Intergroup Contact in Romania: When Minority [600510]

Intergroup Contact in Romania: When Minority
Size is Positively Related to Intergroup Conflict
VASILE CERNAT *
‘Petru Maior’ University of Ta ˆrgu-Mures ¸, Romania
ABSTRACT
Contexts in which minority size is positively related to intergroup conflict are challenging for the
contact hypothesis. In such situations, if opportunities for contact increase prejudice, the contacthypothesis may seem less credible, but if they reduce prejudice, the contact hypothesis may seem less
useful for improving intergroup relations. Based on path analyses run on a Romanian national
probability sample ( N¼733), the current research shows that the contact hypothesis can nevertheless
be relevant. Because the Hungarian minority is concentrated in Transylvania, a region with a longhistory of conflict between Romanians and Hungarians, Transylvanians have more opportunities for
out-group contact than other Romanians. However, the analyses also detected significant differences
within Transylvania: Urban Transylvanians have more opportunities for contact with Hungariansthan rural Transylvanians and, consequently, are less negative towards them. The results, which
closely match recent historical events, suggest that a proper application of the contact hypothesis at a
societal level has to take into account that minority size is not necessarily equivalent to opportunitiesfor contact and that inter-regional comparisons in opportunities for contact can hide significant intra-
regional differences. Copyright #2008 John Wiley & Sons, Ltd.
Key words: contact hypothesis; prejudice; ethno-political beliefs; minority; intergroup conflict
INTRODUCTION
People who are interested in the reduction of intergroup prejudice and conflict may end up
being confused and disconcerted by the existing literature. While some authors recommendincreased contact with out-group members as a solution (Pettigrew & Tropp, 2006), othersperceive contact as a cause of intergroup friction (Forbes, 2004). The present paper seeks to
make a contribution to this under-researched issue by testing the explanatory power of the
contact hypothesis in a challenging real world context in which minority size andintergroup tensions go hand in hand.Journal of Community & Applied Social Psychology
J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
Published online 19 December 2008 in Wiley InterScience
(www.interscience.wiley.com) DOI : 10.1002/casp.1001
* Correspondence to: Vasile Cernat, N. Iorga St. 1, Ta ˆrgu-Mures ¸, 540088, Romania.
E-mail: [anonimizat]
Copyright #2008 John Wiley & Sons, Ltd. Accepted 24 September 2008

THE CONTACT HYPOTHESIS
The contact hypothesis is considered one of the most important, simple and optimistic
social psychological approaches to prejudice reduction (Dovidio, Gaertner, & Kawakami,2003; Hewstone, Rubin, & Willis, 2002; Pettigrew, 1998; Rothbart, 1996). Initially,researchers thought that prejudice could be reduced if contact situations met a series of
essential conditions, like equal status or common goals (Allport, 1954). Subsequent
research has shown that contact can improve stereotypes and attitudes in various settings,and the current view is that Allport’s conditions are not necessary for prejudice reduction,but rather act as catalysts (Pettigrew & Tropp, 2006).
While successful, the contact approach has also been criticized theoretically for
individualism (Dixon, Durheim, & Tredoux, 2005) and empirically on the grounds that ithas small, slow, short-lived and societally insignificant effects (Chirot & McCauley, 2006;Forbes, 2004). The skepticism toward the application of this micro-level approach at a
macro societal level is understandable. After all, a plethora of factors shape intergroup
perceptions and evaluations and have the potential to dwarf the effects of positive contactwith out-group members: Competition over scarce or socially valued resources (Bobo,1999), socio-political ideologies (Billig, 2002; Sidanius, Levin, Federico, & Pratto, 2001),social representations of history (Rouhana & Bar-Tal, 1998), social events (Bar-Tal &Labin, 2001), socialization (Guimond, 2000) and so on. Single factor approaches fail toexplain the intricate and diverse nature of intergroup relations and are thus avoided,especially by the researchers who focus on real world cases (Mays, Bullock, Rosenzweig,
& Wessells, 1998).
Nevertheless, in order to cope with this complexity and build better models, we have to
know which factors are important and which not. Recent research, which focuses on largescale differences in opportunities for contact, suggests that contact can be relevant at theintergroup level. More specifically, Wagner and his colleagues found that differences inethnic prejudice between East and West Germany could be meaningfully explained bydifferences in opportunities for intergroup contact. The incidence of both attacks andprejudice against foreigners is greater in East Germany than in West Germany, whereas the
percentage of foreigners is five times higher in the Western regions of the country than in
the former Eastern German Democratic Republic. Analyses of several national probabilitysamples found support for a model in which the larger proportion of foreigners in the Westis related to more contact opportunities, which leads to more intergroup friendships, whichlessens prejudice (Wagner, Christ, Pettigrew, Stellmacher, & Wolf, 2006; Wagner, vanDick, Pettigrew, & Christ, 2003).
Whereas these studies focused on a country in which minority size is negatively related
to intergroup violence, there are frequent and more problematic cases in which minority
size is positively associated with intergroup conflict and violence. For example, the first
post-communist inter-ethnic clashes in Eastern Europe, which took place in Romania at thebeginning of 1990, involved Romanian and Hungarian ethnics from Transylvania, theregion where most Hungarian ethnics live (Brubaker, Feischmidt, Fox, & Grancea, 2006).Not only did the conflict take place in a region with a large minority population, but also thedeadliest fights were recorded in the most ethnically diverse county (i.e. Mures ¸, where
Hungarians make up 39% of the total population, and there is also a sizeable Romaminority of 7%, the largest in the country).
The positive association between minority size and intergroup conflict seems hard to
reconcile with the basic tenets of the contact hypothesis. In such contexts, this approach is
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/casp16 V. Cernat

