Realistic Group Conflict Theory and Economic Threats: A Study of Anti-Foreign Sentiments in 16 European Countries Using the European Social Survey Data 2002-2012 INTRODUCTION The following paper aims to answer the question of whether deteriorating economic conditions are likely to cause an increase in anti-immigrant prejudice in Europe. One of the major findings in many of the cross-sectional studies is the fact that contextual variables such as declining economic prosperity or high unemployment rates, coupled with certain individual-level factors, have a positive effect on anti-immigrant attitudes in Europe. More specifically, individuals with a particularly low socio-economic status are most likely to experience the effects of environmental factors on their levels of prejudice expression in relation to foreign nationals. These findings are most often explained in relation to realistic group conflict theory (hereafter RGCT), which states that two groups with divergent goals which compete over limited resources, be they tangible or intangible, are likely to experience a feeling of threat from one another. These feelings of threat are then likely to contribute to the increase in discriminatory attitudes and prejudice expression between the two communities. RGCT has on many occasions been used to explain why in times of relative economic decline, majority groups might develop negative attitudes towards foreign workers. The perceived competition over material and nonmaterial resources such as jobs or financial opportunities is likely to result in feelings of threat and fuel anti-immigration attitudes. The major contribution of this paper comes from its distinction between subjective (individual-level) and objective (contextual) ratings of economic performance. While research on contextual determinants of prejudice is rife, little attention has so far been given to subjective rating of economic conditions, which might not necessarily depend on the real level of prosperity within the country. This distinction is particularly important given that individual-level satisfaction with economy might have declined in many European states post 2008 not as a result of deteriorating economic conditions, but the rhetoric of crisis often employed by media and various right-wing parties. The following study will attempt to determine whether real economic changes are a more reliable predictor than subjective views of economic performance. It is crucial to note that the following study gauges its dependent variable, the extent of prejudice toward immigrants, by measuring anti-immigrant attitudes among respondents. It is assumed that those who report that immigrants tend to case deteriorating economic, cultural and living conditions are extremely likely to hold discriminatory views on foreign workers.
While many studies have previously shown that economic conditions might indeed contribute to discriminatory attitudes, they usually either use data collected before the 2008 crisis, or are restricted to a single wave of major cross-sectional surveys such as the European Social Survey (hereafter ESS). Apart from distinguishing between subjective and contextual measures of economic performance, the following study further contributes to the debate on the effects of contextual variables on prejudice expression. Unlike previous studies, it utilizes all available waves of the ESS between 2002 and 2012 in order to observe whether the extent of prejudice expression in Europe has changed as a result of the 2008 economic crisis. EXPLANATIONS OF PREJUDICE Traditionally, the most influential theory accounting for prejudice expression is thought to have been formulated by Gordon Allport (1954/79). He identified the formation of individuals negative affects toward ethnic and cultural minorities as a result of their insufficient exposure to what is located beyond their immediate social surroundings. Allport (1954/79: 29-33) theorized that familiarity, fundamental to human survival, becomes a value shared with those who happened to occupy our closest environment. Over time, individuals become members of in-groups, or clusters of people who can use the term we with the same significance (Allport, 1954/79: 37). Besides the shared value of familiarity, the subjects must possess a basic awareness of their in-group membership (Tajfel, 1982: 2) This approach, termed the social contact theory, asserts that insufficient socialization or lack of familiarity with members of foreign communities will result in negative attitudes and generalizations of entire out-groups: cultures, nations, creeds or classes. These antagonistic feelings can only be reduced by extensive, controlled interaction between in-group and out-group members, resulting in the reduction of social distance and increased tolerance (Allport, 1954/79; Pettigrew, 1998; Pettigrew and Tropp, 2011). Although compelling, the theory fails to account for prejudice and its absence in certain settings. Assuming the theory s soundness, it renders all members of extensive, homogenous groups such as nations with few cultural and ethnic minorities invariably and equally prejudiced toward all out-groups; an assumption which is clearly implausible. Secondly, the theory fails to explain why multiculturalism is often thought to result in heightened ethnic tensions instead of contributing towards reduction of conflict (Lentin and Titley, 2011). It has been widely reported that increase in proportion of outside workers has a positive impact on the anti-immigrant sentiments among the citizens of countries affected by the inflow of foreign labour.
