Canadian Journal on Aging / La Revue canadienne ...

Canadian Journal on Aging / La Revue canadienne du vieillissement http://journals.cambridge.org/CJG Additional services for Canadian

Journal on Aging / La Revue canadienne du vieillissement: Email alerts: Click here Subscriptions: Click here Commercial reprints: Click here Terms of use : Click here

Formal Assistance Among Dutch Older Adults: An Examination of the Gendered Nature of Marital History Andrew V. Wister and Pearl A. Dykstra Canadian Journal on Aging / La Revue canadienne du vieillissement / Volume 19 / Issue 04 / January 2000, pp 508 - 535 DOI: 10.1017/S0714980800012496, Published online: 29 November 2010

Link to this article: http://journals.cambridge.org/abstract_S0714980800012496 How to cite this article: Andrew V. Wister and Pearl A. Dykstra (2000). Formal Assistance Among Dutch Older Adults: An Examination of the Gendered Nature of Marital History. Canadian Journal on Aging / La Revue canadienne du vieillissement, 19, pp 508-535 doi:10.1017/S0714980800012496 Request Permissions : Click here

Downloaded from http://journals.cambridge.org/CJG, IP address: 130.115.93.121 on 03 Feb 2016

Formal Assistance Among Dutch Older Adults: An Examination of the Gendered Nature of Marital History Andrew V. Wister, Simon Fraser University and Pearl A. Dykstra, Netherlands Interdisciplinary Demographic Institute* S'inspirant d'une theorie du cours de la vie, cet article examine les differences entre les sexes reliees a l'aide formelle accordee a des aine(e)s hollandais(e)s non autonomes sur le plan fonctionnel. On a examine cinq groupes distincts quant a la situation matrimoniale: gens maries une seule fois, celibataires, divorces (non remaries), veufs ou veuves (non remaries) et gens remaries. On a effectue des analyses de regression logistique hierarchique de chaque categorie matrimoniale pour verifier les hypotheses concernant les rapports entre les sexes et trois ensembles de variables: 1) mesure de l'age et de l'etat de sante; 2) mesure de la situation socio-economique et 3) mesure du reseau social. Les resultats font ressortir une configuration par sexe de l'aide formelle utilisee chez les gens maries une fois, qui n'ont jamais ete maries et chez les veufs ou veuves. Conformement aux autres etudes, les femmes agees qui ont ete mariees une seule fois sont environ trois fois plus enclines a recevoir de l'aide formelle que les hommes, un ecart qui demeure soutenu meme apres un examen statistique d'autres facteurs, notamment la fragilite du conjoint. Cependant, on remarque aussi que les femmes qui n'ont jamais ete mariees sont 66 pour cent moins enclines a recevoir des soins formels que les hommes qui n'ont jamais ete maries, ce qui pourrait etre un indice des preferences en ce qui concerne l'utilisation de soins formels. Chez les veufs et les veuves, on constate que les homines dont la sante est plus fragile auront plus tendance a recevoir des soins formels que leurs homologues feminins, ce qui laisse * This paper was written while the first author was a visiting scholar at the Netherlands Interdisciplinary Demographic Institute. The data for this paper were collected as part of a research program entitled "Living Arrangements and Social Networks of Older Adults," conducted in the Departments of Sociology and Social Research Methodology of the Vrije Universiteit in Amsterdam, and the Netherlands Interdisciplinary Demographic Institute in the Hague. The program is funded hy NESTOR, the Netherlands Program for Research on Aging. Key Words: Formal Assistance, Gender, Marital History. Mots cles: Aide structuree, genre, situation matrimoniale. Manuscript received August 21, 1998; manuscrit recu le 21 aout 1998. Manuscript accepted December 1, 1999; manuscrit accepts le 1 decembre 1999. Requests for offprints should be sent to:/ Les demandes de tires-a-part doivent etre adressees a: Andrew V. Wister Gerontology Program Simon Fraser University at Harbour Centre Vancouver, BC V6B 5K3 ([email protected]) S08

Canadian Journal on Aging / La Revue canadienne du vieillissement / Vol. 19 no. 4 2000, 508-535

Gender, Marital History and Formal Assistance La Revue canadienne du vieillissement 509

croire a une configuration differente du comportement a l'egard de la recherche de l'aide et de la vulnerability sociale. On remarque d'autres differences de variables chez les groupes, en fonction de leur situation matrimoniale, et elles sont interpretees dans une perspective de cycle de vie. ABSTRACT

Drawing from life course theory, this article examines gender differences in formal assistance among functionally dependent Dutch older persons withinfivedistinct marital history groups - first-married, never-married, divorced (not remarried), widowed (not remarried) and the remarried. Hierarchical logistic regression analyses are performed for each of the marital history groups to test hypotheses regarding the interrelationships among gender and three sets of variables: 1) measures of age and functional health; 2) measures of socio-economic status; and 3) measures of the social network. The results indicate gendered patterns of formal help use among the first-married, never-married and widowed. Consistent with other studies, older first-married women are approximately three times more likely to receive formal help than are men, a difference that remains robust after statistically controlling for other factors, including frailty of spouse. However, we alsofindthat never-married women are about one-third as likely to use formal help than are never-married men, which may be reflective of different preferences regarding formal service use. Among the widowed, we find that men with poorer functional health are more likely to receive formal help than are their female counterparts, suggesting contrasting patterns of help-seeking behaviour and social vulnerability. Additional differences are observed among the marital history groups in terms of the other independent variables, which are also interpreted from a life course perspective.

Introduction The majority of social support in old age travels through informal channels. There is ample evidence, for example, that spousal support comprises the largest segment of informal support in old age among married persons, and that adult children are primary sources of emotional and instrumental support for widowed parents and provide supplementary assistance in situations where parents are still alive. However, the long term consequences of increased longevity, and a rise in rates of singlehood, divorce, and remarriage, suggest a change in the traditional pathways of support flowing to older persons (Dooghe, 1992; Eggebeen, 1992; Kobrin, 1976; Watkins, Bongaarts, & Menken, 1987). Thus, the interrelationship between the informal and the formal care systems may be becoming more complex than suggested in early conceptualizations, such as the hierarchical compensatory and the task specificity models (see Chappell, 1992, for review). According to the first model, formal help is called in as the last resort. When the needs of the older adult exceed the informal network's capabilities, supplementary care is sought in formal services. The model of task specificity proposes that different types of support providers can

