UvA-DARE (Digital Migrant votes ‘here’and ‘there’: Transnational electoral behavior of Turks in the Netherlands

Whether there is a trade-off between ‘here’ (country of settlement) and ‘there’ (the country of origin) is one of the key political questions and concerns regarding political attitudes and behaviors of immigrant minorities. We take this issue by the horns and study three components of political attitudes and behavior within a transnational framework among Dutch-Turkish citizens in the Netherlands: turnout, political trust and interest, and party choice. The empirical data draws on original exit polls held during the Turkish presidential and parliamentary elections at a polling station in 2014 ( n =791) and in 2015 ( n = 456). We ﬁnd that that gender and country of birth inﬂuence electoral participation; social class (working class background as labor migrants) inﬂuences voting behavior. While there is a trade-off for political trust and voting behavior, there is no trade-off for political interest. These ﬁndings call for a more nuanced approach to transnational political behavior that is attentive to processes of convergence between ‘here’ and ‘there’ and the diversity within migrant groups.


Introduction
'. . . (You) are our power outside the country [. . .] For us you are not only emigrants, you are our strength in foreign countries. . . I would request you to make the best use of the power in your hands. I am expecting you to fill the ballots in Germany and in Europe to fill with your favor and determination.' -Turkish President Tayyip Erdogan in a speech to German-Turkish citizens in the German city of Karlsruhe prior to the 2015 elections in Turkey (Ö zay, 2015) Political leaders in West European countries are far from pleased with Turkey's political campaigning on their soil. Although such campaigning among immigrant populations by state and political actors is nothing new, its potential impact has grown since Turkish electoral law was amended in 2012. The new law allows Turkish citizens living abroad to cast their votes in Turkish elections on the territory of their countries of settlement (Supreme Election Council 2012). The active engagement in homeland politics that this law is feared to trigger is at loggerheads with immigration countries' expectations that migrants and their descendants fully integrate into political life in the country of settlement. Dual nationality or political loyalties towards a former or ancestral homeland generally do not fit this ideal. To paraphrase Waldinger (2008), the underlying belief is that involvement over 'there' obstructs integration over 'here'-particularly when that homeland is considered conservative, or worse, authoritarian, as is increasingly the case for Turkey. The 2017 diplomatic crisis with the Netherlands, which expelled the Turkish Minister of Family and Social Policies campaigning among Dutch-Turks for an upcoming referendum, is but one example in a series of clashes between Turkey and West European immigration countries. Two decades of interdisciplinary scholarship on migrant transnationalism has shown that immigrant or dual citizen loyalties are not necessarily mutually exclusive, but often exist side by side (e.g. Levitt and Glick-Schiller 2004). Waldinger (2015) depicts this process as an inter-societal convergence; migrants link the states they leave with those they enter in political and territorial terms. We study the interplay between political attitudes and behaviors of migrants in both sending and receiving political systems to further unravel this process. How do political behaviors and attitudes interact with each other in the transnational space when citizens of immigrant origin are allowed to vote in the elections of more than one country? Taking Dutch-Turks as its case, this article tackles this issue that haunts European immigration countries.
Despite the existence of external voting rights in 115 countries (IDEA 2007), our understanding of how electoral attitudes and behaviors in sending and receiving countries interact remains limited. Extant studies generally focus on one side of external voting: how the political participation of immigrants and dual citizens influences homeland elections and homeland politics more broadly (Duquette-Rury 2016). There is thus, a lacuna in our understanding of how political attitudes and behaviors in countries of origin and settlement interact (for a recent exception in Germany, see Goerres et al. (2018)).
The unidirectional approach does not suffice to understand voting transnationally or, indeed, to assess the impact of transnational voting on political integration in countries of residence. We examine the convergence of political attitudes and behaviors in sending and receiving states to study whether there is a trade-off between emigrants' turnout levels, trust and interest in politics in the homeland and in the country of residence. We demonstrate how political attitudes and behaviors are linked to political relations between the sending and receiving states.