REALISTIC GROUP CONFLICT THEORY In order to account for these issues, the realistic conflict theory developed by Muzafer Sherif (1966) has supplemented Allport s (1954/79) argument with a third dimension: competition for material or symbolic capital. Distinct groups will only stigmatize each other if found in a direct, zero-sum contest over resources (Turner, 1975). Moreover, according to the argument, prejudices serve a clear function of justifying the dominance of the superior groups over the weaker one (Young-Bruehl, 1996: 50). In their Robber s Cave Experiment, Sherif et al. (1966) have shown competition to be a leading factor in bias formation. They assembled two groups of young boys from similar white Protestant, two-parent middle-class backgrounds that had not previously known each other and divided them into two separate groups unaware of each other s existence; after a period of bonding, the groups were exposed to each other in a setting of various competitive disciplines. Within days, derogatory terms and songs relating to competitors had been invented and desire of segregation expressed, along with raiding of the other group s property, by which goods were damaged and stolen (Sherif et al., 1966: 96-113). Various other experimental studies further confirmed that competition tends to increase intergroup hostilities (Sherif, White & Harvey, 1955; Rabble and Horwitz, 1969; Blake and Mouton, 1962). RGCT AND ANTI-IMMIGRANT ATTITUDES Realistic group conflict theory is one of the most widely used explanations for the high prevalence of anti-immigrant attitudes. RGCT assumes that groups which find themselves locked in a zero-sum contest over resources will tend to experience a high degree of threat, and that the threat is then likely to result in stigmatization and development of discriminatory practices and prejudice expression among the contesting groups (Sherif, 1966; Turner, 1975). This competition might concern material resources, such as employment or housing opportunities, as well as less tangible capital such as power, values or social status. The threat of losing vital resources results in negative attitudes towards competitors. It follows that worsening economic conditions, leading to higher rates of unemployment, are likely to result in hostilities between groups which directly compete over jobs and material resources. Furthermore, prejudice might also serve as a means of preserving the social position of dominant groups against what they consider as a threat to their current status. Individual interests of the in-group members, although relevant, are not as crucial as the privileges of the group as a whole (Bobo, 1988). It has been shown that dominant group members
whose interests are not directly or immediately threatened in the competition are as likely as others to develop discriminatory attitudes toward group rivals (Sears and Funk, 1991). As a result, Quillan (1995) has theorized about the importance of contextual factors in accounting for individual prejudice expression, stressing the importance of economic conditions such as unemployment rates and GDP per capita in explaining the rise of antiforeign sentiments in Europe. Although deteriorating economic conditions do not affect the entire population of a country, they are nonetheless likely to result in increased antiimmigrant prejudice. The majority will be more concerned with the threat foreign workers pose to the group as a whole, rather than to its individual members. For this reason, individual-level variations are not enough to satisfactorily explain the prevalence of antiforeign prejudice. Contextual variables such as economic conditions need to be considered if discriminatory behaviour is to be fully accounted for. PREVIOUS RESEARCH Research on context-related determinants of prejudice expression has become widespread in recent years (Coenders, 2001; Semyanov et al., 2006; Sides and Citrin, 2007; Semyanov et al., 2008). The rising of anti-immigrant sentiment across Europe in connection with the 2008 economic crisis has led to renewed interest in how economic conditions might affect the public s attitude toward foreign workers (Mueleman and De Witte, 2014). Although Quillian s (1995) research has found that contextual variables indeed tend to increase anti-foreign attitudes, the relationship between economic conditions and discriminatory opinions has since been largely contested. While some research confirms the effects of economic conditions on negative perceptions of immigrants (Coenders, 2001; Semyanov et al., 2006; Semyanov et al., 2008), others find little or no effect when including contextual variables in their models (Strabac and Listhaug, 2008; Sides and Citrin, 2007). However, given the recent rise in unemployment and economic hardship following the 2008 Eurozone crisis, it is likely that these effects will become more prominent in multilevel analysis. The release of the sixth wave of European Social Survey data from 2012 should allow for a more detailed investigation into how economic conditions might have affected prejudice expression in Europe. Although a recent study by Billiet, Mueleman and De Witte (2014) has tested the assumptions of RGCT by analysing the 2012 edition of the ESS, their research has been limited to one wave only, making it difficult to test for change in immigrant perception before and after the 2008 economic crisis. The following paper will investigate all six waves of the ESS data, including both individual and contextual variables.
