Voting home or abroad? Comparing migrants' electoral participation in countries of origin and of residence

Abstract

The political participation of immigrants has received increased scholarly attention over recent decades. However, comparisons between the electoral behavior of immigrants in their countries of origin and of residence are still limited. This article addresses this gap in the literature and seeks to identify the determinants of Romanian immigrants’ electoral participation in the local elections of four West European countries (Germany, France, Italy, and Spain) as compared to their turnout in their home country's legislative elections. Looking through the lenses of exposure theory, we hypothesize that contact with institutions, people, and values from the countries of residence are likely to have different effects in the two types of elections. We test the explanatory power of four main variables – time spent in the host country, social networks, degree of involvement in the local community, and the type of relationship with citizens of their host countries – to which we add a series of individual-level controls such as age, education, gender, and media exposure. To assess our claim, we employ binary logistic regression to analyze original web survey data collected in the summer of 2013. The result supports the empirical implications of exposure theory.

Keywords:

• immigrants
• voting
• country of origin
• country of residence
• exposure
• resilience

View correction statement:

Corrigendum

ORCID

Sergiu Gherghina http://orcid.org/0000-0002-6627-5598

Notes

1. The study also tested for other SES factors (e.g. occupation, interaction effects between age, education, and occupation) and knowledge of the language in the country of residence. They were not reported since they had no strong or significant effect.

2. There are several studies that provide estimates regarding the profile of Romanian immigrants to some of the four countries investigated here. In the absence of precise official data, we preferred not to rely on estimates because this could further distort the interpretation of results.

3. We have run models with voting as an ordinal variable corresponding to all possible answers. The results were fairly similar to those in the binary logistic regression and these are reported here for a more straightforward interpretation.

4. Here the odds-ratio is defined as the ratio of the probability of Y = 1 relative to the probability of Y = 0. More substantively, in our models, odds-ratio so that the coefficient estimate for variable k having odds-ratio means that, other things equal, the odds that respondents voted in a given election is times larger than the odds for not voting.