Contextual Effects On Subjective Nationa

NATIONS AND J O U R N A L O F T H E A S S O C I AT I O N AS
NATIONALISM
bs_bs_banner
FOR THE STUDY OF ETHNICITY

A N D N AT I O N A L I S M EN
Nations and Nationalism 20 (4), 2014, 701–720.
DOI: 10.1111/nana.12080
Contextual effects on subjective

national identity
JOAN BARCELÓ
The Wilf Family Department of Politics, New York University, New York,
NY, USA
ABSTRACT. Does the interaction between context and individual-level features

affect political attitudes? By using the case of Catalonia, a receiver region of interna-
tional and national immigration since the fifties, this paper intersects a classic accul-
turation model and a newly reemerging literature in political science on contextual
determinants of political behaviour to analyze how context affects subjective national
identity. Results reveal that environment matters. The Percentage of Spain-born
population in the municipality is statistically significant to account for variance in the
subjective national identity, even after controlling for age, sex, origin, language and
left–right orientation and other contextual factors. This conclusion suggests that
researchers should not underestimate the direct effect of the environment on feelings of
belonging in contexts of rival identities.
KEYWORDS: acculturation, assimilation, immigration, integration, multilevel

models, neighborhood effects
1. Introduction
The analysis of acculturation processes has been one of the topics with more
attention in the social sciences. It has been the concern of hundreds of scholars
from varying fields such as economists, sociologists, political scientists, psy-
chologists, among others, and its salience in the literature has been increasing
in the last decades due to the exponential trends of international migration.
Furthermore, the accentuated cultural differences between the sending and the
receiving societies that characterize the new waves of immigration (e.g.
Western individualism vs. Latin-American or African collectivism; Triandis
1995) have required intense efforts for theoretical and empirical studies of
what has sometimes been conceptualized as ‘cultural shock’ (Oberg 1960) or
‘acculturative stress’ (Berry 1970).
Acculturation is defined as comprehending ‘those phenomena which result
when groups of individuals having different cultures come into continuous
first-hand contact with subsequent changes in the original culture patterns of
either or both groups’ (Redfield et al. 1936: 149). However, the impact tends to
be more prominent for one of the two groups (Berry 1990). At the same time,
there exists a distinction between collective and psychological acculturation.
© The author(s) 2014. Nations and Nationalism © ASEN/John Wiley & Sons Ltd 2014
702 Joan Barceló
While the former refers to the general changes that occur in the ethnic or
identity group, the latter describes cultural shifts that take place at individual
level and that usually affect identity, values, attitudes and behaviour (Berry
and Sam 1997; Graves 1967).
For its historical migration trends, its complex situation with diverse and
opposite ethnic-national identifications and the availability of data on key
variables, Catalonia constitutes an excellent case that gathers fundamental
elements for better understanding how identity-based acculturation works.
Historically, Catalonia has been the center of attraction of national and inter-
national migration. Its privileged situation in Spain, as the gateway to Europe,
and its particular economic development in the last two centuries are the main
reasons why Catalonia has become a clear net receptor of migrants (Solé
2000). Immigration has been seen as a double-edged sword, an incoming cheap
labor that creates new economic opportunities for the native population but,
at the same time, a real threat for the Catalan cultural, social, national and
linguistic identity (Zapata-Barrero 2008). Traditionally, the massive entrance
of migrant population to Catalonia was made up of people from other regions
of Spain. Particularly important was the wave of national migrants that came
in the 60s, which coincided with high rates of economic growth and develop-
ment. This tendency shifted over the 90s, when the national migration rates
slumped and the international migration monopolized the recently arrived
population. All in all, Catalonia presents essential features that serve as the
basis for further studying acculturation and its impact over subjective national
identity (SNI) for both the native and the migrant population.
My main theoretical framework is Berry’s identity-based model of accul-
turation. It incorporates a new component usually forgotten in the study of the
migrants’ national identity, the context of reception. It is common to find in
the literature applications of acculturation models and analysis about how the
national identity of immigrants evolves over time where the context of recep-
tion is taken as a constant for the entire population. However, we know that
people’s characteristics are unevenly distributed and that people tend to live
near those that are perceived to be similar. Is then the context of reception
irrelevant for explaining acculturation processes or citizens’ SNI is affected by
context around them? This research gives an answer to these questions by
using quantitative techniques in the Catalan case. The use of quantitative over
qualitative analysis is driven by the research questions posed above, as they are
directed to identify evidence regarding a causal relationship between two
factors: context and choice, and so the emphasis is on understanding what is
occurring, instead of why or how it occurs.
The article proceeds as follows. First, Berry’s acculturation model is pre-
sented as well as the literature that surrounds it. Second, the importance of the
context of reception in sociology and political science literature is illustrated by
reviewing its historical roots and development in the field. However, it also
shows the need to follow up studying the impact of the contextual predictors
in the ethnic identity literature. Third, a new variant to the main theory is
Contextual effects on subjective national identity 703
provided to accommodate an extensive literature in political science about the

sensitivity in front of cross-pressure influence. The following section is devoted
to the specification of the general and specific hypotheses in its application to
Catalonia. After the hypotheses, the methodology is detailed by showing the
variables that will be used as well as the way the analysis will be carried out.
The results section presents the findings derived from testing the individual
and contextual predictors of Catalan vs. Spanish national identity. Lastly, the
significance of the results are discussed.
2. Theoretical framework
Models of acculturation
Psychological acculturation was first understood as a lineal and
unidimensional process whereby immigrants decide whether to lose their cul-
tural heritage in favor of the new society or to keep their previous traditions,
values and beliefs. Thus, the old culture and the new one were conceptualized
as opposing ends of a straight continuum (Gordon 1964).
In this sense, Berry’s contribution to the field with his two-dimensional
approach has entailed a great advance over earlier models and has dominated
the field over the last two decades (Berry 1970, 1990; Berry et al. 2002). He
pointed out that individual wishes to keep his heritage culture do not clash with
his wish to have contacts with the host society. The position of the individual
in both dimensions, maintaining original culture and having contacts with
receiving society, classifies the integration strategies into four possible catego-
ries: integration – if both they want to maintain their culture and have contacts
with the new society; assimilation – they contact host society but not maintain
their heritage; separation – when migrants want to keep their original culture
and having no contact with host majority; and marginalization – when they do
not want neither (Berry 1990). The two dimensions of the model refer to two
items of different natures, while the former is an attitude the latter is a behaviour
that casted some doubts over its consistency. To solve the problem some
authors suggested replacing contact by willingness to adopting the majority
culture (Bourhis et al. 1997; Matera et al. 2011). Table 1 depicts a two-by-two
cross table with the different possible acculturation strategies.
Table 1. Ideals of acculturation strategies
Own culture
Keeping Losing
Majority culture Adopting Integration Assimilation
Not adopting Separation Marginalization
704 Joan Barceló
The concept of acculturation has a broad meaning and multiple different

