The tie that binds? A comparison of ethnicity-based party ties among emigrated and resident citizens

Recent decades have seen a trend toward enfranchising emigrated citizens in home country elections. Political parties have also become increasingly interested in connecting with emigrant voters. That said, we still know little of what voters think of the parties in the home country and how party preferences may change because of migration. On the one hand, research shows that the experience of migration and the context of the host country have a significant impact on the political behavior of migrants. On the other hand, party ties are known to be resistant to change. In this paper, we study how what is generally assumed to be the strongest of party ties, namely ties to an ethnic party, is affected by migration. Utilizing two highly comparable surveys of resident and non-resident citizens, we study how identifying with an ethnic minority party among Finland-Swedes in Finland, where they constitute a linguistic minority, compares with emigrated Finland-Swedes in Sweden, where they speak the majority language. We find that party ties, even with an ethnic party, tend to be weaker for emigrated citizens. However, the difference is relatively small and only materializes after an extended stay abroad.

Page 2 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 Emigrated citizens diverge as voters from citizens living in their home country in two important ways. First, they are residing within a new social and political reality, and a growing body of research on migrants shows that the experience of migration and the context of the host country has a significant impact on the political behavior of migrants (Lafleur & Sanchez-Dominguez, 2015;Ghergina, 2016;Ciornei & Østergaard, 2020;Dahlberg & Linde, 2018;Escobar et al., 2015). Second, non-resident citizens receive distinctive and more limited information about the political situation in the home country, compared to those residing there (Turcu & Urbatsch, 2019). This affects the ability of emigrated citizens to understand the latest developments in home country politics and may result in a different perspective on home country politics compared to resident citizens.
Research on party ties has, nevertheless, established that party ties are resistant to change. Partisans tend to stand by their preferred party, even when the candidates and the issues change (e.g., Dalton, 2016;Dassonnville & Hooghe, 2017). We also know that some party ties are more durable than others. Ethnicity-based party ties are generally assumed to trump ties based on other political preferences (Chandra, 2011). This also raises the question concerning how party ties are affected by emigration, meaning which factors remain influential for party ties after migration. Hence, the aim of the study is not only to examine whether party ties change because of migration but also how they change.
To provide answers to these questions, we examine the party ties of a linguistic-ethnic minority, the Swedish-speaking Finns, also known as Finland-Swedes, a linguistic minority in Finland. More specifically, we look at Finland-Swedes who have migrated to Sweden, the most common destination for emigrating Finland-Swedes, and compare them to Finland-Swedes in Finland. There are two main reasons why Finland-Swedes emigrating to Sweden offer an interesting case for studying ethnicity-based party ties among emigrated citizens. First, two-thirds of the Swedish-speaking population in Finland vote for an ethnic minority party mainly based around Swedish language rights and language policy, the Swedish People's Party of Finland (Bengtsson & Grönlund, 2005;Bengtsson et al., 2007). In other words, ethnicity plays a significant role for the party preferences of Finland-Swedes in Finland. Second, what makes the migration of Finland-Swedes to Sweden particularly interesting is that migrating from bilingual Finland (where Swedish is the second official national language) to monolingual Sweden elevates their mother tongue from a minority to a majority position. As Hedberg and Kepsu (2008) have pointed out, the hierarchy of identities changes as Finland-Swedes migrate to Sweden.
Hence, we seek to examine the persistence of party identification based on ethnic identity by studying the party ties of emigrated minority group citizens in a context that, in principle, removes the rational choice based on language for identifying with an ethnic minority party. We make use of two recent and highly comparable surveys, one conducted among a random sample of Finland-Swedes in Finland (in winter 2019), the other among a random sample of Finland-Swedes living in Sweden (in summer 2019). The two surveys present us with data that provide a rare perspective of party identification among non-resident citizens. Not least because representative survey data on migrated populations is scarce, and most studies of non-resident citizens rely on convenience samples (Ahmadov & Sasse, 2016).
