There are differing views of the family experience of middle age, retirement and ageing. Hagestad (1987) sees the older groups in society as a rich resource both for society at large and for individual families, since family networks now have more multigenerational links, resulting in more variety and longer duration of family relationships. However, while retirement brings a decline in work and outside relationships which may lead to an increase in time spent with family members, there is also a risk of overinvolvement with grandchildren and other kin who need help (Szinovacz and Ekerdt 1995). This has the potential to affect adversely the retirement experience of Laslett's idyllic notion of the 'third age' in life - the time of personal fulfilment (Laslett 1992). This third age (or stage) of life follows the first two stages - dependency and responsibility - but precedes the fourth stage - physical decline. So while some maintain that increased age brings a gradual withdrawal from family connections (Troll, Miller and Atchley 1979), others say this time of life brings increased involvement with family (Szinovacz and Ekerdt 1995).

This paper explores the intergenerational family relations of people aged from 50 to 70 years. This is equivalent to Laslett's third stage of life and is referred to as 'later life' in this paper. The analysis examines interactions between three generations of the same family. It explores various dimensions of family cohesion - family structure, association, exchange, emotional relationships and values. Elements such as family size, availability or degree of interaction between family members, as well as socio demographic and personal characteristics, are investigated as predictors of intergenerational family support for and by people at this later life stage.

It is important to explore family relations and supports in an economic climate of redistribution of family support services. Contemporary economic strategies favour a reduction in government provision of services and the privatisation of services. These strategies may affect expectations about family and community responsibilities. Therefore, the identification of factors that contribute to actual or anticipated family support, particularly exchange between younger and older generations, will facilitate an understanding of the interplay between private and public supports for families. Who has access to what sorts of private resources, and what difference does such access make?

These changes in service provision and questions about who will provide family support lead to a number of social and public policy issues for people in the 'later life' stage. The main social issues that will be considered are the ageing of the population; the question of intergenerational obligations; prolonged dependency of adult children; gender inequality in the shouldering of family care responsibilities; the effects of divorce or relationship breakdown on the availability of family support networks; the experience and responsibilities of grandparenting; and care for the elderly. 

A number of issues raised in both the Australian and overseas literature set the broader theoretical context for analysis of, and discussion about, the dynamics and significance of family relations and exchange. These issues are the broad social policy debate; the dimensions of family cohesion; the social contract; the ageing of the population; gender inequality; and the effects of marital separation.

Social policy debate

Variations in the propensity of family members to provide inter-generational family support, for whatever reason, are important in informing policy debates about the right mix of private and public social and economic supports for individuals. For instance, public provision decisions based upon assumptions of support from adult children to elderly parents may be ill-founded where divorce has occurred (Thompson and Li 1992; Furstenberg, Hoffman and Shrestha 1995). The balance of private and public support resources is particularly relevant to people in their fifties and sixties. Not only will this group's own needs for certain public services (particularly health services) start to increase in the next two decades; they may also have responsibility for dependent or semi-dependent adult children, or for elderly parents in need of care or assistance.

In recent years, many Western societies have moved away from a public welfare style of government, and economic considerations have played a more dominant role in policy-makers' decisions. Because social and health services can be seen as a 'public burden' that drains the economy, the social objectives of care, equity and wellbeing of citizens become subordinate to 'efficiency' and cost-containment objectives. Thus, care of the very young, the ill, the disabled or the elderly is seen to be the responsibility of the community. Thus, family care is largely supported by unpaid labour, mostly provided by women, and often with little help from public services (Hooyman 1992).

Given that indvidualism is encouraged by neo-classical economic theory, then the assumption is that individuals should be responsible for their own welfare rather than relying on the state (Bessant 1996). Under these circumstances there is an expectation that if individuals fail to provide for themselves their families rather than the state should provide for them. However, this expectation is based upon the assumption that the family can and will care for all members. The extra responsibilities implied by prolonged periods of education for young people, rising unemployment rates and an ageing Australian population may mean that this expectation is increasingly difficult to satisfy.

The age group that is the focus of this paper - those in their fifties and sixties - will likely be the ones to bear this extra responsibility for both young adult children and elderly parents.

Dimensions of family cohesion

How can we tell which families are in a position to take on extra responsibilities, and which are already doing so? The ability to identify family resources and propensities to provide intergenerational support becomes critical. A number of dimensions of family cohesion have been discussed in the literature as being important to integenerational exchange within families.

According to Roberts, Richards and Bengtson (1991) the main elements of family cohesion are:

• structure - which relatives exist and where they live;
• association - communication and activities shared with relatives;
• function - transfers and exchange of money, goods and services;
• affect - feelings of affection, closeness, satisfaction with relationships;
• value systems - beliefs about family culture and obligations;
• identity/similarity - social or economic mobility among relatives.

An examination of the interactions between the various dimensions of family cohesion will also help to identify which elements are important in predicting the likelihood of intergenerational family support, whether this assistance be to older or to younger family members. An obvious example is the link between proximity and amount of contact.

For example, people may feel close to a broad range of family members or only a few specific family members, which could affect levels of exchange. Levels of involvement may also vary widely due to the diversity of family sizes and forms produced by rapid social and demographic change. Some newer family forms are the 'beanpole' family, which is a result of greater longevity and the consequent multiple generations, and which extends the duration of intergenerational family roles; the 'age-gapped' family structure, which is due to delayed child-bearing and greater age differences between generations; missing generations, which are increasingly due to voluntary childlessness; and complications brought about by divorce or separation, which give rise to the 'step-relation' family (Bengtson 1993). Such different patterns have implications not only for the availability of support in times of need but also for the degree to which value systems are shared.

Although some families will have the means to provide goods and services to their members while others will not, the propensity to either provide or draw upon family resources for support could also be related to a prevailing family culture. Some families may endorse a traditional gender-based, home-centred mode of caring for relatives, while others may favour a more modern mode of care, using both formal and informal resources outside the home (King and Chamberlayne 1996). Nevertheless, the extent to which individual families are involved in internal transfers and exchange, and the extent to which they espouse the values of mutual obligation and support, have implications for the degree to which they can cover the needs gaps left by diminished public provision. Knowledge of the nature of intergenerational support within families can therefore help in our understanding of the demands families make on public services.

The social contract

Family and societal values have consequences for the intergenerational 'social contract'. This has been defined as a set of social norms 'involving the unwritten informal expectations and obligations that create solidarity or conflict, and the negotiation of these in everyday life' (Bengtson 1993, p. 4). In the broader social context, it means that members of the older generation, having maintained and succoured the next generation, will, theoretically, expect the younger generation to support them in old age. But do people still expect the younger generations - their own or others' children - to accept this responsibility? It could be that these unwritten family obligations are becoming superseded by the ideology of individualism.

Indeed, Motenko et al. (1995, p. 461) maintain that: 'societal conflict is created by a philosophy of privatisation and individualism that espouses the belief that people should be responsible for meeting their own basic needs for housing, health services, and mental health services, with minimal help from the collective taxpayers.

Since the term 'generation' will be used in this paper in its sociological sense, meaning generations of family lineage, rather than the more historical concept of age cohorts, the intergenerational contract refers here to interaction between generations within the same family. Therefore, if minimal help from 'the collective taxpayers' is to be expected, then a person's own family members appear to be the obvious source of support for older citizens who cannot (or will not be able to) provide for themselves. Alternatively, it may be the case that isolated or 'nuclear' family households impede the operation of mutual exchange between family generations. The important issues, therefore, are to what extent younger generations are already providing support for their seniors and to what extent they are expected to do so in the future.

The ageing of the population

The financial support of the aged is increasingly perceived to be a problem for future social policy planning (Officer 1996). One reason for this is that the proportion of citizens paying taxes to support welfare services for the older generation is dwindling, while the proportion of people in the older generation increases in comparison. This results in increases in the 'dependency ratio' - the ratio of dependent aged to taxpayers.

In the past, the population distribution in Australia has resembled a pyramid, with the smallest proportion in the oldest age groups and the largest proportion in the youngest age groups. The size of projected population sections for Australia over the next 50 years indicates that the population distribution will become (and is already becoming) more square-shaped. There is a projected increase of people aged 65 years and over from 12 per cent of the population in 1995 to 22 per cent in 2041 (ABS 1996a). This means that the ratio of wage-earners and other taxpayers to elderly people will be reduced. Also, younger people available to attend to the care of the elderly will make up a smaller proportion of the population. When the current lower fertility rates (de Vaus, Wise and Soriano 1997) are added, it becomes clear that the future elderly will have fewer adult children, or no adult children, to help them in older age.

In the United States, Bengtson (1993) foresees limits being placed upon medical care or pensions for the elderly by the second or third decades of next century. The corollary is increasing political pressure for families to provide for the growing elderly population in Australia as in the United States.

Gender inequality

Practical problems may result from these pressures if insufficient human or economic resources exist within individual families to enable them to shoulder an increasing burden. Furthermore, personal care for family members (excluding financial care) has been observed to be provided by women (Finch 1989; Glezer 1991; Millward 1996; Howe and Schofield 1996); while men do provide some spousal care, they rarely care for disabled family-of-origin members or for impaired parents. Hence, the term 'family care' may be a misnomer. Furthermore, Hooyman (1992) maintains that social policy perpetuates gender inequality because social service cost-cutting relies on women's unpaid labour. Therefore, the social norms around caring roles disadvantage women, because long-term care is seen as a family - that is, female - responsibility rather than as a societal responsibility.

It should not be assumed that there is an endless supply of unpaid female labour. Women's participation in the paid workforce is increasing, as is the number of sole mothers. However, women's domestic caring responsibilities do not recede as their paid work increases, so they may experience the conflict and burden of a 'double duty' (Moen 1992). This can cause physical or mental stress, especially since the actual time available to spend on domestic and caring tasks does decrease, which may lead to an increase in domestic tensions and a reduction in the level of satisfaction women experience in their family role (Bittman and Pixley 1997).

Effects of marital separation

A further factor pertinent to the availability of private family resources is the family network disruption that may follow divorce or separation. This may affect two or more generations of the same family and can result in fewer family members being accessible (Millward 1996). Decisions about the public provision of support that are based upon the assumptions that adult children will support elderly parents may be ill founded where divorce has occurred (Furstenberg, Hoffman and Shrestha 1995).

The present Australian divorce rate raises concerns about future intergenerational relations and exchange. Australian divorce rates throughout the 1980s and 1990s have been fairly stable at between ten and twelve divorces per thousand of the married population for any given year (de Vaus 1997). Although around one-third of divorced people remarry, slightly more second marriages than first marriages also end in divorce (ABS 1996b). As a consequence, the number of children living in one-parent families has steadily increased since the late 1980s (Kilmartin 1997); de Vaus (1997) notes that the pool of Australian people with divorced parents 'is increasing by 40 to 50 thousand children each year' (p. 32).

Such changing marital patterns have implications for the balance of public and private support resources for people in later life since traditional flows of exchange and possibly also emotional ties with adult children may have been disrupted in a sizeable proportion of families. The effect of divorce among grandparents has also been found to be a factor in determining kin network availability and contact (Thompson and Li 1992), because adults whose parents divorce also report a lower quality relationship with their parents (Lye et al. 1995).

These effects suggest there may be less intergenerational involvement when the grandparent generation has been through marital breakdown. This might lessen involvement with both adult children and with grandchildren.

Indeed, Aquilino (1994, p. 908) found that: 'later life parental divorce lowered relationship quality and contact between adult children and parents. The effects were stronger for father-child than mother-child relations and stronger for father-daughter than for father-son relations. Widowhood had negative effects on father-child but not on mother-child relations . . . Divorce disrupted parent-son more than parent-daughter support exchange.'

The Later Life Families Study conducted by the Australian Institute of Family Studies was designed to examine the place of family in people's lives as they approach older age. It was a random national sample collected in late 1996 via computer-assisted telephone interviewing, in conjunction with the larger Australian Lifecourse Study. Structured telephone interviews lasting approximately 45 to 60 minutes were conducted by trained interviewers with 721 respondents. One male or female respondent, aged between 50 and 70 years, was interviewed per selected household. Their spouses or other family members were not interviewed.

In the Later Life Families Study, intergenerational relations are examined with the older parent as the reference person, rather than the adult child. This contrasts with two large US studies: Furstenberg, Hoffman and Shrestha (1995), based on the Panel Study of Income Dynamics of 1988, and Webster and Herzog (1995), based on the Americans' Changing Lives survey of 1989. The Australian study allows for the emphasis to be on the perceptions and behaviours of these older people, rather than the perceptions of younger adults, as is the case with the two US studies. Also, since the three studies cover very similar subject matter, comparisons from the point of view of two different generations are possible.

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Sample characteristics

Table 1 displays personal characteristics of the Later Life Families Study sample. The 721 respondents were aged between 50 and 70 years, with 57 per cent being women and 43 per cent being men. This is an overrepresentation of women by about 6 per cent, and was due to women being more likely than men to agree to be interviewed. Other dimensions of the sample were very close to the proportions in the general population within the age range (Mills 1997).