aspects of study. Schwartz et al. (2010) identified three different layers:
‘behavioral acculturation’, ‘value acculturation’ and ‘identity based accultura-
tion.’ The first deals with the differences of practices, customs or gastronomy,
the second with collectivism and individualism, and the third with the identi-
fication with the sending and the receiving society. In the case of international
migration, this model applies in all three dimensions with changing intensity
depending on the gap between the community of origin and destiny that marks
the immigrant’s ‘cultural shock’.
In the case of Catalonia and Spain some reservations should be discussed
before starting the analysis. First, it is evident that the application of the
acculturation model is not reasonable when the difference between both com-
munities is low. If the difference is nearly negligible there is no space for
changes in any of the communities. Although the differences of behaviour and
values between Spain and Catalonia were fairly similar, they differed more
significantly in terms of national identification, which enables researchers to
use the acculturation model in this case.
The second reservation is about prejudices, post-migration influences and
attitudes. Unlike most of the international migration processes, national
migration movements are characterized by an important preexisting knowl-
edge of the migrants about the receiving territory. At the same time, even
though most international immigrants remain in contact, at least to some
extent, with the reality and the news of their country of origin, this is abso-
lutely essential in internal migration, as the sources of information may almost
be identical before and after the migration process. Also, it is possible that as
the migrant does not change country, he or she may feel legitimated in avoid-
ing any identity change whatsoever. Despite all the additional difficulties for
applying an international migration model to a national context, this paper
attempts to examine to what extent the context of reception may be relevant
for understanding the patterns of acculturation.
The relevance of the context of reception for individual’s attitudes

Although some degree of acculturation always takes place in the migrant
communities, the literature has paid less attention on how patterns of accul-
turation differ across receiving societies. The interaction between migrants and
society may be a key factor for explaining how the newcomers feel about the
place they come to live (Rohmann et al. 2008). As receiving societies differ, the
result of the interaction between migrants and context should also differ. It is
known that the context of reception may fundamentally interfere on migrant’s
reception of host culture, and, therefore, on its acculturation process
(Rumbaut 2008). In particular, a negative local context characterized by
native’s attitudes of discrimination against migrants may become a major
source of stress and affect the acculturation process to the society at large
(Segal and Mayadas 2005).
The idea that context plays a role in the shaping of people’s attitudes,
values, beliefs, and as a partial determinant of behaviour, is not new in politi-
cal science. The neighborhood effect, so the importance of social interactions
with the rest of people around the individual to understand its political behav-
iour, dates back to the genesis of the electoral studies in the French school
(Siegfried 1949, 1995) and the most prominent classic in the field, The Ameri-
can Voter (Campbell et al. 1960). Although the individualistic approach taken
in it has usually been the focus of the analysis, they did assert that to under-
stand electoral behaviour is indispensable to comprehend the context in which
it occurs. Therefore, places or context are not neutral for political behaviour,
they are not empty containers, but full of content able to be significant in
final political outcomes through an impact on people’s attitudes, beliefs and
eventually behaviours.
Even though early contributions from social psychology attempted to
explain context influences on individual choices (Asch 1948), Huckfeldt’s work
is usually taken as the reference for the reemerging importance of the social
networks in politics (Huckfeldt 1979, 1983). Despite the variety and richness of
social networks does not match with geographical space (Baybeck and
Huckfeldt 2002), current literature focuses on the salience of geographically
bounded places (e.g. Baybeck 2006, Oliver and Mendelberg 2000). Context is
especially relevant in interethnic relations, as shown in American literature,
accounting for the great impact of intergroup beliefs within a social space. In
particular, the larger the black community in a neighborhood the higher the
negative stereotypes against blacks, a relationship even more powerful than
the North–South differences (Oliver and Wong 2003).
It seems to be accepted that heterogeneity of the population in a given
context matters for the understanding of how individuals feel accepted or
discriminated in that context. Thus, national attachment of the individuals
living in the same place is expected to affect individual attitudes, either nega-
tively or positively, in one direction or in the other. In the Catalan case, an
extensive literature has somehow dealt with the concept of homogeneity and
national identification. The link from ethnic/linguistic features, largely based
on origin and language, to national identification flows naturally in Catalonia
as reported by a number of studies that consistently show the salience
of the relationship (Chernyha and Burg 2012; Font et al. 1998, 2000;
Martínez-Herrera 2002; Montero and Font 1991; Pallarés and Font 1994;
Pallarés et al. 1997, 2010; Pérez-Nievas and Fraile Maldonado 2000). Thus, I
refer to ethnic homogeneity to the prevalence of both or either of the two key
factors, which function as a fine proxy of contextual variables with a potential
impact on individual’s national identity.1
Even though literature has set a consensus that context matters for accul-
turation and, specifically for the object of study here, for identity-based accul-
turation, it is still unresolved the direction and intensity to which context
drives people’s identity. In this sense, two theories have been argued in the
literature that relates to the Berry’s acculturation model: assimilationist, so
706 Joan Barceló
people move toward the average position of the context, vs. reactive or oppo-
sitional identification, so people react against the context and move in the
opposite direction (Cross et al. 1999; Ogbu 2004; Portes 1994). Berry’s theory
will be examined under a dynamic perspective by using these directional
theories. In addition to these contributions, political science has provided
some orientation about how the impact of contextual factors may be mediated
by cross-pressured effects on identity, which is subsequently analyzed.
Cross-pressure effects on identity

