Upload
kent
View
0
Download
0
Embed Size (px)
Citation preview
Intergroup Friendship, the Far Right and Attitudes toward
Immigration and Islam
David Cutts, Matthew Goodwin, Miles Hewstone and Simon Lolliot1
Abstract. Direct cross-group friendships have been widely found to
not only diminish prejudice but induce greater empathy, promote
tolerance, enable members of majority groups to feel at ease with
minorities as well as providing a gateway to wider cross-group
networks. Put simply, as the quality of cross-group friendships
increases, so does its prejudice-reducing effects. But do these
relationships exist among those with more extreme or less tolerant
views, such as supporters of parties on the extreme right of the
political spectrum? Using a unique sample of self-identified
supporters of three distinct types of right-wing groups, we examined
whether direct cross-group friendships with immigrants reduce
prejudice towards immigrants and Islam and whether diversity beliefs
mediate this relationship. Our findings provide further evidence
that building direct cross-group friendships is a highly effective
tool in reducing prejudice towards both immigrants and Islam, among
all three right-wing groups. While there were similar reductions
among those supporting radical and extreme right parties, counter-
Jihad movements experienced the largest decline in prejudicial
attitudes.
1 David Cutts, University of Bath. Matthew Goodwin, University of Kent, Miles Hewstone, University of Oxford. Simon Lolliot, University of British Columbia. The corresponding author for this version is Prof. Matthew Goodwin [email protected]
1
It is well established that intergroup contact, especially direct
cross-group friendships, can have substantial positive effects on
reducing prejudice toward a wide range of ethnic, religious and
other outgroups. Numerous studies show that individuals who report
having direct cross-group friendships are less intolerant than those
who lack these relationships. But to what extent does this hold true
among supporters of right-wing parties and social movements that
across Western states continue to attract significant support by
overtly campaigning against migrants, ethnic minorities and,
increasingly, Islam? Our research builds on and expands the existing
psychological literature by investigating the effects of direct
cross-group friendships with members of ethnic minorities among
supporters of radical and extreme right-wing groups on their
attitudes toward immigrants and Islam. Based on a unique sample of
self-identified supporters of three distinct types of right-wing
groups, we contribute to this literature by undertaking a far
tougher test for intergroup contact theory than the traditional
reliance upon undergraduate samples with relatively mild attitudes.
To date, few studies examined these relationships and their
effects among ‘real world’ and self-identified supporters of right-
wing groups, and none have been able to compare the impact of cross-
group friendship among members of distinctive groups across the
right-wing political spectrum. Among these supporters, we explore
3
whether and how this more substantive form of intergroup contact
does indeed reduce prejudice towards outgroups, and specifically
whether any impact of cross-group friendships on prejudice reduction
is mediated via so-called diversity beliefs, namely belief in the
perceived benefits of diversity for enriching economic and cultural
life (Tropp & Bianchi, 2006), which can function as individual-level
beliefs and also, when aggregated, social norms (Christ et al.,
2014). We explore whether these effects differ according to three
distinct types of supporters, who variously endorse the populist
radical right, extreme right-wing, and the so-called counter-Jihad
movement that exclusively campaigns against Islam. This
differentiation augments the existing literature by simultaneously
enabling us to disentangle the effects of cross-group friendships on
different types of movements that espouse prejudice. We contribute
to the extant literature by investigating whether contact is not
only a viable technique for reducing intergroup animosity among
general populations, but also for more highly-prejudiced people who
actively support movements that oppose to multiculturalism and
rising diversity.
Intergroup Contact and Friendships
4
In social psychology, a large literature consistently stresses the
ability of intergroup contact to reduce prejudice towards out-groups
(Brown & Hewstone 2005; Pettigrew & Tropp 2006). Around six decades
of research lends impressive support to Allport’s (1954) contact
hypothesis—intergroup contact, especially under the conditions of
equal status, cooperative work towards a common goal, and authority
support will reduce prejudice. We now know that cross-group contact
affects a multitude of outcomes (i.e., not just prejudice) towards a
variety of out-groups (Pettigrew & Tropp, 2011). Cross-group contact
is especially effective at reducing prejudice by lowering intergroup
anxiety and increasing empathy (for meta-analytic support, see
Pettigrew & Tropp, 2008), and by fostering a common group identity
that transcends original group boundaries (Brown & Hewstone 2005;
Gaertner & Dovidio, 2000). Not all forms of intergroup contact are
the same, however. Pettigrew and Tropp’s (2006) meta-analysis of
over 500 intergroup contact studies found that as the quality of
contact increased, so did its ability to reduce prejudice, thus
highlighting the importance of contact quality over contact
quantity.
Cross-group friendships are regarded as a more substantive form
of interaction, as they denote sustained positive contact across
different social contexts that incorporate most of Allport’s
original facilitating conditions (Pettigrew, 1997). Since the
5
reformulation of the ‘contact hypothesis’ (Pettigrew et al., 2011),
this literature has focused heavily on the perceived power of direct
cross-group friendship as a powerful conduit for maximizing the
beneficial effects on attitudes. Friendships are believed not only
to encourage prejudice reduction (Davies et al. 2011) but induce
greater understanding and empathy, enhance tolerance, trust, mutual
respect and cooperation (Brown & Hewstone, 2005; Turner, Hewstone, &
Voci, 2007). Davies et al. (2011), in their meta-analytic review of
cross-group friendships and prejudice, replicated Pettigrew and
Tropp’s (2006) finding that as the quality of the cross-group
friendship increased, so did its prejudice-reducing effects. Their
meta-analysis, using a larger data base and including several
longitudinal studies, also confirmed that cross-group friendships
can have wider effects by providing a gateway to cross-group
networks (Pettigrew et al., 2007), or encouraging further contact
(Page-Gould, Mendoza-Denton, and Tropp, 2008).
The positive contact–lower prejudice relationship has been
found to be significant even after accounting for selection bias,
and existing evidence suggests that the contact-prejudice path is
stronger than the reverse path from prejudice to contact (Pettigrew
& Tropp, 2006). Moreover, these contact effects have been found to
hold over time (Van Laar et al, 2005) even when the possibility of
participant selection has been removed, and hardly vary across
6
countries where intergroup contact research has been conducted
(Pettigrew et al., 2011). They also generalize to the entire groups
involved, new situations, and even to out groups not initially
engaged in the original contact situation (Lolliot, et al., 2013;
Pettigrew et al, 2011). Studies even show how indirect contact –
simply having in-group friends who have out-group friends – can
weaken prejudice (Wright et al. 1997; for a review see Vezzali et
al., 2014). This is of great significance to those individuals in
more segregated locations, with no (possibility for) out-group
contact, as such indirect influences may act as a mechanism
preparing these individuals for later contact (Christ et al., 2010;
Pettigrew et al, 2011).
Since its inception, intergroup contact has proved effective
at reducing prejudice towards a host of outgroups, including groups
based on racial and ethnic distinctions, sexual orientation,
physically and mentally disabled, and the elderly (Pettigrew &
Tropp, 2006). A few studies have looked at the relationship between
intergroup contact and attitudes towards Muslims. These provide a
demanding test of the contact hypothesis, given that prejudice and
discrimination against Muslims has increased dramatically in the
last decade (Halliday, 1999) and they suffer worse discrimination
than members of other religious (Pew Forum on Religion and Public
Life, 2009) or immigrants (Strabac & Listhaug, 2008). For instance,
7
Pettigrew et al. (2007), basing their findings on a probability
sample of German respondents aged 16 or older, found that contact
with foreigners as friends and good acquaintances was significantly
correlated with lower Muslim-prejudice. They also found that
respondents who had friends that had foreigners as friends (i.e.,
extended contact, Wright et al., 1997), also reported less Muslim
prejudice than those who did not. Pettigrew (2009), using data from
two national probability telephone surveys, reported that contact
with foreigners was significantly negatively correlated with
prejudice towards Muslims. Savelkoul et al. (2011), using a Dutch
sample survey from the Social and Cultural Developments in the
Netherlands survey, found that contact with ethnic minorities as
colleagues was associated with less Muslim antipathy. Finally,
Hewstone and Schmid (2014), in a neighbourhood-level analysis of a
random sample of English respondents, tested both direct and
indirect (via intergroup contact) effects of diversity on outgroup
orientations toward Muslims. They found that individuals living in
more ethnically diverse areas - regardless of whether they are White
British members of the majority or non-Muslim members of ethnic
minorities - had more positive contact with Muslims, with positive
consequences for intergroup relations with Muslims.
Thus, research has investigated and confirmed the effect of
intergroup contact, especially cross-group friendships, on anti-
8
Muslim sentiment. These studies, however, investigated the effects
of contact on anti-Muslim prejudice amongst the general population.
A stricter test of contact theory would be to see if it still has
prejudice-reducing effects for samples that are overtly prejudiced.
