Immigrant presence,group boundaries,and support for the welfare state in Western European societies

Abstract

The intersection of group dynamics and socioeconomic status theories is applied as a framework for the puzzling relationship of immigration and support for the welfare state in Western Europe. Group dynamics theories suggest that how individuals define their group boundaries moderates the impact of immigration on support for the welfare state. Immigrant presence should have the strongest effects for those with exclusive national group boundaries; weaker for those with conditionally inclusive boundaries based on reciprocity; and weakest or non-existent for those with inclusive group boundaries. Group boundaries should interact with material self-interest, leading individuals with less material security who are more likely to face social risks to be more supportive of the welfare state. Using data from the 4th European Social Survey linked to regional and national data, we find that group boundary salience plays a large moderating role in the relationship between immigration and native support for the welfare state, and that this role is intricately linked to material self-interest. Group dynamics should therefore be viewed in conjunction with existing structural welfare state theories as opposed to an alternative or isolated mechanism.

Keywords

Immigration welfare state policy preferences group boundaries material self-interest Western Europe

Welfare states and immigration

The national welfare states of Western European countries are often a source of citizen pride (Evans and Kelley, 2002). The relative success of these societies at staving off poverty and inequality is difficult to deny (Brady, 2009; Kenworthy, 2004), offering model systems for countries seeking European Union membership (Pierson and Leibfried, 1995). Yet their sustainability potentially wanes under permanent austerity and the dramatic rise of immigration. Western European countries are now more ethnically and culturally diverse than at any point in their respective histories (Castles and Miller, 2003). Rising immigration coincides with a peak and decline in overall welfare spending across these countries, most notably in ideal-typical Sweden (OECD, 2012). Immigration may be a catalyst for this.

The burning question is whether immigration reduces commitment to welfare policies in Europe as does racial/ethnic diversity in the United States (Fox, 2004; Luttmer, 2001). Some Swedish studies suggest this is the case (Dahlberg et al., 2012; Eger, 2010). However, a number of cross-national studies suggest that it is not (Brady and Finnigan, 2014; Hjerm and Schnabel, 2012; Mau and Burkhardt, 2009). Some even find that immigration increases support among certain individuals (Burgoon et al., 2012). These ambiguous results call for further consideration of this topic.

Previous research identifies two general modes for immigration to shape support for the welfare state, as discussed in the following sections. One is the theory of group dynamics, suggesting that immigrants threaten and in turn activate salience of group boundaries leading to reduced native willingness to distribute resources. Another is that the need for material security motivates support for the welfare state. We argue that these two theoretical propositions most likely interact. Thus, support for the welfare state depends on how individuals define group boundaries and their material securities.

The salience of group boundaries

There are unprecedented levels of foreign-born persons in Western Europe. This may lead to an enhanced native-born awareness of group boundaries existing between themselves and various immigrant groups, which may in turn spur natives to change or re-orient policy preferences to defend what they perceive as their native group’s own resources, or as a means of abandoning solidarity with a national group that does not fit their preferred ideal composition. This theoretical framework of group dynamics is often invoked to explain immigration and political attitudes (Finseraas, 2012; Mau and Burkhardt, 2009).

According to permutations of social identity theory, individuals categorize the world into social groups and identify to which they belong (Allport, 1954; Festinger, 1954; Tajfel, 1970). Belonging requires defining different groups based on group composition (see review by Lamont and Molnár, 2002). The increased presence of out-group members motivates in-group members to clearly define and defend who is in versus out of the group (Tajfel and Turner, 1979; Turner, 1975).

In-group members prefer to allocate their perceived in-group resources to their own subjectively defined group. When this is not possible, in-group members may abandon group sharing to the detriment of both subjectively defined in- and out-group members (Brewer, 1979; Mullen et al., 1992; Tajfel, 1982; Tajfel et al., 1971). This logic applies to immigrant presence and welfare state distribution in Western Europe. If natives subjectively exclude immigrants from their in-groups then theoretically they will stop supporting the welfare state for everyone as either a way to prevent resource movement toward the out-group of immigrants whom tend to be net beneficiaries as opposed to contributors (Boeri, 2010), or as a lack of solidarity with a group that does not fit their subjective ideals (discussion in Bay and Pedersen, 2006: 422).

Native-born is a broad category but it implies a birth-right attachment to a national group whose boundaries give a sense of individual belonging in a world of groups competing for legitimacy, power, and resources (Brubaker, 1996). Natives tend to percieve themselves as being entitled to distributive justice and welfare protection as members of a subjective national in-group (Arts and Gelissen, 2001; Taylor-Gooby, 2011). National group boundaries often subordinate other potentially immigrant-salient boundaries such as family, class, religion, and ethnicity due to the overwhelming symbolic-rhetorical, legal, and militaristic power of nation states (Kaufmann, 2008; Renan, 1882). The national level is also where the largest share of welfare state resources are collected and distributed.

We thus focus on national subjective group boundaries. There are at least two types of exclusive group boundaries, usually drawn along ethnic and civic, or solely civic dimensions (Putnam, 2007; Reeskens and Wright, 2013) and specifying either full or partial exclusion of immigrants. Often those completely exclusive persons see immigrants as not at all ethnically or civically fitting with their subjective national in-group, whereas conditionally exclusive persons focus on immigrants’ abilities to act according to national group maxims regardless of ethnicity. A third, opposite type is persons who are inclusive. For example, some individuals define group boundaries as (supra)national in character, potentially including all human beings (Habermas, 2003; Mau et al., 2008). Although simplified, this tripartite typology offers exemplary differences in subjective group boundary definitions among natives useful for studying group dynamics. Under conditions of greater immigrant presence, the degree of exclusivity of individuals’ group boundaries should shape their attitudes (Brewer, 1979; Hogg and Turner, 1987; Tajfel et al., 1971).

Thus we derive hypotheses based on these three ideal-types. Those who are strictly exclusive should have intense in-group bias when there is greater immigrant presence within their national boundaries, similar to an ethno-civic national identity. Those who are conditionally exclusive should have increased in-group bias but not as dramatic, similar to a civic national identity. Finally, natives who define immigrants as already in their in-groups without reservation have inclusive boundaries, not unlike cosmopolitans who are open to a globally or supra-nationally defined group. They should have boundaries that are not salient or even non-existent under greater national immigrant presence. Dividing the population into these three broad types of natives allows us to frame group boundaries as altering the hypothesized effect of immigration on support for the welfare state (i.e. moderating this effect) leading to the following:

H1: Exclusive. The negative impact of immigrant presence on support for the welfare state will be largest for individuals with exclusive group boundaries.

H2: Conditionally Inclusive. The negative impact of immigrant presence on support for the welfare state will be moderate for individuals with conditionally inclusive group boundaries.

H3: Inclusive. Immigrant presence has no effect on support for the welfare state among individuals with inclusive group boundaries.

The salience of group boundaries should also moderate the impact of immigration at the regional-level where the opportunity for intergroup contact increases. Proximate immigrant presence in a native’s region should increase the salience of group boundaries due to increased opportunity for group comparison (Reicher, 2004). The sub-national level is also desirable as a testing area because it yields a greater number of geographic cases to analyze, and has proven fruitful elsewhere (Eger, 2010; Hjerm, 2009). Therefore, we expect H1 through H3 to apply to the regional-level.