Societies are not dichotomized between natives and immigrants, but their
mixing process inevitably increases over time. Catalan society is not an excep-
tion in these terms and is composed by a large number of people with non-
aligned group identities. While literature has shown that origin and language
are two key determinants for identity, and so for the definition of contextual
homogeneity in Catalonia, the mixed character of the Catalan population
implies that there is a lack of alignments of identity groups of many individuals
who receive a combination of contradictory influences throughout their lives.
Those with more inconsistent influences are more comfortable with differ-
ences, more tolerant with others and, in general, have more positive feelings
and perceptions about out-groups (Brewer and Pierce 2005). Early contribu-
tions to the study of cross-pressure on individual’s voting behaviour by
Lazarsfeld et al. (1968) and Campbell et al. (1960) reached similar conclusions
and predicted that those individuals with more contradictory external influ-
ences would be more likely to have less specific group identity attachments,
have a more independent worldview and receptive attitudes.
Individual’s non-alignment impedes the psychological incorporation of
in-group biases and negativity against the other group, which makes them
undefended of the filters used to protect individuals from contextual
influences. The non-negative predisposition to a specific group may make
them more vulnerable to the opinion of the majority. In addition to the
abovementioned potential expectations of the impact of context over individ-
ual identity, those who have a less aligned identity may be more subjected to
the mainstream values than those who have more consistent and neatly defined
identity groups.
3. Hypotheses
The impact of the contextual factors on the individual’s national identity

requires further empirical research to examine not only whether it impacts on
individuals behaviour but also how it does. The literature is not clear about
how places affect identity-based acculturation, and the aim of this paper is to
fill this empirical gap. Thus, while the general hypotheses are that levels
of homogeneity in the context matter for the understanding of individual’s
1. Assimilation Model 2. Reactive/SeparationModel 3. Integration Model

Catalan-oriented SNI
Catalan-born population Catalan-born population Catalan-born population
(low Spanish-born) (low Spanish-born) (low Spanish-born)
Catalan Spanish Catalan Spanish Catalan Spanish
Figure 1. Hypothetical acculturative scenarios in Catalonia.

Note: Lines represent hypothetical average SNI by Catalan-born population of the
municipality and conditional on the cultural framework of reference (Catalan vs.
Spanish).
subjective national identification. The specific hypotheses reflect the direction

of movements of the communities and differentiate between assimilation,
integration and reactive models of acculturation. Figure 1 presents the direc-
tion of the specific hypotheses expected by each of the three acculturation
strategies.
1. Assimilation model
General hypothesis
When the host society presents high (low) levels of ethnic homogeneity,
members of this society tend to be more similar to the context than they were
expected to be accounting for their individual features.
Specific hypothesis
When a given municipality presents low (high) levels of Spain-born population
(predictor related to Catalan [Spanish] national identification), members of
this municipality tend to be more identified with Catalonia (Spain) than they
were expected to be according to their individual features.
2. Reactive/separation model
General hypothesis
When the host society presents high levels of ethnic homogeneity, members of
the minority ethnic group tend to be more dissimilar or indifferent to the
context and the members of the majority group tend to be more similar to the
context than they were expected to be accounting for their individual features.
Specific hypothesis
When a given municipality presents low levels of Spain-born population (pre-
dictor related to Catalan national identification), members of the minority
708 Joan Barceló
ethnic group of the municipality (Spain-oriented group) tend to be equally or

less identified with Catalonia than they were expected to be accounting for
their individual features.
3. Integration model
General hypothesis
When the host society presents high levels of ethnic homogeneity, the members
of the minority ethnic group tend to be more similar to the context and the
members of the majority group tend to be more dissimilar to the context than
they were expected to be accounting for their individual features.
Specific hypothesis
When a given municipality presents low levels of Spain-born population (pre-
dictor related to Catalan national identification), the members of the minority
ethnic group of the municipality tend to be more identified with Catalonia and
the members of the majority group tend to be less identified with Catalonia
than they were expected to be accounting for their individual features.
4. Methodology
Data analysis
Multilevel models were used for the analysis of the data from Institut de
Ciències Polítiques i Socials (ICPS) for the individual-level characteristics and
data from Institut d’Estadística de Catalunya (IDESCAT) for the
municipality-level contextual analysis. This level of analysis, municipality vs.
other alternatives (comarca, province, etc.), constitutes the operationalization
of host society for several reasons. Methodologically, municipalities are the
smaller political units in Catalonia and so taking them allow researchers to
maximize the second-level observations. Theoretically, municipalities capture
the most immediate individual context with available data, they best approxi-
mate to immigrant enclaves (Logan et al. 2002) and are responsible for a
number of unique political and economic processes (Alesina and Ferrara
2005). Even though many opinion polls are released in Catalonia every year
(e.g. CIS, CEO, ICPS, etc.), the ICPS gives access to the linkage between
observations and contextual variables by postcode matching. This survey aims
at the study of political attitudes in Catalonia and, therefore, incorporates the
key variables required for the study of acculturation and national identity. It
has a sample size of 2,000 respondents with an extended geographic represen-
tation of the four provinces of Catalonia, carried out in September 2010 and
with a margin of error for the whole sample of 2.7 per cent. The survey design
includes 64 different municipalities representative of the Catalan population,
what constitutes a sufficient second-level sample size, and embraces a wide
range of municipalities. Among these, there are big cities such as Barcelona
(N = 450; 22.8 per cent of the observations), Mataró (N = 142; 7.1 per cent)
and Hospitalet del Llobregat (N = 134; 6.7 per cent), and small local commu-
nities such as Almatret or Os de Balaguer, among others. Figure 2 shows the
distribution of the municipalities in the sample by population size.
As far as the methodological technique is concerned, the usage of multilevel
modeling is becoming a more commonly available method, and is potentially
an excellent way to address contextual determinants of individual behaviour
(Gelman and Hill 2006). The SPSS (IBM Corp. 2011) and hierarchical linear
model (HLM) were used to fit multilevel models with an ordinal distribution
assumption with five categories and a logit link. The ordinal nature of the
dependent variable is consistent with the usage of the ordered multinomial
logit models. The method of estimation was a restricted maximum likelihood
procedure.
First, an empty model is fitted to generate the null model, which constitutes
the empirical baseline. In the second stage, the individual-level characteristics
are incorporated, which are selected based on prior literature (Chernyha and
Burg 2012; Font et al. 1998, 2000; Martínez-Herrera 2002; Montero and Font
1991; Pallarés and Font 1994; Pallarés et al. 1997, 2010; Pérez-Nievas and
Figure 2. Number of municipalities by population size in the sample, 2011.
710 Joan Barceló
Fraile Maldonado 2000). In the third stage, models included both individual
and municipality level effects, contextual-level predictors are introduced based
on their availability for all the groups. Finally, cross-level interaction between
individual- and municipality-level characteristics are added to the models in
order to assess whether contextual variables were modified by language at
individual-level. Prior to presenting the results, the operationalization of the
concepts into variables are described below.
Dependent variable
As a natural derivation of the theoretical framework and the hypotheses above
suggested, the key dependent variables that will be studied is SNI. For SNI, we
use the subjective relative identity scale commonly used in the literature (the
so-called ‘Moreno question’) so as to capture the dual or balancing identities
traditionally existing in Catalonia (e.g. Moreno and Arriba 1996; Moreno
et al. 1998). The question has been categorized as 0 to 4 positions, being 0 if the
respondent states to feel exclusively Spanish, 1 if more Spanish than Catalan,
2 if as Catalan as Spanish, 3 if more Catalan than Spanish, and 4 if exclusively
Catalan.
Independent variables
The main independent variables of the analysis are the contextual-level vari-
ables regarding potential predictors of national identity. Accounting for the
limitations of the quality of second-level variables, the variable that has better
suitability for the analysis is the percentage of municipality population born in
the rest of Spain in 2011. This variable measures the density of the Spain-born
population and, as a good indicator of national identity, it captures a sense of
environmental national orientation at municipality-level. From the 64 munici-
palities from which the ICPS sample was generated, the percentage of Spain-
born population ranged from a four per cent in Isona i Conca Dellà to just
above twenty-seven per cent of the population in Hospitalet del Llobregat,
Sant Feliu del Llobregat and Viladecans. The Spain-born population consti-
tutes a 15.7 per cent of the total population of the municipalities in the sample.
An alternative potential indicator is the percentage of the population that
states to be proficient in Catalan: understand, read, speak or write properly.
However, this indicator has not been updated in the Idescat database at
municipality-level since 2001, which limits its usefulness.
Individual- and municipality-level controls

