|Home | About | Journals | Submit | Contact Us | Français|
Frequency of contacts with the family is an indicator of the strength of intergenerational exchange and potential support for older people. Although the availability of children clearly represents a constraint on potential family support, the extent of interaction with and support received from children depends on factors other than demographic availability alone. This study examined the effects of socio-economic and demographic variables on weekly contacts with children in Great Britain, Italy, Finland and The Netherlands using representative survey data which included information on availability of children and extent of contact. Our results confirm the higher level of parent adult-child contact in Italy than in northern European countries, but levels of contact in all the countries considered were high. Multivariate analysis showed that in most countries characteristics such as divorce were associated with a reduced probability of contact between fathers and children; in Finland this also influenced contact between mothers and children. Analyses are also included of possible future scenarios of contact with children that combine the observed effects of the explanatory variables with hypothetical changes in population distribution.
The European population has an age structure older than that of any other world region and within the next few decades is set to age further. By 2020 close to one quarter of the population in several European countries will be aged 65 years or over, and those aged 85 or over will at least as numerous as those aged 75 or over in the 1950s or 1960s. This demographic shift coupled with changes in family-related behaviour such as divorce and non-marital childbearing has raised concerns about the availability of family support for older people. Research on indicators of potential family support, such as intergenerational coresidence, proximity, frequency of contacts, and exchange of help and money, has therefore become an important topic in sociological and demographic research as well as in the broader socio-political debate.
Dramatic increases in the proportion of older people living alone and decreases in the extent of coresidence multi-generational households have been documented in many European countries (Grundy 1996; Keilman 1987; Pampel 1983; Sundström 1994). Sundström (1994), for example, reported that in Norway and Finland 44% and 55%, respectively, of older people lived with a child during the 1950s compared with 11% and 14% during the 1980s. There is considerable diversity between European countries in the living arrangements of older people, with generally much higher proportions of older people living with a child in southern and eastern than in northern Europe (Bartiaux 1991; Grundy 1996; Pampel 1983; van Solinge 1994; Tomassini et al. 2004). Tomassini et al. (2004) analysed trends in the proportions of elderly women living alone in a number of European countries between 1970 and 2000. They found that in all the countries examined the propensity to live alone increased and then stabilised, but that the rankings between countries remained largely stable. The notion of the familistic culture has been advanced to explain apparently stronger family ties in southern than in northern Europe (Banfield 1958; Reher 1998).
Frequent family contact as well as coresidence also seems to be more usual in southern than in northern Europe (Farkas and Hogan 1995; Glaser et al. 1998; Höllinger and Haller 1990; Murphy 1996; Reher 1998; Lowenstein et al. 2003). A 1992 Eurobarometer survey, for example, reported that 70% of older people in Italy had daily contacts with relatives and friends, compared with 19% of their Dutch counterparts (Walker 1993). More recently Lowenstein and colleagues (2003) involved in the OASIS project collected comparable data on intergenerational family solidarity in four European countries and in Israel and found that 90% of older Spaniards had weekly contacts with their children, compared with only 56% of their German counterparts. In this study information was collected on kin availability and on proximity, coresidence, contact, emotional closeness between family members and care provided and received. This is an important advantage as the paucity of data sources that include information both on kin availability and on family exchanges involving non-coresident adults of different generations, although it is known that there are wide variations in Europe regarding the extent of childlessness (Grundy 1996; Prioux 1993). Moreover, comparative studies are sparse, although there is now a growing interest in intergenerational exchange within the family, and a burgeoning literature based on analyses of data for single countries or regions within countries (Attias-Donfut 1995; Dewit et al. 1988; Grundy et al. 1999; Grundy and Shelton 2001; Kaufman and Uhlenberg 1998; Lawton et al. 1994).
Apart from the need to consider whether older people have children at all and, if so, how many, in analyses of intergenerational interaction there are other methodological issues which make comparisons difficult. An important issue is how to quantify contacts when parents and children live together. In several studies intergenerational relations are measured only for dyads of parents and children not living together (Lye 1996; Lowenstein et al. 2003). For international comparisons this approach could be restrictive. High coresidence (as found in southern Europe) could be a proxy for strong family ties and higher level of satisfaction in family relations (Tomassini 1998; Zunzunegui et al. 2001), although it may be argued that the higher southern European rate of coresidence is a consequence rather than a cause of the low supply of public services and benefits, especially for children and older people.
