (SEM) with latent variables
showed good fit for the model in two large, European-origin
(Pakeha), New Zealand student samples (Duckitt, 2001, Study 2
and 3), with in-group and out-group attitudes being pro-European
and antiminority attitudes. All the predicted causal paths were
confirmed, except an originally suggested direct path from tough-
mindedness to SDO, which was then dropped from the model and
is therefore not shown in Figure 1 here. However, these two
analyses did reveal two relatively weaker but significant unpre-
dicted effects on prejudice. One was a direct effect of dangerous
world beliefs in increasing antiminority prejudice, and the second
was an inconsistent, direct effect of social conformity in reducing
Both these effects were interpreted as probably context depen-
dent. Thus, dangerous world beliefs seemed likely to have a direct
effect, increasing prejudice only when out-groups were in a di-
rectly threatening relationship with the in-group. This seemed
view on the two ideological attitude dimensions of right-wing authoritari-
anism (RWA) and social dominance orientation (SDO) and their impact on
intergroup attitudes. The dashed lines indicate context-dependent causal
A causal model of the impact of personality and social world
IDEOLOGY AND PREJUDICE
plausible in terms of the tensions bedeviling contemporary
majority–minority relations in New Zealand, with the Pakeha
(European origin) majority confronted by radical Maori activism
and a major influx of culturally different Asian and Polynesian
immigrants. This interpretation is also consistent with a second
finding from these two studies. In both samples, promajority
attitudes had a powerful and highly significant impact on antimi-
nority attitudes, a pattern of covariation that several prior studies
have suggested seems characteristic only for in-group–out-group
relations that are directly competitive and threatening (Brown et
al., 1992; Duckitt & Mphuthing, 1998; Struch & Schwartz, 1989).
The nonpredicted but inconsistent effect of social conformity in
reducing prejudice was interpreted as possibly reflecting social
norms against antiminority prejudice among traditionally liberal
New Zealand students, with the inconsistency due to fluctuations
in the strength and salience of these norms with changing social
and political circumstances.
The current research set out to test the generality of the theo-
retical model relating personality and world view to ideology and
prejudice by using SEM with latent variables in two very different
samples: American college students and White Afrikaans South
African students. Second, it also set out to investigate several
hypotheses suggested by the nonpredicted effects obtained in the
two earlier studies. Thus, it attempted to investigate whether belief
in a dangerous world and pro-in-group attitudes would directly
impact on anti-out-group attitudes only when out-groups seemed
to have a directly competitive and threatening relationship with the
in-group. It also set out to assess whether social conformity would
have direct effects, reducing negative out-group attitudes that
should be nonnormative in the social contexts studied.
Study 1 used a sample of American college students. For in-
group attitudes, we used a measure of American nationalism, and
for out-group attitudes, we used a measure of generalized prejudice
toward a variety of ethnic and national out-groups (e.g., Arabs,
Africans, Asians, Japanese, Indians, Blacks, and Hispanics). Be-
cause these groups collectively did not appear to be in a directly
competitive and threatening relationship to the United States at the
time of the research during 1999, it seemed likely that neither
Americans’ belief in a dangerous world nor their nationalism
would have direct impacts on these out-group attitudes. However,
because prejudice toward these groups should be generally viewed
as nonnormative among American college students, we hypothe-
sized that social conformity would have a direct effect in reducing
generalized prejudice to these out-groups.
Participants and Procedure
The research participants were 146 undergraduate students at a Mid-
western American university who completed a survey questionnaire during
1999. Fifty-seven were male, and 89 were female, and they had a mean age
of 19.60 years (SD ? 2.62). Expectation maximization (Schafer, 1997) was
used to replace isolated missing values so that a covariance matrix based on
the entire sample could be generated for the SEM analysis.
RWA and SDO.
meyer’s (1996) RWA Scale, and 12 items were sampled from Pratto et al.’s
(1994) SDO Scale so as to give equal numbers of pro- and contrait items
in each of the two shortened scales. The alpha coefficients for this sample
were .88 and .78, respectively.
Social conformity and toughmindedness.
personality constructs by asking participants to rate the degree to which
personality trait adjectives were characteristic or uncharacteristic of their
“personality and behavior.” The 14 social conformity items (with 7 protrait
and 7 contrait) were drawn from a 20-item personality trait adjective rating
scale—originally developed by Saucier (1994)—after judges’ ratings had
excluded any items that might pertain to attitudes or beliefs. Item and factor
analyses showed these 14 items provided a unidimensional and reliable
scale (Duckitt, 2001). The alpha coefficient in this sample was .80, and the
highest loading items were “obedient” and “predictable” (protrait) and
“nonconforming” and “unconventional” (contrait).
The Toughmindedness Versus Tendermindedness Scale had been devel-
oped earlier (see Duckitt, 2001) from a core set of trait adjectives used by
Goertzel (1987) to measure that construct. Item and factor analyses re-
vealed 24 items that were unambiguously personality or behavior trait
descriptors of the construct and provided a highly reliable and unidimen-
sional scale. The alpha coefficient in this sample was .91, and the two
highest loading items were “harsh” and “uncaring” for protrait and “kind”
and “gentle” for contrait. All the items of these two scales are shown in
Social world view: Belief in a dangerous world and belief in a
The concept of social world view is defined as
a coherent set of beliefs about the nature of the social world, specifically
about what people are like, how they are likely to behave to oneself, and
how they should be responded to and treated. The world view of a
dangerous and threatening social world is conceptualized as having two
opposing poles. At one pole is the belief that the social world is a
dangerous and threatening place in which good, decent people’s values and
way of life are threatened by bad people, whereas at the other pole is the
belief that the social world is a safe, secure, and stable place in which
almost all people are fundamentally good. A balanced 10-item scale was
used to measure belief in a dangerous world (see Duckitt, 2001). Eight of
these items were drawn from Altemeyer’s (1988) Belief in a Dangerous
World Scale, with 2 new, specially written items added. The alpha coef-
ficient for this 10-item scale was .80.
We measured belief in a competitive-jungle world using a balanced
20-item scale (see Duckitt, 2001). The construct was defined as the belief
that the social world is a competitive jungle characterized by a ruthless,
amoral struggle for resources and power in which might is right and
winning everything (protrait pole) versus a view of the social world as one
of cooperative harmony in which people care for, help, and share with each
other (contrait pole). We selected items that were judged to conform to this
definition from closely related existing measures such as Altemeyer’s
(1998) PP-MAD and E-MAD Scales, both of which drew on items from the
original Machiavellianism Scale, and some specially written items. The
resulting 20-item balanced scale was found to be unidimensional and
highly reliable (see Duckitt, 2001). The alpha coefficient in this sample
was .84. The highest loading items were “It’s a dog-eat-dog world where
you have to be ruthless at times” (protrait) and “The best way to lead a
group under one’s supervision is to show them kindness, consideration, and
treat them as fellow workers, not as inferiors” (contrait). All the items of
both the social world view scales are shown in Appendix B.
American nationalism and generalized out-group prejudice.
sured American nationalism using 11 items previously used by Pratto et al.
(1994) and Kosterman and Feshbach (1989) to measure nationalism. This
item set was not balanced against acquiescence, as there were 10 protrait
and only 1 contrait item. The alpha coefficient was .83.
Twenty items were randomly sampled from Alte-
We measured both of these
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
We measured generalized out-group prejudice using McFarland’s (1998)
12-item balanced scale. The items express negative or positive sentiments
toward a variety of out-groups relevant to Americans, such as Asians,
Africans, Arabs, Bosnians, Hispanics, Blacks, Indians, Japanese, Russians,
“other races,” “minorities,” and “foreign religions like Buddhism, Hindu-
ism, and Islam.” The alpha coefficient was .82.
Latent Variable Analyses
For the SEM analysis, the items of each scale were randomly assigned
to create three manifest indicators, with equal numbers of pro- and contrait
items for each manifest indicator wherever possible. Item parcels were
used as manifest indicators rather than individual items to keep the model’s
degrees of freedom to a reasonable level. The use of at least three indicators
has been recommended because models with fewer indicators for a latent
variable may, in some cases, be underidentified (Bentler & Chou, 1987).
Item parcels also had the considerable advantage over individual items of
providing more reliable manifest variables and so reducing random error as
well as reducing systematic error due to method effects such as direction of
wording (pro- vs. contrait item formulation), which has been shown to
substantially affect factor analyses of measures such as the RWA Scale
(Altemeyer, 1981, 1988). In the analysis, the three manifest indicators for
each scale were allowed to relate only to their one specific latent variable.
Maximum likelihood estimation was used for these analyses and the
following SEM analyses. We assessed overall model fit using Hu and
Bentler’s (1999) recent recommendations. Their investigation of the opti-
mal cutoff values of standard maximum likelihood fit indices suggests that
good fit is best indicated by values close to or better than .06 for root-
mean-square error of approximation (RMSEA), .08 for standardized root-
mean-square residual (SRMR), and .95 for the comparative fit index (CFI)
and goodness-of-fit index (GFI). In the case of GFI, Steiger (1989) and
others (Maiti & Mukerjee, 1990) have noted that Jo ¨reskog and So ¨rbom’s
(1996) widely used index of sample GFI is biased downward when degrees
of freedom are large relative to sample size and proposed a correction to
GFI that provides a more robust population estimate. Because the SEM
analyses in this and the next study tested very large models relative to
sample size, we used Steiger’s (1989) corrected population GFI (also
referred to as population gamma index) rather than Jo ¨reskog and So ¨rbom’s
sample GFI. Finally, the widely used rule of thumb for very large models,
in which chi-square values tend to be large, of viewing a ?2/df ratio of less
than 2 as a criterion of good fit was also used (Shumacker & Lomax, 1996).
