Comparative studies play an important role in our understanding of the interdependencies between different family members and their impact on intergenerational relationships. Employing typologies assists researchers and others to better understand certain factors in such relationships between and within different countries. Thus, the focus of this article is on multidimensional intergenerational family relationships, viewing and comparing them from European and Israeli perspectives.
In an era of global aging, individual life courses and family forms are diversifying (Lowenstein, Reference Lowenstein, Johnson, Bengtson, Coleman and Kirkwood2005). Additionally, the changing graphic representation of the age structure of our societies, from a triangle to a rectangle, has important implications for generational ties (Lowenstein & Bengtson, Reference Lowenstein, Bengtson, Bengtson and Lowenstein2003). Increase in longevity and decrease in fertility have caused the age pyramids to become rectangularized in most industrialized societies, with almost similar numbers in each age category – children, middle aged and those above 60 as well. Population aging results in fewer family members and caregivers being available for older people. Parallel to the aging of populations, changes in family structures occur that create uncertainty in intergenerational relationships and expectations, and a shrinking pool of family support. Population aging creates a changing balance in care provision between older and younger people in society and between families and the state. Fewer younger people mean fewer children and grandchildren, and fewer family members and caregivers looking after older people in need of care. These trends create challenges for seeking new ways of generational communication, social inclusion, and social integration.
Demographic shifts represent an ever-present structural change in modern society. Increased life expectancies imply that an individual will be a member of a three- and/or four-generation family for a longer period. A collapse in fertility, changes in timing of family transitions – especially marriage, and parenthood and grandparenthood – suggest that others will never be members of such multigenerational families. Additionally, changes in patterns of family formation and dissolution, and the ensuing new styles of family and household forms, lead to more complex and “atypical” household structures. This diversity of family forms creates uncertainty in intergenerational relationships and expectations, and has specific effects on life-course role transitions such as retirement and grandparenthood.
The family arguably constitutes the most basic social institution, representing the very first group into which one enters at birth, and their ties to it remain primary throughout life (Hoff & Tesch-Römer, Reference Hoff, Tesch-Roemer, Wahl, Tesch-Römer and Hoff2007). The modern family is a family of relationships, more “centered on people rather than things” (Durkheim, Reference Durkheim1933), and has to ensure the conditions for construction of personal and social identity for each of its members. The structural organization of the family is particularly critical for those in middle age, a phase in life when individuals are likely to play multiple roles (Lowenstein & Katz, Reference Lowenstein, Katz, Dannefer and Phillipson2010).
Aging can become either a risk factor or an opportunity for realizing new possibilities. Changes in social policies suggest less government responsibility for support of older adults, with associated pressures downloaded onto families. Given these trends, our goal is thus threefold: (1) to explore the various dimensions of generational solidarity between older parents and their adult children using the second wave of SHARE (Survey of Health, Ageing and Retirement in Europe), comparing European countries and Israel; (2) to replicate and compare the results of Dykstra’s and Fokkema’s (Reference Dykstra and Fokkema2011) typologies of generational relationships from the first to second wave data; and (3) to analyze and compare previous data on generational relationships from the OASIS study (Old Age and Autonomy: The Role of Service Systems and Intergenerational Family Solidarity) (Katz, Lowenstein, Phillips, & Daatland, Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005). Even though OASIS included only data collected from four European countries and Israel, it was one of the first studies to be conducted on relationships between older parents and middle-aged children from a comparative perspective.
Review of the Literature
The literature review includes three main parts: the first outlines and discusses the multidimensionality of generational relationships; the second presents the importance of conducting comparative cross-national research; and the third highlights the importance of constructing typologies for understanding similarities and differences within and between countries.
Generational Relationships
Generational bonds between adult family members may be even more important today than in previous decades since individuals are living longer and thus can share more years and experiences with other generations (Bengtson, Giarrusso, Silverstein, & Wang, Reference Bengtson, Giarrusso, Silverstein and Wang2000; Connidis, Reference Connidis2001; Lowenstein, Katz, & Gur-Yaish, Reference Lowenstein, Katz and Gur-Yaish2007). There is a wealth of theoretical frameworks for intergenerational family relations emerging from diverse academic disciplines such as psychology, sociology, social work, gerontology, and economics. There is, however, little theoretical work combining these corpora of knowledge in a comparative view. The present article attempts to fill some of these gaps by combining insights from sociology, gerontology, and economics within a cross-national perspective.
