Trust in People and Attitudes Towards Immigration

We use data from the European Social Survey (ESS) Rounds 1-7 to investigate the relationship between trust in people and attitudes towards immigrants and immigration. Our analysis is based on large longitudinal comparative survey data (ESS), where the immigrant attitudes are operationalized by two groups of items: the attitude toward immigrants (“To what extent do you think [country] should allow people of the same race or ethnic group as most of [country]’s people to come and live here?“; “To what extent do you think [country] should allow people of a different race or ethnic group as most of [country]’s people to come and live here?“; “To what extent do you think [country] should allow people of poorer countries in Europe to come and live here?“) and the attitude toward immigration (“Would you say it is generally bad or good for [country]’s economy that people come to live here from other countries?”; “Would you say that [country]’s cultural life is generally undermined or enriched by people coming to live here from other countries?”; Is [country] made a worse or a better place to live by people coming to live here from other countries?”). We constructed a structural equations model comprised of three intercorrelated latent variables. Each latent variable was supported by three items from the ESS questionnaire. Latent variables represented factors and factor loadings for all items that were higher than 0.65, suggesting satisfactory factor stability for all three factors. These results show that the questions in the ESS questionnaire that targeted respective attitudes and values (immigrants, immigration, trust) are consistent and that each triplet measures a common underlying factor. We performed structural equations modelling (SEM) with three latent variables – attitudes towards immigrants, attitudes towards immigration and trust in people. We studied the correlations between the variables. The model fit for this model was satisfactory. RMSEA was 0.028, and CFI was 0.995. In the second step, we performed a multi-group SEM analysis. We introduced 32 groups, with each group representing data from one country. First, we tested an unconstrained model. We obtained 32 triplets of correlations. The fit of this model was satisfactory. RMSEA was 0.006, and CFI was 0.992. The strengths of the correlations varied, but all of the correlations had matching signs. Additionally, in every country, the correlation between attitudes towards immigrants (F1) and attitudes towards immigration (F2) was always the highest of the three in absolute terms. The correlation between attitudes towards immigration (F2) and trust in people (F3) was the second highest, and the correlation between attitudes towards immigration (F1) and trust in people (F3) was always the lowest. We confirm the significance of every one of the three correlations and discuss the possible factors behind these differences. This indicates a consistent pattern of relationships between the three latent variables across the wide range of countries and adds strength to our finding that levels of trust in people are universally related to attitudes towards immigrants and immigration. Človek a spoločnosť [Individual and Society], 2019, Vol. 22, No. 1, pp. 34-43. Trust in People and Attitudes Towards Immigration 35 We then put constraints on every correlation coefficient to see whether the countries differ significantly in the strengths of correlations. We built three models, forcing covariances between pairs of latent variables to be equal across the countries, and compared the models using the Chi square test. All three models compared to the unconstrained model had p<0.001, indicating significant differences between the constrained and the unconstrained models. We can conclude that there are significant differences in correlation strengths among the countries. Our analysis shows that although correlation signs and strengths are consistent among people in various countries, significant differences remain in terms of how strongly trust in people and attitudes towards immigrants and immigration are correlated. When we look at an ordered list of the countries according to the strength of correlation between attitudes towards immigration, we find that counties with higher correlations are mostly "older" countries of the EU like Germany, Italy, Belgium, the Netherlands and France. Counties with lower correlations include countries that are not in the EU (Turkey, the Ukraine) or countries that joined the EU later (Estonia, Lithuania, Croatia, Bulgaria). We can hypothesize that immigration is not perceived as emotionally in the countries that joined later (most of the data have been collected before the current migration crisis); therefore, the fear of immigrants may not be as intense, and the correlation with the underlying factor of feeling secure is not as high.


