Electoral Participation of Non‐National EU Citizens in France: The Case of the Nord

Since the 1993 Maastricht Treaty, EU citizens have the right to vote in European and local elections in the member state they reside in. In France, only about a quarter do so. Our article considers what factors explain the registration and participation of non‐national citizens for the French Department of the Nord where around 35,000 non‐French European citizens of voting age are living. Among them, 11,638 are registered to vote in the French municipal elections. Following the 2020 municipal elections, we have consulted the electoral rolls in each of the 648 communes to know who actually cast a vote. Based on detailed census data on each EU nationality and on other information contained on the electoral lists and rolls (age, gender, place of birth, etc.) and also contextual variables, this article seeks to identify the main factors associated with registering in the first instance and turning out to vote in the second. Our results confirm wide variation in registration and voting rates according to nationality. They also show that beyond voters’ nationality and the “usual suspects” of electoral participation, contextual factors are important predictors.


Introduction
The transnationalisation of electoral rights is a major challenge for contemporary electoral studies and migration studies alike (Bauböck, 2006;Lafleur, 2013).From this perspective, the European political space offers an excellent field of observation.Indeed, although the Maastricht Treaty granted EU citizens the right to vote and run for office in both European and local elections in the member state where they reside (Shaw, 2007;Strudel, 2009), we know little about the ways in which citizens make use of such rights.With the expansion of intra-European migration, these political rights concern a growing number of individuals; there are now around 15 million EU citizens of voting age living in a member state other than the one(s) in which they hold the nationality.These mobile EU citizens form a highly disparate population in terms of both their social characteristics and their migratory trajectories, be they Erasmus students, low-cost manual workers, highly skilled and cosmopolitan employees of major metropolises, border residents, or heliotropic retirees (Recchi & Favell, 2019).Among the various member states, France is a major host of intra-European immigration, with around 1.2 million non-French European citizens of voting age, making it the second-largest host country in Europe.
Quantitatively, only a small proportion of non-national EU citizens (NNEUCs) make use of these political rights.
It is estimated that, on average, a minority of NNEUCs are actually registered to vote in their host country (Hutcheson & Russo, 2019); 27.2% of NNEUCs living in France are registered to vote in municipal elections and 22.6% in European elections (Gouard & Lombard, 2023).This article examines which social and political factors contribute to NNEUC registration and turnout behaviour.To answer this question, we make use of census data on each EU nationality and information contained on the electoral rolls as well as contextual data.We conceptualise electoral participation as a two-step process (registration is a first signal of political engagement, but may not lead to an actual vote).While previous studies have addressed participation at the macro and meso levels, we are also able to examine individual-level data for the vote.To do so, we draw on a new and unique dataset that records the actual turnout behaviour of all NNEUCs registered to vote at the 2020 municipal elections in the French Department of the Nord (hereafter the Nord).We consulted the electoral lists and systematically compared them with the signatures on the electoral rolls.The Nord was selected because it is the largest in France in terms of its population and also shares a border with Belgium, allowing us to work with a population of around 35,000 non-French European citizens of voting age, a third of whom are Belgian nationals.Our study thus addresses a theoretical issue in the literature by explaining registration and voting by different EU nationalities within the same country.This approach mirrors and complements other studies in this thematic issue which by contrast look at the enfranchisement and vote of migrants of a single nationality in their home country (Finn & Ramaciotti, 2024;Gherghina & Basarabă, 2024).In addition, we address a major empirical gap as most individual-level studies use survey data (Koc Michalska & Strudel, 2012) which are declaration-based and involve a recall bias.Our empirical data, on the contrary, is based on an objective measure of voter registration and recorded voting through access to the post-election rolls.