Material self-interest

As long demonstrated in welfare state research, attitudes are highly contingent upon individual access to material security (Dallinger, 2010; Hasenfeld and Rafferty, 1989). Less access to materials equates with higher social risk and more to gain from welfare, while greater access to materials means more to lose (e.g. in taxes). This material self-interest hypothesis posits that those lower in socioeconomic status (SES) are more supportive of the welfare state (Bean and Papadakis, 1998; Breznau, 2010; Jæger, 2006; Svallfors, 1997).

Individuals having much less, such as fending for the most basic needs of food and shelter, should face trade-offs with subjective group boundary salience. Those with exclusive boundaries might need welfare state protection for survival and thus cannot afford to reduce support merely to punish immigrants or as an expression of lost national group solidarity. This conflict is more difficult to reconcile into a coherent set of policy preferences for those lower in SES (Bobo and Licari, 1989) who often lack political capital (Bourdieu, 1984). A Norwegian survey-experiment demonstrates this, in which respondents were asked about minimum income protection and then asked again if they still support minimum income when it applies equally to non-Norwegian immigrants. The result was that the lower the respondents were in SES the more likely they were to change an initial supportive attitude into an unsupportive attitude (Bay and Pedersen, 2006), whereas those higher in education tended to stick to their initial policy positions. Thus, material security is a kind of trump card against the impact of immigration:

H4a: The negative impact of socioeconomic status on support for the welfare state will be largest for individuals with exclusive group boundaries.

However, for those who are less at material risk, the impact of immigration should be greatest (i.e. uninterrupted by need-based self-interest). Given our three subjective boundaries, we would expect this great impact to only exist for those who already define immigrants outside of their group. Thus:

H4b: Among those with exclusive group boundaries, the negative impact of immigrant presence on support for the welfare state will be largest for individuals with higher socioeconomic statuses.

Material self-interest extends to social context as well as individual conditions. We expect regions that are more economically depressed to have greater support for the welfare state net of individual SES. Labor markets are also organized regionally, and when immigrant presence increases, competition for jobs often follows, leading to a destabilization of wages and employment (Nannestad, 2007). Burgoon et al. (2012) demonstrate this phenomenon, where a greater share of foreign-born in a specific occupation leads the workers to be more supportive of the welfare state out of need.

H5: Individuals in economically depressed regions are more supportive of the welfare state.

Because of H1 we expect this effect to be particularly pronounced for those with exclusive group boundaries because they should otherwise have the lowest support under greater immigrant presence, such that they will have the largest trade-off between defending group boundaries and material needs.

Data and methods

To test our hypotheses, we utilize data from the 4th European Social Survey (ESS4) fielded between 2007–2010, with a welfare state module and representative regional-level samples. We use 14 Western European countries in the survey, with Germany split into two, for a total of 24,604 individuals in 105 regions within 15 societies¹; we refer to these societies as “countries” for ease of discussion.

Measuring support of the welfare state

We measure our dependent variable support for the welfare state using six ESS4 questions asking about the role of the government in providing for basic needs and insurance against social risks, e.g. health care, pensions, and unemployment. These items have theoretical face value, and form one major latent welfare attitude (Roosma et al., 2014). We use a second dimension of welfare policy outcomes as an independent variable measured from five ESS4 questions that measure quality assessment of current welfare provisions.

We do not assume that a group of welfare state-related questions measure one latent attitude because, first, welfare attitudes are shown to be multidimensional (van Oorschot and Meuleman, 2012). Second, there is known to be a bias in support for the welfare state by individual evaluations of its current performance (Roosma et al., 2013). This is an argument of political attitude theories where an individual’s absolute preferences for government welfare provisions are biased by current levels of provisions (Blekesaune and Quadagno, 2003; Wlezien, 1995). Thus, support for the welfare state and welfare policy outcomes are captured in a structural measurement model to correct this bias to some degree following the concepts outlined in Roosma et al. (2013). Our descriptives and multilevel linear models refer only to predicted factors taken from the results of the measurement model whose structural loadings are presented in Table 1.

Table 1.

Support for the Welfare State Measurement Items, ESS4.

Observed Variables	N	Mean	SD	Min(by)Max	Factor Loadings^a
Role of government to:					Support of the Welfare State	Welfare Policy Outcomes
Ensure adequate health care	26,023	8.49	1.65	0(1)10	0.586	−0.119
Ensure standard of living for old	26,016	8.19	1.71	0(1)10	0.730	−0.030
Ensure jobs for all	25,827	6.34	2.56	0(1)10	0.767	−0.079
Ensure standard of living for unemployed	25,888	6.80	2.15	0(1)10	0.656	−0.083
Ensure child care for workers	25,703	7.61	2.07	0(1)10	0.652	−0.064
Provide paid care leave	25,714	7.44	2.12	0(1)10	0.653	−0.035
Quality of welfare in:
Health care	25,920	5.78	2.39	0(1)10	0.047	0.501
Standard of living for pensioners	25,800	5.13	2.15	0(1)10	0.117	0.620
Job opportunies for young workers	25,576	4.73	2.30	0(1)10	0.131	0.609
Standard of living for unemployed	25,494	4.30	1.95	0(1)10	−0.047	0.531
Child care for workers	23,675	5.08	2.31	0(1)10	0.077	0.600
Latent Policy Preferences
Support for the Welfare State^b	26,060	0	1	−4.92(.)1.67	1.000	−0.171
Welfare Policy Outcomes^b	26,060	0	1	−3.35(.)3.31	−0.171	1.000

^aFactor loadings from exploratory measurement model; bottom two rows are the correlations between predicted factor scores.

^bPredicted latent variables from a confirmatory structural measurement model, FIML & mean replacement. Roosma, van Oorscot and Gelissen (2014) use this same ESS4 data to demonstrate that these two factors are scalar invariant across our same countries except Greece. We find that they are also scalar invariant across our groups in Table 2 (see author appendix https://sites.google.com/site/nbreznau/)).

Measuring immigrant presence

We identify immigrant presence as rates of foreign-born persons by country and by sub-national region. Immigrant presence increases the visibility of immigrants directly in an individual’s region and indirectly via the public sphere (Brader et al., 2008; Price, 1989), thus media and political discussion should lead to an impact of country-level immigrant presence for individuals who live in areas where there are none, few, or many immigrants alike. Our country-level measure of percent foreign-born comes from the OECD (2009), and we interpolate the division between East and West German data using regional measures.

We construct our own cross-regional database of percent foreign-born along NUTS (Nomenclature of Territorial Units for Statistics) divisions within countries.² The ESS representatively sampled these regions. Although there are other potential measures of immigrant presence such as percent foreign, ethnic fractionalization, or percent Muslim, for example, it is first generation immigrants of any source that are the most likely to appear, behave, and speak differently than natives. Furthermore, this is the only directly comparable regional measure available.