Page 3 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 Party identification and ethnic parties We approach the theoretical underpinnings of our study in two steps. First, we briefly discuss the nature of partisan ties and why ethnicity-based party ties are considered particularly unlikely to change. Second, in the next section, we discuss how migration and the context of the host country tend to affect political behavior. Together they present us with an interesting interaction between a kind of political behavior that is fairly stable (party preference) and a circumstance where political behavior often changes (migration). In general, a party tie is considered to be a long-term, affective attachment based on one's social identity (see, e.g., Dalton, 2016). Partisan ties are known to develop at an early stage in life. Initial party ties, generally transmitted by the parents, are further reinforced by others from one's social class, ethnicity, or region. Hence, party identity becomes enmeshed in a web of social identities that tend to persist through life (Dalton, 2016). Nevertheless, research on partisanship based on long-term election studies in western democracies indicates that there has been a weakening of party identities taking place over the last few decades (Dalton & Wattenberg, 2000;Fiorina, 2002). Dalton (2012) argues that the patterns of weakened partisan ties can be described as partisan dealignment, i.e., a persistent decline in partisanship among citizens. Others (e.g., Thomassen, 2005) have argued that it is more a question of realignment and that partisanship is organized more along another dimension (the socio-cultural divide) than it used to be.
The increasing erosion in the linkages between voters and parties has also been seen as an indication that longer-term structural forces, i.e., social background and political values shaping partisanship and voting behavior, are giving way to shorter-term determinants, e.g., evaluations of government performance and the image of party leaders (Dalton, 2012;Dassonnville, 2016). While the evidence supporting the latter part of the argument is mixed (Dassonnville, 2016;Söderlund, 2015), it seems as though longerterm structural forces have become less important and that party ties have become more flexible over time. However, it is worth noting that they still are much more stable than political attitudes, which are notoriously unstable for large portions of the electorate (Converse, 1964(Converse, /2006. Despite a development toward more flexible party ties among voters in general, certain party ties are considered particularly durable. According to Chandra (2011, p. 153), it is widely assumed that ethnicity exerts an irresistible pull, i.e., ethnicity exerts a pull that is deeper than economic or other social preferences. Existing literature also suggests that ethnic parties can boost political participation of the ethnic group they represent domestically (Jiglau & Ghergina, 2011;Ishiyama, 2009). In addition, ethnicity is often backed by societal institutions or even legal status, making it more well-grounded than identities based only on interests or values. Hence, party ties based on ethnicity are expected to be stronger and more durable than other social identities.
At the most basic level, two broad models of ethnicity-based party ties can be distinguished. The first model involves ethnic participation almost exclusively within the most popular existing center-left or liberal party of the state, while the second model involves ethnic groups forming political parties, i.e., ethnic parties that exclusively represent their interests (Koev, 2019). In this study we will focus mainly on the latter, as the data we rely Page 4 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 on concerns party ties with an ethnic minority party. An ethnic party derives its support overwhelmingly from an identifiable ethnic group and serves the interest of that group. However, an ethnic party does not have to be the exclusive party of that group, as the group might split its support among several parties (Ishiyama & Breuning, 2011). It is worth noting that the definition above is based on who supports the ethnic party. Ethnic parties can also be defined in terms of who they claim to represent. In this case, a party represents itself to voters as a champion of the interests of an ethnic group and makes such a representation central to its mobilizing strategies (Chandra, 2004). Early literature on ethnic politics often assumed that ethnic identity was hardwired and its salience intrinsic (Rabushka & Shepsle, 1972), but more recent scholarship suggests that political and socio-economic factors influence the salience of ethnic identities (e.g., Laitin, 1998). Being a nominal member of an ethnic group does not always lead someone to have a strong attachment to ethnicity. Nominal ethnic categories are a necessary but insufficient condition for a strong ethnic identity (Chandra, 2012;Higashijima & Nakai, 2016). This suggests that Fiorina's (1981Fiorina's ( , 2002 rational choice approach to partisanship, where attachment to a party is treated as a running tally of accumulated partisan experiences, can also be applied to ethnicity-based party ties. According to Fiorina (2002), partisanship is strongly shaped by early life socialization, but it is continuously updated by successive experiences. This represents a dynamic view of partisanship, where the connection with a particular party is likely to be strengthened over time as an individual supporting a party is further socialized into a partisan voter. However, it also allows party ties to change if this socialization pattern is broken for some reason. This could be the result of losing trust in the party one feels closest to, for whatever reason, or because their life has changed to the degree that their priorities are substantially different from what they were, e.g., because of migration to another country and a new cultural context.