Theory
There are three generally acknowledged opposing theoretical views explaining migrants' political (re)socialisation (M.Voicu & Rusu, 2012;White et al., 2008) that can be applied to their electoral participation.First, pre-migration attitudes and behaviours may be resistant to change, leading migrants to reproduce practices acquired via socialisation in their home country.Alternatively, they may adopt the standards of the host country following exposure to the new political system.Third and somewhat as a midway, their political resocialisation may be bound to the transfer of their beliefs and predispositions to the host social context and polity.We consider these three strands of inquiry and apply them to the electoral participation of NNEUCs: resistance is associated with the prevalence of the political culture of origin, exposure suggests that contextual elements in the host political system are likely to affect turnout, while transfer applies to situations whereby resources allowing participation at home remain largely unchanged abroad.Thus, their probability to vote is explained by the political socialisation process of migrants not only in their country of origin (Eckstein, 1988) but also in that of residence (B.Voicu & Comsa, 2014) or in the interaction between the two (Ciornei & Østergaard-Nielsen, 2020).
The convergent literature finds that distinct political cultures of origin or diverse migration pathways (settlement vs. temporary/lifestyle vs. economic) explain differing outcomes and potential resistance to seizing voting rights (Blokland et al., 2023;de Rooij, 2012;Favell, 2008).Because preferences regarding politics may be specific to a given political community (Inglehart, 1988), we argue that specific characteristics of the country of origin are likely to influence the use of political rights post-migration.
Building on this literature, and given that NNEUCs all come from democracies (with minor exceptions), we should not expect differences in participation to be based on the lack of a voting culture, nor that they are "less culturally developed than local workers and therefore also less politically active" (Martiniello, 2006, p. 86).Nevertheless, differences have been witnessed across countries-and in particular Central and Eastern European countries' vote is tendentially lower (Kostelka, 2017;Thomassen, 2005)-leading us to expect similar trends in the destination country.For example, those citizens coming from countries where compulsory voting is applied and those where trends in participation are higher may have developed a "habit of voting" (Gerber et al., 2003).In that regard, Collard's (2010Collard's ( , 2016) )  Exposure theory predicts that the longer immigrants stay in a new country and the more they develop contacts with the institutions, people, and values of that country, the more they are likely to adapt their attitudes and behaviour to this new political environment.If such is the case, the political and electoral participation of foreigners could also be favoured or, on the contrary, disfavoured, by various contextual factors at the national or local level.In particular, the procedural and practical barriers to electoral participation as identified in the literature include the migrants' initial lack of knowledge of their political rights and interest in their host country's political life (Collard, 2016;Ostling, 2019;Shaw, 2010), but also language barriers and the reluctance or passiveness of local authorities.Targeted public policies have been evidenced to encourage the participation of foreigners, particularly those put in place by local authorities (Morales & Giugni, 2011).Political forces situated on the left of the political spectrum are usually seen as more supportive of migrants and thus inclined to grant them participation rights (Koopmans et al., 2005;Van Heelsum, 2001) and to put in place more initiatives in that regard (Nikolic, 2017).More generally, the political "climate" of the local environment can be more or less welcoming or, on the contrary, hostile to the electoral participation of foreigners.Earlier studies highlight a mobilisation of immigrant populations linked to the presence of the extreme right (Richard, 1998).Césari (1993), for example, points to a surge in participation by immigrant populations in Marseille in response to the perception of such a "political threat."Similarly and based on theories of descriptive and symbolic representation, the role of political parties, and more precisely, their choice to designate candidates of specific nationalities can boost participation.We, therefore, expect that participation will be higher in those communes where European candidates are running, following a type of "candidate effect" (Fiva & Smith, 2017): H2a: The longer their stay in the host country, the higher the electoral participation of NNEUCs.
H2b: Left-led communes are linked to a higher electoral participation of NNEUCs.
H2c: More radical right-leaning communes lead to higher electoral participation of NNEUCs.
H2d: The presence of a co-national candidate(s) leads to higher electoral participation of NNEUCs.
H2e: More politically active communes are associated with higher electoral participation of NNEUCs.
Finally, civic resources and experiences developed in the home country may be transferred to the host one.
As the link between individual resources and individual political participation is well established, there is in fact little reason to a priori dismiss the "usual suspects" of political and electoral participation in the case of NNEUCs.To be sure, when moving, migrants may change their socio-economic status (this may even be one of the reasons for migration) but this is unlikely to change the main tenets of electoral participation.
Several factors explaining participation are often put forward in the literature to explain low turnout, including age, the level of education, and the socio-economic situation of individuals (Braconnier et al., 2017;Lazarsfeld et al., 2021;Mayer & Boy, 1997).In the migration literature too, the level of socio-economic resources is often given as a main explanatory factor of political participation (Blokland et al., 2023;de Rooij, 2012).Civic skills may further be transferred as part of social bonds.Some studies have found national population density to have a positive effect on the vote (Tam Cho et al., 2006), perhaps because the individual's social network in the host country may play a decisive role in raising awareness of their political rights (Rea et al., 2015).Qualitative outlooks have also underlined the role of local associative structuring in boosting participation (Raffini, 2012).Last, the effect of perduring social bonds may be particularly prevalent when the population can easily move back and forth between the country of origin and of residence as in the case of cross-border migrations.Our third hypothesis is therefore multipart, expecting traditional socio-economic factors to influence participation as well as an influence on the anchorage of the community of origin: H3a: Older and more educated NNEUCs are associated with higher rates of participation.
H3b: The stronger anchorage of a given community in a given commune leads to higher participation for that community.