caught in an apparently insurmountable paradox. On the one hand, opposite to what the
contact theorists maintain, intergroup contact could have negative consequences onintergroup relations. As argued by Horowitz (1999), conflicting groups ‘know each otherrather well, and they do not like what they have experienced of each other.’ On the otherhand, contact with out-group members may have positive but insignificant effects onintergroup relations. According to Forbes (2004), contact with out-group members can
improve attitudes at an individual level, but at the group level, existing cultural differences
between groups generate a conflict of interest that increases proportionally with the degreeof contact. More contact means more conflict, and the best conflict management solutionwould be to keep the groups apart as much as possible.
The main purpose of the present paper is to test whether this ‘contact does not work, or it
works but does not matter’ paradox can be solved. Its main solution is the hypothesis thatinter-regional differences in opportunities for contact can conceal significant intra-regionalvariability.
MODEL AND HYPOTHESES
The present research will investigate the explanatory power of the contact hypothesis in the
context of Romanian–Hungarian relationships. The model will test the main hypothesesderived from the contact approach, but it will also control for alternative hypotheses thatfocus on the limits of this theory or run against it.
Theoretical core
The backbone of the model advanced by the current paper is based on the main hypotheses
tested by Wagner and his colleagues in Germany. Generally speaking, negativity againstHungarians should be less prevalent where opportunities for contact are greater. Since,according to the last national census, Hungarians represent about 20% of the Transylvanianpopulation but less than .1% in the rest of the country, there should be less prejudice in
Transylvania than elsewhere in Romania. More specifically, we should expect that: (a) The
greater proportion of Hungarians in Transylvania should lead to more opportunities forcontact (i.e. more Hungarian neighbours, coworkers etc.) than in the rest of the country;(b) more opportunities should lead to more intimate contact (i.e. closer relationships withHungarians); and (c) intimate contact should be related to less negativity againstHungarians.
A crucial demographic idiosyncrasy
This theoretical core is qualified by an additional hypothesis. The assumption that a larger
minority size means more opportunities for contact may be misleading and could lead to animproper application of the contact hypothesis at a societal level. Given the complex natureof real world cases, we should take into account how particular geographic, demographicor residential factors shape opportunities for contact. In the present case, there are reasonsto believe that inter-regional comparisons could conceal significant intra-regionalvariability.
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/caspIntergroup contact in Romania 17

For historical reasons that will not be mentioned here, at the beginning of the last century
rural Transylvania was numerically dominated by Romanians, whereas urban centres werelargely Hungarian. After World War I, Transylvania became Romanian territory and manyRomanians began moving to urban areas. As a consequence, most towns are now ethnicallymixed (Brubaker et al., 2006). On the contrary, the rural population remains relativelyhomogenous in the sense that many villages tend to be ethnically monochromatic.
Therefore, although general statistics show that the proportion of Hungarians is similar in
urban and rural areas, they hide significant differences in ethnic composition variance. Forexample, data from the last national census show that Hungarians represent 38% of theurban population and 41% of the rural population in the Transylvanian county of Mures ¸. In
some areas Romanians represent the majority, whereas in other areas Hungarians representthe majority. However, the minority group is bigger than 10% in all urban localities butsmaller than 5% in 35% of the rural localities. Furthermore, this only happens if we focuson ‘comunes’, the smallest Romanian administrative units, which are made up of several
villages. If we do the math for these villages, we find that in 62% of them one of the two
ethnic groups represents less than 5% of the total population (National Institute ofStatistics, 2008).
Thus, urban Transylvanians have more opportunities for contact with Hungarians than
rural Transylvanians, and as a consequence, they should be less negative againstHungarians. This will be modelled by using as exogenous variables region (Transylvaniavs. rest of Romania), place of residence (urban vs. rural) and an interaction term. If, as aconsequence of increased opportunities for contact, urban Transylvanians are less negative
against Hungarians than the rest of Romanians, including rural Transylvanians, then the
interaction between region and place of residence should significantly predict thisnegativity via the contact indicators.
Differential effects of contact on evaluation and beliefs
Usually, contact has stronger effects on affective and evaluative processes than on cognitive
processes (Paolini, Hewstone, Voci, Harwood, & Cairns, 2006; Tropp & Pettigrew, 2005).This rule suggests that even if the qualified theoretical core would account well for theextent of Romanians’ negative evaluations of Hungarians, it would not be enough for agood argument in favour of the usefulness of the contact hypothesis at a societal level.Ethnic tensions in Transylvania have revolved around critical beliefs regarding the
autonomy of predominantly Hungarian regions, the use of Hungarian language in
education, justice, administration and so on (Brubaker et al., 2006). It may well be thatcontact with Hungarians improves Romanians’ evaluation of this out-group but leavesunaltered their beliefs about the appropriate ethno-political arrangements (Jackman &Crane, 1986). Therefore, in order to test for this potential limitation, the model will includemeasures of both out-group evaluation and ethno-political beliefs.
Negative consequences of opportunities for contact
Finally, from a critical perspective, it could be argued that opportunities for contact could
also have negative effects on out-group perception and evaluation. Simple contact with anout-group member can activate a self-fulfilling prophecy mechanism that may strengthenthe perception of out-group threatening traits (Bargh, Chen, & Burrows, 1996; Chen &
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/casp18 V. Cernat