Furthermore, although contextual variables are likely to have a significant effect on antiimmigrant prejudice, it is crucial to note that individual perception of economic performance is likely to be as important as unemployment levels or GDP growth. While to a large degree the subjective views of economic conditions might be correlated with the environmental factors, it is likely that individuals might change their views depending on the information they are receiving from their immediate surroundings. The crisis rhetoric, prevalent in many European countries after the 2008 financial crash, has been adopted by both the media and right-wing parties and had likely changed the way individuals perceive their state s economic performance. A further discussion of subjective ratings of economy and how it might affect prejudice expression is included in the methods section. In conclusion, the following study contributes to the body of quantitative literature on the RGCT in three ways. Firstly, it employs a double measure of economic performance and compares its effects on anti-foreign attitudes. Secondly, it employs all six waves of the ESS in order to investigate whether prejudice expression has changed as a result of the 2008 crisis. Finally, through employing all six waves of the ESS, the paper contributes significantly to the debate on whether contextual economic variables do indeed affect prejudice expression. The sudden spike in unemployment and stagnating economic growth observed in many European countries after the 2008 financial crisis are the first opportunity to fully investigate whether sudden changes in economic prosperity does indeed affect anti-immigrant attitudes. METHODS The data for the current study has been obtained from all six waves of the European Social Survey conducted every two years between 2002 and 2012. All individuals subjected to the survey have been selected through random probability sampling and include all persons aged 15 and above, regardless of their nationality, language or citizenship. In order to capture the opinion of the majority group within the country, the sample has been restricted to citizens of a given country only. Furthermore, given that the present study is primarily concerned with measuring attitudes towards immigrants, respondents who have identified themselves as foreign workers have also been dropped from the sample 1. In order to collect data from all waves between 2002 and 2012, the number of countries 1 In order to be identified as foreign nationals, respondents had to answer positively to a question regarding potential prejudice against a group they consider themselves to be a member ( Would you describe yourself as being a member of a group that is discriminated against in this country? ) and provide with nationality as a reason for discrimination ( On what grounds is your group discriminated against? )
available for analysis has been dropped to 16, due to limited ESS coverage. Allowing for other modifications such as elimination of missing values and restricting the age group to 15-90 years of age, the sample size has further decreased to 154,736 respondents. MEASURING PERCEIVED THREAT Three items from the ESS survey have been used to construct a tolerance scale for this study: the perceived impact of immigrants on a state s economic conditions ( Would you say it is generally bad or good for [country] s economy that people come to live here from other countries? ); the perceived impact of immigrants on a state s culture ( And, using this card, would you say that [country] s cultural life is generally undermined or enriched by people coming to live here from other countries? ); and the overall impact of immigration on the living conditions within a state ( Is [country] made a worse or a better place to live by people coming to live here from other countries? ). The remaining items referring to immigration, such as whether more or less immigrants ought to be allowed in the country, have been left out from the scale, given that answers to these questions were likely to have been highly dependent on the perceived impact of foreign workers on the state s economy, culture and living conditions. All three items have been shown to measure the same concept through use of confirmatory factor analysis and have been previously used by a number of studies concerned with measuring perceived threat from immigrants (Billiet, Mueleman and De Witte, 2014; Schneider, 2008; Billiet and Philippens, 2004; Coenders et al., 2005). The scale is highly reliable, with Cronbach s alpha = 0.8366, a very high score considering how few items have been used to construct the measurement. The lower the score on the 11 point item, the more prejudiced the respondent is considered to be. Figure 1 presents the average scores for all 16 countries on the tolerance scale, with results ranging from 4.36 for Hungary to 6.26 for Sweden. The overall mean score on the scale for all countries is 5.27.
Figure 1. Mean tolerance scores, by country INDIVIDUAL LEVEL VARIABLES Following previous studies, the paper uses socio-economic indicators as individual-level predictor variables in the model. It is assumed that those with a lower socio-economic status (measured by highest educational achievement and net household income levels) will feel more vulnerable and therefore more threatened by foreign workers who they might perceive as potential competitors. Since resource competition tends to be much higher in low-status occupations which do not pay well and require little to no formal education, and immigrants, on average, tend to compete with the majority group members for these positions, these members of the population might feel more threatened by immigrants than others (Schneider, 2008: 55) Moreover, those with medium and higher incomes are more likely to express positive views on immigration due to having more potential for developing support systems in case of unexpected or prolonged periods of unemployment. For this reason it is hypothesised that individuals with higher
educational and income levels will express more favourable views of foreign workers impact on their country, culture and economy. There are also demographic variables such as age and gender that are expected to have a significant impact on the tolerance scale scores. Previous studies suggest that discriminatory attitudes are more pronounced among women and older people (Semyanov et al. 2002; Semyanov et al., 2008). Since men tend to have a better access to education, an interaction effect for male gender and tertiary education has been included in the model. Due to the high number of younger respondents in the dataset, an interaction between age and tertiary education has also been taken into account. Given that group threat theory assumes that those who feel economically vulnerable or disadvantaged tend to have a higher perception of threat from foreign workers, a dummy employment variable has been included in the analysis. Finally, those who have experienced potential economic hardship due to unemployment in the past ( Have you ever been unemployed and seeking