Intergroup Friendship and Supporters of Prejudiced Groups
To what extent, if at all, do intergroup friendships improve
attitudes among people who belong to openly intolerant political
movements? As Hodson (2011) summarised, early contributors to
intergroup contact theory were initially pessimistic about the
potential for contact to benefit ideologically-oriented, intolerant
and highly-prejudiced persons, whether because these strongly
‘negative’ personalities would overshadow the positive benefits of
contact, or that more highly prejudiced people would inflame
prejudices in face-to-face contact settings, or such bigots would
avoid opportunities for intergroup contact (Allport 1954; Amir
1969). However, building on more recent work (Dhont & Van Hiel 2009;
Hodson, Harry & Mitchell 2009), Hodson (2011) argued that there is
strong reason to anticipate positive contact effects among the
highly prejudiced because contact diminishes anxiety and enhances
empathy, while simultaneously increasing inclusion-of-other-in-the-
self (“psychological closeness”). Indeed, based on a review of nine
9
studies, Hodson (2011) concluded that respondents high on prejudice-
related individual difference variables clearly benefitted from
increased contact quantity and from positive contact experiences in
eight of the nine studies, whereas individuals who were equivalently
high in intolerant ideologies but who did not experience the same
level and contact quality (i.e., fewer friendships) held more
negative attitudes toward out-groups. Other studies also note how
cross-group friendships can be forged among highly prejudiced
people, who in some cases exhibit significantly less out-group bias
as a linear function of the number of direct and indirect
friendships with members of an out-group (Hodson et al., 2013).
While most of these tests have been performed using self-reported
and explicit measures of intolerance, Page-Gould et al. (2008)
showed that cross-group friendships had a stronger effect at
reducing cortisol activity—a physiological measure of anxiety—during
cross-group interactions for individuals who scored high on implicit
measure of prejudice relative to those less implicitly prejudiced.
In a related study of imagined, rather than actual, intergroup
contact, one study has also found that right-wing authoritarians
show less negative emotions toward out-groups and are more willing
to engage with out-groups after imagined positive contact (Asbrock,
Gutenbrunner & Wagner 2013).
10
Much of the research investigating the beneficial effects of
intergroup contact has primarily focused on student or adult samples
(Pettigrew & Tropp, 2006) in relatively benign settings (Hewstone et
al, 2014). While a growing body of research has started to
investigate the effects of intergroup contact in settings that are
still characterised by conflict (Maoz, 2004) or with groups who have
recently emerged from protracted conflict (Hewstone et al., 2014),
these studies are in the minority. While the research reported above
finds stronger prejudice-reducing effects of intergroup contact for
prejudiced individuals, they were also conducted on samples drawn
from the general population. Hodson’s (2008) study represents one of
the few, if not the only, investigations into the effectiveness of
contact with an atypical sample. He utilized a group of White Class
B and C inmates (i.e., inmates that are moderate risk to public
safety; 40% of the sample in Study 1 reported having committed a
violent crime) at one small (Study 1) and one large (Study 2) UK
prison. Hodson asked white inmates about their contact with and
attitudes towards Black inmates; he also obtained Social Dominance
Orientation (SDO) scores for the inmates (SDO assesses people’s
desire to promote intergroup hierarchies, and for their in-groups to
dominate their out-groups; see Sidanius & Pratto, 1999). In Study 1,
Hodson found that contact with Black inmates was associated with
reduced intergroup bias, even after controlling for SDO. He also
11
found that the contact-intergroup bias relationship was significant
only for inmates high in SDO and not for those low in SDO. This
study is one of the only studies to our knowledge that tests contact
as a means of prejudice reduction in an atypical sample. The
research we present here investigates whether cross-group friendship
is associated with a reduction in prejudice towards immigrants and
Muslims amongst three samples that are overtly prejudiced, namely,
members of the (a) English Defence League, (b) British National
Party, and (c) the UK Independence Party.
Right-Wing Movements and Islam
Exploring the effects of cross-group friendships on supporters of
right-wing political groups is particularly intriguing given recent
developments. Across Europe, the rise and continued success of
populist radical right political parties has been one of the most
noticeable developments in recent years. There has emerged a new
consensus among political scientists that the defining features of
the populist radical right are a combination of (at least) nativism,
authoritarianism and populism (Mudde, 2007). The populist radical
right is often hostile to Islam but this intolerance is part of a
broader exclusionary platform that encompasses opposition to
multiculturalism and rapid social change, including the arrival of
12
migrants, asylum-seekers, and religious minority groups, of which
Muslims are only one, and often sexual minorities (Betz & Johnson
2004; Rydgren 2008). For example, Zúquete (2008) notes how such
parties infused their existing discourse with anti-Muslim prejudice,
with ‘opposition to visions of a Muslim takeover’ leading them to
often stress Christian identity themes and embrace pro-Jewish and
pro-Israel positions. Influenced by politicians like Pim Fortuyn and
Geert Wilders in the Netherlands, Mayer (2013: 163) observes how
parties like Marine Le Pen’s National Front frame their supporters
as ‘the defenders of equality, liberty and tolerance against their
main enemy, Islam, described as a religion of fanaticism and
intolerance, incompatible with democratic values and Western
culture’. However, while some political scientists have sought to
explore the role of intergroup contact on electoral support for
populist radical right parties, very few, if any studies have been
able to systematically investigate the effects of cross-group
friendships on self-identified supporters of these parties.
Yet the populist radical right has not been the only movement
to target public anxieties over rising ethnic and religious
diversity. Two other types of movements also campaign on these
issues. Further along the spectrum are traditional extreme right-wing
political parties, such as Golden Dawn in Greece and the National
Democratic Party of Germany that are often rooted in overtly neo-
13
Nazi or neo-fascist currents. Unlike the populist radical right, the
extreme right is defined by its adherence to so-called ‘classical’
or ‘biological’ racism, its complete rejection of human equality and
hostile stance toward liberal democracy (Carter, 2005). Meanwhile,
since 2009 several states have also witnessed the rise of more
explicitly anti-Islam social movements, which are single-issue in
focus. Unlike the populist radical right and the extreme right,
these so-called counter-Jihad movements like the ‘defence leagues’
shunned electoral politics in favour of confrontational
demonstrations and marches against the perceived ‘Islamification’ of
Western states (Goodwin, 2013). Their obsession with the single
perceived threat from Islam was reflected in one small-scale
qualitative study (Busher, 2013), which found that self-identified
counter-Jihad activists heavily emphasised the perceived threat from
‘militant Islam’, noting: ‘It is in these broader narratives that
some activists make use of concepts such as a ‘clash of
civilizations’ between Islam and the West (Huntingdon, 1993) or make
reference to ‘Eurabia’ theories (Ye’or, 2005) and the on-going
‘Islamification’ of Europe through immigration and through higher
birth rates among Europe’s Muslim populations’.
Such groups offer an interesting opportunity to test the
potential prejudice-ameliorating effects of intergroup friendship.
In Britain, these three distinct varieties of right-wing politics
14
are represented by three groups. Founded in 1993, the populist radical
right UK Independence Party (UKIP) was originally a Eurosceptic party
that campaigned exclusively to end Britain’s continuing membership
of the European Union. However, since 2010 the party broadened its
ideological message to include the populist radical right’s
opposition to immigration and a populist critique of the established
parties (Ford & Goodwin, 2014). In 2010, UKIP steered closer to
overt hostility to Islam by calling for a ban on the burqa or veiled
niqab in public buildings and for deporting radical Muslim clerics,
although such policies stood alongside its broader call for a five-
year freeze on immigration and an end to the ‘active promotion of
the doctrine of multiculturalism by local and national government’
(UKIP 2010). While the party would move away from these explicitly
anti-Islam policies, in the aftermath of the ‘Charlie Hebdo attacks’
in Paris in 2015 the UKIP leader, Nigel Farage, claimed that a
‘fifth column’ was living in Britain, a reference to perceived
threats from some settled British Muslims.
The extreme right-wing British National Party (BNP) was founded in
1982, and is rooted not in Euroscepticism but an undercurrent of
neo-Nazism (Goodwin, 2011). Unlike UKIP, the BNP advocated a more
openly racist and Islamophobic programme, calling for the voluntary
repatriation of all migrants and going further in its opposition to
Islam, issuing a call to ‘outlaw all overt signs of Islamic
15
colonisation such as the burka, ritual slaughter, mosques and
minarets (which are little more than symbols of territorial
conquest’ (BNP 2010). In its literature the BNP, whose leader was
charged for inciting racial hatred in 2006 after describing Islam as
‘a wicked and vicious faith’, presents the religion as being
incompatible with modern secular western democracy and as ‘one of
the most deadly threats yet to the survival of our nation’. Between
2010 and 2015 the BNP suffered a sharp decline in support although
the party continues to campaign against child sexual exploitation in
northern towns, which it traces to settled Muslim communities (BNP,
2010).