The impact of migration on political behavior
Research on homeland politics, i.e., the political activities of migrants or refugees pertaining to political decision-making in one's country of origin (Østergaard-Nielsen, 2003), is a relatively recent phenomenon. However, there is growing interest in this field for several reasons. First, transnational mobility has increased in recent decades, and new communication technologies allow those who have emigrated to easily stay in touch with their kin and homeland. Second, emigrating citizens are increasingly likely to have extensive political rights in their homeland, e.g., the right to vote in homeland elections (Bauböck, 2005(Bauböck, , 2015Lafleur, 2013).
While research on the voting behavior of non-resident citizens in home country elections (e.g., Lafleur & Sanchez-Domínguez, 2015;Ghergina & Tseng, 2016, Ciornei & Østergaard, 2020 has become more commonplace, there are still significant gaps in the research regarding political behavior among non-resident citizens. The way identification with home country parties changes as a result of migration represents such a gap. The formation and adaption of social identities are crucial for understanding homeland politics among migrants. As the migrant moves away from one 'home' of identity to a different socio-cultural context, their identity is realigned both in relation to the country of residence and their home country (Hedberg & Kepsu, Page 5 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 2008). Political socialization theory also presents us with different expectations for how migrants adapt to a new political setting. According to White et al. (2008), some aspects of political life are more transferable than others. Political interest is highly transferable, while the transfer of partisanship is a more complicated affair.
To gain a better idea of how political attitudes and behavior change because of migration, several studies have relied on comparisons of migrants and non-migrants from the same home country population. Careja and Emmenegger (2012) use a large survey data set of Central and Eastern European respondents to compare political attitudes among return migrants and non-migrants. They find that while these groups have the same attitudes toward domestic politics, the return migrants are more likely to value democracy, participate in politics, and have a greater interest in EU and foreign politics. A study by Chauvet, Gubert, and Mesplé-Somps (2016) show that Malian migrants have different perceptions and political behavior than their nonmigrant counterparts. They also find that the host country's institutional context matters for the adoption of political norms among migrated Malian citizens.
The latter finding is supported by research on a very different group of migrants, Swedes. Using a unique representative survey of emigrated Swedes, together with two cross-country surveys, Dahlberg and Linde (2018) investigate how a move from a context of high institutional quality to countries characterized by low institutional quality affects institutional trust and satisfaction with democracy. They find that Swedes living in countries with low levels of institutional quality display significantly lower levels of trust and support for democracy. Another study with similar findings regarding the political behavior of migrants by Escobar et al. (2015) shows that Colombian expatriates' electoral participation is greatly influenced by the variation of the local context in which the voters are embedded. This finding regarding variation in electoral participation based on place of residence is also supported by evidence from Peltoniemi's (2016) study of Finnish non-residents in 6 different countries and by Ahmadov and Sasse's (2016) research on electoral engagement among emigrated Ukrainians.
Together, these studies show that the host country context plays an important role in emigrated citizens' political behavior. The results also imply that political behavior tends to change as people migrate from one country to another. However, none of the studies above examine home country party ties of migrated external voters. In a rare study of home country party ties among non-resident citizens, Turcu and Rubatsch (2019) examine whether Turkish migrants across 23 European countries show loyalty to the political parties that granted them enfranchisement. Their findings suggest that emigrant-enfranchising parties tend to garner lasting support among non-resident voters. While research on such diaspora policies gives us some idea of what the party ties of non-resident citizens may look like, it does not tell us very much about how established party ties are affected by migration. What we do know is that the tie to home country politics tends to weaken after migration. Not only is turnout much lower among external citizens (e.g., Ciornei & Østergaard, 2020), which partly can be explained by practical challenges related to external voting, but research also suggests that migrants tend to become less likely to vote in home country elections over time (Ahmadov & Sasse, 2016;Peltoniemi, 2016). While participation in elections cannot Page 6 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 be compared directly to transnational party identification, it suggests that engagement in home country politics becomes weaker over time. In the next section, we take a closer look at a linguistic ethnic group and discuss the potential implications of migration on their party identity.