work for a period of more than three months? ), individuals with history of unemployment, are also expected to be significantly more likely to score lower on the 11 point tolerance scale. An interaction between being currently unemployed and having been unemployed for at least 3 months in the past has also been included, as it suggests a more consistent history of economic hardship. CHANGES IN PREJUDICE EXPRESSION BETWEEN 2002 AND 2012 Firstly, utilizing descriptive statistics, the study will investigate whether the level of antiimmigrant prejudice has increased or decreased over time. Since the economic conditions in most European countries have deteriorated or their economic growth has slowed down, it is assumed that, in suit with RGCT, the levels of prejudice have increased as a result of rising competition over resources between the majority group and the out-group (the foreign workers). Therefore: H1: The respondents are more likely to score lower on the tolerance scale in the waves four (2008), five (2010) and six (2012) then in the first three waves (2002, 2004, 2006). INDIVIDUAL PERCEPTION OF THE CURRENT STATE OF ECONOMY RGCT predicts that individuals are likely to express prejudice regardless of whether their economic position is immediately threatened or not. It is therefore crucial to include an individual level variable which could reflect the feelings of the majority group about
current economic conditions. For this reason, subjective assessment of the state s current economic performance will be included in the model. It is important for two reasons. Firstly, individuals dissatisfied with the state of economy in their country, regardless of its actual performance, are more likely to feel threatened by immigrants, even if their wellbeing is not directly threatened. Perception of poor economic performance might reflect the perception of hardship faced by other members of the group. The awareness of economic difficulties is likely to increase the perceived threat the immigrants might pose to the majority group. Secondly, the measure is likely to have bearing on individual-level differences in prejudice expression. More specifically, low economic satisfaction might be due to some subjective experience of hardship which has not been captured by employment status or income variables. Given that economic satisfaction is likely to be partially influenced by the actual economic condition within the state, it will be allowed to vary both at the individual and state level in the model. The economic satisfaction variable is normally distributed, and is a scale from 0 to 10, where 0 indicates the lowest level of satisfaction, and 10 the highest level of satisfaction (see table 1 in appendix 1 for more details). H2: The higher the respondents place themselves on the economic satisfaction scale, the more likely they are to score higher on the tolerance scale. CONTEXTUAL VARIABLES As stated by Quillian (1995: 591), it is expected that the weather the country, the lower the amount of individuals who find themselves in direct conflict over scarce resources with the foreign workers. The level of competition is likely to have impact on the overall perception of immigrants among the in-group members. High GDP per capita and low unemployment rates are expected to have a negative impact on discriminatory attitudes across all countries. Average rates for both economic performance indicators have been calculated to make sure that small variations in either unemployment rates or GDP per capita do not influence the results. The averages are for three years prior to the ESS data being collected. For example, in order to calculate the mean of unemployment for the first wave of the ESS (2002), the mean GDP per capita (in $1000) and mean unemployment rate (%) has been calculated from years 2000, 2001 and 2002. It is expected that: H3: The higher the mean GDP per capita ($1000) in a country, the higher the scores on the tolerance scale in the county, and:
H4: The higher the mean rate of unemployment (%) in a country, the lower the scores on the tolerance scale in the country. ANALYSIS AND FINDINGS Figure 2 demonstrates the changes in mean tolerance score by ESS wave. While tolerance towards immigrants is highest during the fourth wave (2008), it drops in the subsequent wave, and rises again to similar levels in wave 6 (2012). However significant, this variation does not seem to be large enough to be attributed to the 2008 economic crisis, as a similar drop can be observed between wave one (2002) and wave 2 (2004). Contrary to the predictions formed in hypothesis 1, the graph suggests an upward trend in tolerance scores. Figure 2. Change in mean tolerance score by ESS wave Figure 3 demonstrates the changes in tolerance score distribution before and after the 2008 crisis. While a certain level of variation is obvious, the results are not conclusive. It is clear from the plot that there were major differences in respondents attitudes toward foreign workers depending on the questionnaire wave. In 2002, the respondents seemed to
have a much more unified view of the impact immigrants have upon their country, with over 25% of people scoring below 5 on the tolerance scale. The lower the score on the tolerance scale is, the less favourable the view of foreign workers in general. In 2004, negative attitudes toward immigrants have become more prevalent, with almost 50% of all respondents scoring 5 or below on the scale. Furthermore, the minimum observation has decreased relative to the first wave, indicating a polarizing trend in anti-immigrant attitudes. In the third wave (2006) this trend seems to be even more prominent, with greater amount of people reporting more extreme views on immigrants. While the maximum and minimum values remain relatively similar in the 2008 wave, the 2010 wave indicates that more and more respondents express positive views of immigrants relative to previous waves. The 2012 wave has seen the median tolerance score reaching its highest point since the ESS begun in 2002, indicating that over 50% of respondents held a positive opinion on the impact of foreign workers on their country. These results are somewhat surprising. It was expected that anti-immigrant prejudice would increase, rather than decrease, as a result of the 2008 crisis. On the contrary, people seem to have gained a more positive outlook on foreign workers with time. These results contradict the assumptions of RGCT. According to its premises, deteriorating or stagnating economic conditions, which have been observed in most of European states after the 2008 economic crisis, should have a positive effect on the rise in prejudice as a result of direct economic competition between the majority and minority groups. These initial results suggest that the social contact theory, discussed in the initial section of this paper, is a more plausible explanation of the prevalence of prejudice. The spike in migration might have contributed to the increase in socialization between the groups and decreased, rather than increased, the level of prejudice expression among the respondents.