510 Canadian Journal on Aging Vol. 19 no. 4 2000

Andrew V. Wister & Pearl A. Dykstra

best perform the tasks that are consonant with the structures. Formal helpers can better handle care needs that are routine, predictable and require some level of technical expertise. Informal helpers can better perform tasks that require proximity and flexibility. In an attempt to assess the adequacy of the two models, researchers have investigated the effects of various social network characteristics on patterns of formal support (e.g., Connidis & McMullin, 1994; Dykstra, 1995; Mutchler & Bullers, 1994; Penning, 1990; Stollar & Cutler, 1993). In light of the prospect of increasing variability and complexity in the primary informal support system, it is not surprising that the influence of family change and continuity remains a dominant theme in the formal-informal support literature (Brubaker, 1990). An important specific area of study, and one that guides this paper, is the role of marital history in affecting formal assistance with daily living in old age, and the gendered nature of this experience. Yet, the ways in which marital status has been incorporated into analytic strategies have often underplayed, or even masked, its importance. Most research has been based on current status only, combining those in first and subsequent marriages, amalgamating non-married groups, and sometimes excluding altogether marital status groups of less frequent occurrence (Karney & Bradbury, 1995). Since marital history pertains to marital experiences across the life course, using current marital status or grouping marital statuses together may obscure important past events. Few studies include non-spousal partners for the non-married groups, that is, persons living common-law or those who report having a partner living outside of the household (Jong Gierveld, 1999). Research has, therefore, overlooked the ways in which marital history, current marital status, and relationship formation in old age influence formal assistance patterns of frail elderly persons and, in particular, the ways in which gender affects these important connecting points in the support system. The Salience of Gendered Marital History Our point of departure is that the consideration of marital history will contribute to the understanding of the integral and often complex relationship between the formal and informal support networks. This can be elaborated by using a life course perspective. Life course theory "concentrates on age-related transitions that are socially created, socially recognized, and shared" (Hagestad & Neugarten, 1985, pp. 35-36). Of particular relevance to the present study is the notion that earlier life decisions and circumstances affect outcomes experienced later in life (Elder, 1978; Hagestad, 1990). When applied to channels andflowsof formal assistance to frail elderly persons, the life course perspective helps to unveil the influence of the gendered nature of marital history. It also clarifies interrelationships with other demographic and social factors affecting use of


formal services - in particular, functional health, social network characteristics, and socio-economic status (see especially, Connidis & McMullin, 1994; Mutchler & Bullers, 1994). Marital histories directly shape the size, composition, intensity, and thus, the configuration of "convoys of support" (Kahn & Antonucci, 1980) over the course of a lifetime. Stability (i.e. remaining married or single) and change (i.e. separation, divorce, widowhood, and remarriage) in marital status over the life course have implications for family and non-kin ties, income level, and other factors that affect formal support in old age. For example, the never-married (ever-single) tend to substitute support from the nuclear family with support from the larger social network, especially siblings (Connidis, 1989; Kendig, Coles, Pittelkow, & Wilson, 1988; Strain & Payne, 1992) and friends (Dykstra, 1990a, 1990b). It is also well known that marital dissolution decreases the size of friendship networks and reduces contact with certain family members (Broese van Groenou & Tilburg, 1997). Moreover, marital status is directly linked to living arrangement, thus influencing the proximity and consistency of informal support (Chappell, 1990). For example, widowhood removes spousal support from the household, an experience more common for women. In this way, marital disruption affects the propensity for frail elderly persons to employ formal services - a weakened support network may increase the need for assistance from outside informal circles, while simultaneously deteriorating the communication linkage (resource) between the individual and service agencies. Furthermore, economic disadvantage tends to be heightened for virtually all non- married categories of older women, but less so for the ever-single (Liefbroer & Jong Gierveld, 1995). The trajectories of convoys of support are thus shaped by marital history, but not necessarily in the same manner for men and women. Spousal support is relied on more heavily by older men than by older women (Peters & Liefbroer, 1997; Spitze & Logan, 1989; Stollar & Cutler, 1992; Ward, La Gory, & Sherman, 1985). However, disruptions in marital status at any point in the life course can have different consequences for the quality of family relationships by gender. Aquilino (1994), for example, reports that later life parental divorce and widowhood lowered relationship quality and contact between older parents and their adult children, and that the effects were more pronounced for father-child than for mother-child relationships. Separated/divorced and ever-single older men report being more isolated from family than their female counterparts (Keith, 1986). And widows tend to have more varied sources of support, including close female friends and supportive neighbours, than their male counterparts (Stevens, 1995). Turning to formal assistance, Connidis and McMullin (1994) find that, overall, unmarried older persons seek support from outside the informal network more than the married, but differently across gender. For example, elderly married men are also more likely to use formal help than thenfemale counterparts. In addition, preferences for formal care appear to



differ by gender and marital history. The relatively low utilization rates of formal services by older persons have been shown to be partly related to preferences for independence, which vary by gender and marital status (Solinge, 1994; Wister, 1992). Not only do we find differences between men and women in the consequences of marital experiences, but also in the likelihood of their following particular marital trajectories. Women are more likely to be widowed because of sex differences in mortality and age differences between spouses in which husbands are typically older than their wives. Research (e.g., Antonucci, 1994; Bernard, 1973) suggests that men gain more from marriage than do women in terms of support and coping with stressful life events. This may partly explain higher remarriage rates for men than for women over the life course (Wu, 1995), and a greater propensity for older men to have a non-spousal partner, especially one in better health (Bozon, 1991). In sum, marital history can be viewed as an important thread that weaves together many of the major determinants of formal support in old age. These can be understood as comprising a complex relationship among informal and formal support needs, preferences, and resources - all of which are affected by life course dynamics that may differ for men and women. Given that measures of several support network characteristics (e.g., frailty of a spouse or presence of a partner) differ for persons with different marital statuses, there are advantages to analysing this latter variable separately. For these reasons, patterns of formal help by gender, as well as measures of the support network and socio-economic status, are examined separately for the first-married, ever-single, divorced (not remarried), widowed (not remarried), and remarried (and currently married). Using these more refined marital status groups allows for a better examination of the gendered nature of marital history. The testing of models separately for each marital history group is similar to that used in other studies (e.g., Mutchler & Bullers, 1994) that have separated married and non-married older persons for analytical purposes. We turn now to a discussion of relevant literature for each marital history grouping, for the purpose of developing specific hypotheses for each group. The First-Married Among older couples in their first marriage (referred to here as first-married), it is well established that men tend to rely on their spouse for affective and instrumental support, while women rely on a more extensive informal network (Antonucci & Akiyama, 1987; Jong Gierveld, 1986; Lowenthal & Haven, 1968). Older women are also more likely to be recipients of formal care (Mutchler & Bullers, 1994; Stollar & Cutler, 1992). Using United States data, Mutchler and Bullers (1994) find that married women are about twice as likely as married men to use formal assistance. However, it is not clear whether women need more formal help than do men, whether


they have a stronger preference for formal care, or whether husbands are reluctant or unable to assume a caregiving role, for health or other reasons. To adequately address the research question of gender differences among the first married in formal care usage, one ideally needs information on social networks, functional health status, as well as spousal frailty. We hypothesize that there will be a gender effect and that it will not be accounted for by these other factors. If the gender difference persists after the inclusion of measures of functional health and spousal frailty, as well as after controlling for network characteristics, then indirect support is provided for two interpretations. One is a preference for formal help among women; the second is the possibility that husbands are reluctant to provide care for their wives not because of their own frailty, but because they do not want or are unable to provide such care. The Never-Married Earlier stereotypes of the never-married as an isolated and homogeneous group have been largely laid to rest (Rubenstein, 1987). Nevertheless, never-married men have the smallest social networks (Mugford & Kendig, 1986; Tilburg, 1995; Wenger, 1984). Their situation stands in contrast to that of their female counterparts, who tend to have relatively expansive social networks. Strain and Payne (1992) show that although ever-single older women have a larger friendship network, they do not have more contact with these individuals, and neither are there gender differences in contact with various family members. Moreover, Connidis and McMullin (1994) find an absence of gender differences among the never-married in the use of paid help. Yet, there is some evidence that never-married older women fare better than never-married older men in terms of socio-economic status factors (Arber & Ginn, 1991; Liefbroer & Jong Gierveld, 1995). Past research suggests that socio-economic factors are positively associated with utilization of formal assistance (Chappell, 1992). It is therefore expected that measures of socio-economic status (e.g., education, income, and home ownership) will be more important than gender per se in affecting patterns of formal care among the never-married. We test this hypothesis, controlling for social network characteristics. The Divorced (Not Remarried) Research has consistently found that older divorced women are less socially isolated than their male counterparts (Keith, 1986; Strain & Payne, 1992). Divorced fathers tend to exhibit lower rates of social interaction with their adult offspring and tend to have less rewarding relationships (Cooney, 1993; Cooney & Uhlenberg, 1990; Dykstra, 1998; Strain & Payne, 1992). It has also been shown that poor relationships between middle-aged parents and adult children normally do not improve when parents age, except when children feel a moral obligation to stay in touch (Hagestad, 1987). And compared to non-divorcees, older divorcees are less likely to