Seventy-two per cent were currently in registered or de facto marriages (not necessarily their first marriage), 12 per cent were divorced or separated, and 11 per cent were widowed. Most respondents were born in either Australia or the British Isles (85 per cent) and only 5 per cent spoke a language other then English at home. Half were still in the workforce and nearly half had an education level of year 10 (fourth form or Intermediate Certificate) or less, which reflects common educational standards of the 1930s to 1950s when they were likely to be at school.

Table 2 shows respondents' family and household characteristics. The majority of respondents (71 per cent) lived with a partner and most had children (92 per cent). Of those with children, just over a quarter (27 per cent) still had children living at home with them. Nearly two-thirds were grandparents (62 per cent), and around half still had at least one elderly parent living (35 per cent had their own father or mother and 29 per cent had a partner's parent alive). Nearly all (94 per cent) had at least one sibling living (76 per cent had their own brother, 78 per cent had their own sister and 78 per cent had partners with siblings)

In the investigation of which family factors are linked with intergenerational exchange, the first family cohesion factors to be examined are those of intergenerational family structure, geographic proximity and levels of association.

Where the younger and older generations live

Respondents were asked where adult children, mothers or step-mothers, fathers or step-fathers and partner's parents or step-parents lived, using the following categories: live with me; live within 30 minutes away; live between 30 minutes and 2 hours from me; live elsewhere in Australia; live overseas.

This part of the survey deals with those respondents who had adult children who had left home and were living independently. The effects of co-residence with adult children will be discussed later in this paper in the context of aid exchange.

Most respondents had children living independently (82 per cent) and, of those with independent children, over half had at least one adult child living within a 30-minute drive. Also, women were more likely to have independent children than men (86 per cent compared with 76 per cent of men), which suggests that the men in this age group tend to have younger children than the women.

Elderly parents are included in Table 3 regardless of co-residence, as this applies in only about 3 per cent of cases. Only 35 per cent of respondents had either a mother or father still living, but it was much more common to have an elderly mother (220) than father (90). Around half of those still in their fifties had a parent alive, compared with only 15 per cent of those aged 60 to 71 years. However, very few (3 per cent) had an elderly parent living with them and fewer than 20 per cent had an elderly parent living within a 30-minute drive of them. It is clear, therefore, that the majority of elderly parents are living independently, and not just around the corner. Only 12 per cent of all respondents had an elderly parent (or parent-in-law) in a nursing home or special accommodation home or hostel.

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Contact between generations

Respondents were asked how often they spoke to adult children and to elderly parents, using the following categories: every day; a few times a week; once a week; once a fortnight; monthly; less often; special occasions; not at all.

Seventeen per cent of respondents had adult children (over 18 years of age) still living with them. Living nearby correlated with more frequent contact: if relatives lived nearby they were likely to be contacted either in person, or by telephone, more often than if they lived further afield. Nearly all respondents (92 per cent) with an adult child living independently spoke to them at least once a week, and there was often more than one such child (see Table 4). Nearly two-thirds of respondents with an elderly mother spoke to her weekly, while elderly fathers and partner's parents were not contacted so often.

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Women generally had more frequent contact with both independent children and elderly parents than did men. Ninety-five per cent of women spoke to an independent child at least once a week, compared with 85 per cent of men, but differences were more striking regarding elderly parents. Women were much more likely to have frequent contact with both elderly mothers (73 per cent had weekly contact, compared with 55 per cent of men) and elderly fathers (63 per cent of women had weekly contact, compared with 36 per cent of men). The gender difference worked in the opposite direction for contact with parents-in-law, with men more likely than women to have weekly contact with in-laws (45 per cent weekly, compared with 30 per cent of women). This is consistent with a man seeing his wife's parents more often than a woman seeing her husband's parents.

There were no differences in frequency of contact with either independent children or elderly parents according to respondents' workforce participation. This is somewhat surprising, since it could be expected that those who had retired or were not in the workforce would have more time to contact relatives than would those with employment commitments, but this was not found to be the case.

Although respondents from non-English-speaking backgrounds are less likely than those from English-speaking backgrounds to have elderly parents living in Australia (Millward 1996), it might be expected that, due to cultural norms, there would be more frequent contact with close relatives who had also migrated to Australia. This was found to be the case in a limited sense: all respondents from such a background with an elderly mother living here had contact with her at least weekly, compared with three-quarters of those with English-speaking backgrounds. However, there were not enough elderly fathers or parents-in-law from non-English-speaking backgrounds living in Australia for meaningful comparisons to be made regarding these relatives, and there was no significant difference between these and other respondents in contacting at least one adult child weekly.

The marital status of respondents made some difference to contact rates. Respondents who had been divorced or separated had less frequent contact with adult children. Seventeen per cent of those divorced or separated contacted independent children less often than weekly, compared with only 6 per cent of legally married and 10 per cent of widowed people. So it was not whether people lived alone or with a spouse or partner that made the difference; rather, it was a history of marital separation.

In other studies, parental divorce or separation has also been observed to decrease levels of contact between older parents and their adult children (Roberts, Richards and Bengtson 1991; Webster and Herzog 1995). Such intergenerational dynamics have been of particular interest in cultures where there are high divorce rates, as divorces and remarriages clearly change the structure of the kinship system (Marks and McLanahan 1993).

However, despite a history of marital separation making a difference to contact with children, it did not affect contact with elderly parents. In this case, the important factor was whether or not respondents lived alone or with a partner. Those who were living with a partner were less likely to contact elderly mothers or fathers weekly than were those living alone. Perhaps this is linked to the greater availability of unpartnered people at this later life stage to socialise with elderly parents, or to cater to their more personal needs.

Discussion

A main issue that arises from the consideration of proximity and association is that of gender differences. As seen in Table 3, very few respondents had elderly parents living with them; those who did were more likely to have elderly mothers living with them (2 per cent). Respondents were more likely to have mothers than fathers living within a 30-minute drive (11 per cent) and were more likely to be in fairly frequent contact with mothers than fathers (65 per cent in contact at least weekly) (see Table 4). Female respondents also had more to do with both their parents and independent children, while the men had more to do with their parents-in-law, ostensibly through association between their wife and her parents.

This female 'kin-keeping' phenomenon has been attributed to female socialisation into family roles of organisation, nurture and mediation (Chodorow 1978), which are seen to complement male roles in public life (Bottomley 1983). Female respondents appear to be more connected with both younger and older generations of family members than male respondents, but further analysis will test the hypothesis that the strength of feminine culture within the family milieu translates to more extensive exchange between female relatives than male relatives.

The direction of family ties also seems to be important. While levels of association were generally fairly high, the ties with children appeared to be stronger than those with elderly parents. This was the case regardless of ethnic background, suggesting that downward links are stronger than upward ones. The next section, dealing with actual exchange between generations, will further test the hypothesis that the parent-adult child relationship is the most significant by examing levels of intergenerational exchange both up and down generations from the point of view of the respondents.

The effect of the marital situation of respondents on intergenerational contacts is also further explored later in this paper. The three family dynamics - proximity, contact rates and levels of exchange - have been shown to be correlated, both in recent Australian research (Millward 1996) and in overseas work (Roberts, Richards and Bengtson 1991; Finch 1989). Therefore, the hypothesis is that the marital and partnering status of people in later life will also significantly influence actual intergenerational exchange.

The next dimension of family cohesion, and the main focus of this paper, is that of intergenerational exchange. Three main types of aid exchange between respondents and both adult children and elderly parents are examined: financial assistance, practical help and emotional support or advice. Associated demographic and social factors are expected to vary and different policy issues arise depending on the direction of the intergenerational flow of aid - up to or down from respondents - and the life stage of the 'givers' and 'receivers' of support.

The first issue to be explored is the exchange of aid between respondents and their adult children. This involved interviewing a sub-sample of 632 parents who had children over the age of 18 years, whether these children were living independently or not.

This group of parents was asked: 'Do your adult children provide you with emotional support or advice?' 'Do your adult children help you financially if you need it?' 'Do your adult children provide you with any practical support (doing things for you)?'

These questions pertain to generic types of assistance; no specific tasks, amounts of help or sums of money were mentioned in the survey. This was partly because the survey was a sub-section of a much more extensive and broad-ranging study. Also, because it was a telephone survey, the length of interviews had to be tightly contained. Nevertheless, Table 5 shows that well over three-quarters of these respondents received some emotional support (81 per cent) or some sort of practical help (78 per cent) from their adult children (two-thirds (68 per cent) received both). However, general financial aid from adult children was much less common (23 per cent), so only around a fifth of the parents received all three kinds of help from adult children.

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

As will be seen later in this paper, the big difference between the nature of support given to adult children and received from adult children was that financial help generally flowed in one direction - from older parents to adult children and not vice versa. Nevertheless, there is considerable evidence of a wide-based commitment on the part of adult children towards their parents.

Effect of co-residence

Although most of these respondents (73 per cent) did not have any adult offspring still living at home, the influence of co-residence with adult children for the remaining parents (27 per cent) should be considered, in that more exchange might be expected when adult children are living with their parents (White and Rogers 1997). Table 6 shows that proportion of parents receiving three types of assistance from resident compared with non-resident adult children.

Co-residence did generally enhance financial exchange. Chi-square significance tests showed a weak to moderate difference in the receipt of financial assistance between those with and those without resident adult children. There was a higher instance of receiving money from resident adult children than from non-resident adult children. This contribution to household finances on the part of co-resident children probably takes the form of board and other expenses paid from children's wages or study assistance payments.

However, parents were just as likely to receive practical help from their adult children whether they were co-resident or not, and in fact less emotional support was received by respondents from co-resident children than from those who were living independently (see Table 6). The greater emotional support and advice given to parents by adult children living independently may well be related to their age and maturity, as they could perhaps act as a friend or adviser to parents. Because many were themselves parents, the shared understanding of the parenting role may also provide for more of an equal status with their later life parents.

Parental characteristics and receipt of aid

The characteristics and socio-demographic situation of parents may be a further influence on the amount of help forthcoming from the younger generation. To avoid the confounding effects of co-residence, Table 7 shows the proportions of parents receiving assistance from adult children according to parental characteristics, but only from adult children living independently.

Mothers were more likely than fathers to report receiving all three types of aid from their children, which accords with past research findings that mothers give to adult children and receive from them more help than fathers (Marks and McLanahan 1993). This also further illustrates the correlation between communication and exchange, as mothers also had more frequent contact with independent children than did fathers.

Not surprisingly, those in the workforce and with the highest incomes were less likely than those with lower incomes to receive financial assistance from children. In contrast, those with non-English-speaking backgrounds or low incomes were the most likely to receive financial help. Parents on lower incomes also received the most practical help, which suggests that giving by adult children to their parents is needs-based.

Parents with more than one independent adult child were also more likely than those with only one child to receive emotional and practical aid, as these parents can draw upon a bigger younger-generation 'resource base' in times of need (of course, the sixty-one respondents without children would have no such resource base).

As explained earlier, the effects of marital status and history are explored here in two dimensions. The first is a comparison of those respondents who have had a marital separation with those who have not; the second is a comparison of those currently with a partner and those without.

* sig at p=0.01; Phi = 0.10; ** sig at p=0.0001; Phi = 0.17. 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

* Significant at p<.0001 ; ** Significant at p <.01 ; # Significant at p<.05 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Regarding the first dimension, a past marital separation decreased the probability of emotional support or practical aid from non-resident adult children, which reflects a lower level of involvement and possibly more detached relationships. However, a past marital separation did not seem to affect financial assistance to the later life parents, in contrast with findings from the US Panel Study of Income Dynamics, which found that parental marital separation negatively affects economic aid from adult children (Furstenberg, Hoffman and Shrestha 1995). This contradiction may be linked to the fact that many of the parents reported on by Furstenberg, Hoffman and Shrestha were older than those in the Later Life Families Study, so may have needed more financial assistance from adult children for health care and the like.

On the second dimension, living alone rather than with a partner had no clear effect upon receipt of help. Nevertheless, if both the gender and past divorce or separation of parents are considered simultaneously, a 'pecking order' for receipt of aid from adult children emerged.

Starting with the group of parents most helped, this pecking order was as follows:

• mothers who were widowed and had not repartnered;
• mothers in one intact marriage;
• mothers who were divorced or separated, but not repartnered;
• fathers in one intact marriage;
• mothers who had divorced or separated and were repartnered;
• fathers who were divorced or separated, but not repartnered;
• fathers who had divorced or separated and were repartnered.

Apart from the effects of marital history, this pecking order illustrates several other dynamics. First, people generally 'do not give unconditional assent to the principle of kinship obligations' (Finch and Mason 1993). Rather, this is tempered by the nature and duration of personal ties. Second, gender is important, as mothers in general appear to reciprocate support more than fathers (particularly in interchanges with daughters) and this is an important factor in striving for 'a proper balance' in receiving and donating relationships (Finch and Mason 1993). Third, personal relationship circumstances compound the effects of past divorce or separation. Widows and divorced mothers may be regarded as more in need of children's support than are fathers in general. However, it becomes clear that older men are more likely to experience diminished younger generation support following marital change.