Whereas in most modern welfare states the family continues to bear the major responsibility for elder care (e.g., Katz et al., Reference Katz, Daatland, Lowenstein, Bazo, Ancizu, Herlofson, Bengtson and Lowenstein2003), patterns of generational support are becoming more complex. Esping-Andersen (Reference Esping-Andersen1990, Reference Esping-Andersen1997) developed the best-known typology of welfare regimes. He differentiated between welfare states according to three types: market-liberal, conservative-corporatist, and social-democratic. The typology provides a solid base for comparing European countries. Reher (Reference Reher1998) suggested that European countries are located along a north-south axis, and also along a dimension from presumably more collectivist family traditions in the south to more individualist traditions in the north, and modern welfare states have developed in different ways that reflect their distinctive cultures. Not only patterns of co-residency in old age, but also policy options, account for some of the patterns of informal care in Europe. The prevalence of informal care (both provided and received) is higher in countries with developed care services, whereas in those countries where services are not available, or which rely on cash benefits, the intensity of informal care resembles that of a full-time occupation (Rodrigues, Huber, & Lamura Reference Rodrigues, Huber and Lamura2012).
The conceptual framework of intergenerational family solidarity represents one of several enduring attempts to examine and develop a theory of family relationships for adult family life (e.g., Rossi & Rossi, Reference Rossi and Rossi1990). The solidarity model is a conceptual scheme for describing sentiments, behaviors, and attitudes in parent-child and other family relationships (see Giarrusso, Silverstein, Gans, & Bengtson, Reference Giarrusso, Silverstein, Gans, Bengtson, Johnson, Bengtson, Coleman and Kirkwood2005; Roberts, Richards, & Bengtson, Reference Roberts, Richards, Bengtson, Pfeifer and Sussman1991; Silverstein & Bengtson, Reference Silverstein and Bengtson1997). As such, it suggests that generational relationships within families represent complex social bonds, and that family members are linked by multiple kinds of solidarity that may be complementary (Katz et al., Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005; Lowenstein, Reference Lowenstein2007).
Bengtson and Schrader (Reference Bengtson, Schrader, Mangen and Peterson1982) codified six principal dimensions of solidarity: associational, affectual, consensual, functional, normative solidarity, and family structure. It was empirically supported that each of the multiple dimensions of solidarity is distinct (orthogonal), and each represents a dialectic. The solidarity model contains statistically independent components that divide substantially into two general dimensions: (1) structural-behavioral (associational, functional, and structural); and (2) affective-cognitive (affectual, consensual, and normative) (Lowenstein, Katz, & Daatland, Reference Lowenstein, Katz, Daatland and Silverstein2004).
In the 1990s, however, this framework was challenged because of its normative underpinnings (Marshall, Matthews, & Rosenthal, Reference Marshall, Matthews, Rosenthal, Maddox and Lawton1993). The solidarity paradigm was consequently modified to become the “family solidarity-conflict” model, which incorporates conflict and also focuses on the possible negative effects of too much solidarity (Bengtson, Giarrusso, Mabry, & Silverstein, Reference Bengtson, Giarrusso, Mabry and Silverstein2002; Lowenstein, Reference Lowenstein2007). In the late 1990s, the theoretical framework of solidarity-conflict was again challenged with the introduction of family ambivalence, a newly revived concept for studying parent-child relationships in later life (Connidis & McMullin, Reference Connidis and McMullin2002; Lüscher, Reference Lüscher, Pillemar and Lüscher2004), which suggests that intergenerational relationships may generate ambivalence between family members. The term intergenerational ambivalence is proposed to reflect the contradictions in relationships between older parents and their adult children along two dimensions: at the level of the macro-social structure of roles and norms, and at the psychological-subjective level of cognition, emotions, and motivation (Lüscher & Pillemer, Reference Lüscher and Pillemer1998).
Comparative Cross-National Research
Comparative studies are driven by two contrasting goals: the search for generalities and the search for distinctiveness. Employing both approaches might facilitate finding cross-national similarities as well as differences and national idiosyncrasies. Using this approach helps foster a critical dialogue between different countries and cultures to enhance our understanding of generational solidarity in various societies. This was noted by Bengtson et al. (Reference Bengtson, Giarrusso, Mabry and Silverstein2002), who indicated that we must look outside national borders to construct global conceptualizations of families and aging. This is particularly important because most family studies, since they are conducted in a single country, may have an unacknowledged ethnocentric bias. From comparative findings, we might map some future directions and key challenges that generational relationships currently present (Katz et al., Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005).