have supportive relationships with adult children. Conversely, older divorced women's likelihood of experiencing more financial difficulties than divorced men may be a more influential factor than informal support in determining patterns of formal support use, leading to lower utilization rates of formal care. Our hypothesis for the divorced is similar to that for the never-married. Controlling for social network characteristics, we expect to find that formal service use is determined more by socio-economic status than gender. Widowed (Not Remarried) Similar to the other unmarried groups, widows tend to have stronger informal support networks than widowers (Martin Matthews, 1991; Stevens, 1995; Wister & Strain, 1986) and have lower incomes (Liefbroer & Jong Gierveld, 1995). Research by Mutchler and Bullers (1994) finds no differences between unmarried men and women in the receipt of formal care. However, this research combines all unmarried older persons, and does not include the availability of a (new) partner or other relationships in the social network, such as the availability of adult children. An understanding of the differences in formal care usage between widows and widowers requires knowledge about their respective support networks, their incomes, and about whether there are other confounding effects linked to the presence of a (new) partner, either within or outside the household. We hypothesize that attributes of the social network (including size, presence of a partner and number of children), and socio-economic factors (income, ownership status and education) will be more important than gender in explaining formal help. The Remarried (Currently Married) Research on remarried elderly persons has concentrated mainly on the determinants of remarriage (see e.g., Wu, 1995), and on the decreasing propensity for remarriage among older populations, especially among women (Connidis, 1989). Some research addresses the issue of support. Connidis (1989), for example, notes that remarriage in old age results in greater time spent with friends and children. Since most remarriage occurs among middle-aged persons, it may be that older remarried couples resemble the first-married. In fact, examination of patterns of informal and formal support have tended to combine remarried and married groups — given the availability of a spouse within both marital types - even though remarried elderly persons exhibit several differences from the first-married. First, changing spouses not only involves a previous divorce or widowhood, but also the combination of a new family. These marital status changes directly affect relationships with family and friends. Second, men tend to remarry women who are younger than themselves to a greater degree than in their first marriages (Bozon, 1991), increasing the likelihood that they will have spousal support in old age. For these reasons, we expect


that a gender effect will be present among the remarried, but that it will be partly accounted for by the effect of spousal frailty. Although several studies (e.g., Connidis & McMullin, 1994; Mutchler & Bullers, 1994) have demonstrated significant differences in patterns of informal and formal care among older married, single, divorced and widowed men and women, research has tended to combine marital status groups. In response to this research gap, the focus of this paper is the examination of gender differences in patterns of formal help for each marital history group among older persons who have a functional limitation. Several other factors shown to be important have been included in the analyses, including health of spouse (for married and remarried persons), availability of a non-marital partner (for the other marital status groups), number of living adult children, network size, income, education, home ownership and functional limitations. These have been grouped into three sets of variables: demographic/health; socio-economic status; and social network characteristics. In addition, to determine whether the gendered nature of marriage is part of a broader gendering of man-woman relationships, we also test for interactions between gender and presence of a partner for the unmarried groups, and between gender and having a frail partner for the married groups. Finally, interactions between gender and functional limitations, income, and the other social network characteristics are examined. Methods Sample Characteristics The data used here are a subset of those collected as part of a 1992 Dutch study titled, Living Arrangements and Social Networks of Older Adults (NESTOR-LSN). Face to face interviews were conducted with 4,269 older persons and 225 proxies. The overall response rate was 61.7 per cent.1 The mean age of the sample was 72.8 years. Institutionalized respondents were excluded here, since our specific purpose was to investigate patterns of informal and formal support among older adults who have a health need and are therefore "at risk" of requiring formal assistance. The analysis was therefore restricted to respondents experiencing one or more functional limitations with personal activities of daily living. The following four questions were used to assess limitations: "Can you walk up and down stairs, . . . walk for five minutes outdoors without resting, .. . get up from and sit down in a chair, . . . dress and undress yourself (including putting on shoes, doing up zippers, fastening buttons)?". The response key included: not at all, only with help, with a great deal of difficulty, with some difficulty, and without difficulty. The four functional limitation items formed a hierarchically homogeneous scale (H = .68) which was reliable (Cronbach's Alpha = .87). The scale ranged from 4 (numerous problems) to



20 (no problems). Respondents with a score of 20 were excluded from the analysis, resulting in a sub-sample of 1,378. Separate analyses were performed for each of five distinct marital history groups. These included: first marriages (615, 44.6%); the nevermarried (99, 7.2%); the divorced (not remarried) (63, 4.6%); the widowed (not remarried) (536, 38.9%); and the remarried (currently married) (65, 4.7%). Non-negligible proportions of the never-married, the divorced, and the widowed had partners, usually in consensual unions (this was the case for 7.1% of the never-married, 14.3% of the divorced, and 2.6% of the widowed), less often outside the household (this was the case for 4% of the never-married, 14.3% of the divorced and 2.1% of the widowed). Small proportions of the first-married and of the remarried were not living with their partners because the latter were institutionalized (1.6% and 4.6%, respectively). Measurement Formal Help. The dependent variable dichotomizes respondents into those who receive no formal assistance and those who receive some formal help. It should be noted that either category may include support from informal sources, since the majority of older adults receive some support from family or friends (in line with research findings, see e.g., Soldo, Wolf, & Agree, 1990). Formal sources of help include: help from a private nurse or a community nurse, social services such as home help or serviceflat personnel, or volunteers. 2 Informal sources of assistance are: 1) a partner, 2) household members other than a partner, 3) non-resident children, and 4) other informal sources such as other family members, friends, acquaintances, and neighbours. The frequencies and codes for the dependent variable and independent variables are shown in Table 1. Age, Sex and Functional Status. Age is measured in exact years. For gender, men are coded 0, women 1. The functional limitation score is used as our measure of functional status; a higher score indicates better functional health. Only persons with less than a perfect functional limitation score (ranging from 4 to 19) are included. Socio- Economic Resources. Three measures of socio-economic resources are incorporated into the analysis. The first is monthly household income. Income categories were converted into an interval scale by assigning the median income value for each income category to individuals in that grouping. In order to make the household incomes of those who live alone comparable to those of older adults co-residing with a partner, a family equivalence factor was used. The monthly incomes of households with a partner were multiplied by a factor of 0.7 in accordance with research conducted by Schiepers (1988). The missing cases for household income were recoded using mean substitution using the ten sex/marital history