Many scholars in the last thirty years have contributed to the understanding of
the causes of nationalism in Catalonia and its main predictors (e.g. Chernyha
and Burg 2012, Font et al. 2000, Pallarés et al. 1997). As the contribution that
this paper seeks is to put forward the importance of the nationalist context
even after taking into account individual-level variables, the models have to be
built by adjusting for these individual predictors as well as other potential
second-level factors. All the models have been first controlled for the following
variables provided in the database: age, squared age, sex, origin (recoded to
dummy variables for each category with native and native parents as the
category of reference), language of common usage (recoded to dummy vari-
ables for each category with common Catalan speakers as the reference value)
and left–right scale position (extreme left takes the value of 0 and extreme right
the value of 6). Table 2 provides the descriptive statistics for the individual-
level controls included in the analysis. In the empirical section, models are
built excluding foreigner respondents with foreigner parents, and those with
missing values in relevant variables.
The inclusion of other individual-level variable related to political attitudes
such as support for a nationalist/non-nationalist political party, satisfaction
with national/regional government, preferred level of decentralization, most
Table 2. Descriptive statistics of individual-level controls and outcome
Mean (SD) Min. Max. Obs.

Age 47.27 (18.12) 18 94 2,000
Sex
Female 50.2% 0 1 1,003
Male 49.8% 0 1 997
Origin
Native (Catalan-born, Catalan-born parents) 33.5% 0 1 670
2nd generation mixed (Catalan-born, one 17.2% 0 1 345
Catalan-born parent)
2nd generation (Catalan-born, parents born 10.9% 0 1 218
outside Catalonia)
Immigrant (born outside Catalonia) 19.9% 0 1 370
Foreigner (responder and parents born 17.2% 0 1 398
outside Spain)
Language
Catalan 42.6% 0 1 852
Both 7.4% 0 1 148
Others 9.3% 0 1 187
Spanish 40.6% 0 1 811
Left–right scale 2.30 (1.26) 0 6 1,515
National identity
Only Catalan 18.3% 0 1 365
More Catalan than Spanish 22.8% 0 1 456
As Catalan as Spanish 37.4% 0 1 749
More Spanish than Catalan 5.5% 0 1 109
Only Spanish 6.4% 0 1 128
712 Joan Barceló
relevant political arena (regional or national) may cause severe problems to the
model. On the one hand, most of these variables are derived from precedent
variable in the funnel of causality, which are already introduced in the model.
For instance, if an individual supports a nationalist political party, this might
be largely due to the effect of precedent variables in the causal relationship
such as its age, sex, origin, language and left–right position (independent
variable included in the models) and, even a result of its national identity
(dependent variable). Therefore, the inclusion of this variable seems problem-
atic since they are at a different level in the causality chain. At the same, they
might artificially overshadow the effect of the rest of independent variable,
both individual and contextual variable, since it lies in a closer position to the
dependent variable. In fact, some of them may be even beyond – not as
independent but as a dependent of the SNI – the dependent variable.
On the other hand, the inclusion of these variables would also generate
endogenous problems due to the bidirectional causality of these variables with
the SNI. For all these reasons, the best approach is to use individual and
contextual levels that are equally close in the causal chain with the dependent
variable, so they do not capture excessive variance from each other, and, at the
same time, variables that clearly precede the dependent variable and from
which causality can be claimed. However, it is advisable that control variables
are retained as long as they are at a similar distance from the outcome as the
main independent variables, otherwise the former would artificially over-
shadow the latter.
All in all, contextual factors must be at the same level in relation to the
outcome as the individual controls and this is subject to the data available in
the Idescat at the municipality-level. As far as the contextual-level controls
(Table 3 displays their main descriptive statistics), a number of variables have
been introduced in the models: population and population density and its
various functional forms in 2011, and gross domestic product (GDP) per
capita in 2008. The measure of the GDP per capita has the shortcoming that
it is only provided by the Idescat for median and large municipalities. For
those small, its GDP has been estimated by using the extrapolation of the
relationship between the tax base (available for small municipalities) of the
Table 3. Descriptive statistics of contextual-level control variables
Mean (SD) Min. Max. Obs.

