Introduction

Despite historical, cultural, and socio-economic variation between European countries (e.g., Coleman 2002; Reher 1998), demographic trends in all European countries involve a decrease in fertility and mortality rates, and an increase in divorce and stepfamily formation. Family ties in Europe are affected by demographic trends associated with parenting and partnering (Billari 2005). Life courses and living arrangements have become less standardized and the structure of families has changed. In this article, we examine to what extent contemporary family structure and composition resulting from demographic changes that have occurred in the past decades, affect emotional exchange between children and parents.

A first contribution we hope to make in our study is that we will focus on emotional exchange between children and parents rather than on other dimensions of intergenerational solidarity most of which have extensively been studied (Bengtson and Roberts 1991). Some studies investigated the emotional closeness or the “affectional dimension” of intergenerational solidarity (Bengtson and Martin 2001; Lye 1996; Suitor et al. 1996; Wilson et al. 2004), but a focus on emotional exchange running from parents to children and the other way around is still lacking in the literature.

Secondly, we will specifically focus on Eastern Europe. Living standards and working conditions are higher in Western than in Eastern European countries, which might negatively affect family life in Eastern Europe (Immerfall and Therborn 2010). The general level of well-being (e.g., self-rated health and life satisfaction) is lower in Eastern Europe than in Western Europe (Böhnke and Kohler 2010; Kogan et al. 2008). A variety of factors, from economic affluence and the level of democracy to living arrangements, have been found to be associated to the low level of well-being in Eastern Europe (De Jong-Gierveld 2009; De Jong-Gierveld et al. 2009). To our knowledge, it has not yet been investigated how family ties are affected by these specific characteristics of Eastern European countries.

Our research question, then, reads: How do demographic characteristics related to family structure and disruption affect emotional exchange between parents and children in Eastern Europe?

Theory and hypotheses

We use the “conservation of resources theory” (Hobfoll 1989, 2001) to shed light on how family ties may affect intergenerational emotional exchange. The theory states that people aim to obtain and retain material and non-material resources throughout their lives. With these resources they can fulfill their physical and psychological needs, and family members are likely to play an important role in the fulfillment of these needs. Three main categories of resources have been distinguished: material, social, and personal (Diener and Fujita 1995). Social resources are expected to provide people with affection, social support, and social integration. The “conservation of resources theory” assumes that people strive to maintain resources of affection, and experience the loss of those resources as a “constraint.” We would like to extend the theory by assuming that people, out of a need to regain resources of affection, may be inclined to compensate for lost or absent family ties by intensifying emotional exchange within existing family ties. Another extension of the theory is our assumption that not only the loss of resources of affection can act as a constraint but also the greater number of family members in need of care. We will include these additional assumptions in our hypotheses.

Family members may engage in emotional exchange in the form of care, advice, and aid, but can also experience stress caused by the felt obligation to comply with certain expectations to provide support (Ross et al. 1990). Often exchange with family members will be a mixture of positive and negative aspects. Here, we are mainly interested in the question whether the new family structures affect the extent of the emotional exchange between the generations, making no distinction between positively and negatively experienced exchange.

Decreasing fertility and mortality rates and emotional exchange

Decreasing fertility and mortality rates may influence the potential number of family members who can engage in emotional exchange. First, decreasing fertility rates horizontally affect the potential number of exchange candidates, because people have fewer children than in the past. Second, decreasing mortality rates vertically affect the potential number of exchange candidates. The increase in life expectancy leads to an increase in the number of generations living at the same time. Members of all generations will share more years of life together than ever before in history (Bengtson and Martin 2001). Together, decreasing fertility and mortality rates caused a shift from a pyramid-formed population structure, with few old persons at the top and many children at the base, toward a more vertical structure, with many older people still living but with fewer children following them, the so-called beanpole structure (Bengtson 2001).

The decrease in the number of siblings may imply a decrease in the number of potential emotional exchange candidates (Van Gaalen and Dykstra 2006). However, despite the fact that parents with fewer children have to rely on a lower number of potential emotional exchange candidates, children of these parents have been found to be more likely to visit their parents and provide instrumental support (Wolf et al. 1997; Spitze and Logan 1991). On the basis of our extension of the “conservation of resources” theory we expect that children born in smaller families will compensate for this by actively engaging in emotional exchange with their parents. Therefore, our decreased supply compensation hypothesis reads: people with fewer siblings are more likely to engage in emotional exchange with their parents.