This study uses broadly comparable data from surveys undertaken in four European countries—Finland, Great Britain, Italy and The Netherlands—which include information on availability of children and on contacts with children to examine variations in the proportions of older parents with at least weekly contact with at least one child. We also analyse socio-demographic differentials in this indicator of family exchange and use the results to estimate future scenarios under varying assumptions about changes in parameters associated with contact. This analysis is based on the findings of the European project Future Elderly Living Conditions in Europe (FELICE, European Union 5th Research Programme Framework, “Quality of Life”). The main aim of this project is to produce projections for the population aged 75 years or over in selected European countries by marital status, health conditions, socio-economic status and living arrangements in order to inform policy makers concerned with planning for the elderly population of the near future. The FELICIE project involves several additional countries not included in the analyses presented here because of data constraints
We first analysed the frequency of contact between older parents and their children in the four European countries. Additionally, this study examines similarities and differences in the effects of demographic and socio-economic determinants on the probability of having weekly contact between parents and children in the countries considered. Overall differences in the level of contact could be due to differences in the distribution of these determinants of older people or to differences in the way in which these factors influence the probability of intergenerational contacts, or to other external causes. We then use these results in combination with information on projected or possible characteristics of the future elderly populations in these countries to estimate some possible future scenarios for intergenerational contacts.
We used data from recent surveys including relevant data from Finland, Great Britain, Italy and The Netherlands. The geographic range thus includes countries from northern, western and southern Europe, but unfortunately we were unable to find comparable data from any eastern European country. As noted above for our outcome variable, we chose the proportion of parents aged 65 years or over with at least weekly face-to-face contact (any contact for The Netherlands) with a child as an indicator of intergenerational exchange. We counted those coresident with a child within the group with at least weekly contact. Numerous studies have suggested a gender dimension to family contact with generally higher levels of contact between mothers and children than between fathers and children, and therefore we analysed data for mothers and fathers separately. Covariates used were age group, parental marital status, number of living children, educational level, home ownership and an indicator of health status. Age is obviously important both because older age may be associated with support needs which might prompt more frequent contact, and because the youngest of those in our age range might still have children at home (especially in southern Europe as age at leaving the parental home is high). Previous studies have shown that divorced parents, particularly divorced fathers, have fewer contacts with children than their married and widowed counterparts (Barrett and Lynch 1999; Grundy and Shelton 2001; Lye et al. 1995); therefore marital status of the parents was also included in the analysis. Number of living children is another important indicator, as clearly higher parity older people have a greater pool of children with whom they can meet. Several national studies have shown that those with a higher level of education have less frequent contacts with children, probably because higher level of education is positively associated with longer distance migration, and because at least in some countries friends rather than relatives may be more important in the social networks of the more highly educated (Clark and Wolf 1992; Grundy and Shelton 2001; Lawton et al. 1994). Housing tenure is an indicator of wealth and socio-economic status, although it must be recognised that there are wide differences between European countries in predominant tenure patterns, and thus the meaning of owning a home is not the same in all the countries we consider. Finally, parent’s health status is important, as children may respond to parental poor health by increasing contact (conversely, however, parents in poor health may be less able to visit their children).
For the descriptive analysis we used the weights provided by the data collectors to expand the results to the target populations. We used logistic regression to model variations in the proportion of parents with weekly contact with children. All the surveys used in this study except the Finnish one are nationally representative of the country considered. Details of the national surveys and variables used in our analysis are given below.
A specialist module on kin and kin contact designed by Grundy et al. (1999) was fielded in the 1999 Omnibus Survey carried out by the Office for National Statistics. This survey of 3,651 individuals (749 aged 65 years or over) collected information on numbers of relatives, including children and parents, and on contacts and exchanges of help with specified relatives. Resource constraints and the need to keep the questionnaire to a manageable length meant it was not possible to ask for information about contacts with all relatives, and the detailed questions on exchanges with children were asked only for exchanges with the eldest child. ‘Eldest’ was used as a way of identifying a specific relative to ask about as requiring the interviewers to choose one at random, which would have been preferred, but would have added considerably to fieldwork costs. The education variable in the British sample is based on the age at which the person left school/education. The highly educated are regarded here those left education at age 19 years or over. Similarly, medium education is for ages “minimum” to 18 and low education for ages “minimum” or below. “Minimum” is the age a person completes compulsory education. Information on health was not available.