The correlations among prejudice, nationalism, the two person-
ality scales, the two social world view scales, and the two ideo-
logical attitude scales are shown in Table 1. As expected, social
conformity was highly significantly associated with higher RWA
but not SDO, whereas toughmindedness was associated with
higher SDO but not RWA. Both RWA and SDO had significant
positive correlations with both nationalism and prejudice, which in
turn were positively correlated with each other. There was a weak,
though significant, positive correlation between RWA and SDO.
Figure 2 shows the standardized path coefficients for the struc-
tural equation model. To simplify the diagram, we do not show the
manifest variables and paths from latent to manifest indicators.
However, the paths from each of the latent variables to its manifest
indicators were all highly significant and powerful, with the weak-
est coefficient being .66. The overall fit indices indicated good fit
for a model of this size, ?2(236, N ? 146) ? 319.8, ?2/df
ratio ? 1.35, RMSEA ? .049 (90% confidence limits [CL] ?
.035, .063), SRMR ? .066, population GFI ? .95, CFI ? .96. Fit
for this model was clearly better than that for the null model, for
which ?2(276) ? 2,636.4. Overall, the model accounted for sub-
stantial proportions of the variance in RWA (51%), SDO (46%),
generalized out-group prejudice (67%), and nationalism (31%).
The basic paths predicted by the model were all clearly signif-
icant. Thus, social conformity had a significant path to dangerous
world beliefs, and both social conformity and dangerous world
beliefs had relatively strong paths to RWA. Toughmindedness had
a powerful, significant path to competitive world beliefs, and
competitive world beliefs had an equally powerful path to SDO.
Just as in the earlier New Zealand research, toughmindedness did
not have a significant, direct path to SDO (when modeled, ? ?
?.10, z ? ?.94, p ? .35). The links expected among these
personality, world view, and ideological attitude variables were
also supported. Social conformity had a significant negative path
to toughmindedness, and competitive world beliefs had a margin-
ally significant positive path to dangerous world beliefs. RWA and
SDO did have reciprocal positive paths to each other, but both
were weak, with a nonsignificant path from RWA to SDO and a
marginally significant path from SDO to RWA.
The effects of personality, world view, and ideology on in-group
and out-group attitudes were also as hypothesized. Both RWA and
SDO had significant, direct, positive paths to both nationalistic
attitudes and out-group prejudice. Moreover, as predicted for this
sample, neither dangerous world beliefs nor pro-in-group nation-
alistic attitudes had significant effects on anti-out-group prejudice.
Correlations Among the Variables for Study 1 (N ? 142–146)
3. Social conf
4. Tough mind
5. Dang world
6. Comp world
conformity; Tough mind ? toughmindedness; Dang world ? dangerous world; Comp world ? competitive-
* p ? .05. *** p ? .001.
RWA ? right-wing authoritarianism; SDO ? social dominance orientation; Social conf ? social
IDEOLOGY AND PREJUDICE
As expected, the path from nationalism to out-group prejudice was
also nonsignificant. Finally, as predicted, social conformity had a
significant, direct effect, reducing out-group prejudice.
Modification indices (Lagrange multiplier tests) were used to
test for additional nonpredicted effects that might be significant
and improve overall fit, in particular to check for evidence of any
further unexpected direct effects of personality and world view on
prejudice or nationalism. However, no other effects were found
that improved fit or would have been statistically significant if
included in the model.
Thus, the only direct effect of personality on the two outcome
variables, prejudice and nationalism, was for social conformity,
which had its predicted direct effect of reducing prejudice (? ?
?.25, z ? ?2.58, p ? .005). In addition to this, social conformity
also had its expected indirect effect (primarily through RWA) of
increasing prejudice (? ? .19, z ? 2.19, p ? .03). Thus, the two
opposite effects for social conformity on prejudice almost exactly
cancelled each other out (total effect from the SEM analysis was
? ? ?.06, z ? ?.71, p ? .48). In all other instances, the two
personality variables had their expected indirect effects through
RWA or SDO of increasing prejudice and nationalism. The total
effects from the SEM analysis were therefore positive and signif-
icant for social conformity on nationalism (? ? .18, z ? 2.79), for
toughmindedness on nationalism (? ? .17, z ? 3.52), and for
toughmindedness on prejudice (? ? .30, z? 4.35).
The correlational findings are entirely consistent with previous
research. Thus, the association of social conformity with RWA but
not SDO and of toughmindedness with SDO but not RWA repli-
cates the findings reported for samples in Canada, England, and
New Zealand (Duckitt, 2001). A number of previous studies have
reported positive correlations for RWA and SDO with both na-
tionalism and prejudice and significant correlations between na-
tionalism and prejudice in America (e.g., Altemeyer, 1998; Duc-
kitt, 2001; McFarland, 1998; McFarland & Adelson, 1996; Pratto
et al., 1994; Sidanius & Pratto, 1999). Finally, the correlation
between RWA and SDO was positive and, although significant,
relatively weak, as has been the case for previous findings with
student samples in the United States (Altemeyer, 1998; Duckitt,
2001, see Table 1 for a review of findings; McFarland, 1998,
McFarland & Adelson, 1996; Pratto et al., 1994).
The findings for the SEM analysis clearly support the model.
The overall fit was good, the paths predicted were significant and
broadly of the magnitude expected, and the model accounted for
substantial proportions of the variance in ideological and inter-
group attitudes. These findings closely replicate the findings ob-
tained previously from SEM analyses in two large New Zealand
samples (Duckitt, 2000, 2001), indicating that the model seems to
have cross-national validity. One minor difference from these prior
findings is that RWA and SDO had significant positive impacts of
moderate magnitude on each other in both the New Zealand
samples. In this American sample, there also appear to have been
positive reciprocal effects, but, as expected, they were weaker, so
that the effect for SDO on RWA was marginally significant and
that of RWA on SDO was nonsignificant. However, the two
effects were reasonably similar in magnitude. Thus, these findings
are not inconsistent with the expectation that RWA and SDO
model with the following latent variables: two personality, two world view, two ideological attitude, nationalism,
and generalized out-group prejudice. To simplify, we do not show manifest variables or the paths from latent to
manifest variables. Dashed lines indicate nonsignificant paths. RWA ? right-wing authoritarianism; SDO ?
social dominance orientation.?p ? .10. * p ? .05. ** p ? .01.
Standardized maximum likelihood coefficients from Study 1 (N ? 146) for the structural equation
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
should have weak reciprocal effects on each other, and both of
these effects might have proved significant with greater statistical
power. This finding is consistent with the generally weak associ-
ation between RWA and SDO in North American student samples,
as opposed to the stronger associations for societies in which
politics tends to be more highly ideologized, such as New Zealand,
and for older, nonstudent American samples (cf. Duckitt, 2001, see
Table 1 for a summary of prior North American studies).
As predicted, dangerous world beliefs and pro-in-group nation-
alistic attitudes had no direct effects on out-group prejudice in this
American sample, in contrast to their positive significant effects in
the earlier New Zealand research. This is consistent with the
hypothesis that dangerous world beliefs and pro-in-group attitudes
only increase prejudice against out-groups that are in a directly
competitive and threatening relationship with the in-group. The
SEM analysis also showed that the significant correlation obtained
between nationalism and prejudice in this sample and in previous
American research (Altemeyer, 1998; McFarland, 1998; McFar-
land & Adelson, 1996; Pratto et al., 1994) appears to be noncausal,
reflecting the impact of RWA and SDO on both.
Social conformity also had its predicted direct effect of reducing
prejudice, in contrast to its indirect effect, mediated through RWA,
of increasing prejudice. The direct effect of reducing prejudice
replicates the effect obtained in prior New Zealand research. This
seems plausibly explained by the counternormative nature of the
negative attitudes being studied for both American and New Zea-
land college students. Thus, in liberal social settings in which
prejudice is nonnormative, social conformity seems to have two
directly opposing effects on prejudice: an indirect effect of increas-
ing prejudice, mediated through more authoritarian attitudes, as
well as a significant, direct effect of reducing prejudice.
Finally, one caveat that might be noted derives from the use of
item packages for the SEM analyses. Although the use of item
packages for the SEM analyses was necessary to keep the model to
a reasonable size, this did mean that the SEM measurement model
did not provide as strong a test of the factorial differentiation of the
variables as would be provided by item-level analyses. The issue
of factorial differentiation of the constructs seems particularly
relevant in the case of the two sets of relatively closely associated
personality, world view, and ideological attitude variables, the first
being RWA, dangerous world beliefs, and social conformity and
the second being SDO, competitive-jungle world beliefs, and
toughmindedness (Duckitt, 2001). We therefore used item-level
factor analyses in Study 2 to check the factorial differentiation of
these two sets of variables.
The second study tests the theoretical model using a very
different sample and intergroup setting; that is, White Afrikaans
college students in South Africa. In-group attitudes were pro-
Afrikaner attitudes, and we assessed out-group attitudes toward
two different Black groups that seemed to vary considerably in the
degree to which they stood in a directly competitive and threaten-
ing relationship to White Afrikaners. The first of these groups was
the politically and numerically dominant Africans, who are Afri-
kaners’ historical rival for power (e.g., Thompson, 1995) and who
now, after majority rule, pose an ongoing threat to the social and
economic advantages still enjoyed by Afrikaners in South Africa
(W. Johnson, 1994). The second group was the numerically small,
socially marginal, and politically conservative Indian minority,
who, after majority rule, tended to give electoral support to the
White Afrikaner-dominated National party opposition (Guelke,
1996). Thus, if in-group attitudes and dangerous world beliefs only
increase prejudice toward groups seen as directly competitive and
threatening, then this should be the case for White Afrikaners’
attitudes to Africans but not to Indians.