Data Collection
We consider two facets of electoral participation: registration and turnout behaviour (voting).In France, as in most other countries, electoral participation requires individuals to be registered in advance.While French citizens are automatically registered upon the legal majority, NNEUCs must make the active choice to register, after which they remain registered for the next municipal elections.Once registered, however, there is no requirement that an individual exercise their right to vote.We can therefore consider both registration and voting as forms of participation.
One unique aspect of the French system is that the French Ministry of the Interior makes voter registration information available.This allows us to determine the nationality and commune of all registered voters.
The Nord is particularly interesting to study for two main reasons: (a) it is one of the French departments with the highest number of NNEUCs, with a majority national group from a Northern European country (Belgians), whereas on average in France, the Portuguese, Italians, and Spaniards make up two-thirds of NNEUCs (see Table 1); (b) this department has a very large number of communes (648), which makes it possible to carry out robust statistical analyses using the commune level.

Registration Data
Among the around 35,000 non-French European citizens of voting age in the the Nord, 11,638 are registered on the lists for the French municipal elections.Analyses of this file were coupled with analysis of census data from the French National Institute for Statistics and Economic Studies (INSEE) for the various European populations of voting age.Census data allowed us, for each commune of more than 5,000 inhabitants, to know the socio-demographic composition of the NNEUCs population in 2019 (sex, age group, length of residence in France, diploma, occupation) and commune-level data.Calculating registration rates at the communal level can offer one difficulty: A person who has changed residence may remain registered in her former commune because each voter is expected to update her situation following a change of residence.It is estimated that 15% of French voters are registered in a commune where they do not currently reside (Braconnier et al., 2016).
Data was obtained via the network for research Quételet-Progedo.
Table 1 presents the main demographic and electoral data for the Nord, comparing them with national averages for each of the nine best-represented European nationalities.It reveals important inequalities in registration rates for local elections according to the country of origin.Schematically, both in France and in the Nord, we notice three blocks of countries.The first block corresponds to citizens from Northern Europe (Belgian, Dutch, German) who display high registration rates.The intermediate group is composed of citizens from Southern Europe (Italians, Portuguese, Spanish).The last group is composed of citizens from Central and Eastern European countries (Bulgarians, Polish, Romanians) for whom registration rates are the lowest.
Inequalities in registration rates between the different groups of citizens according to their country of origin could be explained by other factors, notably social characteristics.Our research attempts to unveil the different profiles behind these inequalities in registration.