The counter-Jihad English Defence League (EDL) emerged in 2009
and in response to protests by radical Islamic extremists at a
homecoming parade for military forces returning from Iraq. Over the
next three years the mainly working-class movement became one of the
most significant counter-Jihad groups in Western states, with around
80,000 followers on Facebook and more than fifty demonstrations that
attracted between one and three thousand supporters. Such activity
sought to rally public opposition to Islam, and Islamic
institutions, and claimed that ‘Muslim gangs’ were coordinating the
sexual exploitation of young children; the EDL generally portrayed
the religion of Islam as a threat to Britain and its people. Unlike
the extreme and radical right, however, the EDL focused exclusively
16
on Islam and claimed allegiance to ethnic and sexual minorities,
such as Hindus, Sikhs and LGBT groups that opposed the spread of
Islam. The EDL presented itself as a ‘human rights organization’
that wanted ‘to protect the inalienable rights of all people to
protest against radical Islam’s encroachment into the lives of non-
Muslims’.1
Hypotheses
In summary, we extend previous research on intergroup contact and
prejudice by systematically examining the effects of intergroup
friendship on blatant prejudice towards immigrants and attitudes
toward Islam among a sample of self-identified supporters of radical
and extreme right-wing groups. Because we were interested
specifically in attitudes towards two outgroups targeted by right-
wing ideology, immigrants and Muslims, and to avoid shared method
variance (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003) between
predictor (contact measure of cross-group friends) and outcomes
(prejudice towards imigrants and Islam), we used a measure of cross-
group friendships with ethnic minorities in general. Then, we
examine whether this relationship is mediated by diversity beliefs.
Lastly, we innovate further by investigating whether these effects
vary according to whether people belong to a more ‘moderate’
17
populist radical right political party, a hard-line extreme right
party or an extra-parliamentary anti-Islam street movement. To
recap, our hypotheses are as follows:
1) Direct cross-group friendships with ethnic minorities will have a
significant negative direct effect on prejudice to immigrants and
attitudes to Islam. More precisely, there is an expectation that
supporters from all three groups who report cross-group
friendships will experience reductions in prejudice to Islam and
migrants.
2) Direct cross-group friendships with ethnic minorities will be
associated with more favourable beliefs regarding ethnic and
cultural diversity. More precisely, supporters from all three
groups who have cross-group friendships are significantly more
likely to view migrants as enriching the economic and cultural
life of Britain; the direct effect of diversity beliefs will in
turn be negatively associated with blatant prejudice towards
migrants and Islamic threat.
3) Direct cross-group friendships will significantly reduce blatant
prejudice towards immigrants and Islamic threat via the indirect
effect of diversity beliefs. More precisely, among supporters of
these right-wing political groups diversity beliefs will mediate
the effect between cross-group friendships and prejudice toward
immigrants and Muslims. However, we would expect the size of
18
these mediated effects to differ according to whether supporters
self-identify with populist radical right, extreme right, or a
counter-jihad street movement. Our expectation is that supporters
of the more extreme neo-Nazi BNP and counter-Jihad EDL, which
have made opposition to Islam its guiding mission, who have
direct cross-group friendships, will be the most likely to report
larger reductions in overall prejudice toward minorities. In
particular, we would expect supporters of the EDL, whether as a
result of direct effects or of mediated heightened beliefs in
diversity, to report the sharpest reductions in prejudice toward
Islam.
Data and Methods
Participants
The sample was drawn from the YouGov online panel and was collected
in 2012. It consisted of N = 2152 self-identified right-wing
supporters. Of this total, 186 identified themselves as supporters
of the EDL (Mage = 45.96, SD = 16.23, 63.4% male); 626 as supporters
of the BNP (Mage = 50.71, SD = 14.67, 70.9% male); and 1340 as
supporters of UKIP (Mage = 58.97, SD = 12.92, 72% male). All analyses
were conducted in MPlus 6.1 (Muthén & Muthén, 1998-2010) using the
maximum likelihood estimator with robust standard errors. We used
19
bootstrapping with 5000 re-samples and bias corrected confidence
intervals to test for mediation given its ability to handle skewed
data while returning accurate Type I and II error rates (Fritz &
MacKinnon, 2007). Evidence of a mediated effect is indicated by the
95% (or 99%) confidence interavals not including zero.
Measures
Respondents from the three groups completed an online YouGov survey
from which we derived the following measures to test our key
hypotheses. Intergroup contact was measured with one item.
Participants were asked, “When thinking about your friends, would
you say you have many friends, some friends, or no friends at all
among people from ethnic minority groups?” and they indicated their
answer on a three point rating scale (1 = Many, 2 = Some, 3 = No
friends). This item was reverse coded so that higher scores indicate
more friendships with people from ethnic minority groups. Two items
were used to tap diversity beliefs. On two 11-point scales ranging
from 0 (Cultural life undermined / Extremely bad for economy) to 10 (Cultural life
enriched / Extremely good for economy), participants were asked, “On a scale
of 0 to 10, would you say that Britain’s cultural life is generally
undermined or enriched by migrants” and “On a scale of 0 to 10,
would you say it is generally bad or good for Britain’s economy that
20
migrants come to Britain from other countries.” These two items were
strongly and significantly correlated with each other in all three
samples (EDL r = .73, p < .001, α = .84; BNP r = .58, p < .001, α
= .73; UKIP r = .59, p < .001, α = .74).
We used four items to assess blatant prejudice towards
immigrants. On a 5-point rating scale, participants indicated the
degree to which they disagreed / agreed with the following
statements, “Immigrants should be sent back to their home country
whether or not they break the law”, “Family members of immigrants
who were born in the UK should be sent back to their country of
origin”, “Immigrants are the main cause of crime in society”, and
“Immigrants are the main source of disease in society” (1 = Strongly
Disagree, 5 = Strongly Agree). These four items displayed good reliability
in all three samples (EDL α = .89; BNP α = .83; UKIP α = .80).
Higher scores reflect more prejudice towards immigrants. Islamic
threat was measured using two items. On a 5-point rating scale,
participants indicated their agreement with the following two items:
“Islam does not pose a serious danger to Western civilization”
(reverse scored) and “There will be a ‘clash of civilizations’
between Islam and the West” (1 = Strongly Disagree, 5 = Strongly Agree).
Following re-coding of the first item, these two items correlated
significantly with each other in all three samples (EDL r = .48, p
< .001, α = .64; BNP r = .52, p < .001, α = .67; UKIP r = .49, p
21
< .001, α = .65). In our full models we controlled for respondent
age, gender (-1 = Male, 1 = Female), social grade (1 = AB, 2 = C1, 3
= C2, 4 = D), and terminal age of education (1 = 15 years or under to 5
= 20+).
All participants reported having some cross-group friendships
(EDL, BNP, and UKIP Mode = 2). BNP participants scored significantly
above the scale middle with regards to their blatant prejudice
towards immigrants (M = 3.68, SD = 1.02, t(622) = 16.70, p < .001)
whereas UKIP respondents reported attitudes equal to the scale
middle (M = 3.00, SD = 0.98, t(1334) = 0.14, p > .05). EDL
participants, on the other hand, scored significantly below the
scale mid-point (M = 2.66, SD = 1.09, t(185) = -4.20, p < .001). A
between-subjects analysis of variance (ANOVA) indicated that the
blatant prejudice scores for all three samples were significantly
different from each other, F(2, 2141) = 122.78, p < .001, partial η2
= .10. All three groups reported high levels of Islamic threat with
each sub-sample scoring significantly above the scale mid-point (BNP
M = 4.48, SD = 0.81; UKIP M = 4.37, SD = 0.80; EDL M = 3.66, SD =
1.02; all ts > 8.87, all ps < .001). A between-subjects ANOVA
indicated that all three mean Islamic threat scores were
significantly different from each other, F(2, 2141) = 72.58, p
< .001, partial η2 = .06. With regards to diversity beliefs, EDL
respondents reported the most favourable diversity beliefs, that
22
were significantly above the scale mid-point (M = 5.40, SD = 2.45,
t(184) = -3.35, p = .001). Both UKIP and BNP respondents reported
diversity beliefs that were below the scale midpoint (UKIP M = 3.80,
SD = 2.11, t(1336) = -38.19, p < .001; BNP M = 2.77, SD = 2.00, t(623)
= -40.34, p < .001). A between-subjects ANOVA indicated that all
three diversity belief scores were significantly different from each
other, F(2, 2143) = 121.78, p < .001, partial η2 = .10.
Factorial Invariance of the Measurement Model
We performed a series of confirmatory factor analyses (CFA) to
assess if the factor structure of the three latent variables -
diversity beliefs, attitudes towards immigrants, and attitudes
towards Islam - was invariant between the populist radical right
party (UKIP), the extreme right-wing party (BNP) and the counter-
Jihad political movement (EDL). First, we first tested for
configural invariance (i.e., do the latent variables have the same
underlying factor structure between the two samples) for the extreme
right (UKIP; Model 1a), radical right (BNP; Model 1b), and counter-
Jihad movement (EDL; Model 1c) respondents. We then tested for
configural invariance for all samples combined (Model 2). Second, we
tested for metric invariance between the samples (see below) by
constraining the factor loadings of the respective items to equality
23
(Model 3). Partial metric invariance is the minimum requirement when
wanting to compare the structural parameters between groups
(Vandenberg & Lance, 2000). If (at least) partial metric invariance
is achieved we will then test for scalar invariance (Model 4;
Meredith, 1993). This level of model equivalence tests the
tenability that the item intercepts are invariant across the two
samples (Vandenberg & Lance, 2000). If (at least) partial scalar
invariance is achieved, we will then test for uniqueness invariance
(Model 5). Tests for uniqueness invariance determine whether the
explained variance for every item is the same for both groups (Model
5; Vandenberg & Lance, 2000; van de Schoot, Lugtig, & Hox, 2012).