Measuring group boundaries

Investigating immigration impacting support for the welfare state via group boundary salience requires identification of the specific subjective boundaries that each individual uses to define the national in-group and its resources. To do this we use the ESS4 question asking if and when an immigrant should get the same social benefits as natives (see Table 2). At one extreme we identify group boundaries that are inclusive of immigrants. We use either the first or second answer categories as evidence of inclusive boundaries because these individuals prefer to extend social rights immediately or without reciprocity from immigrants. At the other extreme we identify exclusive boundaries characterizing those who would never extend social benefits to immigrants based on the second-to-last category in Table 2. They define their subjective national in-group boundaries as closed to non-natives. The other two middle categories identify those with group boundaries that are conditionally inclusive based on reciprocity of immigrants. Some answered “don’t know” to this question, but the case numbers are few so we dropped them.

Table 2.

Measuring group boundaries through the European Social Survey (ESS4) question on social rights for immigrants.

Thinking of people coming to live in [country] from other countries, when do you think they should obtain the same rights to social benefits and services as citizens already living here?
Response	Group basis	N	Group boundary definition
Immediately on arrival	Inclusive	2,596	Immigrants granted access to in-group benefits and services with or without contributing; group boundary open
After living in [country] for a year, whether or not they have worked	Inclusive	2,885
Only after they have worked and paid taxes for at least a year	Conditionally Inclusive	11,753	Immigrants granted access conditional upon material or civic contribution to the in-group; boundary fickle
Once they have become a [country] citizen	Conditionally Inclusive	8,419
They should never get the same rights	Exclusive	1,718	Immigrants excluded from in-group benefits and services; boundary closed
Don’t know	Unknown	788	Respondent does not define boundary

ESS4, 15 countries surveyed in Western Europe.

As the group boundaries question refers to social benefits, it may be endogenous with our dependent variable and our independent variable welfare policy outcomes, i.e. that different types of welfare attitudes are also expressed in response to the group boundary question. In order to check that this is not the case we test the measurement of our dependent variable separately for each type of group boundary; put in statistical terms, we look for measurement invariance by group (van de Schoot et al., 2012). To do this we first run a model where all parameters are free to vary by group. This model sets a comparative baseline fit.³ We then constrain the freely estimated factor loadings each to be equal across the three groups, leaving the means and variances to vary. This model fits slightly better. Finally, we fix all parameters to be equal across all groups, and this model does not fit well at all. Given that we have scalar invariance across groups, we conclude that individuals in these three groups have similarly constructed attitudes toward the welfare state and this construct does not pattern or correlate with these group boundaries, thus we have evidence against endogeneity. However, if there is endogeneity, then this should reduce the impact of our immigration variable due to collinearity between it and categorical group boundaries. If anything, this should make an impact of immigrant presence harder to identify and our tests more conservative.

Measuring material self-interest

Education, occupation, and income are the typical markers of SES. However, they are difficult to disentangle even when we have thousands of cases. We measure education in years, and proxy class via a linear occupational status measure (Kelley, 1990) plus dichotomous variables for supervisor of workers and capitalist owners with employees. We take these measures despite their collinearity, and do not argue for priority of one over others. Essentially, all are markers of material self-interest, and we do not seek to uncover exactly what drives this. We omit income as it introduces a large missing data problem, and it probably is not missing at random leaving us with a more accurate chance of measuring socioeconomic positions of individuals using occupation and education. With 28% missing data on income in the ESS4, imputation likely will attenuate the impact of income and inflate the impact of other SES and demographic regressors (Landerman et al., 1997).

Finally, we control for individuals who are vulnerable to social risks because they were unemployed or unable to work due to sickness in the last seven days, and thus have self-interest in supporting the welfare state (Pfeifer, 2009).

Independent variables

In addition to SES, demographics are the other most common explanatory variables for welfare state attitudes (Arts and Gelissen, 2001; Bean and Papadakis, 1998; Svallfors, 1997; Taylor-Gooby, 2001). Based on this research we use the following essential controls. Females are more supportive on average. Older and younger persons are more supportive alike and thus we take age and age-squared.⁴ We also utilize subjective left–right to control for the known overlap between partisan or political ideology and welfare policy preferences (Huber, 1989). At the regional and country-levels, we measure gross domestic product (GDP) per capita at parity and the employment rate as a percent of the labor force to identify the material resources necessary to provide, and potential need of, welfare across regions and countries (Eurostat, 2014; OECD, 2009).

Table 3.

Descriptive statistics for all variables analyzed in the multilevel linear models and by group.

Individual-level				Overall		Exclusive				Inclusive
				N=24,604		N=1,621		N=18,253		N=4,730
Variable	Measure	Min	Max	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Support for the welfare state^a	See Table 1	–4.92	1.70	–0.01	1.00	0.04	1.24	–0.04	0.98	0.08	0.98
Female	(Male)	0	1	0.52	0.50	0.54	0.50	0.51	0.50	0.53	0.50
Age	In years	15	85	47.74	17.90	50.13	17.76	48.01	17.98	45.86	17.47
Subjective left–right^b	Political orientation	0	10	5.00	1.97	5.35	2.10	5.11	1.93	4.47	2.00
Welfare policy outcomes^b	See Table 1	–4.92	1.70	–0.01	1.00	0.04	1.24	–0.04	0.98	0.08	0.98
Education^b	In years	0	23	12.70	3.95	10.99	3.63	12.66	3.89	13.42	4.10
Occupational status^b	Linear (Kelley, 1990)	0	10	4.59	2.49	3.52	2.06	4.58	2.47	4.99	2.58
Capitalist	Self-emp. w/ employees	0	1	0.05	0.21	0.05	0.22	0.05	0.22	0.04	0.19
Supervisor	Supervises others on the job	0	1	0.30	0.46	0.23	0.42	0.31	0.46	0.30	0.46
Vulnerable	Sick or unemp. past 7 days	0	1	0.29	0.45	0.38	0.48	0.29	0.46	0.25	0.43

Regional-level^c	N=105
Foreign-born	%	1.98	37.16
GDP	per capita; in €1,000	14.90	120.60
Employment rate	employed as % of LF	53.97	81.63
Country-level	N=15
Foreign-born	%	4.10	26.90
GDP	per capita; in €1,000	20.96	87.20
Employment rate	employed as % of LF	58.77	79.70

ESS4, 15 countries surveyed in Western Europe, N restricted to those non-missing on the dependent and demographic variables.

^aMeasurement models suggest that across the three groups, loadings are invariant while means vary. This suggests that structural loadings (i.e. individual understandings of the questions) for support of the welfare state and welfare policy outcomes are not endogenous with group boundaries. For measurement of groups, see Table 2.

^bImputed using stochastic single hierarchical linear imputation to preserve cases; imputed 0.5% for education, 7.8% subjective left–right, and 8.6% occupational status.

^cSix regions were dropped due to lack of observations for the Exclusive group.

GDP: gross domestic product; LF: labor force.

Methods

We use a three-level general linear regression model to predict support for the welfare state. We see group boundaries as moderating the impact of immigrant presence on welfare state attitudes (H1–H3) and group boundaries as interacting with SES at both the group level (H4a) and at the country-level (H4b). We run this as a multigroup model, taking the focus away from comparison with an overall population mean and interaction terms, to a regression to the mean of each group with fixed and varying coefficients by group. We do, however, employ a cross-level interaction with education and country-level foreign-born.