Population, 2011 52,635 (204,676) 378 1,615,400 64
Population density, 2011 1,470 (3,320) 6,6 20,650 64
Province: Barcelona 29.7% 0 1 64
Gross domestic product per 25,003€ (9,518) 11,000€ 75,700€ 64
capita, 2008
Spain-born population, 2011 15.7% (6.3) 4% 27% 64
median and large and its GDP. Therefore, none of the variables available are
related to political attitudes but, likewise the set of independent variables
precede the SNI and are equally close to or far from it.
5. Results
In this section, I present the results of the multilevel analysis to test the
abovementioned hypotheses. The order follows an incremental approach from
the null model to a model with all the predictors available in the table. I build
the equations by using ordered multinomial logistic regression models with
two levels of analyses.
Individual and contextual predictors of SNI

To confirm the existence of cross-level effects between first-level subjective
identity predictors and second-level local municipality predictors, a null model
of multilevel analysis is first tested. Even though the ordinal logit model does
not provide an estimate of level 1 variance, and so the interclass correlation
cannot be calculated, it yields a between-group variance estimate from which
an examiner can assess whether there is sufficient variance across groups to
justify a multilevel analysis. Table 4 reports the estimates of the individual and
contextual-level effects on SNI. The null model (shown column 1 of Table 4)
indicates a statistically significant between-municipality variance (τ00 = 0.21),
P < 0.01). As a consequence, there is sufficient group-level variance to support
the need for following-up multilevel linear modeling analysis to study the
variation in SNI at a municipality-level.
The initial model (column 2 of Table 4) includes demographic variables
such as age, sex and origin. At the same time, a social variable such as language
of common usage and a clearly politically oriented variable such as position in
the left–right scale. The individual-level model (column 3 of Table 4) confirms
the importance of socio-political variables. First, the odds of being in the
lowest category in the SNI, the only Spanish, increase if the respondent was
born in Catalonia with either one of the parents in Spain (γ04 = 0.87,
OR = 2,39; t-ratio = 5.02), or both in Spain (γ05 = 0.61, OR = 1.84; t-ra-
tio = 2.98) or the three (respondent and her parents) have been born in Spain
(γ06 = 1.98, OR = 7.2; t-ratio = 9.46) as compared with when it is categorized as
Catalan native (the three of them born in Catalonia).
Similarly, the likelihood of being in the only Spanish category increases as
the respondents’ language is Spanish (γ09 = 2.19, OR = 8.94; t-ratio = 10.08),
both (γ08 = 1.58, OR = 4.85; t-ratio = 6.43) or others (γ07 = 2.68, OR = 14.6;
t-ratio = 4.2) as compared with Catalan-speakers. Furthermore, ideology
appears to have a substantial effect on the expected SNI. A movement to the
right increases the likelihood of being in the only Spanish category by a
significant amount (γ07 = 0.31, OR = 1.36; t-ratio = 6.5).2 The first two are
714 Joan Barceló
Table 4. Individual and contextual determinants of SNI. Ordinal logistic

regression
Ordered logit model

Dependent variable: SNI (1) (2) (3) (4)
Individual-level predictors
Intercept 1 −2.94*** −6.20*** −6.14*** −6.14***
Threshold 2 0.78*** 0.92*** 0.92*** 0.92***
Threshold 3 2.81*** 3.93*** 3.94*** 3.94***
Threshold 4 4.07*** 4.07*** 5.79*** 5.79***
Age 0.00 ns 0.00 ns 0.00 ns
Age Squared 0.00 ns 0.00 ns 0.00 ns
Sex (male) 0.00 ns 0.00 ns 0.00 ns
Origen (ref. category: Catalan Native)
2nd Generation Mixed 0.87*** 0.83*** 0.83***
2nd Generation 0.61*** 0.58*** 0.58***
1st Generation 1.98*** 1.95*** 1.95***
Language (ref. category: Catalan)
Both 1.59*** 1.57*** 1.57***
Others 2.69*** 2.65*** 2.65***
Spanish 2.20*** 2.17*** 2.17***
Left–right scale 0.31*** 0.32*** 0.32***
Contextual-level predictors
Population (log) 0.14** 0.13**
Population density (log) −0.09ns −0.07 ns
Province: Barcelona −0.25* −0.26*
Gross domestic product −0.50** −0.57**
per capita (log)
Spain-born population 3.75*** 4.44**
(%)
Cross-level interactions
Spain-born population −10.7**
(%)* Language both
Spain-born population −3.44 ns
(%)*Language others
Spain-born population −2.90 ns
(%)*Language
Spanish
Covariance parameters
Between groups 0.21*** 0.04* 0.00 ns 0.00 ns
Observations level 1 1,389 1,389 1,389 1,389
Observations level 2 64 64 64 64
Category of reference: Only Catalan. Significance levels: P < .1*; P < .05**; P < .01***
significant and in the expected direction. The importance of ideology as a