4 4.5 5 5.5 6 6.5 Figure 3. Changes in tolerance scale score distribution by ESS wave Sweden 15 Sweden Source: EVS 2002-2012 2002 2004 2006 2008 2010 2012 ESS wave Upon estimating the null model, the interclass correlation coefficient of 0.07 has confirmed that a considerable difference in scores on a tolerance scale (7%) can be explained by between-country variance. The χ² of the likelihood-ratio test comparing the standard model to a two-level one has shown the latter model to be significantly better than standard OLS regression, justifying the need for multilevel analysis. Table 1 shows that the full individual-level model explains about 42% of between country variance in tolerance scores, with economic threat variables (income, employment status, employment history and perceived economic performance of the country) adding most explanatory value to the model. The results of from the full individual-level model are displayed in table 2.
TABLE 1. Variance components for model with individual level explanatory variables Null model Individual level model Demographic variables Education level Variables Economic threat variables Individual level variance 3.722 3.675 3.435 3.276 Between country variance Explained individual level variance Explained country level variance 0.290 0.283 0.252 0.168 0 0.012 0.077 0.12 0 0.025 0.13 0.42 In the initial model, it appears that the younger the individuals, the more tolerant they were likely to be. The tolerance levels decrease with age until the respondents reach the mean age of 47.3 years, when their tolerance seems to increase as they get older. This effect seems to be inversed once educational variables are accounted for and once reaching a mean age of 47.3 the respondents tend to become less tolerant with the passage of time. This is likely due to the fact that to a certain point, the older the respondents, the more likely they are to have reached full secondary and tertiary education. However, once they reach the 38-45 age category at which the educational level is highest, there is a steep decrease in the amount of respondents with an educational level higher than primary (see figure 2 in appendix 1). Similarly, the initial effect of being male on tolerance level (increase in 0.1 points on the tolerance scale) seems to disappear with the introduction of education variables. This is most likely because men, on average, tend to be more educated than females. This finding contradicts that of Semyanov et al. (2008) who find that women, on average, tend to be more prejudiced than men. Secondary education tends to increase overall tolerance by 0.4 point and tertiary education by a little over 1 full point even when other variables are controlled for in the full individual level. This suggests, as predicted, that education increases tolerance levels among individuals. This finding is not surprising, given that
educational attainments are likely to improve one s economic security by increasing employment opportunities. Furthermore, it is also likely that education has a positive correlation with tolerance through promulgation of specific values, such as development of pro-democratic attitudes and appreciation for diversity (Hyman and Wright, 1979). Educated respondents are also more likely to appreciate the potential economic benefits of foreign labour and feel less threatened by their presence. It is clear from table 2 that economic satisfaction is a strong and significant predictor of tolerance, as expected in hypothesis 2. An increase of one point in economic satisfaction level results in almost 2 point increase on the tolerance scale. These effects are visible even when accounting for factors such as income, unemployment history or being currently unemployed and seeking for work. This confirms the expectation that low levels of satisfaction are likely a reflection of perceived economic threat from immigrants to the country s general population, not only individual s self-interest. Finally, neither being unemployed, having a history of unemployment, or the interaction between the two seems to have any significant impact on tolerance score in the model.