With data from five national samples (Norway, England, Germany, Spain, and Israel), OASIS was one of the first cross-national studies that empirically investigated the theoretical frameworks of solidarity-conflict and ambivalence (Katz et al., Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005). SHARE is a more recent international study with rich data on family relationships. In both studies, researchers identified various types of family relationships.
Construction of a Typology
Typologies are systematic classifications of phenomena that have characteristics or traits that are similar in some respects and dissimilar in others. Since the early days of empirical social sciences, the construction of typologies has been an important method in both qualitative and quantitative analysis (Kluge, Reference Kluge2000; Weber, 1949/1904). Employing typologies assists researchers and others to better understand certain conditions or factors. The social scientists’ construction of typologies is probably an expression of a more general human disposition to make the distinctions and group phenomena described in phenomenological research (Schutz & Luckmann, Reference Schutz and Luckmann1974).
The literature provides some evidence of ways in which the solidarity dimensions might serve to distinguish types of families, even though we cannot anticipate the number of family types that might emerge. As mentioned earlier, several cross-national studies constructed typologies of generational relationship types such as OASIS (e.g., Katz et al., Reference Katz, Daatland, Lowenstein, Bazo, Ancizu, Herlofson, Bengtson and Lowenstein2003; Lowenstein et al., Reference Lowenstein, Katz, Daatland and Silverstein2004). Later OASIS data were compared with data from the Longitudinal Study of Generations (LSOG) (Silverstein, Gans, Lowenstein, Giarrusso, & Bengtson, Reference Silverstein, Gans, Lowenstein, Giarrusso and Bengtson2010). Recently, Dykstra and Fokkema (Reference Dykstra and Fokkema2011) analyzed data from SHARE and developed another typology of family relationships based on the solidarity dimensions. In the study discussed in this article, we further addressed the question of whether a typology of generational solidarity, using multiple dimensions of solidarity, could be empirically distinguished.
From the literature review, it seems that some basic, cross-cultural questions must be addressed concerning family relationships and generational solidarity in modern and post-modern societies. We aimed to answer the following questions: (1) Are there similarities/differences between various societies regarding intergenerational family relationships? (2) Are there similarities/differences between different societies at different points in time?
Method and Data
Data Sources
We based our analyses on data from the second wave of SHARE (SHARE project: http://www.share-project.org/), conducted in 2006–2007 (for a more detailed description, see Börsch-Supan, Hank, & Jürges, Reference Börsch-Supan, Hank and Jürges2005). SHARE is a multidisciplinary, cross-national bank of microdata on health, psychological and economic variables, family and social support of individuals aged 50 and older, living in the community. The first wave took place in 2004–2005 in 11 countries. Israel joined the project some time later, and the first SHARE wave took place there in 2005–2006. Further data were collected in all 11 countries and in the Czech Republic and Poland (in 2006–2007) and in Israel (in 2009–2010; see http://www.share-project.org/home0/wave-2.html). This article focuses on the same 11 participating countries in the first wave of SHARE, and thus we have broadened our discussion to changes in intergenerational family relationships and explanatory variables over time. To test the robustness of the findings in this study, we estimated the same econometric analyses, including Israel. All the measures we used in the analyses (socio-demographic and generational solidarity) were based on Dykstra’s and Fokkema’s (Reference Dykstra and Fokkema2011) study.
Measures of Socio-demographic Attributes
Socio-demographic characteristics of the parents included gender (coded 0 = male, and 1 = female), age (50–59, 60–69, and over 70), marital history (three categories: living with a partner, single after being widowed, single after divorce), health problems (1 = yes if subject reported difficulties performing one or more activities of daily living, reported severe limitations in performing usual activities for at least the past six months because of a health problem, or rated general health as poor), household income (quartile measure: ≤ €16,115 for bottom 25%, ≥ €62,933 for top 25%), educational attainment on three levels: low, intermediate, and high, and religiosity (based on the question, “Thinking about the present, about how often do you pray?”, with four categories: daily, weekly, less than weekly, never).
The measures of the socio-demographic characteristics of adult children were aggregate indicators. They included the number of children (coded as 1, 2, 3, and ≥4), having one or more daughters (1 = yes), one or more children living with a partner (1 = yes), one or more children with a paid job (1 = yes), one or more divorced children (1 = yes), and one or more children with high educational attainment (1 = yes). Following Dykstra and Fokkema (Reference Dykstra and Fokkema2011), we studied indicators of the need to provide help in kind, the availability to get help in kind, and the readiness to provide financial support or any support exchange.