predictor for SNI is not a surprise, yet it usually results to be a significant
factor in related literature on national identity in Catalonia. Most likely,
Spain’s history and the conservative dictatorship play an explanatory role for
this phenomenon. Thus, the historically deep-rooted connection between
Spanish identity and rightist ideology is a reasonable explanation for the
significant predictive capability of the left–right scale (Muñoz 2009, 2012).
A strictly individualistic approach may hide phenomena that go beyond the
individual level and belong to the individual’s context. To test for this, column
3 of Table 4 incorporates the potential municipality-level predictors to explain
individual national identity (all the predictors from Table 2). These results
support the thesis that the larger the Spanish-origin population in the respon-
dent’s municipality the lower is the predicted individual Catalan national
identity for those who live in the municipality even after adjusting for origin,
language, ideological orientation of the individual and the population, popu-
lation density, GDP per capita and the location province of Barcelona of the
municipality (γORIGINMUNICIPALITY = 3.75; P < 0.01). Additionally, this provides
evidence to support the general assimilationist for the Catalan population by
which individuals’ SNI is affected by the characteristics of the population in
which they live. Apart from this interesting result, other contextual-level vari-
ables also have an impact on individual’s SNI such as municipality’s popula-
tion (positively related with Spanish-oriented SNI), GDP per capita and
located in Barcelona province (negatively related with Spanish-oriented SNI).
Although the effect of being in the Barcelona province was expected to be
positive, this negative effect must be understood as a coefficient adjusted by
population size and Spain-born population. With all this controls in the
model, it is not surprising that population density is not distinguishable from
zero.
All in all, results in the first model indicate that there has been a remarkable
improvement of the null model with the introduction of the individual-level
and municipality-level variables. There has been a reduction of the between-
municipality variance and its significant level from 0.21 (P < 0.01) to 0.04
(P > 0.10), which implies that individual’s SNI is no longer significantly dif-
ferent across municipalities when the set of individual and contextual variables
are included.
Apart from this general assimilationist tendency, it is theoretically and
empirically interesting to analyze whether this effect is moderated by first-level
variables such as individual’s language. In the cross-level effects model
(column 4 of Table 4), individual and contextual predictors remain qualita-
tively identical to the previous main effects equations, yet some interactions
are substantively interesting and others statistically significant. First, the
model continues to indicate a positive and significant effect of the Spain-born
percentage in the municipality on individual’s SNI at the baseline (with no
interactions), which strengthens the assimilationist effect for those who speak
Catalan. However, this effect is not the same across all the language groups.
716 Joan Barceló
The moderating effect is negative, yet nonsignificant, for Spanish speakers as

well as speakers of other languages. In both cases, the main effect of the
municipality (γORIGINMUNICIPALITY = 4.44; P < 0.01) is almost entirely compen-
sated by one’s Spanish language (γORIGINMUNICIPALITY = −3.44; P > 0.1), or par-
tially compensated by another language (γORIGINMUNICIPALITY = −2.90; P > 0.1),
which diminishes the main effect coefficient to 1.0 (OR = 2.5) and 1.54
(OR = 4.67), respectively. Despite this reduction, the coefficient of the main
effect remains large in magnitude and the reduction is statistically nonsignifi-
cant for both groups.
For all this, results lend credence to the assimilationist theory for most of
the speaking groups (Catalan-speakers, Spanish-speakers and speakers of
other languages). The general model is only broken by those who state to
speak both Catalan and Spanish. Even though this coefficient should be taken
with caution due to the small sample size of this group (7.4 per cent of the
sample), the interaction between speaking both languages and the percentage
of Spain-born population in the municipality is highly negative and statisti-
cally significant, which is consistent with the reactive/separation model for
those who speak Catalan as well as Spanish.
6. Discussion
I examine the process of acculturation in a complex case characterized by

competing ethnic-national identifications. Multiple waves of national and
international immigration in Catalonia have been steady throughout the last
fifty years, which has left an extremely diverse social picture. Added to the
preexistent national rivalry, these migratory movements have increased even
further the political complexity in the region. At the same time, it constitutes
an excellent scenario to see what factors determine the orientation of
people’s national identification. Literature has centered in the study of vari-
ations such as origin, language or ideology, among others, but always at the
individual-level (Chernyha and Burg 2012; Font et al. 1998, 2000;
Martínez-Herrera 2002; Montero and Font 1991; Pallarés and Font 1994;
Pallarés et al. 1997, 2010; Pérez-Nievas and Fraile Maldonado 2000). This
paper contributes to the literature by analyzing the effect of the context on
SNI.
For this purpose, multilevel models have been implemented to analyze the
impact of individual- and municipality-level variables on national identity.
Results are sharp. Individual-level variables predominate and seem to be
stable in all the models. Particularly, origin, language and left–right position
are the main predictors of national identity. The same pattern of relationship
between sociodemographic individual variables and national identity has been
a constant in the literature about identity in Catalonia mentioned above. At
the contextual level, it has been established a negative relationship between the
percentage of population born in the rest of Spain and identification with
Catalonia as opposed to Spain as a general trend. Thus, the second-level

predictor means that citizens are more nationally identified with Spain as the
number of people born in Spain in their municipality increases. However, this
is partially moderated by the language of habitual usage of the respondent.
While the effect is significant for Catalan speakers, whose likelihood of being
more identified with Spain augments as the percentage of Spain-born popula-
tion in their municipality also increases, those who usually speak Spanish and
other their assimilationist process to the average of the context is more attenu-
ated. By contrast, bilingual respondents that habitually use both languages at
a similar frequency present a high tendency to react to the context, although
these results should be taken carefully due to the small number of observations
in this category.
Nevertheless, there are some considerations to bear in mind in the inter-
pretation of the results. A potential objection to the estimation of contextual
effects is to argue that the process of residential clustering is not random,
and that it is often endogenous to individuals’ preferences or regional or
local institutions (Adelman 2005; Fong and Chan 2010). Thus, it might be
the case that citizens with a Spanish-oriented identity would settle in munici-
palities with a more heterogeneous (Spain-born) context or with more
favorable policies, while those more Catalan-oriented might tend to live in
municipalities closer to their own features or policies more valued for them.
Notwithstanding the commonality of the residential clustering in migratory
flows, it is typically a function of observable characteristics such as origin or
language rather than on national identification itself. For instance, emigrants
look for settlement areas where people from their own origin already locate
because it constitutes an advantage in their initial endeavor in the new
country (e.g. better prospects of getting a residence or a job, preexistent
social network of support in these places, feeling like home, etc.), eases the
settlement process as the learning costs required to know the new language
of the place are diminished and policies may be already favoring the
newcomers.
However, it is less reasonable that the clustering process is done through
non-observable features as they are hard to identify and could only provide
little advantage for the settlement process such as national identity once the
independent effect of origin and language are taken into account. For all
this, the models presented here are robust against this potential objection
since they adjust for those individual characteristics that are relevant for the
nonrandom clustering. In other words, if people with certain origin or lan-
guage prefer to move to certain places with more people with similar origin
and language, then these variables should capture the variation in the
contextual variable and, thus, place should have no statistical effect on
individual identity. By contrast, the municipality-level predictor and the
cross-level interactions related to it do have an independent effect on SNI
after adjusting for relevant individual factors and other municipality-level
predictors.
718 Joan Barceló
7. Conclusion
Results should encourage further research in the study of the contextual

