How are attitudes towards immigrants in Europe shaped by regional contexts?

The relationship between the spatial location of immigrants and attitudes towards immigration remains a key issue in the migration research agenda (Kaufmann and Harris, 2015; Tam Cho and Baer, 2011) and, more generally, in the public and policy debate. Specifically, a key question is whether an increase in the physical presence of immigrants – in the streets, in the neighbourhoods, at work, on the bus, at school – exerts a positive or a negative effect on how the majoritarian populations in destination countries perceive these immigrants.

Studies find that individuals who reside in neighbourhoods or small areas (i.e. provinces, small regions) with a higher immigrant concentration have, in general, more positive views towards immigration compared to individuals who live in native majoritarian areas (for an overview see Kaufmann and Harris (2015)). Contact theory, which states that prejudice can reduce by means of (positive) contact with others, is a commonly used framework for explaining this finding (Pettigrew, 1998; Pettigrew and Tropp, 2006). However, this effect might not always occur. Following the arguments posed by Quillian (1995), we argue that the socioeconomic characteristics of the area might affect this relationship (see also Branton and Jones, 2005; Hjerm, 2009). While contact with immigrants might be positive for attitudes in areas that are better off – e.g. with more availability of jobs, better infrastructure, etc. – a higher presence of immigrants might bring no or even a negative effect on attitudes in areas with poorer (i.e. scarcer) resources. Competition or threat – rather than positive contact – would be the main explanatory effect (Blumer, 1958).

Using data from the European Social Survey (2014), a cross-national dataset covering the most common destination countries in Europe, in combination with aggregated data collected at the NUTS31 level (small regions), the study answers the following questions:

What is the relationship between attitudes towards immigration and immigrant concentration in the area of residence?
Does this relationship vary according to the socioeconomic levels of the areas? If so, in which way?

Our study is innovative in different ways. First, to our knowledge, this question has not been explored yet at the European level and with such small geographies. On the one hand, previous studies based on cross-national data from Europe in combination with aggregated data collected at different NUTS levels (including NUTS3) explore only the effect of immigrant concentration (Semyonov and Glikman, 2009; Weber, 2015; Markaki and Longhi, 2013), but not how it interacts with the socioeconomic characteristics of areas. On the other hand, while the highly cited work by Quillian (1995) and, more recently, the work by Hjerm (2007) – also based on cross-national data – do explore the interaction between immigrant concentration and socioeconomic characteristics of areas, this is done only at the country level.

The second innovative aspect of this study refers precisely to the type of the geographical areas we study. Thanks to permission granted by the board of European Social Survey, we use NUTS3 rather than countries or other intermediate geographies (which are often a much commonly used by researchers, given their accessibility). A recent study shows that, in general, the lower the geographical measure we use to measure the concentration of immigrants, the more positive attitudes are (Kaufmann and Harris, 2015), a phenomenon that is attributed to contact theory (Pettigrew, 1998). Indeed, Weber (2015) finds a positive effect at the NUTS3 level with data from the European Values Study (2008). By using NUTS3 geographies, we intentionally wanted to confirm the finding that concentration plays a positive role on attitudes (which we do), and then see how (and if) this relationship changes when we move from more to less socioeconomically affluent contexts (which we also do). In other words, if contact theory appears as a potential relevant mechanism, then the next step – and main aim of this article – is to disentangle in which contexts it is.

The third value added by our study, finally, regards how we deal with the conditionality, which is not only relevant from a methodological perspective, but also from a theoretical point of view. The striking majority of research base their findings on the interpretation of the interaction between immigrant concentration and socioeconomic context (e.g. Quillian, 1995; Branton and Jones, 2005; Hjerm, 2007; Hjerm, 2009; Oliver and Mendelberg, 2000). In other words, they generalise the findings to the entire distribution. However, when studying interaction terms it is important to evaluate whether the effect is statistically significant throughout the distribution or only in one part of it (Brambor et al., 2006). This is relevant, because it allows to conclude in which contexts relationships work and in which contexts they do not.

Our findings reveal an interaction effect between immigrant concentration and socioeconomic conditions of neighbourhoods, which goes in the expected direction. However, and in line with the third value added of this study, we show that this interaction effect is significant only after a certain threshold of socioeconomic conditions has been crossed. In other words, while the increase in the immigrant concentration has a more positive effect on attitudes in better-off areas compared to areas with more intermediate socioeconomic conditions, in the poorest areas, the increase in the concentration of immigrants has no effect on attitudes.

The paper is organized as follows. Section 2 presents the theory and previous studies, with a particular focus on cross-national studies and those that have dealt with the interaction between immigrant and socioeconomic neighbourhood compositions. In section 3 we present the data and methodology. Sections 4 and 5 present, respectively, our empirical findings and robustness checks. Finally, we conclude and briefly discuss policy implications of our study in section 6.

This is an introduction to a working paper by Rezart Hoxhaj and Carolina V. Zuccotti.