Voting Data
French law allows for the voter rolls to be examined for a very limited period following an election (usually for 10 days).Each voter is required to sign next to their name on the roll when they cast a vote; thus, we were able to determine voter turnout based on these rolls.We considered the rolls of 508 of the 648 communes in the Nord.112 communes were excluded because there were no NNEUCs registered and for 28 communes the rolls were unavailable.In some communes, pages were missing which excluded a further 845 individuals.
Nevertheless, we were able to determine voting behaviour for 9,999 of the 11,638 NNEUCs registered voters.This labour-intensive process involved photographing and examining each page of the voter rolls.
Municipal elections in France have the following basic characteristics which have consequences in terms of our analysis.In communes of over 1,000 inhabitants, there is a two-round propotional representation system with a bonus (which allows the candidate list with a relative majority of votes to gain a majority of seats overall).In smaller communes (less than 1,000 inhabitants), municipal councillors are elected following a two-round plurinominal majority vote where panachage is possible.Of the 648 communes, 279, or 43.1%, are smaller than 1,000 residents, while 94.2% of the department's population reside in communes greater than 1,000 residents.1).For privacy reasons, data is only provided for communes of at least 5,000 residents, meaning that communes below this threshold are excluded.We further exclude from the analysis any nationality-commune dyads where there are fewer than 20 national residents of voting age.We make this choice to avoid a single registration having too large an impact on the overall registration rate within the commune.In total, there are 234 nationality-commune dyads that fulfil the study criteria.These represent 82 unique communes with 2.9 nationalities per commune on average.In sum, the dependent variable is the registration rate of a specific nationality in a specific commune.The number of cases per nationality is as follows: 59 for Belgian and Italian, 47 for Portuguese, 24 for Spanish, 15 for Romanian, 11 for Polish, 10 for German, five for Dutch, and four for Bulgarian.
The census data breaks down the population of each nationality by age, education, socio-professional category, gender, and length of residence.We use these data to determine the characteristics of the national population in each commune.We break down education as the percentage of having completed at least two years of post-secondary education.For the socio-professional category, we consider the share of the working class among the active population.For gender, we consider the percentage of women among the adult population.
For age, we calculate the percentage aged 60 years or more.
At the municipal level, we also gather contextual information about the commune itself to account for local anchorage.We include a dichotomous variable for communes which are located on the Belgian border (coded 1,  = 76) with the aim of incorporating a territorial variable that would reflect the presence of a particularly participatory civic culture just across the border (i.e., compulsory voting in Belgium).
We conducted an exhaustive search of the National Associations Register to identify associations targeting one or more of the relevant nationalities and generate a dichotomous variable indicating the presence of national group associations.In the Nord, we identified 245 national group associations from the National Associations Register.We also calculate the density of each nationality retained in each commune.
At the municipal level, we also include information on the political climate of the commune which could impact registration.Using the results of the 2017 presidential election, we include a variable measuring the percentage of votes in the first round for the far-right candidate Marine Le Pen.We include a variable for the political orientation of the municipal majority put in place after the last French municipal elections in 2014 for communes of over 1,000 inhabitants (using the labels provided by the Ministry of the Interior) since we expect left-wing majorities to be more proactive in encouraging participation.Finally, we consulted the candidate lists to detect if there is one or more conational candidates in the commune.On top of the above-described gender and socio-professional categories, we include a control variable for commune size regrouped into four modalities (5,000-7,499; 7500-9,999; 10,000-19,999; ≥20,000) since participation is generally higher in rural than in urban areas, leading us to expect a commune size effect (Nevers, 2008).