Bargh, 1997), generate anxiety (Blascovich, Mendes, Hunter, Lickel, & Kowai-Bell,
2001), and lead to avoidance of further contact (Plant, 2004). It takes more and especiallybetter contact with out-group members in order to feel less threatened by them (Aberson &Haag, 2007; Blascovich et al., 2001, Experiment 3).
Consequently, although opportunities for contact may usually lessen intergroup anxiety
by stimulating intergroup friendships (Paolini et al., 2006; Stephan et al., 2002; Tropp &
Pettigrew, 2005), we should not rule out the possibility that they could also cause threat in
some situations and thus increase prejudice. In the present case, it could be argued thatthreatening ideological and historical beliefs could undermine the positive effects ofopportunities for contact. The Romanian–Hungarian conflict has centred on the disputeover the legitimate ‘owner’ of Transylvania, a territory that last century alone changedsides several times, and led to the development of opposite political–historical theories(Boia, 2001). Consistent with the theories developed by Romanian elites, many Romaniansbelieve that Hungarians conquered Transylvania centuries ago and oppressed their
peaceful ancestors, and now, after Transylvania has been reunited with the rest of Romania,
are conspiring to bring it back to Hungary (Mungiu-Pippidi, 1999). When a Romanianinteracts with a Hungarian, such beliefs could be activated and generate threat. Therefore,the model should include a measure of threat in order to test whether opportunities forcontact are positively related to threat. This would also offer the opportunity to test whetherthe perception of out-group threat mediates the effect of intimate contact on out-groupevaluation and ethno-political beliefs.
METHOD
Sample
The present study analyses data from an ethnic relations barometer conducted by two
survey agencies for the Ethnocultural Diversity Resource Centre in 2002. Metro MediaTransylvania was responsible for the field research, and The Gallup Organization Romaniacontrolled the data. The survey collected information from 733 Romanian ethnics based onface to face interviews. The three-stage stratified probabilistic sample was representative ofthe adult population of Romania. Three hundred eighty six of the respondents were femalesand 347 males, 348 were from rural areas and 384 from urban areas, 207 were
Transylvanians and 526 non-Transylvanians. Participants’ ages ranged between 20 and
91 years, with a median of 49 years. Missing data ranged from 0 to 15% for individualvariables, with an overall rate of 6.6% ( Ethnocultural Diversity Resource Centre , 2002).
Because a listwise approach would have led to a final sample of 480 individuals, data
imputation was used instead (expectation–maximization). In support of this decision, itshould be stated that the average difference between the means obtained by listwisedeletion and EM imputation was small (it equalled 0.01 and ranged from 0.001 to 0.06),and Little’s missing completely at random (MCAR) test was non-significant
(x
2¼1920.115, degrees of freedom (DF) ¼1925, p¼0.53).
Measures
Contact : Three items measured respondents’ contact with Hungarians in the neighbour-
hood and at the workplace (i.e. ‘I have Hungarian coworkers’, ‘There are Hungarians Iknow’, ‘I sometimes shop at Hungarian-run stores’; a¼0.81, M¼0.93, SD ¼1.16).
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/caspIntergroup contact in Romania 19