The mere aging of populations is often supposed to produce a new condition for middle-aged cohorts, especially women living in three- or even four-generational families, the so-called sandwich generation (Agree et al. 2003; Brody 1981; Miller 1981; Grundy and Henretta 2006). The increase in the number of generations living at the same time implies an increase in the number of potential emotional exchange candidates. Our second extension of the “conservation of resources” theory leads to our expectation that the greater number of living members of the elderly generation in need of care, combined with the presence of young children, will pose constraints on the availability for emotional exchange by the generation in-between. Our sandwich generation hypothesis reads: people living in a family with more generations are less likely to engage in emotional exchange with (a) their parents and (b) their children.

Partnership dissolution, stepfamily formation and emotional exchange

Nowadays people are more often prepared to end an unsatisfactory relationship and live on their own again. A growing number of people do no longer spend their life with one partner, but instead have a series of subsequent stable relationships. As a consequence of repartnering there is also an increase in the merging of families, the so-called patchwork families (Latten 2004). In the case of joint custody, children will grow up in two different families. When new unions are formed between parents and their new partners, children will have a stepmother and/or stepfather who also can play a key role in their upbringing.

Partnership dissolution and stepfamily formation may weaken ties between children and parents, as a consequence of “constraints” on resources of affection such as diminished contact between children and parents, or a lack of clear guidelines and divided loyalties (Giddens 1992; Beck 1997; Beck and Beck-Gernsheim 1995; Silverstein and Bengtson 1997; Kaufman and Uhlenberg 1998; Shapiro 2003; Tomassini et al. 2004; Daatland 2007). Single-parent families are found to show more internalizing and externalizing problem behavior (Coleman and Ganong 1990) and experience more stress due to attachment disruption and periods of diminished parenting (Brody et al. 1988). Remarriage and stepfamily formation following partnership dissolution are found to negatively affect parent–child ties (Lutz 1983; Coleman and Ganong 1990). Our weakening ties hypothesis reads: (a) ever divorced people and people with a stepfamily are less likely to engage in emotional exchange with their children, and (b) people who experienced a parental divorce are less likely to engage in emotional exchange with their parents.

Our first extension of the “conservation of resources” theory leads to an alternative to the weakening ties hypothesis which stresses the possibility of strengthening family ties occurring within the new family structures. Although most studies report negative outcomes from parental divorce for parent–child relationships (Amato and Booth 1991; Kaufman and Uhlenberg 1998), some studies found that emotional bonds between mothers and children get stronger following a divorce (Arditti 1999; Arditti and Madden-Derdich 1995; Cooney et al. 1986; Guttman 1993). The strengthening ties hypothesis reads: (a) ever divorced people and people with a stepfamily are more likely to engage in emotional exchange with their children; (b) people who experienced a parental divorce are more likely to engage in emotional exchange with their parents; (c) because existing research shows that divorce often has more negative outcomes for ties with the father than for ties with the mother (Riggio 2004; Seltzer 1991), we expect the strengthening effect of parental divorce to apply more strongly to mothers than to fathers.

Finally, we focus on the effects of people’s own divorce on emotional exchange with their parents. From the point of view of the “conservation of resources” theory divorced people and people with stepchildren face a “constraint,” due to the loss of former resources of affection; as a consequence they may be more concerned with their own feelings and family situation in comparison with never divorced people and people with a traditional family life. Therefore, these people will less often engage in emotional exchange with their parents. We thus formulated the emotional preoccupation hypothesis: people who have to cope with difficulties regarding their own household are less likely to engage in emotional exchange with their parents.

Value orientations and structural characteristics

Cultural and structural changes, such as secularization, educational expansion, and an increase in female labor market participation, have been running parallel to trends in partnering, childbearing and parenting, and even have influenced them (Neels 2006). The process of individualization has diminished the control of people’s beliefs and behavior by family, church, and community and the process of secularization has affected people’s choices in partnering and parenting (Dykstra 2004; Lesthaeghe and Van de Kaa 1986). For example, religious people have more children than non-religious people (Frejka and Westoff 2008) and are less likely to divorce (Clydesdale 1997). Religion also provides moral values that encourage family solidarity (D’Antonio et al. 1982; Dobbelaere 1999; Thornton 1989). Education is associated with, for instance, the frequency of intergenerational contact and fertility rates (Jensen 1995; Brewster and Rindfuss 2000; Rindfuss et al. 1996; Heck et al. 1997; Kalmijn 2006; Hank 2007; Di Giulia et al. 1999).