According to these findings, the fairly high and stable divorce rates over the 1980s and 1990s may compromise the availability of younger- generational support for the age cohorts who will be aged over 60 years in the second and third decades of the twenty first century. This implies that older parents with a history of marital separation, particularly those living alone, may need to rely more on public nursing, home help or accommodation services than those who have not experienced marital separation. Therefore, long-term planning of public support services may need to compensate for shortfalls in filial resources, particularly for divorced older men. Also, the need for an expansion of aged services in the future is suggested for the low-income, not employed and non-English-speaking background groups, who may need to rely more and more on offspring for support as they age and whose offspring may be unavailable or unable to provide support.

Expectations of aid

As well as being asked about the help they received from their children, the later life parents were asked who they would turn to for help in certain situations. Table 8 gives the proportions of respondents with different characteristics who felt they would turn to independent adult children for care when ill, to confide in, and for help in an emergency.

Generally, the proportions of respondents who would expect these forms of support from adult children were smaller than the proportions actually receiving emotional and practical help (see Table 7). This could in part be due to definitional differences, as parents may view 'confiding in children' as narrower than receiving general 'emotional support', and 'practical help' would cover a much wider range of tasks or services than the special cases of illness and emergencies. However, the difference could also reflect these parents' feelings that they should not expect to (or would not desire to) call upon adult children in such circumstances.

As was the case with receipt of aid, mothers were more likely than fathers to expect all three types of aid from their children. Those in the workforce were less likely to expect all three types of aid from children (see Table 8). Those on the lowest incomes were the most likely to expect help from children when they were ill. Parents with English speaking backgrounds were less likely to expect help in emergencies and were less likely to confide in adult children than were those with non-English-speaking backgrounds. Age was related to expectations too, with those over 60 more likely to expect help in illness and emergencies than those in their fifties.

The picture that emerges is one of parents with fewer material or social resources (due to low income, migrant status, or lack of a partner) relying more on their adult children than other parents. However, having more than one independent child over 18 years did not increase the likelihood of seeking these types of help, despite the existence of a larger potential resource pool. The age-related effect may indicate perceptions of higher or more immediate need by the older respondents, to whom the possibility of requiring such extra help may be more salient.

* Significant at p < .0001 ; ** Significant at p <.01 ; # Significant at p<.05 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

As was the case with actual aid, a past marital separation was related to expectations about aid, but only for those who were living alone: the widowed group were much more likely to expect help from their children than were the divorced or separated (see Table 8). Of course, this is partly related to gender because there were far more women among the widowed group. It should also be noted that parents with partners, regardless of their previous marital history, were most likely to report turning to their partner for help rather than to other family members, including adult children.

Relationships and family values

So far we have seen that the extent of intergenerational exchange is influenced by some personal and social characteristics, as well as by the more structural dimensions of family cohesion, such as family structure, proximity of relatives and contact frequency between family members. However, there are more subjective aspects of family cohesion that might be just as important in determining exchange of aid. In particular, the following analysis explores satisfaction with relationships between respondents and adult children, as well as beliefs about family culture and intergenerational obligations.

Satisfaction with parent-child relationships

First, to indicate the quality of filial relationships, a scale measuring satisfaction with the relationship with adult children was used. Again, it was not possible to gauge closeness with individual children, as the question was asked: 'How satisfied are you with your relationship with your children?'

On a five-point scale from '1 = Very satisfied' to '5 = Very dissatisfied', 75 per cent of parents whose adult children all lived away independently said they were very satisfied with their relationships with their children. A further 20 per cent were satisfied. Table 9 shows that the highest levels of satisfaction with parent-child relationships were associated with gaining emotional support or practical help from adult children, and with expectations of help.

However, although only a small group of twenty-five parents were dissatisfied or had mixed feelings about their relationships with adult children, these feelings were linked with a lower probability of gaining assistance from those children. For example, only 33 per cent of those who were not satisfied with filial relationships received any practical help from adult children, compared with 80 per cent of those who were very satisfied. Not surprisingly, it appears that subjective feelings about the nature of intergenerational relationships make a difference to the flow of aid, but it is unclear whether poorer relationships led to there being less chance of gaining help or whether they resulted from receiving less help.

* sig at p=0.0001 (Cramer'sV=0.23); ** sig at p=0.001 (Cramer'sV=0.20); 
# sig at p=0.05 (Cramer'sV=0.12). 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Furthermore, mothers (81 per cent) were more likely than fathers (67 per cent) to have very satisfactory relations with adult children, as was also the case for those who had not experienced marital separation (81 per cent) compared with those who had (59 per cent). This is consistent with mothers, people in intact marriages and the widowed also receiving more help from, and having more contact with, adult children than did those with a history of marital separation.

Attitudes to intergenerational obligations

A second measure was constructed to probe views about intergenerational family obligations. This was a normative scale indicating parents' views on the obligations of adult children toward older parents. The scale is based on a four-item factor derived by performing principal components analysis on nine family values propositions.

Respondents were asked whether they strongly agreed, agreed, had mixed feelings about, disagreed, or strongly disagreed with the following statements: 'Adult children should help their parents financially if they need it. Children should let ageing parents live with them if they need to. It is not the responsibility of adult children to take care of ageing parents. I would rather use outside services than have help from family or friends.'

The raw 'obligations' scale had a range from 3 (denoting very high support) to 19 (no support) and was reasonably reliable (Chronbach Alpha = 0.65). When making the scale, the last two items were reverse-coded for consistency, as they were worded in a negative way.

Parents receiving financial assistance from adult children more strongly supported the belief in adult children's responsibility toward older parents than did those not receiving financial assistance. However, there were no differences regarding the receipt of emotional or practical aid from children, nor were there differences in the expectations of help in illness, help in emergencies or sharing of confidences. There were also no significant differences in this intergenerational obligation scale according to marital history.

In contrast with satisfaction with relationships, stronger support for filial obligation was more likely to be reported by fathers than mothers (42 per cent compared with 26 per cent). Older men more strongly endorsed the proposition that adult children should care for older parents and support them financially. Perhaps some men were more traditional in their views or were worried about who would look after them if they lost their wife. However, the gender difference in views could also reflect the fact that many women knew more fully the commitment needed to care for elderly parents and did not believe that children should feel obligated to shoulder such a burden. (A detailed discussion of elder care and responsibility is provided in the section examining respondents and their elderly parents).

The degree of satisfaction with relationships with adult children seems important for ongoing support from them, while there is only limited and somewhat unclear evidence that parents' attitudes toward filial obligations are linked with actual or expected support from adult children.

Of course, aid given to later life parents could depend more on the children's attitudes towards intergenerational obligations than the parents' own views, but unfortunately the views of adult children are not available for this paper. Furthermore, although Webster and Herzog (1995) note that members of the older generation tend to perceive parent-child relations more positively than do their children (a form of 'generational bias'), it cannot be assumed that parents view filial obligations as more important than do children. Neither can it be assumed that the views of the two generations will coincide. Indeed, Finch and Mason (1993) point out that the sense of commitment between parents and children (and other family members) develops and changes over time, and may be viewed differently by members of the same family as they move through different stages and circumstances.

As well as receiving some assistance from their adult children as they grow older, parents generally continue to help their children, no matter what their ages (Finch 1989; Towsend 1989; Greenberg 1991).

Regarding assistance given to adult children, respondents were asked: 'Do you provide your adult children with emotional support or advice?' 'Do you help your adult children financially if they need it?' 'Do you provide your adult children with any practical support (doing things for them)?'

Psychological support, social interaction and practical help through money, goods and services can all be important resources for today's younger generation, particularly in a political climate that is increasingly adopting a 'user-pays' philosophy concerning social infrastructures and welfare services. Certain changes to federal legislation - such as restricting eligibility for Austudy or unemployment benefits for young adults - are based upon the premise that parents are still responsible for the welfare of adult children who are unemployed or studying. Prolonged dependency of increasingly large sub-groups of adult children may result, but some children may be more likely than others to receive help from parents.

To explore the extent and correlates of aid, as well as reasons for aid, provided by older parents to adult children, the parents in this study were asked what type of support they gave to their adult children (those over 18 years): financial, practical or emotional. Because these questions were asked about adult children in general, it was not possible to determine the characteristics of those who did or did not receive help from parents, but just that some help was given. However, it was possible to examine the characteristics of the parents who were providing the assistance (see Table 10).

The great majority of parents were helping their adult children. The most common form of assistance was emotional support and advice, which attests to the intimate nature of parent-child relationships. Although almost universal, the provision of emotional support and advice to adult children was related, to a small degree, to the parent's gender, age and current relationship status. Emotional support was more likely from mothers than fathers, from those under the age of 60 than those over 60, and from those living with a partner (whether a registered or de facto marriage) than those living alone (whether widowed, divorced or never married). This emphasises the closeness of mother-child relationships - an intimacy clearly demonstrated by previous Institute research which found that both men and women have more contact with, and tend to confide in, their mother above any other relative (Millward 1996). There was also a suggestion that younger parents (those under 60) may be more in tune with the problems of their adult children than older parents. Those who themselves had emotional 'back-up' in the form of an intimate relationship with a partner or spouse also appeared better placed to support adult children emotionally (see Table 10).

The least common form of aid to adult children was financial help, but this was still given by over two-thirds of parents. Overall, financial assistance was related mainly to the economic and employment status of parents or basically to their means to assist. Those with higher incomes (particularly over $60 000 per year), higher education and those still in the paid workforce were most likely to help adult children financially. Of interest here is the fact that, as with emotional support, parents who were married or de facto appeared better placed to help support adult children financially, perhaps because of the availability of double incomes (from pensions, wages, etc.) compared with parents who were unpartnered (see Table 10).

Again, the giving of practical assistance was widespread, with more than three quarters of the parents interviewed giving this type of assistance. This was not significantly affected by parental characteristics.

Overall, 60 per cent of parents reported giving all three types of assistance to children. This seems a big commitment on the part of these parents, half of whom also had elderly parents to consider.

** Significant difference between groups at p < .01 ; 
* Significant difference between groups at p < .05. 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Why do some children get more help than others?

More specifically, parents were asked: 'Do any of your adult children need more help from you than others?'

If the answer was yes, they were then asked 'Who?' and 'Why?' Where extra help was given to a particular adult child, the design of the study made it possible to examine that adult child's personal circumstances as well as those of the parent. Parents' own explanations for any differential treatment of their children were also elicited to provide a deeper understanding of circumstances.

In total, the 630 parents supplied information about the personal circumstances of 1958 adult children. This is an average of 3.1 children per respondent, which reflects the relatively high fertility rate of this age cohort (de Vaus, Wise and Soriano 1987). The children concerned could be those of the respondent or of their spouse or of both.

Three per cent of these adult children had no adult siblings to enable comparisons of aid given by parents and so were dropped from this analysis. Nearly 10 per cent (n = 175) of the remaining adult children were reported by parents to need more help than their siblings. It is this group that will be the focus of the next section of analysis.

First, Table 11 compares the characteristics of adult children who received extra help from parents with those who received no extra help.

Where adult children live

Extra help was related to where adult children lived in two ways: those who were still living with parents and those who had moved out but were living nearby tended to received more help than siblings further afield.

* Excludes those with no adult siblings. 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Co-residency

Nearly twice the proportion (17 per cent) of co-resident adult children as non-resident ones (9 per cent) received extra help from respondents (see Table 11). Regarding type of help given (not shown in the Table), a greater proportion of parents (80 per cent) gave financial assistance to resident children than to non-resident ones (69 per cent), and more parents (90 per cent) gave practical aid to resident children than to non resident ones (80 per cent). This is not surprising, for, as one mother put it,

'[My son] lives at home so he gets his washing and cooking done.'

However, while more help was exchanged in co-residence, this also resulted in lower levels of satisfaction with relationships with adult children: 76 per cent of parents were very satisfied with relationships when all adult children lived away, compared with 57 per cent being very satisfied when at least one child over 18 was still at home. This concurs with work by White and Rogers (1997) which found that increased interaction via co-residence could exacerbate parent-adult child tensions, particularly if co residence is involuntary. Relationship difficulties may also be due in part to the extra physical load on parents whose adult children are still at home.

These examples reflect a current Australian trend in co-residency, where children are staying with parents for longer periods of time than was the case in the past. As discussed in de Vaus and Qu (1997), the ABS Families Survey (1992) showed that around 43 per cent of young people aged 20 to 24 years were still living with parents, as were around 12 per cent of those aged 25 to 29 years .

This is quite a high percentage of those under 25 years and reflects the trends toward later age at first marriage (Qu 1998), as well as continuing educational participation. Furthermore, co-residence on the part of a youngest child does not preclude older siblings having also enjoyed a great deal of support from parents before they left 'the nest', so the overall period of time spent by later life parents with adult children in the home could be quite considerable.

Proximity

However, for 73 per cent of these parents, all children over 18 years of age were living away independently. Of parents with non-resident children, just over half (54 per cent) had at least one child living within a 30-minute drive. Nearly all (90 per cent) spoke with at least one child weekly or more frequently (either in person or on the telephone). As has been found in other large studies (Millward 1996), and is noted in the analysis above, close proximity and frequent contact were strongly correlated, so the nearer adult children lived, the more interaction occurred between the two generations.

