Intergenerational ties across borders: a typology of the relationships between Polish migrants in the Netherlands and their ageing parents

ABSTRACT

The question of how intergenerational relationships are maintained when family members reside in different countries has been gaining scholarly attention. However, those studies focus mostly on the so-called old migrant groups. The focus on the ‘new migrants’ from Central and Eastern Europe is still scarce. In this paper, we examine the transnational ties between Polish migrants in the Netherlands and their parents living in Poland. To identify types of transnational ties, we performed a latent class analysis using data on 970 men and women from the Families of Poles in the Netherlands (FPN) study. Following earlier studies on adult child–parent relationships in transnational context, we combined information on upward and downward emotional support, upward financial and practical support and frequency of contact (face-to-face and via communication technologies) and commitment to norms of filial obligation. Three types of transnational child–parent relationships were distinguished: harmonious, detached and obligatory. Multinomial regression analyses showed that that background characteristics of the adult children and their parents rather than the time elapsed since arrival in the Netherlands accounted for variability in relationship type. The relatively high probability of face-to-face contacts even in detached ties is characteristic of the strong commitment to family life among people of Polish descent.

KEYWORDS:

• Intergenerational relationships
• transnational ties
• typology
• polish migrants
• the Netherlands
• latent class analysis

Introduction

Studies on intergenerational family solidarity have repeatedly confirmed the crucial role of transfers up and down family lines for the lives of non-migrants in a wide range of European countries (Albertini, Kohli, and Vogel 2007; Hank 2007; de Jong Gierveld, Dykstra, and Schenk 2012; Albertini, Mantovani, and Gasperoni 2019). In the context of contemporary migration flows, the question of how intergenerational relationships are maintained when family members are separated by national borders becomes highly relevant and is, as yet, largely unanswered.

Migration disturbs safety nets. When adult children leave the country of origin, ageing parents are deprived of potential care and support (King and Vullnerati 2006; Bordone and de Valk 2016), and this is especially disruptive in societies where families play an indispensable role in care and welfare provisions (Zhou 2012). However, increasingly accessible communication means like telephone and social media (Dekker and Engbersen 2014) and low-cost travel undeniably simplify the maintenance of transnational relationships and the exchange of (financial) support across borders (Baldassar 2007; Baldassar et al. 2016). The question arises as to which view deserves more credibility: ties broken by migration or close ties maintained despite the distance.

Though transnational family relations have been receiving increasing scientific attention in Europe and beyond, most work focuses on transnational child-rearing (Parreñas 2005) rather than intergenerational ties in adulthood. Insofar as quantitative studies on transnational families have been carried out, they have examined dimensions of support separately (Attias-Donfut and Wolff 2008; Baykara-Krumme 2008), neglecting the nuances of intergenerational solidarity as a multifaceted construct (Bengtson and Roberts 1991). Notable exceptions are a study by Rooyackers, de Valk, and Merz (2016) – who simultaneously studied different aspects of intergenerational support and distinguished types of child–mother relationships of Turkish, Moroccan, Antillean and Surinamese migrants in the Netherlands and their mothers living in the countries of origin – and work by Baykara-Krumme and Fokkema (2019), who compared intergenerational solidarity patterns of child–parent relationships among Turkish stayer and migrant families.

Although those first attempts offer invaluable insights into transnational family relations, they focus on so-called old migrants in Europe (e.g. guest workers and migrants from previous colonies) and often also on second generation migrants. Much less is known about how ‘new’ migrants, i.e. recent migrants from Central and Eastern European countries, maintain their family relationships across borders. This group differs from ‘old’ migrants: they have a less restricted legal status, are geographically closer to their country of origin, and have greater options to engage in return migration (Engbersen et al. 2013). Conceivably, these structural conditions better enable supportive exchanges between migrant children and their ageing parents. A focus on Polish migrants is particularly relevant, given that they are the largest group in the post EU-enlargement migration wave. According to recent estimates, Poland has become the main sending country to the Netherlands, and the number of Polish migrants entering the country exceeds the number of ‘old’ migrant groups taken together (Gijsberts and Lubbers 2013).

The first attempts to describe the intergenerational ties of Polish migrants and their ageing parents have been based on small-scale in-depth studies and describe strategies that migrants develop to fulfil their obligations towards ageing parents living in Poland. In some cases, migrants are engaged in strenuous transnational care-giving for frail parents (Ryan et al. 2008), while in others remittances replace care by the adult migrant child (Krzyżowski and Mucha 2013). There is a lack of representative quantitative information on the family relations of Polish migrants that takes various dimensions of intergenerational relationships into consideration. We address this gap by constructing a typology of adult child–parent relations of Polish migrants in a transnational context. We study recent migrants, thus adding to our understanding of the dynamics of early stages of migration. Our main research question is: What types of relationships bind recent Polish migrants to the Netherlands with their parents living in Poland? Moreover, we ask whether the distribution of those types is also dependent on the socio-demographic characteristic of child and parent. Our second question is thus: Does the incidence of those types differ according to child’s and parent’s characteristics?