influences on identity and behaviour. The extent and direction to which
context matters are still a source of discussion, and they are likely to be
contingent on case-specific features. To return to Berry’s theoretical back-
ground, it seems that the Catalan case most faithfully resembles an
assimilationist strategy of acculturation for most of the speaking groups,
although the intensity varies by individuals’ language.
On the one hand, when the context presents high levels of ethnic homo-
geneity (Catalan-born predominance), the minority ethnic group (Spain-born
minority) tends to be more similar to the context, and so more Catalans-like
identity than they would be otherwise. On the other hand, in this same sce-
nario, the majority ethnic group, constituted by the Catalan-speaking group
tends to be more influenced by the characteristics of the context and adapt to
it, and thus even more Catalans than they would be otherwise.
In sum, context is found to be as powerful as to be able to influence not only
through individual-level factors but also as an independent effect on individual
identity. As a consequence, results advise the incorporation of second-level
controls by location to gain efficiency of the models in future research on SNI.
Acknowledgements
I am grateful to Francesc Pallarés, Enric Martínez-Herrera, Lucía Medina and

three anonymous reviewers for their comments and suggestions. I also grate-
fully acknowledge financial support from the Institut de Ciències Polítiques i
Socials (ICPS) in Barcelona.
Notes
1 Availability of data limits the empirical results to the contextual effect of origin. Language is,
thus, used as an individual-level variable only. More details on the operationalization are provided
in the methodology section.
2 Notice that ordered logit coefficients in HLM go in the opposite direction to what ordered logit
coefficients are in other standard statistical packages. They show how the odds of being in a lower
category change, while standard ordered logit coefficients show how the odds of being in a higher
category change. Hence they would be the same number but with an opposite sign.
References
Adelman, R. M. 2005. ‘The roles of race, class, and residential preferences in the neighborhood
racial composition of Middle-Class Blacks and Whites*’, Social Science Quarterly 86, 1: 209–28.
Alesina, A. and Ferrara, E. L. 2005. ‘Ethnic diversity and economic performance’, Journal of
Economic Literature 43, 3: 762–800.
Asch, S. E. 1948. ‘The doctrine of suggestion, prestige and imitation in social psychology’,
Psychological Review 55, 5: 250–76.
Baybeck, B. 2006. ‘Sorting out the competing effects of racial context’, The Journal of Politics 68,
2: 386–96.
Baybeck, B. and Huckfeldt, R. R. 2002. ‘Spatially dispersed ties among interdependent citizens:
connecting individuals and aggregates’, Political Analysis 10, 3: 261–75.
Berry, J. W. 1970. ‘Marginality, stress and ethnic identification in an acculturated aboriginal
community’, Journal of Cross-Cultural Psychology 1, 3: 239–52.
Berry, J. W. 1990. ‘Psychology of acculturation’ in J. J. Berman (ed.), Nebraska Symposium on
Motivation, 37. Lincoln, NE: University of Nebraska Press.
Berry, J. W. and Sam, D. L. 1997. ‘Acculturation and adaptation’ in J. W. Berry (ed.), Handbook
of Cross-Cultural Psychology, vol III: Social Behavior and Applications. Boston: Ally & Bacon.
Berry, J. W. et al. 2002. Cross-Cultural Psychology: Research and Applications. 2nd edn. Cam-
bridge, UK: Cambridge University Press.
Bourhis, R. Y. et al. 1997. ‘Towards an interactive acculturation model: a social psychological
approach’, International Journal of Psychology 32, 6: 369–86.
Brewer, M. B. and Pierce, K. P. 2005. ‘Social identity complexity and outgroup tolerance’,
Personality and Social Psychology Bulletin 31, 3: 428–37.
Campbell, A. et al. 1960. The American Voter. New York: Wiley.
Chernyha, L. T. and Burg, S. L. 2012. ‘Accounting for the effects of identity on political behavior:
descent, strength of attachment, and preferences in the regions of Spain’, Comparative Political
Studies 45, 6: 774–803.
Cross, W. E. et al. 1999. ‘African American identity development across the life span: educational
implications’ in R. H. Sheets and E. R. Hollins (eds.), Racial and Ethnic Identity in School
Practices: Aspects of Human Development,. Mahwah, NJ: Lawrence Erlbaum.
Fong, E. and Chan, E. 2010. ‘The effect of economic standing, individual preferences, and
co-ethnic resources on immigrant residential clustering 1’, The International Migration Review
44, 1: 111–41.
Font, J. et al. 1998. L’abstenció en les eleccions al Parlament de Catalunya. Barcelona:
Mediterrània.
Font, J. et al. 2000. ‘El comportament electoral dels catalans: factors i pautes’ in M. R. Virós et al.
(eds.), Eleccions i Comportament Electoral a Catalunya, 1989–1996. Mediterrània., Barcelona:
Fundació Jaume Bofill.
Gelman, A. and Hill, J. 2006. Data Analysis Using Regression and Multilevel/Hierarchical Models.
Cambridge; New York: Cambridge University Press.
Gordon, M. M. 1964. Assimilation in American Life the Role of Race, Religion, and National
Origins. New York: Homepage of Oxford University Press. Online.
Graves, T. D. 1967. ‘Psychological acculturation in a tri-ethnic community’, Southwestern Journal
of Anthropology 23, 4: 337–50.
Huckfeldt, R. R. 1979. ‘Political participation and the neighborhood social context’, American
Journal of Political Science 23, 3: 579–92.
Huckfeldt, R. R. 1983. ‘Social contexts, social networks, and urban neighborhoods:
environmental constraints on friendship choice’, The American Journal of Sociology 89, 3:
651–69.
IBM Corp. 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp.
Lazarsfeld, P. F. et al. 1968. The People’s Choice; How the Voter Makes up His Mind in a
Presidential Campaign. New York: Columbia University Press.
Logan, J. R. et al. 2002. ‘Immigrant enclaves and ethnic communities in New York and Los
Angeles’, American Sociological Review 67, 2: 299–322.
Martínez-Herrera, E. 2002. ‘From nation-building to building identification with
political communities: consequences of political decentralisation in Spain, the Basque
Country, Catalonia and Galicia, 1978–2001’, European Journal of Political Research 41, 4:
421–53.
720 Joan Barceló
Matera, C. et al. 2011. ‘The role of immigrant acculturation preferences and generational status in
determining majority intergroup attitudes’, Journal of Experimental Social Psychology 47, 4:
776–85.
Montero, J. R. and Font, J. 1991. ‘El voto dual en Cataluña: Lealtad y transferencia de votos en
las elecciones autonómicas’, Revista de Estudios Políticos 73: 7–34.
Moreno, L. and Arriba, A. 1996. ‘Dual identity in Autonomous Catalonia’, Scottish Affairs 17:
78–97.
Moreno, L. et al. 1998. ‘Multiple identities in decentralized Spain: the case of Catalonia’, Regional
& Federal Studies 8, 3: 65–88.
Muñoz, J. 2009. ‘From National-Catholicism to Democratic Patriotism? Democratization and
reconstruction of national pride: the case of Spain (1981–2000)’, Ethnic and Racial Studies 32,
4: 616–39.
Muñoz, J. 2012. La construcción política de la identidad española: ¿del nacionalcatolicismo al
patriotismo democrático? Madrid: Centro de Investigaciones Sociológicas.
Oberg, K. 1960. ‘Cultural shock: adjustment to new cultural environments’, Practical Anthropol-
ogy 7: 177–82.
Ogbu, J. U. 2004. ‘Collective identity and the burden of “Acting White” in black history, com-
munity, and education’, The Urban Review 36, 1: 1–35.
Oliver, E. J. and Mendelberg, T. 2000. ‘Reconsidering the environmental determinants of white
racial attitudes’, American Journal of Political Science 44, 3: 574.
Oliver, E. J. and Wong, J. 2003. ‘Intergroup prejudice in multiethnic settings’, American Journal
of Political Science 47, 4: 567–82.
Pallarés, F. and Font, J. 1994. ‘Las elecciones autonómicas en Cataluña 1980–1992’ in P. Del
Castillo (ed.), Comportamiento Político y Electoral. Madrid: Centro de Investigaciones
Sociológicas.
Pallarés, F. et al. 1997. ‘Non state-wide parties in Spain: an attitudinal study of nationalism and
regionalism’, Publius 27, 4: 135–69. Toward Federal Democracy in Spain.
Pallarés, F. et al. 2010. ‘State-level politics and regional electoral competition: first- and second-
order effects in Spain’s Autonomous Elections in 2007’, Regional & Federal Studies 20, 4–5:
561–75.
Pérez-Nievas, S. and Fraile Maldonado, M. 2000. ‘Is the Nationalist Vote Really Nationalist?
Dual Voting in Catalonia 1980–1999’, Fundación Juan March, Estudio/Working Paper, vol.
2000/147.
Portes, A. 1994. ‘Introduction: immigration and its aftermath’, The International Migration
Review 28, 4: 632–9. Special Issue: The New Second Generation.
Redfield, R. et al. 1936. ‘Memorandum for the study of acculturation’, American Anthropologist
38, 1: 149–52.
Rohmann, A. et al. 2008. ‘When attitudes do not fit: discordance of acculturation attitudes as an
antecedent of intergroup threat’, Personality Social Psychology Bulletin 34: 337–52.
Rumbaut, R. G. 2008. ‘Reaping what you sow: immigration, youth, and reactive ethnicity’,
Applied Developmental Science 12, 2: 108–11.
Schwartz, S. J. et al. 2010. ‘Rethinking the concept of acculturation: implications for theory and
research’, The American Psychologist 65, 4: 237–51.
Segal, U. A. and Mayadas, N. S. 2005. ‘Assessment of issues facing immigrant and refugee
families’, Child Welfare 84, 5: 563–83.
Siegfried, A. 1949. Géographie électorale de L’Ardèche sous la IIIe République. Paris: A. Colin.
Siegfried, A. 1995. Tableau politique de la France de l’ouest. Paris: Imprimerie nationale.
Solé, C. 2000. ‘Inmigración interior e inmigración exterior’, Papers: revista de sociologia 60:
211–24.
Triandis, H. C. 1995. Individualism and Collectivism. Boulder: Westview Press.
Zapata-Barrero, R. 2008. La inmigración en naciones minoritarias: Flandes, Quebec y Cataluña en
perspectiva. Barcelona: Icaria Editorial.
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and rea