The physical proximity of children to parents also made a difference to the likelihood of additional aid being given. That is, where children were not actually residing with their parents, the closer they lived, the more likely they were to gain additional help from parents. One mother said she helped her daughter more 'just because she lives closer than the son'.

Indeed, nearly half (47 per cent) of those being given more help than their siblings lived within a half-hour drive of their parent or parents, while only a fifth (19 per cent) of those given additional assistance lived more than 2 hours away. Geographical closeness can be viewed as an important element of 'structural solidarity' in families (Roberts, Richards and Bengtson 1991); these findings are consistent with other studies showing the importance of proximity of family members in facilitating family exchange (Millward 1996; Finch 1989; Glezer 1991).

Personal characteristics of parents

The gender of parents did not affect the likelihood of them giving extra help to certain children; nor did marital history, age or employment status. It was anticipated that perhaps older parents and those who had retired might give more practical help to adult children, as they might have more time available, but this was not found to be the case.

Personal characteristics of adult children

While the characteristics of parents had no impact on the likelihood that they would give extra aid to particular children, the characteristics of the adult children themselves certainly did explain why some received additional support from their parents.

Age

First, those receiving 'extra help' were likely to be younger. Table 11 shows that a higher proportion of 18- to 24-year-olds received extra help than did not, but a lower proportion of over 35 year olds received extra help than did not. Some parents explained that they gave more advice to very young adults (often their youngest child) because they were just more psychologically dependent on parents:

'My son is the youngest and needs a little more support at times.' 
'Need more support and advice because they are young and inexperienced.'

More practical assistance was also reported to go to the youngest child because they were just starting out and had fewer resources than their older siblings, who were generally better established financially and considered to be more psychologically independent. Therefore, youth was a factor in determining need. Even though older siblings were more likely to have children (and more children) than their younger siblings, the assistance expected via grandparenting did not appear to cancel out the age effect.

Occupation

Somewhat related to age and experience was the adult child's occupational status, as extra financial support was frequently said to be needed by adult offspring who were studying, particularly if they were undertaking full-time university courses. Indeed, a number of parents commented that adult students (some in their twenties or thirties) were largely or wholly financially dependent upon them:

'He's a student and 100 per cent financially dependent.' 
'She's still studying at university and at this stage not getting any government support.'

Also, some children were unemployed or in precarious employment, so their parents provided a financial 'cushion' for them. One father said he helped his youngest son the most because,

'He never stays in jobs for very long. There are financial difficulties so we need to support him.'

It is not surprising that the financial needs of adult offspring are linked to their occupational activities. Table 11 shows that those in paid employment were not as likely as the others to receive extra help. Nevertheless, some parents explained that they gave both additional financial and practical help to a particular child because they had just started their own business; for example,

'My son has just bought his own farm in partnership with his wife.'

Also, some grandparents said they provided work-related child care because their daughter was in paid work and so needed more help with babysitting than did her siblings:

'I mind [this daughter's] children a lot more because she's working.'

There was a larger proportion of homemakers among those adult children receiving extra help than those not receiving extra help (see Table 11). These were likely to be daughters (whether 'lone' mothers or not) and to be caring for children, which tended to increase the amount of time or other resources being invested by their parent or parents. This finding may partly reflect the higher levels of grandparental involvement found elsewhere when grandchildren are below school age (Millward and Matches 1995).

Gender

A much higher proportion of daughters than sons were given extra assistance by parents (see Table 11). Indeed, female social roles of 'kin-keeping', domestic organisation and child nurturing imply greater family involvement, and it has been shown that assistance flows more through maternal kin lines than paternal ones (Finch 1989; Rosenthal 1985). This means that women tend to be more closely linked within families than are men. It was particularly the later life mothers who reported giving additional domestic help to daughters rather than to sons, acknowledging their daughters' greater domestic responsibilities:

'She needed help because she was the girl in the family.' 
'I do more for my daughters - they do not have wives.'

In contrast, the father of a man who was a long-distance truck driver reported providing extra practical support to him because, while he is away:

'I need to keep an eye on his house and family for him.'

A further point is that daughters are much more likely than sons to be lone parents. According to the 1991 Census (ABS 1993a), 86 per cent of one-parent households were headed by a mother. Such households tend to have low incomes and female lone parents are more likely to have young children than are male lone parents. Not surprisingly then, if one child was a lone parent they tended to need more help than their siblings, according to some parents; for example:

'She's on her own on a single parent's pension. It's hard for her to make ends meet.'

However, not all lone parents mentioned were daughters. For example, one older mother explained how she provided extra babysitting for her son because

'He has three kids to bring up by himself.'

Marital status

The amount of support received by adult children who were lone parents is one consequence of adult children's marital or relationship status. Indeed, a greater proportion of those receiving extra help from parents than of those not receiving extra help were divorced or separated (see Table 11 ). Often parents gave emotional support to help adult children through marital breakdowns as well as through other relationship problems; for example:

'For emotional reasons. Both [daughter] and her son still bear scars from an abusive relationship.'

'[Daughter] has split up with her partner and has a child from that relationship. She is going through a traumatic period because they split up only a few weeks ago.'

The other significant effect of adult children's marital status was that the largest proportion of those receiving extra parental help were single and never married (see Table 11). Thus, some parents provided friendship and support for children who lived alone or did not have an intimate relationship. For example, one mother said:

'[My daughter] grew up without a father, and she likes to stay closer to me. She needs me as a friend - we discuss things a lot.'

Others gave additional practical aid to unpartnered children, especially if they were living alone, since they had no-one to assist them with domestic work or maintenance jobs:

'She's single - she needs more help with practical tasks.'

Being single and never married is generally associated with youth and inexperience, so may essentially equate to a state of continuing dependency, whether via co residence or not. However, with later age at first marriage being more common, and an increased propensity not to marry at all (de Vaus 1997), there is some evidence that the phenomenon of living alone or 'being unattached' is becoming more common in older adult children. The present findings suggest that single adults draw more upon their parents' resources than do married children, thus prolonging the term of parental responsibility beyond the time of parents' retirement and entry into old age.

Parental status

There was no evidence that adult children being parents necessarily resulted in additional help being received from the children's grandparents (see Table 11). Two thirds of the later life parents interviewed were grandparents, so it could well be that this was such a generic state among them that it was not viewed as a reason for giving special help to the adult children. Nevertheless, the need to give practical help to one child more than another was explained by some older parents in terms of the needs of grandchildren, and it tended to be linked to the age and number of grandchildren. While some later life parents said one adult child needed more help because they had the youngest grandchildren (often babies), others said it was because a particular adult child had the most children (generally four or five). For example:

'[My daughter] needs help because of her babies. They are very tiny and only one year apart.'

'[My daughter] has five children [and] is always needing a bit of help around the house...and looking after the children.'

Obviously the life stage of the adult child is important. The parents of the youngest grandchildren will be inexperienced in child care and so need extra help, though they themselves are not necessarily the youngest child of the respondents.

Other explanations for extra help given

A number of other explanations were provided by parents for giving particular adult children an extra helping hand.

Sometimes more guidance and emotional support was given because of the adult child's personality traits; for example, to those considered by the parents to be emotionally insecure, immature or even 'irresponsible'. For example:

'He's having trouble settling down. Has a lot of growing up to do.'

'He is emotionally immature - cannot get on his feet and become independent. He needs to make his own way in life.'

These emotional needs may or may not be linked with a clinical psychological problem or condition. However, problems such as depression, attempted suicide, aggression, instability or addictions to drugs, alcohol or gambling did require extra support from some older parents. For example:

'He needs help because of his current drug addiction. He ended up in a psychiatric hospital two years ago.'

Other health problems of children or grandchildren also required extra support from some older parents; for instance, a physical or mental disability or handicap on the part of the adult child or grandchild:

'[My daughter] is disabled and needs full-time care.'

'[My daughter] has an intellectually disabled child and frequently needs more support and freedom.'

Others gave the short- or long-term illness of an adult child or grandchild as the reason for giving extra help:

'[My son] has epilepsy and cannot drive.'

'[My daghter] has hurt her back.'

Discussion

The characteristics and life stage of adult children, as well as other special circumstances such as illness, clearly influence the amount of time and other resources being 'donated' by the older generation to their adult offspring. The findings also concur with Greenberg's (1991) work, which illustrates the extent of emotional involvement of older parents with their adult children, and their consequent efforts to try to help children overcome difficulties, whatever their nature. Although particular problems or characteristics of children led parents to feel more responsible for some than for others, it is important to remember that there was also a high level of general support overall (see Table 10).

From the parents' point of view, prolonged periods of education, increased incidence of lone parenthood and fewer employment opportunities for young people often necessitate the extended support of adult offspring. The changing nature of public welfare provision is restricting financial support for young adults, which is also resulting in an increased load on older parents.

The fact that later life parents tended to give more help to children who were either single without children, divorced with children or experiencing personal problems also reflects the needs of young adults which are inherent in current social trends toward semi-permanent relationships, delayed family formation and increased rates of marital dissolution.

A sizeable proportion of the parents interviewed, roughly half of whom were retired, were still involved with meeting the needs of their children. Past studies have shown that most family support flows from older to younger generations (Finch 1989; Glezer 1991). The present findings support this observation, but most apparently in the area of financial support, as there was a good deal of reciprocity of emotional support and practical help between the generations. Unfortunately, the scope of the study did not allow for the amount of help exchanged to be quantified (for example, in terms of hours per week, or dollars per month given or received). Therefore, although we know that a greater proportion of parents were supporting adult children than vice versa, we do not know exactly how much help was actually being given to individual children.

Older parents are not necessarily dissatisfied with their commitment to adult children, but the findings underline the fact that parenting responsibilities do not cease just because children become adults. Many of the respondents are shouldering, to varying degrees, the responsibility for younger adults in a variety of ways including co residence, financial support of students and back-up services for lone parents.

As established above, two-thirds of the respondents who were parents were also grandparents, and this status brings another dimension to intergenerational relations. Settles (1993) maintains that most people look forward to spending time with their grandchildren. A number of studies have also shown that the majority of grandparents tend to live in the same city as their adult children and grandchildren, facilitating a considerable degree of aid to parents and active involvement with grandchildren (Shanas and Streib 1979; Lee 1980; Kendig and Rowland 1983; Glezer 1991; Millward 1996).

Because the period of child-bearing in the late twentieth century is much shorter than it once was, today's grandparents are freed earlier from full parenting obligations and so, potentially, can devote more time and energy to their grandchildren. In addition, Hagestad (1988) proposes that grandmothers exert a stabilising effect on mothers, which allows them to feel more confident and competent with their own children. Grandmothers' experience, back-up and advice can, therefore, be seen as an important 'second-order' generational effect. Indeed, both British and Australian studies have shown that between a quarter and a half of employed women have their children looked after by a grandmother while they are at work (Finch 1989; Glezer 1991; Millward and Matches 1995).

However, there may also be a negative side to grandparental involvement. Bergquist, Greenberg and Klaum (1993) found that grandmothers who felt obligated to provide long periods of child care were under stress and reported diminished enjoyment of the time spent with grandchildren. Townsend (1989) says it is unfair of people to expect their parents to provide child care after retirement because older people deserve some time to themselves to develop new skills or interests, to travel or to spend more time with their spouse or friends.

The important questions for the purposes of this study are, Did the grandparents see their role as a positive one? How involved were they with their grandchildren? How did they feel about looking after grandchildren, and what factors affected such feelings?

Sample of grandparents

There were 449 grandparents in the sample: thirty-five per cent were men and 65 per cent were women. This sex bias was due partly to the sex bias in the group of people who agreed to be interviewed, as women were more likely to agree than were men. It is also consistent with there being more women in these age cohorts in the general population, and to women becoming grandparents at an earlier age than men. In this sample, more women than men aged in their fifties were grandparents.

As was the case with aid to adult children, questions about grandparenting referred to grandchildren in general, not to individual grandchildren. However, 147 of the grandparents had only one son or daughter who had children, so for this group the characteristics of the adult child who was the parent of the reference grandchildren were identifiable, as were the ages of the grandchildren.

Therefore, when analysis uses respondents' characteristics, it is based upon 449 grandparents, but when analysis also considers adult children's characteristics, it will be based upon a sub-set of 147 grandparents with one adult child who is a parent.

Mediating factors in the experience of grandparenting

The experience and level of involvement of grandparenting may be affected by a number of personal factors. The first three to be considered here are the gender, workforce participation and marital or relationship status of both the respondents and their adult children who are themselves parents. Thus three generations are taken into account. The fourth main factor considered is the proximity of grandchildren to their grandparents.

Gender

There is assumed to be an intergenerational bond of motherhood between mothers and daughters due to the existence of 'comparable tasks and responsibilities' (Fischer 1986). However, because of the matrilineal nature of exchange, there is generally more help with the grandchildren from the maternal grandmother and less involvement from paternal grandmothers (Millward 1996; Fischer 1986; Chodorow 1978). Indeed, for women, having children has been found to lead to less conflict with their own mother but more with their mother-in-law (Fischer 1986).