Our data are from the Families of Poles in the Netherlands (FPN) study, a recent survey on Polish migrants, who registered in the country after the EU enlargement in 2004 (Karpinska, Dykstra, and Fokkema 2016). Participants reported, among others, on flows of support and the relationship with their parents. The sample selected for this study consists exclusively of migrants with parents in Poland.

Family ties in Poland

Our goal is to identify patterns of child–parent exchanges in families where members are separated by national borders. As a starting point, we describe intergenerational family patterns in the country of origin. According to Iacovou and Skew (2011), Poland, together with other Central and Eastern European countries, has an extreme form of what they called a ‘Southern’ type: a family system with extensive co-residence of adult children and their parents and a low number of lone-parents households (and consequently, larger households). The country’s familialistic care regime, with a deficit of state-regulated formal care services, obliges family members to step in (Bordone, Arpino, and Aassve 2017) and reinforces a high level of intergenerational interdependence (Krzyżowski and Mucha 2013). Consequently, family ties of Poles are strong, and family members engage in frequent contact and exchanges of support (Iacovou and Skew 2011; Krzyżowski and Mucha 2013). Similarly to residents of other Central and Eastern countries, Poles abide by strong family obligation norms (Muresan and Haragus 2015). The role of kin keeper traditionally falls upon women (Krywult-Albanska 2016). Although Polish families have been undergoing considerable changes after the collapse of communism, with increasing divorce rates, postponement of marriage, and dropping fertility rates (Krywult-Albanska 2016), Poles consistently declare that family is the most important value in their lives (preceding health, career, respect or friendships: Boguszewski 2013; Titkow and Duch 2004), and the patterns of high exchanges between generations remain stable.

Context of migration

Poland experienced a massive emigration flow after the country joined the EU in 2004. According to recent estimates, approximately 2.4 million Poles left the country between 2004 and 2014 (Statistics Poland 2015). The majority moved to the United Kingdom (one of the three countries that lifted labour market restrictions for new migrants immediately after the EU accession) or Germany (a neighbouring country). Polish migrants usually migrated alone and had their partner and children join them at later stages. Parents of migrants mostly stayed behind (Szawarska 2014).

This massive migration flow has led many publicists to sketch a pessimistic view of the future of Polish families and the transmission of cultural practices (Barglowski, Krzyżowski, and Wiątek 2015). Public discussions emphasised the disturbing consequences of migration, with ‘orphaned children’ and ‘abandoned parents’ becoming the exemplification of the disruptive changes. There is, however, little scientific support for this view. On the contrary, the evidence suggests that families develop strategies to maintain close ties and even to be involved in care across borders, both for the old (Ryan et al. 2009) and for the young, with ‘mobile grandmothers’ as an example of temporary care for grandchildren in the destination countries (Barglowski, Krzyżowski, and Wiątek 2015).

Family solidarity in transnational context

The model of family solidarity developed by Bengtson and his colleagues (Bengtson and Roberts 1991; Roberts, Richards, and Bengtson 1991) often serves as the guiding framework in analyses of intergenerational relations in migrant and non-migrant families (e.g. Senyurekli and Detzner 2008). The model posits that solidarity consists of six dimensions of child–parent interaction: affectual (emotional attachment); consensual (agreement on values and attitudes); functional (patterns of resource sharing and exchanges); associational (frequency of contact between parent and child); structural (opportunities for exchanges and – related – geographic proximity of family members); and normative (strength of commitment to familial roles and feelings of family obligation).

Solidarity is shaped by the geographical proximity of family members: frequent face-to-face contact increases emotional closeness and facilitates other forms of exchange because it reduces the costs of support giving and also helps to make support-givers aware of recipients’ needs (Dykstra and Fokkema 2011).

The current study focuses on the ties that bind generations separated by national borders and not on co-resident dyads (de Valk and Bordone 2019). Not every dimension of intergenerational solidarity as proposed by Bengtson and Roberts (1991) can be put into practice by migrants in the same way non-migrants can. Migrants cannot engage in frequent face-to-face contact, and offering practical support or personal care is severely restricted, or at least very strenuous. However, ample research stresses that proximity is not an absolute prerequisite for solidarity in families. In the 1960s evidence already emerged that members of the extended family can maintain cross-generational cohesion through modern means of communication and transportation (Silverstein and Bengtson 1997). Forms of support that can bridge distances include exchanges of contact by telephone or other current communication technologies (Parreñas 2001; Faist 2004; Baldassar et al. 2016), exchanges of emotional support, and exchanges of financial support (Baldock 2000; Guo, Chi, and Silverstein 2011) and practical help offered during visits (Krzyżowski and Mucha 2013) – the associational, affectual and functional dimensions of the model of intergenerational solidarity. Engagement in support exchanges is strongly linked to norms of family obligation in non-migrant families (Gans and Silverstein 2006). We assume that the normative dimension of the model of intergenerational solidarity also predicts exchanges between migrants and their parents, although the practical possibilities for intergenerational support are more restricted in migrant families.