Intimate contact was also measured by three items (i.e. ‘I pay visits to Hungarians’, ‘When
it comes to personal problems, I often talk with a Hungarian’, ‘Sometimes I ask for helpfrom a Hungarian’; a¼0.86, M¼0.46, SD ¼0.94). Each item was coded 0 if the answer
was ‘no’ or 1 if the answer was ‘yes,’ and the answers were then summed to form the scales.
Evaluation : Because no item asked the respondents about their feelings toward
Hungarians, an alternative measure was used instead, which is similar to the evaluative
indices built from trait or attribute ratings (e.g. Brown, Croizet, Bohner, Fournet, & Payne,
2003; Gaertner, Mann, Murrell, & Dovidio, 1989; Gonzales & Brown, 2003; Klauer &Meiser, 2000). Whereas these indices are usually based on attribute ratings, the presentindex was based on attribute selection. More specifically, respondents had to choose threeattributes they thought best characterize Hungarian ethnics out of a list of 24 items (e.g.kind hearted, egoistic, hostile, intelligent). Respondents’ choices were coded as/C01 (positive attributes) or ț1 (negative attributes). Evaluatively ambiguous items (e.g.
religious, united) were discarded from the analyses. The resulting values were added up to
form an evaluative index that could range between /C03 andț3, with higher scores meaning
more negativity against Hungarians. This measure had a mean of /C00.46 and a SD of 1.65.
Ethno-political beliefs : Ethno-political beliefs were measured by averaging responses to
four items that focused on participants’ opinions on several sensitive issues regarding theHungarian minority (i.e. ‘The Romanian state should support the cultural organizations ofHungarians from Romania’, ‘The Hungarian state should grant some rights to Hungariansfrom Romania’, ‘Hungarians should be able to have double-citizenship—Romanian andHungarian’, ‘The Romanian state should offer Hungarian language education to Hungarian
children’; a¼0.76, M¼2.72, SD ¼0.76). Responses ranged from 1—completely agree to
4—completely disagree. Higher scores reflect ethno-political conservativeness, whereaslower scores reflect a more liberal attitude toward the rights that should be granted to theHungarian minority.
Threat : According to the revised intergroup threat theory, there are four basic types of
threat: realistic group threats, symbolic group threats, realistic individual threats andsymbolic individual threats (Stephan & Renfro, 2002). The ethnic relations barometer didnot make a distinction between realistic and symbolic threats, focusing exclusively on the
individual-collective distinction. Romanian respondents were asked whether they feel that
Hungarians could be a threat to their national group/country/family. The three items had afour point answer scale ranging from 1—to a great extent to 4—not at all. The item thatmeasured threat at an individual level was not significantly related to the variables includedin the model and, consequently, was discarded from further analyses. The remaining twoitems, which measured collective threat, were reverse coded so that higher scores wouldmean more perceived threat ( a¼0.96, M¼3.05, SD ¼0.92).
Region and place of residence were dummy coded, with Transylvania and urban
residence coded as 1 and rest of Romania and rural residence coded as 0.
RESULTS
Preliminary analyses
Data were tested for normality. The degree of univariate skewness varied between /C00.70
and 1.91, univariate kurtosis ranged between /C01.99 and 2.14., and Mardia’s multivariate
kurtosis was significant (6.47). Although these values do not exceed warning limits (West,
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/casp20 V. Cernat

Finch, & Curran, 1995), they are high enough to raise suspicion. In order to address this
potential problem, the model was run using both maximum likelihood (ML) and robustML, which is less distributionally dependent. Fit indices and parameter estimates weresimilar across the estimation methods, thus eliminating the suspicion of practical non-normality.
The relationships between the variables included in the model and a series of socio-
demographic factors were analysed to determine potential confounds. Compared to the rest
of Romanians, Transylvanians did not differ with respect to education (i.e. years ofeducation), income (i.e. monthly income), age and national identification (i.e the averagedresponses to six items that asked respondents to what extent being a Romanian makes themfeel proud, ashamed, etc.; a¼0.75). Place of residence had significant effects on education
(b¼0.44, p<0.001), income ( b¼0.34, p<0.001), age ( b¼/C00.21, p<0.001) and
national identification ( b¼0.18, p<0.001). That is, urban Romanians were more
educated, earned more, were younger, and identified less with their national group than
rural Romanians. After controlling for these variables, most of the relationships posited by
the model remained unaltered. The only exception was that the effect of place of residenceon opportunity for contact became non-significant after controlling for education.
The Wald and Lagrange Multiplier tests were used to exclude non-significant paths and
to include paths that may contribute significantly to the model. The final model fit the datawell: x
2¼13.07, DF ¼12,p¼0.36, comparative fit index (CFI) ¼0.99, root mean square
error of approximation (RMSEA) ¼0.01, with a 90% CI for RMSEA of 0.00, 0.04 (see
Figure 1).
Regional differences in prejudice
As predicted by the contact hypothesis, the zero-order correlations (see Table 1) indicated a
significant negative relationship between region and out-group evaluation (r ¼/C00.10) and
a non-significant relationship between region and ethno-political beliefs (r ¼/C00.02).
However, intimate contact tended to be more strongly associated with ethno-political
beliefs than with out-group evaluation ( r¼/C00.17 and /C00.08, respectively, z¼/C01.75, the
Figure 1. Path analysis results for the general sample. All paths are significant ( p<0.05). Values in
parentheses represent the R2. Model fit: x2¼13.07, DF ¼12,p¼0.36, CFI ¼0.99, RMSEA ¼0.01,
the 90% CI for RMSEA ¼0.00–0.04.
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/caspIntergroup contact in Romania 21