Party identification and migration among Finland-Swedes
This study focuses on Finland-Swedes, Finland-Swedes that have emigrated to Sweden in particular. Due to the crucial role the case at hand has for our research design, it is paramount to have an understanding of the Swedish-speaking minority in Finland and what migration to Sweden implies for this ethnic minority group. Finland has two national languages, Finnish and Swedish. The majority of the population speaks Finnish as their native tongue, but around five percent of the Finnish population (approximately 290 000) speaks Swedish as their first language (Statistics Finland, 2019). The relatively prominent role of the Swedish language in Finland today is explained by the fact that Finland constituted the eastern half of the Kingdom of Sweden for a period of six hundred years, between the 13 th and the beginning of the nineteenth century, and Swedish being the administrative language in Finland until the second half of the nineteenth century (Tandefelt & Finnäs, 2007). With Finland's transition to universal suffrage in 1906, the Swedish-speaking population in Finland coalesced around the Swedish People's Party in Finland (SFP 1 ). The party was a result of the Swedish-speaking elite reaching out to the Swedish-speaking common people to mobilize in defense of the Swedish language in Finland. Apart from a few minor disputes between the language groups during the first decades of the twentieth century when the Finnish language was growing into its current role as the dominant language, the linguistic cleavage in Finland has remained comparatively uneventful (Karvonen, 2014). The SFP can be described as a moderate protectionist ethnic party (de Winter 1998, Ishiyama & Breuning, 2011). Like most ethnic parties, it attempts to appeal to all, or at least most co-ethnic voters, and the party receives around 70 percent of the Finland-Swedish vote in each election (Grönlund, 2011;Sundberg et al., 2005). While the party has an ideological outlook (liberal on social issues and leaning to the right on economic issues), the common denominator for SFP voters is the Swedish language (Djupsund & Carlsson, 2005;Sundberg et al., 2005).
The principal role of linguistic identity might sound self-evident. Still, one needs to keep in mind that the Finland-Swedes are quite a heterogeneous group. Despite being few in number, the Finland-Swedes are geographically fairly spread out along Finland's coastal regions to the west and south (Hedberg & Kepsu, 2008, p. 97). As a group, Finland-Swedes are both very urban, over-represented in the central parts of the capital city Helsinki, and very rural, residing in small municipalities along the coast of Finland (Finnäs, 2015;McRae, 1999). The presence of the Swedish language in Finland-Swedes' daily lives is also very different, depending on where they happen to live in Finland. In multiple small towns in the countryside, Swedish is the majority language. In large bilingual municipalities, like Helsinki and Turku, Swedish is the second official language, but Swedish speakers make up only a small proportion of the inhabitants (Statistics Finland, 2019). Despite these differences, support for the SFP among Swedish-speakers is relatively similar in all the bilingual areas in Finland (Bengtsson et al., 2007).
The factors mentioned above are essential to our understanding of Finland-Swedes and their relationship with SFP as an ethnic party. However, the emigration of Finland-Swedes is interesting in and of itself. Finland-Swedes are a very mobile group of citizens. According to Statistics Finland (2019), on average, 1500 Finland-Swedes emigrated from Finland every year during the last decade. This number may seem rather trivial in an international context, but it is the equivalent of 0.5 percent of the Swedish-speaking population in Finland emigrating every year. In fact, approximately 13 percent of all Finland-Swedish voters live abroad (Harjula and Himmelroos (2020). Second, a large part of the Finland-Swedish migration goes in the direction of Sweden. As a result, 75 percent of the non-resident Swedish-speaking Finnish citizens are living in Sweden. This migration pattern is largely explained by the fact that Sweden lies geographically close to Finland and that it is both linguistically and culturally close to the Finland-Swedes (Hedberg & Kepsu, 2008;Höckerstedt, 2000). Global migration patterns suggest that people prefer to move and settle down in a country that is close to the country of origin or has a significant population of the same ethnic group (Pedersen et al., 2008). Moreover, Finland-Swedes emigrating to Sweden is also representative of a general global migration pattern, where the flows tend to follow a 'network effect' of people preferring to move to societies with an entrenched language and cultural population (Ivlevs, 2013;Pedersen et al., 2008).

Hypothesis
From the perspective of partisanship and party identification, migration presents the voter with a situation where long-term structures affecting party choice become more distant and potentially gain a different meaning. Considering this idea of identity reformation and drawing on the research on party identity and the effects of migration on political behavior discussed above, we formulate three main hypotheses. First, we expect that emigration and the change of linguistic context will result in a weaker party tie with the ethnic minority party since research on emigrated citizens repeatedly show that they differ substantially from resident citizens in their political behavior (Chauvet et al., 2016;Dahlberg & Linde, 2018) and the migration to Sweden represents a change in the hierarchy of identities for Swedish-speaking Finns (Hedberg & Kepsu, 2008).
H1 Identification with an ethnic (minority) party is lower for migrated citizens than for resident citizens.