In addition to people’s age and gender, we will include family attitudes, religiosity, educational level, and labor market participation as control variables since these characteristics may affect both demographic characteristics associated with family structure and intergenerational emotional exchange.

Data and measurement

To test our hypotheses we use the Generations and Gender Surveys (GGS), which are part of the Generations and Gender Program (GGP). The primary aim of the GGP is to improve the knowledge-base for policy-making in UNECE countries. The GGS is a panel survey of an 18–79-year-old resident population which is held in a number of European countries and is designed for a face-to-face interview. It aims to survey nationally representative samples of the population. The GGS contains information about the most important societal aspects of demographic choices in contemporary developed societies, focusing on the processes of childbearing, partnership dynamics, home-leaving, and retirement. In this article, we use the GGS data for the Eastern European countries Bulgaria (2004), Georgia (2006), Romania (2005), and Russia (2004).

Emotional exchanges between children and parents are measured by asking respondents whether or not there was any exchange about anyone’s personal experiences and feelings over the last 12 months. If there was any exchange, respondents were asked with whom they did talk: their partner, parents, parents-in-law, children, stepchildren, grandparents, grandchildren, siblings, other relatives, or non-relatives. Respondents could mention at most five persons with whom he or she exchanged any personal experiences and feelings. With this information we constructed two dependent dichotomous variables: emotional exchange with at least one child (stepchildren not included) and emotional exchange with at least one parent. Although we cannot make a distinction between positively and negatively experienced exchange, the question in the questionnaire was labeled “emotional support.” It is important to note that we can distinguish between emotional exchange about the other person’s feelings from emotional exchange about the respondent’s own feelings. Respondents were also asked whether or not anyone talked to them about their personal experiences and feelings. Although we did not use this question, it becomes more plausible that respondents distinguished between emotional support given and emotional support received.

The number of surviving parents, siblings, children, and the number of generations within families indicates the number of potential emotional exchangers. Regarding the number of surviving parents we distinguish between: 1. both biological parents are alive, 2. only the biological father is alive, 3. only the biological mother is alive, and 4. neither is alive. The number of surviving children (biological, adoptive, and foster) is determined by adding the number of children within the household with the number of non-resident children. The number of generations above and below the generation of the respondent indicates the number of generations alive.

Divorce and stepfamily formation are measured with three variables: own divorce, parental divorce, and having stepchildren. For each past partnership the respondent was asked whether or not this partnership ended by divorce. Together with information about the timing of the divorce, we constructed a variable measuring whether or not the respondent ever experienced a divorce, independent of their current marital status: 1. never divorced, 2. divorced 1–5 years ago, 3. divorced 6–15 years ago, and 4. divorced more than 15 years ago. “Parental divorce” is a dichotomous variable in which respondents with divorced parents form the reference category. Stepfamily formation is measured by distinguishing whether or not the respondent has stepchildren.

To exclude the possibility of a spurious relationship between family structure and emotional exchange patterns, we control for people’s value orientations which can precede both factors. As measures for family values we included people’s opinions about caring for elderly and helping children; two scales were constructed on the basis of several statements on which respondents agreed or disagreed on a five-point scale (strongly disagree–strongly agree). Because information about people’s church membership and church attendance was lacking, we measured people’s religiosity by the importance they attach to religious ceremonies. A five-point scale was constructed on the basis of three statements regarding religious ceremonies (see Table 5 in appendix for more specific information).

In our analysis we control for people’s structural characteristics. As measures of educational level we included country-specific scores to the International Standard Classification of Education (ISCED). The scale “educational level” ranges from ISCED level 0 (pre-primary education) to ISCED level 6 (second stage of tertiary education). On the basis of information on the respondents’ present work and daily activities we constructed a dummy-variable “economic activity” with six categories: 1. fulltime employed, 2. part-time employed, 3. unemployed, 4. retired, 5. homemaker, and 6. other. Students and people who are on maternity leave, among others, are included in the category “other.”

In our analyses we also control for the respondent’s age (18–79) and gender. The latter is a dichotomous variable where women receive score 1. Because emotional exchange between children and parents is more likely when they live in the same household, we control for co-residence. In the analysis for emotional exchange with children we control for the presence of resident children (0–1) and in the analysis for emotional exchange with parents we control for co-residence with parents (0–1).