Introduction
In 2015, according to the International Organization for Migration (IMO), the number of migrants was the highest ever recorded, reaching 244 million. Previous estimates were150 million in 2000 and 214 million in 2010. The results of Gallup's World Poll survey carried out for the IMO (Esipova et al., 2015), based on interviews conducted between 2012 and 2014 with over 183,000 adults from more than 140 countries, were different from the prevailing negative perceptions of migration in the media highlighting that, "people are more likely to want immigration levels in their countries to either stay at the present level or to increase, rather than to decrease -with the important exception of Europe" (2015, p. 1). Hainmueller & Hopkins (2013, p. 2), in their review that included approximately one hundred studies on immigration attitudes, recognize two broad traditions: the first tradition, called the "political economy", explains immigration attitudes "with reference to native-born citizens' individual self-interest" and focuses on topics such as labor market competition and the fiscal impacts of immigration. The second tradition, called "socio-psychological", is less unified than the previous tradition and "emphasizes the role of group-related attitudes and symbols in shaping immigration attitudes". Immigrants, seemingly different from the native population's perspective, are particularly easy targets for ingroupoutgroup stereotypes and prejudices. However, this does not always mean that positive sentiments toward ingroup features are automatically inversely accompanied with hostility toward outgroup features, as proposed by Sumner (1906) and his predecessors (Bizumic, 2014). Brewer's (1999, p. 433) review of findings from both cross-cultural research and laboratory experiments, based on Allport's (1954) recognition, argues that humans, "as a species have evolved to rely on cooperation, rather than strength, and on social learning rather than instinct as basic adaptation". When contact becomes regular (Intergroup Contact Theory, Pettigrew, 1998;Pettigrew & Tropp, 2006), the levels of acceptance become higher. Despite these expectations, the relationships among the majority of immigrants obviously incorporates a serious extent of rivalry and competitiveness, consistent with the propositions of Sherif's Realistic Conflict Theory (1953) and Tajfel's Social Identity Theory (1979). In addition, some amount of thoughtfulness is moderated by the level of perceived cultural similarity as expected according to the Belief Congruence Theory (Rokeach, Smith & Evans, 1960) and the Terror Management Theory (Solomon, Greenberg, & Pyszczynski, 1991).
The "constrict theory" (Putnam, 2007) is more straightforward, and states that ethnic diversity (Putnam used US data), especially with targets that are immigrants, may foster social distance to the point of isolation and reduce trust among the ethnically dissimilar populations and also among homologous communities and their corresponding neighbors. Although Lancee & Dronkers (2008) confirmed Putnam's hypothesis in a large sample of respondents in the Netherlands, Gesthuisen, van der Meer & Scheepers (2009) did not find support for this proposition in their analyses based on Eurobarometer 62.2/2004 data collected in 28 European countries. Instead, they found that economic inequality and a national history of democracy were more important in the explanation of cross-national differences. Hooghe et al. (2009)  In our study, we will concentrate on the connection between trust in people and attitudes towards immigrants. Smith's (2010, p. 455) review of the contributions to the concepts of moral philosophers, political scientists, psychologists, and sociologists reveals three primary conceptualizations: generalized, particularized, and strategic trust. In the ESS questionnaires (like in the General Social Survey or the World Values Survey), generalized (social) trust is the object of interest. Rotter called it "interpersonal trust" and defined it as "a generalized expectancy held by an individual who the word, promise, oral or written statement of another individual or group can be relied on" (Rotter 1980, p. 1). The usual means of measuring generalized trust represents the direct question; whether people can be trusted. The first systematic critique of this method of measurement (Glaeser, Laibson, Scheinkman, Soutter, 2000) came from the discovery that behavior in a trust game did not correspond with survey measures of trust. Several replications of this study engendered arguments either for or against this result and also prompted efforts to improve the measurement (Naef, Schupp, 2009). To cope with the problem, generalized trust in people in the ESS core module questionnaire is approached from three different perspectives: "Generally speaking, would you say that most people can be trusted, or that you can't be too careful in dealing with people? Please tell me on a score of 0 to 10, where 0 means you can't be too careful and 10 means that most people can be trusted." "Do you think that most people would try to take advantage of you if they got the chance, or would they try to be fair? Please tell me on a score of 0 to 10, where 0 means that most people would try to take advantage on you and 10 means that most people would try to be fair".
"Would you say that most of the time people try to be helpful or that they are mostly looking out for themselves? Please tell me on a score of 0 to 10, where 0 means that people mostly look for themselves and 10 means that people mostly try to be helpful".