Strong mother-daughter relationships might also imply less involvement with grandchildren for grandfathers. However, men interviewed by Bergquist, Greenberg and Klaum (1993) said that being a grandfather often gave them an opportunity to enjoy their grandchildren in a way they had not experienced with their own children because they lacked time. This sometimes caused resentment on the part of their own children, who saw the big difference between the attention their children were receiving and the attention they had received.

In this study, grandfathers and grandmothers were compared according to how often they saw grandchildren, how often they looked after them, who took the main responsibility for caring, the level of satisfaction with caring and the importance of grandparenting.

Grandparents were asked the following questions:

• 'How often are you in touch with your grandchildren? How often do you look after your grandchildren?' (The options were: once a week or more often; fortnightly, or a few times a month; about once a month; every few months; occasionally; not at all.)
• 'When looking after grandchildren, do you, or your [spouse or partner], or both take the main responsibility in caring for them?'
• 'Overall, when you're looking after grandchildren, do you find: It is positive and satisfying? You have mixed feelings? It is quite a strain?'
• 'How important is being a grandparent to you? Is it: Very important? Important? Not at all important?'

Grandmothers were in touch with at least one of their grandchildren significantly more often than were grandfathers, but, rather surprisingly, there was no significant difference between grandmothers and grandfathers regarding how often they looked after a grandchild (see Table 12). However, when grandfathers reported looking after grandchildren, only 4 per cent of the grandfathers interviewed said they took the main responsibility of care, while around half shared it with their partner or spouse and nearly half said their partner or spouse was solely responsible for the grandchildren. In contrast, more than half (54 per cent) of the grandmothers reported taking the main responsibility of care for grandchildren, with just over one-third saying they shared the responsibility with their partner or spouse.

Perhaps related to this tendency for grandmothers to take responsibility is the fact that only 7 per cent of grandfathers were not completely satisfied with caring for grandchildren, while 14 per cent of grandmothers had some reservations about the merits of caring (either 'mixed feelings' or thought it was 'quite a strain'). The burden of care, therefore, appears to be heavier for grandmothers, which is consistent with studies from the 1970s and 1980s. The nucleus of family care does not seem to be shifting from women to men, at least in this generation of grandparents. It will be interesting to see whether or not today's fathers take on a greater role as grandfathers than their own fathers have done.

Not surprisingly, significantly more grandmothers (86 per cent) than grandfathers (62 per cent) reported grandparenting to be very important to them. Because grandmothers are more involved with grandchildren, their grandparenting role probably contributes more to their sense of identity than is the case for grandfathers, particularly if they see themselves in a mentoring role for their daughters or daughters-in-law, or have made raising a family their main life's work.

Both grandfathers and grandmothers were likely to see their daughters' children more often than their sons' children. For example, 78 per cent of grandmothers saw a daughter's children at least weekly, compared with 64 per cent who saw a son's children this often. However, there was no difference in how often they actually looked after the children of daughters rather than sons. This suggests that there is a closer relationship and tendency to socialise more with daughters and their children, while the provision of child care appears to be linked to need rather than gender-specific relations.

Workforce participation

Another lifestyle factor for grandparents is their attachment to the paid workforce. A major change for people in middle age and early old age can be retirement from paid work, for themselves, their partner or both. Among the respondents there was a much higher proportion of grandmothers (68 per cent) than grandfathers (47 per cent) in the retired or non-employed group.

At the same time, an increasing number of grandmothers are participating in paid employment. At the 1991 Census, 63 per cent of married women aged 45 to 54 years and 34 per cent of those aged 55 to 59 years were in paid employment, compared with 36 per of those aged 45 to 54 years and 23 per cent of those aged 55 to 59 years at the 1971 Census (ABS 1993a). However, most practical help is provided by grandmothers - the age group of women whose labour force participation rate has been increasing. Hagestad (1987, p. 419) maintains that this may lead to an overload for middle generation women, who assist adult children, grandchildren and parents 'in addition to facing the demands of workday life and household maintenance'.

Since the grandmothers interviewed were the ones giving most of the care to grandchildren and thus expending more time and energy on the grandparenting role, it might be expected that those who were in the workforce would look after grandchildren less often than those not in the workforce. However, this was not found to be the case.

First, grandmothers in the paid workforce saw their grandchildren slightly more often than did those who were not in the workforce or who were retired (84 per cent compared with 72 per cent weekly). This appeared to be related to the age of grandchildren, as working grandmothers were more likely to have grandchildren aged less than 2 years.

Second, retired or non-employed grandmothers did not look after grandchildren more often than those in paid work - 34 per cent of both groups did this weekly. Nor were they more satisfied with caring for grandchildren. Around 85 per cent of both groups were very satisfied, so employed grandmothers did not find caring for grandchildren less satisfying due to the possibility of conflicting demands on their time.

Furthermore, when considering the effect of the workforce participation of the grandchildren's parents, it was found that those in the workforce were no more likely to have their children minded by grandparents than were those not in the workforce. There was also no significant difference in the frequency of looking after grandchildren according to the workforce participation of daughters. In fact, a higher proportion of grandparents looked after grandchildren fairly regularly when the mother was not in the workforce (53 per cent weekly or monthly) than when she was (39 per cent weekly or monthly). This seemed to be related to the fact that 53 per cent of the grandmothers looked after babies or toddlers at least once a week, while only 17 per cent looked after primary school-aged children that often, and daughters were more likely to be at home with very young children.

Finally, workforce participation made no difference at all to the frequency of grandfathers seeing their grandchildren or caring for them (whether work-related or not), or to the feelings of satisfaction grandfathers gained from caring for their grandchildren. This is not surprising since in most cases their partner or spouse was actually providing the care.

Marital or partnering status of grandparents and adult children

Earlier analysis suggests that the marital or partnering status of both grandparents and their adult children might also bear upon the experience of grandparenting. As has been shown earlier in this paper, the fathers interviewed for the study who had experienced marital separation had much less to do with their adult children. There was a similar outcome if men were without marital partners due to being widowed, although this applied to very few men in the study.

When grandparents are divorced and repartnered, grandchildren acquire extra grandparents, so there can be few grandchildren and many grandparents, making involvement complex and possibly contentious. Alternatively, if parents separate and repartner, grandparents can acquire new step-grandchildren, so there can be few grandparents and many grandchildren. Thus, loyalties, time management and relationships can become complex and strained (Bergquist, Greenberg and Klaum 1993).

Our study found that grandparents who were divorced or separated saw grandchildren less often than did married or widowed ones. Partnered grandparents were then compared with unpartnered ones to see whether or not being on their own made a difference to people's interaction with their grandchildren (the partnered grandparents were generally a bit younger).

Table 13 shows that the partnering status of grandmothers made no difference to the frequency of seeing grandchildren, but that partnered grandfathers saw their grandchildren significantly more often than those without a partner (63 per cent compared with 42 per cent weekly).

* Significant difference for grandfathers only (Phi= .17; p= .06) 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

A similar and slightly stronger association held between gender and partnering status for frequency of looking after grandchildren. Partnered grandfathers were three times as likely as unpartnered ones to look after grandchildren, which is consistent with the spouse or partner of the grandfathers actually taking care of the grandchildren and probably also organising the visits with them.

Being partnered or not made no difference to how often grandmothers looked after grandchildren, except that partnered grandmothers were likely to provide work-related care for grandchildren (20 per cent) slightly more often than were grandmothers living alone (8 per cent). For both grandmothers and grandfathers, having a partner made no difference to the perceived importance of grandparenting. However, grandfathers without partners were more inclined to feel that caring for grandchildren was a bit of a strain (12 per cent compared with 2 per cent).

An adult child's marriage breakdown has also been shown to disrupt kin networks and ties such that post-divorce maternal grandparent-grandchild ties are stronger, while paternal grandparent-grandchild ties are weaker (Thompson and Li 1992; Funder 1989). This is because children generally live with their mother after divorce, so kinship links are more active on the mother's side. However, Funder (1989) stresses that grandparent roles are flexible and negotiable: while children may be less likely to see paternal grandparents after divorce, as they grow into teenagers they are likely to want to see less of all grandparents.

There was information on only five divorced or separated adult children in this sub sample of respondents with grandchildren by only one adult child, so meaningful comparisons with married offspring could not be made. Of these five, however, two had children cared for weekly and one had children cared for monthly by the grandparents.

Proximity of adult children and grandchildren

We have already seen that proximity affects contact and involvement between generations. Therefore, the final factor considered in grandparent-grandchild attachments is that of proximity. The relocation of older people to more desirable holiday-type areas has been termed 'retirement migration'. Although this may, in one sense, be seen as withdrawing from family obligations, it can also attract extended visits from family - notably grandchildren (Szinovacz and Ekerdt 1995). Retirement relocations can also involve moving closer to family members, thus strengthening ties. Either way, such relocations may vary the amount of interaction between grandparents and grandchildren.

Half of the respondents' sons and daughters who themselves had children lived very near to their parents. In fact, nine were actually living with the respondent. Table 14 shows a very strong relationship between close proximity and frequency of both seeing and caring for grandchildren. This might suggest that grandparents who move away on retirement will have less involvement with grandchildren. However, there were no significant differences according to retirement status. Only 29 per cent of those who were retired (which for some women related to the retirement of their spouse or partner) compared with 26 per cent of those not retired were living further than a 2-hour drive away from their son or daughter and their grandchildren.

* Significant difference (Gamma> .40; p< .001) 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

It appears that difference in proximity rather than any specific effect of the transition to later life or retirement status per se was responsible for difference in contact. It should be noted that adult daughters did not generally live closer to their parents than did adult sons, so close proximity would not account for mothers of adult daughters seeing the grandchildren more often than mothers of adult sons.

Discussion

Overall, interaction with grandchildren was an important and satisfying part of the later life experience of these men and women. However, since being a grandmother rather than a grandfather was associated with more contact and greater responsibility for the care of grandchildren, the perpetuation of gender roles is evident. The mother-daughter bond is also evident, since mothers of daughters with children saw their grandchildren more often than did mothers of sons with children.

According to these findings, the responsibility for family care does not seem to be shifting from women to men in this generation of Australian grandparents. However, with the increasing social emphasis on father-child involvement (Amato 1996), it will be interesting to see whether or not today's fathers will take on a greater role as grandfathers than their own fathers appear to have done. So far, however, research indicates that men are still generally reluctant to commit more time to the care of children (Daly 1996) and this may continue to hold for both fathers and grandfathers.

While some grandmothers had mixed feelings about looking after grandchildren, most were very positive about doing so. They also seemed more likely than grandfathers to identify themselves in terms of their grandparenting role, since they felt it was more important than did the men. Grandfathers without partners appeared much more isolated from their grandchildren than were those with a spouse or partner. No inference can be made, however, as to whether this is due to choice or to social roles and circumstances.

The time of greatest need of assistance with young grandchildren is when many grandmothers are still in paid employment themselves. The demand for care of grandchildren by grandparents may also be on the increase if projected demands for work-related child care into the next century cannot be satisfied by child care centres due to escalating costs to parents. Further withdrawal of government subsidies for child care therefore implies an increasing reliance on informal sources of care: chiefly middle-aged grandmothers, many of whom are still in the paid workforce.

Retirement from the paid workforce did not seem to engender more involvement with grandchildren. On the contrary, the findings suggest that when grandmothers are retired, and therefore older, their grandchildren are also generally older and do not need as much care. It will be interesting to see whether or not this trend will be reversed with the next generation of grandparents who, due to their adult children becoming parents at a later age, might commonly have grandchildren born after their retirement.

Finally, the close proximity of adult children and grandchildren is consistent with the values of family solidarity, and indicates the potential availability of grandparents. However, the so-called 'army' of unpaid domestic and child care labour, in which middle-aged grandmothers feature prominently, may become less available with future generations, requiring substantial modifications of government policies predicated on the assumption of 'family care'.

The last section of this paper focuses on relations between the respondents and their own elderly parents. Due to the advanced age of many parents of those people interviewed (who were themselves aged up to 70 years), much of the discussion will centre around the need to provide filial care by those with elderly parents in need.

'Care-giving work' for elders has been defined as: 'custodial or maintenance help or services, rendered for the wellbeing of older individuals who, because of chronic physical or mental illness, or disability, cannot perform such activities for themselves' (Waerness, quoted in Hooyman 1992, p. 182).

It is difficult to obtain exact statistics on the prevalence of caring for the elderly, as it depends very much on the definition of caring used, and this is often subjectively defined. In the United States, it was found that around 7 per cent of the adult population were spouses or adult children of disabled elderly people needing long term care (Stenberg-Nichols and Junk 1997). However, the proportion of these people who were actually providing care is more difficult to determine. In Australia, it has been estimated that around 17.5 per cent of households contain at least one carer for a person who is disabled, handicapped or frail elderly, around 5 per cent of adults are the 'main' carer for an elderly relative and around 40 per cent of all carers are adult offspring caring for an elderly parent (Howe and Schofield 1996).