Tentative predictions could also be formulated for two other
context-dependent effects proposed by the original model. First, in
the new, post-Apartheid South Africa, anti-Black and particularly
anti-African prejudice is counternormative. Thus, even among
White Afrikaners, being higher in social conformity should have
direct effects in reducing anti-Black and particularly Anti-African
prejudice. Second, the relationship between RWA and SDO should
depend on how ideologized politics is in a particular society. In
South Africa, politics has always been ethnically rather than ideo-
logically organized (Guelke, 1996; R. Johnson & Schlemmer,
1994). Thus, the correlation between RWA and SDO should be
relatively weak, as in the United States, and not powerful, as in
New Zealand or Western European societies. Moreover, it has
been argued that White Afrikaner nationalism in South Africa has
been primarily characterized by threat and authoritarianism arising
out of a historical experience of extreme insecurity (de Kiewiet,
1957; MacCrone, 1937; see also the discussion in Duckitt, 1992,
pp. 98–100). This suggests that ideological beliefs in group in-
equality and the right to dominate other groups might be secondary
to a threat-driven, ideological emphasis on group control, cohe-
sion, and security in this culture. This suggests that RWA should
affect SDO more powerfully than the reverse.
Participants and Procedure
A survey questionnaire was administered in late 1999 to introductory
psychology students at the University of Pretoria, a predominantly White
Afrikaans-language university in South Africa. The questionnaire was
translated into Afrikaans and then back-translated to check equivalence. It
was completed by 233 White Afrikaans students, of whom 145 were
female and 88 male, with an average age of 19.90 years (SD ? 1.87).
Altemeyer’s (1981) RWA Scale, which had previously been developed
from an item analysis of responses from a White South African student
sample (Duckitt, 1993). The reliability and validity of this shortened RWA
scale was supported in a series of studies using South African samples
(Duckitt, 1993; Duckitt & Farre, 1994). In the present sample, however, 4
of these items had nonsignificant item–total correlations and had to be
discarded. These 4 items were all contrait and all involved expressions of
support for antigovernment activism and radical dissidence. Thus, agree-
ment with these items could have reflected not just antiauthoritarianism but
also support for the right-wing conservative, radical Afrikaner activists
opposing the new left-wing Black government. The remaining 10-item
RWA Scale was therefore only partially balanced, with 7 protrait and 3
contrait items. The alpha coefficient in this sample was .67.
Ten items from the SDO Scale (Pratto et al., 1994) were ran-
domly sampled to give equal numbers of pro- and contrait items. The alpha
We measured RWA using a balanced 14-item short form of
IDEOLOGY AND PREJUDICE
Belief in a dangerous world.
using the same 10 items as in the previous study, with an alpha in this
sample of .82.
Belief in a competitive-jungle world.
competitive-jungle world using a balanced 14-item subset of the items used
in the previous study, with an alpha of .76.
We measured social conformity using the same 14
items as in the previous study, with an alpha of .72.
We measured toughmindedness using 20 of the
items (with equal numbers of pro- and contrait items) used in the previous
study, with an alpha of .88.
The same eight generalized intergroup attitude
items (four protrait and four contrait), selected to be appropriate to assess
attitudes toward any social group, were used to assess attitudes toward
White Afrikaners, Africans, and Indians. Examples of the items are “I have
a very positive attitude to Indians/Africans/White Afrikaners” and “I can
understand people having a negative attitude to Indians/Africans/White
Afrikaners.” The alpha coefficients were .72 for attitudes toward White
Afrikaners, .84 for attitudes toward Africans, and .77 for attitudes toward
We measured belief in a dangerous world
We measured belief in a
Factorial Differentiation of the Measures
We used item-level factor analyses to check the factorial differentiation
of the two relatively closely associated sets of personality, world view, and
ideological attitude measures, the first being RWA, dangerous world
beliefs, and social conformity, and the second being SDO, competitive-
jungle world beliefs, and toughmindedness. However, item-level explor-
atory factor analyses of both the RWA and the SDO Scales have typically
shown strong direction-of-wording method effects, with protrait and con-
trait items loading on separate method factors (Altemeyer, 1981, 1988,
1998). Exploratory factor analyses of the scales used in this and the
previous study also revealed strong method factors, which tended to
preclude clear-cut solutions.
We therefore used item-level confirmatory factor analyses (CFAs), with
pro- and contrait items loading both on their content factors and on separate
pro- and contrait method factors, to control direction-of-wording effects,
using the data from both Study 1 and this study. For each set of related
personality, world view, and ideological attitude items, the analyses com-
pared the fit of the hypothesized three-factor model against the one-factor
model and the 3 two-factor models, reflecting the three possible combina-
tions of the three scales. As Marsh and Grayson (1995) have recom-
mended, all three of the CFA models that have been used to analyze
multitrait–multimethod data (correlated traits with uncorrelated methods,
correlated traits with correlated methods, and correlated traits with corre-
lated uniqueness) were used for each analysis, but these produced essen-
tially the same findings.
Table 2 shows the fit indices for the CFAs with correlated trait and
correlated method factors for the three-factor, one-factor, and best fitting
two-factor models for the two personality, world view, and ideological
attitude item sets. The three-factor solutions for both the SDO,
competitive-jungle world beliefs, and toughmindedness item set and the
RWA, dangerous world beliefs, and social conformity item set produced
reasonably acceptable solutions given the magnitude of these very large
models, with RMSEA close to .06. The one- and two-factor solutions were
less satisfactory, with RMSEA clearly exceeding .06. The chi-square
difference test indicated that the superiority of the three-factor solution to
all other solutions was highly significant (the smallest chi-square differ-
ence was 93.2, with df ? 2, p ? .0001).
The correlations among all the variables are shown in Table 3.
The correlation between RWA and SDO was .21, which, although
significant, is relatively weak, as expected (using Cohen’s, 1988,
conventions of effect size strength). Social conformity was signif-
icantly associated with RWA but not SDO, and toughmindedness
Fit Indices for the Confirmatory Factor Analyses With Correlated Trait and Correlated
Direction-of-Wording Method Factors for the Personality, World View,
and Ideological Attitude Item Sets
RMSEA SRMRGFI CFI
SDO, competitive-jungle world beliefs, and toughmindedness item set
American sample (Study 1)
Best fitting two-factor model
South African sample (Study 2)
Best fitting two-factor model
RWA, dangerous world beliefs, and social conformity item set
American sample (Study 1)
Best fitting two-factor model
South African sample (Study 2)
Best fitting two-factor model
RMSEA ? root-mean-square error of approximation; SRMR ? standardized root-mean-square residual; CFI ?
comparative fit index; SDO ? social dominance orientation; RWA ? right-wing authoritarianism.
Goodness-of-fit index (GFI) is population GFI (Steiger, 1989). Study 1, N ? 146; Study 2, N ? 233.
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
was significantly associated with SDO but not RWA. RWA and
SDO were both significantly associated with pro-Afrikaner, anti-
African, and anti-Indian attitudes. Pro-Afrikaner attitudes were
significantly correlated with attitudes toward both the out-groups,
Africans and Indians, which were strongly associated with each
The covariance matrix for the SEM analysis was based on a
sample size of 233, with isolated missing values estimated by
expectation maximization (Schafer, 1997). As in Study 1, we
constructed three manifest item package indicators for each hy-
pothesized latent variable by randomly assigning the items of each
scale to three subsets, with, where possible, equal numbers of pro-
and contrait items in each subset. Each latent variable was permit-
ted to load only on its manifest indicators. A partial correlation
(correlated error) was modeled between anti-African and anti-
Figure 3 shows the standardized path coefficients from the
analysis. To simplify the diagram, we do not show the manifest
variables and paths to them. However, all loadings of latent vari-
ables on manifest indicators were substantial, with the weakest
standardized loading being .68. The overall fit indices for the
model indicated good fit for a model of this size, with ?2(302, N ?
233) ? 436.1, ?2/df ratio ? 1.44, RMSEA ? .044 (90% CL ?
.033, .052), SRMR ? .053, population GFI ? .96, CFI ? .95. Fit
for this model was clearly better than that for the null model, for
which ?2(351, N ? 233) ? 3,187.5. The model accounted for 61%
Correlations Among the Variables for Study 2 (N varies between 215 and 231)
3. Social conformity
4. Tough minded
5. Dangerous world
6. Competitive world
* p ? .05.
RWA ? right-wing authoritarianism; SDO ? social dominance orientation.
** p ? .01.*** p ? .001.
model with the following latent variables: two personality, two world view, two ideological attitude, pro-in-
group attitudes, and negative attitudes toward two out-groups. To simplify, we do not show manifest variables
or the paths from latent to manifest variables. Dashed lines indicate nonsignificant paths. RWA ? right-wing
authoritarianism; SDO ? social dominance orientation.?p ? .10. * p ? .05. ** p ? .01.
Standardized maximum likelihood coefficients from Study 2 (N ? 233) for the structural equation
IDEOLOGY AND PREJUDICE
of the variance in RWA, 59% of SDO, 55% of pro-Afrikaner
attitudes, 31% of anti-Indian attitudes, and 58% of anti-African
All the paths predicted by the model on ideological attitudes
were significant and closely replicated the findings in Study 1.