Participants and procedure
We used data from the European Social Survey, particularly the ESS1-7 cumulative file. The sample for the present study included adults (individuals aged over 15) in representative samples from 32 countries. In total, the sample consisted of N = 331,871 individuals who participated in 7 rounds of the ESS between the years 2002 and 2014. The mean age was 47.7 years (standard deviation [SD] = 18.58), and 46% of the participants were male.

Measures
Attitudes towards immigrants. Participants were asked about their beliefs according to the 3 statements described above using a 4-point response scale (1 = allow many; 2 = allow some; 3 = allow few; 4 = allow none). The Cronbach's α reliability coefficient of this 3-item scale was 0.872.

Attitudes towards immigration.
Participants were asked about their beliefs according to the 3 statements described above using an 11-point response scale (e.g., "Would you say it is generally bad or good for [country]'s economy that people come to live here from other countries?" where 0 = bad and 10 = good). The Cronbach's α reliability coefficient of this 3item scale was 0.851. Trust in people. Participants rated the 3 statements described above using an 11-point response scale. The Cronbach's α reliability coefficient of this 3-item scale was 0.780.

Data analysis
Structural equation modeling is a multivariate method that enables researchers to conduct a more precise analysis of the data than linear regression or factor analysis. SEM uses latent variables to account for measurement errors and allows for the simultaneous evaluation of multiple relationships in the model, which is not possible using linear regression.
We used the maximum likelihood estimation method to estimate the parameters of the model. We did not modify any parameters to improve the fit of the model. For the model fit, we used CFI and RMSEA indices. We did not use Chi square statistics for the model fit because of the large sample size. We did not exclude any data from the original sample and did not modify the original ESS dataset in any way.
We used multi-group modeling to allow data from different countries to be handled independently. We did not differentiate between rounds because every round was administered to a different set of respondents.

Results
We constructed a structural equations model comprised of three intercorrelated latent variables. Each latent variable was supported by three items from the ESS questionnaire. Latent variables represented factors and factor loadings for all items that were higher than 0.65; suggesting satisfactory factor stability for all three factors. Individual factor loadings are shown in Figure  1.
These results show that the questions in the ESS questionnaire that targeted respective attitudes and values (immigrants, immigration, trust) are consistent and that each triplet measures a common underlying factor.
We performed structural equations modeling (SEM) with three latent variablesattitudes towards immigrants, attitudes towards immigration and trust in people. We studied the correlations between the variables as depicted in Figure 1. This shows a correlation between trust in people and the respondents' attitudes towards migrants and migration. These constructs correlate positively, meaning that lower levels of trust in people correlate with more negative attitudes towards immigrants and immigration. More negative attitudes towards immigrants correlate with more negative attitudes towards immigration. (Note that polarity of items differs.) In the second step, we performed a multi-group SEM analysis. We introduced 32 groups, with each group representing data from one country. First, we tested an unconstrained model. We obtained 32 triplets of correlations as shown in Table 1. The fit of this model was satisfactory. RMSEA was 0.006, and CFI was 0.992. The strengths of the correlations varied, but all of the correlations had matching signs. Additionally, in every country the correlation between attitudes towards immigrants (F1) and attitudes towards immigration (F2) was always the highest of the three in absolute terms. The correlation between attitudes towards immigration (F2) and trust in people (F3) was the second highest, and the correlation between attitudes towards immigration (F1) and trust in people (F3) was always the lowest.
This indicates a consistent pattern of relationships between the three latent variables across the wide range of countries, and adds strength to our finding that levels of trust in people are universally related to attitudes towards immigrants and immigration.
We then put constraints on every correlation coefficient to see whether the countries differ significantly in the strengths of correlations. We built three models, forcing co-variances between pairs of latent variables to be equal across the countries, and compared the models using the Chi square test. All three models compared to the unconstrained model had p<0.001, indicating significant differences between the constrained and the unconstrained models. We can conclude that there are significant differences in correlation strengths among the countries.
This implies that although there is an overall pattern of correlations between the three latent variables, there are significant differences between respective countries. In some countries, the trust and attitudes towards immigrants/immigration do not seem to correlate as strongly as in other countries.