Much of the literature on caring for elderly parents focuses on gender inequality in such caring work. For instance, Hooyman (1992) maintains that 'family care' is a misnomer, as men rarely care for impaired parents. Others have also noted that daughters, not sons, are expected to care for parents (Wilkinson 1988; Lehr 1984; Marsden and Abrams 1987).

Hooyman (1992) contends that, in the United States, the cost-cutting that is part of social policy around aged services is predicated on long-term care of the elderly being a familial rather than a societal one. This perpetuates the reliance on women's unpaid labour. According to McDonald (1997), there is also the fear in Australia that if more funding were to be put into formal supports for the elderly, then the family would abdicate its responsibility, placing a severe load on the public purse.

There have been claims that elder care responsibilities may lead to middle-aged daughters' careers, finances and health being compromised by the ongoing burden of caring, often with very little help from other family members (McDonald 1997). The close contact between elderly resident parents and their children has also been found to cause resentment in the children, particularly where elderly mothers are living with daughters. Lower levels of physical and emotional wellbeing have also been noted in carers of elderly people living at home (Marsden and Abrams 1987; Howe and Schofield 1996). It will be interesting to see whether or not the trend toward earlier retirement for men will result in their assuming more responsibility for elderly parents than has been the case up to now.

Before the analysis of intergenerational exchange between respondents and their elderly parents is presented, two provisos should be noted. The first relates to gender; some respondents, particularly men, clearly answered questions from the point of view of their support of parents or parents-in-law as a couple, not as an individual. The second is that only a certain percentage of elderly people need time-consuming or extensive care from their adult children. This, coupled with the fact that many people in their fifties and sixties no longer have any living parents, means that the problems or 'burdens' of elder care will apply to a minority of respondents, in contrast with the responsibilities for adult children, which (as seen above) applied to the majority of the respondents.

Who helps elderly parents?

Just over a third of the 50- to 70-year-old respondents had a living parent (254) and just under a third (206) had a partner with at least one parent alive. Overall, therefore, half of them (368) had either one of their own parents or their partner's parents still living.

This group was asked: 'Do you provide your parents with emotional support or advice?' 'Do you help your parents financially if they need it?' 'Do you provide your parents with any practical support (doing things for them)?'

As with exchange with adult children, these questions pertained to generic types of assistance and no particular tasks, amounts of help or sums of money were specifiied. However, there was a series of questions about being an elder carer, which will be discussed a little further on.

The majority of those respondents with elderly parents reported giving emotional support or advice (73 per cent) and practical help (68 per cent) to their parents or parents-in-law. However, respondents were less likely to help their parents or parents in-law financially than they were to help adult children in this way. Consequently, only one-quarter of those who had parents or parents-in-law gave all three types of assistance to them.

Some characteristics of respondents were related to the likelihood of assisting elderly parents, as shown in Table 15. Those with a higher income and higher level of formal education were a little more likely to be giving emotional support and advice to elderly parents, and those who had never experienced marital separation or divorce were a little more likely to give financial assistance. Respondents with more education may feel more obliged to give their elderly parents advice - and they may be in a better position to do so (this group includes those in professional occupations). Also, being in an intact marriage was associated with higher income - and presumably greater assets, since these had not been subject to division through divorce - thus giving respondents more scope to assist elderly parents financially.

** Significant difference at p < .01 ; * Significant difference at p < .05. 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Female respondents were a little more likely than male respondents to give practical help to elderly parents. This is in keeping with the expectation of daughters providing domestic assistance to elderly parents - for example, help with tasks such as shopping, cooking and housekeeping, and the provision of chauffeuring and of care when parents are sick. Furthermore, although 73 per cent of these women (n = 134) helped a parent (or parents) in practical ways, very few said that other people also helped. In fact, only sixteen women said a husband or other male relative helped, and a further sixteen mentioned help being received from a sister, sister-in-law or daughter.

Caring for elderly parents

Those with an elderly parent or parent-in-law were asked if they had the added responsibility of being a 'main carer': 'Are you the main carer of a parent or parent-in law with health problems or disabilities, or who is aged?'

Caring was whatever tasks and commitments the respondent defined as care, but those answering 'yes' had to be taking the main caring responsibility. That is, no-one else spent more time than they did caring for this elderly person. Provision of this more specialised or constant care for the elderly was found to be gender-related. As well as giving general practical help to parents, women were also much more likely than men to be acting as the main carer for an elderly parent. Of all respondents, 11 per cent of women and 4 per cent of men were the main carer for a parent or parent-in-law (a total of forty-four women and fourteen men). Of those with elderly parents, 21 per cent of the women and 8 per cent of the men were the main carer.

It appears that the greater propensity of daughters (or daughters-in-law) than sons to provide instrumental care to elderly parents may be linked more to the social expectation that women's family responsibilities take precedence over paid work than to their lack of commitment to, or perhaps more precarious attachment to, the paid workforce. Table 16 shows that half (52 per cent) of the female elder carers were either not employed or retired and a further third (32 per cent) were engaged in paid work for less than 40 hours per week. In contrast, ten out of the fourteen male elder carers (71 per cent) were in paid work for 40 plus hours per week. Furthermore, comments made by some of the men suggest that in fact their wives were sharing or taking the main part in such parental care responsibilities. Thus there is more evidence of an obligation to care and the tendency to arrange working hours to fit with other family responsibilities on the part of later life women than men.

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne

Respondents who did not live with a spouse or partner were also more likely to take on the main caring role (23 per cent) than were those who were partnered (12 per cent). This may also be somewhat related to gender because more women (77 per cent) than men (67 per cent) were living alone. Women in this older age group are more likely than men to be widowed and are also less likely to remarry after divorce (McDonald 1997), so they may have more time available to care for relatives other than a spouse.

Nursing homes

Half (n = 186) of the later life respondents with elderly parents or parents-in-law had at least one parent suffering from a long-term illness or disability. For most, this was an ill or disabled mother (n = 92) or mother-in-law (n = 69). Fewer fathers (n = 26) or fathers in-law (n = 26) were reported to be ill or disabled, reflecting the fact that fewer respondents had surviving fathers.

About half of the ill or disabled elderly parents were in a nursing home or special accommodation home. Again, this group was overrepresented by women: 80 per cent of those in homes were mothers or mothers-in-law, but only 20 per cent were fathers or fathers-in-law. Nevertheless, only about 12 per cent of all later life respondents had at least one elderly parent or parent-in-law in a nursing home.

Although most respondents who were main carers were caring for an elderly parent who was not in a nursing home, just over a third were responsible for the care of an elderly parent who was actually living in a nursing home. This could involve regular - even daily - visiting; taking care of clothing and personal effects, and personal business or finances; conferring with medical staff; buying pharmaceuticals or other needs; and possibly taking the parent on outings.

Support services

Outside these formal residential situations there was a range of government or community services used by some respondents who were carers of elderly parents in the home situation. While 72 per cent of carers described their parent as being long term ill or disabled, only half of the 58 carers involved said that their elderly parent needed 'a lot of' help (compared with 'some' or 'a little' help). However, only around half whose parent needed a lot of help were actually using any support services (as were around a third of those whose parents were said to need less help).

The main services used were home help (14 per cent of carers), the district nursing service (10 per cent of carers), meals-on-wheels (9 per cent of carers) and respite care or day care (8 per cent of carers). A few had used particular services provided by hospital departments or other specialist institutions for things such as wheelchair loans or guide dogs for the blind. All users of services said they were 'very satisfied' or 'satisfied' with the services used and nobody registered any direct complaints.

Since more than half of the carers did not use services, they were asked if they felt they had service needs that were not being met. However, only seven carers (12 per cent) said 'yes'. Three of them needed respite or day care services and four needed either home help, home nursing, home maintenance or chauffeur services.

The general lack of reliance on public services suggests that most later life carers were relying on their own or their family's resources to carry out their caring role, as was also found in Howe and Schofield's (1996) carers study. This raises the question of what effect the task of caring for elderly parents might have on the health or happiness of the helpers concerned, particularly those caring for long-term disabled parents.

Differential outcomes for carers and non-carers

Howe and Schofield's (1996) carers study found that carers had less satisfactory physical health, emotional wellbeing and lower life satisfaction than non-carers. To test for these outcomes, satisfaction with health, time management, main personal relationships and employment were explored as indicators of carer wellbeing. The respondents who were caring for elderly parents were compared with non-carers to see if there were disparate outcomes for the two groups.

Health and stress

There was no discernable relationship between the self-ratings of 'overall health' (rated as 'excellent', 'good', 'fair' or 'poor' health) and the fact that respondents were helping or caring for elderly parents. Furthermore, those either providing general help to elderly parents or acting as elder carers were no more likely than non-helpers to report that they or their partner suffered from long-term health problems or disabilities, which applied to around 45 per cent of respondents.

However, in a more psychological vein, two other questions were asked to probe for perceptions of 'time squeeze' and constraints on the freedom to make future life decisions, both of which might cause psychological stress.

All respondents were asked the following two questions: 'Is the amount of time you have for yourself enough, too much or not enough?' 'Have, or will, the needs of your parents or parents-in-law influence your decisions about later life?'

A significant effect was found regarding the demands placed upon the time of people caring for elderly parents. Nearly half of them agreed that they did not have 'enough time' for themselves, compared with only a quarter of non-carers. Carers were also more likely to agree that the needs of their elderly parents had influenced, or would influence, their decisions about their own retirement and ageing than were non-carers (44 per cent compared with 30 per cent). So, even though overall physical health did not seem to be a problem, there was certainly more of a feeling among carers than others that their time was not their own and that their future plans would have to take account of the needs of parents. This suggests some evidence for potential psychological stress among the elder carers.

Marital and work satisfaction

To see whether helping elderly parents or being a main carer for an elderly parent substantially affected respondents' satisfaction with other key aspects of wellbeing, the mean scores of those providing assistance and those not providing assistance were compared on marital satisfaction and work satisfaction measures. Analysis involved women only, because they were more likely to give practical aid to parents and were more likely to be the main carer of a parent than were men, and the main issue raised in the literature is the possible effect on women of gender inequality in caring for the elderly.

For women in a registered or de facto marriage, a measure representing feelings of marital 'togetherness' was created. This scale was derived from four items about marital relationships. On a scale of 1 to 4, where 1 = often, 2 = sometimes, 3 = rarely, 4 = never, the women were asked to indicate, how often, as a couple: 'Do you spend time doing things together?' 'Do you talk over things that concern you?' 'Have you considered ending your relationship?' 'Have you regretted the relationship?'

The combined measure also had a scale of 1 to 4, where 1 = a positive or 'high' score and 4 = a negative or 'low' score, and the female respondents generally scored very positively (average = 1.4), meaning most felt a sense of togetherness and security in their marital relationship.

If a strain was being placed on the relationship by helping or caring for elderly parents, it was expected that this might be reflected in lower levels of marital satisfaction. However, there was no significant difference in average ratings on this measure between those who were giving aid or caring for an elderly parent and those who were not. In fact, those giving some practical assistance to parents had a slightly more positive score (1.34) on this marital togetherness measure than those not doing so (1.52; f = 3.74; p = .05).

Thus, if this is seen as an indicator of emotional wellbeing, women who were carers did not fare worse than those without responsibility for an elderly parent. It may be that a secure personal relationship means the women feel better able to give their time to help parents. In addition, some women might be helping elderly parents together with their spouse, as a couple, thus strengthening their feelings of togetherness and shared purpose in their marital relationship.

As seen in Table 16, women's caring responsibilities do not recede despite the increasing acknowledgement of women's contribution to the paid workforce. Thus the 'double duty' of work and family obligation may lead some women to hold less positive attitudes about their job, and this may particularly be the case when responsibility for an elderly parent is added to general family responsibilities.

For women in the workforce, a scale was constructed to assess the effect on job satisfaction of devoting time and energy to helping or caring for elderly parents. The three scaled items used (where 1 = very satisfied to 5 = very dissatisfied) were the following questions: 'How satisfied are you with your rate of pay? your job security? the job you have?'

The overall scale was 1 = 'high' work satisfaction and 5 = 'low' satisfaction. On all three measures, the average ratings of work satisfaction were quite positive at around 2.1, and these average scores did not differ between women who were main carers or helpers of elderly parents and those who were not. Therefore, there is no indication that women in paid work who had responsibilities for elderly parents were less satisfied with their jobs than were those who did not report such responsibilities. Had there been less satisfaction, it may have suggested less choice, security or reward for elder helpers or carers compared with other women, but this was not the case.

An earlier Australian study of workers with family responsibilities found that 17 per cent of workers aged in their forties took days off to care for elderly parents, as did 13 per cent of workers in their fifties (VandenHeuvel 1993). Although no sex difference was found in the mean number of days taken for elder care, male employees were more likely than females to take time off to deal with 'administrative, business or legal tasks' on behalf of aged parents, while women were more likely to take time off for personal care of elders (VandenHeuvel 1993). Nevertheless, the Later Life Families Study has found that the bulk of family care is undertaken by women, who are less likely than men to be employed full-time. Many women may decide to take fewer hours of paid work per week to allow them to juggle work and caring responsibilities, but this can also be at the expense of future financial security due to lost accrual of superannuation benefits (Wolcott and Glezer 1995).