Measures of Generational Solidarity
The following solidarity measures were used. Geographic proximity indicated whether the parent had at least one child living within five kilometers (0 = no, 1 = yes). Frequency of contact pertained to whether the parent had more than weekly contact with one or more children either in person, by telephone, or mail (0 = no, 1 = yes). The family obligation norms variable was based on items assessing opinions on state versus family responsibility for elder care, combined with items assessing opinions on the duty to care for children and grandchildren. With regard to support exchange, three dichotomous variables (0 = no, 1 = yes) were constructed: (a) downward help in kind: whether the parent had provided personal care, practical household help, or help with paperwork, or had looked after the grandchildren “almost every month” in the past year; (b) upward help in kind: whether one or more of the adult children had provided personal care, household help, or help with paperwork “almost every month” to the parent, and (c) downward financial transfers: whether the parent had given any financial or material support amounting to €250 or more to any of the adult children from outside the household during the past 12 months.
Based on Dykstra’s and Fokkema’s (Reference Dykstra and Fokkema2011) latent class analyses, using the aforementioned measures revealed four types of generational solidarity that were robust across European countries and Israel. The family types were (a) descending, (b) ascending, (c) supportive-at-distance, and (d) autonomous. Descending familialism is characterized by living close to and having frequent contact with children. In this solidarity type, belief in family norms overrides state responsibility to support elders. Also, parents are greater providers of in-kind support to their adult children. Ascending familialism is similarly characterized by living close to children with frequent contact and belief in family norms; however, in this type, the children are the main providers of in-kind support. The third family type, supportive-at-distance, is characterized by not living close to children but having frequent contact with them. Belief in state responsibility to support elders overrides family norms. Parents are the greater providers of financial support to their adult children. The fourth family type, autonomous, is characterized by not living nearby the children, having a low level of contact; with belief in state responsibility to support elders and with few support exchanges. Each family type in the study data was present in each country, but the distributions varied (for further details, see Dykstra & Fokkema).
Data Analysis and Sample
We followed the mode of analyses presented by Dykstra and Fokkema (Reference Dykstra and Fokkema2011) using the second wave of SHARE data. The analyses used these data for all the European countries that originally joined the first wave of the SHARE project, and in so doing we could address changes over time.
Data analyses refer to respondents who had at least one living child (24,262 cases out of 27,835 individuals aged 50 and older in the second wave). We further restricted the analyses to parents who had no children living at home (14,581 cases in the second wave. Including Israel brought the relevant figures to 26,617, 30,299, and 15,975 respectively).
We applied a multinomial logit regression analysis, which is an extension of the logistic regression model for a dichotomous response variable, and is used in situations wherein the response variable is composed of more than two categories and there is no natural ordering of the categories. If the response variable has j categories, then in order to construct the logits (log of the odds) in the multinomial case, one of the categories must be considered as the reference level and all the logits are constructed relative to it. Following a latent class analysis, we used this type of econometric model to determine the associations between family types and the socio-demographic characteristics of parents and their offspring. The multinomial logit model (MNLM) is appropriate because the categories of the four late-life family types in our model – descending familialism, ascending familialism, support-at-distance, and autonomous – are discrete, nominal, and unordered in our reference group: elders in Europe and Israel.
Table 1 presents the descriptive information about the second wave of SHARE for the European countries that originally participated in its first wave. We also added descriptive information on all these countries including Israel. The results appear in the last column of Table 1.
Table 1: Descriptive characteristics of parents and adult children in the analysis sample, European countries that originally participated in both waves, SHARE Wave 2 (weighted percentages)
Notes: Weighted percentages, SHARE Wave 2.
a Dykstra & Fokkema (Reference Dykstra and Fokkema2011).
b Sample size: 14,581.
c Sample size: 15,975.
On comparing the socio-demographic data of both waves, we observed some differences. The second wave data were composed of a younger age group; a higher rate of older respondents who lived with a spouse; a higher level of religiosity (i.e., more prayed daily and fewer never prayed); an increased ratio of number of children (i.e., increase in having three and four children); a decrease of adult children who lived with a partner; and an increase in number of adult children with higher education.
Our analysis was based on the four-typology model described above, using latent class analysis and multinomial logit regression.
Results
Our main goal was to compare data on generational solidarity from the second wave of SHARE with Dykstra’s and Fokkema’s (Reference Dykstra and Fokkema2011) first-wave data, where a four-type typology was developed, and to follow the Dykstra and Fokkema model to ensure no possibility of any omitted-variable bias. In an attempt to compare differences from the second-wave data of SHARE with results based on its first wave, in the analyses we included the same countries that participated in both waves. Additionally, in an attempt to compare second-wave SHARE data with and without Israel, we included Israel as one of the surveyed countries.