The 1998 Indagine Multiscopo sulle Famiglie e Soggetti Sociali is a quinquennial social survey delivering data on family relations for people living in the community carried out by the National Institute of Statistics. The survey has a total sample of over 59,000 persons. The survey covered a wide variety of topics including questions on household structure, demographic background, housing and life course. Relevant to this study is the section on family exchanges, including: household compositions, kin (children, parents, grandchildren, grandparents, siblings and other relatives) availability, interaction with non-resident children, siblings, parents, care provision and care received. The Italian education variable includes: high (general secondary education, up to age 18 years, and higher vocational education, college education, university education), medium (general intermediate education, up to 13 years old) and low level (illiterate, primary not completed and primary education). The health variable is based on a question on the presence of disability that limited daily activities. The sample size is 6,802 parents aged 65 years or over.
Data were collected in the context of the 1992 research programme Living Arrangements and Social Networks of Older Adults (LSN; Knipscheer et al. 1995). The programme used a stratified random sample of 3,805 men and women born between 1908 and 1937 and living in the western, northeastern and southern parts of The Netherlands. The oldest individuals, particularly the oldest men, were over-represented in the sample. In 1992–1993 (n=3,107), 1995–1996 (n=2,545), 1998–1999 (n=2,076), and 2000–2001 (n=1,691) follow-ups were carried out in the context of the Longitudinal Aging Study of Amsterdam (LASA; Deeg et al. 2002; van Tilburg and Broese van Groenou 2002). LASA is an ongoing study on the physical, social, cognitive and psychological functioning of Dutch older adults. The education variable used in this study was coded as follows: low (elementary not completed and elementary education) medium (lower vocational education, general intermediate education and intermediate vocational education) and high level (general secondary education, higher vocational education, college education and university education. Regarding health, for six types of daily activities (e.g. using transport, dressing, walking), one reported with how much difficulty the activity was performed. The dichotomised total score indicates 0=no difficulty at all, 1=difficulty with at least one of the six activities.
In the analyses reported here we used the 1,691 respondents from the 2000–2001 cycle, excluding 297 respondents who were not living in private households, 62 who were younger than 65 years, and 150 who had no (living) children, leaving a sample of 1,182. LASA data do not distinguish between face-to-face and phone contacts.
The Finnish data derive from the longitudinal Good Ageing in the Area of Lahti (GOAL) survey. Participants were recruited in the Lahti region, an area about 100 km north of Helsinki. The main demographic, social and economic indicators for this region are close to the national average, so the data can be treated as fairly representative of the country. The population-based sample included men and women born in 1926–1930, 1936–1940 and 1946–1950. Altogether 2,815 men and women participated in the baseline survey in 2002. The survey included questions on sociodemographic characteristics, health, health-related behaviours, and a section of questions on the need and availability of care and support. The level of education was operationalised as follows: low (primary education, i.e. 6 years of schooling), medium (intermediate or lower secondary, 7–9 years), high education (high school graduates and above). The health variable used was based on the self-rated health question. We categorised those reporting bad or very bad health as being in bad health.
For the purposes of this study participants in the two oldest cohorts, who were aged 62–66 years and 72–76 years at the time of the survey, were included. From a total of 1,908 respondents we excluded 19 who were not living in private households, 356 who had no children and 16 with missing values, leaving a sample of 1,517 for the analyses reported here.
Here we present descriptive statistics for the variables used in the multivariate analysis for this work. Table 1 summarises frequencies (weighted when needed) for the dependent and independent variables used in the models for each country. We analysed men and women separately. The sample size is presented at the bottom of the table. The distributions of older parents in the different categories used for the explanatory variables show some interesting similarities and differences between countries, due mainly to the different design of the surveys. The age distribution of the sample is similar for Great Britain and Italy but older in The Netherlands (since those included are the survivors of the original sample collected in 1992). Because of the study design, the Finnish sample has a younger age distribution.
With regard to the number of living children, Italy and Great Britain again show very similar parity distributions, while The Netherlands has a higher proportion of high-parity older people, and Finland has the highest proportion of older people with two children. These differences reflect known variations in historic fertility patterns. The Netherlands has among the highest fertility levels in Europe for cohorts born before 1930 (Tomassini et al. 2004), while the same cohorts in both Italy and Great Britain were characterised by high proportion of women with one or two children (Grundy et al. 2004). The Netherlands and Italy have a similar marital status distribution with high proportions of widowed women. The Finnish data show a high proportion of married and single/divorced older people and a low proportion of widowed people, mainly reflecting the younger age structure of the sample. Great Britain is characterised by a higher proportion of older married women then the other countries, possibly due to the smaller gender gap in further life expectancy at age 65 years there.