Table 2. Full individual model Estimates of fixed parameters β + Demographic variables + Education variables + Economic variables (indiv.) Constant 8.429*** 4.496*** 4.876*** Age, in years -0.067*** 0.004-0.145 Square rooted age, centred 0.768*** -0.154*** -0.034 Male gender 0.092*** 0.073*** 0.020 Secondary education 0.420*** 0.387*** Tertiary education 1.241*** 1.093*** Age*tertiary education 0.002* 0.002** Male*tertiary education 0.024-0.016 Satisfaction with economy 0.175*** Ever unemployed for a period of at least 3 months 0.0119 Unemployed, actively seeking work -0.040 Ever unemployed for at least 3 months*unemployed, actively seeking work -0.105 Medium income 0.098*** High income 0.265***
3 4 5 6 7 Figure 4 provides a more detailed view of how the subjective view of economic performance affects the mean tolerance score. There is a very clear positive relationship between economic satisfaction and tolerance. Poland and Sweden are clear outliers where positive attitudes towards foreign migrants seem to correlate with economic satisfaction to a much lesser degree. It is likely that the coefficient in the first model (table 2) would be much higher if the outliers were to be removed. Figure 4. The relationship between perceived satisfaction with state s economy and mean scores on the tolerance scale Sweden Poland 4 4.5 5 5.5 6 6.5 Mean score on the tolerance scale Figure 5 below demonstrates the strength of the relationship between subjective economic satisfaction and tolerance scores by ESS wave. Apart from the first wave (2002) where the slope is visibly less steep than in other waves, the relationship seems to be rather stable. Scores on economic satisfaction scale were lower in 2008, 2010 and 2012 wave, which corresponds with the deteriorating economic conditions after the 2008 financial crisis.
2 3 4 5 6 7 Figure 5. The relationship between perceived satisfaction with state s economy and scores on the tolerance scale, by wave 0 2 4 6 8 10 Score on the tolerance scale 2002 2004 2006 2008 2010 2012 Source: ESS 2002-2012 Table 3 shows the result from a model with the addition of two contextual variables (mean GDP per capita and mean unemployment rate) and the satisfaction with economy variable into the random-coefficient model. The likelihood ratio tests suggest that addition of each country-level variable at level two significantly improves the model. The results from the full two-level model are presented in table 4. Age, education and income remain highly significant predictors of tolerance. Mean unemployment rate does not seem to have a significant effect on tolerance levels. Surprisingly, the increase in mean GDP per capita tends to decrease, rather than increase the scores on the scale. A possible explanation for the effect is the fact that despite the 2008 economic crisis, GDP per capita has been steadily increasing most countries. Another explanation could be the fact that the change in mean GDP per capita scores has not been prominent enough to have the predicted effect on discriminatory attitudes. This contradicts hypothesis 3 which states that the higher the GDP per capita, the higher the scores on the tolerance scores. Similarly, hypothesis 4 stating that higher unemployment rates would lead to decrease in tolerance scores has also been disproved. Nevertheless, economic satisfaction remains a
highly significant predictor of anti-immigrant attitudes. The visual representation of these results can be seen in figures 2 and 3 in the appendix, which show the effects of mean GDP per capita change and economic satisfaction change on tolerance for specific countries. These results are surprising, and might be due to one of the following factors. Firstly, it is possible that subjective perceptions of economic satisfaction are unrelated to real changes in economic performance and are simply a more powerful predictor of antiimmigrant prejudice. This could suggest that the rhetoric of economic crises, whether substantiated by actual changes in performance or not, affects respondents to a much greater degree than real changes in economic growth or unemployment level. These latter two factors might simply be unperceivable to the respondents, who might not always follow the current economic trends of their countries. In future research, it would perhaps be useful to control for the level of political interest expressed by survey participants. Secondly, it is equally likely that GDP per capita and unemployment level are not good predictors of real economic performance. For example, it is possible that relative change in unemployment would be a much more helpful measure of the current economic conditions than the percentage of people who are currently out of work. A 10% level of unemployment might have a different significance for Poland than it does for Sweden. Some countries might have a naturally high level of informal economy, where officially unemployed citizens still engage in economic activity without officially participating in it. Similarly, in some countries it might be more natural for women to stay at home, making the unemployment rate higher without reflecting potential economic hardship. A more focused cross-country comparison in the future could offer some insights into which countries should be examined together, and which indicators would best predict economic performance. Table 3. The final model Full model Estimates of fixed parameters β Std. error t Constant 7.43*** 0.657 11.31 Age -0.008** 0.003-2.84 Square rooted age, centred 0.007 0.035 0.21
Male gender 0.016 0.011 1.45 Secondary education -0.678*** 0.034-20.06 Tertiary education 1.062*** 0.035 30.21 Age*tertiary education 0.002** 0.001 68.53 Male*tertiary education -0.007 0.021 4.45 Satisfaction with economy 0.181*** 0.011 15.81 Ever unemployed for a period of at least 3 months 0.014 0.011 1.22 Unemployed, actively seeking work -0.048 0.062-0.77 Ever unemployed for at least 3 months*unemployed, actively seeking work -0.105 0.067-1.57 Medium income 0.097*** 0.012 8.45 High income 0.251*** 0.013 20.02 Wave 1-0.786*** 0.109-7.19 Wave 2-0.714*** 0.069-10.20 Wave 3-0.460*** 0.037-12.36 Wave 5-0.152*** 0.022-6.94 Wave 6-0.012 0.029-0.41 Square root of GDP per capita -0.010*** 0.002-3.96 Natural logarithm of mean unemployment rate 0.063 0.148 0.43 Finally, it is worth mentioning the effects of the wave in which the respondents have been interviewed and the possible effects of the economic crisis on the perception of threat from foreign workers. The reference category in the group was year 2008 (wave 4). Tolerance towards immigrants in 2008 is statistically greater than tolerance in 2002 and 2006 waves. The tolerance in the subsequent wave (2010) is also significantly lower than that in 2008. This confirms the initial findings from figures 2 and 3 earlier in this section.