Voting
Voting is considered a dichotomous variable taking on the value of (1) when there was a signature (evidencing participation) of a given registered voter on the electoral rolls of her commune.We focused on the first round only, so each voter is accounted for only once.Only 92 communes held a second round and those are mainly large urban municipalities, not allowing us to test several of our hypotheses.In our dataset, among the registered voters, 3,750 voters voted and 6,249 abstained.
To explain (non-)participation at the individual level, we consider data included on the electoral lists and rolls, such as nationality, gender, and date and place of birth.Because we believe that the relationship between voting and age is represented by an inverse U, we also include the quadratic term for age.Neither the voter lists nor the electoral rolls allow us to know how long registered voters have lived in France.However, their country of birth is indicated, and this variable is therefore taken as an approximation of the length of residence in France.
At the municipal level, we include-as for registration-the three transfer variables (border, associations, and density of the national group), but also the exposure ones (the percentage of Le Pen vote in 2017 in the commune, the municipal majority put in place after the last French municipal elections in 2014, and whether there was one or several candidates of the same nationality of the voter running in the municipal election).
To this category, we add the turnout level in the election.Although the Covid-19 epidemic caused low turnout in the 2020 French municipal elections, the social logic of voting remained broadly the same (Audemard & Gouard, 2022), with only the effect of age being undermined (Haute et al., 2021).We also recode communes touching the Belgian border ( = 76).At the municipal level, commune size has been regrouped into six sizes (<1,000; 1,000-4,999; 5,000-9,999; 10,000-19,999; 20,000-49,999; ≥50,000).In the analyses at the individual level, we are able to include a more detailed examination of commune size due to the larger number of observations.A summary of the hypotheses, levels of analyses, and data used is presented in Table 2.

Models
To assess the factors impacting electoral registration, we employ a generalised linear model with Gaussian distribution and identity link.To account for heteroscedasticity and the non-independence of variation among nationalities, we cluster standard errors by nationality.One challenge presented by this model is the relatively low number of observations ( = 234) which is linked to the data constraints described above.
Overall, fewer observations are linked to higher standard errors, increasing the risk of type II errors but not type I which is reassuring for the reliability of results obtained.However, aggregating also raises the question of ecological fallacy.While we should be cautious not to assume that population characteristics are true of any particular registered voter, this does not prevent us from using this aggregated data in an attempt to falsify our hypothesis that socio-economic factors are linked to electoral registration among NNEUCs.
Second, voting is assessed on an individual basis using a generalised linear model with binomial distribution and logit link.Standard errors are again clustered by nationality.We considered a mixed-effects model with random effects for nationality, but the residual intraclass correlation for nationality was close to zero, and the likelihood-ratio test showed that the mixed-effects model did not perform better than the fixed-effects model.
Because INSEE does not share municipal-level census data for municipalities of less than 5,000 residents, we are not able to calculate the density of the nationality for individuals residing in communes less than 5,000.
In order to not exclude these individuals, we run a set of analyses with and without the density variable.
In total, 486 municipalities are represented among the 9,827 electors included in this analysis (172 voters were excluded due to missing data).When considering only communes with over 5,000 residents, there are 6,268 European electors in the analysis, representing 101 unique communes.