Table 2 provides information on the distinguished family types. It presents latent class analysis with factor loading in each class. As can be seen in the last row of Table 2, 32 per cent of families were of the first type (descending familialism), 29 per cent of the second (ascending familialism), seven per cent of the third (support-at-distance), and 32 per cent of the fourth (autonomous). These percentages are the cumulative probabilities for all families of the respective types.
Table 2: Latent class analysis of solidarity between parents aged 50 or older and their non–co-resident children – European countries that originally participated in both waves
Significance levels: *** p < .01; ** p < .05; * p < .1
Note: Sample size: 12,178 (age 50+).
Source: Survey of Health, Ageing and Retirement in Europe (SHARE) – Wave 2.
Data in Table 3 show the socio-demographic, health, and economic attributes of respondents across the four family types. Of older parents in supportive-at-distance or descending familialism family types, 51–53 per cent were female, respectively, while only 40-42 percent of older parents in autonomous or ascending familialism were female, respectively. Among all types there was a higher likelihood for older parents living with spouses. The likelihood of having health problems was higher in older parents in autonomous or ascending familialism types, while the likelihood of a higher level of education was higher among older parents in supportive-at-distance or autonomous types.
Table 3: Descriptive characteristics of parent and adult children in the analysis sample – parents aged 50+, European countries that originally participated in both waves, SHARE Wave 2 (%)
Notes: Weighted percentages. Sample size: 14,581 (age 50+).
Source: Survey of Health, Ageing and Retirement in Europe (SHARE) – Wave 2.
Older parents in a descending familialism type are characterized by having a relatively higher level of religiosity. Older parents in descending familialism or supportive-at-distance types are characterized by having higher household income. With regard to adult children, the probability of having children and living with a partner is lower in parents in a supportive-at-distance type, whereas the probability of having children with a paid job or higher education is higher in older parents in the ascending familialism or autonomous family types. Older parents with at least one divorced child have a higher probability of being categorized as a descending familialism type. The inclusion of Israel has left the statistical situation intact.
The results of the multinomial logit regression of the linkages between characteristics of parents, adult children, and countries regarding the four family types in our study are presented in Table 4. The econometric analysis used data from the European countries that originally joined the first wave of the SHARE project, and in so doing we could address changes over time, compared with the results obtained by Dykstra and Fokkema (Reference Dykstra and Fokkema2011). We added a period variable to control for the time lapse between the waves.
Table 4: Multinomial logit regression model of the four type of late-life families –age 50+, European countries that originally participated in both waves: Marginal effects (reference group – Type 4: Autonomous)a
Significance levels: *** p < .01; ** p < .05; * p < .1
Notes: Sample size: 12,178 (age 50+).
Source: Survey of Health, Ageing and Retirement in Europe (SHARE) – Wave 2.
a A period effect was included in the model. It was found to have a positive effect on the ascending familialism and the supportive-at-distance types and a negative effect on the descending familialism type.
The autonomous family type was used as the reference group. To interpret the MNLM results, we estimated marginal effects (Liao, Reference Liao1994). The marginal effect provides the change in probability by one unit change in an explanatory variable, when all other variables are held constant at sample mean values.
The first family type, descending familialism, held a greater likelihood that single parents with a higher household income and a higher level of education would be included in this family type. These results were consistent with the aging process and people’s tendency to help their children when their income is high and they can afford it. In Dykstra’s and Fokkema’s (Reference Dykstra and Fokkema2011) typology (Table 5, p. 560), the likelihood of single parents to be included within this type was negative. Household income and education did not have any effect. Children with attributes such as living with a partner, being in the workforce, or having higher education, had a lower probability of being included in this type, which differed from the previous study. Differences existed among the various European countries: whereas the highest probability was found in Sweden, Denmark had the lowest likelihood (compared to Austria as the reference country) of being included.
Table 5: Distribution of late-life family types by country: Comparing SHARE Wave 1 (11,181) and Wave 2 (15,975) (weighted percentages)
Notes: Wave 1, based on the 11 European countries in the Survey of Health, Ageing and Retirement in Europe. Sample size: 11,181. Source: Dykstra and Fokkema, Reference Dykstra and Fokkema2011; Wave 2 based on the 12 European countries in the Survey of Health, Ageing and Retirement in Europe, including all European countries that originally participated in Wave 1, including Israel. Sample size: 15,975.