The literature gives us some hints as to how migration influences the ties between migrants and their parents, but we must note that the available evidence focuses mostly on migrants residing in the same country and largely adopts the parent’s perspective. In her analysis of uni-national second generation migrants in Germany, Baykara-Krumme (2008) suggested two patterns: particularly cohesive families (given the dominant value orientation in their countries of origin) or disintegrated families (given the separation across national borders). We elaborate on the two contrasting images to develop our predictions.

One point of view is that migrants continue to maintain the close family ties of their socialisation in the country of origin. Thus, Polish migrants preserve their traditional model of family ties based on a commitment to strong norms of filial obligation and frequent reciprocal exchanges. Frequent contact via modern media offers affordable means to bridge geographic distances and is the basis for mutual expressions of moral support and care (de Bruine et al. 2013). Even relatively frequent face-to-face contact is a possibility. The distance between the Netherlands and Poland does not prohibit (holiday) get-togethers, participation in key family events (such as festivities related to religious ceremonies), and trips in times of emergency or death. Such visits offer opportunities for help around the parental house or help with paperwork, etc. In fact, there is evidence that suggests involvement of Polish migrants in such practices (Krzyżowski and Mucha 2013).

Close contact is likely to be accompanied by frequent exchanges of financial support. As noted earlier, higher incidences of contact make the support-givers more aware of the needs of the support-receivers, which translates into higher financial support provision. The literature on non-migrant samples in Western and Northern Europe shows that financial support flows usually down the family line (Attias-Donfut, Ogg, and Wolff 2005; Albertini, Kohli, and Vogel 2007). However, families in ‘old’ origin countries generally show the reverse pattern. Aged people in countries with poor welfare state provisions need to rely on their families for support, implying an upward wealth flow (Baykara-Krumme 2008). A similar mechanism is likely among Polish migrants, as the welfare provisions in this country are not generous. Therefore, in our analysis, we assume that financial support will flow up the family line – from migrant children to parents – rather than in the opposite direction and that this form of support characterises close intergenerational ties.

Contrary to this positive view of close family ties preserved despite migration, is the view of family disintegration resulting from geographic separation. Migration is an opportunity for the adult child to redefine the terms of the relationship with the parent (Szawarska 2014). For those left behind, the child’s decision to migrate might be interpreted as an act of deliberate distancing from the parent. Polish cultural principles prescribe that support, assistance and care to ageing parents should be delivered by children (Krzyżowski and Mucha 2013). Thwarted parental expectations might lead to conflict (Szawarska 2010). Among the leavers, an uncomfortable discrepancy between internalised norms of filial obligation and the inability to act on them is likely to result in a downward adjustment of those beliefs (Dykstra and Fokkema 2011) and might serve as an incentive for withdrawal.

Even if the child receives the license to leave (Baldassar 2007), the ties can fade over time, as the social relations that are established in the destination country gain increasing importance (Creese, Dyck, and McLaren 1999). Senyurekli and Detzner (2008) show that migrants censor information passed to their parents in the country of origin: discussing the child’s problems would burden parents. When parents apply the same strategy and conceal information on troubling events in the country of origin (such as unexpected deaths), the child feels alienated and mutual trust is abused. Although this behaviour arises out of concern, estrangement is likely. Financial support might also diminish over time, especially if the migrant child does not intend to return to the country of origin (Burholt 2004; Wolff 2019) or when the pressure to remit becomes too high (Schmalzbauer 2004).

The previous considerations suggest that two types of intergenerational ties will emerge from the analyses. On the one hand, there is the possibility that intergenerational ties remain close, with a high frequency of contact and emotional exchanges and a strong commitment to norms of filial obligations. Given the awareness of parental needs and a clear sense of duty, the incidences of financial and practical support are high. On the other hand, there is the possibility that migration leads to the weakening of intergenerational ties, defined in terms of low incidences of contact and low emotional support and, also, weak commitment to norms of filial obligation. The disruption of the ties is accompanied by a disruption of financial help and practical support. Thus, we expect that two types of ties will be distinguished, defined in terms of high and low levels of solidarity.