Thus, social conformity had a weak to moderate effect on danger-
ous world beliefs, and both social conformity and dangerous world
beliefs had strong effects on RWA. Tougmindedness had a pow-
erful effect on competitive-jungle world beliefs, and competitive-
jungle world beliefs a powerful effect on SDO. Again, there was
no significant direct effect of toughmindedness on SDO, but there
was a moderate to strong indirect effect mediated through com-
petitive world beliefs.
The links between the two personality, world view, and ideo-
logical attitude systems were also as expected. Social conformity
had a negative effect on toughmindedness, whereas competitive
world beliefs had a positive effect on dangerous world beliefs. The
effects of RWA and SDO on each other differed from those
obtained in the American sample in Study 1, but in the direction
that we predicted might be the case for this White Afrikaner
sample. Thus, these two ideological attitudes did not have recip-
rocal positive effects on each other. RWA had a significant posi-
tive effect on SDO (? ? .28, z ? 3.42), but SDO did not have a
positive effect on RWA (? ? ?.05, z ? ?.56).
The effects on pro-in-group and anti-out-group attitudes were
also as expected. Thus, RWA and SDO both had significantly
positive paths to pro-Afrikaner, anti-African, and anti-Indian atti-
tudes. As predicted, dangerous world beliefs and pro-Afrikaner
attitudes did not significantly increase prejudice against Indians,
the nonthreatening out-group, but did have significant direct ef-
fects of increasing prejudice against Africans, the more threatening
out-group. Social conformity had direct negative effects on anti-
African and anti-Indian attitudes, reducing prejudice in both cases,
but the effect, although significant for anti-African attitudes, did
not reach significance for anti-Indian attitudes (? ? ?.15, z ?
?1.28, p ? .21). Just as in the previous study, social conformity
also had significant indirect effects in the opposite direction of its
direct effects, which were mediated mainly through RWA, of
increasing both anti-African prejudice (indirect effect ? .36,
z ? 3.50, p ? .001) and anti-Indian prejudice (indirect effect ?
.27, z ? 2.60, p ? .009). As a result of the opposing direct and
indirect effects for social conformity on prejudice against both
Africans and Indians, social conformity had nonsignificant total
effects on both anti-African (.09, z ? 1.25) and anti-Indian prej-
udice (.13, z ? 1.60).
In all other instances, the personality variables had only indirect
effects mediated through RWA or SDO of increasing pro-in-group
or anti-out-group attitudes. Thus, toughmindedness had no direct
effects on pro-in-group and anti-out-group attitudes, but its indirect
effects through SDO produced significant total effects, increasing
pro-Afrikaner attitudes (.18, z ? 3.96), anti-African prejudice (.23,
z ? 4.77), and anti-Indian prejudice (.18, z ? 4.16). Social con-
formity also had significant indirect effects through RWA of
increasing pro-Afrikaner attitudes (.39, z ? 4.86).
Modification indices (Lagrange multiplier tests) were again
used to test for additional nonpredicted effects that might be
significant and improve overall fit. However, no other effects were
The findings again clearly support the theoretical model. Overall
fit for the SEM analysis was good, all critical predicted paths were
significant and broadly of the magnitude expected, and the model
accounted for substantial proportions of the variance in ideological
and intergroup attitudes. These findings closely replicate those
obtained for the American sample in Study 1 and those obtained
for research using two large samples of New Zealand students
One interesting difference from the previous studies, which had
been tentatively predicted, was for the effects of RWA and SDO
on each other. As predicted for a society in which politics is
ethnically rather than ideologically organized, the correlation be-
tween these two ideological attitude dimensions was relatively
weak and similar in magnitude to that obtained in the United States
in Study 1 and in previous research (see Duckitt, 2001). However,
unlike the prior American and New Zealand SEM analyses, which
had suggested reciprocal positive effects between them, these
findings suggest only a unidirectional impact of RWA on SDO.
This fits the characterization of Afrikaner society as a threat-
control culture in which the ideological belief in inequality and the
right to dominate other groups is secondary to and stems from
threat- and insecurity-driven authoritarianism (de Kiewiet, 1957;
Duckitt, 1992, 2000; MacCrone, 1937).
This interpretation that threat-driven authoritarianism is the
dominant ideology in Afrikaner culture also suggests that the
White Afrikaner students should be much higher than the Amer-
ican students in dangerous world beliefs and RWA but not neces-
sarily higher in SDO or competitive world beliefs. To check this,
we used those items that were used in both studies (6 RWA, 10
dangerous world, 14 competitive world, and 10 SDO items) to
compute item means for these scales in the two samples, which
were then statistically compared. These means are shown in Ta-
ble 4. As expected, the Afrikaner sample was markedly higher than
the American sample on dangerous world beliefs and RWA but no
different in SDO and significantly lower in competitive world
The SEM findings confirm the prediction that belief in a dan-
gerous world and pro-in-group attitudes would increase negative
attitudes to the more threatening and competitive out-group, Af-
ricans, but not to the apparently less threatening out-group, Indi-
ans. This finding clearly supports the hypothesis that these effects
are context dependent. Although a moderating role for intergroup
threat was not directly demonstrated by the research because threat
itself was not directly measured, the characterization of Africans as
a more threatening out-group than Indians is supported by the
out-group attitude scale means. The anti-African attitude mean
(M ? 5.67, SD ? 12.88) was higher than the anti-Indian attitude
mean (M ? 1.44, SD ? 9.91), and the difference was highly
significant, F(1, 230) ? 37.73, p ? .0001.
Social conformity was expected to have direct effects of reduc-
ing anti-African and anti-Indian prejudice, as all racist attitudes
should be generally viewed as socially nonnormative in the new
nonracial and democratic South Africa. Both effects were as pre-
dicted, although the effect for Indians was weaker than that for
Africans and was not statistically significant. The weaker effect for
Indians could be due to norms for nonprejudice to Indians, a
relatively peripheral and marginal minority in South Africa, being
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
somewhat weaker and less salient than norms of nonprejudice for
the politically and numerically dominant Africans. Again, as in the
previous study, the significant direct effects of social conformity
reducing prejudice and its indirect effects of increasing prejudice
mediated primarily through RWA largely cancelled each other out
overall for both anti-African and anti-Indian prejudice.
The results of both the SEM studies indicate good fit of the
empirical data to the theoretical model of personality, world view,
ideological attitudes, and intergroup attitudes. This finding seems
encouraging, as the difficulty of obtaining good fit indices for large
and complex models has frequently been commented on (Potthast,
1993; Yung & Bentler, 1994), and the model tested here is clearly
a large and complex one. In addition, all the basic causal paths
proposed by the model were significant in both samples, and the
critical paths were strong enough to explain substantial variance in
ideological and intergroup attitudes. These findings also closely
replicate those obtained in SEM analyses in two large New Zea-
land samples (Duckitt, 2001), supporting the cross-national gen-
erality of the model.
Ideally, the model would have been compared with plausible
alternative models. In these two studies, the alternative model was
the null or independence model simply because no clearly plausi-
ble alternative models seemed apparent. However, reviewers of
this article suggested several possible alternatives. One is a full
alternative model in which each of the two sets of personality,
world view, and ideological attitude variables might be caused by
a single, unmeasured causal variable. Thus, social conformity,
dangerous world beliefs, and RWA might be caused by Latent
Variable X, and toughmindedness, competitive-jungle world be-
liefs, and SDO might be caused by Latent Variable Y, with X and
Y affecting in-group and out-group attitudes. However, when we
tested this full alternative model using the data from Studies 1
and 2, the fit indices, which are shown in Appendix C, indicated
poor fit in both cases.
A second alternative proposal was derived from social domi-
nance theory and suggests that the competitive-jungle world view
can be viewed as a legitimizing myth through which people who
are high in SDO justify holding prejudiced attitudes. This suggests
a linear causal model, with SDO causally affecting competitive-
jungle world beliefs, which in turn affect prejudice, in contrast to
the current model, which expects competitive-jungle world beliefs
to affect SDO, which then affects prejudice. However, when this
alternative, linear social dominance model was tested for general-
ized out-group prejudice in Study 1 and then separately for anti-
Indian and anti-African prejudice in Study 2, the fit indices tended
to be poor and were clearly inferior to fit for the equivalent linear
models proposed by the current model, for which fit was good in
all three cases (see Appendix C).
Overall, therefore, the SEM analyses from both the studies
reported here and two prior, large-sample New Zealand studies
(Duckitt, 2001) provide clear support for the theoretical model of
how two personality and world view dimensions might influence
the dual ideological attitude dimensions of RWA and SDO and,
through them, intergroup attitudes. Three findings seem to merit
particular comment: first, the effects of the two personality and
world view dimensions on the ideological attitude dimensions of
RWA and SDO; second, the effects of personality and world view
on in-group and out-group attitudes; and, third, findings pertaining
to the hypothesized context dependence of certain effects.
Thus, in the first instance, the SEM analyses indicated strong
effects for the two personality and world view dimensions on the
two ideological attitude dimensions of RWA and SDO. The find-
ings also confirm an interesting asymmetry noted in the two
previous New Zealand studies (Duckitt, 2001) for the effects of the
personality and world view dimensions on the two ideological
attitudes. Although social conformity had a relatively strong, direct
effect on RWA and a weaker, indirect effect mediated through
dangerous world beliefs, the effect for toughmindedness on SDO
was entirely indirect and mediated through competitive-jungle
world beliefs. These two effects seem intuitively plausible. Being
toughminded—that is, dispositionally hard, ruthless, and unfeel-
ing—seems highly likely to elicit reactions from others that pro-
vide powerful confirmation that the social world really is a com-
petitive jungle, and this should generate a very strong effect.