Registration
Table 3 presents the results of analyses of electoral registration of NNEUCs.Model 1 considers the border effect for all nationalities while Model 2 limits the border variable to Belgians only.The results provide support for H1 which expects nationality to impact participation.We see strong nationality effects on registration; all else being equal, Belgians have the highest registration rates and Polish have the lowest.Moreover, the variation in predicted registration rate by nationality is quite large, with a 33-point difference between the Belgians and Polish.Figure 1 displays the predicted registration rate for each nationality when all other covariates are at their means.It is also interesting to note that the geographic grouping observed in Table 1 largely  As regards our exposure-related hypotheses, the coefficient on the length of residence variable is positive and significant, indicating that the higher the percentage of NNEUCs having resided for more than 20 years in France, the higher the registration rate for that nationality group in a given commune.As there are no automatic deregistration procedures, it is logical that the registration rate increases with the length of residence in France.For the political contextual factors, there is more limited evidence that they impact registration rates, possibly due to the cumulative nature of registration rates.There is no statistical difference between municipalities governed by centre, right, or left parties.Although there is lower registration when the municipal majority is non-classified, it is difficult to draw meaningful conclusions because this category is used for a wide range of candidate lists.Also, the presence of co-national candidate(s) and participation rates have no observable relationship to registration rates.The failure to observe an effect related to the presence of a co-national candidate may be linked to the fact that the increase in registration caused by the presence of a co-national candidate represents very few individuals compared to those who have already registered on the electoral lists prior to this election.Finally, the Le Pen vote share in 2017 is positive and significant in Model 1 but not Model 2, suggesting that the relationship is not robust to the specification of the border variable.
We find support for our third hypothesis concerning the variables for age and education which both have positive, significant coefficients.As the proportion of the population who is aged 60+ or has obtained at least two years of post-secondary education increases, the registration rate also increases.This is consistent with the current understanding of the determinants of electoral participation.In terms of our third hypothesis on contextual factors, we find mixed results.We find that as the size of the commune increases, the expected NNEUC registration rate decreases up to a point, falling from 38.1 for communes under 7,500 to 27.9 for communes between 10,000 and 20,000 inhabitants.However, the expected registration rate increases to 29.4% in communes larger than 20,000, though the relatively few communes of this size make the estimation less reliable.
As for transfer-related variables, the most surprising finding is perhaps the negative association between national group density and registration rate.That said, the presence of a national-group association is strongly associated with higher registration rates, suggesting that community-based socialisation is still important.Besides, registration rates are positively impacted in communes bordering Belgium and the magnitude of the predicted relationship increases when only considering the impact on Belgian nationality-perhaps because the compulsory nature of voting on the other side of the border has helped develop a habit of voting in communities who live on either side of the frontier.Other traditional determinants of political participation do not have an observable relationship with electoral registration.
The proportion of women and the share of the working class among the population do not significantly affect registration rates in this model, as we had already noticed for the national level.

Explaining the Vote Versus Abstention of NNEUCs
Table 4 reports the results of analyses on determinants of voting.Models 3 and 5 include a variable for the density of the nationality group whereas Models 4 and 6 exclude this variable.Additionally, we employ two specifications of the border variable; in Models 3 and 4, we consider the border residence for all individuals regardless of nationality whereas in Models 5 and 6 we only consider border residence for Belgians.Strong nationality effects remain, providing further evidence in support of our first hypothesis.Interestingly, we do not observe the same hierarchy of participation in voting as we do in registration.While Belgians were the highest-registered national group, their probability of voting was the lowest.Conversely, the Dutch, whose expected predicted registration rate in the Nord was rather low, have a relatively higher rate of participation.
Figure 2 shows the probability of voting by nationality and demonstrates that the geographical groupings observed in the results reported in Tables 1 and 3 largely do not hold.This suggests a different logic of voting versus registration, but we should also consider the potential role that the higher registration rate of Belgians may play.Because the Belgian registration rate is so high (46.2%), it may include voters who are not very politicised and, as a consequence, vote less.Conversely, it is possible that the small proportion of Bulgarians on the electoral lists (5.1%) form a particularly politicised segment who, once registered, vote widely.When considering our second group of hypotheses pertaining to exposure factors, we find mixed results.
Political factors perform better when predicting voting than registration.This is not surprising considering that voting occurs at a specific moment in time whereas registration may have occurred at any time.
We observe that vote share for the extreme right and right-led local governments are both associated with a lower probability of voting.In other words, when right-wing politics are more prevalent, NNEUCs vote less.
One reason for this may be that left-led communes engage more in "get out the vote campaigns" including foreign voters.We also see a strong effect associated with the presence of co-national candidate(s); NNEUCs are more likely to vote when a compatriot is on the ballot.This provides further evidence for the importance of exposure factors.
Turning to our third block of hypotheses, unlike in the analysis of registration, neither national group associations nor population density have a statistically significant relationship with voting.Similarly, living in a border commune does not seem to have an effect on voting, even for Belgians.We do not have information on education at an individual level, so we are not able to test its impact on voting.However, we do find that significant relationship between age and voting.The probability of voting increases with age up until the age of about 56 after which it begins to decline.Naturally, the overall participation rate is positively associated with the probability of voting, suggesting that the factors that influence French nationals to vote or abstain also influence NNEUCs.