All individual-level variables are grand mean centered. For regional and country-level variables we standardize x-variables at their respective levels, which allows us to test for the relative impact of immigrant presence, GDP, and employment rate (Andreß et al., 2013). As long noted by social researchers, both the salience of group boundaries and the effect of material self-interest are subject to social comparison. Individuals compare their current situations to the situations of others within their contexts (Festinger, 1954). It is important to note that standardized measures may only be interpreted in comparison to the variance at each level, so a net effect of 0.50 for a regional-level variable is large, but the gross size is relative to the amount of the total variance at that level which we will show is small (>4%) in the case of the regional level (Hox, 1995). Our main goal is not to determine what explains the most, but to test for significant effects.

If group boundaries moderate the effects of independent variables, we expect that the addition of our three group boundaries as parameters in our multilevel regression leads to a better empirical model of the dependent variable. This can be tested with the chi-square difference test, with the model selection criteria of AIC offering a measure of the trade-off between fit and complexity, and with slope tests comparing the coefficients and standard errors (like a t-test) between groups to determine if slopes are significantly different (Baron and Kenny, 1986; Pedhazur, 1997).

If the salience of group boundaries explains the relationship between immigration and attitudes, we should see that the effect of immigrant presence will be greatest for those whose group boundaries are most exclusive of immigrants; i.e. the category we label exclusive. If the effect of immigration on welfare attitudes does not differ across groups, then the theoretical mechanism of group boundary salience does not appear responsible for the relationship. We can test this by running models in sequences that add variance-components and variables in stages and then compare them following this order:

M0. Empty-model with intercepts (means) allowed for individuals, regions, and countries.

M1. Intercepts and errors in M0 are allowed to vary across our three groups.

M2. All individual-level variables added to M1 but constrained to equal coefficients (effects) across groups.

M3. Regional and country foreign-born, GDP, and unemployment added to M2 constrained across groups.

M4. Percent foreign-born, GDP and unemployment vary by group. Formally tests H1, H2, H3, and H5.

M5. The coefficients for individual self-interest variables vary by group. Formally tests H4a.

M6. An interaction within each group for education and country foreign-born. Formally tests H4b.

Results

Table 4 demonstrates the size and significance of standardized independent variable coefficients predicting our latent measure of support for the welfare state by group. The exclusive category is presented in the first four columns, conditionally inclusive next, and inclusive last.

Table 4.

Multigroup, multilevel linear models predicting support for the welfare state.

Group boundary	Exclusive				Conditionally inclusive				Inclusive
Variable^a	M3	M4	M5	M6	M3	M4	M5	M6	M3	M4	M5	M6
Education	–0.01***	–0.01***	–0.03***	–0.04***	–0.01***	–0.01***	–0.01***	–0.01***	–0.01***	–0.01***	0.00	0.00
	0.00	0.00	0.01	0.01	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Occupational status	–0.02***	–0.02***	–0.03*	–0.03*	–0.02***	–0.02***	–0.02***	–0.02***	–0.02***	–0.02***	0.00	0.00
	0.00	0.00	0.01	0.01	0.00	0.00	0.00	0.00	0.00	0.00	0.01	0.01
Capitalist	–0.12***	–0.12***	0.07	0.08	–0.12***	–0.12***	–0.15***	–0.15***	–0.12***	–0.12***	–0.04	–0.04
	0.03	0.03	0.11	0.11	0.03	0.03	0.03	0.03	0.03	0.03	0.07	0.07
Supervisor	–0.03*	–0.03*	–0.02	–0.02	–0.03*	–0.03*	–0.02	–0.02	–0.03*	–0.03*	–0.07*	–0.07*
	0.01	0.01	0.06	0.06	0.01	0.01	0.02	0.02	0.01	0.01	0.03	0.03
Vulnerable	0.08***	0.08***	–0.02	–0.01	0.08***	0.08***	0.09***	0.09***	0.08***	0.08***	0.06	0.06
	0.02	0.02	0.05	0.05	0.02	0.02	0.02	0.02	0.02	0.02	0.03	0.03
Group intercept	–0.22	–0.24	–0.26	–0.26	–0.10	–0.11	–0.11	–0.11	0.00	0.00	0.00	0.00
Regional foreign-born	–0.08**	0.06	0.07	0.07	–0.08**	–0.08**	–0.08**	–0.08**	–0.12***	–0.12***	–0.13***	–0.13***
	0.03	0.05	0.05	0.05	0.03	0.03	0.03	0.03	0.03	0.03	0.03	0.03
Regional GDP	0.04	–0.10*	–0.11*	–0.12*	0.04	0.03	0.03	0.04	0.11**	0.11**	0.10**	0.11**
	0.03	0.05	0.05	0.05	0.03	0.03	0.03	0.03	0.03	0.03	0.03	0.03
Regional employment rate	–0.04*	–0.01	0.00	0.00	–0.04*	–0.03	–0.03	–0.04*	–0.08**	–0.08**	–0.08***	–0.08**
	0.02	0.03	0.03	0.03	0.02	0.02	0.02	0.02	0.02	0.02	0.02	0.02
Country foreign-born	–0.20***	–0.40***	–0.39***	–0.45***	–0.20***	–0.19***	–0.19**	–0.19**	–0.20**	–0.20**	–0.20**	–0.20**
	0.07	0.08	0.08	0.08	0.07	0.07	0.07	0.07	0.07	0.07	0.07	0.07
Country GDP	0.09	0.05	0.06	0.08	0.09	0.10	0.10	0.10	0.07	0.07	0.07	0.07
	0.08	0.09	0.09	0.09	0.08	0.08	0.08	0.08	0.08	0.08	0.08	0.08
Country employment rate	–0.14	–0.13	–0.12	–0.13	–0.14	–0.16	–0.16	–0.16	–0.08	–0.08	–0.08	–0.08
	0.09	0.09	0.09	0.09	0.09	0.09	0.09	0.09	0.09	0.09	0.09	0.09
Country foreign-born *individual education				–0.03***				0.00				0.01*
Country foreign-born *individual education				0.01				0.00				0.00

*p<0.05, **p<0.01, ***p<0.001; individual variables grandmean centered; regional variables are within-country-mean centered and then divided by within-country SD; country-variables centered and divided by across-country SD. Coefficients in standard typeface are constrained to be equal across groups; coefficients in bold are allowed to vary. Models 0, 1 and 2 omitted to save space (for fit statistics see Table 5).

^aAs they are intended to be control variables we do not show results for the variables age, age², subjective left–right, and welfare policy outcomes in order to save space; these variables are present in Models 3–6.

GDP: gross domestic product.

In Table 4 we omit Models 0–2 but report their fit statistics as baselines for comparison in Table 5. Also, we do not report demographic and control variable effects to keep the focus on our hypotheses. We first see the impact of SES starting with M3. For every year of education, individuals across all groups are –0.01 standard deviations less supportive of welfare. Those who are one point higher (out of 10) in their occupational status are –0.02 less supportive. Those who are capitalist are –0.12 less, and so on. All effects in this and the following models may be understood in this way as the dependent variable is also standardized. Generally, those who have less material security due to lower education, income, occupation status, or current employment status (vulnerable) are more supportive of welfare. M3 also gives the first evidence of regional and country-level effects. For all groups, in regions where there is one standard deviation more immigrants than the country’s average, individuals are –0.08 individual-level standard deviations less supportive of the welfare state; this might be labeled a large effect: if we extrapolate this impact to the regional-level it explains over half of regional variance in individual support for the welfare state; however, the overall regional variance is small so relative to the individual or country-level this is not a relatively large effect. Individuals who live in countries that are one standard deviation above the Western European mean on immigration are –0.20 standard deviations less supportive. Measures of regional and national GDP have no impact, but regions with more employment have less support on average.