Changes to party identification after migration are, nevertheless, unlikely to happen overnight. Especially considering that ethnic party ties generally are quite strong. However, research on resocialization (White et al., 2008) and turnout levels among external citizens (Ahmadov & Sasse, 2016;Peltoniemi, 2016) show that the relationship with political entities changes over time. There will likely be a hysteresis effect, i.e., a slowed adaption to a new system or context due to psychological factors and processes of desocialization from the home country context and socialization into the host country Page 8 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 context. We expect party ties based on ethnic identity may start to diminish over time when one has lived in the host country for a longer period and become more integrated into the new context. Hence, creating a reaction with lag instead of a sharp disruptive effect.
H2 Identification with an ethnic (minority) party weakens over time for emigrated voters.
However, emigration also presents a situation where voters may experience a realignment in accordance with new or previously suppressed preferences. Supporting a party based on ethnicity could mean having to set aside ideological preferences that would otherwise be pertinent to your vote choice. Hence, moving from a context where the ethnic identity is taken for granted to a context alien to that identity can potentially activate or amplify ideological preferences previously suppressed in an ethnic minority context. Following this logic, we expect that voters who have political preferences (e.g., on economic or social issues) which are less closely aligned with the political outlook of the ethnic party are more likely to experience dealignment. In our case, this would mean that more left-leaning or socially conservative voters who identify with SFP mainly because of the language issue would experience dealignment after migration. According to this assumption, the ideological outlook of the SFP, self-placement to the right, and being socially progressive should be more emphasized among migrated SFP voters.
Despite the highly cohesive political behavior of Finland-Swedes in the home country, we also expect that non-resident citizens originating from (smaller) municipalities with a Swedish-speaking majority are more likely to identify with SFP after migration. Migrants coming from smaller Swedish-speaking communities where SFP is the dominating party are less likely to have come in close contact with the politics and party organizations of other parties. Thus, they may simply have a less clear picture of the available options if the ethnic party starts to feel less relevant after migration. Based on these assumptions, we formulate three sub-hypotheses relating to central factors predicting support for the SFP.
H3a Ideological self-placement to the right predicts identification with the SFP better among emigrated Finland-Swedes than among resident Finland-Swedes.

H3b
Socially progressive values predict identification with the SFP better among emigrated Finland-Swedes than among resident Finland-Swedes.
H3c Emigration from a municipality where Swedish is the majority language predicts identification with the SFP better among emigrated Finland-Swedes than among resident Finland-Swedes.

Survey data
The study relies on two highly comparable surveys carried out among two populations, the Finland-Swedes living in Finland and the emigrated Finland-Swedes, with random Page 9 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 samples of the populations. The survey among Finland-Swedes living in Finland was part of the 5th wave of the European Values Study (EVS). It included the full EVS questionnaire with a few additional questions relating specifically to the Swedish-speaking minority in Finland. The invitation to take part in the survey among the Swedish-speaking Finns in Finland was distributed by mail to a random sample of 3,600 Finland-Swedes together with a paper copy of the survey. There were, altogether, 1341 completed responses, resulting in a response rate of 37 percent. 2 The survey among non-resident Finland-Swedes was also based on the EVS questionnaire, complemented with additional questions relating specifically to living abroad as a foreign citizen and a few related to the Swedish-speaking minority in Finland. The invitation to participate in an online survey was distributed by mail to a stratified random sample of 4800 Finland-Swedes in 15 countries, drawn from the Digital and Population Data Services Agency registry on Finnish citizens living abroad. A reminder was sent to respondents who had not answered the survey within one month. There were 1971 completed responses or a response rate of 41 percent. However, for this study, we are only interested in non-resident Swedish-speaking Finnish citizens living in Sweden. Moreover, we only include Finnish citizens who have lived in Finland at some point during their lives (emigrants). We do not include citizens who have emigrated from the Åland Islands, as the party system in this autonomous region is different from the one in mainland Finland. This leaves us with 708 emigrant respondents.
The mechanism of a language context mitigating the language minority group's party identification is analyzed with logistic regression analysis. By combining the two data sets and treating Finland-Swedes in Finland as a control group, we can estimate the effect of emigration on party identification among the language minority group of Finland-Swedes. We also conduct several post-estimates to test the robustness of the results.