Moreover, being toughminded only seems to lead to beliefs in
inequality and social dominance when this kind of competitive-
jungle view of the world is adopted. On the other hand, although
highly conforming persons tend to identify with conventional
Social World View and Ideological Attitude Means for Afrikaners (n ? 230–234)
and Americans (n ? 146)
Cohen’s dM SDM SD
Social world views
authoritarianism; SDO ? social dominance orientation.
**** p ? .0001.
The means are item means with a theoretical range from 4 (highest) to ?4 (lowest). RWA ? right-wing
IDEOLOGY AND PREJUDICE
society and therefore to be more sensitive to threats to it, it seems
plausible that this effect is no more than weak to moderate in
magnitude. The relatively strong, direct effects of social confor-
mity of personality on RWA should therefore reflect a strong
tendency for people who are dispositionally inclined to behave in
a socially conforming manner to also adopt conventional, tradi-
tional, and proauthority social attitudes and values congruent with
In the second instance, the SEM analyses confirmed the expec-
tation that the effects of personality and world view on pro-in-
group and anti-out-group attitudes would be largely indirect and
mediated through ideological attitudes. Thus, the SEM analyses
indicated that both the two personality and the two world view
dimensions had statistically significant indirect effects of increas-
ing pro-in-group and anti-out-group attitudes in every case, with
the magnitude of these effects varying between a low of .17 for
toughmindedness on nationalism in Study 1 and a high of .48 for
competitive-jungle world beliefs on prejudice in Study 1. In addi-
tion to these indirect effects, however, the analyses indicated two
direct effects of personality (social conformity) or world view
(dangerous world beliefs) on anti-out-group attitudes, but both
seemed to be context or out-group dependent rather than consistent
effects, as had been predicted from the model.
Thus, it had been hypothesized that social conformity would
reduce prejudices that were nonnormative in the samples stud-
ied—an effect directly opposite to social conformity’s expected
general effect of increasing anti-out-group prejudice indirectly by
raising RWA. The present research clearly demonstrates both these
opposing effects, with significant indirect effects increasing prej-
udice against all three the target groups in the two studies and
direct effects reducing prejudice against all the three target groups,
although the effect in one case, for anti-Indian prejudice, was not
significant. Because these prejudiced attitudes were all nonnorma-
tive, these direct effects of social conformity in reducing prejudice
are consistent with expectation, except for the nonsignificance of
the direct effect reducing anti-Indian prejudice. However, even the
weaker effect here seems reasonably easily explained, because
norms against anti-Indian prejudice in South Africa should be
much less salient than norms against anti-African prejudice, be-
cause Indians are a marginal and relatively peripheral minority,
whereas Africans are the politically dominant majority.
Context-dependent effects were also clearly supported for pro-
in-group attitudes on anti-out-group attitudes and for dangerous
world beliefs on out-group attitudes. In both cases, these effects
varied as hypothesized according to the degree that out-groups
seemed to be in directly threatening and competitive relationships
with the in-group. Thus, as predicted, dangerous world beliefs and
pro-in-group attitudes significantly increased anti-African preju-
dice, as Africans are an out-group that is clearly threatening to
White Afrikaners, but did not significantly affect attitudes to the
nonthreatening Indian out-group. Moreover, neither dangerous
world beliefs nor pro-in-group attitudes significantly affected prej-
udice in Study 1, in which the target out-groups would generally
not be seen as directly competitive or threatening to Americans at
the time of the study. Several previous findings have also found
that the effect of pro-in-group attitudes on anti-out-group attitudes
seems to be context or out-group dependent and moderated by
perceived threat from the out-group (Brown et al., 1992; Duckitt &
Mphuthing, 1998; Struch & Schwartz, 1989).
It also seems theoretically plausible that out-group threat should
influence the impact of dangerous world beliefs and pro-in-group
attitudes on out-group attitudes. Dangerous world beliefs should
index individuals’ reactivity to threat and therefore should deter-
mine attitudes to threatening groups and not to nonthreatening
groups. Similarly, pro-in-group attitudes should generate negative
attitudes toward out-groups that threaten the in-group but have
little effect on out-groups that are not threatening to or directly
competitive with the in-group (cf. also Brewer, 1979, 1999).
A cautionary note is in order here, however. Although the
results are consistent with a moderating role for out-group threat
on the effects of dangerous world beliefs and pro-in-group atti-
tudes on out-group attitudes, they do not demonstrate this moder-
ating effect directly. This would have required us to directly
measure degree of threat from the different out-groups, and we did
not do this in this research. In similar fashion, direct measures of
the degree to which particular prejudiced attitudes are nonnorma-
tive in particular social contexts would be necessary for us to
demonstrate more definitively that this moderates the direct effects
of social conformity in reducing negative attitudes toward partic-
The model had also expected reciprocal but context-dependent
causal effects between RWA and SDO. The current findings are
also broadly consistent with this. First, the correlation between
RWA and SDO, though significant, was relatively weak in both
the American and the South African samples. In the latter case, this
is to be expected, because politics in South Africa tends to be
ethnically rather than ideologically organized. The expected recip-
rocal effects of RWA and SDO in the American sample, though
both positive, were weak, with the path from SDO to RWA being
marginally significant and that from RWA to SDO nonsignificant.
The finding is therefore not entirely clear but is not inconsistent
with weak reciprocal effects, which would be expected in a society
in which politics is not strongly organized ideologically and in
which neither of the two ideological dimensions seemed particu-
larly dominant. The finding for South Africa is clearer and seems
to illustrate the context dependency in the relationships between
RWA and SDO. Here, the effects were clearly different, with the
effect of RWA on SDO positive and significant and the effect of
SDO on RWA nonsignificant and close to zero. This is consistent
with the expectation that in a culture in which authoritarian–
conservative ideological beliefs are dominant and therefore
primary in socialization, cognitive pressures for ideological con-
sistency should flow from the dominant ideological belief di-
mension to the less dominant one. Thus, for White Afrikaners,
authoritarian–conservatism ideological beliefs should influence
the adoption of beliefs in social dominance and group inequality,
and not the reverse. The relative weakness of the effect therefore
reflects the relatively weak degree of ideological organization of
politics in a society where politics is primarily organized in terms
In evaluating the findings from these two studies, we find that
several important methodological issues merit comment. One issue
pertains to the use of self-report psychometric measures of the
constructs proposed, specifically to their factorial independence
from each other and their construct validity. A risk here is that
content overlap between subsets of items in different scales might
build in spurious relationships between the scales. However, a
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
number of steps were taken to control for this in the development
of the scales used here.
Thus, in pilot testing, items that did not clearly correlate more
powerfully with their own scales than with other scales were
deleted or rewritten. Independent judges carefully checked all the
measures used for item-content overlap between scales and dis-
carded any dubious items. For example, Saucier (1994) had in-
cluded items such as “liberal,” “conservative,” “traditional,” “re-
ligious,” and “nonreligious” in his original Norm Orientation
personality scale, from which the social conformity scale was
derived. These were excluded from the social conformity scale
because they could be used to describe attitudes as well as behav-
ior and might have spuriously inflated the relationship between
conformity and RWA. In the development of the scales, we used
factor analyses to assess scale unidimensionality and the factorial
independence of the scales. In addition, as Altemeyer (1981) has
recommended, we checked the correlations between individual
items and other scales to confirm that scale intercorrelations de-
rived from all the items of the scales and were not just created by
particular subsets of items.
Finally, the good fit obtained in the SEM analyses for the two
samples reported here also supports the factorial independence of
the latent variables. Although item packages rather than items were
used in these analyses, the items of each scale were randomly
assigned to their packages so that each package would represen-
tatively sample the items of the scale. Item-level CFAs, however,
also provided more rigorous support for the factorial differentia-
tion of the two sets of related personality, world view, and ideo-
logical attitude measures.
An issue closely related to that of the factorial differentiation of
the scales is that of their construct validity, or the degree to which
the item sets used measured the constructs specified. Considerable
evidence has been reported for the construct validity of the RWA,
SDO, and Belief in a Dangerous World Scales (Altemeyer, 1981,
1988, 1996; Pratto et al., 1994). The competitive-jungle world
view items were largely drawn from the older Machiavellianism
Scale, but here the items were content validated to a much
more precisely defined and more specific construct: that of a
competitive-jungle world view.
A good deal of prior research had validated Machiavellianism as
a predictively useful individual-differences construct (cf. Wilson,
Near, & Miller, 1996), but the scales to measure Machiavellianism
appeared to be complexly multidimensional (e.g., Allsopp, Ey-
senck, & Eysenck, 1991; Hunter, Gerbing, & Boston, 1982).
However, the most comprehensive investigation of these Machia-
vellianism measures concluded that the central core of their items
and of the construct is a cynical and competitive view of and
orientation toward others (Hunter et al., 1982). This conclusion is
entirely congruent with the current conceptualization of a
competitive-jungle world view, which drew on these core items for
its measure. The highest loading item for this competitive-jungle
world view scale expresses this clearly: “It’s a dog-eat-dog world
where you have to be ruthless at times.” The SEM analyses in both
the samples and in prior New Zealand studies suggest that this
construct plays the critical role in mediating the effect of tough-
mindedness as a personality construct on the ideological attitude
dimension of SDO, with no direct effects of toughmindedness on
SDO. This suggests that it is only when toughminded personalities
see the social world as a competitive jungle at an interpersonal
level that they adopt ideological beliefs in inequality and social
The construct in the current research that has had the least prior
attention is that of social conformity. The items for this construct
were developed by Saucier (1994), who used factor analysis to
derive nonevaluative personality scales using trait adjective de-
scriptors and who had named this scale Norm Orientation because
its items (the highest loading in Saucier’s research being “rebel-
lious” and “conforming”) seemed to express a dispositional ten-
dency to behave in conformity with social and societal norms and
rules. However, several of Saucier’s original items could pertain to
attitudes and beliefs (e.g., “conservative,” “liberal”) as well as
personality, and we therefore excluded them to avoid possible
content overlap with the attitudinal items of the RWA Scale. We
also renamed the scale Social Conformity because this seemed a
more appropriate conceptualization of its item content. A reviewer
of this article, however, suggested that other items that had re-
mained in the scale, such as “unorthodox,” “free-living,” and
“old-fashioned,” did not seem to be classic personality trait de-
scriptors and might also reflect social attitudes. If so, this might
have built in spurious content overlap between this scale and the
RWA Scale. However, we feel this is not likely for several reasons.