Table 5.

Model statistics, fit, and slope tests.

Fit statistics	M0	M1	M2	M3	M4				M5				M6
Log-likelihood	–32,720	–32,668	–32,127	–32,114	–32,065				–32,043				–32,033
Chi2 diff. test	***	***	***	***	***				***				***
AIC	65,449	65,348	64,285	64,272	64,199				64,173				64,161
Slope tests					Exclusive (X)^a		Inclusive (I)^a		Exclusive (X)^a		Inclusive (I)^a		Exclusive (X)^a		Inclusive (I)^a
Slope tests					b	p-value by group	b	p-value by group	b	p-value by group	b	p-value by group	b	p-value by group	b	p-value by group
Education										C; I		C**; X	–0.04	C; I*	0.00	C*; X*
Occupational status													–0.03		0.00
Capitalist													0.08	C*	–0.04
Supervisor													–0.02		–0.07
Vulnerable										C*			–0.01		0.06
Regional foreign-born					0.06	C; I*	–0.12	X**	0.07	C; I**	–0.13	X***	0.07	C; I*	–0.13	X***
Regional GDP					–0.10	C; I**	0.11	X***	–0.11	C; I**	0.10	X***	–0.12	C; I*	0.11	X***
Regional employment rate					–0.01	I*	–0.08	X*	0.00	I*	–0.08	X*	0.00	I*	–0.08	X*
Country foreign-born					–0.40	C^†; I^†	–0.20	X^†	–0.39	C^†; I^†	–0.20	X^†	–0.45	C; I	–0.20	X*
Country GDP					0.05		0.07		0.06		0.07		0.08		0.07
Country employment rate					–0.13		–0.08		–0.12		–0.08		–0.13		–0.08
Residual variance
Individual	0.828	0.824	0.789	0.789		0.786				0.782				0.781
Regional	0.033	0.034	0.030	0.025		0.024				0.025				0.025
Country	0.147	0.149	0.133	0.058		0.055				0.066				0.067

^aThe conditionally exclusive category is not mentioned due to redundancy. All slope tests between the three groups is captured in these two groups’ test statistics.

^† p<0.10 (only for country-level), *p<0.05, **p<0.01, ***p<0.001; all models have individual N=24,604, regional N=105, and country N=15.

GDP: gross domestic product; C: conditionally inclusive.

Next we test group dynamics. M4 shows whether the effect of immigration is greater for those whose group boundaries should become more salient under increased immigrant presence (H1–H3). Our first result counters our hypotheses. The impact of regional percent foreign-born is not significant for the exclusive category. There is a negative and significant impact for both conditionally inclusive and inclusive to the degree of –0.08 and –0.12. However, at the country-level the impact of percent foreign-born confirms our predictions for the exclusive category with a very large –0.40 coefficient. Thus, in a country with one standard deviation higher immigrant presence, residents are nearly half a standard deviation less supportive of the welfare state, independent of GDP and regional and individual effects. The size of this effect is less for the conditionally inclusive category at –0.19 and for inclusive at –0.20. We also test H5 in this model, predicting that higher regional GDP or lower employment rates will lead to lower support for the exclusive category. GDP does have the expected effect at the regional-level but is not significant at the country-level, and employment rate is suggestive of this effect at the country-level at –0.12, but is not significant. In a model without GDP, employment rate is significant as they are strongly correlated like all measures of development (Breznau et al., 2011), but we prefer to have both in the model to theoretically control for ability to provide welfare (GDP) and direct need of welfare (employment rate).

M5 allows individual-level SES measures to vary by group offering a test of H4a. We find that material security captured by education and occupation has the largest negative impact on support for the welfare state in the exclusive category, although these also have an impact for the conditionally inclusive category; along with a much larger effect for capitalists being –0.15 less supportive. There is very little evidence of an SES effect in the inclusive category. Our preferred model, M6, adds an interaction between education and percent foreign-born at the country-level. We see that this interaction is highly significant for the exclusive group. This supports H4b, suggesting that those higher in SES do not face a material security trade-off with their subjective group boundaries and are far less supportive of the welfare state where there are more immigrants; however, this is only true at the country-level. We tested the regional-level and it was insignificant across all groups. Interestingly, we see a small but significant positive effect of this interaction for the inclusive group.

Table 5 gives model statistics. Based on the chi-square and AIC statistics, models M0–M6 each fit relatively better than the previous, suggesting that our tests are robust and our multigroup approach meaningful. However, in order to ensure that there is variation across groups we conducted slope analyses comparing the coefficients between each category of group boundary. We find that when coefficients vary (M4–M6), the impact of percent foreign-born is significantly different for the exclusive category compared with both the conditionally inclusive (“C” in the table) and inclusive (“I”) categories at both the regional and country-levels. Meanwhile, its impact for the inclusive category is only significantly different from the exclusive category (“X”). Thus, immigrant presence by region and country is not significantly different in impact between conditionally inclusive and inclusive. GDP follows a similar pattern but only at the regional-level, and the effect of employment is only different between the exclusive and inclusive groups at the regional-level.⁵

We combine into a sheaf the derivative effects of education, occupational status, capitalist, supervisor, vulnerable, GDP, and employment in Figure 1. This provides a sense of effect sizes and allows comparison across groups. It highlights the dramatic effect predicted by H4b and shows how material self-interest and immigrant presence can work together against support for the welfare state.

Figure 1.

Predicted marginal support for the welfare state by group boundary, immigrant presence and socioeconomic status, ESS4.

Predicted margins at one standard deviation above and below the mean or 0 and 1 for dummy-variables show that the largest impact of SES alone occurs for the conditionally inclusive type, where almost half of a standard deviation in support for the welfare state exists between low and high SES persons; this effect is a tiny bit smaller for the exclusive group and does not exist at all for the inclusive group. The largest impact of immigrant presence alone occurs for the exclusive group, where a 0.7 standard deviations gap exists between those living in low regional and country-level immigrant presence versus high. However, roughly 0.6 standard deviations in support for the welfare state occurs in the inclusive group, contrary to what we hypothesized. Looking at the combined impact of immigrant presence and SES reveals that the exclusive and conditionally inclusive categories have a huge combined effect of about one standard deviation in welfare state support when comparing those with low SES and low immigrant presence to those with high and high, respectively. This is smaller for the inclusive category at roughly 0.7 standard deviations.