Table 1 Frequency distributions, means and standard deviations for dependent and independent variables* Dependent Variable

Categories

Formal Help

No formal help Some formal help Total

Frequency

Percentage

880 498 1378

63.9 36.1 100.0

Categorical Independent/ Control Variables Marital Status

First married Never married Divorced (not remarried) Widowed (not remarried) Remarried

615 99 63 536 65

44.6 7.2 4.6 38.9 4.7

Gender

Male Female

586 79

42.5 57.5

Home Ownership

Non-owner Owner Missing

970 385 23

70.4 27.9 1.7

Frailty of Spouse

Healthy 1 or more ADL problems Missing/NA

421 270 687

30.6 19.6 49.9

Partner Status

No spouse, cohabitation, and/or outside partner Partner present

644 734

46.7 53.3

Interval Independent Variables

Missing Cases

Range

Mean

Standard Deviation

Age

—

54-89

75.7

9.2

ADL

2

4-19

16.7

3.0

Dutch guilders**

32

963-4025

1742.9

759.9

Years of Education

-

5-18

7.9

3.1

Network Size

35

0-40

8.4

7.7

# Living Children

13

0-12

2.8

2.2

Household Income in

* Frequencies and percentages are provided only for categorical variables. ** 1 Dutch guilder is approximately $0.7 CDN.

groups. It is recognized that these data provide only an approximation of differences in household income. The second measure is home ownership, an indicator of accumulated assets and convertible resources. Together, household income and home ownership measure financial capital. Educa-



tional attainment is the third measure of socio-economic resources, measured by number of years. After statistically controlling for household income and home ownership, it is contended that education measures human capital. This would entail knowledge of services, how to access them, and familiarity with using bureaucratic systems. In addition, education reflects attitudes and preferences towards using formal help, but it is recognized that education is only a proxy for subjective dimensions of formal service decisions. The intercorrelations among these three variables did not indicate evidence for multicollinearity. Social Network Measures. Three social network characteristics are included; however, not all are included in each analysis of the marital history groups, for reasons detailed below. The first measure is network size. The procedure by which network members were identified is an adaptation of one developed by Cochran, Lamer, Riley, Gunnarson, and Henderson (1990); details on the network delineation procedure can be found in Van Tilburg (1995). The aim was to have information on those with whom the older adult interacted regularly, and on the diversity of those contacts. The survey focussed on the interactive network. Seven relationship domains were specified: household members; children and their partners; other kin; neighbours; colleagues; organizational contacts; and "others". For each domain, the respondent was requested to specify the names of those with whom they were "in touch regularly" and those who were "important" to them. The definitions of "regular contact" and "important" were left to the respondents. To be selected, network members had to be 18 years old. A limit of 80 was set on the number of network members that could be chosen. Network size is the total number of reported names, with the exception of partners, children and children-in-law. These latter relationships were excluded from the computation of network size to avoid contamination with the availability of a partner and children which, for analytical purposes, were introduced as separate variables. The second characteristic is the availability of a partner, defined differently for the currently married and the nonmarried. Among the currently married, which includes both the first-married and the remarried, the frailty of the spouse is considered, under the assumption that a partner who is ill or frail is less likely to be a source of help. Those with a spouse who has no functional limitations (i.e. a functional capacity score of 20) were contrasted with those with a spouse who has functional limitations (i.e. a functional capacity score between 4 and 19). It is recognized that this is only a rough indication of functional limitations. Among the nonmarried (the never-married, divorced and widowed), the health of the partner is not measured because of small numbers. Rather, the distinction is drawn between those with a partner (either within or outside the household) and those without such a relationship (partner status). Preliminary analyses revealed that the helping patterns of older adults with partners outside


the household more closely resemble those of the cohabitating than those of older adults who had no partners at all. The number of surviving children is the third network characteristic. It includes biological, adoptive and step children. Logistic Regression Analyses We use logistic regression techniques to investigate the effect of gender and several other independent variables on whether the older adult received support strictly on an informal basis, including those who received no support (coded 0), or whether some formal care was received alone or in combination with informal support (coded 1). This approach conforms to previous research (e.g., Stoller & Cutler, 1993; Mutchler & Bullers, 1994). Logistic regression is a statistical technique constructed for the purpose of analysing dichotomous dependent variables (see especially DeMaris, 1995). The results of the logistic regressions will be interpreted as predicting the odds of using formal help (with or without informal support) versus using only help from informal sources or none at all. The results of the logistic regression analyses are presented in Tables 3 to 7. Regression coefficients, their standard error, and level of statistical significance are presented for each of the independent variables. The associated odds ratios are reported for statistically significant coefficients only. The regression coefficients produced by logistic regression represent the change in the log of the expected odds in the dependent dichotomous variable for each unit change in an independent variable (DeMaris, 1995). For categorical variables, the odds ratio [exp(B)] gives the odds of receiving formal help, given one category of an independent variable compared to a reference (omitted) category, while controlling for the effects of all other variables in the statistical model. For interval variables, the odds ratio gives the odds of receiving formal help for each unit increase in the independent variable, net of other variables statistically controlled. A hierarchical logistic regression procedure is used to test the hypotheses. We examine the effect of gender separately and in combination with background factors (age and functional limitations), financial resources (household income, home ownership, and education), and social network characteristics (network size, health status or presence of a partner, and number of living children). Since different hypotheses are posed for the various marital history groups, different hierarchical models are employed. However, all five of the logistic regression analyses follow a similar rationale for hierarchical modelling. Model 1 incorporates age and all other control variables. Models 2 and 3 include gender and/or the variables measuring financial resources and network characteristics, depending on the hypothesis. Gender is entered in Model 2 by itself for those marital history groups for which we hypothesize a decrease in a gender effect after the introduction (in Model 3) of financial status or social network variables (for the first-married, widowed, and remarried groups). Gender is entered

20 (46.5) 9 (37.5) 58 (45.7) 9 (22.5)

23 (53.5)

15 (62.5)

69 (54.3)

31 (27.5)

423 (72.2)

Never Married

Divorced (not married)

Widowed (not remarried)

Remarried

Total 163 (27.8)

67 (19.0)

285 (81.0)

First Married

Some n (Row %)

None n (Row %)

Male

Marital Status

Table 2 Formal help by marital status and gender

586 (100)

40 (100)

127 (100)

24 (100)

43 (100)

352 (100)

Total n (Row %)

Formal Help

Gender

457 (57.7)

12 (48.0)

202 (49.4)

25 (64.1)

29 (51.8)

189 (71.9)

None n (Row %)

335 (42.5)

13 (52.0)

207 (50.6)

14 (35.9)

27 (48.2)

74 (28.1)

Some n (Row %)

Female

792 (100)

25 (100)

409 (100)

39 (100)

56 (100)

263 (100)