Turkish emigrants in the Netherlands are considered an excellent case for a study of transnational political behavior in Europe. The Turkish state actively mobilizes Turkish voters abroad. At the same time-due to the growing strength of anti-immigrant populist parties in Western Europe-political parties in Western Europe increasingly problematize presumed attachments with countries of origin among immigrant origin communities.
Their message is clear: dual loyalties are an obstacle for political integration. The European-Turkish communities, and especially those members who are deemed to be religious and conservative, are the focal point of these heated political discussions in various European countries (Vermeulen 2019). Developments, in both Turkey and the Netherlands have effect on the transnational political behavior of Dutch-Turkish citizens.
Next, we review national and transnational approaches to the electoral attitudes and behaviors of immigrants and their descendants and pose our research questions. The empirical body of the article draws on original exit polls held during Turkish presidential and parliamentary elections, at the polling station in the city of Rotterdam in 2014 (n = 791) and in Rijswijk (n = 456) in 2015. The exit polls inquired about electoral behaviors and attitudes in the Netherlands as well as in Turkey. We complement our survey data with official records of Turkish citizens voting in the Netherlands and other European countries. Our analysis focuses on three key components of political attitude and behavior: (1) turnout rates, (2) trust and interest in Turkish and Dutch politics, (3) party choices in the Netherlands and in Turkey. The interaction between these components across Dutch and Turkish political contexts lies at the core of our study.

Transnational electoral behavior
Although continued loyalty to the homeland provokes questions regarding one's (political) integration in the country of settlement, the assumed negative relationship remains understudied in the literature. Extant studies tend to focus on one side of this transnationalism by studying participation in either homeland or country-of-residence elections. The transnational literature discusses the normative implications of allowing or disallowing external voting and how this is linked to the quality of an emigration country's democracy (Bauböck 2007;Smith 2008;Brand 2010;Lafleur 2013;Burgess 2014;Collyer 2014;Caramani and Grotz 2015). Central in this debate are the strategies of sending states to encourage or discourage their citizens abroad to participate in elections and the extent to which these efforts are symbolic. The main actors of inquiry in this field are sending state governments.
Empirical studies on the relationship between the political participation in the country of origin and the country of residence remain inconclusive (Mügge 2016a). Some studies show that immigrants who participate in the political system of the country of origin are less attached to the political system of the country of residence and as a consequence of that less involved and less integrated politically (Staton et al. 2007). Other studies, however, do not find such relationship (Ramakrishan 2005;Chadhary 2018;Pilati and Herman 2018). Waldinger (2015) emphasizes that international migration produces intersocietal convergence that cannot be understood from a geographical angle. While emigrants are no longer physically in the sending state they are still part of it through a complex package of feelings of belonging, citizenship, memories, and family relations. Emigrants do not only migrate with their body but bring their political ideas and attachments with them. In this process, they create a zone of intersocietal convergence. They link 'here' and 'there' in a variety of domains, including politics (Waldinger 2015). Emigrants who demand full citizenship rights in both the sending and the receiving state produce a stronger link, than those without dual citizenship. To participate in both spaces emigrants, need to be recognized as citizens of the country they left, but they also need to be recognized as legitimate citizens in the states they enter. Recognition is often contested on both ends. Transnational political engagement is thus shaped by the entire migration process. This includes acceptance by natives in the receiving state political system, sending state policies regarding citizens abroad as well as the international relationship between sending and receiving state (Mügge 2010).
Drawing on subsets of the literature on political behavior and attitudes, we formulate three research questions concerning turnout, political trust and interest, and political choice. In these elements we capture the process of intersocietal convergence, as well as factors that determine political integration of citizens with a migration background in receiving societies.