Predictors of relationship type

Previous work on family solidarity in non-migrant populations reveals consistent differentiations by gender, age, and sibship size. Mothers and daughters, who traditionally are more likely to take on the role of kin keepers, demonstrate greater responsibility for maintaining contact and organising family life compared to father and sons (Gerstel and Gallagher 2001; Komter and Vollebergh 2002; van Gaalen and Dykstra 2006). Involvement in family relations also differs by age, with ties to parents being most intense in young adulthood and levels of exchange dropping when children enter middle age and become more involved in families of their own (Rossi and Rossi 1990). In our analysis, we control for whether migrants have children of their own, as supporting offspring can occur at the expense of support to parents, especially when resources such as time or finances are limited. As regards sibship size, a number of studies have shown that per-child exchanges are less frequent in larger families, and having a higher number of sisters lowers parent–child contact more than having brothers (van Gaalen and Dykstra 2006; van Gaalen, Dykstra, and Flap 2008; Emery 2013). We assume that the differentiations by gender, age and sibship composition that are observed for the family ties of non-migrants also hold for migrants. We also consider differences in the distribution of child–parent relationship type by level of educational attainment. The highly educated tend to have less intensive contacts with their parents, partly because they live further away from them, and partly because they tend to have weaker feelings of obligation (Kalmijn 2014). Of interest is whether such an educational gradient shows up in the ties of Polish migrants who are separated from their parents.

We also take a number of parental characteristic into account. We consider parent’s frail health, arguing that the migrant child will be offering more support, in the form of higher contact and emotional exchanges, to compensate for the inability to offer hands-on care (Krzyżowski and Mucha 2013). We also take parental marital status into account, given that divorced parents, and divorced fathers, in particular, tend to have fewer interactions with their offspring than parents who are married (Kalmijn 2014).

Finally, we consider the role of migration. More specifically, we take into account the duration of residence in the Netherland. Following Baykara-Krumme (2008) one might predict that ties with parents are particularly close during the initial phase of residence: family can be a haven and a source of support in the difficult times of settlement. An alternative prediction, which follows from the work of Creese, Dyck, and McLaren (1999), is that ties with parents are neglected during the early period of residence because migrants are devoting time and energy to establishing themselves in the country of destination. There are also contradictory expectations as to how transnational ties evolve as migration continues: some argue that migrant integration in the new country coincides with the loosening the ties with societies of origin (Alba and Nee 1997), while the existing evidence shows that migrants are able to maintain transnational ties over sustained periods of time (de Haas 2010). Given these competing views, we refrain from formulating a prediction about intergenerational relationship type and the duration of residence in the Netherlands. The analyses also take into consideration whether the respondent has a partner living in the Netherlands. Gijsberts and Lubbers (2013) suggest that family unification in the country of destination is an indicator of settlement intention, which could imply lower involvement in family relations in the country of origin. Similarly, a relationship with a native partner strengthens links to the receiving country and the ties with the homeland may lose their importance. Being single, on the contrary, provides freedom for a potential return and the motivation for maintaining kin networks back in Poland.

Data and method

Data

The data stem from the recent FPN survey (Karpinska, Dykstra, and Fokkema 2016). The data were collected between October 2014 and April 2015 among Polish migrants aged 18–59, who registered in the municipality where they were living in the Netherlands after the EU enlargement in 2004. The sample was drawn from the population registers via simple random sampling, offering national coverage. Web and computer-assisted personal interviews were conducted among a total of 1131 respondents. Both Polish and Dutch versions of the questionnaire were offered. The response rate was 51%, one of the highest among surveys of Polish migrants in the Netherlands.

The blueprint of the FPN survey is the 2015 questionnaire of the Gender and Generations Surveys (GGS, Aassve et al. 2013). Next, to questions about migration history and intentions to return or to move to another country, the FPN has questions about the family of origin, exchanges of money, practical and emotional support, espoused family obligations, and marital and parenthood histories. We restricted our sample to respondents with at least one surviving parent living in Poland. Parents living with the migrant child in the Netherlands or residing in a country other than Poland were not considered. If only one parent was alive or only one parent was living in Poland, this parent was included in the analyses. If both parents satisfied the selection criteria, we randomly selected one to avoid within-family dependency. In total, 970 child–parents dyads were included in the analyses.

Measures of intergenerational solidarity

Latent class analysis (LCA) was applied to construct the typology of child–parent relationships. Typically, variables used in LCA are dichotomised for reasons of parsimony and to increase manageability of the data (Dykstra and Fokkema 2011). Analyses were based on the following measures of family solidarity.

Two measures assessed the frequency of contact: how often per year a child saw his/her parent face-to-face and how often per year a child was in touch with his/her parent via telephone, Skype or other means of modern communication. The possible answers for both questions ranged from (1) ‘More than once a week’ to (6) ‘Seldom or never’. Given the efforts and costs involved in transnational visits and the facility of contact via communication technologies different time units were used for both variables. Consequently, contact face-to-face, pertained to whether the child saw the parent at least a few times a year (0 = no; 1 = yes), whereas contact via communication technologies denoted whether the child was in touch with the parent via telephone, Skype or other social media at least weekly (0 = no, 1 = yes).

