Protecting Us, Protecting Europe? Public Concern about Immigration and Declining Support for European Integration in Italy

Abstract

This article investigates the effect of immigration on attitudes towards Europe in Italy. This question resonates throughout the European Union (EU), but does so particularly in Italy – one of EU's most exposed external borders, especially under recent global challenges such as the economic crisis and the Arab Spring. Over the last two decades, Italy received more than four million migrants and refugees, which has put immigration on top of the political agenda ever since. Using data from the European Values Study (EVS), we find that concerns about immigration, especially in the context of the economic crisis of 2007, is a factor of declining support for European integration and that these effects vary across regions. In particular, we demonstrate the mediating effect of context by showing that the effect of being concerned about immigration is more corrosive for support in regions with higher immigration rates than in regions with lower rates.

Key words:

• Support for European integration
• immigration
• Italy
• regional level
• economic crisis

View correction statement:

Erratum

Notes

This article was originally published with errors. This version has been corrected. Please see erratum (http://dx.doi.org/10.1080/15705854.2014.900982)

¹ EVS ( 2010): European Values Study 2008: Italy (EVS, 2008). GESIS Data Archive, Cologne. ZA4755 Data file Version 1.0.0, doi:10.4232/1.10031.

² EVS (2006) (2012): EVS – European Values Study 1999 – Italy. GESIS Data Archive, Cologne. ZA3780 Data file Version 3.0.1, doi:10.4232/1.11542.

³ Questions traditionally used in analysis of support for Europe, such as citizens' opinion about membership to the EU, image of the EU or whether the country benefited from its membership, are not available in the 2008 EVS data.

⁴ The results of the factor analysis are not shown here, but are available on request.

⁵ The results of the factor analysis are not shown here, but are available on request.

As a robustness check, the analysis was replicated using Mokken scaling, and lead to the same results: two latent traits, one capturing attitudes towards immigration related to the economy, the other related to ethnicity.

⁶ Marginal effects here were calculated using the method described by Brambor, Clark & Golder (2006).

For instance, for Model 2A: attitudes towards immigration (economic dimension) and regional immigration are the two independent variables of interest, attitudes tw. immigration*regional immigration is the interaction term:

Marginal effect of attitudes towards immigration on support for Europe=b _{attitudes tw. immigration} +b _{attitudes tw. immigration*regional immigration} *regional immigration

Standard error of the marginal effect=(var. _{attitudes tw. immigration} +var. _{attitudes tw. immigration*regional immigration} *(regional immigration²) + 2*covar(_{attitudes tw. immigration, attitudes tw. immigration*regional immigration}) *regional immigration²).

⁷ Predicted value calculated for attitudes tw. immigration (economic dimension) and regional immigration at their maximum values, all other variables at their means, p value of .000 and confidence intervals of [1.46; 1.93].

⁸ Predicted value calculated for attitudes tw. immigration (economic dimension) and regional immigration at their maximum values, all other variables at their means, p value of .005 and confidence intervals of [.27; 1.49].

⁹ Predicted value calculated for attitudes tw. immigration (ethnic dimension) and regional immigration at their maximum values, all other variables at their means, p value of .000 and confidence intervals of [.80; 1.28].

¹⁰ Predicted value calculated for attitudes tw. immigration (ethnic dimension) and regional immigration at their maximum values, all other variables at their means, p value of .016 and confidence intervals of [–.54; –.006].

¹¹ Out of the 1,104 respondents included in the analysis, only 10% belong to the unskilled workers' category.

¹² R-squared obtained using the mlrsq command in Stata 12.