For the variable educational level, we created a binary indicator variable that is coded as 1 when the value is missing and 0 when the value is present. Subsequently, the people with missing values are assigned the mean for the specific variable. For the other variables with less than 2.5% missing values we applied a listwise deletion.

Table 1 presents the descriptives of our variables per country. It can be seen that a substantial percentage of the respondents engage in emotional exchange. Russians most often have emotional exchanges with their children (20.23%) and Bulgarians most often have emotional exchanges with their parents (14.26%). People from Romania least often engage in emotional exchange with both their parents and children. Table 1 brings us to conclude that there is a substantial level of emotional exchange between parents and children, running in both directions, in these Eastern European countries.

Table 1 Descriptives by country

Results

We perform two logistic regression analyses, the first one focusing on emotional exchanges by parents with their children, while the second one focuses on emotional exchanges by adult children with their parents. In our first analysis, we include people with at least one child, either biological children or adoptive children (N = 34.485). In our second analysis, we include people with at least one surviving parent (N = 27.232). To take into account people’s common descent we included dummy-variables for the four countries.

Decreasing fertility and mortality rates

In Table 2, the results of two logistic regression analyses on emotional exchange between children and parents are presented. In model 1, we investigate the direct relationship between demographic characteristics and emotional exchanges between children and parents. In models 2 and 3, we, respectively, add the value orientations and structural characteristics as control variables to our analysis. In all models we control for people’s age and gender, which are strongly intertwined with the other variables in the analyses. We see that older people are more likely to engage in emotional exchange with their children but this effect appears to be curvilinear. This suggests that after a certain age, people become less inclined to talk about the personal experiences and feelings of their adult children. Younger people are more likely to have emotional exchanges with their parents. Women are more likely than men to engage in emotional exchanges with both their parents and their children. To control for co-residence we take into account whether or not parents have resident childrenFootnote 1 (first analysis) and whether or not adult children and their parent(s) live in the same household (second analysis). Parents with co-resident children are more likely to have emotional exchanges with their children, and adult children who live together with their parents are more likely to have emotional exchanges with their parents. To control for the number of potential recipients, we include in the first analysis the number of children, and in the second analysis the number of surviving parents.

Table 2 Logistic regression analysis on emotional exchange with parents and children by demographic characteristics, value orientations, and structural characteristics

Models 1a and 1b show that demographic characteristics explain around 16 and 7% of the emotional exchange with, respectively, children and parents. Although these percentages are substantial, apparently there are other predictors of emotional exchange between children and parents. Demographic characteristics are more important in explaining emotional exchange initiated by parents with their children than exchange going the other way around.

In model 1a, one can see that there is no significant influence of the number of surviving parents on emotional exchange with children. This finding does not support the assumption that people with both parents and children, the so-called sandwich generation, have to divide their attention between two generations. People in families with more generationsFootnote 2 are more likely to have emotional exchanges with their children. Apparently, having a multi-generational family positively affects emotional exchange. In the case of emotional exchange from adult children to parents, model 1b provides some support for the decreased supply compensation hypothesis. People with two or more siblings are less likely to have emotional exchanges with their parents than are people without siblings. This supports our expectation that people with more siblings can divide parental care among their brothers and sisters. The sandwich generation hypothesis is not supported by our findings in model 1b. People with children are equally likely to have emotional exchanges with their parents as people without children. Also, the number of generations does not affect emotional exchange from adult children to their parents.

Partnership dissolution and stepfamily formation

In model 1a, we see that ever divorced people are as likely to engage in emotional exchanges with their children as are people who never divorced. This finding does not support the strengthening ties hypothesis or does it support the weakening ties hypothesis. Apparently, parents are just as likely to have emotional exchanges with their children after their divorce as before. Furthermore, model 1a demonstrates that people with stepchildren are less likely to have emotional exchanges with their own children compared with people without stepchildren. This finding supports the weakening ties hypothesis: stepfamily formation weakens ties between parents and their own children. However, when we control for people’s value orientations and structural characteristics in models 2a and 3a, the effect is no longer significant at a significance level of 1%.