In the second type, ascending familialism, fathers were more likely to be in this type than mothers. However, in Dykstra’s and Fokkema’s (Reference Dykstra and Fokkema2011) typology, gender had no effect. Parents with a lower household income or lower educational level were also more likely to be included in this type. In their typology, some differences were found in these variables. Parents with a medium level of religiosity were less likely to be included. In Dykstra’s and Fokkema’s typology, level of religiosity had no effect. Parents who had daughters were more likely to be included. In Dykstra’s and Fokkema’s typology, having daughters was not associated with the ascending familialism type. Children with attributes such as living with a partner, being in the workforce, or having higher education showed a higher probability of being included in this type, whereas being divorced had a negative probability of inclusion. These results illustrate the extent of parents’ economic dependence on their children and their spouses. They also reflect the ability of children to support their parents. All these results differed from those of the previous study.
In Dykstra’s and Fokkema’s typology, children with spouses or with a higher level of education were less likely to be included within this type, whereas being divorced or in the workforce were not associated with the ascending familialism type. Differences existed among the various European countries, whereas Spain had the highest likelihood, and France, Denmark, and Switzerland had the lowest likelihood (compared to Austria as the reference country) of being included in ascending familialism.
In the third type, supportive-at-distance, there was a higher likelihood of parents with spouses, a higher household income, higher education, and having at least one daughter, to be included in this family type. These results were consistent with the aging process and people’s tendency to help when they could afford it. In Dykstra’s and Fokkema’s typology – as in our study – children with attributes such as living with a partner or belonging to the workforce had a negative probability of being included in this type.
In the present study, children with higher education had a lower probability of being included, while divorced children had a higher probability of being included in the supportive-at-distance type. However, in Dykstra’s and Fokkema’s typology, children with higher education had a higher probability of being included, whereas being divorced had no effect. Differences existed between the various European countries. Spain had the highest likelihood of being included in this family type, whereas in the previous wave Sweden and Denmark had the highest likelihood. In Switzerland, the probability of being included was found to be negative (compared to Austria as the reference country).
In the fourth family type, autonomous, older fathers were more likely to be included than mothers. There was a higher likelihood for older parents with spouses, higher education, a relatively higher level of religiosity, or at least one daughter, to be included in this family type. Children with a partner, in the workforce, or having higher education, had a higher probability of being included (see Table 4).
Adding Israel did not change the descriptive trends. However, we found that, in Israel, the highest probability of being included in any of the four types of late-life families was in the descending familialism type, and it had a negative probability of being found in the ascending familialism type. In Israel, the probability of being included in the supportive at distance type was found to be negative.
Distribution of the Four Family Types across Europe and Israel
In Table 5 we present the distribution of family types by European countries based on the two waves, and Israel was added in the second wave.
As we can see, the distribution of the four family types was quite similar across the 11 original countries in Wave 1 compared to Wave 2. In only two countries – Sweden (35 in Wave 1; 29 in Wave 2) and Greece (19 in Wave 1; 26 in Wave 2) – a five per cent difference was observed within the autonomous family type. In Israel, about a third of respondents were in the descending familialism type where solidarity is strong, and in-kind transfers flow downward, and a further third in the autonomous type with few support exchanges. The similarities in the distribution of the four types between the countries were reflected in the European means. We compared three different means: that of the first wave (Dykstra & Fokkema, Reference Dykstra and Fokkema2011), with 11 countries; that of the second wave with the same 11 countries; and another one with 12 countries (including Israel). Two thirds of the respondents were almost equally divided between the descending familialism and autonomous types, around 30 per cent were in the ascending familialism type, and the rest, less than 10 per cent, were in the supportive-at-distance type.
Results of Comparison between the Present and an Earlier Study
In addition to the first wave of SHARE data from 2004 (Dykstra & Fokkema, Reference Dykstra and Fokkema2011) as shown above, the data presented in our current study were compared to another data set from a multidisciplinary cross-national research project: OASIS (Norway, England, Germany, Spain, Israel), with a wide range of data on family generational solidarity.
OASIS Typology
In order to understand the complexity of generational relations in the five OASIS countries, we used correspondence and cluster analyses and obtained a four-cluster typology. The analyses were based on the solidarity dimensions and on conflict and ambivalence. We did this by identifying the groups of parents whose responses were as closely related as possible within each group, and as different as possible between each group. It yielded the following category types: close, steady, ambivalent, and distant (Katz et al., Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005; Phillips, Ogg, & Ray, Reference Phillips, Ogg, Ray, Lowenstein and Ogg2003), as presented in Table 6.