We tested the assumptions for metric, scalar, and uniqueness
invariance in a multi-group CFA whereby we compared the chi-square
statistic for the more restrictive model to the chi-square statistic
for the less constrained model using the scaled chi-square
difference test (Satorra & Bentler, 2001). The latent variables were
scaled using the reference-group method whereby all factor loadings
were allowed to be freely estimated; the latent variance was
constrained to 1; and the latent mean was constrained to 0.
Satisfactory model fit will be taken to be indicated by a x2/df ratio
≤ 3-4; a comparative fit index (CFI) ≥ .95, a root mean square error
of approximation (RMSEA) ≤ .06; and a standardized root mean square
residual (SRMR) ≤ .08.
24
We first tested to see if we could collapse across the BNP and
EDL samples given that they both represent the more extreme right
wing whereas UKIP is a more of a conventional populist radical right
party. Collapsing across the BNP and EDL samples would also give us
a more comparative sample size (combined N = 812) to the 1340 UKIP
sample. All three tests for configural invariance indicated that we
met this assumption (see Model 1a, 1b, 1c). Next, the model
constraining all factor loadings to equivalence between the BNP and
EDL samples (Model 2.1) resulted in a significant drop in model
fit, (Δx2(8) = 24.70, p = .002). We next tested to see if we could
obtain at least partial metric invariance between the two samples.
We could achieve partial metric invariance for the Islamic threat
and blatant prejudice scales, but neither of the factor loadings for
the diversity beliefs items could be successfully constrained to
equality, both Δx2(1) ≥ 5.79, both ps < .02. Further evidence
suggesting that we could not combine the two groups together comes
from the ANOVA results reported above; BNP participants reported
significantly higher levels of blatant prejudice towards immigrants
and more Islamic threat than EDL participants. Furthermore, EDL
respondents reported significantly more favourable diversity beliefs
when compared to BNP participants (Model 2.2).2
Comparing the factor loadings between the BNP and UKIP samples
resulted in a chi-square statistic that was statistically equivalent
25
to the unconstrained model, Δx2(8) = 9.10, p = .33 (Model 2.3; for
all subsequent model fit indices, see Table 3). Given that we could
satisfy metric invariance between BNP and UKIP participants, we
continued to test for scalar and uniqueness invariance between these
two samples. We also continued the multi-group analysis of the
structural relationships between these two samples. However, we will
report the structural model for the EDL sample, but because we were
not able to demonstrate metric invariance, we cannot be sure that
the questionnaire items were interpreted similarly by the BNP and
UKIP samples and thus we will not be able to confidently compare the
structural relationships for EDL respondents to BNP and UKIP
respondents.
We were not able to meet the requirement for full scalar
invariance (Model 3, Table 3), Δx2(5) = 13.55, p = .02; the intercept
for the item “Islam does not pose a serious danger to Western
civilization” was significantly larger for UKIP participants (4.55)
than it was for BNP participants (4.42), Δx2(1) = 7.46, p < .01.
Allowing this item to be freely estimated while constraining the
other intercepts to equality resulted in a model (Model 3.1) that
fit the data statistically equivalently to the metric invariance
model, Δx2(4) = 6.43, p = .17. Lastly, constraining the two groups to
uniqueness invariance (Model 4) did not result in a significant drop
in model fit when compared to the model specifying scalar
26
invariance, Δx2(8) = 6.29, p = .62. Thus, we met the assumptions for
metric, (partial) scalar, and uniqueness invariance between BNP and
UKIP supporters (see Table 4 for the unconstrained factor loadings
for each of the latent variables by sample).
Insert Table 1 about here
Results: Structural Equation Modelling and Multi-group
Comparisons
We tested for three types of structural invariance. First, we tested
for structural invariance between the two samples for the control
variables and, second, we tested for differences in the structural
paths between the two samples excluding diversity beliefs; this
second model would allow us to test hypothesis 1. We did so by first
specifying a model in which no structural constraints were placed
(Model 5; see Table 3) and then where the relationships between the
control variables and the outcomes variables were constrained to
equality between the two samples (Model 6). Next, we specified a
model whereby the remaining structural paths were constrained to
equality between the two samples (Model 7). We then included
diversity beliefs as a mediator and tested for structural invariance
by comparing the freely estimated model (Model 8) to one where all
the structural paths for the control variables were constrained to
27
equality between the two samples (Model 9). Lastly, we specified a
model whereby the remaining structural paths were constrained to
equality between the two samples (Model 10). As with the measurement
model, we tested for the various levels of structural invariance by
comparing the chi-square statistic for the less constrained models
to the chi-square statistic for the more constrained model. Because
we are not interested in the group differences in the effects of our
covariates, we do not report them here, but their fit statistics can
be found in Table 3, and the corresponding regression weights can be
found in Table 5.
For BNP participants, the model explained 64% (95% CI
[.61, .67]) of the variance in blatant prejudice towards immigrants,
17% (95% CI [.13, .21]) of the variance in diversity beliefs, and
38% (95% CI [.34, .42]) of the variance in Islamic threat. For the
UKIP participants, the model explained 61% (95% CI [.56, .66]) of
the variance in blatant prejudice towards immigrants, 9% (95% CI
[.05, .13]) of the variance in diversity beliefs, and 33% (95% CI
[.27, .39]) of the variance in Islamic threat. For the EDL
respondents, the model explained 56% (95% CI [.47, .65]) of the
variance in blatant prejudice towards immigrants, 25% (95% CI
[.14, .36]) of the variance in diversity beliefs, and 49% (95% CI
[.39, .59]) of the variance in Islamic threat.
28
Hypothesis 1: Cross-Group Friendship reduces prejudice towards immigrants and Muslims
Constraining the paths between the independent and dependent
variables to equality between samples (Model 7) and comparing the
resultant fit statistics to those for Model 6.1 indicated a
significant drop in model fit, Δx2(2) = 7.44, p < .05. The
relationship between cross-group friendships and blatant prejudice
towards immigrants could not be constrained to equality, Δx2(1) =
6.11, p = .013. The relationship between cross-group friendships and
blatant prejudice towards immigrants was significant and negative
for BNP participants (b = -.63, SE = .109, β = -.32, p < .001) whereas
it was also negative but significantly weaker for UKIP participants
(b = -.37, SE = .06, β = -.20, p < .001). The direct relationship
between cross-group friendships and Islamic threat was significant
and negative for both samples (b = -.24, SE = .05, β = -.13, p
< .001). For the EDL participants, the structural model showed
adequate model fit, χ2 (27) = 57.890, p < .001, χ2/df = 2.14, SCF =
0.952, CFI = .94, RMSEA = .08 [.05, .11], SRMR = .04. Cross-group
friendships were directly negatively associated with blatant
prejudice towards immigrants (b = -.48, SE = .17, β = -.24, p < .01)
but were not directly associated with Islamic threat (b = -.14, SE
= .18, β = -.08, p > .05). Our findings mostly support hypothesis 1
that direct cross-group friendships were associated with reduced
prejudice among EDL, BNP and UKIP supporters toward immigrants
29
whereas direct cross-group friendships were only directly associated
with lower Islamic threat for BNP and UKIP supporters.
Next, we ran the same set of models but this time included diversity
beliefs as a mediator. Constraining the paths between the
independent, mediator, and dependent variables to equality between
samples (Model 10) and comparing the resultant fit statistics to
those for Model 9.1 indicated a significant drop in model fit, Δx2(5)
= 13.62, p < .02. The relationship between cross-group friendships
and blatant prejudice towards immigrants, once again, could not be
constrained to equality, Δx2(1) = 8.75, p = .003. The relationship
between cross-group friendships and blatant prejudice towards
immigrants remained significant and negative for BNP participants (b
= -.53, SE = .12, β = -.18, p < .001) once the mediator was entered
into the model. The relationship between cross-group friendships
remained negative but was non-significant for UKIP participants (b =
-.11, SE = .08, β = 04, p > .05). The direct relationship between
cross-group friendships and Islamic threat also became non-
significant for both samples (b = -.05, SE = .06, β = -.02, p > .05).
For the EDL participants, the structural model showed adequate model
fit, χ2 (41) = 73.525, p < .002, χ2/df = 1.79, SCF = 0.980, CFI = .96,
RMSEA = .07 [.04, .09], SRMR = .04. The association between cross-
group friends and blatant prejudice towards immigrants became non-
significant (b = -.19, SE = .20, β = -.07, p > .05) once diversity
30
beliefs was entered into the model. The relationship between cross-
group friendships and Islamic threat remained non-significant (b
= .23, SE = .19, β = .09, p > .05).