Turnout
Electoral turnout is a key measure of political behavior. Several authors have found participation in homeland elections to be closely tied to the availability of compatriot media and the existence of ethnic and national organizations (see Leal et al. (2012) for Mexicans; Morales and Pilati (2014) for Ecuadorians; Ahmadov and Sasse (2016a) for Ukrainians). Burgess (2014) finds that emigrants are more likely to participate in homeland elections when homeland parties actively campaign abroad. The overall idea is that if there is a dense infrastructure of immigrant organizations and media sources that disperse information on homeland politics, emigrants will be more inclined to vote. Among Colombians, Escobar et al. (2015) find that those who have previously voted in the homeland are more likely to vote in homeland elections when they reside abroad. For Central and East European post-communist democracies, Kostelka (2017) concludes that transnational voting rates are lower than domestic ones. Whether or not emigrants and their descendants will continue to be involved with homeland politics depends on the strength of homeland nationalism, homeland policies to include or exclude participation of citizens abroad and the actual political climate (Mügge 2013). The latter may fuel outbursts of homeland directed protests on the soil of the receiving country. Examples of the past include the Irish, Jewish, and Lebanese lobby in the USA and the Kurds in European countries (Berkowitz and Mügge 2014).
Studies on the political participation of citizens of immigrant origin underline the importance of social capital. Ethnic groups with dense internal organizational networks-are more likely to participate in local elections in the country of residence. Comparisons between first-generation ethnic groups in the Netherlands show that Dutch-Turks have the most ethnic social capital and turn out in higher numbers in municipal elections (Fennema and Tillie 1999). Quintelier's (2009) study on the political participation of immigrant youth in Belgium finds higher participation rates than among their parents. Given that the second generation is politically socialized in the Netherlands, we expect Dutch-born Turkish citizens to have greater Dutch social capital-in access to information about and participation in civil society organizations and local politics-than the generation of their parents.
Finally, there is a paucity of studies examining the relationship between voting behavior in homeland elections and elections in the country of residence. The extant work suggests that there is no trade-off. On the contrary, in a comparative study of Ecuadorians, Moroccans and Turks in European cities, Morales and Morariu (2011: 167) find that voting in homeland elections is strongly and positively associated with the propensity to vote in elections in the country of residence.

Interest and trust
Political trust is generally understood as the trust citizens invest in their government to develop and implement fair laws in a way that is considered just (André 2014). People who have higher levels of political trust are expected to participate more in conventional political activities such as voting and campaigning. Those with high levels of distrust not only participate less, but also feel less connected to the political system and more often feel politically alienated (Levi and Stoker 2000). Political trust is therefore, an important democratic resource for citizens more generally (André 2014). However, empirical research on the relationship between political trust and different forms of political participation has drawn inconsistent, and at times even contradictory, conclusions (Gabriel 2017).
Political interest is another important precondition for democratic citizenship (Van Deth 1989). Interested citizens are more knowledgeable about political affairs and have higher levels of political participation (Verba et al. 1995). Theoretically, political interest is expected to be highly associated with political participation by functioning as a stepping stone (Verba et al. 1978;Fischer-Neumann 2014).
For citizens of immigrant origin, political trust and interest are often related to feelings of loyalty to the political system in the country of residence. A high level of trust is seen as a milestone in becoming a full member of (mainstream) society, an indicator of attachment and loyalty to the political system of the receiving country (Maxwell 2010). Since a large part of the political behavior of (e)migrants revolves around recognition (Waldinger 2015), we expect political trust and interest to correlate with feelings of recognition in both the homeland and host country political systems. Wals and Rudolph (2018) expect that trust among immigrants declines when as assimilate in an increasingly more distrustful contemporary Western political culture. Here, we may expect a trade-off as well: More trust and interest in the political system of the homeland are triggered by distrust and decreasing interest in the political system of the Western countries of residence. Or vice versa. The illiberal development of the Turkish political system in the period under study revolves around the inclusion of particular political, ethnic, or religious groups (Ö ktem and Akkoyunlu 2016). Groups that feel excluded will most likely display lower levels of trust than included groups.