Table 6: Cluster parent-child relationships by country – OASIS study (%)
Note: Numbers in parentheses show the percentages of the specific type among all respondents.
p < .01
Source: Katz et al., Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005 (based on the OASIS Final Report, chapter 7, 2003).
The close relationship type represents 24 per cent of respondents, and was characterized mostly by parents feeling extremely close to their child. The steady relationship type, the largest category, represented 32 per cent of parents. These parent-child relationships were more emotionally distant than the first group, although they could still be distinguished as close. Parents generally got on well with their children, but perhaps they liked to keep some emotional distance. The third was the ambivalent group, representing 27 per cent of parents. This group showed signs of a generational gap emerging between parents and their children. These parents tended to feel neither emotionally close nor distant from their adult child. There was conflict in these relationships, which seemed to result more from the distancing of their relationships rather than from their closeness. Mixed feelings began to show significantly when compared to the close and steady groups. The distant group was the smallest, representing 17 per cent of parents. This was clearly the group where relationships showed signs of emotional distancing and were more likely to have conflict, mixed feelings, and to express different norms regarding family responsibility.
Inter-country differences, as presented in Table 6, showed important differences in relationship styles between the five countries participating in the study. Ambivalent relationships were most evident in Germany, Spain, and Norway, distant relationships in England, and close relationships in Israel. German parents differed from those in all other countries since a majority of parent-child relationships were either ambivalent or distant (for further details, see Katz et al., Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005; Lowenstein, Reference Lowenstein2007).
Discussion
In this investigation, we analyzed and compared generational solidarity between parents and their adult children based on several international databases: SHARE (first two waves) and OASIS. The main conceptual framework used in the analyses of these studies was the solidarity paradigm developed by Bengtson et al. (Bengtson & Roberts, Reference Bengtson and Roberts1991; Silverstein & Bengtson, Reference Silverstein and Bengtson1997). Empirical analyses of SHARE (first wave) yielded four family relationship types: ascending, descending, supportive-at-distance, and autonomous (Dykstra & Fokkema, Reference Dykstra and Fokkema2011). An earlier study (OASIS) included dimensions of conflict and ambivalence in parent-child relationships that were added to the solidarity framework (Lowenstein, Reference Lowenstein2007; Lüscher & Pillemer, Reference Lüscher and Pillemer1998). The results of that study also yielded family relationship types such as ambivalent and distant.
Four conclusions can be drawn from the comparative analyses. First, regarding personal resources, baseline comparisons revealed differences in the resources older parents and adult children were able to draw on when being associated with each family type. For example, higher household income of older parents or older parents with a higher level of education were related to the descending and supportive-at-distance types, whereas higher age of parents and health problems were more associated with the ascending type.
Personal resources of either parents and/or children that relate to basic needs such as income, education, having a partner, and/or having a paid job, were found to be important for patterns of generational solidarity (e.g., Pinquart & Sorensen, Reference Pinquart and Sorensen2000). In the OASIS study, for example, the importance of these variables to elders’ quality of life was emphasized (Katz & Lowenstein, Reference Katz and Lowenstein2003). Social support and reciprocal exchange between generations were also found to be related to well-being (Lowenstein et al., Reference Lowenstein, Katz and Gur-Yaish2007). The comparison between the various data sets clearly attests to the importance of including a more comprehensive array of personal resources and transfers (i.e., in-kind, financial, and time), and family relationship dimensions (i.e., solidarity, conflict, and ambivalence).
The second conclusion relates to cultural differences. Our multivariate findings suggest that cultural differences in the patterns of intergenerational support are reflected in similarities and dissimilarities between the various European countries and Israel. The degree of representation varied across countries. The descending family type was most strongly represented in Belgium, and least strongly represented in France. The ascending family type was most strongly represented in Spain, and least strongly represented in Denmark. The supportive-at-distance family type was most strongly represented in Denmark, and least strongly represented in Spain. The autonomous family type was most strongly represented in France and Switzerland, and least strongly represented in Italy and Spain. Parents aged 50 and older in Sweden, Denmark, Belgium, and Israel were more likely to be part of the descending familialism type, while those in Germany were less likely to be part of this late-family type. European parents aged 50 and older in the Netherlands and Spain were more likely to be part of an ascending familialism type, while the likelihood was smaller for their counterparts in Sweden, Denmark, Belgium, Germany, France, Switzerland, and Israel. The likelihood of families being of the supportive-at-distance type was greater in Sweden, Denmark, and the Netherlands, but smaller in Belgium, France, Switzerland, Italy, Spain, Greece, and Israel.