Table 1. Correlation matrix of exogenous and endogenous variables. Values in parentheses represent the covariance
Variables Region Residence Interaction Opportunity Friendship Threat Evaluation
Residence 0.00 (0.00)
Interaction 0.68 (0.11) 0.39 (0.07)
Opportunity 0.62 (0.32) 0.22 (0.13) 0.56 (0.23)Friendship 0.51 (0.21 /C00.08 (/C00.03) 0.57 (0.19) 0.66 (0.71)
Threat /C00.08 (/C00.03) /C00.15 (/C00.07) /C00.13 (/C00.04) /C00.13 (/C00.13) /C00.18 (/C00.15)
Evaluation /C00.10 (/C00.08) 0.08 (0.06) /C00.09 (/C00.05) /C00.07 (/C00.13) /C00.08 (/C00.13) 0.29 (0.43)
Beliefs /C00.02 (/C00.01) /C00.08 (/C00.03) /C00.07 (/C00.02) /C00.09 (/C00.08) /C00.17 (/C00.12) 0.29 (0.20) 0.27 (0.34)
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/casp22 V. Cernat

two-tailed p¼0.08). This suggests that the zero-order correlations between region and out-
group evaluation are misleading. Indeed, effect decomposition revealed that the total effectof region on evaluation was only /C00.01, which means that most of the relationship between
region and evaluation is spurious. On the other hand, region had a significant indirect effecton ethno-political beliefs ( /C00.06) that was cancelled out by a direct effect (0.09).
Hypothesized relationships
As predicted by the main hypotheses, region was linked to opportunities for contact,
opportunities for contact related to intimate contact, intimate contact was negativelyassociated with perceived out-group threat, and increased out-group threat was associatedwith more negative out-group evaluations and more conservative ethno-political beliefs. Theanalyses failed to detect any negative effects of opportunities for contact on out-group threator out-group perception, and, contrary to the general findings in the literature, intimate con-tact tended to have a stronger effect on ethno-political beliefs than on out-group evaluation.
The interaction between region (Transylvania/rest of Romania) and place of residence
(urban/rural) was significant. The interaction term (which contrasted urban Transylvania to
the remaining levels) was significantly related to opportunities for contact and intimatecontact. It had significant indirect effects on both out-group evaluation ( /C00.02) and ethno-
political beliefs ( /C00.07) and a significant direct effect on out-group evaluation ( /C00.12).
To obtain further information on this interaction, separate analyses were conducted on
four subsamples based on respondents’ region or place of residence. Specifications of themodels were similar to those depicted in Figure 1, except that when region was theexogenous variable, the models also included a direct path from region to intimate contact.
Results are illustrated in Figures 2 and 3.
In the Transylvanian and urban subsamples, the results are consistent with the
hypothesized relationships. However, in the non-Transylvanian and rural subsamples,the effects of intimate contact on perceived out-group threat and ethno-political beliefswere non-significant. This is probably because of low variability of the contact indicators inthe subsamples that do not include the urban Transylvanians. For example, for intimatecontact in rural areas outside Transylvania, M¼0.08, SD ¼0.37; in urban areas outside
Transylvania, M¼0.21, SD ¼0.61; in rural Transylvania, M¼0.69, SD ¼1.03; whereas
in urban Transylvania, M¼1.69, SD ¼1.24. The adjusted comparisons between these
means were significant (all ps<0.001), except for the difference between urban and rural
non-Transylvanians. Furthermore, the correlation between intimate contact and perceivedout-group threat was significant only among urban Transylvanians, whereas the correlationbetween intimate contact and ethno-political beliefs was significant among both urbanTransylvanians and urban non-Transylvanians.
The results also shed light on the direct relationships between place of residence and out-
group evaluation and between region and ethno-political beliefs, which were found in the
general sample. In the first case, the effect was identified in the subsamples that contrasted
urban non-Transylvanians to urban Transylvanians and rural non-Transylvanians, whereasthe second effect was found in the subsample that contrasted rural Transylvanians to ruralnon-Transylvanians and almost reached significance in the subsample that contrasted ruralTransylvanians to urban Transylvanians. Further analyses showed a significant direct effectfrom the exogenous variable to out-group evaluation in a subsample that contrasted urbannon-Transylvanians to rural Transylvanians and a direct effect on ethno-political beliefs ina subsample that contrasted rural Transylvanians to urban non-Transylvanians. Thus, the
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/caspIntergroup contact in Romania 23

Figure 2. Path analysis results for the Transylvanian ( x2¼4.66, DF ¼4,p¼0.32, CFI ¼0.99,
RMSEA ¼0.03, CI for RMSEA ¼0.00–0.11) and non-Transylvanian ( x2¼9.42, DF ¼4,p¼0.05,
CFI¼0.98, RMSEA ¼0.05, CI for RMSEA ¼0.00–0.09) subsamples. Values in parentheses
represent the parameter estimates for the non-Transylvanian subsample.
An asterix (*) indicates a significant path ( p<0.05).
Figure 3. Path analysis results for the urban ( x2¼4.14, DF ¼5,p¼0.53, CFI ¼1.00,
RMSEA ¼0.00, CI for RMSEA ¼0.00–0.06) and rural ( x2¼6.11, DF ¼5,p¼0.30, CFI ¼1.00,
RMSEA ¼0.03, CI for RMSEA ¼0.00–0.08) subsamples. Values in parentheses represent the
parameter estimates for the rural subsample.
An asterix (*) indicates a significant path ( p<0.05).
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/casp24 V. Cernat