In a comparative study of Ecuadorians, Moroccans, and Turks in European cities, Morales and Morariu (2011: 167) find a strong correlation between sustained interest in homeland politics and interest in the politics of the country of residence. For political trust, we lack such comparative research. Do voters have similar levels of trust and interest in the Dutch and Turkish political systems?   shows that the Labor Party (PvdA) has been the most popular party among citizens of immigrant origin. This holds especially for voters of Turkish descent (Vermeulen et al. 2014). Another explanation for the popularity of the Labor Party is its traditionally strong representation of politicians of Turkish descent at both local and national levels (Mügge 2016b). Support for the Labor Party from the immigrant origin constituency, however, dropped significantly in the last elections. In 2006, more than 85 per cent of voters of Turkish descent in Amsterdam voted for the Labor Party; in 2014 this was just over 40 per cent (Vermeulen 2019). New, small religious and ethnic parties with strong links to the Turkish constituency profited from the loss of the Labor Party, and gained 30 per cent of the Dutch-Turkish electorate's votes in Amsterdam (Vermeulen et al. 2018). On the national level we observe a similar pattern. In the 2017 national elections, a new party, DENK (which means 'think' in Dutch and 'equality' in Turkish), founded by two former Dutch-Turkish Labor Party parliamentarians, gained three seats. Their electorate consists predominantly of Dutch-Turkish and other immigrant-origin groups. At the time of the Turkish elections in 2014 and 2015, DENK had not yet been formed.
Party choice and political ideologies are rarely studied transnationally; the few existing studies addressing the ideological correlation between party choices in home and host countries have arrived at different conclusions. According to Wals (2013), Mexican immigrants tend to import their homeland ideology, which remains decisive for their electoral behavior in the host country. The same study also finds that length of residence in the USA does not have a significant impact on party preference in American politics. In contrast, Lafleur and Sánchez-Domínguez (2014) in their study of Bolivians in Spain, the USA, Argentina, and Brazil argue that emigrants treat home and host country politics as entirely separate arenas. Voters may thus support a left-wing party with a pro-immigrant program in the country of residence and a right-wing party in the homeland. The few studies that examine party choice in the country of origin show that preferences mirror the voting behavior of nonmigrants with similar socioeconomic backgrounds in the country of origin (see Ahmadov and Sasse (2016b) for Poland; Lafleur and Sánchez-Domínguez (2014) for Bolivia). Do people vote for similar parties (in terms of left-right ideology) in Dutch and Turkish elections?

Context, methods, and operationalization
The current study focuses on Turkish citizens settled in a West European country, the Netherlands. Turks are the largest non-western immigrant group in the Netherlands, with a population of 397,471 in 2016 (Central Bureau of Statistics 2016a). This includes first-generation immigrants born in Turkey as well as the second generation, defined as citizens who have at least one parent born in Turkey. In many ways, Turks in the Netherlands are illustrative of Turks in Western Europe; with Austria, Germany, France, and Belgium, the Netherlands is among the European countries with the largest group of Turkish voters (Abadan-Unat et al. 2014). Like Turkish migration to other European countries, large-scale emigration to the Netherlands began in the late 1960s in the context of bilateral guest worker agreements; in the ensuing decades, guest workers were followed by political refugees and those emigrating for family reunification and formation reasons (Akgündüz 2008). Since the 1990s, Dutch-Turks are also highly active in Dutch politics as voters, candidates, and politicians. Until 2014, Dutch-Turks had no formal opportunities to participate in Turkish electoral politics on Dutch soil; dense, networked immigrant organizations were their main vehicles for involvement in transnational party politics (Mügge 2010).
In response to emigrant requests to attain external voting rights, the Turkish parliament Turkey is among the most active sending countries and has been managing the loyalties of its citizens abroad for decades (Danis¸and Parla 2009). Recently, under Erdogan, political ties with the 'diaspora have been increasingly formalized to strengthen the government's political leverage abroad (Ü nver 2013; Aksel 2014; Mencutek and Baser 2018). West European immigration countries with large Turkish populations, like Germany and the Netherlands, continue to be critical of Turkish involvement with 'their' citizens, and fear conservative influences that do not correspond with views on integration in the country of settlement (Mügge 2012). Diplomatic relations are part of the transnational electoral context. During our data collection diplomatic relations between the Netherlands and Turkey were tense. Issues of conflict included the political influence of Turkey on Turkish-origin citizens in the Netherlands, such as religious affairs (Ö ztürk and Sözeri 2018), the Turkish-Armenian genocide dispute, foster care families for Turkish origin children and Islamophobia in Western Europe. These tensions led to a diplomatic crisis in 2017 (after the period under study) when a Turkish minister was expelled from the Netherlands as an 'unwanted foreigner'.
The Institute for Migration and Ethnic Studies (IMES) and Türkevi Research Centre conducted exit polls at the two polling stations in the Netherlands. The surveys were  (1) to 'Very high trust' (5). Since these questions concern different aspects of political trust (generalized versus parties), we conducted robustness checks to ensure that the results hold when we analyze the years separately. The results for the separate years are mostly similar to the main results we present below, where we collapse the years (if otherwise, this is indicated in the text).

Turnout
In total, 2,798,726 Turkish citizens around the world were entitled to vote externally for the 2014 presidential elections, the first election in which Turkish citizens could vote in their countries of residence. Yet, the turnout was low: only 19 per cent of the eligible voted (Supreme Election Council 2014b, c). In Turkey the turnout was 77 per cent (Supreme Election Council 2014e). An explanation for the low turnout abroad is that Turkish emigrants could only vote at a particular time on appointment. Voters complained about different dates offered to members of the same family who had to travel hundreds of kilometers to consulates. Appointments were no longer necessary with the amendments in 2015 (Abadan-Unat et al. 2014; S¸ahin-Mencütek and Erdogan 2016). In 2015, the turnout was higher: 37 per cent of eligible voters casted their ballots at Turkish consulates, embassies, and borders (Supreme Election Council 2015a). Yet, turnout levels among Turkish emigrants was still low especially when compared with the participation levels in Turkey, which was 86 per cent (Supreme Election Council 2015d).
In the Netherlands, 87 ballot boxes were available for 240,315 Turkish voters, of whom 49,294 or 20 per cent turned out (Supreme Election Council 2014b). The majority cast their votes at the Turkish consulates in Deventer and Rotterdam (31,958 persons), with the remainder (17,336 persons) casting their votes at the border (Supreme Election Council 2014d). In line with the trend in other countries, turnout among Turkish citizens in the Netherlands was higher (30.09 per cent; 76,502 out of 254,221 eligible voters) for the national parliamentary elections than for the presidential elections in the previous year (Kranendonk et al. 2015). Table 1 details the demographic characteristics of respondents who participated in the survey during the presidential elections of 2014 and the parliamentary elections of 2015. While the demographic variables are comparable in the two samples, they differ from the demographic make-up of Dutch-Turks and Turks living in the Netherlands. The percentage of women, young voters (below the age of 30 years) and voters born in the Netherlands is significantly lower in our sample than in the Turkish and Dutch-Turkish population. Additionally, highly educated Turks and Dutch-Turks are over-represented in our sample (compared to their numbers in the Netherlands) (Central Bureau of Statistics 2016b). Taken together, the Turkish and Dutch-Turkish citizens most likely to vote in Turkish elections were: men, citizens above the age of 30 years, people born in Turkey (first-generation immigrants), and the highly educated. Women, people born in the Netherlands (second generation), and less educated Turks were less likely to cast their votes. Figure 1 illustrates that trust in the Turkish political system increases the probability of voting for Erdogan, but decreases the probability of voting for I˙hsanoglu and Demirtas¸. Voters thus base their judgment of trust on their attachment to the ruling party led by Erdogan.