Several attempts at comparing welfare regimes and family culture have been made, such as Esping-Andersen’s welfare regimes typology (1990; 1997) and Reher’s family cultures (1998). For example, in the present study, we observed variations between the ascending and descending family types. The ascending family was most strongly represented in Spain – a Mediterranean welfare state model – and least strongly represented in Denmark, a social-democratic welfare regime. In the OASIS study, the variation found in the strength of normative solidarity (one of the six solidarity dimensions) was congruent with Reher’s north-south division of European family types, but the division was not replicated in the other solidarity dimensions (Lowenstein & Daatland, Reference Lowenstein and Daatland2006). When comparing OASIS and LSOG, Silverstein et al. (Reference Silverstein, Gans, Lowenstein, Giarrusso and Bengtson2010) found “that once social involvement with children was considered, cross-national differences in the prevalence in each of the types roughly followed expected patterns, particularly at the extremes of family culture and state functions” (p. 20). Also, Attias-Donfut, Ogg, and Wolff (Reference Attias-Donfut, Ogg and Wolff2005) found “some evidence of the expected north-south European gradient” (p. 171). However, Glaser, Tomassini, and Grundy (Reference Glaser, Tomassini and Grundy2004) as well as Dykstra and Fokkema (Reference Dykstra and Fokkema2011) indicated that the “distribution of family types across countries clearly does not fit the north-south divide that has commonly been suggested” (p. 562).
Since there were differences between the components upon which the typology of Dykstra and Fokkema (Reference Dykstra and Fokkema2011) and that of OASIS rests, in a way their typology reflects a more macro-perspective, whereas the OASIS typology focuses more on the meso-level of the family. Thus, each typology has its strengths and shortcomings. Comparing the typologies, however, provides a more comprehensive picture of intergenerational family relationships in different societies.
The third conclusion relates to within-countries comparisons. Similar to the first wave of SHARE, no particular European country or Israel can be described by a single dominant family relationship type, albeit the degree of representation varied across countries. As Dykstra and Fokkema (Reference Dykstra and Fokkema2011) concluded from their study, each type of family relationship “is prevalent in each country, suggesting that scholars need to move beyond the idea that a particular country can be characterized by a single dominant type of late-life family” (p. 562). Building typologies is thus a recommended method of comparative research that contributes to understanding both similarities and differences between and within countries.
Finally, the findings of the above-mentioned studies show that, even in the modern era, generational solidarity is still strong in most European countries and Israel. Around half of the respondents were clustered in the descending and ascending relationship types. These types are characterized by adult children living in close proximity to their older parents, having frequent contact with them, believing in the norms of filial obligations, and providing and/or receiving in-kind support. Only a third of respondents (except for the southern countries of Greece, Italy, and Spain) were clustered in the autonomous type characterized by geographical distance between parents and children, and few support exchanges.
What are the implications of these results for family and old-age policy? They offer a prism through which to consider how society might change over the coming decades, and how society might be better prepared for those changes. It will help policy makers to identify upcoming issues and stimulate the debate on long-term family policy for enhancing older parents–adult children relationships, and altering the balance between state and family support in post-modern societies. Understanding variations in policy environment across various European countries and Israel might present a wider picture of the interplay between factors at the societal, familial, and individual levels (Wolf & Ballal, Reference Wolf and Ballal2006), potentially informing family policy development.
Intergenerational exchange is an important social issue because families in modern societies are still the main source of care and support for older people. Increased longevity, and the inability or unwillingness of societies to continue to meet the needs of elderly cohorts, place the family in a central caregiving role.
Changing norms and behaviors pose significant challenges for societies, families, and individuals. One of those challenges is to maintain intergenerational support that is based on the notion that each generation invests in the human capital of the next, and is taken care of at the end of life by the generations in which it has invested (Biggs & Lowenstein, Reference Biggs and Lowenstein2011). Katz et al. (Reference Katz, Lowenstein, Phillips, Daatland, Bengtson, Acock, Allen, Dilworth-Anderson and Klein2005) suggested that the typical site for social gerontological discourse on generations in later life is “relations between older parents and their adult children when the parents become frail” (p. 394).
While the family continues to bear major responsibility for elder care, patterns of generational solidarity and support become more complex and have to be renegotiated. It is through the lens of the family that multifaceted developments can be explored – and perhaps anticipated – in housing, health, work, welfare, leisure, and the economy.