Contextual Effects On Subjective Nationa

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Contextual Effects On Subjective Nationa

Uploaded by

Copyright:

Available Formats

NATIONS AND J O U R N A L O F T H E A S S O C I AT I O N AS

FOR THE STUDY OF ETHNICITY

Contextual effects on subjective

ABSTRACT. Does the interaction between context and individual-level features

KEYWORDS: acculturation, assimilation, immigration, integration, multilevel

provided to accommodate an extensive literature in political science about the

Table 1. Ideals of acculturation strategies

The concept of acculturation has a broad meaning and multiple different

The relevance of the context of reception for individual’s attitudes

Cross-pressure effects on identity

The impact of the contextual factors on the individual’s national identity

1. Assimilation Model 2. Reactive/SeparationModel 3. Integration Model

Catalan Spanish Catalan Spanish Catalan Spanish

Figure 1. Hypothetical acculturative scenarios in Catalonia.

subjective national identification. The specific hypotheses reflect the direction

ethnic group of the municipality (Spain-oriented group) tend to be equally or

Figure 2. Number of municipalities by population size in the sample, 2011.

Individual- and municipality-level controls

Table 2. Descriptive statistics of individual-level controls and outcome

Mean (SD) Min. Max. Obs.

Table 3. Descriptive statistics of contextual-level control variables

Mean (SD) Min. Max. Obs.

Individual and contextual predictors of SNI

Table 4. Individual and contextual determinants of SNI. Ordinal logistic

Ordered logit model

significant and in the expected direction. The importance of ideology as a

The moderating effect is negative, yet nonsignificant, for Spanish speakers as

I examine the process of acculturation in a complex case characterized by

Catalonia as opposed to Spain as a general trend. Thus, the second-level

Results should encourage further research in the study of the contextual

I am grateful to Francesc Pallarés, Enric Martínez-Herrera, Lucía Medina and

You might also like