Discussion

We delineated natives’ subjective group boundaries into three types in most Western European countries. We identify individuals who prefer to have exclusive national in-group boundaries with respect to immigrants, those who will be conditionally inclusive of immigrants only if immigrants contribute to the state and society, and those who are unconditionally inclusive of immigrants within their national in-group boundaries. We identified that support for the welfare state is not a different attitude construct across these three groups based on tests of measurement invariance.⁶ Using group dynamics and social identity theories, we predicted that group boundary salience produced by increasing immigrant presence would be the highest for the exclusive type (H1). This higher salience should lead to a reduction in willingness to share group resources measured via support for the welfare state. Here we find that, with respect to welfare state resources, our hypotheses are true at the country-level. Countries with greater immigrant presence lead those in the exclusive category to be less supportive of the welfare state. This impact exists independent of controls for sex, age, political ideology, perceptions of current welfare policy outcomes, and regional and country-level prosperity and employment. The impact of increased immigrant presence at the country-level is smaller for those in the conditionally inclusive and inclusive groups – partial support of H2 and H3. The impact is significantly less but does not disappear altogether. Thus, group boundaries do not entirely explain the effect of immigrant presence on support for the welfare state. Immigration still has an impact for the inclusive group, suggesting that their welfare state support may be negatively impacted by immigration. We speculate that their reported openness to immigrants may not be so agape. Also, there may be other pathways by which immigration indirectly reduces welfare state support. This should be probed in future research.

At the regional-level we found ostensibly counter-intuitive evidence. Immigrant presence has no effect for those with exclusive group boundaries. It was only significant for conditionally inclusive and inclusive groups. Although we would like to revise our hypotheses, we were honestly surprised. We speculate, however, that those with the most exclusive subjective group boundaries are most affected by immigration at the country-level. Our findings point toward this with the largest impact of immigrant presence on support for the welfare state occurring for this exclusive category at the country-level, with a standardized effect of –0.40 (H4b). Our finding of an interaction between individual SES and immigrant presence suggests that this speculation has merit. The exclusive group under greater country-level immigrant presence is less supportive than others with less subjectively restrictive group boundaries. However, this is not true for the least educated (lowest SES) members of this exclusive group, who likely struggle with a kind of trade-off between self-interest and group boundaries (e.g. Bay and Pedersen, 2006). We tried a similar interaction with regional-level immigrant presence and it produced insignificant results. Thus, the significant interaction points toward the relevance of country-level immigrant presence as most salient for individuals with exclusive subjective boundaries. In addition to a Bayesian credible interval robustness check, the interaction gives support to country-level foreign-born individuals, revealing a true impact of immigrant presence, and not some luck given that we had a small-N problem (Breznau, 2016; Stegmueller, 2013). Thus, we cautiously conclude that country-level immigrant presence has a large impact on reducing suport for the welfare state, and that this varies by subjective group boundary and SES for exclusive group boundaries.

A surprising non-effect of immigrant presence for the exclusive category at the regional-level leads us to speculate that the welfare state attitudes of those who have subjectively exclusive group boundaries are not strongly impacted by intergroup contact, and are only impacted by national-level trends. This claim is supported by strong anti-immigrant movements such as PEGIDA in Germany, which started in Dresden where there are far less immigrants than the average in the country’s other major cities. Other scholarship might help shed light on this finding. Shayo (2009) claims that the degree of national identity will decrease support for policies that target a specific class, due to national group solidarity (i.e. no one in the group should get special treatment). It is possible that more foreign-born can increase or decrease this degree (nationalism), leading to a shift toward policy favoring working-class groups among those lower in SES; a claim potentially supported by our interaction of education and country-level foreign-born but not further testable in these models. Ultimately, there are likely to be simultaneously conflicting effects especially given that a dualist or chauvinist welfare state is often not possible in practice although it is ideally preferred by many natives with exclusive group boundaries (Bay et al., 2013).

Our results also support a long tradition of research on material self-interest associated with support for the welfare state (H4a; H5). However, we see that material self-interest is highest for those in the exclusive and conditionally inclusive categories, i.e. the non-inclusive. We speculate that their group boundary definitions are in conflict with their need for welfare and this leads to a larger impact of SES. The desire to restrict the welfare state for everyone crumbles in the face of a desire to protect individual, if not familial, welfare security. On the contrary, inclusive boundaries may indicate post-materialist values (Inglehart and Flanagan, 1987) or cosmopolitanism (Habermas, 2003), and such values may be steadfast despite material well-being. This might explain why those with inclusive boundaries show very little impact of SES variables on support for the welfare state. Moreover, inclusive boundaries want to reduce resource sharing under greater regional immigrant presence, suggesting that their redistributive attitudes might be susceptible to change given contact. Perhaps the most inclusive and open attitudes belong to persons who are more gated from experiencing contact with immigrants. They may change their support for the welfare state when immigrant presence arrives in their neighborhoods as easily as the exclusive boundary types who experience material hardship.

Theoretically, our analysis provides an empirical application of group boundary salience leading to in-group bias. Although we see our findings as supporting this theoretically causal mechanism, we do not wish to overstate causality in strict terms (Pearl, 2009). Our evidence is nonetheless supportive of what we see as sound theoretical reasoning: the group boundary mechanism might be as powerful as a material self-interest in shaping support for the welfare state. Furthermore, self-interest is especially tricky when it comes to those with subjectively exclusive boundaries who face a trade-off between material security and preventing redistribution to or acceptance of immigrants. So long as those with exclusive subjective group boundaries remain a marginal slice of the overall European welfare states’ populations, we do not expect that support for the welfare state will dissolve given continued increases in immigration. Nevertheless, the activation of group boundaries may manifest in a variety of forms and holds the ominous potential to undermine the welfare state among the non-exclusives as well, as evidenced by the sudden and surprising 2016 closing of European borders in reaction to the Syrian refugee crisis.

Our results contribute to theories of welfare state retrenchment. Individuals in Western Europe live in countries consisting of anywhere from 4–28% foreign-born immigrants and, within those countries, reside in regions with 2–38% foreign-born. The more immigrants visible or known to be present within natives’ national borders, the more likely that group boundaries become salient and support for the welfare state may wane; as it turns out, this may be true even for those proclaiming to have inclusive group boundaries. After the prosperous “golden era” of European welfare states ending sometime between 1980 and 2000 in most countries, neoliberal agendas and conditions of permanent austerity are retrenching social policies (Pierson, 2002). Welfare state dismantling comes in general at the hands of both right and left parties, despite right parties taking a much stronger oppositional position (Schumacher et al., 2013). Further, mobilizing both welfare chauvinism and anti-immigrant rhetoric (Eger and Valdez, 2015), anti-immigrant parties traditionally form coalitions with right and right-center parties who target welfare, in order to push their own “radical” agendas (Mudde, 2013). With the exceptions of Ireland and Spain, these parties have increased electoral gains throughout Western Europe. Perhaps most surprisingly, in Germany, where populist parties have been off the map since WWII, there were significant gains for the Alternative für Deutschland party, which not only mobilizes populist anti-immigrant rhetoric but explicitly aims to reduce social policy provisions. This is perhaps a manifestation of the societal forces driving our findings herein, which explain a piece of the welfare state retrenchment puzzle at the turn of the millennium. However, these conclusions are tentative because we are limited by the cross-sectional nature of our data and more research is certainly in order. We look forward to the next round of the ESS in 2016 to analyze the impact of changes in immigrant presence over time.

Footnotes

Notes

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Appendix

Table A1.

Regional database construction.