relationship between place of residence and evaluation and between region and ethno-
political beliefs is explained by the effects of urban non-Transylvania and ruralTransylvania on the respective measures.
Alternative models
Tests were run on several alternative models in which the paths from intimate contact to
out-group evaluation and ethno-political beliefs were reversed for one or both dependent
variables and the paths were direct, indirect (via threat) or both direct and indirect.
Akaike’s information criterion (AIC) suggested that these models are a poor alternative tothe main model: on average, their AIC was bigger by 36.99 than the AIC of the main model(the difference ranged between 7.22 and 102.77). The models that approached the twocontact indicators or the two dependent variables as manifestations of one underlyingfactor also proved to be worse alternatives ( DAIC¼30.27 and 15.49, respectively). Finally,
two models allowed direct and indirect links (via threat) between out-group evaluation andethno-political beliefs. Although the AIC of the main model was again smaller than the
AIC of the evaluation to beliefs and beliefs to evaluation models, these differences were
negligible ( DAIC¼1 and 0.91, respectively).
DISCUSSIONPrevious research on the effects of contact on intergroup relations is divided between those
who maintain that contact leads to more conflict (Forbes, 2004) and those who argue thatcontact improves intergroup relations (Wagner et al., 2003). The first line of research
suggests that minority size is positively linked to intergroup tension, whereas the second
line of research predicts the opposite. The present study addressed this contradiction byfocusing on the case of Romanian–Hungarian relationships, where the region with biggerminority size is also the region with stronger inter-ethnic frictions.
Path analyses run on a national probability sample found that Romanians living in
Transylvania had more Hungarian neighbours or coworkers than Romanians living in otherareas of the country; in addition, they reported more opportunities for contact related tocloser relationships with Hungarians, which were negatively associated with perceived
out-group threat. Decreased out-group threat related to less negative out-group evaluations
and less conservative ethno-political beliefs. Crucially, there were also significant intra-regional differences: Romanians from urban Transylvania had more opportunities forcontact and intimate contact than other Romanians, including rural Transylvanians, andthey were also less threatened by Hungarians and less biased against them.
These findings show that the contact approach can be useful for understanding
intergroup relations even where general minority size goes hand in hand with intergrouptensions. In order to see this, we have to accept that inter-regional differences can conceal
significant intra-regional variability. Thus, one of the keys to a proper application of the
contact hypothesis at a societal level is to take into account how geographic, demographic,residential and other similar factors shape the nature and amount of opportunities forcontact.
In the present case, most of the Hungarian minority is concentrated in Transylvania, a
region with a long history of intergroup conflict between Romanians and Hungarians.However, as a rule, Transylvanian towns are ethnically mixed, whereas Transylvanianvillages tend to be ethnically homogenous. Thus, urban Transylvanians have more
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/caspIntergroup contact in Romania 25

opportunities for contact with Hungarians than rural Transylvanians and, consequently, are
less biased against this ethnic out-group. In this respect, it is interesting to mention that,although the most violent clashes in 1990 took place in a town with nearly equal numbersof Romanians and Hungarians (i.e. Ta ˆrgu Mures ¸, with 50.34% Romanian ethnics and
46.73% Hungarian ethnics), violence was ignited by peasants from nearby ethnicallyhomogeneous villages (i.e. Hodac and Iba ˘nes¸ti, with negligent Hungarian minority
populations of 0.002 and 0.003%, respectively) who were misled by local elites to come to
the town and defend it against Hungarians (Helsinki Watch, 1990). It would not seem farfetched to think that a significant difference in negativity against Hungarians betweenurban and rural Transylvanians could be one of the factors that made this possible.
Obviously, contact is just one of the factors that shape inter-ethnic relations, and the
present study shows this quite clearly. For example, independent of the indicators ofcontact, rural Transylvanians expressed more conservative ethno-political beliefs thanother Romanians. Because this direct effect remained significant after controlling for the
possible confounds measured by the survey, we can only speculate about its nature.
The relationship between rural Transylvania and ethno-political beliefs could be
explained by historical beliefs. To make a long story short, Hungarian troops committedatrocities against Transylvanian peasants during World War II, the communist regimeheavily exploited these events for political purposes, and their historical recollection is stillalive in present day Transylvania (Boia, 2001; Mungiu-Pippidi, 1999). It is possible thatsuch historical beliefs are more prevalent in Transylvania than elsewhere in Romania andstronger in rural Transylvania than in urban Transylvania. Given the significant difference
in national identification between rural and urban residents, one may wonder whether
national identification mediates this relationship. That is, rural Transylvanians wouldexpress more conservative beliefs because their national identification is stronger, andthose who identify more strongly with their national group should have more negativehistorical beliefs about Hungarians. Nevertheless, this very plausible hypothesis was notsupported by the current data.
We should also notice that compared with urban Transylvanians, rural Transylvanians
may interact differently with Hungarians not only quantitatively but also qualitatively. In
Romania, rural areas are dominated by small-scale farming. The typical farmer has less
than 1.7 hectares and often works the land individually with rudimentary tools. Even ifRomanian farmers have Hungarian neighbours, they probably spend less time interactingwith them than urban Romanians, who not only have Hungarian neighbours but probablyalso have Hungarian coworkers with whom they may team up to solve common problemson a daily basis. That is, urban Romanians not only know more Hungarians than ruralRomanians, but they probably know them in better circumstances. Consistent with thisinterpretation, when ethno-political beliefs were regressed simultaneously on the three
items that formed the simple contact scale, only the item asking respondents to indicate
whether they have Hungarian coworkers had a significant effect on this measure. Moreover,the negative link was mediated by intimate contact. This implies that future research shouldtake into account that not all opportunities for contact have similar consequences onintergroup perceptions.
Limitations
The present paper is based on secondary data analyses. While the sample is of high quality
compared with the usual convenience student samples employed by social psychologists,
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/casp26 V. Cernat