The dependent variable
Our dependent variable measures identification with the Swedish People's Party of Finland. The variable is derived from the European Values Study survey question: "Which Finnish political party appeals to you most?". 3 The answer was given by choosing from a list of the parties represented in the Finnish parliament with the additional options "Other party" and "Don't know. " The dependent variable consists of a dummy capturing identification with the Swedish People's Party of Finland. The "Don't know" answer is included in the baseline category of identification with other parties so that we get a dichotomous variable on whether one identifies with the SFP or not. The "Don't know" answer is included in the dummy's baseline category because of the high share of responses on "Don't know, " which means its exclusion would make the statistical models susceptible to the spurious effect.

Independent variables and covariates
Independent variables are factors that we, based on the theoretical framework, suggest will explain the emigrated Finland-Swedes propensity to back the SFP compared to 2 See Declarations for data statement and availability. Finland-Swedes residents in Finland. On the other hand, covariates or control variables are known to have the potential to explain identification with a political party but are not theoretically of interest in the study. By controlling for these covariates in the analysis and comparing their effect to the independent variables, we can determine the relative effect size of independent variables and exclude alternative explanations to the observed phenomenon. We measure Finland-Swedish identity with the question: "People have different views about themselves and how they relate to the world. How close do you feel to … [Swedish part of Finland]?" using a 4-graded item, where 1 equals very close and 4 not close at all. The operationalization meets the idea of social identity as in-group self-identification.
The rational choice explanation for Finland-Swedes' identification with the Swedish People's Party is securing welfare service in one's native language. Skill in the majority language, Finnish, is an indicator for dependency on service in Swedish. Finnish language skills are measured with the question "How good is your command of the Finnish language?" on a scale of 1 to10, where 1 equals a complete lack of Finnish language skills and 10 excellent language skills. The variable has been inverted to allow for a more straightforward interpretation since we expect lower Finnish language skills to increase the likelihood of identifying with the SFP. The variable has also been normalized to vary from 0 to 1.
We also measure political orientation on a left-right scale to control for political ideology as an alternative explanation for backing the Swedish People's Party, which has traditionally been on the right of center in the political landscape. The measure is based on the question: "In political matters, people talk of 'the left' and 'the right' . How would you place your views on this scale, generally speaking?" The political orientation was measured on a scale of 1 to 10, where 1 equals 'left' and 10 'right' . In addition, the SFP stands for socially progressive values on questions of same-sex marriage or welcoming refugees. As these immaterial values are not captured by the traditional left-right ideological orientation, we measure the degree of conservative or liberal stance on social values with the question "Do you agree or disagree with the following statement? Homosexual couples are as good parents as other couples. " The respondents were asked to answer whether they agree strongly, agree, neither agree nor disagree, disagree, or disagree strongly. The answers were coded as 1-5 and inverted so that 1 reflects the most conservative stance and 5 the most liberal stance of the respondents.
To analyze whether migration matters for party identification, we measure migration with a dummy variable of 0 for living in Finland and 1 for living in Sweden. Further, to explore if time spent in the new context affects party backing, we measure time lived abroad from the official register data of the respondent's emigration date. The variable is coded with Finnish-based respondents as 0, 0-10 years lived abroad as 1, and 11 years or more as 2.
Interest in politics is also treated as a covariate to control how general interest in politics affects identifying with the Swedish People's Party or other parties regardless of the ideology or cause. It is measured with the question "How interested would you say you are in politics?" on a 4-graded item. These covariates have also been normalized to vary from 0 to 1.
Page 11 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 Further, we control for party membership since it is connected to the strength of the party tie. Party membership is measured with the statement: "Please look carefully at the following list of voluntary organizations and say which, if any, you belong to… [Political parties or groups]. " We also include interest in following Finnish politics on social media, measured with the question: "How often do you follow Finnish politics… […on social media (for example, Facebook, Twitter, Snapchat, Instagram, etc.)]?" Following politics on social media indicates an awareness of the political, social, and economic reality in the home country (Grad et al., 2020), and these information channels are accessible to migrated and resident citizens alike.
Socioeconomic status, which is known as a powerful explanation of party identification (Kroh & Selb, 2009), is controlled for by including dummy variables on education (higher education or not), employment (full-time employed or not), and household income (high earner or not) as covariates. Demographics are controlled for with age and sex from register data. The degree of urbanization is controlled for with a dummy variable on municipality size according to population size on the scale 0-10 000, 10 001-30 000, 30 001-100 000, 100 001-500 000, and 500 000+. The degree of urbanization is used to control the varying presence of the Swedish language in different geographic locations, i.e., a majority language in some smaller towns but a minority language in other larger bilingual cities. Language environment is furthermore controlled for on the macro-level with variables on official municipal language status; monolingual Finnishspeaking, bilingual with a Finnish-speaking majority, or bilingual with a Swedish-speaking majority (see the description of the geographical concentration of Finland-Swedes and legal framework for official language status in the section Party identification and migration among Finland-Swedes). Descriptive statistics for all variables can be found in the "Appendix".