Country	NUTS^a	Geo unit	Count	Year	Sources		Accessed	Mean	SD
Austria	2	Bundesländer	9	2010	Statistics Austria	http://statistik.at	Feb 2014	15.98	7.49
Belgium	1	Gewesten/ Régions	3	2007	Stat Bel	http://statbel.fgov.be	Feb 2014	21.17	18.66
Denmark	2	Regioner	5	2008	Statistics Denmark	www.dst.dk	May 2011	6.23	2.54
Finland	2	Suuralueet	4	2008	Statistics Finland	www.stat.fi	Jun 2011	3.24	1.63
France	1	ZEAT	8	2005	National Institute of Statistics and Economic Studies	www.insee.fr		10.38	5.65
Germany (East)	1	Länder	6	2007	German Microcensus	www.regionalstatistik.de/	Apr 2011	5.89	5.20
Germany (West)	1	Länder	11	2007	German Microcensus	www.regionalstatistik.de/	Apr 2011	14.53	3.10
Greece	2	Perifereies	10	2001^b	Hellenic Statistical Authority	www.statistics.gr	Jun 2011	6.52	2.40
Ireland	3^c	Regional Authorities	5	2005	Ireland’s Central Statistics Office	www.cso.ie	Apr 2011	16.30	2.31
Netherlands	2	Provincies	12	2007	Centraal Bureau voor de Statistiek	http://statline.cbs.nl	Apr 2011	9.20	3.74
Norway	2	Regioner	7	2008	Stat Nord	www.h2.scb.se/grs/	Jun 2011	8.97	4.05
Spain	2	Comunidades y ciudades autónomas	18	2007	Instituto Nacional de Estadística	www.ine.es	Jun 2011	9.45	4.75
Sweden	2	Riksområden	8	2007	Statistics Sweden	www.ssd.scb.se/	Apr 2011	11.50	4.59
Switzerland	2	Regionen	7	2010	Federal Swiss Statistics	www.pxweb.bfs.admin.ch	Apr 2012	27.32	6.84
UK	1	Regions	12	2007	Office for National Statistics	www.statistics.gov.uk	Apr 2011	9.02	7.81

^aNUTS 1: Population = 3–7 million; NUTS 2: Population = 800,000–2.99 million; NUTS 3: Population = 150,000–799,999.

^bGreece only published percent foreign citizen, which is the proportion of the population born abroad excluding immigrants who have naturalized. We impute using the ESS sampling design, which has a near perfect match between the rate of immigrants sampled and their actual rate by region in all the other countries.

^cIreland’s eight units collapsed into five in ESS.

ESS: European Social Survey; NUTS: Nomenclature of Territorial Units for Statistics.

References

Allport

(1954) The Nature of Prejudice. Cambridge, MA: Addison-Wesley.

Alwin

(1989) Problems in the estimation and interpretation of the reliability of survey data. Quality and Quantity 23(3–4): 277–331.

Andreß

H-J

Golsch

Schmidt

(2013) Applied Panel Data Analysis for Economic and Social Surveys. Berlin: Springer.

Arts

Gelissen

(2001) Welfare states, solidarity and justice principles: Does the type really matter? Acta Sociologica 44(4): 283–299.

Baron

Kenny

(1986) The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51(6): 1173–1182.

Bay

A-H

Pedersen

(2006) The limits of social solidarity: Basic income, immigration and the legitimacy of the universal welfare state. Acta Sociologica 49(4): 419–436.

Bay

A-H

Finseraas

Pedersen

(2013) Welfare dualism in two Scandinavian welfare states: Public opinion and party politics. West European Politics 36(1): 199–220.

Bean

Papadakis

(1998) A comparison of mass attitudes towards the welfare state in different institutional regimes, 1985–1990. International Journal of Public Opinion Research 10(3): 211–236.

Blekesaune

Quadagno

(2003) Public attitudes toward welfare state policies: A comparative analysis of 24 nations. European Sociological Review 19(5): 415–427.

10.

Bobo

Licari

(1989) Education and political tolerance: Testing the effects of cognitive sophistication and target group affect. The Public Opinion Quarterly 53(3): 285–308.

11.

Boeri

(2010) Immigration to the land of redistribution. Economica 77(308): 651–687.

12.

Bourdieu

(1984) Distinction: A Social Critique of the Judgement of Taste. London: Routledge.

13.

Brader

Valentino

Suhay

(2008) What triggers public opposition to immigration? Anxiety, group cues, and immigration threat. American Journal of Political Science 52(4): 959–978.

14.

Brady

(2009) Rich Democracies, Poor People. How Politics Explains Poverty. Oxford: Oxford University Press.

15.

Brady

Finnigan

(2014) Does immigration undermine public support for social policy? American Sociological Review 79(1): 17–42.

16.

Brewer

(1979) In-group bias in the minimal intergroup situation: A cognitive–motivational analysis. Psychological Bulletin 86(2): 307–324.

17.

Breznau

(2010) Economic equality and social welfare: Policy preferences in five nations. International Journal of Public Opinion Research 22(4): 458–484.

18.

Breznau

(2016) Secondary observer effects: Idiosyncratic errors in small-N secondary data analysis. International Journal of Social Research Methodology 19(3): 301–318.

19.

Breznau

Lykes

Kelley

. (2011) A clash of civilizations? Preferences for religious political leaders in 86 nations. Journal for the Scientific Study of Religion 50(4): 671–691.

20.

Brubaker

(1996) Nationalism Reframed: Nationhood and the National Question in the New Europe. Cambridge: Cambridge University Press.

21.

Burgoon

Koster

van Egmond

(2012) Support for redistribution and the paradox of immigration. Journal of European Social Policy 22(3): 288–304.

22.

Castles

Miller

(2003) The Age of Migration. New York: Guilford Press.

23.

Dahlberg

Edmark

Lundqvist

(2012) Ethnic diversity and preferences for redistribution. Journal of Political Economy 120(1): 41–76.

24.

Dallinger

(2010) Public support for redistribution: What explains cross-national differences? Journal of European Social Policy 20(4): 333–349.

25.

Eger

(2010) Even in Sweden: The effect of immigration on support for welfare state spending. European Sociological Review 26(2): 203–217.

26.

Eger

Valdez

(2015) Neo-nationalism in Western Europe. European Sociological Review 31(1): 115–130.

27.

ESS4 (2008) European Social Survey Round 4. Stata Readable Datafile: European Social Survey. Availabe at: www.europeansocialsurvey.org.

28.

Eurostat (2014) Employment rates by NUTS region (%), [lfst_r_lfe2emprt]. Database: Regional employment – LFS annual series . Available at: http://ec.europa.eu/eurostat (accessed 10 April 2014).

29.

Evans

MDR

Kelley

(2002) National pride in the developed world: Survey data from 24 nations. International Journal of Public Opinion Research 14(3): 303–338.

30.

Festinger

(1954) A theory of social comparison processes. Human Relations 7(2): 117–140.

31.

Finseraas

(2012) Poverty, ethnic minorities among the poor, and preferences for redistribution in European regions. Journal of European Social Policy 22(2): 164–180.

32.

Fox

(2004) The changing color of welfare? How whites’ attitudes toward Latinos influence support for welfare. The American Journal of Sociology 110(3): 580–625.

33.

Habermas

(2003) Toward a cosmopolitan Europe. Journal of Democracy 14(4): 86–100.