When the mediator is added to the model, our findings suggest that
direct cross-group friendships did not directly reduce prejudice
among UKIP and EDL supporters toward immigrants and Islam. However,
supporters of the BNP, who reported direct cross-group friendships
reported lower prejudice toward immigrants but not Islam. In sum,
direct cross-group friendships with out-groups did not reduce
prejudice for those on the radical and extreme ends of the political
spectrum, while attitudes towards Islam may well be exempt from
these effects among supporters of the traditional extreme right.
However, we might expect these relationships to become non-
significant when the mediator is added, as our expectation is that
contact still reduces prejudice for all three groups but just
through the mechanism of diversity beliefs. We test whether this is
the case below.
Hypothesis 2: Cross-group friendships will have a significant positive effect on diversity
beliefs
For the BNP and UKIP samples, cross-group friendships were
positively associated with diversity beliefs (b = .39, SE = .05, β
31
= .20, p < .001). More favourable diversity beliefs, in turn, were
significantly associated with lower blatant prejudice towards
immigrants (b = -1.12, SE = .09, β = -.74, p < .001) and lower levels
of Islamic threat (b = -.35, SE = .06, β = -.56, p < .001). For EDL
participants, cross-group friendships were positively associated
with more favourable diversity beliefs (b = .51, SE = .17, β = .25, p
= .003). Diversity beliefs were, in turn, associated with lower
blatant prejudice towards immigrants (b = -.94, SE = .23, β = -.72, p
< .001) and Islamic threat (b = -.86, SE = .20, β = -.71, p < .001).
These findings suggest that direct cross-group friendships have a
significant and positive effect on beliefs in diversity among all
three groups of supporters, indicating that when followers of right-
wing political groups have friendships with members of out-groups
they are more likely to think positively about the economic and
cultural contributions that migrants make to life in Britain. While
we cannot draw direct comparisons between the samples because of the
metric inequality, it is interesting to note that contact was more
strongly related to diversity beliefs for EDL supporters than it was
for BNP and UKIP respondents; relatedly, contact accounted for more
variance in diversity beliefs for EDL respondents than it did for
the other two samples. This is interesting given that EDL
respondents scored the highest on the diversity beliefs scale. This
suggests that, compared to the other right-wing parties, their claim
32
to be a ‘multicultural group’ that is open to the ethnic and
cultural diversity of modern Britain (excluding Islam) appears to
have some validity.
Hypothesis 3: Cross-group friendships significantly reduce prejudice towards immigrants
and Muslims by increasing diversity beliefs
Testing for a mediated effect - which is present when the bias
corrected bootstrap confidence intervals do not include zero -
revealed that diversity beliefs mediated the relationships between
cross-group friendships and blatant prejudice towards immigrants
(point estimate (PE) = -.44, 99% CI [-.64, -.28]; standardized PE =
-.15, 99% CI [-.20, -.10]) and cross-group friendships and Islamic
threat (PE = -.25, 99% CI [-.37, -.16]; standardized PE = -.11, 99%
CI [-.15, -.07]) for both BNP and UKIP respondents. For EDL
participants, diversity beliefs also mediated the relationships
between cross-group friendships and blatant prejudice towards
immigrants (point estimate (PE) = -.48, 99% CI [-1.25, -.03];
standardized PE = -.18, 99% CI [-.34, -.02]) and immigrant
friendships and Islamic threat (PE = -.44, 99% CI [-1.35, -.04];
standardized PE = -.18, 99% CI [-.34, -.01]). Cross-group
friendships reduced prejudice towards immigrants and Muslims by
increasing the view that immigrants contribute positively to the
33
country’s economic and cultural life. Across all three groups,
diversity beliefs mediate the relationship between cross-group
friendship and prejudice towards immigrants and Muslims. When
comparing the standardised coefficients we found that the largest
reduction was for EDL supporters in terms of their attitudes toward
Islam and immigrants. But both the BNP and UKIP reported similar
reductions in prejudice towards the two out-groups although not as
sharply as supporters of the counter-Jihad English Defence League.
Discussion
This study provided a unique investigation of the effects of direct
cross-group friendships on a sample of people who support radical
and extreme right-wing groups that are known for their xenophobic
and anti-Islam political campaigns. Our sample of actual, self-
identified supporters makes an important contribution to the
existing literature that often leans heavily on undergraduate
students, who tend to have less extreme views (Sears, 1986). To our
knowledge there have been few studies that explore such a large
sample of ‘real world’ supporters of prejudiced groups. Given our
unique sample, the study represents a tougher test for intergroup
contact theory than the traditional undergraduate samples in benign
settings (Hewstone et al, 2014). We now summarize the main findings,
34
highlighting some specific issues for further consideration, and
acknowledge some limitations of the research.
We found that, among supporters of right-wing groups, there
was extensive support for our three hypotheses. First, our findings
mostly support hypothesis 1 that direct cross-group friendships with
members of ethnic minorities were associated with reduced prejudice
among all three groups of supporters toward immigrants; however,
direct cross-group friendships were only directly associated with
lower Islamic threat for BNP and UKIP supporters. Second, consistent
with hypothesis 2, direct cross-group friendships were associated
with more favourable beliefs regarding diversity. Our findings
suggest that cross-group friendships have a significant and positive
effect on beliefs in diversity among all three groups, indicating
that when followers of right-wing groups have friendships with
members of ethnic out-groups they are more likely to think
positively about the economic and cultural contributions that
migrants make to life in Britain. In turn, the direct effect of
diversity beliefs is negatively associated with blatant prejudice
towards migrants and Islamic threat. Finally, turning to hypothesis
3, cross-group friendships reduced prejudice towards immigrants and
Muslims by increasing the view that immigrants contribute positively
to economic and cultural life. Across all three groups, diversity
beliefs mediated the relationship between cross-group friendship and
35
prejudice towards immigrants and Muslims. These findings add
substantially to a growing body of research indicating that contact
is a viable technique for reducing intergroup animosity, including
for more highly prejudiced people who need it most, and confirm
prior research on the significance of diversity beliefs as a
mediator of contact effects (Christ et al., 2014; Tropp & Bianchi,
2006).
One interesting finding was that, for BNP and UKIP supporters,
only one path could not be constrained to equality; specifically,
the relationship between cross-group friendships with ethnic
minorities and blatant prejudice toward migrants. This indicates two
possible outcomes. Diversity beliefs as a mediator could not explain
the prejudice-reducing effect of cross-group friendships on
prejudice towards immigrants for supporters of the extreme right
BNP. This suggests that there is another process that our model did
not account for.3 Taking this finding into consideration with the
lower proportion of variance explained in the outcome measures,
especially diversity beliefs, contact may have a stunted effect for
UKIP respondents. While some studies have reported a greater effect
of contact on those who are less tolerant, or more prejudiced (Dhont
& Van Hiel 2009; Hodson, Harry & Mitchell 2009), there is also some
evidence that contact is more strongly associated with reduced
prejudice for more tolerant individuals (Schmid, Hewstone, Küpper,
36
Zick, & Wagner, 2012) and in settings where contact is sanctioned by
authorities (Pettigrew & Tropp, 2006). Furthermore, Pettigrew (1958)
found that sociocultural factors, such as a norm of intolerance,
affect outgroup attitudes over-and-above strong personality
predictors of prejudice, like right-wing authoritarianism. Given
that UKIP has a xenophobic and, at times, Islamophobic agenda (Ford,
Goodwin & Cutts, 2012), this strong norm towards intolerance may
stunt the effects of intergroup contact rather than boost it. It
could be the case, then, that there is a quadratic effect where the
relationship between contact and prejudice increases as prejudice
increases until a tipping point is reached where, in the very
prejudiced, attitues become more solidified. It should be noted,
however, that this interpretation is speculative, given that only
one path differed between BNP and UKIP respondents. Furthermore,
although 8 per cent more variance was explained in diversity beliefs
for BNP respondents than for UKIP respondents, this difference is
not statistically significant given that the confidence intervals
for the R2 values did not overlap.4
The findings reported here are encouraging as they show that
contact is even an effective prejudice-reduction technique for
highly prejudiced, even xenophobic, individuals (for blatant
prejudice towards immigrants, the sample mean for BNP respondents
was significantly above the scale mid-point, and for UKIP and EDL
37
respondents, it was equal to and below the scale mid-point,
respectively; all three groups reported high levels of Islamic
threat, each sub-sample mean being significantly above the scale
mid-point). This study goes beyond the often-used sample of
undergraduate students to test the effectiveness of contact and adds
to the scant body of research that tests contact theory on samples
that are unlikely to meet the conditions underwhich positive contact
is said to have its most beneficial impact. Another important
finding relates to diversity beliefs as a mediator for a sample of
the population who generally hold unfavourable attitudes towards
diversity, individuals on the political far right (whereas student
samples are typically positive, only our EDL respondents reported
diversity beliefs that were significantly above the scale mid-point,
and both UKIP and BNP respondents reported diversity beliefs that
were below the scale midpoint). Holding favourable attitudes towards
diversity has been implicated as a mediator of the effect of
intergroup contact specifically because it is a group-level variable
and is not necessarily group-specific (Lolliot, et al., 2013).