Results
We begin the analysis with a more descriptive look at the party ties of emigrated Finland-Swedes before moving on to the more in-depth explanatory analysis. In our descriptive analysis (Fig. 1), we compare the party preferences of Finland-Swedes in Sweden to the party preferences of Finland-Swedes living in Finland. The results suggest that the two groups are highly similar when it comes to which parties they identify with. The Swedish People's Party is the dominant party in both groups, albeit slightly less so among those having emigrated to Sweden. There are generally very small differences between the two groups. The most noteworthy difference is that a larger share of respondents among the emigrated Finland-Swedes answer 'don't know' when asked about their party preference. Other minor differences include a slightly smaller percentage identifying with the Social Democratic Party and a somewhat larger group identifying with the Green league among Finland-Swedes residing in Sweden. In our next comparison (Fig. 2) of party ties, we look at emigrated Finland-Swedes living in Sweden and the extent to which their party identification differs based on their length of stay abroad. In this comparison, the difference in identification with the Swedish People's Party, between resident and non-resident citizens, is slightly larger than for the comparison between those residing in Finland and Sweden in general. The most notable difference in this comparison is related to the 'don't know' answers to the question used to measure party ties. The respondents who had lived abroad for more than 10 years were twice as likely to not know which Finnish party they identify with. Identification with the Green league is also markedly higher among the more recent emigrants.
In the second part of the analysis, we employ several logistic regression models to test our hypotheses further. The two data sets, with Finland-Swedes living in Finland and in Sweden, have been pooled for these analyses. In the first model (col 1), we examine the difference in identification with the SFP between Finland-Swede emigrants and residents of Finland while controlling for the influence of the other variables. The expatriate dummy is statistically significant and shows expatriated Finland-Swedes to be only 66 percent as likely to identify with the SFP as Finland-based Finland-Swedes. This is in line  Page 13 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 with our first hypothesis (H1), suggesting that emigrated Finland-Swedes are less likely to identify with the Swedish People's Party than those residing in Finland. Since the finding contradicts our previous finding in the descriptive part of the analysis, it underlines the importance of considering the differences between the populations we examine.
We also see that ethnic identity measured as a sense of belonging with Swedish Finland increases the likelihood of identifying with the SFP by 19 times. Similarly, right-leaning political orientation on the traditional left-right axis greatly increases the odds of identifying with the SFP, as do liberal views in social questions. These findings are all in line with existing knowledge about SFP voters. The second model (col. 2) shows the same relation when also controlling for the duration of living abroad. Here we see that a shorter stay of up to 10 years is not changing the odds ratio for how likely one is to identify with the SFP. However, a more extended stay of over 10 years means the Finland-Swede is only half as likely to identify with the SFP. Thus, in line with our second hypothesis it seems that identification, even with an ethnic party, fades over time. The largely similar effect sizes of the other independent variables compared to Model 1 indicate the results without considering time spent abroad were quite robust (Table 1).
The third to fifth models (col. 3-5) test whether political orientation, social values, or majority language in the municipality of residence in Finland explain identification with the SFP when moderated by living abroad or in Finland. However, we find no evidence supporting our sub-hypotheses (H3a-c) regarding a realignment of party preferences after migration.
Among control variables, political party membership, education, and age are correlated with identification with SFP, but the effects are smaller than those of the independent variables. The geographical controls 'municipality size' and 'official language' have a somewhat larger effect size. Respondents living in or originating from the larger cities are less likely to identify with the SFP than their small-town peers, and Finland-Swedes from bilingual Finnish majority municipalities are more likely to identify with the SFP.
Since regression analysis is often insufficient for ruling out endogeneity even when controlling for alternative factors, the analysis is complemented with additional robustness tests (Additional file 1). Treatment effect analysis relying on entropy balancing and Mahalanobis distance matching together with regression analysis using the weighed covariates, provided very similar results to logistic regression model 1. These findings lend further support to the analysis made based on the initial model.