Total n (Row %)


simultaneously with the financial status variables in Model 3 to directly compare the effects of these variables (for the never-married and divorced analyses), subsequent to the inclusion of social network variables in Model 2. Number of children is not included in the never-married analysis because of the small number of persons with children. Separate analyses examine interactions between gender and other key variables. These include interaction terms between gender and 1) partner status (for the non-married groups), 2) frailty of spouse (for married groups), and 3) functional status, income, network size and number of children (for all relevant groups). The correspondence rule used to assess the hypotheses related to gender and the other principal variables is whether there is a statistically significant association with formal help, and whether the association with gender is reduced to non-significance with the introduction of other variables. Results Descriptive Analyses

Of all of the marital history groups, the first-married are least likely to receive formal care, with noted differences between the sexes: 19.0 per cent of the first-married men and 28.1 per cent of the first-married women receive formal help (see Table 2). The widowed and the never-married are most likely to make use of formal services: 45.6 per cent of widowed men, 50.6 per cent of widowed women, 46.5 per cent of never-married men, and 48.2 per cent of never-married women report using these services. The divorced occupy an intermediate position: 37.5 per cent of the men and 35.9 per cent of the women receive formal care. The most pronounced gender difference emerges among the remarried. Whereas formal service usage is relatively low among remarried men, it is highest among remarried women. Subsequent logistic regressions provide insight into the determinants of the use of formal care within the five marital history categories. Logistic Regression Analyses

The First-Married The logistic regression for the first-married, presented in Table 3, shows that all three models are statistically significant (see Model Chi Square). Age and education exhibit statistically significant relationships with use of formal services and these associations are in the expected direction as shown in Model 1. The likelihood of using formal care (with or without informal support) increases (odds ratio=l .08) for each increment in chronological age and for education, after controlling for the other variables entered in Model 1. However, household income, home ownership, network size and number of children do not display statistically significant coefficients (see Table 3).

.05

.18

.07*

.01

-.06

EDUCAT

NETWORK CHILDREN

591

46.91***

—

591

-9.25 68.54***

1.07***

-.03

.00

.12

.09

.09*** -.00

B

.24

.05

.01

.04**

.23

.00

.01

Model 2 S.E.

591

-6.13 98.80***

.70**

-.04 1.15*** -.16***

2.92

.01

.13***

.19

.08*** -.00

B

—

—

1.12

—

—

1.10

Odds Ratio

2.02

.85 .23

3.17 .03

—

—

1.14

—

—

1.08

Odds Ratio

.25

.05

.12

.04

.24

.00

.01

Model 3 S.E.

NOTES: 1. Dependent Variable - formal help (0 = no formal help, 1 = some formal help); Independent Variables - ADL = activities of daily living, HINCOME = household income, OWNER = home ownership (0 = non-owner, 1 = owner), EDUCAT = number of years of education, NETWORK = network size excluding nuclear family, CHILDREN = # of living children, FRAILSP = spouse's health (0 = healthy, 1 = 1 or more ADL). 2. All codes not shown above range from low to high. 3. B = parameter estimate, S.E. = standard error of estimate. 4. *p < .05, **p < .01, ***p < .001.

Model Chi Square Cases in Analysis

FRAILSP Constant

-7.17

.04 .01

-.000

HINCOME OWNER

GENDER ADL

—

.22

.07***

AGE

1.08 —

1.08 —

.01 .00

B

Independent Variables

Model 1 S.E. Odds Ratio

Table 3 Logistic regression for the first married


Gender is added in Model 2. The odds of using formal help are almost three times higher for women than for men (odds ratio=2.92). Functional status and frailty of spouse are included in Model 3. Having fewer functional limitations results in a decline in the probability of using formal help (odds ratio=.85). And, as expected, having a frail partner increases the likelihood of receiving formal assistance (odds ratio=2.02). The inclusion of the functional status and frail spouse variables has little effect on the relationship between gender and formal help. Thus, although the functional status of the spouse affects the propensity to employ formal help, the tendency for women to use more formal assistance than men is not affected by the health status of their marital partner. Moreover, the interaction between frailty of spouse and gender is not supported (results of the separate interaction analyses are not reported in Table 3). In Models 2 and 3, the statistically significant associations of age and education are replicated. The Never-Married As observed in Table 4, only age displays a statistically significant association with formal help in Model 1 (functional status is not statistically significant). There is an increased likelihood of using formal assistance with each increase in age (odds ratio=1.05). Two social network characteristics (size and the presence of a partner) are included in the Model 2; however neither of these variables demonstrates a statistically significant relationship, and the overall model is not statistically significant. In Model 3, gender is compared with the socio-economic variables. Gender, household income and owner status form statistically significant relationships with formal help. It is interesting that the odds of using formal help are about one-third as high for females as for males (odds ratio=.31), after controlling for all other variables in the model. As hypothesized, household income exhibits a positive association with use of formal assistance. For each unit increase in monthly household income, the probability of using formal help increases 1.001. The association for home ownership is in the opposite direction (odds ratio=.29), with homeowners about one-third as likely to use formal care. Education is not statistically significant. A separate analysis was conducted that also included living arrangement (living alone/coresiding), since homeowners have more room to live with other persons who may provide assistance in the home.3 However, the findings for home ownership, as well as the other relationships, were replicated. Furthermore, the interaction between partner status and gender did not result in a statistically significant relationship. The Divorced (Not Remarried) All three models for the divorced marital history group are statistically significant as shown in Table 5 (see Model Chi Square). Model 1 incorporated age and functional status; however, only age resulted in a statisti-

.05* .01

-3.68 .01**

-1.17*

.22

—

.04

.08**

B

—

1.05 —

Odds Ratio

NOTES: 1. PARTNER = partner status (0 = no spouse, cohabitant, or partner outside of the household, 1 = present). 2. B = parameter estimate, S.E. = standard error of estimate. 3. *p < .05, **p < .01, ***p < .001.

95

27.16***

7.78 95

Model Chi Square

95

6.42*

Constant

Case Analysis

.04 -8.74

-3.97

-4.89

EDUCAT

-1.25*

2.32

.20

.09

.02

Model 2 S.E.

OWNER

HINCOME

GENDER

-2.36

—

1.05 .16

.04

ADL

PARTNER

.05*

AGE

B

NETWORK

.02

.08

B



Table 4 Logististic regression for the never-married

.08

.58

.29 —

1.001

.31

.57 .00

—

—

—

1.09

Odds Ratio

2.45

.22

.10

.03

Model 3 S.E.

.17

-.31

CHILDREN

63

-1.74 19.10** 63

-3.67

12.85***

NOTES: 1. B = parameter estimate, S.E. = standard error of estimate. 2. *p < .05, **p < .01, ***p < .001.


EDUCAT Constant

OWNER

GENDER HINCOME

.06 1.17

.08 -1.87

.04

NETWORK PARTNER

—

1.09

Model 2 S.E. .13

.12

.09**

-.19

AGE ADL

B .08* -.22

.03

B



Table 5 Logistic regression for the divorced (not remarried)

— — —

1.09 —

Odds Ratio

63

1.13 26.14**

.14

.001 1.14 -.10

.76 .001* .61

.20

1.5

.08

.04 .14

Model 3 S.E.