Favourable diversity beliefs should affect attitudes towards many
outgroups as the very notion of diversity should not / cannot be
limited to one or two outgroups. This concept relates closely to
Pettigrew’s (1997) notion of deprovincialization, whereby intergroup
38
contact leads people to hold a less provincialized outlook about
their ingroup in relation to outgroups.
Notwithstanding our robust findings, we acknowledge some
limitations of this research, of which we highlight two main ones
here. First, although for all other main constructs we had multi-
item measures, we only measured contact with a single item tapping
cross-group friendships. Moreover, this item asked about cross-group
friends “among people from ethnic minority groups”, rather than
immigrants or Muslims. Although, as we noted earlier, choice of this
item has an advantage in terms of avoiding shared method variance
between predictor and outcome (Podsakoff et al., 2003), it is likely
that the lack of correspondence between measures of contact and
prejudice will result in an underestimate of the true association between
outgroup contact and outgroup prejudice. The second limitation refers to our
survey yielding only cross-sectional, correlational data. This, of
course, limits our ability to draw inferences about causality. In
other words, we have not been able to demonstrate that contact
promotes prejudice reduction, rather than vice versa. Although the vast
majority of studies on intergroup contact rely on cross-sectional
data ( Pettigrew & Tropp, 2006), a burgeoning number of studies
attest to longitudinal (as well as experimental) effects of contact,
but do also acknowledge that evidence exists for the reverse or
self-selection path, although it is typically weaker. Future
39
research should thus use longitudinal (panel) data to allow for
testing of bi-directional relationships, and greater confidence in
inferring causality.
To conclude, this research provides highly-original evidence
of the benefits of intergroup contact for more tolerant social
attitudes, even in sample of right-wing voters, whose chosen parties
espouse intolerant beliefs, including opposition to immigrants and
Islam, both of which were included as outcome variables in our
survey. We do not, of course, argue that positive intergroup contact
is a panacea in any case, let alone for holders of such intemperate
views, but the supportive evidence adds to an already impressive
evidence base showing that where prejudice is the problem,
intergroup contact should form part of the solution.
40
Figures
Cross-group
Friendships
Blatant Prejudi
ce Immigrants
Islamic Threat
Diversity
Beliefs
.20***
-.18*** / -.04
-.74***
-.56***
R2 = .64 / .61
R2 = .38 / .33
R2 = .17 / .09
Figure 1. Structural equation model showing the relationships between the variables under investigation. Regression coefficients (reported before the / for BNP participants and after the / UKIP respondents; reported in bold for constrained values) are standardized. The control variables have been left out of the figure for the sake of clarity (see Table 2). * p < .05, ** p < .01, *** p < .001 (UKIP denotes United Kingdom Independence Party; BNP denotes British National Party).
41
Cross-group
Friendships
Blatant Prejudice Immigrant
s
Islamic Threat
DiversityBeliefs
.25**-.72***
-.71***
R2 = .56
R2 = .49
R2 = .26
Figure 2. Structural equation model showing the hypothesised relationships between the variables under investigation for the EDL sample. Regression coefficients are unstandardized. The control variables have been left out of the figure for the sake of clarity (see Table 2). * p < .05, ** p < .01, *** p < .001 (EDL denotes English Defence League).
42
Tables
Table 1. Pearson product-moment correlations between composite variables, means, and standard deviations.
1 2 3 4
UKIPMeans(SD)
1. Ethnic minority friendships
- .20*** -.20*** -.15*** 2a
2. Diversity beliefs
.23*** - -.56*** -.36*** 3.80(2.11)
3. Immigrant attitudes
-.31*** -.52*** - .32*** 3.00(0.98)
4. Islam attitudes
-.14*** -.47*** .31*** - 4.37(0.80)
BNP Means (SD) 2a 2.77(2.38)
3.68(1.02)
4.48(0.81)
Correlation coefficients below the diagonal are for BNP (British National Party) participants whereas the correlation coefficients above the diagonal are for UKIP (United Kingdom Independence Party) respondents. SD = standard deviation. a = mode. * p < .05, ** p < .01, *** p < .001
43
Table 1 cont.
Pearson product-moment correlations between composite variables, means, and standard deviations.
1 2 3 4
1. Ethnic minority friendships
-
2. Diversity beliefs
.29*** -
3. Immigrant attitudes
-.26*** -.60*** -
4. Islam attitudes
-.15* -.53*** .40*** -
EDL Means (SD) 2a 5.40(2.45)
2.66(1.09)
3.66(1.01)
SD = standard deviation. a = mode. * p < .05, ** p < .01, *** p < .001. EDL denotes English Defence League.
44
Table 2.Comparisons of the Nested Multigroup Models.
Model Model Fit
Model
Compariso
n
Δ χ2
(df)
1a χ2 (16) = 54.855***, χ2/df = 3.43, SCF = 1.099, CFI = .96, RMSEA = .06[.05, .08], SRMR = .03
1b χ2 (16) = 63.703***, χ2/df = 3.98, SCF = 1.152, CFI = .98, RMSEA = .06[.05, .08], SRMR = .03
1c χ2 (16) = 28.545***, χ2/df = 1.78, SCF = 1.003, CFI = .98, RMSEA = .07[.02, .10], SRMR = .04
2.1
Unconstrained χ2 (30) = 67.993***, χ2/df = 2.27, SCF = 1.082, CFI = .98, RMSEA = .05 [.04, .07], SRMR = .03
Constrained χ2 (38) = 93.027***, χ2/df = 2.45, SCF = 1.097, CFI = .97, RMSEA = .06 [.05, .08], SRMR = .08
24.70**(8)
2.2Unconstrained χ2 (30) = 68.770***, χ2/df = 2.29, SCF = 1.095, CFI = .99, RMSEA = .04 [.03, .05], SRMR = .02
Constrained χ2 (38) = 95.227***, χ2/df = 2.51, SCF = 1.088, CFI = .98, RMSEA = .05 [.05 .06], SRMR = .06
26.66***(8)
45
2.3
Unconstrained χ2 (30) = 84.079***, χ2/df = 2.80, SCF = 1.184, CFI = .98, RMSEA = .04 [.03, .05], SRMR = .02
Constrained χ2 (38) = 92.147***, χ2/df = 2.43, SCF = 1.209, CFI = .98, RMSEA = .03 [.03 .05], SRMR = .03 9.10 (8)
SCF = Scale Correction Factor; this was used in comparing the chi-square values of the different models.Model Key: Model 1a = configural invariance for UKIP, Model 1b = configural invariance for BNP, Model 1c= configural invariance for EDL, Model 2.1 = Metric invariance for BNP and EDL samples, Model 2.2 =metric invariance for UKIP and EDL samples, Model 2.3 = metric invariance for BNP and UKIP samples. * p< .05, ** p < .01, *** p < .001; (UKIP denotes United Kingdom Independence Party; BNP denotes BritishNational Party; EDL denotes English Defence League).
Table 3. Metric, scalar, uniqueness, and structural fit indices for the different levels of invariance between the BNO and UKIP samples.
Model Number Fit StatisticsModel
Comparison
Change inchi-square
(df)2.3 (Metric
Invariance)
χ2 (38) = 92.147***, χ2/df = 2.43, SCF = 1.209, CFI = .98, RMSEA= .03
46
[.03.05],SRMR= .03
3
χ2 (43) = 105.468***, χ2/df = 2.45, SCF = 1.184, CFI = .98, RMSEA= .04[.03, .05],SRMR= .03
3 vs.
2.313.55*(5)
3.1
χ2 (42) = 99.074***, χ2/df = 2.36, SCF = 1.190, CFI = .98, RMSEA= .04[.03, .05],SRMR= .03
3.1 vs.
2.36.43(4)
4 χ2 (50) = 98.403***, χ2/df = 1.97, SCF = 1.330, CFI = .98, RMSEA= .03[.02, .04],
4 vs.
3.1
6.29(8)
47
SRMR= .04
5
χ2 (67) = 120.914***, χ2/df = 1.80, SCF = 1.190, CFI = .98, RMSEA= .03[.02, .04],SRMR= .03
5 vs 4 16.71(17)
6
χ2 (75) = 145.748***, χ2/df = 1.94, SCF = 1.172, CFI = .97, RMSEA= .03[.03, .04],SRMR= .03
6 vs. 5 26.37***(8)
6.1
χ2 (73) = 126.500***, χ2/df = 1.73, SCF = 1.176, CFI = .98, RMSEA= .03[.02, .04],SRMR= .03
6.1 vs.
54.78(6)
48
7
χ2 (75) = 133.320***, χ2/df = 1.78, SCF = 1.171, CFI = .98, RMSEA= .03[.02, .04],SRMR= .03
7 vs.