First, the instructions for the scale explicitly and prominently
asked participants “to rate the extent to which you feel each of the
following descriptive adjectives is characteristic or uncharacteris-
tic of your PERSONALITY AND BEHAVIOR”. Second, the item-
level CFAs of the social conformity, RWA, and Dangerous World
Scales supported the hypothesized three-factor solution and
showed it to be significantly superior to a solution in which RWA
and social conformity items loaded together on one factor. Third,
to eliminate any possibility of content overlap, we asked three
independent judges to select only those items of the social con-
formity scale that unambiguously pertained only to behavioral
dispositions and personality and not to social attitudes. Five items
were unanimously rated as such: “obedient,” “predictable,” “re-
spectful,” “erratic,” and “unpredictable.” We recomputed both the
correlational and the SEM analyses for both studies using only
these five items as the social conformity scale (with alphas in the
two studies of .59 and .51). There were no changes in the signif-
icance of effects, and the correlations and path coefficients be-
tween this purged social conformity scale and RWA were only
slightly weaker. For example, the correlation between this short-
ened five-item social conformity scale and RWA, corrected for the
lower reliability of the shortened social conformity scale, was .38
(as opposed to .43 for the full scale) in Study 1 and .35 (as opposed
to .44 for the full scale) in Study 2.
Fourth, an important finding from these two studies seems to
provide compelling evidence for divergent validity between the
social conformity and RWA Scales. In both these studies and in the
two prior New Zealand studies, this social conformity scale had
nonsignificant correlations with all the indices of prejudice used
(correlations varied between ?.07 and .10, with the averaged
correlations over five analyses being .05, with n ? 1,169), whereas
RWA correlated highly significantly with prejudice in each case
(correlations varied between .31 and .49, with the averaged cor-
relations being .39, with n ? 1,176). These different effects
therefore clearly support the conceptual distinction between the
measures of social conformity and RWA. The significant direct
effects of social conformity in reducing prejudice that emerged in
IDEOLOGY AND PREJUDICE
the SEM analyses were also consistent with its conceptualization
as measuring a personality disposition to conform to social and
societal norms and rules, as the prejudiced attitudes measured in
these studies seem to be clearly nonnormative in their societies.
Finally, constructs very similar to social conformity, as concep-
tualized and measured here, have emerged in broader investiga-
tions of personality trait dimensions, usually with somewhat dif-
ferent labels. An example is the Rule-Free Versus Rule Bound
scale from Lorr’s (1986) Interpersonal Style Inventory, which is
defined as “To follow one’s own way of doing things and resist
conventional rules and norms” versus “To respect and comply with
authorities, to follow conventions, rules, and norms.” In Big Five
terms, social conformity seems to subsume three highly specific
(narrow bandwidth) and closely related trait subdimensions within
Conscientiousness (dutifulness and orderliness) and Openness
(creativity) from Goldberg’s (1999) International Personality Item
Pool’s 45 AB5C scales. This would locate social conformity as a
medium-bandwidth trait construct straddling the boundary of C?
and O?. The toughmindedness versus tendermindedness con-
struct, as conceptualized and measured in the current research,
seems to subsume three specific component traits of the Big Five
Agreeableness higher order dimension, such as Goldberg’s (1999)
closely related three AB5C subdimensions of empathy, sympathy,
and tenderness, locating it as a medium-bandwidth construct
A second methodological issue that merits comment is that of
causality. SEM can test the fit of hypothesized causal models to
empirical data but cannot demonstrate causality as experimental or
longitudinal research might (Hoyle & Smith, 1994). Thus, the
evidence from the SEM studies reported here cannot validate the
causal hypotheses proposed by this model. However, there are
experimental or longitudinal findings that do clearly support sev-
eral critical causal paths proposed by the model. For example, a
study by Katz and Hass (1988) bears on the causal directionality
between ideological attitudes and out-group attitudes. Their exper-
iment showed that priming two ideological attitude measures
essentially similar to RWA (a Protestant ethic beliefs scale,
which is an expression of social conservatism) and SDO (a
humanitarianism–egalitarianism scale) causally influenced racial
attitudes but priming racial attitudes did not causally influence
ideological attitudes. Other evidence shows that social environ-
mental changes that are directly reflected in social world view
seem to causally impact on ideological attitudes and prejudice. For
example, Altemeyer (1988) showed that scenarios that depicted
threatening and dangerous social changes markedly increased
RWA. A more recent experimental investigation also used future
scenarios to manipulate social threat and found that this produced
significant increases in RWA that were entirely mediated through
elevated dangerous world beliefs (Duckitt & Fisher, 2001). A
number of other longitudinal or experimental studies have shown
that social threat seems to increase authoritarianism and prejudice
(Doty et al., 1991; Downing & Monaco, 1986; Meloen, 1983;
Sales, 1973; Sales & Friend, 1973).
Finally, although the model tested in this research provides a
theory of the psychological bases of individual differences in
ideological attitudes and prejudice, it also has implications beyond
the research reported here that suggest potentially fruitful future
avenues for investigation. One implication suggested by the model
pertains to social environmental influences on ideological beliefs
and prejudice. Social world views should be influenced not only by
individuals’ personalities but also by social reality. Consequently,
social environments that really are dangerous and threatening or
competitive jungles should both increase prejudice, the former by
generating authoritarian attitudes and the latter through generating
social dominance beliefs. The theoretical model therefore proposes
hypotheses about the impacts of both personal influences such as
personality and social environmental influences such as threat on
ideological beliefs and prejudiced attitudes.
A second implication arises from the core proposition of the
model that two basic motivational goals, threat-driven control and
security motivation and competitively driven dominance or supe-
riority motivation, which are expressed in two different kinds of
social attitudes or ideological beliefs, underlie prejudice. This
suggests that there should be two distinct dimensions or kinds of
prejudice. In the case of threat-control-driven prejudice, out-
groups are disliked and feared because they are seen as threaten-
ing, dangerous, and disruptive to social or group order and secu-
rity. This seems likely to generate a particular kind of prejudiced
social categorization, with the social world categorized into
“them,” who are bad, dangerous, immoral, and deviant and who
threaten “us,” who are normal, morally good, decent people.
In the case of competitive-dominance-driven prejudice, out-
groups are despised and derogated because they are believed to be
inferior, worthless, and inadequate. Here the social world is cate-
gorized into “us,” who are superior, strong, competent, and dom-
inant (or should be) and “them,” who are inferior, incompetent,
and worthless. This motivationally based distinction between two
dimensions of prejudice is consistent with prior accounts that have
differentiated two kinds of racism—that is, aversive versus dom-
inative (Kovel, 1970), hot versus cold (Fiske, 1998), and symbolic
versus traditional (Kinder & Sears, 1981; McConahay & Hough,
1976). It is also consistent with Fiske, Xu, Cuddy, and Glick’s
(1999) recent distinction between two intergroup attitude dimen-
sions of disliking versus disrespecting and with the well-
established existence of two distinct dimensions of evaluative
stereotypes—that is, beneficence or morality stereotypes (good vs.
bad) and competence stereotypes (superior or competent vs. infe-
rior or incompetent; Fiske et al. 1999; Giles & Ryan, 1982; Poppe
& Linssen, 1999).
A third implication is that these two kinds of motivational goals
are activated by characteristics of intergroup relationships. Thus,
intergroup relationships characterized by out-group threat to in-
group values and security should activate threat-driven social
control and security motivational goals and authoritarian ideolog-
ical attitudes generating negative attitudes toward that particular
out-group. When the intergroup relationship is one of inequality in
status and power or competition over relative dominance and
power, this activates competitively driven dominance and power as
motivational goals to generate prejudice toward the lower status or
competing out-group. The model therefore seems to have potential
for integrating individual-level theories of prejudice, such as au-
thoritarianism and social dominance, and group-level perspectives
focusing on intergroup threat (Pettigrew, 1998; Stangor & Cran-
dall, 2000; Stephan, Ybarra, Martinez, Schwarzwald, & Tur-kaspa,
1998), inequality in status and power (Fiske et al., 1999), and
social competition over relative status (Tajfel & Turner, 1979).
Thus, both intergroup and personal processes may influence indi-
viduals’ prejudiced attitudes through the cognitive activation of
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
two different sets of motivational goals and their expression in
In conclusion, therefore, this research has proposed a model to
explain how personality and social world views might influence
individuals’ ideological attitudes and prejudiced intergroup atti-
tudes. The model’s constructs are clearly defined and differenti-
ated and seem to have some heuristic value. Preliminary testing of
the model using SEM analyses across a range of samples from
different countries has produced highly consistent findings that
seem supportive. However, testing thus far has been limited to
cross-sectional survey research using a particular set of self-report
psychometric instruments, several of which can be viewed as
preliminary and requiring further construct validation. New re-
search should test the model using alternative, conceptually equiv-
alent measures of its constructs and, where possible, actual behav-
ioral indicators of the constructs rather than psychometric self-
report scales, which are open to response biases. Longitudinal or
experimental research would help to test the actual causalities
proposed by the model. One such direction that is being currently
developed with promising initial findings uses scenarios and prim-
ing procedures to manipulate experience of the social environment
to observe effects on social world view, motivational goal salience,
and ideological beliefs (Duckitt & Fisher, 2001).