-.37 -.97

-3.05*

.12

-.29*

.05

B

1.001 — —

—

.05 —

—

.75

—

Odds Ratio



cally significant association. The likelihood of using formal help increases with each age (odds ratio=1.09). The network characteristics were entered in Model 2, where we observe no statistically significant associations with the dependent variable. In Model 3, we compare gender simultaneously with the socio-economic factors. However, no gender difference emerges. Among the socio-economic factors, only household income is statistically significant, which is in a positive direction (odds ratio=1.001). The previous relationships in Model 2 are replicated, except that relationships are also supported for functional status and the presence of a partner. The probability of engaging formal help decreases with better functional status (odds ratio=.75) and having a partner (odds ratio=.05), after controls are placed on the other independent variables. None of the interaction effects, including the interaction between partner status and gender, were supported. Widowed (Not Remarried) All three models are found to be statistically significant for the widowed group (see Table 6). Age and functional status are entered in Model 1, where relationships are observed for both variables. The likelihood of using formal help increases with age (odds ratio=l .06), and decreases with higher functional status (odds ratio=.87). Gender is incorporated in Model 2, but an association is not statistically supported. In Model 3, the socio-economic and social network variables are entered. Here wefindthat partner status and number of children both display statistically significant relationships with formal care in Model 3. The presence of a partner among the widowed reduces the likelihood of engaging formal help, net of the other variables. Those without a partner are approximately one-tenth as likely to use formal assistance (odds ratio=.10). Furthermore, the probability of using formal support decreases with each additional living child (odds ratio=.92). However, none of the socio-economic variables exhibit a statistically significant association with formal help. Associations for age and functional status are replicated in Model 3. Tests for gender interactions revealed a statistically significant difference between widowed men and women in regard to functional status. Older widows with poorer functional health are less likely to use formal help than are older widowers with poorer functional capabilities. The Remarried (Currently Married) The results of the logistic regression for the remarried sub-sample are presented in Table 7. None of the models are found to be statistically significant, although gender alone produces a relationship that is statistically significant. The interaction effects, including the one between gender and frailty of spouse, were also examined, but did not have statistically significant associations with formal care.

.06*** -.14***

AGE

.04 .24

-.00

—

.22

.10

NOTES: 1. B = parameter estimate, S.E. = standard error of estimate. 2. *p < .05, **p < .01, ***p < .001. 3. Gender X ADL interaction (B = -.29***, odds ratio = .75).

503

42.17*** 503

42.36***

503

65.71***

-.09* -1.73

CHILDREN


-2.27**

PARTNER

-2.24

.03

Constant

— .04 .04

NETWORK

.04

.76

.92

.10

—

—

.04

—

—

.85

1.05

Odds Ratio

.24

-2.39

.01

.05***

.01

-.17***

Model 3 S.E.

.87

B

.03

1.06

Odds Ratio

.06*** -.14***

Model 2 S.E.

.000

.87

.03

B

.001 -.22

1.06

.01


HINCOME OWNER EDUCAT

ADL GENDER

B


Table 6 Logistic regression for widowed (not remarried)

61

.13

.08 .04 —

— —

—

—

—

—

NOTES: 1. B = parameter estimate, S.E. = standard error of estimate. 2. *p < .05, **p < .01, ***p < .001.

Cases in Analysis

Model Chi Square

-.34 5.22

.13

CHILDREN

GENDER FRAIL SP Constant

.08

.75

-.52

-.01

.00

.00

HINCOME OWNER

NETWORK

.10

-.14

ADL

EDUCAT

.03

.01

AGE


B


Table 7 Logistic regression for remarried

61

-6.05 12.14

1.93**

.14

-.00

.15

-.80

.00

-.11

.06

B

.79

.14

.04

.09

.80

.00

.12

.04

Model 2 S.E.

6.89

—

—

—

.70 .33

61

12.36

-5.71

6.58

—

—

—

.79

.14

.04

.09

—

—

—

—

Odds Ratio

1.88*

.14

-.00

.15

.81

-.81

—

.00 .00

—

.12

-.11

—

.04

.05

Model 3 S.E.

—

B

—

Odds Ratio


Discussion Research into patterns of social support among older adults has been spurred, in part, by continued changes in family structure and dynamics, which transform the configuration of support across the life course. The gendered marital experiences of persons as they age constitute an important theme within this area of study. Drawing on life course theory, the purpose of this study is to extend previous research by examining gender differences in reliance on formal help among older persons with a functional limitation, within distinct marital history groups. Specific hypotheses are developed and tested for each marital history group. Consistent with the literature on this topic (see especially, Mutchler & Bullers, 1994), we observe that first-married older women with functional limitations are approximately three times more likely to use formal help than are their male counterparts. This relationship remains robust even after controlling for all of the other independent variables, including functional health, frailty of spouse, other network characteristics, socioeconomic status, and age. Additionally, first-married persons who are older, who are more educated, who score lower on functional health, and who have a spouse with a health problem, exhibit a greater propensity to rely on formal help. These relationships are in the expected direction. While the absence of support for an association between financial status and formal care among married elderly replicates earlier work (Connidis & McMullin, 1994; Mutchler & Bullers, 1994), the present research also incorporates an education variable and finds a positive effect. Older married persons with more years of education probably have more extensive knowledge about available community and health services and appear to exhibit a more accepting attitude towards service use. In sharp contrast, none of the models for the remarried are statistically significant. Furthermore, none of the gender interaction effects are supported for either of the two married groups. We turn now to the unmarried groups. Research conducted by Mutchler and Bullers (1994) suggests that no gender differences in formal care exist among the currently unmarried. The analyses in the present study disaggregate the unmarried into never-married, divorced (not remarried), and widowed (not remarried) marital histories. Although our findings for the divorced and widowed groups support the conclusion above, they do not for the never-married. Indeed, wefindthat the likelihood of relying on formal help is about one-third lower for never-married females compared to never-married males, after accounting for functional health, socio-economic status, network size and presence of a partner. Note that this association between gender and formal assistance is opposite to that found for the first-married. Among the never-married, it was anticipated that there would be few, if any gender differences in patterns of formal help. Moreover, it is surprising that there was a gender effect despite controlling



for socio-economic status and social network characteristics. Other research demonstrates that never-married women tend to have higher socioeconomic status than never-married men (Arber & Ginn, 1991; Liefbroer & Jong Gierveld, 1995), and are embedded within larger social networks (Mugford & Kendig, 1986; Strain & Payne, 1992; Tilburg, 1995; Wenger, 1984). Our finding, while unexpected, may be indicative of the cumulative impact of different lifestyles on never-married men and women that are not measured using the variables here. For example, it is possible that single older women exhibit greater levels of independence and self-reliance than single men, and that these normative, preference and lifestyle differences contribute to lower formal help. Conversely, perhaps the lower socio-economic statuses of older never-married men, and their greater use of various social services throughout their life increase their propensity to rely on formal help in old age, apart from the direct influence of socio-economic status. This unexpected finding for the never-marrieds demands future study and also reinforces the importance of separately examining non-married groups. There are several other salient differences among these marital groups pertaining to patterns of formal help that deserve attention. For the never-married, being older, with a higher household income, and not owning a home are associated with higher levels of formal help. The income finding corroborates previous research, suggesting that socio-economic status is more influential than family support for the lives of single elderly persons when compared to the married. Surprisingly, however, homeowners are less rather than more likely to use formal services. This finding remains even after statistically controlling for living arrangement, which was undertaken in a supplementary logistic regression analysis. One explanation for this association is that when formal support becomes necessary, single individuals move out of owned homes into smaller places that can be managed more easily, thus limiting the amount of formal support that is required. Thus, single persons may remain in their owned homes only when they are able to function independently. The divorced (not remarried) are also distinguishable from the first-married, remarried, widowed and never-married. The divorced are less likely to use formal help if they are in better functional health, have a partner, and have a higher household income. Contrary to the never-married, it seems that aspects of the social network are at least as important for the divorced as are socio-economic resources. It would appear that in so far as divorced elderly persons are embedded within a supportive informal network, there is little reliance on formal help. For those divorced older persons who have been unsuccessful in maintaining supportive informal ties, there is a greater need for formal services. However, no differences are observed between women and men in the predictors of receipt of formal care.