34.

Hasenfeld

Rafferty

(1989) The determinants of public attitudes toward the welfare state. Social Forces 67(4): 1027–1048.

35.

Hjerm

(2009) Anti-immigrant attitudes and cross-municipal variation in the proportion of immigrants. Acta Sociologica 52(1): 47–62.

36.

Hjerm

Schnabel

(2012) How much heterogeneity can the welfare state endure? The influence of heterogeneity on attitudes to the welfare state. Nations and Nationalism 18(2): 346–369.

37.

Hogg

Turner

(1987) Intergroup behaviour, self-stereotyping and the salience of social categories. British Journal of Social Psychology 52(4): 325–340.

38.

Hox

(1995) Applied Multilevel Analysis. Amsterdam: TT-Publikaties.

39.

Huber

(1989) Values and partisanship in left-right orientations: Measuring ideology. European Journal of Political Research 17(5): 599–621.

40.

Inglehart

Flanagan

(1987) Value change in industrial societies. The American Political Science Review 81(4): 1289–1319.

41.

Jæger

(2006) What makes people support public responsibility for welfare provision: Self-interest or political ideology? A longitudinal approach. Acta Sociologica 49(3): 321–338.

42.

Kaufmann

(2008) The lenses of nationhood: An optical model of identity. Nations and Nationalism 14(3): 449–477.

43.

Kelley

(1990) The failure of a paradigm: Log-linear models of social mobility. In: Clark

Modgil

(eds) John H. Goldthorpe: Consensus and Controversy. London: Falmer, pp.319–346.

44.

Kenworthy

(2004) Egalitarian Capitalism. New York: Russell Sage Foundation.

45.

Lamont

Molnár

(2002) The study of boundaries in the social sciences. Annual Review of Sociology 28(1): 167–195.

46.

Landerman

Land

Pieper

(1997) An empirical evaluation of the predictive mean matching method for imputing missing values. Sociological Methods & Research 26(1): 3–33.

47.

Luttmer

EFP

(2001) Group loyalty and the taste for redistribution. Journal of Political Economy 109(3): 500–528.

48.

Mau

Burkhardt

(2009) Migration and welfare state solidarity in Western Europe. Journal of European Social Policy 19(3): 213–229.

49.

Mau

Mewes

Zimmermann

(2008) Cosmopolitan attitudes through transnational social practices? Global Networks 8(1): 1–24.

50.

Mudde

(2013) Three decades of populist radical right parties in Western Europe: So what? European Journal of Political Research 52(1): 1–19.

51.

Mullen

Brown

Smith

(1992) Ingroup bias as a function of salience, relevance, and status: An integration. European Journal of Social Psychology 22(2): 103–122.

52.

Nannestad

(2007) Immigration and welfare states: A survey of 15 years of research. European Journal of Political Economy 23(2): 512–532.

53.

OECD (2009) Society at a Glance 2009: OECD Social Indicators. Available at: www.oecd.org (accessed 1 October 2012).

54.

OECD (2012) Social Expenditures Database. Available at: www.oecd.org (accessed 1 October 2012).

55.

Pedhazur

(1997) Multiple Regression in Behavioral Research: Explanation and Predicton. New York: Harcourt Brace.

56.

Pearl

(2009) Causality: Models, Reasoning and Inference (2nd ed.). New York, NY, USA: Cambridge University Press.

57.

Pfeifer

(2009) Public opinion on state responsibility for minimum income protection: A comparison of 14 European countries. Acta Sociologica 52(2): 117–134.

58.

Pierson

(2002) Coping with permanent austerity: Welfare state restructuring in affluent democracies. Revue française de sociologie 43(2): 369–406.

59.

Pierson

Leibfried

(1995) Multitiered institutions and the making of social policy. In: Leibfried

Pierson

(eds) European Social Policy: Between Fragmentation and Integration. Washington D.C.: The Brookings Institution, pp.1–40.

60.

Price

(1989) Social identification and public opinion: Effects of communicating group conflict. The Public Opinion Quarterly 53(2): 197–224.

61.

Putnam

(2007) E Pluribus Unum: Diversity and community in the twenty-first century: The 2006 Johan Skytte Prize Lecture. Scandinavian Political Studies 30(2): 137–174.

62.

Reeskens

Wright

(2013) Nationalism and the cohesive society: A multilevel analysis of the interplay among diversity, national identity, and social capital across 27 European societies. Comparative Political Studies 46(2): 153–181.

63.

Reicher

(2004) The context of social identity: Domination, resistance, and change. Political Psychology 25(6): 921–945.

64.

Renan

(1882) Qu’est-ce qu’une nation. Ourves Completes I (Paris, 1947): 887–907.

65.

Roosma

Gelissen

van Oorschot

(2013) The multidimensionality of welfare state attitudes: A European cross-national study. Social Indicators Research 113(1): 235–255.

66.

Roosma

van Oorschot

Gelissen

(2014) The preferred role and perceived performance of the welfare state: European welfare attitudes from a multidimensional perspective. Social Science Research 44: 200–210.

67.

Schumacher

Vis

van Kersbergen

(2013) Political parties’ welfare image, electoral punishment and welfare state retrenchment. Comparative European Politics 11(1): 1–21.

68.

Shayo

(2009) A Model of Social Identity with an Application to Political Economy: Nation, Class, and Redistribution. American Political Science Review 103(2): 147–174.

69.

Stegmueller

(2013) How many countries for multilevel modeling? A comparison of frequentist and Bayesian approaches. American Journal of Political Science 57(3): 748–761.

70.

Svallfors

(1997) Worlds of welfare and attitudes to redistribution: A comparison of eight western nations. European Sociological Review 13(3): 283–304.

71.

Tajfel

(1970) Experiments in intergroup discrimination. Scientific American (November): 96–102.

72.

Tajfel

(1982) Social psychology of intergroup relations. Annual Review of Psychology 33(1): 1–39.

73.

Tajfel

Turner

(1979) An integrative theory of intergroup conflict. In: Austin

Worchel

(eds) The Social Psychology of Intergroup Relations. Monterey, CA: Brooks-Cole, p.49.

74.

Tajfel

Billig

Bundy

. (1971) Social categorization and intergroup behaviour. European Journal of Social Psychology 1(2): 149–178.

75.

Taylor-Gooby

(2001) Sustaining state welfare in hard times: Who will foot the bill? Journal of European Social Policy 11(2): 133–147.

76.

Taylor-Gooby

(2011) Security, equality and opportunity: Attitudes and the sustainability of social protection. Journal of European Social Policy 21(2): 150–163.

77.

Turner

(1975) Social comparison and social identity: Some prospects for intergroup behaviour. European Journal of Social Psychology 5(1): 1–34.

78.

Van de Schoot

Lugtig

Hox

(2012) A checklist for testing measurement invariance. European Journal of Developmental Psychology 9(4): 486–492.

79.

Van Oorschot

Meuleman

(2012) Welfarism and the multidimensionality of welfare state legitimacy: Evidence from the Netherlands, 2006. International Journal of Social Welfare 21(1): 79–93.

80.

Wlezien

(1995) The public as thermostat: Dynamics of preferences for spending. American Journal of Political Science 39(4): 981–1000.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB