Transnational Activities and Identifications –A population-based study on three immigrant groups in Finland

ABSTRACT Our understanding of the lived experience of transnationalism is limited by the scarcity of high-quality quantitative evidence. Representative evidence on the emotional and identification aspects of transnationalism is especially missing. To address this gap, we investigated the transnational involvement of three migrant groups – Somali, Kurdish and Russian – in Finland. Using high-quality random sample survey data, the study examined transnational activities – remittance sending, following media, keeping in touch with and visits to the country of origin – as well as transnational identification with significant others living abroad. Multiple correspondence analysis (MCA) demonstrated that transnational involvement can be empirically divided into transnational activities and identifications, as suggested by prior theoretical works. Regression analysis showed that transnational activities and identifications varied strongly between the three groups. Russians were the most active in transnational activities while Somalis showed the highest levels of identification with their country of origin. Activities increased with the longer stay, while identification decreased. Family ties played a role in transnational identification. Our results point towards the complexity in the relationship between transnational involvement and economic integration in the host society.


Introduction
The development of communication technologies and ease of travel have been fundamental in changing the lived realities of migrants (Baldassar, 2007;Levitt, 2003;Madianou, 2012;Vertovec, 2001Vertovec, , 2004. Migrants keep their feet in both worlds -the societies of settlement and origin -living their everyday lives locally, but also connected with a transnational social field (Erdal & Oeppen, 2013;T. Kemppainen et al., 2020;Levitt, 2003;Levitt and Glick Schiller, 2004). In contrast to classic assimilationist theory, integration to the majority culture is not necessarily in conflict with transnational involvement or identification with one's own ethnic community (Alba and Nee, 1997;Berry et al., 2006;Hällsten et al., 2018;Lauglo, 2017;Nandi and Platt2015;Nguyen & Benet-Martínez, 2013;Sam & Berry, 2010; however, see Battu et al., 2007). Bicultural identification may even benefit migrants' psychological well-being (Berry et al., 2006; for review, see Hällsten et al., 2018). Building on these theoretical developments, T. Kemppainen et al. (2020) have emphasised the multifocal and multidimensional nature of migrant integration (see also Schunck, 2014). Levitt and Glick Schiller (2004) conceptualise transnationalism by distinguishing the actual doing, such as sending money or travelling (ways of doing) from emotional attachments and identifications (ways of being). Noting this, and drawing from the social psychological and anthropological studies on social identities, Snel et al. (2006) define general transnational involvement as consisting of transnational activities and transnational identifications. With transnational activities, the authors refer to migrants' economic, political and socio-cultural cross-border activities and practices (see also Al-Ali et al., 2001;Portes et al., 1999), and with identifications, to migrants' level of identification with their compatriots outside the host country.
Recent research has shown that both individual life circumstances in the host society and the social and political situation in the country of origin are associated with migrants' transnational involvement. While there is a growing body of studies on transnational activities, high-quality and representative evidence on the emotional and identification side of transnationalism is lacking, which limits our understanding of the transnational lived experience. Moreover, more detailed data and analysis on the different dimensions of transnational activities have been called for (Schunck, 2014). In addition, some consider that studies on transnationalism are suffering from inconsistent methods, data and/or analysis (Schunck, 2014;Snel et al., 2006). Finally, the research literature is contextually unbalanced with most of it focused on the United States (Itzigsohn & Giorguli-Saucedo, 2005;Portes et al., 1999Portes et al., , 2002Waldinger, 2008) and only a few studies examining the situation in Europe (Schans, 2009;Schunck, 2014;Snel et al., 2006).
This study responds to these research gaps by providing high-quality quantitative evidence on migrants' transnational involvement in the European context. We use representative face-to-face survey data on three migrant groups -Russian, Somali and Kurdish -in Finland. Our study extends prior studies in theoretical scope and data quality with a detailed and comprehensive investigation of the determinants of transnational activities and transnational identifications (see Levitt & Glick Schiller, 2004;Snel et al., 2006). In particular, we examine the role of socio-economic resources obtained through structural integration in the host society. Furthermore, we study how the length of stay, internet skills, and negative experiences in both the host society and the country of origin are associated with transnational involvement. Finally, we look at the geography of family ties, which has been shown to be an important force in transnational involvement (e.g., Schunck, 2014). Thus, we aim to produce theoretically and methodologically consistent evidence on migrants' transnational involvement.
Our research questions are as follows:

How common is transnational involvement?
Transnational activities are more widely studied than transnational identifications. The former can be divided into economic, socio-cultural and political activities (Al-Ali et al., 2001;Portes et al., 1999). Prior studies show that, in general, transnational activities are rather common among migrants and that family ties are an important driving force (Bloch, 2008;Schunck, 2014). Schunck (2014) argues that less costly activities are especially common among migrants, while 'deeper' transnational modes of living are rare. The authors also call for more detailed evidence on the different dimensions of transnational activities and their associations. Recent studies have especially examined economic activities, such as sending remittances (Carling et al., 2012;Lacroix, 2013;Schans, 2009), investments (Kuuire et al., 2016) or business (Ley, 2013) and entrepreneurship (Itzigsohn & Giorguli-Saucedo, 2005;Portes et al., 2002). While transnational business and entrepreneurship are rather uncommon, sending remittances is one of the most common transnational practices (see Mügge, 2016;Schunck, 2014). However, most of the evidence in Europe is non-representative. In previous representative studies from Norway, Germany and the Netherlands, the sending of remittances varied between 3.3% and 74% depending on the migrant group (Carling et al., 2012;Schans, 2009;Schunck, 2014). Carling et al. (2012) found that in Norway, 74% of migrants of Somali origin sent remittances. In previous studies from the United States, money transfers ranged from 47% to 74% (for review, see Schunck, 2014).
The most common socio-cultural transnational activities are keeping in touch with relatives abroad and visits to the country of origin. In the Netherlands, almost 90% of respondents kept in touch with their relatives who were left behind (Schans, 2009;Snel et al., 2006). Visits to countries of origin ranged from 69% in Germany (Schunck, 2014) to 76% in the Netherlands (Snel et al., 2006). However, the types of contact often differ in different migrant groups according to their economic and legal abilities to travel (see also Bloch, 2008).
Migrants' political transnational activities include, for example, political party membership, voting, lobbying and civic activities (Bilgili, 2014;Chaudhary, 2018;Khayati & Dahlstedt, 2014;Mügge, 2016;Østergaard-Nielsen, 2001;Van Bochove et al., 2010). In Europe, Kurds from Turkey have been especially politically active about the situation in their homeland by organising protests and lobbying in their new host societies (Khayati & Dahlstedt, 2014;Mügge, 2016;Toivanen, 2014). However, low-cost activities such as reading newspapers and keeping in touch with the politics of the country of origin are the most common forms of political activity, while party memberships, lobbying and protesting are less common (Schunck, 2014;Snel et al., 2006;Van Bochove et al., 2010). In Snel et al.'s (2006) study, 74% of migrants reported reading newspapers from their country of origin. Recent studies have argued (Bilgili, 2014;Bloch, 2008;Chaudhary, 2018) that political activities can be seen as complementary: if one is active in politics, they tend to be active in both their country of origin and the new host country.
There are fewer studies on transnational identifications and on the emotional side of transnational lives despite the importance of emotions in this context (e.g., Skrbiš, 2008;Toivanen, 2014). Emotions have been studied in relation to transnational families and care (Baldassar, 2007;Skrbiš, 2008). Some recent studies have approached transnational belonging from the point of view of emotional attachment and imaginative practices (Baldassar, 2007;Buffel, 2017;Klok et al., 2017). Imaginative transnational belonging can include nostalgia and longing for one's country of origin, as well as wishes to return. In Waldinger's (2008) study from the United States, 35% of respondents planned to return to their home country, while in De Haas and Fokkema (2011) study on Spain and Italy, the number was 28%.
The prior evidence is mixed and often suffers from a lack of representativeness. Our study fills an important gap in the European evidence by documenting the prevalence of several transnational practices. The same critique applies to the determinants of transnational involvement, which we turn to next.

Determinants of transnational involvement
In this section, we review the key literature on determinants that previous research has shown to be associated with transnational involvement, and this study focuses on socioeconomic resources, length of stay, negative experiences in the new host society, the situation in the country of origin, geography of family ties and internet skills.
The theoretical works of Tsuda (2012), Carling and Hoelscher (2013), Itzigsohn and Saucedo (2002), and Itzigsohn & Giorguli-Saucedo (2005) provide helpful conceptual tools for organising the prior literature. Tsuda (2012)distinguishes different kinds of relationships between integration in the host society and transnational involvement. In a zerosum relationship, engagement in one society decreases participation in the other, while co-existence refers to the independence of engagements -i.e. one not influencing the other. Furthermore, engagements can be in a positively or negatively reinforcing relationship. Considering this typology, recent studies have found evidence about the positive relationship between host society integration and transnational involvement (Bakker et al., 2014;Bilgili, 2014;Kuuire et al., 2016;Levitt & Glick Schiller, 2004;Schunck, 2014;Şimşek, 2018;Snel et al., 2006;Van Bochove et al., 2010). However, there is also some evidence on their competing nature (Ley, 2013). Carling and Hoelscher (2013) elucidate on the agentic and contextual aspects of the matter by arguing that engagement with transnational practices is a matter of capacity and desire, both of which are shaped by the situation in the country of origin and one's circumstances in the country of residence (see also Al-Ali et al., 2001). Furthermore, Itzigsohn and Saucedo (2002); Itzigsohn & Giorguli-Saucedo (2005)) describe different rationales that encourage or discourage transnational activities. On the one hand, resources obtained in the host society may enable transnational activities (resourcebased transnationalism), but on the other hand, migrants may react to negative experiences in the host society by seeking engagement from the transnational sphere (reactive transnationalism) (Castañeda et al., 2014;Snel et al. 2016; however, see Bakker et al., 2014). The study by Şimşek (2018) adds to this framework by outlining a further rationale for transnational involvement: it argues that refugees may use transnational activities strategically to adapt to their new host society.
Several studies have shown that economic integration in the new host country -and corresponding socio-economic resources -are positively associated with sending remittances (e.g., Bloch, 2008;Carling & Hoelscher, 2013;Lacroix, 2013;Schans, 2009;Snel et al., 2006), which supports the idea of resource-based transnationalism. However, there were mixed results on this. For example, Carling and Hoelscher (2013) found in their sample of different migrant groups in Norway that Somali migrants were the most active in remittance sending despite their low level of economic integration. Furthermore, Snel et al. (2006) found no evidence of the association of economic integration with transnational activities, but found an association between weaker employment status and transnational identification. In line with the idea of reactive transnationalism, there is evidence that people with less economic, social and cultural resources in the host country might be forced to rely more on transnational contacts (Levitt, 2003;Şimşek, 2018).
Relatedly, it is suggested that dissatisfaction with the new host country and negative perceptions by the host society's locals about one's own culture and religion are associated with the importance of identification with one's country of origin (Ehrkamp, 2005;Mügge, 2016;Schans, 2009). In their study of middle-class migrants in the Netherlands, Snel et al. (2006) found that the respondents were more transnationally involved both in activities and identifications if they had experienced discrimination.
Concerning the length of stay in the host country, Schunck (2014) found that transnational practices diminish over time, but Snel et al. (2006) showed that while transnational identification may weaken over time, actual transnational activities (e.g., sending remittances, visits) hardly diminish with an increased length of stay (see also McMichael et al., 2017). Furthermore, Schans (2009) found that length of stay was not significantly associated with sending remittances, but the frequency of other types of contact with relatives in the country of origin diminished over time.
In addition to migrants' circumstances in their new host society, the economic, political and social situation in their country of origin is an important factor shaping transnational involvement (Bakker et al., 2014;Carling et al., 2012;Chaudhary, 2018;Østergaard-Nielsen, 2001). Carling et al. (2012) found that migrants from conflict or recent post-conflict countries sent remittances more often than migrants from conflict-free countries. They argued that remittance sending becomes more important when one's relatives live in a conflict zone and have an acute need of help. On the other hand, war or conflict can prevent visits to one's country of origin while limited telephone and internet access could make keeping in touch with relatives difficult (Bakker et al., 2014). However, even if actual transnational activities are hindered by the situation in the country of origin, Snel et al. (2006) found no significant differences in the dimension of identification in transnational involvement.
Turning to the geography of family ties, family-related reasons have been shown to be important for both transnational activities (see Schunck, 2014) and the emotional side of transnationalism. De Haas and Fokkema (2011) found that having family in one's country of origin was associated with nurturing the wish to return (De Haas & Fokkema, 2011). On the other hand, Buffel (2017) and Klok et al. (2017) found that family (children and grandchildren) living in the host country is also the reason for older migrants choosing to 'age in place' -stay in their new home country. Thus, the location of family members seems to be an important factor for transnational identification.
Finally, recent studies have suggested a de-territorialised sense of belonging through a virtual or imagined transnational community (Kok & Rogers, 2017). The internet and new media technology have widened the possibilities for transnational activities and identifications by enabling stronger and richer social, cultural and political connections to one's country of origin (Chen, 2006;Panagakos & Horst, 2006). Studies have shown that the internet facilitates transnational business activities (e.g., Chen, 2006). However, in their online networks analysis of Somali political activities, Kok and Rogers (2017) found that Somalis' political online engagement was mainly directed towards their host society and local Somali community instead of transnational advocacy for their homeland. The internet may be especially important for transnational identification since it enables the widening of social space and new forums for social interaction and identity across distance (Madianou, 2012;Panagakos & Horst, 2006). However, the availability of the internet, technologies and devices along with individual user skills can restrict different groups' access to the virtual sphere and contribute to the so-called digital divide (e.g., Castells, 2004;Madianou, 2012).

Three migrant groups in Finland: background
Finland is a Nordic welfare state characterised by a universal right to social welfare and health-care services. Finnish migration policy is based on government objectives, the common migration and asylum policy of the European Union and various international agreements. In a European comparison, the integration policy of Finland has been characterised as multicultural since it has implemented relevant legislation and policy programmes and provides some resources for activities that maintain minority cultures and languages. However, according to Saukkonen (2013), on the practical level, the integration policies are based on migrants' individual adaptation to Finnish society, instead of on the development of Finnish society towards actual multiculturalism. On the symbolic level, the national identity is still constructed upon the ideals of a homogeneous nation based on language and historical traditions, which is reflected in the anxiety and hesitation towards multiculturalism (Saukkonen, 2013). Finland has a long history of emigration, but a relatively short history of net inward migration. The size of the foreign-born population in Finland has tripled from 1998 to 2018. Today, the foreign origin population constitutes about 7.3% of the total population, or roughly 400,000 people in 2018 (Statistics Finland, 2019a). This study examines Russian, Somali and Kurdish origin populations in Finland, which together comprise more than a quarter of the foreign-born population in the country (Statistics Finland, 2019a).
Migration from Russia to Finland has generally been voluntary and much of it is explained by labour migration and personal relationships. With a population of approximately 79,000, Russian-speakers are the biggest migrant group in Finland (Statistics Finland, 2019b). People originating from Russia are considered to be less visible and culturally more proximal to the Finnish population than many other migrant populations (Liebkind & Jasinskaja-Lahti, 2000). However, there is evidence of discrimination against Russians and job applicants with Russian-sounding names (Liebkind et al., 2016). One specific type of migration from Russia is the return migration of the Ingrians, who are of Finnish descent and have been living on the Russian side of the Finnish-Russian border. The state programme for remigration of the Ingrian Finns began in 1990 and ended in 2016, resulting in the return of more than 30,000 Ingrians (Finnish Immigration Service, 2016).
Migration from Somalia to Finland began in the early 1990s following the political conflict in Somalia. The size of the Somali-speaking population in Finland is approximately 21,000 (Statistics Finland, 2019b). This group has been found to bare a significant degree of prejudice (Jaakkola, 2009) and faces a wide range of discrimination in Finland (Sotkasiira & Haverinen, 2016). Not feeling welcomed or accepted as full members of the Finnish society is considered a major obstacle to the integration of Somalis in Finland (Tiilikainen et al., 2013). Moreover, compared to the native population, Somalis suffer more often from various health and well-being related issues, ranging from psychological distress to problems in functional capacity (Mölsä et al., 2014: cf. Rask, 2018. People of Kurdish origin have generally moved to Finland as refugees, asylum seekers or for family reunification. The size of the Kurdish-speaking population in Finland is approximately 14,000 (Statistics Finland, 2019b). Like the Somali, Kurdish migrants also face different kinds of discrimination (Rask, 2018), which has been found to be associated with mental health issues, a poor quality of life, feelings of insecurity, decreased societal trust and poor self-rated health (Castaneda et al., 2015;Rask et al., 2018). Separation from one's primary family has been demonstrated to be significantly associated with the well-being and integration of migrants of Somali and Kurdish origin in Finland (Rask et al., 2016). Recent studies have drawn attention to multifocal marginalisation risk (T. Kemppainen et al., 2020) and the co-occurrence of substance use, affective symptoms and suicidal ideation (Salama et al., 2020) within the Kurdish migrant population in Finland.

Data
This study was based on a Finnish survey project on migrants' health and well-being (Maamu) and focused on migrants with a Russian, Somali or Kurdish background (Castaneda et al., 2012). The target population was defined as follows: aged between 18 and 64 years, residing in Finland for at least 1 year, and currently residing in one of the following six cities: Helsinki, Espoo, Vantaa, Turku, Tampere or Vaasa. Further definitions were group-specific: for the Russian sample, birth in the Soviet Union or Russia and a mother tongue of Russian or Finnish, for the Somali group, birth in Somalia, and for the Kurdish group, birth in Iraq or Iran and a mother tongue of Kurdish (Sorani) were required. A random sample of 3,000 people was drawn from the National Population Register and was stratified by city and migrant group. The data were collected in 2010-2012 by trained staff who spoke Finnish and the language of the respective target group. Structured face-to-face interviews were used. All in all, 70% of the Russian sample (n = 702), 51% of the Somali sample (n = 512) and 63% of the Kurdish sample (n = 632) participated in at least one part of the survey. The Coordinating Ethical Committee of the Helsinki and Uusimaa Region approved the Maamu study and each participant provided written informed consent. For a detailed account of the methodology of the Maamu study, please see Castaneda et al. (2012).

Indicators
The outcome variables on transnational involvement were operationalised with the following items, which were entered into multiple correspondence analysis (MCA; see sections 3.3 and 4.2) to obtain composite indicators for use in the regression analysis. These items were chosen on theoretical grounds based on prior studies (see T. Kemppainen et al., 2020)  The key explanatory variables were the following (see Table 1 for details): • Socio-economic resources (main activity, education in Finland, household income) • Length of stay • Geography of family ties (place of residence of father, mother and siblings) • Negative institutional experiences in the country of origin: war, imprisonment, torture (yes to any/no) • Experiences with discrimination in Finland (yes/no) • Self-assessed internet skills (yes/no) • Migrant group (Russian, Somali, Kurdish) • Basis of residence permit The general socio-demographic control variables were age, gender, city and living with a partner. In the case of education, income and siblings living abroad, cases of missing responses were more numerous compared to other variables. Thus, the missing cases were included in the regression analysis as categories of their own, which is an oft-used solution to retain analytical power.

Methods
MCA was used to examine the covariation of the five items on transnational involvement. It is largely analogous to principal component analysis and enables data reduction for categorical indicators, which holds the advantage of increasing content validity and reliability when operationalising complex concepts, such as social integration (L. Kemppainen et al., 2018). MCA was performed with Burt's approach and standard normal coordinates were extracted (Greenacre, 2007;Le Roux & Rouanet, 2010;StataPress, 2015). Adjusted sampling weights were used in managing the variation in inclusion probabilities and unit nonresponse. Response probability adjustment was based on age, gender, migrant group, city and marital status. Prevalence and regression analyses were performed while accounting for the adjusted sampling weights, sample stratification and finite population correction. Stata 15.1 was used.

Results
Tables 1 and 2 present descriptive statistics for explanatory variables and the indicators of transnational involvement, respectively.

Prevalence of transnational activities
To answer RQ1, we estimated the prevalence for different transnational activities. Table 3 presents these results both for the pooled data and separately for each migrant group. In the case of all five activities, the differences between groups were substantial and statistically significant (p ≤ 0.001). In all three groups, more than half followed what happens in their country of origin daily. Among the Kurdish, the rate was as high as 71%. However, the Somali and Kurdish migrants were more polarised in this respect compared to the Russians, as sizeable segments in both groups very rarely or never followed the events of their countries of origin.
There are clear differences between groups regarding travel to the country of origin. The Russian migrants frequently visited their home country, as almost 80% visited their country of origin more than five times while living in Finland. In contrast, 73% of the Somali migrants had never gone back to Somalia. The Kurdish migrants were between these extremes.
Fairly frequent contact with friends or relatives living abroad was typical among all groups. The Russians were most active in this respect, since around half of them were in transnational contact with their friends or relatives at least weekly. The Kurdish migrants were the least active, but still around one-third of them were in such contact at least weekly.
The rate of sending remittances varied from 24% among the Kurdish to almost 50% among the Somali. The pattern of sending material help was different, as it was most frequent among the Russian (53.4%) and rarest among the Somali (6.2%).

Dimensions of transnational involvement: a multiple correspondence analysis
Relating to RQ2, we wanted to see how many dimensions we could distinguish empirically in the pattern of items on transnational involvement. The motivation was to see whether the prior theoretical distinctions made in the literature had empirical grounds in our data.
To this end, we performed MCA for the variables on transnational involvement. The results showed that there are two strong underlying dimensions that together express around 70% of the total variation in the set of items (46.6% and 23.5%, respectively). Moreover, the interpretation of the dimensions is theoretically meaningful as it largely coincided with the theoretical distinction between transnational activities and identifications. Consider Table 2, which shows how the classes of the variables entered in MCA are positioned along these two dimensions. Looking at the scores in the first dimension, we see that travelling to the country of origin had a clear gradient-like pattern, where those who never travelled back scoring −0.78 units, those who made one to five trips scoring −0.21 units and those who tended to travel more frequently scoring 1.1 units. The MCA indicators were standardised to the mean of zero and a standard deviation of one, therefore the pattern was strong.
In similar fashion, sending material or economic help showed a clear pattern along the first dimension, ranging incrementally from a score of −0.54 units among those who did not send any kind of help to 1.59 units among those who helped the most. Also, following the country of origin's media and maintaining interpersonal transnational contacts were logically related to this indicator, but the pattern was not completely incremental. All in all, the first dimension can be interpreted as indicating transnational activities, because it shows concrete travel and help-sending patterns.
Turning to the second dimension, there was a clear pattern regarding the subjective importance of family and friends living abroad. Those who did not consider these social ties especially important in their life scored −1.86 units, those who said they are rather important scored −0.80 units and those for whom transnational contact with family and friends was very important scored 0.46 units. Also, following the country of origin's media showed an easily interpretable pattern, extending incrementally from the score of −0.61 units among the least interested to 0.25 units among the most interested.
Interestingly, travelling to the country of origin was in a contrasting relationship with the second dimension: those who travelled the least had the highest mean score (0.50) and vice versa. Again, interpersonal contact was largely in an interpretable relationship with the composite indicator, but not in a completely incremental manner. Considering all this, it is justified to interpret the second MCA dimension as an indicator of transnational identification, characterised by the subjective importance of family and friends living abroad and interest in what happens in the country of origin.

Determinants of transnational involvement: regression analysis
Next, let us consider the determinants of transnational involvement (RQ3). We separately analyse the two dimensions of involvement that are identified above in section 4.2.

Socio-economic resources
Concerning transnational activities, those with a secondary vocational education in Finland (ISCED levels 3 and 4) were more active than the highly educated (Table 4).
Moreover, there was a clear income gradient where higher income levels were associated with stronger tendencies for transnational activities. The main activity did not have an independent relationship with transnational activities. In the case of transnational identification, household income was the only statistically significant determinant: those with a higher income identified more strongly with the transnational sphere compared to the less well off, but the result was weaker and less consistent than in the model of transnational activities.

Negative experiences in the country of origin and host society
Negative institutional experiences in the country of origin seemed to curb transnational activities, but not transnational identification. Moreover, experienced discrimination in the host society showed non-significant but possibly important weak signals for both outcomes: positive association with transnational activities (0.08; p = 0.066) but negative association with identification (−0.10; p = 0.056).

Length of stay
The results on the length of stay showed that those migrants with the shortest time of stay in Finland were the least involved in transnational activities. In contrast, those with the longest had the weakest transnational identification.

Geography of family ties
The residence of parents or siblings abroad did not make a difference in transnational activities. In contrast, having a father living in Finland was associated with weaker transnational identification.

Internet skills
Self-assessed capacity to use internet was positively and strongly associated with both forms of transnational involvement. In the case of transnational activities, the coefficient reached 26% of the outcome variable's standard deviation, while for identification, the coefficient amounted to 35%. In both cases, these were fairly strong regression results.

Other variables
Finally, let us consider other variables of the model, including the migrant group and residence permit. The tendency for transnational activities was strongest among the Russian migrants and lowest among the Somali. The Kurdish migrants were closer to the Somali in this respect. The largest difference between migrant groups (Russian and Somali) reached 125% of the standard deviation of the outcome, which is a very strong regression finding. In contrast, the Somali migrants had the strongest transnational identification. Moreover, there was no statistically significant difference between the Russian and Kurdish migrants. The largest between-group difference (Somali and Kurdish) represented around 90% of the outcome variable's standard deviation; again, a strong result.
A work-based residence permit was positively associated with transnational activities. The Ingrian Finns as well as the partners or children of a native Finn had the weakest transnational identification.
Living with a partner was positively associated with transnational practices and almost reached the conventional level of significance for transnational identification (0.12; p = 0.054). Moreover, women had stronger transnational identification compared to men. Finally, there were also significant differences between the six cities in terms of transnational activities.

Discussion
In this study, we examined migrants' transnational involvement by using high-quality random sample survey data on three migrant groups in Finland. We found that transnational activities and identification varied strongly between the three groups. Following media of the country of origin was the most common practice, especially for Kurdish migrants. Also sending remittances was a common practice.
Our analysis showed that transnational involvement can be empirically divided into transnational activities and identifications. The activities dimension is formed around the actual economic, social and political practices of sending remittances, keeping in touch with relatives abroad and following the media in one's country of origin. The dimension of identification is based on the importance of experience of family members abroad and following the country of origin's media. With regard to group-based differences, Russians were the most active while Somalis showed the highest levels of identification with their country of origin. These results can be interpreted in the context of the geographic proximity of Russia as well as the cultural and geographic distance between Somalia and Finland. Also, the specific migration history of the Ingrian Finns, who are considered of Finnish heritage, explains the negative association of the Ingrian background with identification with the country of origin. Interestingly, women showed higher levels of identification with the country of origin.
Our results pointed towards the complexity in the relationship between transnational involvement and economic integration to the host society. We found evidence for the coexistence of the two as well as positive and negative relationships between them (Tsuda, 2012). Empirically, this complexity has manifested itself in mixed results in prior literature. Moreover, we found that income level was positively associated with transnational activities, which is consistent with the idea of resource-based transnationalism (Itzigsohn & Giorguli-Saucedo, 2005;Itzigsohn & Saucedo, 2002) -i.e. the interpretation that once migrants have established a sufficient source of income in the host society, they are better able to, say, send remittances. Similarly, the working residence permit was associated with transnational activities. This might be explained by one's original expectations and reasons for migration: when entering the country to work (compared to family reasons or refugee status), one reason for migrating may be to provide monetary and material support to those left behind.
With regards to length of stay, we found that the temporal pattern of transnational activities may differ from the pattern of identification. While activities showed an increase with longer stay, identification had a decreasing temporal pattern (cf. Snel et al., 2006). As a methodological remark, there was no conclusive evidence on the causality of this matter. The observed decreasing temporal pattern of transnational identification was consistent with at least three different processes. Over time, the migrants with the strongest transnational identification may end up moving back, with the result that the remaining population identifies less transnationally with each passing year. However, time may weaken identification due to, say, increased identification with the host society (Tsuda, 2012). Third, there may also be, in theory at least, cohort differences such that the earlier cohorts are less inclined towards transnational identification, which could depend on a specific historical situation in the country of origin. However, the latter explanation is less credible because the findings were consistent with our three contexts of origin and with Snel et al.'s (2006) empirical case.
Prior literature has highlighted the geography of family ties for the wish to return (Buffel, 2017;De Haas & Fokkema, 2011;Klok et al., 2017). We were able to examine the role of family ties for transnational activities and identifications by considering separately the places of residence of the mother, father and siblings. We found that having a father living in the host society was fairly strongly associated with weaker transnational identification. Further qualitative inquiry could shed light on the interpretation of this finding: does it have to do with the father's cultural position in these -or some of these -cultures?
Considering the context of origin, there are studies on the role of wars and conflicts for transnational involvement (Bakker et al., 2014;Carling et al., 2012). We offered novel evidence on this topic as we approached the subjective side of institutional tensions and conflicts, such as experienced torture and war. More specifically, we found that such negative experiences may hinder transnational activities, whereas identification seems not to be affected (see also Snel et al., 2006). Concerning prior discussion on the matter, our findings were in line with the argument made by Bakker et al. (2014) about the difficulties that conflicts imply for actual activities, such as phone calls or money transfers (cf. Carling et al., 2012).
Prior discussions have referred to the facilitating role of technological developments (e.g., Vertovec, 2001) and to a de-territorialised sense of belonging (Kok & Rogers, 2017). Our findings largely support these views, since self-assessed internet skills were associated with higher levels of both transnational activities and identification. In other words, those who considered themselves unable to use the internet were less involved in transnational matters, which may be seen as a manifestation of digital inequalities.
It is well known that cross-sectional observational data limit causal inference, basically due to possible unobserved confounders and, often, an inability to analyse the direction of causality. Additionally, the groups in this study have been defined a priori by the researchers. This does not necessarily mean that the studied subjects would consider the groups formed in this study to be meaningful for them. For example, the categorization of the Russian origin group can be contested, as this group also included Ingrian Finns who might not identify as either Finnish or Russian (Rask, 2018).
Finally, the merits of our study included the use of representative high-quality survey data, empirical analysis of the patterns of transnational involvement and a comprehensive set of indicators covering transnational involvement and its determinants. Moreover, we provided evidence on the role of negative subjective experiences in the country of origin. All these points extend our understanding of the complexity and dynamics of transnational involvement in this contemporary world, which is characterised by many ongoing conflicts and tensions.

Disclosure statement
No potential conflict of interest was reported by the authors.

Notes on contributors
Laura Kemppainen, PhD, is a University Researcher at the University of Helsinki. She does research on sociology of health, migration and transnationalism studies and ageing. Currently, she works in the Finnish Centre of Excellence on Research in Ageing and Care (CoE AgeCare) and in the project 'DigiIn -Towards socially inclusive society' funded by the Strategic Research Council at the Academy of Finland.
Teemu Kemppainen, PhD, is Senior Lecturer in Urban Geography at the University of Helsinki and an associated fellow at Centre Maurice Halbwachs (CNRS/EHESS/ENS), Paris. In addition to immigration, he has been studying well-being as a contextual phenomenon from various aspects, including theory of recognition, welfare regimes, urban contexts and insecurity. His other current research interests include mixed methods methodology, critical theory and urban studies. Shadia Rask, PhD and occupational therapist, is a research manager at the Finnish Institute for Health and Welfare (THL) and a member of the Expert Group of Cultural Diversity (MONET) at THL, Finland. Her main research interests are related to diversity and health inequity in the population. Her published works focus on migration and health, particularly the mental health and physical functioning of populations of migrant origin and the association between experiences of discrimination and health and well-being.