Conclusions
This study set out to shed more light on how emigration affects the party identity of ethnic minorities. Focusing on the Swedish-speaking linguistic minority in Finland, we examined how party identification with ethnic-minority is affected by migration to Sweden. We expected that identification with the ethnic minority party, the Swedish People's Party in Finland (SFP), would be lower among emigrants since the minority identity is less relevant in the Swedish context (H1). We also expected that this would be a lagged effect, i.e., that identification with the ethnic party would wane over time (H2). Our third and final hypothesis (H3a-c) was that the composition of the voters identifying with an ethnic party would be different among emigrants than among resident Finland-Swedes. Based on our analysis, we find relatively robust support for the first two hypotheses. There is indeed less support for the SFP among the emigrated Finland-Swedes (H1). We also find that the party identity is lower among Finland-Swedes who migrated to Sweden more than 10 years ago, whereas those who had emigrated more recently show no difference to resident Finland-Swedes when it comes to identifying with the SFP. Hence, it appears that dealignment is mainly a result from a hysteresis effect, where attitudes, values, and habits change slowly in a new system (H2). We were, however, less successful in teasing out the rationale behind the diminishing support for the SFP among Finland-Swedish emigrants in Sweden. There were no meaningful differences between emigrated citizens and resident citizens regarding identification with the Swedish-speaking region in Finland or political orientation (left-right and liberal-conservative) to explain identification with the SFP.
In sum, time appears to be the most important factor for explaining dealignment among Finland-Swedes. The longer a migrant has spent outside the linguistic minority context in Finland, the less likely they are to identify with the Swedish People's Party. This is likely the result of the naturally slow process of partisan resocialization (e.g., White et al., 2008), and it suggests that even strong party identities based on ethnicity will weaken over time among emigrated citizens. And even though the effect is small and change is slow, the pattern seems to fit Fiorina's (1981Fiorina's ( , 2002 assumptions about accumulated partisan experiences, where partisanship is expected to decrease if the socialization pattern is broken. However, based on the findings from our analysis, the weakened partisan identity seems to be less due to a realignment (Thomassen, 2005) according to new values and more due to a dealignment (Dalton, 2012) from home country politics. There is a notable increase in the number of emigrants who do not identify with any of the home country political parties when comparing those who have lived abroad for less than 10 years to those who emigrated long ago. Page 16 of 20 Himmelroos and Vento Comparative Migration Studies (2022) 10:19 It should also be noted that even though the party tie is loosened somewhat over time, it is still highly resilient. While the party tie may not be as strong as for those residing within the minority context in Finland and more recent emigrants, the connection to the ethnic minority party remains surprisingly strong even among the migrants who emigrated more than 30 years ago. All in all, our findings suggest that even life-changing events, such as emigration will not radically undermine strong partisan ties. However, it is worth noting that due to the geographical proximity of Finland and Sweden, maintaining a connection with the home country is easier than in many other cases of migration. This could partly explain why Finland-Swedes identifying with the Swedish People's Party do not experience greater dealignment despite the radical change to their hierarchy of identities migrating to Sweden would imply.
At the outset of this article, we mention that most research conducted on the relationship between political parties and external citizens has focused on how parties try to connect with external citizens (Østergaard-Nielsen & Ciornei, 2019a;Østergaard-Nielsen & Ciornei, 2019b;Kernalegenn & van Haute, 2020). Based on the findings from this study, it seems the SFP would benefit from engaging with Finland-Swedes abroad, as they still identify strongly with the party after having migrated. Especially if we consider that the decidedly lackluster turnout among external Finnish citizens may be improving with the recent introduction of postal voting from abroad.
It should also be pointed out that this study is based on a single case and, therefore, contains certain limitations when it comes to generalizability. The mechanism of diminishing identification with an ethnic party over time can be assumed to have a more general bearing for migrating populations, especially for similar cases where the basis for identification with a party may have little relevance outside the country of origin. However, it is less clear that the same pattern can be found for individuals identifying with a social democratic or conservative party, i.e., parties that exist in most multi-party systems. Perhaps a question that future research on party identification and migration can help answer. The findings from this study would benefit from being analyzed comparatively with other similar ethnic groups from multilingual countries, such as Canada or Switzerland, in the future. This would help us understand whether the explanatory factors revealed in this study indeed present a more general mechanism for how (ethnic) party ties are affected by migration.

Appendix
See Tables 2 and 3.