Multigenerational responsibilities

Around half of the female respondents caring for elderly parents were in paid employment, and there was also some evidence of psychological strain and 'time squeeze' among elder carers. In addition to the double role presented by employment, respondents had other commitments to family members. Many were still providing a home for children, or supporting independent adult children in some way. Many were involved with grandchildren and some were also the main carer for a spouse or other ill or disabled family members. Bergquist, Greenberg and Klaum (1993) found that care for elderly parents was often a problem for those in their fifties because of competing priorities: wanting simultaneously to care for an elderly parent (or parents), maintain or improve relationships and exchange with adult children and grandchildren, and regenerate or improve their relationship with their partner in later life.

In the United States, it has been estimated that 40 per cent of adults aged between 45 and 54 years have at least one child aged 19 years or younger and a living parent aged 65 years or older (Stenberg-Nichols and Junk 1997). In the Later Life Families Study, 36 per cent of the 345 respondents with an elderly parent (or partner's parent) also had at least one child living at home, as did a third of those with an ill or disabled parent and 22 per cent of those who were the main carer for an elderly parent. These proportions are quite substantial, but, as we have seen above, it is by no means only those children who still live with parents who are relying on them for assistance and support. To define dependency of children as co-residence, therefore, does not present the whole picture.

Thus, when considering the burdens on respondents, we will examine not only respondents' own situations and those of their elders, but also their responsibilities toward children still living at home, children who have left home and grandchildren. Therefore, the following analysis gives a profile of the multi-generational responsibilities being undertaken by those respondents who were in fact providing financial or practical support to an elderly parent, or who were acting as the main carer of an elderly parent. The sphere of emotional support is omitted here, partly because it was ubiquitous, but also to focus on the more practical tasks involved in assuming responsibility for elderly parents.

First, analysis was conducted to explore the respondents' own health needs and likely responsibilities for other family members. Table 17 shows the percentage of elder carers or helpers who themselves had health problems or whose spouse or children had health problems. Around a fifth of those who were helping or caring for elderly parents also had health problems themselves or had a spouse or child with such problems. The proportions who considered themselves the main carer for either a spouse or other relative or friend were smaller, but it is interesting to note that the people who designated themselves 'main carers' for elderly parents were nearly twice as likely as those giving more general support to also report being the main carer for someone else (see Table 17).

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Although helpers or carers of elderly parents were not more likely than non-carers to have health problems overall, those who were shouldering responsibility for elderly parents as well as suffering ongoing health problems themselves clearly had an extra burden when compared with those who either had no parents alive or did not assist or care for their parents. Up to a fifth of carers had a combination of their own and/or their spouse's health problems or disabilities, as well as responsibility for elderly parents, which has been shown to predispose some to psychological stress. The same applies to nearly a fifth of respondents who had a child suffering from ill health. Probably the biggest load would be for the smaller group who were main carers for a spouse or other person in addition to shouldering elder care.

Table 18 shows the profile of helpers and carers of elderly people according to the type of general assistance they were also giving to children and grandchildren. As can be seen, over a third of those giving financial or general practical help to elderly parents still had at least one of their own children living at home. In fact, of those giving practical help to elderly parents, twenty-nine (12 per cent) still had children at school. A fifth of elder carers also still had a child living at home, which may translate to extra domestic labour as well as considerable responsibilities toward their elderly parent or parents.

* Those who looked after grandchildren at least once a fortnight 
** That is "often" or "occasional" care of grandchildren while parents were working. 
Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

While few respondents were actually caring for an ill or disabled child (five people), Table 18 shows that a great deal of help was being given to adult children by parents who were also elder carers or helpers. Furthermore, some of these adult children were being helped more than their siblings (this applied to 30 per cent overall) due to their health problems.

Of the types of support to elderly parents shown in Table 18, being the main carer was most strongly associated with providing emotional and practical assistance to adult children. A higher proportion of main carers also cared for grandchildren, but then more of this 'carer' group had grandchildren (62 per cent, compared with 44 and 54 per cent). Three-quarters of these elder carers were women, whom we have seen are also more likely than men to be highly involved with children and grandchildren. Indeed, having at least one daughter has been described as the key to elderly people gaining support (Stenberg-Nichols and Junk 1997). These authors have also noted that a common complaint among care-giving daughters is the lack of support forthcoming from 'male adult children of care recipients' (that is, from their brothers or brothers-in-law).

Discussion

Overall, it appears that for some people - principally women - there is indeed a 'culture of caring' where two generations are catered for at once by the same individual. Despite this finding, when aggregate or summary measures of 'family relations' and 'family responsibility' were tested, they related to the nature of the intergenerational relationship rather than conforming to an overall 'multigenerational' family measure. That is, there were independent underlying factors that contributed to measures of relations with adult children and elderly parents respectively (see the appendix for full details of the aggregate measures constructed and tested).

Therefore, certain people are more involved with either the younger or older generation. Furthermore, even respondents who had adult children and grandchildren and elderly parents tended not to have an evenly distributed level of intergenerational involvement or responsibility across family generations. Variations in the age, life stage and needs of the relatives in question, as well as idiosyncratic personal relationships, would contribute to this outcome. For example, as was shown above, often more help is given to adult children who are the youngest, have more vulnerable personalities or are in particular need, such as students or lone parents. Additional care is also needed by both children and parents with ill-health or a disability.

Nevertheless, people who help ageing parents while still caring for dependent or semi dependent children appear to form the backbone of family care. They have been referred to as the 'sandwich' generation because they are sandwiched between simultaneous demands from both younger and older generations (Stenberg-Nichols and Junk 1997).

Such dual roles and demands have been seen in the United States as problematic mainly for women because they experience 'more health stress' as well as 'significantly higher absenteeism', and 'difficulty combining work and family' (Neal 1993). This appears to be consistent with the present findings, as nearly half of the elder carers had either personal health difficulties or a partner with long-term health troubles. The implication is that the difficulties faced by these people are exacerbated by their dual or treble loads of family responsibility. However, no causality can be implied, as we do not know whether or not a caregiver's ill health is a result of their helping role. Nevertheless, we can surmise that their own ill health adds to the overall strain, particularly as they tend to feel they do not have enough time for themselves, and that they are constrained in their own future lifestyle choices by responsibilities towards elderly parents.

The ramifications for employment opportunities are a little mixed for the respondents, partly because half of them, being over 60 years of age, were not in the paid workforce. Nevertheless, half of the women who were main carers for elderly parents were in paid employment, and recent Australian analysis showed that female employees were more likely than men to suggest the need for part-time work, job-sharing opportunities or special leave, as they perceived the 'juggling' of work-family roles to be more problematic than did men (VandenHeuvel 1993). This did not occur just among those workers in their forties and fifties, but also among younger age groups, who had dual responsibilities toward younger and older generational family members.

Although women in VandenHeuvel's study did not have higher absenteeism rates than men, and employed female carers in the Later Life Families Study were not dissatisfied with pay, security and the nature of their job, the findings from VandenHeuvel's study clearly suggest that job flexibility rather than conditions or security might be more relevant to older working women who take responsibility for elderly parents.

At each stage of the family life cycle, intergenerational dynamics take on a different perspective. To complete the picture of family exchange, assistance from elderly parents and the ways in which they influence the decision-making and future plans of people in later life are now considered.

Help received from elderly parents

First, respondents with elderly parents were asked, again generally: 'Do your parents provide you with emotional support or advice?' 'Do your parents help you financially if you need it?' 'Do your parents provide you with any practical support (doing things for you)?'

Of respondents with elderly parents, the percentages who received financial, practical and/or emotional support from the parents are shown in Table 19. There was less assistance from elderly parents to respondents than the other way around, which is not surprising considering that all the respondents were at least 50 years old, so their parents' ages ranged from the seventies upward (indeed, some people had elderly parents in their nineties).

Source: Later Life Families Study, 1996, Australian Institute of Family Studies, Melbourne.

Being in Laslett's (1992) 'fourth age' stage (mostly over 75 years) of physical decline, quite a high proportion of respondents' elderly parents had long-term illnesses or disabilities. This is consistent with them being unable to give a great deal of practical assistance to the respondents. Nevertheless, nearly 30 per cent of respondents with elderly parents indicated that they received practical help from them, which is a strong testimony to intergenerational sharing when considering the age of such parents.

Some authors are adamant that ageing parents should not be seen as a major drain on wider family resources or on taxpayers' money because middle-aged people and their children 'are more likely to get money from ageing parents than to give it to them' (Stenberg-Nichols and Junk 1997). However, the respondents in this survey, particularly those over 60 years old, can themselves be seen as ageing, and while there is certainly some financial assistance coming from their elderly parents to them, there is more in the opposite direction. The age group of respondents' elderly parents suggests that they incur heavy health care expenses and the historical context of their younger lives means that many will be on fixed incomes or pensions. Thus, many may not have the means to help younger family members financially.

By way of intergenerational comparison, it is interesting to note that the same proportion of respondents (around a fifth) received financial assistance from elderly parents as from adult children, while more than twice the proportion of respondents were giving financial assistance to children (70 per cent) than to elderly parents (33 per cent). The nature of financial assistance on the part of children and elders may also differ because the latter may give indirect assistance to respondents via their advice and counselling on the economic aspects of retirement and ageing.

Indeed, over 40 per cent of respondents received advice or emotional support from their elderly parents. This seems to be the most important form of support given by the most senior generation to the next and signifies the worth of learning from one's elders. It also attests to the continuation of the parent-child emotional bond throughout life.

Influence of elderly parents on future plans

As previously noted by Wolcott (1997), elderly parents were still influencing the lives and future plans of the respondents. Interestingly, the ways in which respondents felt their future plans or expectations were influenced by elderly parents did not seem to differ according to whether they were carers or non-carers.

One main influence was in showing the importance of financial planning for retirement and old age, for 'securing your future, setting up to be comfortable'. This could include expectations of inheritance; for example, 'My mother-in-law said she'll leave us the house and all the furniture, plus a portion of her estate.' However, it can also be by way of negative example: 'They never had anything at all, so I just felt that I wanted to be better off than they were.'

Another influence was the example shown by elderly parents of how to make the most out of older age. Burns (1996) found that women in particular felt their mothers were their main 'blueprints' for ageing. Our respondents were most likely to see their mothers as active, competent and independent - even well into their eighties. Some mothers were held up as models for ageing; for example, 'Both my mother and mother in-law have enormous independence of spirit...They have been able to do the things they want as they grew older, more than when young. '

Linked with the increased awareness among respondents of the future is the importance of keeping active and 'the need to look after health in later life [so as] to enjoy personal pursuits while you are able to'. Such views mirror Burns' findings that women in later middle age are very conscious of maintaining good health as well as having a positive mental attitude toward ageing.

Another major influence noted by Wolcott (1997) is the need that elderly parents have, or might have, for personal care. We have seen above that this applies to a minority of the population. Although estimates vary, probably less than 10 cent of people will become elder carers. Nevertheless, the possibility exists for anyone with older parents still alive, and this may include the possibility of having an elderly parent co-residing. According to one respondent, 'If they become dependent my life will be altered, [but] my father might have to come and live with me.' Considerations of parents' needs could also involve having to live near parents, as well as having to restrict future travel plans, as appeared to be the case for some respondents who were already caring for an elderly parent.

Despite the acknowledgement of parental needs, close contact (particularly co residence) has been shown to cause resentment among carers of elderly parents (Marsden and Abrams 1987). One respondent commented that having a parent living with her adversely affected her marriage and 'put me off the idea of living with my children when I'm older'. Although both carers and non-carers were equally likely to express a strong commitment to the notion of filial obligations towards ageing parents ('I've always been brought up to take parents' needs into account in this regard'), some people disapproved of shouldering responsibility for parents: 'It was left to the children to look after them [the parents]. I don't want my children to be left in that situation - to be obligated to look after me.'

Some elderly parents were held up as examples of what not to do and some were certainly considered to be a burden. However, overall, most respondents appeared to be committed to helping or caring for elderly parents, either now or in the future. Indeed, many saw elderly parents in a very favourable light, as good role models of successful ageing - being independent, being good financial planners and having a positive outlook on life.

Furthermore, most respondents' elderly parents were living independently. Hagestad (1987) notes that independence is easier for elderly women than men to maintain because of their links to the family network. Women often invest time and energy in maintaining relationships and assisting family members, and reap the benefits of this in old age by receiving assistance from family members, particularly adult children. Indeed, according to Laslett (1992), most elderly people are not isolated from kin and, even though they live separately, they enjoy 'intimacy at a distance'. Nevertheless, the element of gender cannot be ignored, as most isolated elderly people tend to be male.

In general, the analysis of exchange between the older generations paints a picture of older people - from the 50-year-old respondents through to octogenarian parents - providing a rich source of support for younger generations. Not surprisingly, Bergquist, Greenberg and Klaum (1993) have noted the psychological trauma experienced by the late middle aged when their elderly parents die. Not only are they suddenly the 'senior' generation in the family, but they also lose the advice, support or approval of their parents, upon which they may have relied all their lives.