Discussion and Conclusions
Some political scientists have described the post-Cold War world as a "clash of civilizations" (Huntington, 1996), where cultural and religious differences will become the primary source of conflict rather than economic and political interests; in particular between the Muslim and Western worlds. According to Inglehart and Norris (2003, p. 63), based on the World Values Survey 1995-2001 data, "Samuel Huntington was only half right. The cultural fault line that divides the West and the Muslim world is not about democracy, but sex. According to a new survey, Muslims and their Western counterparts want democracy, yet they are worlds apart when it comes to attitudes toward divorce, abortion, gender equality, and gay rights -which may not bode well for democracy's future in the Middle East". It seems necessary to mention this because we need to consider that in the present wave of migration into Europe, the influx comes mostly from the Muslim countries of Northern Africa, the Middle East, Afghanistan, and Pakistan. In this context, the perceived cultural and religious differences between natives and migrants are especially visible and acute (Dummett, 2001). The data from the European Social Survey provide an opportunity to test the interconnections between generalized trust in people and migrant attitudes on a large sample of adults from several countries over the course of a relatively long time period (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014). Our results indicate a firm relationship between trust in people and attitudes towards immigrants and immigration. Similar research has been completed by Halapuu et al. (2013Halapuu et al. ( , 2017 regarding trust towards institutions and attitudes towards immigrants, but on a smaller data sample. These findings indicate that trusting people perceive immigrants more positively. We may hypothesize that an underlying factor of feeling secure may exist that enables people to trust and not to fear. Our analysis also shows that although correlation signs and strengths are consistent among people in various countries, significant differences remain in terms of how strongly trust in people and attitudes towards immigrants and immigration are correlated. When we look at an ordered list of the countries according to the strength of correlation between attitudes towards immigration, we find that counties with higher correlations are mostly "older" countries of the EU like Germany, Italy, Belgium, the Netherlands and France. Counties with lower correlations include countries that are not in the EU (Turkey, the Ukraine) or countries that joined the EU later (Estonia, Lithuania, Croatia, Bulgaria). We can hypothesize that immigration is not perceived so emotionally in the countries that joined later (most of the data had been collected before the current migration crisis); therefore, the fear of immigrants may not be as intense, and the correlation with the underlying factor of feeling secure is not as high.
Other authors (Hatton, 2017) advise that other factors be taken into consideration. Historically, trust in people has essentially been built on close personal experience; we still typically trust our family members and friends more than strangers. Even though (as theorised back to Hobbes and Locke) the importance of trust in people was repeatedly emphasized as an essential part of civilization, in the process of civilization, "service industries" transfigure immediate interpersonal contacts on their impersonal/functional form in a step-by-step manner. Not surprisingly, considering the rapid process of urbanization (72% of the EU population live in cities), this way of social life is described as a community of strangers (Wirth, 1937;Baumann, 1993). The belief in whether strangers can be trusted or not is seemingly misleading, not only in the traditional sense, but also among other things because both are psychologically expensive; improper distrust is an equally corrosive force as improper trust. Despite the tendencies described above, our results confirm the lasting role of general trust in people as a "social compass" in a changeable world: "Trust is the chicken soup of social life. It brings us all sorts of good things, from a willingness to get involved in our communities to higher rates of economic growth and, ultimately, to satisfaction with government performance, to making daily life more pleasant. Yet, like chicken soup, it appears to work somewhat mysteriously. It might seem that we can only develop trust in people we know. Yet, trust's benefits come when we put faith in strangers" (Uslaner, 2002, p. 1).