Political trust and interest in Turkish and Dutch politics
A similar pattern can be seen in the parliamentary elections of 2015. Figure 2 shows that trust in the Turkish political system increases the probability of voting for the AKP and decreases the probability of voting for all other parties. Although the questions related to political trust were framed differently in 2014 (trust in politics) and 2015 (trust in political parties), they relate similarly to voting choices. Higher levels of political trust are associated with a higher probability of voting for the incumbent party and its presidential candidate.  Citizenship solely affects trust and interest in Dutch politics and not in the Turkish politics. The country of birth affects both trust in Turkish politics as well as in Dutch politics. Being born in Turkey significantly decreases trust in the Turkish political system in our observations in Turkish Presidential Elections in 2014. The robustness check shows that being born in Turkey increased trust in the Turkish political system in 2015, although this was not statistically significant. Individuals born in Turkey also portray higher levels of trust in the Dutch political system. Female voters are more trusting of and interested in Dutch politics than male voters. Educational level affects trust and interest in both Turkish politics and Dutch politics. Being highly educated decreases trust in Turkish politics and increases trust in Dutch politics and interest in both Turkish and Dutch politics. Figure 3 visualizes effects of age on trust and interest in Turkish and Dutch politics. Younger and middle-aged respondents are somewhat more likely to have higher levels of trust in Turkish politics than older voters, although for younger voters, the difference is only marginally significant. In comparison to both younger and older voters, middleaged voters show lower levels of trust in Dutch politics. Age does not affect interest in  Turkish politics, while interest in Dutch politics increases with age before flatlining from middle age. We include trust and interest variables as predictors of each other to research their correlation and control for demographic variables and education. The questions related to political trust were framed differently in 2014 (trust in politics) and 2015 (trust in political parties), we therefore present the years separately. To be clear, we do not assume causality between the trust and interest variables and are solely interested in their associations.
Controlling for demographic variables and education, trust in the Turkish political system appears to be negatively associated with trust in the Dutch political system. Interest in Turkish politics is positively associated, and interest in Dutch politics negatively associated (though not statistically significant in 2015), with trust in the Turkish political system. Interest in Dutch politics is positively associated with interest in Turkish politics, and with trust in the Dutch political system. Our models explain the least variance of trust in the Turkish political system in 2015 (at 9 per cent) and the most variance of interest in Dutch politics in 2015 (at 29 per cent).

Candidate and party choice
How do Dutch-Turks abroad vote and how does this compare to the political behavior of Turks residing in Turkey? In Turkey, the results of the 2014 presidential elections were as follows: despite growing political opposition, Erdogan came out as the winner almost 52 per cent of all votes cast by Turkish citizens. The combined candidate of 14 opposition parties-I˙hsanoglu-received around 38 per cent, while the pro-Kurdish candidate Demirtas¸gained almost 10 per cent of the votes (Supreme Election Council 2014e). In the Netherlands, Erdogan received 78 per cent of the votes cast, Ihsanoglu, 18 per cent, and Demirtas¸, around 4 per cent (Supreme Election Council 2014a). Table 5 illustrates the results according to our exit poll, which closely approximates the official results for the Netherlands. It shows that the support for Erdogan is substantially higher in the Netherlands among Dutch-Turkish voters than in Turkey itself.
Mirroring the results from the presidential elections, the AKP was the most popular party in the 2015 elections for the national parliament among Turkish voters living in the Netherlands. Support for the AKP was significantly higher in the Netherlands (Supreme Election Council 2015c) than in Turkey (Supreme Election Council 2015d). According to the official results, the AKP received 64.31 per cent of the votes cast in the Netherlands and 40.66 per cent of the votes cast in Turkey.