All the countries are characterised by a high proportion of older people who are homeowners, ranging from 53% among Dutch men to 89% among Finnish men. When education is considered, comparing the countries is difficult due to the different classifications used. Italy and Finland appear to have higher proportions of older people with low education than Great Britain or The Netherlands. In Italy this is due to the late introduction of minimum compulsory schooling until age 14 years. In Finland compulsory elementary schooling was introduced early, but opportunities for higher education expanded only after the Second World War. All the countries except Finland show marked gender differences in education, with men being twice as likely as women to have higher levels of education.
Health is also difficult to compare due to differences in the definition used in the survey and to the different age compositions of the samples. More than three-quarters of older people in Italian and Finnish samples are in good health or free from disability, compared with only 46% and 31% of Dutch men and women respectively. This is probably the effect of the older sample or the more sensitive question on health included in the survey.
When contact with children is considered, Italy and The Netherlands both have very high proportions of older people who have weekly contact with their children (over 90% for both mothers and fathers). The definition of contact between parents and children in the Dutch survey also includes contact by mail and by phone, which explains the high proportion for The Netherlands. In Great Britain about one-half of the parents have at least weekly contact, but this is not strictly comparable since the Omnibus Survey considers only contacts with the eldest child, rather than contact with any child as in the other surveys. Weekly face-to-face contact with children is rather lower in the Finnish sample with 64% and 76% of older men and women, respectively, seeing a child at least once a week.
Tables 2 and and33 present the results of logistic regression analyses of the effects of the explanatory variables, illustrated in the previous section, on the probability of having contact with a child at least once a week (including coresidence) for mothers and for fathers. As noted above, in all countries except The Netherlands the form of contact considered is face-face contact.
Results show some interesting cross-national similarities. In Great Britain, Italy and Finland men who are single, separated or divorced have much lower odds of at least weekly contact with a child than those who are married; in Finland this effect is also apparent among women. Widowhood is not significantly associated with contact with children except in The Netherlands, where widowed women have higher odds of weekly contact with children than the married. This difference may reflect the different indicator of contact available for The Netherlands.
Number of children is positively associated with mothers’ and fathers’ contacts with a child in Italy, with fathers’ contact but not that of mothers in The Netherlands, and not with either in Finland. The significant negative effect in Great Britain must be interpreted in light of the fact that we are considering only contact with the eldest child. This suggests that when parents have only a single child, he or she is the eldest child, and therefore there is a higher probability of contact with this specific child. With increasing number of children the probability of having contact with a specific child may decrease. The association between level of education and contact with a child is statistically significant for fathers in Great Britain and Finland and for mothers in Italy and The Netherlands: in all cases higher education is associated with a lower probability of having weekly contact with children. Age, tenure and health status were not significantly related to contact with children in any of the countries analysed.
One interesting way to present data from logistic models (which can also be used for projection purposes) is to calculate the probability of the outcome, while keeping constant the effect of the explanatory variables (coefficients) and changing the mean values of the explanatory variables according to previous projections scenarios or to specific hypothesis of the researchers. Different scenarios have been created for Great Britain, The Netherlands, Italy and Finland, and the five profiles deriving from them are illustrated in Fig. 1. The profiles that we created are:
Given the rather limited effects of the covariates that we chose, changing patterns in the mean values of the variables considered are not going to greatly affect the probability of having contact with children. For example, even if Italian parents in the future have on average only one child, the probability of having weekly contact with the child will change slightly from 0.94 to 0.89 for mothers and from 0.93 to 0.89 for fathers. An increase in education will have a small effect as well. For example, in Finland if we double the proportion of older people with high education, the effect on weekly contact with their children remains basically unchanged at 0.77 for mothers and at 0.64 for fathers. In Great Britain the proportion of older people with medium or higher education is rather low. Doubling both these proportions would decrease the probability of weekly contact with their eldest child by 5% for mothers (from 0.50 to 0.48) and 14% for fathers (from 0.46 to 0.39).
Weekly contact with children is sensitive particularly regarding the proportion of older people who are single, separated or divorced. At baseline the proportion for fathers is low for most countries, especially for Italy and The Netherlands. If the proportion of single or divorced men is increased by ten times in the latter countries (from 2.5% to 25%), the probability of weekly contact with children does not change for Dutch fathers but drops from 0.93 to 0.90 for Italian fathers.