6.17.44*(2)
7.1
χ2 (74) = 126.891***, χ2/df = 1.71, SCF = 1.174, CFI = .98, RMSEA= .03[.02, .04],SRMR= .03
7.1 vs.
6.10.20(1)
8
χ2 (100) = 176.993***, χ2/df = 1.77, SCF = 1.212, CFI = .98, RMSEA =.03[.02, .04],SRMR= .04
9 χ2 (112) = 229.111***, χ2/df = 2.05, SCF = 1.186, CFI = .97, RMSEA =.03[.03, .0
9 vs. 8 59.02***(12)
49
4],SRMR= .04
9.1
χ2 (110) = 190.641***, χ2/df = 1.73, SCF = 1.189, CFI = .98, RMSEA =.03[.02, .03],SRMR= .04
9.1 vs 8 12.68(10)
10
χ2 (115) = 203.801***, χ2/df = 1.77, SCF = 1.184, CFI = .98, RMSEA =.03[.02, .04],SRMR= .04
10 vs.
9.113.62*(5)
10.1
χ2 (114) = 195.006***, χ2/df = .171, SCF = 1.186, CFI = .98, RMSEA =.03[.02, .03],SRMR= .03
10.1 vs.
9.14.17(4)
50
Model key: Model 3 = scalar invariance, Model 3.1 = partial scalar invariance, Model 4 = uniqueness invariance, Model 5 = freely estimated structural model without diversity beliefs, Model 6 = group-invariant structural model for control variables (without diversity beliefs), Model 6.1 = partial group invariance for the structural model for control variables (without diversity beliefs), Model 7 = group-invariant structural equation model for independent- and control variables (without diversity beliefs), Model 7.1 = partial structural invariance between samples for independent- and control variables (withoutdiversity beliefs), Model 8 = freely estimated structural model with diversity beliefs included, Model 9 = group-invariant structural equation model for control variables with diversity beliefs included, Model 9.1 = partial group invariance for the structural model for control variables with diversity beliefs included, Model 10.1 = structural invariance between samples for independent-, mediator-, and control variables, Model 10.1 = partial structural invariance between samples for independent-, mediator-, and control variables
51
Table 4.Factor loadings from the unconstrained configural invariance model (Model 2) by sample group.
Item
Construct and Sample
Blatant Prejudicetowards Immigrants Islamic Threat Diversity Beliefs
EDL UKIP BNP EDL UKIP BNP EDL UKIP BNP
Immigrants should be sent back to their home country whether or not they break the law
.81 .73 .71 -- -- -- -- -- --
Family members of immigrantswho were born in the UK should be sent back to theircountry of origin
.71 .66 .66 -- -- -- -- -- --
Immigrants are the main cause of crime in society .86 .65 .74 -- -- -- -- -- --
Immigrants are the main source of disease in society .76 .62 .66 -- -- -- -- -- --
Islam does not pose a serious danger to Western civilization
-- -- -- .52 .62 .71 -- -- --
There will be a ‘clash of -- -- -- .92 .82 .75 -- -- --
52
civilizations’ between Islamand the West
Britain’s cultural life is generally undermined or enriched by migrants
-- -- -- -- -- -- .86 .75 .74
Britain’s economy that migrants come to Britain from other countries.
-- -- -- -- -- -- .84 .78 .78
All factor loadings are significant at the p < .001 level; (UKIP denotes United Kingdom IndependenceParty; BNP denotes British National Party; EDL denotes English Defence League)..
53
Table 5. Unstandardized regression coefficients for the relationship between the controlvariables and the mediator and outcome variables.
BNP and UKIP Samples
Control
Variable
DiversityBeliefs
b (SE)
BlatantPrejudicetowards
Immigrants
b (SE)
Islamic Threat
b (SE)
Age
.004 / -.02***
(.003) /
(.004)
.02*** /
-.02***
(.004) /
(.005)
.01***
(.003)
Sex-.13*
(.06)
.10
(.08)
-.09
(.07)
Social Grade-.06*
(.03)
.06†
(.03)
-.06†
(.03)
Education.12***
(.02)
-.05
(.03)
.01
(.02)
EDL Sample
Control
Variable
DiversityBeliefs
b (SE)
BlatantPrejudicetowards
Immigrants
b (SE)
Islamic Threat
b (SE)
Age-.004
(.006)
-.02**
(.007)
.01
(.008)
Sex.22
(.20)
.07
(.20)
.03
(.20)
54
Social Grade-.18*
(.08)
-.11
(.08)
.02
(.09)
Education.22**
(.07)
-.09
(.08)
.03
(.08)
Regression coefficients separated by a / could not beconstrained to equality. Those that appear before the / arefor the UKIP sample and those after the / are for the BNPsample. † p < .06, * p < .05, ** p < .01, *** p < .001; (UKIPdenotes United Kingdom Independence Party; BNP denotes BritishNational Party; EDL denotes English Defence League).
References
Allport, G.W. (1954). The nature of prejudice. Reading, MA: Addison-
Wesley.
Amir, Y. (1969). Contact hypothesis in ethnic relations.
Psychological Bulletin, 71, 319–342. doi:10.1037/h0027352
Asbrock, F., Gutenbrunner, L., & Wagner, U. (2013). Unwilling,
but not unaffected-Imagined contact effects for authoritarians
and social dominators. European Journal of Social Psychology, 43, 404-
412. doi: 10.1002/ejsp.1956
Betz, H.G. & C. Johnson (2004). Against the current—stemming
the tide: the nostalgic ideology of the contemporary radical
populist right. Journal of Political Ideologies 9.3, pp.311-327
55
BNP (2010). Democracy, freedom, culture and identity: British National Party
General Election Manifesto. Welshpool, UK: British National Party.
Busher, J. (2013). Grassroots activism in the English Defence
League: Discourse and public (dis)order. In M. Taylor et al.
(Eds.) Extreme right-wing political violence and terrorism. London, UK:
Bloomsbury Academic/Continuum.
Carter, E. (2005). The extreme right in Western Europe: Success or failure?.
Manchester University Press
Davies, L., Tropp, L., Aron, A., Pettigrew, T. F., & Wright,
S. C. (2011). Cross-group friendships and intergroup
attitudes: A meta-analytic review. Personality and Social Psychology
Review, 15, 332-351. doi:10.1177/1088868311411103
Dhont, K., & Van Hiel, A. (2009). We must not be enemies:
Interracial contact and the reduction of prejudice among
authoritarians. Personality and Individual Differences, 46, 172-177. doi:
10.1016/j.paid.2008.09.022
Ford, R. and Goodwin, M.J. (2014). Revolt on the right: Explaining
support for the radical right in Britain. London, UK: Routledge.
Ford, R., Goodwin, M. J., & Cutts, D. (2012). Strategic
Eurosceptics and polite xenophobes: Support for the United
Kingdom Independence Party (UKIP) in the 2009 European
56
Parliament elections. European Journal of Political Research, 51, 204-234.
doi:10.1111/j.1475-6765.2011.01994.x
Fritz, M. S., & MacKinnon, D. P. (2007). Required sample size
to detect the mediated effect. Psycological Science, 18, 233-239.
Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias:
The common ingroup identity model. Philadelphia, PA: Psychology
Press.
Goodwin, M. J. (2013). The roots of extremism: The English Defence League
and the Counter-Jihad challenge. Chatham House.
Goodwin, M.J. (2011). New British Fascism: Rise of the British National Party,
Abingdon, UK: Routledge.
Halliday, F. (1999). Islamophobia reconsidered. Ethnic and Racial
Studies, 22,
892-902.
Hewstone, M., Cairns, E., Voci, A., Hamberger, J., & Niens, U.
(2006). Intergroup contact, forgiveness, and experience of
“The Troubles” in Northern Ireland. Journal of Social Issues, 62, 99–
120.
Hewstone, M., Lolliot, S., Swart, H., Myers, E., Voci, A., Al
Ramiah, A., & Cairns, E. (2014). Intergroup contact and
57
intergroup conflict. Peace and Conflict: Journal of Peace Psychology, 20,
39-53. doi:10.1037/a0035582.
Hewstone, M., & Schmid, K. (2014). Neighbourhood ethnic
diversity and orientations toward Muslims in Britain: the role
of intergroup contact. The Political Quarterly, 85, 320-325.
Hodson, G. (2008). Interracial prison contact: The pros for
(socially dominant) cons. British Journal of Social Psychology, 47, 325-
351.
Hodson, G. (2011). Do ideologically intolerant people benefit
from intergroup contact? Current Directions in Psychological Science, 20,
154-159. doi: 10.1177/0963721411409025
Hodson, G., Harry, H., & Mitchell, A. (2009). Independent
benefits of contact and friendship on attitudes towards
homosexuals among authoritarians and highly identified
heterosexuals. European Journal of Social Psychology, 39, 509-525. doi:
10.1002/ejsp.558
Hodson, G., Costello, K., & MacInnis, C. C. (2013). Is
intergroup contact beneficial among intolerant people?