its measures are below the discipline’s best practice standards. For example, the measure of
out-group threat is not as sophisticated and comprehensive as the scales developed byresearchers working on the integrated threat theory (e.g. Stephan et al., 2002). Similarly,the evaluative index, although based on similar approaches based on trait ratings, is notexactly an ideal measure of out-group evaluation as long as the attributes’ favourability isjudged by the researchers rather than the respondents themselves.
Although a path analysis showed that the proposed model fit the data better than
alternative models, the results are correlational, and it would not be wise to draw strongconclusions about the flow of causality within the model. Further studies in this settingshould use longitudinal designs and, as suggested by the current analyses, they should alsomeasure variables pertaining to national ideology and shared historical memories.
CONCLUSIONS
Although cases in which greater minority size is positively associated with intergroup
tensions are usually regarded as an argument against the contact hypothesis, the currentfindings show that intergroup contact can explain intergroup evaluations and perceptionseven in such challenging contexts. The results suggest that a proper application of thecontact hypothesis at a societal level should take into account that: (a) larger minority sizecan mean merely out-group proximity, not necessarily more opportunities for contact;(b) inter-regional comparisons can hide significant intra-regional differences; and(c) various opportunities for contact may have different effects on intergroup perception
and evaluation.
REFERENCES
Aberson, C. L., & Haag, S. C. (2007). Contact, perspective taking, and anxiety as predictors of
stereotype endorsement, explicit attitudes, and implicit attitudes. Group Processes and Intergroup
Relations ,10, 179–201.
Allport, G. W. (1954). The nature of prejudice . Reading, MA: Addison-Wesley.
Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait
construct and stereotype activation on action. Journal of Personality and Social Psychology ,71,
230–244.
Bar-Tal, D., & Labin, D. (2001). The effect of a major event on stereotyping: Terrorist attacks in Israel
and Israeli adolescents’ perceptions of Palestinians, Jordanians and Arabs. European Journal of
Social Psychology ,31, 265–280.
Billig, M. (2002). Henri Tajfel’s ‘cognitive aspects of prejudice’ and the psychology of bigotry.
British Journal of Social Psychology ,41, 171–188.
Blascovich, J., Mendes, W. B., Hunter, S. B., Lickel, B., & Kowai-Bell, N. (2001). Perceiver threat in
social interactions with stigmatized others. Journal of Personality and Social Psychology ,80, 253–
267.
Bobo, L. D. (1999). Prejudice as group position: Microfoundations of a sociological approach to
racism and racial relations. Journal of Social Issues ,55, 445–472.
Boia, L. (2001). History and myth in the Romanian consciousness . Budapest: Central European
University Press.
Brown, R. J., Croizet, J.-C., Bohner, G., Fournet, M., & Payne, A. (2003). Automatic category
activation and social behavior: The moderating role of prejudiced beliefs. Social Cognition ,21,
167–193.
Brubaker, R., Feischmidt, M., Fox, J., & Grancea, L. (2006). Nationalist politics and everyday
ethnicity in a Transylvanian town . Princeton: Princeton University Press.
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/caspIntergroup contact in Romania 27