Findings from the Later Life Families Study demonstrate that there is a great deal of help flowing between generations within the family. Such findings suggest a norm of reciprocity between family generations in Australia. However, there appears to be political impetus for ongoing movement away from a wider societal or public form of intergenerational support and toward a stronger reliance on familial support. The factors shown in the Later Life Families Study to determine the level and nature of intergenerational exchange within families can contribute to a better understanding of how such ongoing social change may impact upon the current situation.

The socio-demographic and economic resource bases of donors of intergenerational family aid reflect who is most able to take on any additional responsibilities resulting from a shift in the balance of public versus family support resources. The resources of donors and receivers in family exchange are affected by family structure and geographic distribution, finances and workforce attachments, cultural background and age. The receipt of aid depends upon who most needs support services.

Both genealogical and generational relationships determine exchange. The respondents were more likely to help their children than to help their own parents. The most widely exchanged intergenerational aid was emotional support or advice, which attests to the natural closeness of members of the family of origin and the tendency by many to turn to a family member first in times of emotional crisis or difficult decision making. This was particularly noticeable for respondents with a non-English-speaking background, who were more likely to confide in their adult children and expected, as well as received, more help from them.

In contrast with the majority of respondents who had two-way exchange with adult children, those with no children (8 per cent of the sample) will have no recourse to filial supports as they grow older. Of interest is the fact that childless people were not more likely than those with children to be the main carer for an elderly parent. Therefore, there was no suggestion of the stereotypical 'spinster' carer here; rather, there was evidence of an additional load for the group who already had filial responsibilities.

Help from family members was more likely to be received by those living with, or very near, the relative in question, and by those who were in frequent contact. The patterns of proximity, contact and co-residence were consistent with the comparative amounts of help given to children and elderly parents. Thus, problems may ensue for people with no relatives nearby - those who have perhaps moved for personal or work reasons, as well as many migrants. The respondents helped their children above all others, but the children also tended to be living closer than elderly parents, and were also much more likely to be residing with respondents than were the senior generation.

Where adult children were in financial difficulty - just 'starting out' in life, or struggling as students or single parents - they were even more likely to be helped by later life parents. The present social climate makes it harder for young adults to be financially independent as there is an ever-increasing need for higher formal qualifications for the same job. At the same time, there are ever-receding 'goalposts' regarding what constitutes adulthood. Financial responsibility for adult children has crept from 18 to 22 to 25 years, and it is parents in the mid-life and later life years who must keep providing assistance to these young adults.

If the respondents were themselves in a good financial situation, with a good income level - which often related to still being in paid work - they were better able to provide such support. However, today's economic climate may also make this increasingly difficult for the later life parents. A growing number of people are being encouraged to take early retirement or are in early retirement after being made redundant or being long-term unemployed. Therefore, their financial resource base as donors in family exchange may be unequal to the potential demands of younger- or older-generation family members in the years to come. Therefore, in families where members of the younger generation are in financial need, for instance due to unemployment or recent marital separation, but mid-life or later life parents are also unemployed or on low incomes, it cannot be assumed that the latter are in a position to support their adult offspring.

Although aid generally flowed from older to younger generations, the middle generation - those in their fifties and sixties - provided most of the assistance in both directions. However, there was apparently better financial security and more independence among the elderly than among the younger generation. Although elderly parents were much more likely than respondents to have health problems or disabilities, respondents gave much more assistance overall to their adult children than to their elderly parents, which is inconsistent with popular stereotypes of the aged as dependent and burdensome.

The age group and employment status of respondents affected the nature of assistance given. Those who were helping children or elderly parents financially were more likely to be in paid work and in their fifties; younger grandmothers tended to care for younger and more demanding grandchildren more often; two-thirds of the elder carers were in their fifties, and half were still employed. Thus, those in their fifties were the main 'donors' in intergenerational family exchanges, and they also received less help themselves from children than did respondents who were over 60 years old.

Other personal characteristics important to intergenerational exchange were the marital history, marital status and gender of donors or receivers. Later life mothers who lived alone, particularly widows, gained more help from their adult children than did mothers living with a spouse or partner. This seems linked to both the practical and emotional needs of older women living on their own. For fathers, however, having experienced marital separation resulted in less support being received from adult children, regardless of present marital status. If ongoing relationships with fathers suffer after divorce, adult children may feel estranged from them and less disposed to offering support. In contrast, when adult children had experienced marital separation, more help was often forthcoming from parents, as this was seen to be a time of greater need as adult children had financial and child care difficulties.

Any increased expectations for families to shoulder greater social welfare responsibilities impacts upon older women much more than men. Increased physical or psychological health risks, threats to personal and career choice, and limitations on the accrual of superannuation entitlements are all symptoms of gender inequality in intergenerational caring.

Therefore, it is clear that, although families are not generally shirking their responsibility for members' welfare, some people are not in a situation to give aid or receive aid from family - particularly divorced fathers, the childless, those whose families live too far away, recent migrants, those with chronic health problems, and people on low incomes. On the other hand, some people, particularly female relatives, may be unfairly overburdened. Furthermore, while some people expected their children to look after them in old age, many did not.

Since so much is already being done by families, how can society expect more? For example, current encouragement toward self-funding of retirement may pose a problem for family carers, as those who are not in paid work or are only working part time due to family responsibilities may be unable to accrue enough superannuation benefits to support their own retirement. This suggests a logical tendency for carers, who are mainly women, to eschew caring responsibilities in the future in order to prepare for their own independence in old age. Thus, the notion of an intergenerational contract of mutual obligation may become less rather than more prominent within families into the twenty-first century.

Furthermore, while public acknowledgement of the social and economic contribution of families is clearly justified, assumptions of limitless family resources to support and care for the younger and older generations are ill-founded for a number of reasons. Individual family members also have to care for themselves, which may mean commitments to paid work over family work. Family households are becoming smaller, thus reducing internal family resources, while the disruption of divorce can weaken ties between family households and wider kin networks, thus reducing wider family resources. In such a context, publicly provided support services may become more rather than less important to the generation caught in the middle who give most support to younger and older generations.

Aggregate measures were constructed and tested for two indicators of family cohesion: subjective perceptions about family relations and an overall level or index of family responsibility of the respondents.

To overcome problems with scale construction, three separate measures of respondents' perceptions of family relations were used, as well as two indexes to indicate the amount of responsibility taken by respondents, (1) for adult children or (2) for elderly parents.

Family relations or 'salience' measures

The first measures were of perceptions of the importance of family in everyday life. To do this, the following variables were tested: frequent contact with adult children or elderly parents; exchange of some assistance with them; satisfaction of relationships, or closeness with them; the importance of family; the importance of grandparenting.

Three measures of perceptions about family relations were:

• Importance of contact with family members (very important/somewhat important/not at all important)
• Satisfaction with the relationship with adult children (very satisfied/satisfied/mixed feelings/dissatisfied/very dissatisfied)
• Feelings of closeness to mother and to father (very close/close/not close)

First, the analysis probed the interrelationship between these three perceptual measures and between them and the measures of proximity and contact with adult children and older parents.

The importance of contact with family members was moderately correlated with feeling close to mother (r = .234; p = .001), being satisfied with the relationship with children (r = .227; p = .001) and (for grandparents) the importance of grandparenting (r = .336; p = .001). Satisfaction with the relationship with children was moderately correlated with frequency of contact with them (r = .277; p = .001).

Feelings of closeness to mother were strongly correlated with feelings of closeness to father (r = .56; p = .001) and not quite so strongly with frequency of contact with mother (r = .36; p = .001). Feelings of closeness to father were strongly correlated with frequency of contact with him (r = .51; p = .001). This implies that people could feel very close to their mother without having a great deal of contact with her, but frequent contact seemed necessary for feeling very close to father.

Second, the relationship between these three perceptual measures and the socio demographic characteristics of respondents was examined. Although there were no socio-demographic differences for feelings of closeness to parents, women were significantly more satisfied with their relationships with adult children (chisq 17.23; p .002), as were the older group, aged 60 and over (chisq 17.09; p .002), married or widowed respondents (chisq 16.54; p .03) and those not in the workforce (chisq 10.65; p .03). Similarly, women over 60, married or widowed, those with lower incomes and those not in the workforce reported contact with family to be more important than their comparison groups.

Family responsibility index

The second measure sought to construct an index of 'family responsibility' using the following variables: number of children still living at home; provision of help to adult children or elderly parents; frequency of looking after grandchildren; having an adult child or elderly parent with a chronic illness or disability; being the main carer for a family member.

Specifically, in order to construct an index of the amount of responsibility taken by respondents, two sets of variables were considered, first with regard to adult children, second with regard to elderly parents.

Respondents scored zero if they had no adult children and one point for each of the following elements of responsibility for the younger generation:

• For each adult child still living at home
• For giving financial assistance to adult children
• For giving practical assistance to adult children
• For giving emotional support or advice to adult children
• For looking after grandchildren on a regular basis (at least monthly)
• For having a child with a long-term illness or disability
• For being a main carer to that child (only seventeen cases)
• For that child requiring a lot of care
• Respondents scored zero if they had no elderly parents and then one point for each of the following elements of responsibility for the older generation:
• For each parent or partner's parent living with the respondent
• For giving financial assistance to elderly parents
• For giving practical assistance to elderly parents
• For giving emotional support or advice to elderly parents
• For having an elderly parent with a long-term illness or disability
• For being a main carer to elderly parents (58 cases, or 8 per cent)
• For elderly parents requiring a lot of care

Again the analysis probed the interrelationship between these scales and the three perceptual 'family salience' measures, as well as proximity and contact with adult children and older parents.

The index of responsibility toward elderly parents correlated moderately with feeling close to mother (r = .31; p = .001) and feeling close to father (r = .26; p = .02) and also with mother living nearer (r = .285; p = .001). The correlation was most strong with frequent contact with mother (r = .424; p = .001). However, the index of responsibility toward adult children did not correlate with amount of contact with children or with feelings of satisfaction with relationships with children.

Finally, the relationship between these responsibility indexes and the socio demographic characteristics of respondents was examined.

There were no significant differences in mean scores on the index of responsibility toward elderly parents according to the following socio-demographic variables: respondent's sex, age group (fifties or 60 plus), non-English-speaking background, level of education, income bracket, marital status or workforce participation.

However, there were differences between certain socio-demographic groups for mean scores on the index of responsibility toward adult children: women scored higher than men (mean of 3.22 compared with 2.93; Eta = .11; p = .004) and respondents in their fifties scored higher than those aged 60 years and over (mean of 3.23 compared with 2.94; Eta = .11; p = .003).

Concluding remarks

The aggregate measures of 'family salience' and 'family responsibility' related to the nature of the intergenerational relationship rather than conforming to an overall 'multigenerational' family measure. That is, there were independent underlying factors for relations with adult children and for relations with elderly parents. Therefore, certain people were more involved with either the younger or older generation. This finding could be partly due to the fact that respondents who did not have adult children, did not have grandchildren, or did not have elderly parents could not have scores on measures of all three types of relations.

Furthermore, even respondents who had adult children and grandchildren and elderly parents tended not to have an evenly distributed level of intergenerational involvement or responsibility across family generations. Variation in the age, life stage and needs of the relatives in question, as well as idiosyncratic personal relationships, would contribute to this outcome. For example, as was shown above, often more help was given to adult children who were the youngest, had more vulnerable personalities or were in particular need, such as students or lone parents. Additional care was also needed by both children or parents with ill health or a disability.

It was also apparent that many characteristics combined to paint different scenarios of family salience. For the respondents, being older was related to being female, as was lower income, not being in the workforce and being a widow.

• Table 1 Personal characteristics of Later Life Families sample
• Table 2 Family characteristics of Later Life Families sample
• Table 3 Proximity of later life respondents' adult children and parents
• Table 4 Contact between respondents and non-resident children or elderly parents
• Table 5 Later life parents who received financial, practical and/or emotional support from adult children
• Table 6 Aid given to later life parents by adult children according to resident status of children
• Table 7 Characteristics of later life parents by help received from non-resident adult children aged 18 years and over
• Table 8 Characteristics of later life parents by help expected from non-resident adult children aged 18 years and over
• Table 9 Later life parents' satisfaction with relationships by proportions receiving or expecting aid from non- resident adult children
• Table 10 Later life parents who gave financial, practical and emotional support to adult children, by the parents' characteristics
• Table 11 Characteristics of children aged 18 years and over who received, or did not receive, more parental help than their adult siblings
• Table 12 How often grandparents saw or looked after at least one grandchild, by the sex of the grandparent
• Table 13 Grandparents' partnering status by how often they saw grandchildren
• Table 14 How often grandparents saw and cared for grandchildren, by residential proximity
• Table 15 Respondents' characteristics by financial, practical and emotional support given to elderly parents
• Table 16 Employment commitments of later life carers of elderly parents
• Table 17 Health problems and additional caring responsibilities of respondents by type of aid they gave to an elderly parent
• Table 18 Help given by later life parents to children and grandchildren by type of assistance given to elderly parents
• Table 19 Respondents who received financial, practical and/or emotional support from elderly parents