For the widowed, being older, having lower functional health, lacking a partner, and having fewer children increase the propensity to use formal care, but socio-economic status does not. In addition, use of formal help among the widowed is affected by a gender and health interaction, whereby older widows with better functional health are less likely to be users than males with functional limitations. It is not entirely clear how to account for the latter finding. One explanation is that widowers and widows are differently affected by the loss of resources, with widowers exhibiting greater vulnerability. A health decline would have more negative repercussions for widowers' independence. An alternative explanation suggests differences in help-seeking behaviour. Widowers affected by a decline in health would more readily mobilize help from professionals. Additional research is required to ascertain what a decline in health means for the support needs of widows and widowers. Thesefindingspoint to the gendered nature of marital history for formal care in old age. Not only do men and women differ in the likelihood of experiencing various marital histories, but also the consequences of these life events reflect unique elements of that experience. First marriages are clearly gendered - regardless of the frailty level of their spouse, first-married women receive more formal help,first-marriedmen receive less. Thus, first marriages can be treated as the marital status group that reproduces traditional gender relationships in which women provide more informal support to their spouse and must rely on more formal help when support is required. It is probable that the additional burden of providing support to a spouse for older married women contributes to reliance on the formal system to cope when limitations in health occur. The strong gender effect among the first-married also underlines the primacy of spousal support in old age, at least from the perspective of older men. The findings are consistent with Bernard's (1973) contention that men gain more from marriage than do women. Women tend to be in caregiving roles at all points in the life course, roles that are devalued and tend to be stressful (Arber & Ginn, 1991). Another gendered pattern in formal help occurred among the nevermarried - where it is observed that older single women exhibit a lower propensity to use formal help than older men. This supports the view that norms and preferences towards formal assistance develop differently over the life-courses of never-married men and women. Given that we statistically controlled for age, functional health, socio-economic status, and social network characteristics, alternative explanations concerning the social and economic conditions of never-married men compared to women obtain less support. Furthermore, it was found that, among the widowed and the divorced (not remarried), formal help is relied on less if a non-spousal partner is present, regardless of the gender of the individual. This is perhaps suggestive that partnerships outside of marriage are inclined towards more



egalitarian ties, at least among the divorced and widowed. In addition, it is noteworthy that the remarried are the only group for whom none of the models is statistically significant. It would appear that the remarried are unique, in that their use of formal help in old age cannot be explained using the same sets of variables as for the other marital status groups. This issue begs further research into alternative hypotheses that may account for patterns of formal support among remarried persons in old age. Together, the results from this study reinforce the theory that patterns of formal care are influenced by an integral relationship among support needs, preferences and resources, all of which are affected by life course dynamics that vary by gender. Marital histories shape the constellation of informal support and economic resources available to older adults facing restrictions in function, and perhaps even the norms and preferences underlying formal care decisions. There also appears to be some degree of substitution of one type of resource for another. For example, the proclivity to use formal assistance among never-married elderly persons is affected more by their financial resources than their social resources, whereas both factors are important for the divorced, and social network is more important for the widowed. Since this research does not investigate the degree to which there is a "fit" between health needs and patterns of both informal and formal support, we are unable to address the issue of adaptability, that is, whether and to what extent people's needs are being met through various combinations of support. Our findings imply, however, that older people with distinct marital histories have different needs and preferences, as evidenced by the varying patterns of explanatory variables for each marital status group observed here. More specifically, the results indicate that life course trajectories and the circumstances surrounding them, such as the availability and reliance on various support systems, are influenced by gender-specific key life events, such as marital transitions (Antonucci, 1994; Hagestad, 1990). Thus, only by looking into people's past can we uncover the keys to understanding the present. This research does not examine the specific role of other family members in decisions to use formal care, research that may help to elaborate the dynamics involved in service use. There is also a need to incorporate direct measures of norms and preferences connected to these decisions in order to better understand the ways in which the gendered nature of marital history, marriage, and relationship formation in old age influences informal and formal support at later stages of life. Explication of the intricacies of these processes will make it possible to better discern the consequences of family continuity and change on future patterns of formal care among the growing populations of older adults in contemporary societies. Notes 1

For more details on the NESTOR-LSN study, see Broese van Groenou, Tilburg, Leeuw, & Liefbroer, 1995.


2 3

Volunteer help has been included under formal help because it originates from outside the primary informal network. The correlation between home ownership and living arrangement was approximately .32, thus not raising problems of multicollinearity. We also found no evidence of suppression effects.

References Antonucci, T.C. (1994). A life-span view of women's social relations. In B.F. Turner & L.E. Troll (Eds.), Women growing older: Psychological perspectives (pp. 239-269). Thousand Oaks, CA: Sage. Antonucci, T.C, & Akiyama, H. (1987). An examination of sex differences in social support in mid life and late life. Sex Roles, 17, 737-749. Aquilino, W. (1994). Later life parental divorce and widowhood: Impact on young adults' assessment of parent-child relations. Journal of Marriage and the Family, 56, 908-922. Arber, S., & Ginn, J. (1991). Gender and later life: A sociological analysis of resources and constraints. London: Sage. Bernard, J. (1973). The future of marriage. New York: Bantam. Bozon, M. (1991). Women and the age gap between spouses: An accepted domination? Population, 3, 113-148. Broese van Groenou, M.B., & Tilburg, T. van (1997). Changes in the support networks of older adults in the Netherlands. Journal of Cross-Cultural Gerontology, 12, 23-44. Broese van Groenou, M.B., Tilburg, T. van, Leeuw, E. de, & Liefbroer, A.C. (1995). Appendix: Data collection, In C.P.M. Knipscheer, J. de Jong Gierveld, T.G. van Tilburg, & P.A. Dykstra (Eds.), Living arrangements and social networks of older adults (pp. 185-197), Amsterdam: Vrije Universiteit University Press. Brubaker, T.H. (1990). Families in later life: A burgeoning research area. Journal of Marriage and the Family, 52, 959-981. Chappell, N.L. (1990). Living arrangements and sources of caregiving. Journals of Gerontology: Social Sciences, 46, 51-58. Chappell, N.L. (1992). Social support and aging. Toronto: Butterworths. Cochran, M., Larner, M., Riley, D., Gunnarson, L., & Henderson, C.R. (1990). Extending families: The social networks of parents and their children. Cambridge: Cambridge University Press. Connidis, I. (1989). Family ties and aging. Toronto: Harcourt Brace. Connidis, I., & McMullin, J. (1994). Social support in older age: Assessing the impact of marital and parent status. Canadian Journal on Aging, 25(4), 510-527. Cooney, T.M. (1993). Recent demographic change: Implications for families planning the future. Marriage and Family Review, 18, 37-55. Cooney, T.M., & Uhlenberg, P. (1990). The role of divorce in men's relations with their adult children after mid-life. Journal of Marriage and the Family, 52, 677-688. DeMaris, A. (1995). A tutorial in logistic regression. Journal of Marriage and the Family, 57, 956-968. Dooghe, G. (1992). Informal caregivers of elderly people: A European overview. Ageing and Society, 12, 369-380.