Model 1b provides support for the strengthening ties hypothesis and contradicts the weakening ties hypothesis. We find that people who experienced a parental divorce are more likely to engage in emotional exchange with their parents. We also assumed that the positive effect of parental divorce on emotional exchange would apply more strongly to mothers than to fathers. In the additional analysis in Table 3, we calculated different models for emotional exchange from adult children to mothers and fathers. It turns out that parental divorce positively affects emotional exchange from children to mothers, whereas it negatively affects emotional exchange from children to fathers. When we control for co-residence these effects are smaller but still present. One possible explanation may be that children have closer emotional connections with the single parent (in most cases the mother) they grew up with, even if they already have left the parental home. Another explanation might be that mothers are generally more emotionally involved with their children, a phenomenon which might persist also after a divorce has occurred.

Table 3 The influence of parental divorce on emotional exchange transfers from adult children to parents

Furthermore, the results presented in model 1b from Table 2 contradict the emotional preoccupation hypothesis which assumed that people who have to cope with difficulties regarding their own household are less likely to have emotional exchanges with their parents. A recent divorce positively affects emotional exchange from adult children to parents. Apparently, having difficulties in one’s own household reinforces rather than diminishes emotional exchanges between adult children and their parents. However, stepfamily formation does not affect emotional exchanges from adult children to parents. People who have stepchildren are as likely to have emotional exchanges with their parents as are people without stepchildren.

Value orientations and structural characteristics

In models 2a and 2b, we add people’s value orientations and attitudes. People who are more positive about helping children are more likely to engage in emotional exchange with their children. Against our expectation, the importance people attach to religious ceremony does not affect emotional exchanges from parents to their children. In the case of emotional exchanges from adult children to parents, the findings are reversed. People who are more positive about caring for elderly are as likely to engage in emotional exchange with their parents as are people who are more negative. Instead, people who attach more importance to religious ceremony are more likely to talk to their parents about personal experiences and feelings. In both logistic regression analyses, the Nagelkerke R 2 in models 1 and 2 is almost identical. This implies that adding value orientations to our model does not significantly contribute to the explanation of emotional exchange between children and parents.

Models 3a and 3b show that people who work part-time are as likely to have emotional exchanges with both their children and parents as are people working fulltime. Moreover, homemakers and retired people are less likely to have emotional exchanges with their children, compared with people who work fulltime. Apparently, the availability of time is not so important for engaging in emotional exchanges. Models 3a and 3b further show that more highly educated people are more likely to have emotional exchanges with their children and parents. This does not correspond with findings from earlier studies that more highly educated people are less family-oriented (Kalmijn 2006). The negative consequences of the number of siblings on emotional exchanges with the parents diminish after controlling for economic activity and educational level but the effects remain highly significant. We can conclude that our control variables cannot be held responsible for the relationship between family structure and emotional exchange.

Emotional exchange across the four countries

Table 4 shows for each country the impact of family structure and composition on emotional exchange between children and parents, controlled for the characteristics in Table 2, model 3. Emotional exchange does not substantially differ across the four countries. However, there are some differences. We see for example that stepfamily formation does only affect emotional exchange from parents to children in Romania. The number of siblings does affect emotional exchange from children to parents in Georgia and Russia, but not in Bulgaria and Romania. Experiencing a parental divorce does not affect emotional exchanges from adult children to parents in Georgia and Romania. Despite these differences, the impact of demographic characteristics on emotional exchanges is quite similar across countries.

Table 4 Logistic regression analysis on emotional exchange transfers between parents and children by family structure—per country

Conclusion and discussion

A lack of emotional exchange between family members has been shown to be related to loneliness, depression, and bad health conditions in old age (De Jong-Gierveld and Dykstra 2008; Dykstra and De Jong-Gierveld 1999; Prince et al. 1997; Sorkin et al. 2002; Van Tilburg et al. 1998). One potential consequence of demographic changes in the structure and composition of the family (e.g., smaller families, increased divorce rates) is a decrease in intergenerational emotional exchange, which might in turn lead to more loneliness and depression among elderly parents.

The aim of our study was to investigate to what extent demographic characteristics relating to fertility, mortality, divorce, and stepfamily formation affect emotional exchanges between parents and children in Eastern Europe. Our hypotheses were based on the “conservation of resources theory” (Hobfoll 1989, 2001), which assumes that people strive to maintain resources of affection, and experience the loss or absence of these resources as a “constraint.” We extended the theory, first, by assuming that people may be inclined to compensate for lost or absent family ties by intensifying emotional exchange within existing family ties; a second extension of the theory is our assumption that not only the loss of resources of affection can act as a constraint but also the greater number of family members in need of care; in particular the “sandwich generation” might be burdened by this. By studying emotional exchange up and down family lineages, we were able to test this idea.