Exploring individual differences in the benefits of contact on
attitudes. Advances in intergroup contact, 49-80.
58
Huntingdon, S. (1993) ‘The Clash of Civilizations’, Foreign Affairs
72(3): 22–49.
Islam, M. R., & Hewstone, M. (1993). Dimensions of contact as
predictors of intergroup anxiety, perceived out-group
variability, and out-group attitude: An integrative model.
Personality and Social Psychology Bulletin, 19, 700-710. doi:
10.1177/0146167293196005
Lolliot, S., Schmid, K., Hewstone, M., Al Ramiah, A., Tausch,
N., & Swart, H. (2013). Generalized effects of intergroup
contact: The secondary transfer effect. In G. Hodson, & M.
Hewstone, Advances in intergroup contact (pp. 81-112). London, UK:
Psychology Press.
Maoz, I. (2004). Coexistence is in the eye of the beholder:
Evaluating intergroup encounter interventions between Jews and
Arabs in Israel. Journal of Social Issues, 60, 437-452. doi:
10.1111/j.0022-4537.2004.00119.x
Mayer, N. (2013). From Jean-Marie to Marine Le Pen: Electoral
Change on the Far Right. Parliamentary Affairs, 66, 160-178.
Meredith, W. (1993). Measurement invarance, factor analysis
and factorial invariance. Psychomterika, 58, 525-543.
59
Mudde, C. (2007). Populist radical right parties in Europe (Vol. 22, No.
8). Cambridge: Cambridge University Press.
Muthén, L. K., & Muthén, B. O. (1998-2010). Mplus user's guide
(Sixth ed.). Los Angeles, CA: Muthén & Muthén.
Page-Gould, E., Mendoza-Denton, R., & Tropp, L. (2008). With a
little help from my cross-group friend: Reducing anxiety in
intergroup contexts through cross-group friendship. Journal of
Personality and Social Psychology, 95, 1080-1094. doi:10.1037/0022-
3514.95.5.1080
Pettigrew, T. F. (1958). Personality and sociocultural factors
in intergroup attitudes: A cross-national comparison. Journal of
Conflict Resolution, 2, 29-42.
Pettigrew, T. F. (1997). Generalized intergroup contact
effects on prejudice. Personality and Social Psychology Bulletin, 23, 173-
185.
Pettigrew, T.F. (2009). Secondary transfer effect of contact:
Do intergroup contact effects spread to noncontacted
outgroups? Social Psychology, 40, 55-65.
Pettigrew, T. F., & Tropp, L. R. (2008). How does intergroup
contact reduce prejudice? Meta-analytic tests of three
mediators. European Journal of Social Psychology, 28, 922-934.
60
Pettigrew, T., & Tropp, L. (2006). A meta-analytic test of
intergroup contact theory. Journal of Personality and Social Psychology,
90, 751-783. doi:10.1037/0022-3514.90.5.751
Pettigrew, T.F., & Tropp, L.R. (2011). When groups meet: The
dynamics of intergroup contact. Hove, East Sussex: Psychology Press.
Pettigrew, T.F., Christ, O., Wagner, U., & Stellmacher, J.
(2007). Direct and indirect intergroup contact effects on
prejudice: A normative interpretation. International Journal of
Intercultural Relations, 31, 411-425. doi:
10.1016/j.ijintrel.2006.11.003
Pettigrew, T.F., Tropp, L.R., Wagner, U., & Christ, O. (2011).
Recent advances in intergroup contact theory. International Journal
of Intercultural Relations, 35, 271-280.
Pew Forum on Religion and Public Life (2009). Muslims widely
seen as facing discrimination. Pew Research. Retrieved from
http://www.pewforum.org/2009/09/09/muslims-widely-seen-as-
facing-discrimination on 1. September 2013.
Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y., & Podsakoff, N.P.
(2003). Common method biases in behavioral research: A
critical review of the literature and recommended remedies.
Journal of Applied Psychology, 88, 879-903.
61
Rucker, D. D., Preacher, K. J., Tormala, Z. L., & Petty, R. E.
(2011). Mediation analysis in social psychology: Current
practices and new recommendations. Social and Personality Psychology
Compass, 5/6, 359-371. doi:10.1111/j.1751-9004.2011.00355.x
Rydgren, J. (2008). Immigration sceptics, xenophobes or
racists? Radical right‐wing voting in six West European
countries. European Journal of Political Research 47:6, 737-765.
Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-
square test statistic for moment structure analysis.
Psychometrika, 66, 507-514. doi:10.1007/BF02296192
Savelkoul, M., Scheepers, P., Tolsma, J., & Hagendoorn, L.
(2011). Anti-Muslim attitudes in the Netherlands: Tests of
contradictory hypotheses derived from ethnic competition
theory and intergroup contact theory. European Sociological Review,
27, 741-758.
Schmid, K., Hewstone, M., Küpper, B., Zick, A., & Wagner, U.
(2012). Secondary transfer effects of intergroup contact: A
cross-national comparison in Europe. Social Psychology Quarterly, 75,
28-51. doi:10.1177/0190272511430235
Sears, D.O. (1986). College sophomores in the laboratory:
Influences of a narrow data base on social psychology’s view
of human nature. Journal of Personality and Social Psychology, 51, 515-30.
62
Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup
theory of social hierarchy and oppression. Cambridge, UK: Cambridge
University Press.
Strabac, Z., & Listhaug, A. (2008). Anti-Muslim prejudice in
Europe: A multilevel analysis of survey data from 30
countries. Social Science Research, 37, 268-286.
Tropp, L. R., & Bianchi, R. A. (2006). Valuing diversity and
interest in intergroup contact. Jourmal of Social Issues, 62, 533-551.
Turner, R. N., Hewstone, M., & Voci, A. (2007). Reducing
explicit and implicit outgroup prejudice via direct and
extended contact: The mediating role of self-disclosure and
intergroup anxiety. Journal of Personality and Social Psychology, 93, 369-
388. doi:10.1037/0022-3514.93.3.369
Vandenberg, R. J., & Lance, C. E. (2000). A review and
synthesis of the measurement invariance literature:
Suggestions, practices, and recommendations for organizational
research. Organizational Research Methods, 3, 4-70.
doi:10.1177/109442810031002
Van de Schoot, R., Lugtig, P., & Hox, J. (2012). A checklist
for testing measurement invariance. European Journal of Developmental
Psychology, 9, 486-492. doi:10.1080/17405629.2012.686740
63
Van Laar, C., Levin, S., Sinclair, S., & Sidanius, J. (2005).
The effect of university roommate contact on ethnic attitudes
and behaviour. Journal of Experimental Social Psychology, 41, 329-345.
Vezzali, L., Hewstone, M., Capozza, D., Giovannini, D., &
Wölfer, R. (2014). Improving intergroup relations with
extended and vicarious forms of indirect contact. European Review
of Social Psychology, 25, 314-389.
Voci, A., & Hewstone, M. (2003). Intergroup contact and
prejudice towards immigrants in Italy: The mediational role of
anxiety and the moderational role of group salience. Group
Processes & Intergroup Relations, 6, 37–54.
Wright, S. C., Aron, A., McLaughlin-Volpe, T., & Ropp, S. A.
(1997). The extended contact effect: Knowledge of cross-group
friendships and prejudice. Journal of Personality and Social Psychology, 73,
73-90.
Ye-or, B. (2005) Eurabia: The Euro-Arab Axis, Madison, NJ: Fairleight
Dickinson University Press
Zúquete, J.P. (2008). The European extreme-right and Islam:
New directions? Journal of Political Ideologies 13, 321-344.
64
1 ‘The English Defence League: About Us’, available at http://englishdefenceleague.org/about-us (accessed 20 June 2012)2 Constraining all the factor loadings to equivalence for the EDL and UKIP resulted in a significantly drop in model fit, Δx2(8) = 26.66, p < .001but we could obtain partial metric invariance for the blatant prejudice and Islamic threat scales. The diversity beliefs scale showed signs of partial metric invariance. The item DB2 could not be constrained to equivalence, Δx2(1) = 14.09, p < .001. Item, DB3, however, showed marginally different factor loadings, Δx2(1) = 3.82, p = .051. Given this pattern of results is largely consistent with the comparisons between the BNP and EDL sample; the significant differences in the overall scale means for all three items under question; and the patterns of comparisons between the BNP and UKIP samples, we chose to keep these two samples separate. 3 We are not suggesting that diversity beliefs is the only mediator of contact effects for UKIP and EDL respondents. Other variables could explain the indirect of contact on both Islamic threat and blatant prejudice. While we agreewith many methodologists that a lack of a direct effect means there are no othermeaningful indirect effects, the remaining direct effect for BNP participants isinteresting. 4 There was nearly a statistically significant difference. For brevity, BNP variance explained in diversity beliefs: 17% (95% CI [.13, .21]); UKIP variance explained in diversity beliefs: 9% (95% CI [.05, .13]).