For emotional exchange, we constructed two variables that capture support flowing up and down the family line. Emotional support offered denoted whether the parent had turned to the child to discuss matters that are important to him/her in the past 12 months (0 = no; 1 = yes). Emotional support received was constructed in a similar manner, but this time the child indicated whether he or she had turned to the parent to discuss problems in the last 12 months (0 = no; 1 = yes). Financial support indicated whether the child offered the parent any goods or money amounting to €250 or more in the past 12 months (0 = no, 1 = yes). Practical support was whether the child helped the parent with household tasks such as preparing meals, cleaning, small repairs and paper work in the past 12 months (0 = no, 1 = yes).

For norms of filial obligation, we first constructed scale-summing answers to questions on child responsibilities towards ageing parents. Four items used were: ‘Children should take responsibility for caring for their parents when parents are in need’; ‘Children should adjust their working lives to the needs of parents’; ‘Children ought to provide financial help for their parents when parents are having financial difficulties’, and ‘Children should have their parents to live with them when parents can no longer look after themselves’. The answers ranged from (1) ‘strongly disagree’ to (5) ‘strongly agree’. Subsequently, a dummy variable weak norm of filial obligation was constructed, with those scoring in the bottom 20% defined as refusing norms of filial obligation (0 = no; 1 = yes).

Predictors

To evaluate how the distribution of the classes varies by socio-demographic characteristics of respondents and their parent, we included the following measures. Next to gender of the child (daughter = 1) and parent (mother = 1), we also entered child’s age in the analysis. We also controlled for the educational level of an adult child (highest educational degree obtained, coded in International Standard Classification of Education (ISCED-97). We constructed three categories: low (e.g. pre-primary education, primary education or first stage of basic education, and lower secondary or second stage of basic education, educational levels of ISCED 0, 1 or 2), intermediate (upper secondary education, and post-secondary non-tertiary educations, education; ISCED 3 or 4), and high (first and second stage of tertiary education; ISCED 5 or higher).

We also looked at the duration of residence in the Netherlands; it ranged from 0 to 18 years. The higher residence durations deviate from the intended maximum of 10 years for migrants who registered after 2004. Some might have lived in the Netherlands illegally before the EU accession and registered only after Poland had joined the EU. Given that robustness checks showed that the results were very similar when the sample was limited to respondents who lived in the Netherlands up to 10 years, the original measure was used. Partnership status of the child was denoted by four dummy variables: single, having a Polish partner living in the Netherlands, having a non-Polish partner living in the Netherlands, and having a Polish partner living in Poland. Has children denoted whether the respondent has one or more children, living either in the Netherlands or elsewhere. We also looked at the number of sisters and the number of brothers, respectively (range from 0 to 7). Frail health of the parent is a dummy variable indicating whether the parent was limited in his/her daily activities because of physical or mental health problems (0 = no, 1 = yes). Parental partnership history had three categories: married/living with a partner, divorced/separated, and widowed.

Analyses

LCA is a procedure that assumes a probabilistic relationship between the latent concept (in this case, the structure of ties binding the migrant child and the parent living in Poland) and manifest indicators (e.g. measures of contact frequency and support) (van Gaalen and Dykstra 2006). We first estimated a latent class model with only one latent class (assuming no relation between manifest indicators) (Dykstra and Fokkema 2011). In subsequent steps we added one class after the other, determining the model fit based on Bayesian Information Criterion (BIC). We used the Latent GOLD 5.0 programme. After the class membership for all child–parent dyads was defined, we applied multinomial logistic regression analysis to determine the associations between the types of child–parent relationship, on the one hand, and child’s and parent’s characteristics, on the other. To facilitate the interpretation, we estimated marginal effects, which give the mean change in probability by one unit of the predictor, when other variables are kept constant at sample means. This analysis was executed using Stata 14.

Results

Descriptive results

Table 1 provides information on sample demographics and the solidarity measures. Mother–daughter dyads are the largest category in our data (35% of all dyads), followed by mother–son and father–daughter dyads (23% and 24% of all dyads, respectively), whereas father–son dyads are the smallest category (18% of dyads). In general, women are overrepresented in our sample as compared to the population of the registered Polish migrants in the Netherlands: almost 60% of respondents are women, as compared to 51% women among registered migrants.

Table 1. Descriptive characteristics of adult children and parents (N = 970).