We found that people with fewer siblings are more likely to engage in emotional exchange with their parents. Apparently, people without siblings or with few siblings compensate for the low level of emotional exchange supply that is available to their parents. This finding supports our first extension of the conservation of resources theory, which introduced the notion that people compensate for the lower supply of emotional exchange candidates by intensifying exchange within existing ties. People in families with more generations are more, instead of less likely to engage in emotional exchange with their children. Apparently, the so-called sandwich generation does not economize on emotional exchange to their parents and children. This result does not support our second theoretical extension, which assumed that the greater number of family members would act as a constraint on emotional exchanges. Both findings indicate that despite recent demographic changes in fertility and mortality family members still actively engage in emotional exchanges with both their parents and their children.

Most existing research evidence indicates that divorce and stepfamily formation have negative outcomes for ties between children and parents. Our research findings offer some support for the conservation of resources theory, in the sense that the absence of the original resources of affection can be considered a constraint. People who experienced a parental divorce were less likely to engage in emotional exchange with their father but more with their mother. This finding suggests that parental divorce might indeed negatively affect emotional exchange between children and parents, especially when the parent is their father. However, we did not find strong support for the idea that stepfamily formation negatively affects emotional exchange between parents and their own children.

We did not find support for the idea that people who recently faced a divorce are less likely to engage in emotional exchange with their parents due to a potentially increased preoccupation with their own household. On the contrary, these people are more likely to have emotional exchanges with their parents, a result that is in line with earlier studies (e.g., Johnson 1988). It also confirms our first extension of the conservation of resources theory, which stipulated that people are inclined to compensate for lost or absent family ties. The intensified emotional exchange with parents by people who faced a divorce suggests once more that the demographic trend of increased partnership dissolution, like the trends of lower fertility and increased longevity, do not necessarily lead to less emotional exchange within families.

Our findings can be interpreted in the light of two contrasting perspectives on how demographic trends may have affected the family: the “family decline” and the “family resilience” perspective (Amato 2005). Defendants of the first perspective argue that as a consequence of the retreat from marriage and the spread of single-parent families’ people have become increasingly individualistic and preoccupied with their own personal happiness, with less emotional exchange as a consequence. Advocates of the family resilience perspective do not consider the rise in marital instability a reason for alarm. In the past, many unhappy marriages remained intact because of formal and informal barriers against divorce. “Modern” family structures do not present children with more harmful conditions than the more traditional family structures, and do not necessarily prevent them from engaging in emotional exchanges with their parents. Overall, our results are more in favor of the family resilience perspective than of the family decline view.

The reason for our focus on Eastern Europe was that the general level of well-being, and the living and working conditions in this part of Europe are lower than in Western European countries, potentially affecting family life as well. However, considerable levels of emotional exchange between adult children and their parents are still found in these countries. Depending on the country, 10–20% of the people are engaged in emotional exchange with at least one child in the last 12 months. For emotional exchange with parents this is 6–14%. Moreover, our findings demonstrate that parents and children are inclined to engage in emotional exchange when they face difficulties, such as a divorce. However, it should be noted that emotional exchange as such is not necessarily positive in every single situation. For example, when children have emotional exchanges with their divorced parent, this might be rewarding for the parent but present a burden for the children. Our findings merely indicate that parents and children have emotional exchanges in difficult times, and do not reflect on the content of these exchanges.

A few limitations of our study should be mentioned. First, we did not dispose of personal information about the children who engaged in emotional exchange with their parents. The Gender and Generations Survey only contains information about whether or not the respondent had emotional exchanges with a child in the last 12 months, not about which child was involved in the exchange. Therefore, we could not link personal information from the respondents to personal information from their children. A second limitation is that we had no information about how often emotional exchanges occurred between parents and children. We only know whether or not the respondent had engaged in emotional exchanges with at least one child or parent in the last 12 months. It is possible that we would have found different outcomes, had we studied the actual frequency of emotional exchanges. For example, divorce may affect the frequency of emotional exchanges rather than the likelihood of those exchanges.

Although we did not perform a longitudinal analysis, our data permit some tentative generalizations concerning future consequences of demographic changes for family life. As far as the consequences of divorce are concerned, our results in general do not give rise to severe concerns about future emotional exchanges between generations, although adult children tend to have fewer emotional exchanges with their father than their mother after a parental divorce. Finally, our results lead us to expect that lower fertility and increased longevity will only have a limited effect on emotional exchanges between parents and children.