Adorno, T., Frenkel-Brunswick, E., Levinson, D., & Sanford, N. (1950).
The authoritarian personality. New York: Harper.
Allsopp, J., Eysenck, H., & Eysenck, S. (1991). Machiavellianism as a
component in psychoticism and extraversion. Personality and Individual
Differences, 12, 29–41.
Altemeyer, B. (1981). Right-wing authoritarianism. Winnipeg, Canada:
University of Manitoba Press.
Altemeyer, B. (1988). Enemies of freedom: Understanding right-wing
authoritarianism. San Francisco: Jossey-Bass.
Altemeyer, B. (1996). The authoritarian specter. Cambridge, MA: Harvard
Altemeyer, B. (1998). The other “authoritarian personality.” In M. P.
Zanna (Ed.), Advances in experimental social psychology (Vol. 30, pp.
47–92). San Diego, CA: Academic Press.
Bentler, P., & Chou, C. (1987). Practical issues in structural modeling.
Sociological Methods and Research, 16, 78–117.
Bierly, M. (1985). Prejudice toward contemporary outgroups as a gener-
alized attitude. Journal of Applied Social Psychology, 15, 189–199.
Brewer, M. (1979). In-group bias in the minimal intergroup situation: A
cognitive–motivational analysis. Psychological Bulletin, 86, 307–324.
Brewer, M. (1999). The psychology of prejudice: Ingroup love or outgroup
hate. Journal of Social Issues, 55, 429–444.
Brown, R., Hinkle, S., Ely, P., Fox-Cardamone, L., Maras, P., & Taylor, L.
(1992). Recognizing group diversity: Individualist-collectivist and
autonomous-relational social orientations and their implications for in-
tergroup processes. British Journal of Social Psychology, 31, 327–342.
Christie, R., & Geis, F. (1970). Studies in Machiavellianism. New York:
Cohen, J. (1988). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Erlbaum.
D’Andrade, R. (1992). Schemas and motivation. In R. D’Andrade & C.
Strauss (Eds.), Human motives and cultural models (pp. 23–44). Cam-
bridge, England: Cambridge University Press.
de Kiewiet, C. (1957). A history of South Africa: Social and economic.
London: Oxford University Press.
Doty, R., Peterson, B., & Winter, D. (1991). Threat and authoritarianism in
the United States, 1978–1987. Journal of Personality and Social Psy-
chology, 61, 629–640.
Downing, L., & Monaco, N. (1986). In-group/out-group bias as a function
of differential contact and authoritarian personality. Journal of Social
Psychology, 126, 445–452.
Duckitt, J. (1992). The social psychology of prejudice. New York: Praeger.
Duckitt, J. (1993). Right-wing authoritarianism among white South African
students: Its measurement and correlates. Journal of Social Psychology,
Duckitt, J. (2000). Culture, personality, and prejudice. In S. Renshon & J.
Duckitt (Eds.), Political psychology: Cultural and crosscultural foun-
dations (pp. 89–107). New York: New York University Press.
Duckitt, J. (2001). A dual process cognitive-motivational theory of ideol-
ogy and prejudice. In M. P. Zanna (Ed.), Advances in experimental
social psychology (Vol. 33, pp. 41–113). San Diego, CA: Academic
Duckitt, J., & Farre, B. (1994). Right wing authoritarianism and political
intolerance among Whites in the future majority rule South Africa.
Journal of Social Psychology, 134, 735–742.
Duckitt, J., & Fisher, K. (2001). The impact of social threat on worldview
and ideological attitudes. Manuscript submitted for publication.
Duckitt, J., & Mphuthing, T. (1998). Group identification and intergroup
attitudes: A longitudinal analysis in South Africa. Journal of Personality
and Social Psychology, 74, 80–85.
Eysenck, H. (1954). The psychology of politics. London: Routledge and
Feldman, S., & Stenner, K. (1997). Perceived threat and authoritarianism.
Political Psychology, 18, 741–770.
Fink, H. (1971). Fictitious groups and the generality of prejudice: An
artifact of scales without neutral categories. Psychological Reports, 29,
Fiske, S. (1998). Stereotyping, prejudice, and discrimination. In D. Gilbert,
S. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th
ed., Vol. 2, pp. 357–411). New York: McGraw-Hill.
Fiske, S., Xu, J., Cuddy, A., & Glick, P. (1999). (Dis)respecting versus
(dis)liking: Status and interdependence predict ambivalent stereotypes of
competence and warmth. Journal of Social Issues, 55, 473–489.
Giles, H., & Ryan, E. (1982). Prolegomena for developing a social psy-
chological theory of language attitudes. In E. Ryan & H. Giles (Eds.),
Attitudes toward language variation (pp. 208–223). London: Arnold.
Goertzel, T. (1987). Authoritarianism of personality and political attitudes.
Journal of Social Psychology, 127, 7–18.
Goldberg, L. (1999). A broad-bandwidth, public-domain, personality in-
ventory measuring the lower-level facets of several five-factor models.
In I. Mervielde, I. Deary, F. de Fruyt, & F. Ostendorf (Eds.), Personality
psychology in Europe (Vol. 7, pp. 7–29). Tillburg, the Netherlands:
Tillburg University Press.
Guelke, A. (1996). Dissecting the South African miracle. Nationalism and
Ethnic Politics, 2, 141–154.
Hartley, E. (1946). Problems in prejudice. New York: King’s Crown Press.
Hoyle, R., & Smith, G. (1994). Formulating clinical research hypotheses as
structural equation models: A conceptual overview. Journal of Consult-
ing and Clinical Psychology, 62, 429–440.
Hu, L., & Bentler, P. (1999). Cutoff criteria for fit indexes in covariance
structure analysis: Conventional criteria versus new alternatives. Struc-
tural Equation Modeling, 6, 1–55.
Hunter, J., Gerbing, D., & Boston, F. (1982). Machiavellian beliefs and
personality: Construct invalidity of the Machiavellianism dimension.
Journal of Personality and Social Psychology, 43, 1293–1305.
Johnson, R., & Schlemmer, L. (1994). Launching democracy in South
Africa. New Haven, CT: Yale University Press.
Johnson, W. (1994). Dismantling apartheid. Ithaca, NY: Cornell Univer-
IDEOLOGY AND PREJUDICE
Jo ¨reskog, K., & So ¨rbom, D. (1996). LISREL 8: User’s reference guide.
Chicago: Scientific Software International.
Katz, I., & Hass, R. (1988). Racial ambivalence and American value
conflict: Correlational and priming studies of dual cognitive structures.
Journal of Personality and Social Psychology, 55, 893–905.
Kinder, D., & Sears, D. (1981). Prejudice and politics: Symbolic racism
versus racial threats to the good life. Journal of Personality and Social
Psychology, 40, 414–431.
Kosterman, R., & Feshbach, S. (1989). Toward a measure of patriotic and
nationalistic attitudes. Political Psychology, 10, 257–274.
Kovel, J. (1970). White racism: A psychological history. New York:
Lorr, M. (1986). Manual of the Interpersonal Style Inventory. Los Angeles:
Western Psychological Services.
MacCrone, I. (1937). Race attitudes in South Africa. London: Oxford
Maiti, S., & Mukerjee, B. (1990). A note on the distributional properties of
the Joresko ¨g-So ¨rbom fit indices. Psychometrika, 55, 721–726.
Marsh, H., & Grayson, D. (1995). Latent variable models of multitrait-
multimethod data. In R. Hoyle (Ed.), Structural equation modeling:
Concepts, issues, and applications (pp. 177–198). Thousand Oaks, CA:
McConahay, J., & Hough, J. (1976). Symbolic racism. Journal of Social
Issues, 32, 23–45.
McFarland, S. (1998, July). Toward a typology of prejudiced persons.
Paper presented at the annual meeting of the International Society of
Political Psychology, Montreal, Canada.
McFarland, S., & Adelson, S. (1996, July). An omnibus study of person-
ality, values, and prejudice. Paper presented at the annual meeting of the
International Society of Political Psychology, Vancouver, Canada.
Meloen, J. (1983). De autoritaire reaktie in tijden van welvaart en krisis
[The authoritarian response in times of prosperity and crisis]. Unpub-
lished doctoral dissertation, University of Amsterdam.
Peterson, B., Doty, R., & Winter, D. (1993). Authoritarianism and attitudes
toward contemporary social issues. Personality and Social Psychology
Bulletin, 19, 174–184.
Pettigrew, T. (1958). Personality and socio-cultural factors in intergroup
attitudes: A cross-national comparison. Journal of Conflict Resolu-
tion, 2, 29–42.
Pettigrew, T. (1998). Reactions toward the new minorities of western
Europe. Annual Review of Sociology, 24, 77–103.
Poppe, E., & Linssen, H. (1999). Ingroup favouritism and the reflection of
realistic dimensions of difference between national states in Central and
Eastern European nationality stereotypes. British Journal of Social Psy-
chology, 38, 85–102.
Potthast, M. (1993). Confirmatory factor analysis of ordered categorical
variables with large models. British Journal of Mathematical & Statis-
tical Psychology, 46, 273–286.