Dykstra, P.A. (1990a). Next of(non) kin: The importance of primary relationships for older adults' well-being. Amsterdam/Lisse: Swets & Zeitlinger. Dykstra, P.A. (1990b). Disentangling the direct and indirect gender effects on the supportive network. In C.P.M. Knipscheer and T.C. Antonucci (Eds.), Social network research: Methodological questions and substantive issues (pp. 55—65). Amsterdam: Swets and Zeitlinger. Dykstra, P.A. (1995). Network composition. In C.P.M. Knipscheer, J. de Jong Gierveld, T.G. van Tilburg, & P. A. Dykstra (Eds.), Living arrangements and social networks of older adults (pp. 97-114). Amsterdam: VU University Press. Dykstra, P.A. (1998). The effects of divorce on intergenerational exchanges in families. The Netherlands'Journal of Social Sciences, 33, 77-93. Eggebeen, D. (1992). Family structure and intergenerational exchanges. Research on Aging, 14, 427-447. Elder, G.H. Jr. (1978). Approaches to social change and the family. American Journal of Sociology, 84, S1-S38. Hagestad, G. (1987). Parent-child relations in later life: Trends and gaps in past research. In J.B. Lancaster, J. Altman, A.S. Rossi, & L.R. Sherrod (Eds.), Parenting across the life span (pp. 405-433). New York: Aldine de Gruyter. Hagestad, G. (1990). Social perspectives on the life course. In R. Binstock & L. George (Eds.), Handbook of aging and the social sciences (pp. 151-168). San Diego, CA: Academic Press. Hagestad, G., & Neugarten, B. (1985). Age and the life course. In R. Binstock & E. Shanas (Eds.), Handbook of aging and the social sciences (pp. 35-61). New York: Van Nostrand Reinhold. Jong Gierveld, J. de (1986). Husbands, lovers and lonliness. In R.A. Lewis & R.E. Salt (Eds.), Men in families (pp. 115-125). London: Sage. Jong Gierveld, J. de (1999). New living arrangements: Older adults between independence and companionship. Paper presented at the European Population Conference, The Hague, The Netherlands, Aug. 30-Sept. 3, 1999. Kahn, R.L., & Antonucci, T.C. (1980). Convoys over the life course:Attachment, roles and social support. In P.B. Baltes & O. Brim (Eds.), Life-span development and behaviour (pp. 253-286). New York: Academic Press. Karney, B., & Bradbury, T. (1995). The longitudinal course of marriage quality and stability: A review of theory, method and research. Psychological Bulletin, 118, 3-34. Keith, P.M. (1986). Isolation of the unmarried in later life. Family Relations, 35, 389-395. Kendig, H., Coles, R., Pittelkow, Y., & Wilson, S. (1988). Confidants and family structure in old age. Journals of Gerontology: Social Sciences, 43, S32-S40. Kobrin, F. (1976). The primary individual and the family: Changes in living arrangements in the United States since 1940. Journal of Marriage and the Family, 38, 233-239. Liefbroer, A., & Jong Gierveld, J. de (1995). Living arrangements, socio-economic resources, and health. In C.P.M. Knipscheer, J. de Jong Gierveld, T.G. van Tilburg, & P.A. Dykstra (Eds.), Living arrangements and social networks of older adults (pp. 15—36). Amsterdam: VU University Press. Lowenthal, M.F., & Haven, C. (1968). Interaction and adaptation: Intimacy as a critical variable. American Sociological Review, 33, 20-30. Martin Matthews, A. (1991). Widowhood in later life. Toronto: Butterworths.


Mugford, S., & Kendig, H. (1986). Social relations: Networks and ties. In H. Kendig (Ed.), Ageing and families: A social networks perspective (pp. 38-59). Boston: Allen & Unwin. Mutchler, J., & Bullers, S. (1994). Gender differences in formal care use in later life. Research on Aging, 16, 235-250. Penning, M. (1990). Receipt of assistance by elderly people: Hierarchical selection and task specificity. The Gerontologist, 30, 220-227. Peters, A., & Liefbroer, A. (1997). Beyond marital status: Partner history and well-being in old age. Journal of Marriage and the Family, 59, 687-699. Rubenstein, R. (1987). The never married elderly as a social type: Re-evaluating some images. The Gerontologist, 27, 108-113. Schiepers, J.M. (1988). Huishoudensequivalentiefactoren volgens de budgetverdelings-methode [Family equivalence scales using the budget distribution methods]. Supplement bij de Sociaal-Economische Maandstatistiek, 2, 28-36. Soldo, B., Wolf, D., & Agree, E. (1990). Family, households and care arrangements of frail older women: A structural analysis. Journal of Gerontology: Social Sciences, 45, S238-S249. Solinge, H. van (1994). Living arrangements of non-married elderly people in the Netherlands in 1990. Ageing and Society, 14, 219-236. Spitze, G., & Logan, J. (1989). Gender differences in family support: Is there a payoff? The Gerontologist, 29, 108-113. Stevens, N. (1995). Gender and adaptation to widowhood in later life. Ageing and Society, 15, 37-58. Stoller, E.P., & Cutler, S.J. (1992). The impact of gender on configurations of care among married elderly couples. Research on Aging, 14, 313-330. Stoller, E.P., & Cutler, S.J. (1993). Predictors of use of paidhelp among older people • living in the community. The Gerontologist, 33, 31-40. Strain, L., & Payne, B. (1992). Social networks and patterns of social interaction among ever-single and separated/divorced elderly Canadians. Canadian Journal on Aging, 11, 31-53. Tilburg, T.G. van (1995). Delineation of the social network and differences in network size. In C. P.M. Knipscheer, J. de Jong Gierveld, T.G. van Tilburg, & P.A. Dykstra (Eds.), Living arrangements and social networks of older adults (pp. 83—96). Amsterdam: VU University Press. Ward, R., La Gory, M., & Sherman, S. (1985). Social ties of the elderly. Sociology and Social Research, 70, 102-106. Watkins, S.C., Bongaarts, J., & Menken, J. (1987). Demographic foundations of family change. American Sociological Review, 52, 346-358. Wenger, G.C. (1984). The supportive network: Coping with old age. London: Allen & Unwin. Wister, A.V. (1992). Residential attitudes and knowledge, utilization and future use of home support agencies. Journal of Applied Gerontology, 11, 84—100. Wister, A.V., & Strain, L. (1986). Social support and well-being: A comparison of older widows and widowers. Canadian Journal on Aging, 5, 205-220. Wu, Z. (1995). Remarriage after widowhood: A marital history study of older Canadians. Canadian Journal on Aging, 14, 719—736.