The regional and socioeconomic background of Dutch-Turkish AKP voters overlaps with the profile of the religious, lower and middle-class AKP constituency in Turkey. Economic prosperity for this group is one of the driving forces of Erdogan's success,  while the secular CHP is mostly supported by Turkey's urban elite. Due to its elitist reputation, the CHP has weak support among Dutch-Turks who originate mostly from rural areas (Kalaycıoglu and Ç arkoglu 2006;Toprak 2005). In line with this, the CHP won only 12 per cent of the vote in the Netherlands, compared to 25 per cent in Turkey. Support in the Netherlands for the pro-Kurdish HDP (11 per cent) and nationalist MHP (10 per cent) were a little lower than in Turkey. Table 4 shows the results of our exit poll for the 2015 parliamentary elections (AKP voters were under-represented and HDP voters over-represented in the sample). The survey for the 2014 presidential elections contained questions about voting preferences in Dutch politics. Table 7 shows that in the Dutch national elections, the Labor Party

Discussion and conclusion
Whether there is a trade-off between 'here (country of settlement) and 'there' (the country of origin) in the political behavior of citizens with immigrant backgrounds is a central concern in the immigration countries of Western Europe. We take this issue by the horns. Within a transnational framework of intersocietal convergence (Waldinger 2015), we study three components of political behavior among Dutch-Turkish citizens in the Netherlands: electoral turnout, trust and interest in politics, and most importantly party choice. Our findings show that, highly educated men above the age of 30 years who were born in Turkey are more likely to vote in Turkish elections in the Netherlands than women, the  second generation and/or less educated Dutch-Turks. Voters with greater trust in Turkish political parties have been more supportive of the AKP and its leader, which have been ruling Turkey since 2002. Voters with high levels of trust in Dutch political parties are less likely to support the AKP and its leaders. Political trust among Turkish-origin voters seems correlated with the feeling of acceptance and recognition in both political systems. While we find a trade-off between trust in Turkish and Dutch political parties, there is no such trade-off when it comes to interest in Turkish and Dutch politics. Voters tend to trust one system or the other, but this does not influence their level of political interest. Voters who are more interested in homeland politics tend to be more interested in politics in the country of residence as well. Interest in Turkish politics is positively associated with and interest in Dutch politics negatively associated with trust in Turkish politics. Interest in Dutch politics is positively associated with interest in Turkish politics and with trust in Dutch politics. When we compare party choice in the Dutch and Turkish elections, it appears that most Erdogan supporters voted for the Labor Party (PvdA) in the Dutch national elections in the period under study. They supported a conservative candidate in Turkey and a progressive social-democratic party in the Netherlands. This pattern can largely be explained by the voters' socio-economic backgrounds and social class. While the AKP represents the conservative, religious, lower, lower-middle, and middle classes, the Dutch Labor Party traditionally represents workers, including workers with an immigrant background. Support for Erdogan and the AKP among Dutch-Turkish voters confirms existing research that the AKP appeals to individuals with traditional beliefs and lower socio-economic status. For individual Dutch-Turkish voters this difference in voting behavior in Turkish and Dutch elections does not necessarily present a contrast. It illustrates their 'guest worker' background in the Netherlands and their origins in mostly conservative regions (Central Anatolia) in Turkey (Vermeulen et al. 2018).
Compared to other immigrant origin groups, Dutch-Turks are the most active in Dutch and homeland politics. Dutch-Turks also appear to be more critical of Dutch politics than other immigrant origin groups. Over the past years, several Dutch-Turkish politicians have clashed with the Labor party (PvdA) over issues in Turkish politics and policies regarding integration and Islam. This clash triggered the formation of a new political party, DENKthe only party with representation in the Dutch parliament led by politicians with immigrant backgrounds (Vermeulen et al. 2019). This development underlines how the situation in both sending and receiving states constantly changes and that such dynamic can lead to more convergence. It also illustrates that for the second generation 'here' and 'there' cannot be simply separated. Developments in both sending and receiving societies simultaneously affect the political behavior of citizens with a migration background.
Our results underline that the political behavior of citizens of immigrant origin cannot be captured as a simple trade-off. Transnational political behavior is complex and differs across its key components of voter turnout, trust and interest in politics, and party choice. Nevertheless, differences in turnout based on gender and country of birth point to the importance of considering diversity within immigrant-origin groups when studying transnational political involvement. Overall, we can conclude that there is no trade-off in interest between the politics of both countries.
Yet, our finding that there is a trade-off in trust between Turkish and Dutch political parties and systems, merits further more qualitative studies that go beyond survey material used in this study to explain this difference. One hypothesis is that how voters view their own position in the political system (sending or receiving) influences trust. A negative diplomatic relationship illustrates a more general discontent with either the situation of Dutch-Turkish citizens in Western Europe (among the large group of AKP-voters) or a discontent with the political developments in Turkey (among voters for oppositional parties). Until recently Dutch-Turkish voters saw 'here' and 'there' as two different political spheres as they voted for opposite ideological parties. Due to recent developments in Turkey and Western Europe overlap between 'here' and 'there' increases.
Future research should study what the citizens, who participate in two political systems, consider important for adequate transnational representation. How are 'here' and 'there' related in their views? How does convergence differ over time, per group, and between political contexts?