Pratto, F. (1999). The puzzle of continuing group inequality: Piecing
together psychological, social, and cultural forces in social dominance
theory. In M. P. Zanna (Ed.), Advances in experimental social psychol-
ogy (Vol. 31, pp. 191–263). San Diego, CA: Academic Press.
Pratto, F., Sidanius, J., Stallworth, L., & Malle, B. (1994). Social domi-
nance orientation: A personality variable predicting social and political
attitudes. Journal of Personality and Social Psychology, 67, 741–763.
Ross, M. (1993). The culture of conflict. New Haven, CT: Yale University
Sales, S. (1973). Threat as a factor in authoritarianism. Journal of Person-
ality and Social Psychology, 28, 44–57.
Sales, S., & Friend, K. (1973). Success and failure as determinants of level
of authoritarianism. Behavioral Science, 18, 163–172.
Saucier, G. (1994). Separating description and evaluation in the structure of
personality attributes. Journal of Personality and Social Psychology, 66,
Saucier, G. (2000). Isms and the structure of social attitudes. Journal of
Personality and Social Psychology, 78, 366–385.
Schafer, J. (1997). Analysis of incomplete multivariate data. London:
Chapman & Hall.
Schwartz, S. (1992). Universals in the content and structure of values:
Theoretical advances and empirical tests in 20 countries. In M. Zanna
(Ed.), Advances in experimental social psychology (Vol. 25, pp. 1–65).
San Diego, CA: Academic Press.
Sherif, M. (1967). Group conflict and cooperation. London: Routledge and
Shumacker, R., & Lomax, R. (1996). A beginner’s guide to structural
equation modeling. Mahwah, NJ: Erlbaum.
Sidanius, J., & Pratto, F. (1993). The dynamics of social dominance and the
inevitability of oppression. In P. Sniderman & P. Tetlock (Eds), Preju-
dice, politics, and race in America today (pp. 173–211). Stanford, CA:
Stanford University Press.
Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory
of social hierarchy and oppression. Cambridge, England: Cambridge
Stangor, C., & Crandall, C. (2000). Threat and the social construction of
stigma. In T. Heatherton, R. Kleck, M. Helb, & J. Hull (Eds.), The social
psychology of stigma (pp. 62–87). New York: Guilford Press.
Steiger, J. (1989). EzPATH: A supplementary module for SYSTAT and
SYGRAPH. Evanston, IL: SYSTAT, Inc.
Stephan, W., Ybarra, O., Martinez, C., Schwarzwald, J., & Tur-kaspa, M.
(1998). Prejudice toward immigrants to Spain and Israel: An integrated
threat theory analysis. Journal of Cross-Cultural Psychology, 29, 559–
Stone, W., Lederer, G., & Christie, R. (1993). The status of authoritarian-
ism. In W. Stone, G. Lederer, & R. Christie (Eds.), Strength and
weakness: The authoritarian personality today (pp. 229–245). New
Strauss, C. (1992). Models and motives. In R. D’Andrade & C. Strauss
(Eds.), Human motives and cultural models (pp. 1–20). Cambridge,
England: Cambridge University Press.
Struch, N., & Schwartz, S. (1989). Intergroup aggression: Its predictors and
distinctness from in-group bias. Journal of Personality and Social Psy-
chology, 62, 564–576.
Tajfel, H., & Turner, J. (1979). An integrative theory of intergroup conflict.
In W. Austin & S. Worchel (Eds.), The social psychology of intergroup
relations (pp. 33–47). Monterey, CA: Brooks/Cole.
Thompson, L. (1995). A history of South Africa. New Haven, CT: Yale
Wilson, D., Near, D., & Miller, R. (1996). Machiavellianism: A synthesis
of the evolutionary and psychological literatures. Psychological Bulletin,
Verkuyten, M., & Hagendoorn, L. (1998). Prejudice and self-categoriza-
tion: The variable role of authoritarianism and in-group stereotypes.
Personality and Social Psychology Bulletin, 24, 99–110.
Yung, Y., & Bentler, P. (1994). Bootstrap-corrected ADF test statistics in
covariance structure analysis. British Journal of Mathematical & Sta-
tistical Psychology, 47, 63–84.
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
Items Measuring the Two Personality Dimensions
following descriptive adjectives is characteristic or uncharacteristic of
YOUR PERSONALITY AND BEHAVIOUR.”
The instruction read, “Rate the extent to which you feel each of the
IDEOLOGY AND PREJUDICE
The Construct Definitions and Items of the Two World View Dimensions
Dangerous and threatening social world view:
Belief that the social world is a dangerous and threatening place in which good, decent people’s values and way of
life are threatened by bad people versus belief that the social world is a safe, secure and stable place in which almost
all people are fundamentally good
1. Although it may appear that things are constantly getting more dangerous and chaotic, it really isn’t so. Every
era has its problems, and a person’s chances of living a safe, untroubled life are better today than ever before
2. Any day now chaos and anarchy could erupt around us. All the signs are pointing to it
3. There are many dangerous people in our society who will attack someone out of pure meanness, for no reason at
4. Despite what one hears about “crime in the street,” there probably isn’t any more now than there ever has been
5. If a person takes a few sensible precautions, nothing bad is likely to happen to him or her; we do not live in a
6. Every day as society become more lawless and bestial, a person’s chances of being robbed, assaulted, and even
murdered go up and up
7. My knowledge and experience tells me that the social world we live in is basically a safe, stable and secure
place in which most people are fundamentally good
8. It seems that every year there are fewer and fewer truly respectable people, and more and more persons with no
morals at all who threaten everyone else
9. The “end” is not near. People who think that earthquakes, wars, and famines mean God might be about to
destroy the world are being foolish
10. My knowledge and experience tells me that the social world we live in is basically a dangerous and
unpredictable place, in which good, decent and moral people’s values and way of life are threatened and
disrupted by bad people
Competitive jungle social world view:
Belief that the social world is a competitive jungle characterized by a ruthless, amoral struggle for resources and
power in which might is right and winning everything versus belief that the social world is a place of cooperative
harmony in which people care for, help, and share with each other
1. Winning is not the first thing; it’s the only thing
2. The best way to lead a group under one’s supervision is to show them kindness, consideration, and treat them as
fellow workers, not as inferiors
3. If one has power in a situation, one should use it however one has to in order to get one’s way
4. If it’s necessary to be cold blooded and vengeful to reach one’s goals, then one should do it
5. Life is not governed by the “survival of the fittest.” We should let compassion and moral laws be our guide
6. Money, wealth and luxury are what really count in life
7. It is better to he loved than to be feared
8. It is much more important in life to have integrity in your dealings with others than to have money and power
9. It’s a dog-eat-dog world where you have to be ruthless at times
10. Charity (i.e., giving somebody something for nothing) is admirable not stupid
11. You know that most people are out to “screw” you, so you have to get them first when you get the chance
12. All in all it is better to be humble and honest than important and dishonest
13. My knowledge and experience tells me that the social world we live in is basically a competitive “jungle” in
which the fittest survive and succeed, in which power, wealth, and winning are everything, and might is right
14. Honesty is the best policy in all cases
15. There is really no such thing as “right” and “wrong.” It all boils down to what you can get away with
16. Do unto to others as you would have them do unto you, and never do anything unfair to someone else
17. One of the most useful skills a person should develop is how to look someone straight in the eye and lie
18. Basically people are objects to be quietly and coolly manipulated for one’s own benefit
19. One should give others the benefit of the doubt. Most people are trustworthy if you have faith in them
20. We can make a society based on unselfish cooperation, sharing and people generously helping each other, and
NOT on competition and acquisitiveness
Enemies of Freedom: Understanding Right-Wing Authoritarianism (pp. 74, 78) by B. Altemeyer, 1988, San Francisco:
Jossey-Bass. Copyright 1988 by Jossey-Bass. Reprinted by permission of John Wiley & Sons, Inc. All competitive-jungle
dimension items except Items 13 and 20 were taken or adapted from the Personal Power, Meanness, and Dominance Scale
and Exploitive Manipulative Amoral Dishonesty Scale from “The Other ‘Authoritarian Personality’ ” by B. Altemeyer in
Advances In Experimental Social Psychology, Vol. 30, edited by Mark P. Zanna, copyright 1998 by Elsevier Science (USA),
reproduced by the permission of the publisher.
All items in the dangerous world dimension except Items 7 and 10 are from the Dangerous World Scale from
DUCKITT, WAGNER, DU PLESSIS, AND BIRUM
Received February 8, 2001 Download full-text
Revision received October 24, 2001
Accepted October 31, 2001 ?
Fit Indices for Alternative Models
Study and analysis
RMSEA SRMR GFICFI
Full alternative model
Social dominance alternative model
(SDO 3 CW 3 prejudice)
Study 1 (generalized prejudice)
Study 2 (anti-African prejudice)
Study 3 (anti-Indian prejudice)
Current model’s equivalent to
social dominance model
(CW 3 SDO 3 prejudice)
Study 1 (generalized prejudice)
Study 2 (anti-African prejudice)
Study 3 (anti-Indian prejudice)
alternative model is a latent variable model with social conformity, dangerous world beliefs, and RWA caused by Latent
Variable X and toughmindedness, competitive world beliefs, and SDO caused by Latent Variable Y. RMSEA ? root-mean-
square error of approximation; SRMR ? standardized root-mean-square residual; CFI ? comparative fit index; SDO ?
social dominance orientation; CW ? competitive world; RWA ? right-wing authoritarianism.
Goodness-of-fit index (GFI) is population GFI (Steiger, 1989). Study 1, N ? 146; Study 2, N ? 233. The full
IDEOLOGY AND PREJUDICE