CONCLUSION Although research on RGCT has been growing in recent years, with many studies finding a positive correlation between unfavourable economic conditions and the presence of antiimmigration attitudes, the results of this paper are mixed. Firstly, the hypothesis stating that respondents would be more likely to score lower on the tolerance scale between 2008 and 2012 relative to the 2002-2006 period has not been confirmed. Clearly, the economic crisis itself has not influenced people s views of foreign workers in the manner hypothesized by RGCT. On the other hand, it suggests that social contact theory might be a worthy competitor in terms of explaining the presence or absence of prejudice in European society. The more socialization between immigrants and the in-group as the outgroup population increases, the lower the level of threats expressed by the majority. Secondly, the perceived economic performance was a much better predictor of antiimmigrant attitudes than contextual variables. While unemployment seems to have had no statistically significant impact on the tolerance scale, growth in GDP per capita has actually decreased tolerance scores across all 15 investigated countries apart from Poland. These findings are important because it suggests either that contextual variables are not relevant when accounting for anti-foreign attitudes or that unemployment levels and GDP per capita are not reliable measures of economic prosperity. Future studies could investigate sudden changes in unemployment or GDP per capita levels on anti-foreign prejudice as there are more likely to reflect economic turbulence within the state. Despite contextual variables having no or little effect on prejudice expression, one aspect of RGCT has been confirmed: people who rate the economic conditions in the country unfavourably were most likely to express negative views of immigration, regardless of their educational level or income. This suggests that prejudice towards foreign workers might be independent of any immediate threats to one s economic security, as predicted by Bobo (1986). Prejudice expression might indeed be partially dependent on the relative threat to the entire in-group as opposed to the individual herself. It is also possible that an individual s perception of economic disadvantage, even when unsubstantiated by GDP figures, has a significant impact on anti-foreign sentiments. Finally, individual-level variables seem to have had a pronounced effect on prejudice expression. Particularly, those in more advantaged economic situation (with a high level of education, and high income) seem to have the most favourable views of foreign workers impact on the economy, culture and living conditions within the state. The overall conclusion of the paper is that contextual predictors of prejudice should be
approached with caution and a much more discussion on operationalization of these variables should be included in future research. Furthermore, more attention should be paid to subjective interpretation of environmental factors, as this paper suggests they are a very strong predictor of anti-foreign attitudes. While the RGCT assumptions have not been substantiated fully by this paper, it is likely that this is due to relatively low construct validity rather than incorrect theoretical specifications.
Appendix 1. Table 1. Definition for the individual-level and country-level variables included in the analysis Individual level variables Definition Age Square root of age, centred Male gender Education Secondary education Tertiary education Age*tertiary education Male*tertiary education Satisfaction with economy Unemployed Unemployment history Income In years In years Man = 1 (%) What is the highest level of education you have successfully completed? Secondary = 1 (%) Tertiary = 1 (%) Interaction between age and having tertiary education Interaction between being male and having tertiary education On the whole how satisfied are you with the present state of the economy in [country]? 0-10 scale: 0 = extremely dissatisfied, 10 = extremely satisfied Using this card, which of these descriptions applies to what you have been doing for the last 7 days? Select all that apply. Unemployed = 1 (%) Have you ever been unemployed and seeking work for a period of more than three months? Yes = 1 (%) Using this card, please tell me which letter describes your household's total income, after tax and compulsory deductions, from all sources? If you don't know the exact figure, please give an estimate. Use the part of the card that you know best: weekly, monthly or annual income Wave High Medium Tolerance items Recode of the income variable, country specific. High = 1 (%) Medium = 1 (%) European Social Survey round Would you say it is generally bad or good for [country] s economy that people come to live here from other countries? 0-10 scale: 0=bad for economy, 10=good for economy And, using this card, would you say that [country] s cultural life is generally undermined or enriched by people coming to live here from other countries? 0-10 scale: 0=cultural life undermined, 10=cultural life enriched Is [country] made a worse or a better place to live by people coming to live here from other countries? 0-10 scale: 0=worse place to live, 10=better place to live
Country-level variables Definition Mean GDP (square root of) Mean unemployment (natural logarithm of) Mean Gross Domestic Product per capita, in $1000, calculated for every three years between 2000 and 2012 Mean unemployment rate (%) for all ages, calculated for every three years between 2000 and 2012 Appendix figure 2. A visual representation of the effects of economic satisfaction on tolerance scores, by country.