Conclusion
This article has considered what are sometimes presented as competing explanations of migrants' political socialisation in their host country to examine how this has translated into their actual electoral participation.
Above all, the prevalence of nationality as a major explanatory factor of their differentiated participation undeniably points to the importance of resistance patterns among NNEUCs who tend to reproduce abroad their home practices.It confirms that it is still largely far-fetched to consider EU mobile citizens as a homogenous whole of "pioneers of European integration" (Recchi & Favell, 2009), but calls for a more fine-grained analysis to grasp their differentiated approaches to participation.Registration of NNEUCs-which may happen over a long period and is thus unsurprisingly less sensitive to contextual aspects-is further marked by a clear transfer of previously acquired civic skills.NNEUCs heavily rely in France on their imported personal resources, be they national-based (associations, density, proximity border) or more socio-economic (education and age).Voting, by contrast, is sensitive to contextual elements and implies that some local practices do matter in explaining migrant's participation.Hence, the participation trends, the political orientation of the commune, and the presence of co-national candidates that did not affect registration do impact voting.All in all, our results show that, when explaining migrant's political participation, various socialisation rationales (transfer, exposure, resistance) are not mutually exclusive but respond to different circumstances.The two-step participation process in use in many countries (where voters first have to register and then go to the polls) essentially means that different logics may prevail.
Empirically, our study brings to the forefront participation behaviours.To the best of our knowledge, few research pieces have ever had access to data on actual voting behaviours, most focusing either on recall questions or attitudes.Thus, gaining access to electoral rolls has provided invaluable insights.Our future research will compare the Nord with six other departments where similar data was collected.Including data for other departments will offer several advantages.First, it will allow us to further examine certain variables.
In particular, proximity to the border could be tested for other nationalities (Germans, Italians, Spaniards).
Second, it will make the results more robust thanks to a larger number of cases processed.But above all, a larger number of cases will enable us to consider how national origin may moderate the effects of other variables or be moderated by them.By considering not only civic culture but also the migration paths and social configurations of residence in France of different European citizens, we can deepen our understanding of how European migrants participate in political life in their host countries.
initial research into NNEUC registration in French municipal elections in 2001 and 2008 highlights significant national inequalities.Our first hypothesis, thus, posits that national origins condition electoral participation: H1: National origin will have a relationship with electoral participation among NNEUCs.

Figure 1 .
Figure 1.Predicted registration rate of NNEUCs in France by nationality with 95% confidence intervals.

Table 1 .
Electoral registration and main social characteristics of the nine largest EU nationalities: Comparison between the Nord and France.Because there is not a central database of every NNEUC resident, there is no way to individually compare registered and non-registered citizens.For this reason, variables corresponding to voter registration are aggregated at the communal level.We use data from the 2019 census provided by INSEE to calculate the registration rate as well as population-level characteristics for the nine largest European nationalities in the Nord (see Table Politics and Governance• 2024 • Volume 12 • Article 7507
Registering and voting: National associations that are part of the National Association Registry (counted per nationality); list of communes bordering Belgium (dichotomous); population density of the nationality based on 2019 census data

Table 3 .
Regression models: Analysis of registration rates, by nationality-commune dyad.

Table 4 .
Regression models: Analysis of voting among registered European voters from nine principal nationalities.

Table 4 .
(Cont.)Regression models: Analysis of voting among registered European voters from nine principal nationalities.