If we consider the distribution by marital status obtained from previous projections by marital status (Murphy and Kalogirou 2004) results show a more noticeable change. The proportion of older men having weekly contact with their eldest child in Great Britain changes from 0.46 to 0.32—which is quite a large and potentially important change—while the effect is smaller for women and for Italian parents.
The results presented here show that in Italy parents have more frequent face-to-face contacts with their children than in Finland or Great Britain. The Netherlands also has high frequency of intergenerational contact, but here the definition also includes contact by mail and by phone. The British data used here relate only to contact with the eldest child; however, results of analyses of the 2001 English Citizenship Survey, which includes information on monthly face-to-face contacts with all children, show that when monthly contact is analysed, the proportion of older people having contact with any of their children is still smaller than in the other European countries analysed here (Grundy et al. 2004). The Dutch findings are not consistent with the stereotypical view of weaker intergenerational exchanges in northern than in southern Europe, but there are several distinctive features of The Netherlands which may be important, including the relatively dense population and small size of the country. Previous results from the LASA study show that children live an average of about one hour’s travel time from their parents, and those living within one-half an hour’s travel time have contact with their parent at least twice a month (van Tilburg and Dykstra 1995). As the family size of Dutch parents is rather large, (over one-half have three or more children), it is likely that at least one of the children lives within close proximity and maintains frequent contact with the parent. In this respect Finland may be the opposite case, with a small population in a relatively large country.
For all the countries analysed, with the exception of Finland, men and women tend to have similar frequency of contact with children. The result for Italy confirms previous studies on intergenerational exchanges in countries where strong ties are less likely to create privileged dyads between parents and children (e.g. mother-daughter; Tomassini et al. 2003). It is also possible that this finding is due to the fact that for married couples only one spouse is interviewed, and he/she attributed to the spouse the same level of interactions with children that he/she has.
When the determinants of the probability of having at least weekly contact with children were analysed, it was interesting to observe differences and similarities among the European countries involved in our study. Being divorced, separated or never married has a significant effect in lowering fathers’ contact with children in three of the countries considered. For mothers the effect of being divorced, separated or never married is significant only in Finland. This result confirms previous studies on the topic (Cooney and Uhlenberg 1990; Kaufman and Uhlenberg 1998) of the effect of marital disruption on parent-children relationships. Despite earlier findings on the Dutch 1992 baseline sample supporting the disadvantaged position of ever-divorced fathers towards their children (Dykstra 1997), these differences were not found in the 2001 dataset. This may be due to the low proportion of divorced/single fathers in the LASA sample (2.6%). It is also possible that the association between marital status and contact in countries with a high proportion of remarried older people, such as in The Netherlands, is affected by the fact that remarriage does not in fact fully repair distorted parent-child relationships of divorced fathers (Dykstra 1997). The number of children is significant for fathers’ contact in all countries except Finland, while for mothers it is significant only in Italy and Great Britain.
Since the majority of determinants that we considered show no significant effect on the probability of at least weekly contact with their children, our results suggest that contact is related to other variables that we have not been able to include in our analyses. Even if it is true that in Finland a high proportion of highly educated parents combined with the significant effect of fathers’ education helps to explain the low proportion of men having contacts with their children, this finding is marginal in the explanation of the differences between Finland and the other countries. However, it must also be remembered that some of the sample sizes available were relatively small, and therefore the power of the analyses may not be sufficient to identify relatively small effects. The limited number of common explanatory variables that we had in the surveys available is another important constraint. A further important limitation of the work reported here is that our indicator—at least weekly contact with a child—is a rather crude one. We have no information on the quality or meaning of older parents’ relationships with their children, on types of help and support exchanged or on possible responses to increased need by one or other party.
Regarding future scenarios in intergenerational exchanges our results show that the demographic and socio-economic changes that may occur in the coming decades have a limited effect on the probability of parents having weekly contact with their children, given the current propensity. Lower parity and more divorced or single parents and more education will decrease the probability of intergenerational contact but not in a remarkable way. We are unable to predict what if any changes in family solidarity may occur, but the few attitudinal data at European level show very little change in the sense of responsibility of young generations towards their parents (European Commission 1997).
The research reported here was conducted as part of the European Union project Future Elderly Living Conditions In Europe (FELICIE), European Union Fifth Research Programme Framework, “Quality of Life”). The data for Great Britain were collected by E.G. and M. Murphy with funding from the UK Economic and Social Research Council (grant reference numbers R000237776 and R000237076. Collection of the LASA data was funded largely by the Dutch Ministry of Health, Welfare and Sports. P.M. was supported by the Academy of Finland (70631, 48600).