Appendix figure 3. A visual representation of the effects of the increase in mean GDP per capita on tolerance scores, by country. BIBLIOGRAPHY Allport, G. W. (1954/79) The Nature of Prejudice. 25th Anniversary Edition. New York: Basic Books. Billiet,J., Meuleman, B., and De Witte, H. (2014) The Relationship Between Ethnic Threat and Economic Insecurity in Times of Economic Crisis: Analysis of European Social Survey Data, Migration Studies, forthcoming. Billiet, J. and Philippens, M. (2004) Quality assessment of the registered responses. In Billiet, J. and Philippens, M. (Eds), Work Package 7: Data-Based Quality Assessment in ESS Round 1. URL: http://tinyurl.com/2zx9v2. Blake, R. R., and Mounton, J. S. (1962) The intergroup Dynamic of Wind-Loss Conflict and Problem Solving Collaboration in Union-management Relations, in M. Sherif (ed.), International Relations and Leadership. New York: John Wiley. 94-141.
Bobo, L. (1983) Whites' Opposition to Busing: Symbolic Racism or Realistic Group Conflict, Journal of Personality and Social Psychology, 45 (6), 1196-1210. Coenders, M., Lubbers, M. and Scheepers, P. (2005) Majorities Attitudes towards Minorities in Western and Eastern European Societies: Results from the European Social Survey 2002 2003. Report 4 for the European Monitoring Centre on Racism and Xenophobia, Vienna. Lentin, A. and Titley, G. (2011) The Crises of Multiculturalism: Racism in a Neoliberal Age. London: Zed Books. Quillian, L. (1995) Prejudice as a Response to Perceived Group Threat: Population Composition and Anti-Immigrant and Racial Prejudice in Europe, American Sociological Review, 60 (4), 586-611. Pettigrew, T. F. (1998) Intergroup Contact Theory, The Annual Review of Psychology, 49 (1), 65-85. Pettigrew, T. and Tropp, L. R. (2011) When Groups Meet: The Dynamics of Intergroup Contact. New York and Hove: Psychology Press. Rabble, J., & Horwitz, M. (1969) Arousal of Ingroup-Outgroup Bias by a Chance Win or Loss, Journal of Personality and Social Psychology, 13, 269-277. Schneider, S. L. (2008) Anti-Immigrant Attitudes in Europe: Outgroup Size and Perceived Ethnic Threat, European Sociological Review, 24 (1), 53-67. Sears, D. 0., and Funk, C. L. (1991) The Role of Self-Interest in Social and Political Attitudes, Advances in Experimental Social Psychology, 24 (1), 1-91. Semyonov, M., Raijman, R. and Gorodzeisky, A. (2006) The Rise of Anti-foreigner Sentiment in European Societies, 1988-2000, American Sociological Review, 71, 426-449. Semyonov, M., Raijman, R. and Gorodzeisky, A. (2008) Foreigners Impact on European Societies. Public Views and Perceptions in a Cross-National Comparative Perspective, International Journal of Comparative Sociology, 49(1), 5-29. Semyonov, M., Raijman, R., Yom Tov, A., and Schmidt, P. (2004) Population Size, Perceived Threat, and Exclusion: A Multiple-Indicators Analysis of Attitudes Toward Foreigners in Germany, Social Science Research, 33, 681-701. Sherif, M. (1966). Group conflict and cooperation: Their social psychology. London: Routledge &Kegan Paul. Sherif, M., White, B. J., & Harvey, O. J. (1955) Status in experimentally produced groups, American Journal of Sociology, 60, 370-379. Sides, J., and Citrin, J. (2007) European Opinion About Immigration: The Role of Identities, Interests and Information, British Journal of Political Science, 37, 477-504.
Tajfel, H. (1970) Experiments in Intergroup Discrimination, Scientific American, 223 (5), 96-102. Turner, J. C. (1975) Social Comparison and Social Identity: Some Prospects for Intergroup Behaviour, European Journal of Social Psychology, 5 (1), 5-34.