Chen, M., & Bargh, J. A. (1997). Nonconscious behavioral confirmation processes: The self-fulfilling
consequences of automatic stereotype activation. Journal of Experimental Social Psychology ,33,
541–560.
Chirot, D., & McCauley, C. (2006). Why not kill them all? The logic and prevention of mass political
murder . Princeton: Princeton University Press.
Dixon, J., Durrheim, K., & Tredoux, C. (2005). Beyond the optimal contact strategy. A reality check
for the contact hypothesis. American Psychologist ,60, 697–711.
Dovidio, J. F., Gaertner, S. L., & Kawakami, K. (2003). Intergroup contact: The past, the present, and
the future. Group Processes and Intergroup Relations ,6, 5–21.
Ethnocultural Diversity Resource Centre . (2002). Barometer on ethnic relations, the 2002 edition.
Retreived from World Wide Web on 24 May 2008 from: http://www.edrc.ro//docs/docs/bare/
Bare_2002_spss.zip
Forbes, H. D. (2004). Ethnic conflict and the contact hypothesis. In Y. T. Lee, C. McCauley, F.
Mogghaddam, & S. Worchel (Eds.), The Psychology of ethnic and cultural conflict (pp. 69–88).
Westport, CN: Praeger.
Gaertner, S. L., Mann, J., Murrell, A., & Dovidio, J. F. (1989). Reducing intergroup bias: The benefits
of recategorization. Journal of Personality and Social Psychology ,57, 239–249.
Gonza ´les, R., & Brown, R. (2003). Generalization of positive attitude as a function of subgroup and
superordinate group identifications in intergroup contact. European Journal of Social Psychology ,
33, 195–214.
Guimond, S. (2000). Group socialization and prejudice: The social transmission of intergroup
attitudes and beliefs. European Journal of Social Psychology ,30, 335–354.
Hewstone, M., Rubin, M., & Willis, H. (2002). Intergroup bias. Annual Review of Psychology ,53,
575–604.
Helsinki Watch. (1990). News from Romania: Ethnic conflict in Tirgu Mures. Helsinki Watch
newsletter ,2, 1–8.
Horowitz, D. L. (1999). Structure and strategy in ethnic conflict: A few steps towards synthesis. In B.
Pleskovic, & J. Stiglitz (Eds.), Annual world bank conference on development of economics
(pp. 345–370). Washington, DC: World Bank.
Jackman, M. R., & Crane, M. (1986). Some of my best friends are Black …: Interracial friendship and
Whites’ racial attitudes. Public Opinion Quarterly ,50, 459–486.
Klauer, K. C., & Meiser, T. (2000). A source-monitoring analysis of illusory correlations. Personality
and Social Psychology Bulletin ,26, 1074–1093.
Mays, V. M., Bullock, M., Rosenzweig, M. R., & Wessells, M. (1998). Ethnic conflict: Global
challenges and psychological perspectives. American Psychologist ,53, 737–742.
Mungiu-Pippidi, A. (1999). Transilvania subiectiva ˘. Bucures ¸ti: Humanitas.
National Institute of Statistics. (2008). The 2002 Census—The population of Mures ¸ county by ethnic
group. Retrieved from World Wide Web on 24 May 2008 from http://www.mures.insse.ro/phpfiles/Populatia_stabila_dupa_etnie_rec_02.xls
Paolini, S., Hewstone, M., Voci, A., Harwood, J., & Cairns, E. (2006). Intergroup contact and the
promotion of intergroup harmony: The influence of intergroup emotions. In R. Brown, & D.
Capozza (Eds.), Social Identities: Motivational, emotional, and cultural influences (pp. 209–238).
Hove, England: Psychology Press.
Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Psychology ,49, 65–85.
Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup contact theory. Journal of
Personality and Social Psychology ,90, 751–783.
Plant, E. A. (2004). Responses to interracial interactions over time. Personality and Social
Psychology Bulletin ,30, 1458–1471.
Rothbart, M. (1996). Category-exemplar dynamics and stereotype change. International Journal of
Intercultural Relations ,20, 305–321.
Rouhana, N. N., & Bar-Tal, D. (1998). Psychological dynamics of intractable ethnonational conflicts:
The Israeli-Palestinian case. American Psychologist ,53, 761–770.
Sidanius, J., Levin, S., Federico, C. M., & Pratto, F. (2001). Legitimizing ideologies: The social
dominance approach. In J. Jost, & B. Major (Eds.), The psychology of legitimacy: Emerging
perspectives on ideology, justice, and intergroup relations (pp. 307–331). New York: Cambridge
University Press.
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/casp28 V. Cernat

Stephan, W. G., Boniecki, K. A., Ybarra, O., Bettencourt, A., Ervin, K. S., & Jackson, L. A., et al.
(2002). The role of threats in the racial attitudes of blacks and whites. Personality and Social
Psychology Bulletin ,29, 1242–1254.
Stephan, W. G., & Renfro, C. L. (2002). The role of threat in intergroup relations. In D. Mackie, & E.
Smith (Eds.), From prejudice to intergroup emotions: Differentiated reactions to social groups
(pp. 265–283). New York: Psychology Press.
Tropp, L. R., & Pettigrew, T. F. (2005). Differential relationships between intergroup contact and
affective and cognitive dimensions of prejudice. Personality and Social Psychology Bulletin ,31,
1145–1158.
Wagner, U., Christ, O., Pettigrew, T. F., Stellmacher, J., & Wolf, C. (2006). Prejudice and minority
proportion: Contact instead of threat effects. Social Psychology Quarterly ,69, 180–190.
Wagner, U., van Dick, R., Pettigrew, T. F., & Christ, O. (2003). Ethnic prejudice in East and West
Germany: The explanatory power of intergroup contact. Group Processes and Intergroup
Relations ,6, 22–36.
West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with nonnormal variables:
Problems and remedies. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues and
applications (pp. 56–75). Thousand Oaks, CA: Sage.
Copyright #2008 John Wiley & Sons, Ltd. J. Community Appl. Soc. Psychol., 20: 15–29 (2009)
DOI: 10.1002/caspIntergroup contact in Romania 29