	Range	Mean	Std. dev.
Dyads
Mother–daughter	0–1	0.35
Mother–son	0–1	0.23
Father–daughter	0–1	0.24
Father–son	0–1	0.18
Child’s characteristics
Age migrant child	21–57	32.8	6.71
Education child
High	0–1	0.28
Intermediate	0–1	0.48
Low	0–1	0.23
No. of brothers	0–7	1.04	1.10
No. of sisters	0–7	1.00	1.10
Length of stay in the Netherlands (in years)	1–18	6.37	3.19
Partnership status
Polish partner living in the Netherlands	0–1	0.65
Non-Polish partner living in the Netherlands	0–1	0.16
Polish partner living in Poland	0–1	0.04
Single	0–1	0.14
Has own child(ren)	0–1	0.53
Parent’s characteristics
Frail health	0–1	0.10
Partnership status
Married/living with a partner	0–1	0.56
Separated	0–1	0.24
Widowed	0–1	0.20
Indicators
Emotional support offered to parent	0–1	0.37
Emotional support received from parent	0–1	0.32
Financial support offered to parent	0–1	0.06
Face-to-face contact (at least a few times a year)	0–1	0.77
Contact via communication technologies (at least weekly)	0–1	0.57
Practical support offered to parent	0–1	0.01
Weak norm of filial obligation	0–1	0.19

The mean age of respondents was 32. Almost one in three respondents completed higher education, while close to 50% obtained a diploma at the intermediate level, and 23% were low educated. On average, respondents had been living in the Netherlands for 6 years. The majority lived in the Netherlands with their partner, who was either Polish (65%) or had another nationality (16%); 4% had a Polish partner who still lived in Poland, while 14% were single.

The majority of respondents (77%) reported seeing parents at least a few times a year and 57% had at least weekly contact via telephone, email or other media. The frequency of emotional support exchanges was less high: 37% of respondents offered support to the parent, while 32% received such support from the parent in the past year. In the majority of cases, the support was reciprocal (results not shown). Six percent of respondents reported having provided financial support to the parent in the past year and one% offered practical support. Nineteen percent had weak norms of filial obligation.

Latent class analysis

When determining an optimal number of classes, the preferred model is selected based on the smallest BIC and non-significant L². The BIC statistics of the three-class model was the lowest, with the value of 5973.6 (BIC of two- and four-class models were 6032.4 and 6016.9, respectively). The L² for a three-class solution had a value of 51.6 and was not statistically significant (p > .05), indicating that the optimal number of classes was three.

Table 2 summarises the predicted probabilities of the final model. The first class, as expected, is characterised by a high likelihood of contact both face-to-face and via communication technologies (0.84 and 0.80, respectively). Frequent contact is accompanied by a high likelihood of emotional exchanges both up and down the family line (probabilities of 0.96 and 0.95). The likelihood of financial support is moderate, while the likelihood of practical support and the likelihood of weak norms of filial obligation are low. We labelled this type harmonious, given its resemblance to a type reported elsewhere (Dykstra and Fokkema 2011; Rooyackers, de Valk, and Merz 2016). This type is the least frequent: 25% of all child–parent dyads can be classified as harmonious.

Table 2. Results of LCA of solidarity between migrant children and their parents living in Poland.

	Harmonious	Detached	Obligatory
Manifest variables (mean values on 0–1 scale)
Emotional support offered to parent	0.96	0.08	0.23
Emotional support received from parent	0.95	0.01	0.17
Financial support offered to parent	0.11	0.01	0.06
Contact face-to-face (at least few times a year)	0.84	0.51	0.94
Contact via communication technologies (at least weekly)	0.80	0.02	0.89
Practical support (in the past 12 months)	0.04	0.01	0.00
Weak norm of filial obligation	0.14	0.30	0.13
Frequency (percentage values)	25	34	41

The second type is also in line with our expectations and is labelled detached. The children–parent dyads in this class have a low probability of being involved in emotional exchanges, a moderate probability of at least yearly face-to-face contact and a low probability of contact via communication technologies. The likelihood of offering financial support and practical support to the parent is low, while the likelihood of weak norms of filial norms is substantial. Thirty-three percent of dyads belong to the detached type.

The third and unanticipated type is the largest, with 41% of the dyads belonging to this class. It is marked by a high likelihood of face-to-face contact at least a few times a year and contact via communication technologies at least weekly (0.94 and 0.89, respectively), but a moderate likelihood of emotional exchanges. The likelihood of providing financial or practical support is low, as is the likelihood of weak norms of filial obligation. Although the likelihood of exchanging emotional support is higher than in the detached type, it is considerably lower than for the harmonious type. Following van Gaalen and Dykstra (2006), we labelled this type obligatory.

To assess whether the distribution of relationship type varied by demographic characteristics of the child and the parent, we performed a multinomial logit regression. Table 3 depicts the marginal probabilities of these models. Gender differences are largely in line with our expectations. Daughters are more likely to be in harmonious relationships than sons, and mothers are also more likely than fathers to have this type of ties. Separate analyses revealed (results not shown) that mother–daughter dyads are more likely to belong to the harmonious type than are all other dyad configurations (e.g. mother–son, father–daughter and father–son). Child’s gender is not a predictor of being in detached or obligatory relationships, while mothers are less likely to have detached relationships.

Table 3. Marginal effects of multinomial logit regression predicting three types of relationships of migrant children and their parents (reference categories in parentheses; N = 970).

	Harmonious	Detached	Obligatory
Characteristics of migrant child
Daughter	+0.08**	–0.04	–0.04
Age	+0.00	+0.00	–0.00
Educational attainment (Low)
Intermediate	+0.05	–0.06	+0.01
High	+0.13**	–0.12**	–0.01
No. of brothers	–0.00	+0.00	+0.00
No. of sisters	–0.02	+0.03*	–0.00
Length of stay in the Netherlands (In years)	–0.00	–0.00	+0.00
Partner (No partner)
Polish partner living in the Netherlands	–0.11**	+0.01	0.10*
Non-Polish partner living in the Netherlands	–0.06	+0.05	+0.01
Polish partner living in Poland	–0.13	+0.10	+0.02
At least one child (No children)	–0.03	–0.00	+0.04
Characteristics of parents
Mother	+0.17***	–0.19***	+0.02
Frail health	+0.09*	–0.06	–0.02
Relationship status (Married/living with a partner)
Separated	–0.02	+0.12**	–0.10*
Widowed	+0.01	+0.00	–0.02
Separated × father	–0.05	+0.23**	–0.19

Note: Cox-Snell R² = 0.192. Significance levels: * p < .05, ** p < .01, *** p < .001.

Age does not distinguish between the types, but the distribution of relationship types varies by level of educational attainment. Compared to lower educated individuals, those with higher levels of education are more likely to maintain close harmonious ties with their parents. The likelihood of being detached from parents is lower for those with higher levels of education compared to the lower educated. A higher number of sisters is, as expected, positively associated with the likelihood of being part of the detached type, while having more brothers made no difference. Having own child(ren) did not distinguish between the types. Frail parental health is a predictor of the harmonious type. The likelihood of being part of a detached relationship is greater if parents are separated or divorced than if they are in intact marriages, and more strongly so for fathers than for mothers.

The role of migration showed mixed results. The distribution of relationship types does not vary by the duration of residence. We also tested whether the initial phase of the residence (i.e. up to two years of residence in the Netherlands) distinguished between the relationship types, but it was not the case (results not shown). Living with a Polish partner in the Netherlands is, however, related to a lower likelihood of having a harmonious relationship with the parent and a higher likelihood of being in a relationship based on obligation.

Discussions and conclusions

Family relations of migrants have been receiving increasing scientific attention (Albertini, Mantovani, and Gasperoni 2019). The current study explored the relationships of recent Polish migrants and their parents living in Poland, and in this effort took the perspective of the migrating adult child. Our main research question was: What types of relationships bind recent Polish migrants to the Netherlands with their parents living in Poland? Moreover, we explored whether the distribution of those types is dependent on childrens’ and parents’ socio-demographic characteristics. Acknowledging the complexity of intergenerational family ties, we considered multiple dimensions of solidarity: frequency of contact (face-to-face and via communication technologies), emotional support (given and received), financial and practical support (given) and commitment to norms of filial obligation. Our study has several unique features. It has a large sample size, contrary to most work on transnational family ties (e.g. Baldassar et al. 2016), focuses on adults rather than children (Mazzucato and Schans 2011; Zentgraf and Chinchilla 2012), focuses on ties with both parents rather than only ties with mothers (Rooyackers, de Valk, and Merz 2016), and is among the first to study family relations in a truly transnational context (i.e. all parents of the respondents resided in the country of origin). The analyses were executed using recent data on Polish migrants in the Netherlands.

Contrary to predictions, we found three rather than two types of adult child–parent relationships. We had expected to find the harmonious type, with its high likelihoods of emotional exchanges and contact (both face-to-face and by means of communication technologies) and a strong commitment to norms of filial obligation. The harmonious type also has the highest likelihoods of providing financial and practical support to parents, although the absolute support levels are rather low. We had also expected to find the detached type, with its lowest likelihoods of contact and exchange and low commitment to norms of filial obligation. The emergence of the obligatory type, distinguished by a high likelihood of contact (both face-to-face and via communication technologies) and virtually no exchanges, was unexpected.

Financial and practical support did not distinguish between relationship types, probably because the provision of these forms of support was not a common practice in the sample. We might say that geography matters: proximity facilitates the exchange of practical support (Dykstra and Fokkema 2011). Nevertheless, we should also note that the parents of the migrants are still relatively young and only a few face health problems. There might be no need to offer financial and practical support. Another interpretation is that the migrants financially support their own families rather than their family of origin (de Bruine et al. 2013). The majority of migrants in our sample united in the Netherlands with their Polish partners or started a family with a Dutch resident. They may be less likely to remit to their parents, preferring to invest money in their nuclear family. This interpretation is further substantiated by findings on the predictors of relationship type. Migrants with partners, regardless of the nationality and country of residence of their partners, are less likely to have harmonious relationships with their parents than are migrants who have no partner.

The types of adult child–parent relationships that emerged in our study are similar to those that have been established in previous work. The harmonious type is among the classes reported by van Gaalen and Dykstra (2006) for a Dutch sample, resembles Silverstein and Bengtson–s (1997) ‘tight-knit’ type in an American sample and is akin to the ‘emotional interdependent’ type in ‘old’ migrant groups in the Netherlands (Rooyackers, de Valk, and Merz 2014; 2016). Types resembling detached parent–child relationships have also been reported by the previously cited authors. Though we did not expect to find it, the obligatory type has also been distinguished by Silverstein and Bengtson (1997) and van Gaalen and Dykstra (2006).

In addition, the distributions of relationship types by background characteristics show few surprises. Relationships with daughters and mothers are most likely to be harmonious and least likely to be detached. The larger the number of sisters, the greater the likelihood of being part of a detached relationship. Adult children respond to their parents’ frailty through frequent contacts and exchanges of support. Parents who have separated are more likely to be detached from their adult children than are parents who are still married and more strongly so for fathers than for mothers. After separation, the likelihood that adult children engage in obligatory relations is lower for fathers than for mothers. The educational gradient in the distribution of relationship type is contrary to expectations. Migrants with higher levels of education are most likely to have harmonious relationships with their parents in Poland and least likely to have detached relationships. Yet, these patterns have also been reported elsewhere. Hogan, Eggebeen, and Clogg (1993) showed that adult children with more years of education were more likely to be involved in a high exchange relationship with their ageing parent, with different types of support flowing up and down the family line. Kalmijn (2006) reported that though the lower educated generally were more likely to have at least weekly contact with their parents, they were also most likely to have broken off ties. Taken together, the findings show that background characteristics of the adult children and their parents, rather than the time that has lapsed since the arrival in the Netherlands, account for variability in relationship type.

Given the similarity of our findings to previous work, the question arises as to whether there are any patterns that might be attributable to either transnationalism or to ‘Polishness’ – two unique characteristics of our sample. To properly answer this question we would need longitudinal data on the families of migrants over time and comparative information on families in Poland. Baykara-Krumme and Fokkema (2019) show, for instance, that the transnational relationships of Turkish migrants are different from those of Turks who remained in their country of origin. Similar patterns of selectivity might apply to Polish migrants and Poles in Poland. Nevertheless, we venture to identify results that suggest an effect of migration or a typical Polish family pattern. The finding that harmonious ties are least frequent, whereas obligatory ties are most frequent is, in our view, characteristic for adult children and parents who live in separate countries. There are efforts to keep in touch, but supportive exchanges are hampered by geographic distance. It seems to help to have a Polish partner: the likelihood of maintaining obligatory ties is higher in that case. Moreover, we cannot rule out a possible connection between migration and the high frequency of detached ties: the desire to escape non-rewarding family ties might have served as a motive to leave the country of origin. The finding that, even in detached ties, there is a rather high likelihood of face-to-face contacts is, in our view, characteristic for Polish family life. Get-togethers on the occasion of religious holidays and celebrations are highly valued. Szawarska (2014) notes that in some cases the strong sense of obligation drives visits to the country of origin, but those visits are seen as a burden. Family members visit even though they do not particularly enjoy spending time together.

The current study offers a glimpse into the complexity of the transnational ties that bind adult migrant children and parents living in the country of origin. While the findings suggest that their relationships bear resemblances to the close-knit ties of non-migrant Poles, there are also indications that geographic separation hinders intensive exchanges. Moreover, we cannot rule out that intergenerational autonomy facilitated the decision to migrate. Future research should focus on disentangling possible selection effects (migrants with particular family relationships are more likely to migrate) and the impact of migration on intergenerational family relations (by comparing migrants and non-migrants with the same background). How migrant child–parent relationships develop once parents residing in Poland become frail and require care, and how filial obligations are negotiated in the transnational settings, are other exciting avenues for future research.

Acknowledgements

Thanks to the editors and the anonymous reviewers for their comments on earlier drafts.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

We gratefully acknowledge financial support from the European Research Council, Advanced Investigator Grant ‘Families in Context’ (grant agreement no. 324211).