National Academies Press: OpenBook

Demography of Aging (1994)

Chapter: 5 The Elderly and Their Kin: Patterns of Availability and Access

« Previous: 4 Income, Wealth, and Intergenerational Economic Relations of the Aged
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

5
The Elderly and Their Kin: Patterns of Availability and Access

Douglas A. Wolf

INTRODUCTION

In recent years the field of family demography has developed rapidly. These developments include theoretical and methodological advancements and an especially rapid growth of applied research driven by the availability of large, nationally representative household surveys, as well as several longitudinal data sources (for a survey of the field, see Bongaarts et al., 1987). This chapter addresses issues in what might be called the ''family demography of the elderly." First, and most fundamental, we consider the composition of families containing elderly. Under the heading "kin availability," we address the observability of kin structures, ranging from the simple—counts of living kin occupying specified relationships—to the complex—enumerations of individual surviving kin according to type of relationship, with each member of the kin group described by an array of attributes. The second major focus is summarized by the term "access" to kin, which here has two manifestations: either the face-to-face access implied by coresidence, or the less intense or sustained access implied by close spatial proximity

The author recognizes a debt to his collaborators on several papers—Rebecca Clark, Vicki Freedman, and Beth Soldo—whose efforts are reflected throughout this paper. Thanks also go to Linda Martin, Sam Preston, John Casterline, and Ron Lee for useful comments on an earlier draft of the chapter, and to Lena Rose Orlando for her help in preparing the manuscript.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

(e.g., that which accompanies residence in the same neighborhood or community).

One of the more remarkable demographic trends of recent decades is a reduction in household size, a trend that has been noted for persons of all ages and in numerous countries. Although a number of underlying factors help explain this trend, one of particular importance with respect to older people is a post-World War II trend toward having fewer children with whom to coreside. As illustrated below, however, this trend has recently reversed (or will soon reverse) in many industrialized countries. Household structure (or "living arrangements") is thus in part a consequence of patterns of kin availability and is the second major topic addressed in this chapter.

The third and final topic addressed is the spatial proximity of elderly and their kin, especially their adult children. Throughout, an effort is made to survey, albeit selectively, theoretical, methodological, and empirical contributions to the relevant literature. Some attention is also devoted to enumerating existing data sources that figure prominently in actual (or potential) research.

KIN AVAILABILITY

Demographers and other social scientists have a long history of interest in kinship. Lotka (1931), in an article on the relationship between mortality and orphanhood, developed methods for determining the probability that a person at a given age has a living parent. More generally, demographers have devoted efforts to describing kinship patterns and to the formulation of models that relate kin patterns to underlying demographic forces. These issues are the concern of the present section.

Kin availability is of particular importance with respect to the elderly, since members of the kin group constitute a resource pool—with "resources" construed broadly to encompass space (i.e., shared residential space), time, and money—on which elders in need of care or assistance can call. The kin group also represents, of course, a set of potential claimants on similar resources held by the elderly. Thus the composition of a kin group defines a complex set of potential interpersonal linkages that are of substantive interest.

Conceptual Issues

Inclusiveness of Measures

Before attempting to measure patterns of kin availability, it is necessary to establish the scope of the term "kin." Our concern is with kin groups

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

containing at least one elderly person. The specific labels attached to linkages between individuals differ according to who in the kin group is taken to be the reference person, or "ego"—for example the relationship between two individuals may be "uncle" or "nephew" depending on who is ego. Furthermore it is clear that in common usage of the term, a person is likely to simultaneously be part of more than one kin group; this is especially true of married people.

The range of relationship ties encompassed in the term "kin group" will differ across studies and across disciplinary lines. It should also be noted that descriptions of kin patterns are rarely of purely intrinsic interest; rather, of interest are the roles occupied, and the functions performed, by individuals in the network, and the dynamics through which these roles and functions develop. As has been demonstrated in numerous sociological and anthropological studies, the nature, composition, and functioning of kin groups differ considerably across cultural, ethnic, and/or racial lines, and the inclusiveness of kin groups can extend well beyond ties defined by blood and marriage (see, for example, the classic study of urban blacks in the United States by Stack, 1974).

Nevertheless, this chapter adopts a narrowly demographic perspective, limiting its attention to relatives defined with respect to blood or marriage. This perspective reflects a desire to relate kin patterns to underlying demographic processes. It also reflects a concern with the ability to generalize from empirical findings, which leads to an emphasis on research based on population data or on data from large-scale surveys. Available data, in turn, tend to provide only a limited range of information on the composition of kin groups. In fact, the following discussion is generally restricted to immediate relatives—parents, siblings, and children—and only occasionally extends to more distant kin found along direct lines of ascent or descent, such as grandparents and grandchildren.

Among the married elderly, the spouse is possibly the most important member of the kin group. Moreover, the death of a spouse is a key life course transition, experienced in most cases late in life, and is often accompanied by major shifts in economic circumstances. Nevertheless, this chapter devotes very little attention to spouses: data availability is considerably more problematic for kin relationships other than spouse; modeling issues are more complex for relationships such as child, parent, and sibling (where the number, not just the existence, of such kin is itself a variable); and the consequences considered here—coresidence and proximity—are the result of very different underlying processes for spouses and for other members of kin groups.

Even with a narrow demographic conception of kinship, observed shifts or long-term trends in patterns of divorce, remarriage, and childbearing imply that the kin-availability patterns of successive cohorts of elderly will

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

become more and more complex. The "blended family" so prevalent in some contemporary societies, in which one or more spouses/partners are in second (or higher-order) marriages containing children from two or more marriages, is an example of a phenomenon that will increase the prevalence of half- and stepchildren and siblings (among others), in addition to own and in-law children and siblings, in the kin groups of the elderly. These considerations raise issues of measurement and analysis that have scarcely been addressed in the literature on old-age living arrangements and family support behavior.

Demographic Forces Underlying Kin Availability

It is evident that the essential demographic forces that determine the size, age structure, and gender mix of a population—patterns of birth and death, by age—underlie patterns of kin availability as well. To fertility and mortality rates we must add patterns by age of marriage, divorce, and remarriage as well, since marriage creates linkages among the parents, siblings, and offspring of married couples, whereas divorce and later remarriage modify and further widen the network of kin and the interrelationships between individuals in the kin group.

At a specified age, the size and composition of one's kin group provide a partial record of one's demographic history. The survivorship and ages of parent(s), for example, reflect the parents' ages (and, therefore, their relative ages) at ego's birth—for any selected ego—as well as the specified age of ego and the relevant age- and sex-specific mortality rates. The same is true with respect to ego's spouse. The presence in ego's kin group of living siblings with specified characteristics depends on parental fertility history, ego's position in that history, and the relevant patterns of survivorship, while the number, ages, and gender mix of living children reflect ego's own childbearing history as well as survival patterns within the relevant cohorts. Finally, the presence and age/sex attributes of any in-laws, stepsiblings, stepparents, and/or stepchildren, all reflect the history of divorce, remarriage, and childbearing within specific marriages.

Thus, indicators of the size and composition of kinship networks will be influenced by changes in age- and sex-specific mortality, fertility, marriage, divorce, and remarriage rates, as well as by age differences between spouses. It is beyond the scope of this chapter to attempt any characterization of the many and complex trends in any of these underlying demographic processes. Consider, however, an array of indicators of the size and composition of kin networks. For a randomly selected older woman, for example, we might wish to measure whether a spouse is present; the number of surviving sisters (brothers); the number of surviving married (unmarried) daughters (sons); the number of surviving parents; and so on. Analy-

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

sis of the effects on each such indicator of changes in any of the underlying demographic forces—an upward or downward shift in fertility rates at all ages, or movement to the right of the age curve of first marriage, or a lowering of sex-specific and age-specific death rates, for example—leads to the following general conclusions. First, each of the underlying demographic forces will have consequences for many, if not all, of the selected indicators. Second, each indicator will tend to be affected by changes in many, if not all, of the underlying demographic forces. Finally, the net effect on any indicator of a change in any one of the underlying demographic factors will depend on levels and trends in all the other factors, as well as any interactions among them. Complexities of these sorts have been addressed by using a variety of demographic models; these models are discussed in more detail below.

Empirical Issues

Aggregate Measures of Kin Availability

Population aging is an aggregate phenomenon, one revealed by a change in the age structure of a population. It is possible to construct simple measures of one aspect of kin availability—the availability of children—in the aggregate, by using population data. An early contribution to the modern literature is an often-cited study by Kobrin (1976), who showed that the path of a simple measure of kin availability—the ratio of "daughters" (women aged 35-44) to their unmarried "mothers" (widowed and divorced women aged 55 and over)—closely paralleled that of average household size in the United States during the period 1890-1973.

Measures based on aggregate data, such as those used by Kobrin, have certain limitations. For example, it is impossible to align population counts exactly by age so that they delineate distinct generations. Furthermore, aggregate measures provide information only about the average of kin-availability patterns, omitting important features such as the extent to which the elderly are childless or to which adults are without living parents. Yet aggregate measures have the obvious virtues of demanding only minimal information—population counts by age and sex (and, in the case of Kobrin's series, marital status)—and of showing clearly the relationship between population aging and average kin-availability patterns.

Figure 5-1 illustrates the time path of a variant of Kobrin's index (in particular, its reciprocal) for selected major regions of the world, using regional groupings and data produced by the Population Division of the United Nations.1 The underlying data reflect actual data (including esti-

1  

The data are extracted from the series "Sex and Age 1950-2025" and are described fully in the United Nations (1993).

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Figure 5-1 "Mother-daughter" ratio for selected regions of the world, 1950-2025.

mates) for most countries through 1985 or 1990, and the United Nations' medium variant projections thereafter. The "mother-daughter ratio" shown is the ratio of women 65 and older to women 25 years younger, a rough approximation of the mean length of a generation.2

In both the northern European countries as a whole and North America (Canada and the United States), a distinctive pattern can be seen, in which the mother-daughter ratio has climbed steeply for much of the period 1950-

2  

Specifically, the ratio is calculated as

where nx is the number of women in the indicated age group, and N65+ is the total number of women 65 and older. That is, in each of four "mothers" age groups, the ratio of that group to women in the 5-year age group 25 years younger is calculated; then the weighted sum of these four ratios is obtained, with weights representing the proportions that the mothers' age range represents, relative to all women 65 and older.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

1980. During the years 1990 through 2005, the upward trend is expected to reverse, although in neither region is the ratio expected to fall to levels as low as those seen in 1950; after 2005 the trend is projected to reverse once again, with an even sharper increase to unprecedentedly high levels by 2025, when the projected series ends. These curves clearly portray the late-life consequences of being in generations that successively produced baby "booms" and baby "busts."

The northern European and North American patterns just described are not repeated in the other parts of the world shown in Figure 5-1. However, a pronounced rise in the mother-daughter ratio can be anticipated in eastern Asia after 2010, paralleled by a gentler increase in South America. Africa, with an overall pattern of sustained high fertility, exhibits a near-level mother-daughter ratio throughout the period of historical and projected data.

Measures Based on Individual-Level Data

Censuses and sample surveys have for the most part failed to produce information with which to describe kin-availability patterns, since they have usually focused their questions on individual respondents and the households in which they live. In a typical household-based survey, one or more individuals serve as respondents, generally either enumerating the other individuals with whom they live or providing responses that summarize the composition of the household. This generates information about only the coresident parts of kin groups.

Less systematic is the availability of information on the full set of living kin for even a limited set of kin types. The existence of a spouse, even if not coresident, can often be inferred from survey items on marital status, although nonmarital unions may be poorly measured. Problems grow as we consider children, the category of most interest for studies of the elderly population. The crudest measure of available children is number of children ever born. Several existing data sources include information on the preferred measure, a count of living children, while some go further, eliciting information on each living child. Even less common are questions pertaining to siblings, grandchildren, and more distant relatives.

A Sample of Microdata Sources

As part of its monthly Current Population Survey (CPS), in which a large rotating sample of U.S. households is interviewed, the Census Bureau has included questions on kin availability, asked of all adults in sample households, on four occasions: July 1987, June 1988, November 1989, and June 1991. In these surveys, respondents were asked how many living natural parents, brothers, sisters, and children they had. Some descriptive

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

material from the 1988 survey is reported in Woodrow (1990), whereas findings based on the 1989 survey can be found in Woodrow and Peregoy (1991). From a methodological standpoint, these data (and other data generated by comparable survey items) confront the analyst with the phenomenon of "multiplicity sampling." That is, individuals classified according to a particular kin relationship are represented in the sample in proportion to their multiplicity in the population. For example, "sibships" of size four (represented by a respondent who has three siblings) have four times the chances of being sampled as do only-child sibships. Researchers who wish to present results in which the kin group is the unit of analysis must adjust sampling weights accordingly (for a fuller discussion of this issue, see Woodrow and Peregoy, 1991). These CPS files, which have the potential to support a range of interesting demographic analyses, have received little attention to date.

Highlights of other major sources of U.S. data on the members of kin groups are presented here in brief. The 1984 Supplement on Aging (SOA) to the National Health Interview Survey contained only limited information on kin: counts of living sons, daughters, brothers, and sisters. A subset of these respondents was reinterviewed in 1986, 1988, and 1990 as part of the Longitudinal Study of Aging (LSOA; see Kovar et al., 1992). The kinship questions were not repeated in 1986, although the number of son/daughter questions was repeated in 1988 and 1990.

Shanas's 1962 and 1975 surveys of the aged in the United States obtained, for each living child, the sex, marital status, birth order, work status, and (if not coresident with respondent) distance (expressed in travel time) from respondent (Shanas et al., 1968; Shanas, 1982). 3 The enumeration of a full roster of living children has also been accomplished in the National Long-Term Care Survey (NLTCS) of 1982, 1984, and 1989. In the NLTCS moreover, additional items were obtained for each child: their age, the presence in their household of minor children, and several indicators of each child's helping activities on behalf of the elderly respondent. A limitation of the NLTCS is its restrictive coverage: only individuals with long-term functional limitations were selected for interviewing.

The 1987-1988 National Survey of Families and Households (NSFH; Sweet et al., 1988) used a large (N approximately 13,000) nationally representative household sample, and provided an extensive array of cross-sectional measures of the kin networks of its respondents.4 For adult children

3  

The questionnaire used by Shanas in 1962 was also administered to samples drawn in Denmark and the United Kingdom; extensive results from the three surveys are reported in Shanas et al. (1968).

4  

A follow-up interview of NSFH respondents was conducted in 1992.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

outside the household and for living parents, information on age, sex, marital status, and distance from respondent is available; for parents, the respondent's assessment of parental health is also provided. Only limited information on respondents' siblings, however, was obtained. The spouses of married respondents were asked a parallel set of questions, thus allowing analysis of kin networks containing in-laws and stepchildren. The NSFH has been used in numerous studies on kin relationships involving the elderly, some of which are cited below.

The Panel Study of Income Dynamics (PSID) is an ongoing annual survey of a sample originally containing about 5,000 families in 1968. Individual household members are tracked and interviewed if they depart from a previous sample household, so that over time the sample has grown to include about 7,000 families (Hill, 1992). In 1988, supplementary questions were added to the questionnaire, producing data on the existence and characteristics of nonresident parents and parents-in-law, and on time and money resource flows between the respondent's household and nonresident relatives. Analyses of the resource-flow data have recently begun to appear (Altonji et al., 1992; Furstenberg et al., 1993; Hill et al., 1993).

The Health and Retirement Survey (HRS) is a prospective longitudinal study of a cohort of people approaching retirement age, the first interview of which took place in late 1992 (Juster, 1992). The HRS will provide, albeit for a sample restricted to the age range 51-61 in 1992, the most extensive array of information on kin networks for a large, representative household sample in the United States to date. For all living children, researchers will know age; sex; educational, work, and marital status; own children (i.e., respondent's grandchildren); and (if not coresident) distance from respondent. For all siblings and siblings-in-law, age, sex, and marital status will be recorded; for as many as four each of siblings and siblings-in-law, additional items on work, financial status, and household arrangements will be obtained.5 Similarly detailed information on each of the respondents' living parents and parents-in-law is also being collected. A parallel longitudinal study of the "oldest-old" is also planned, with many questionnaire items replicated from the HRS and a first wave of interviewing scheduled to occur in late 1993. This Asset and Health Dynamics (AHEAD) survey will, like the HRS, provide unusually detailed information about the kin networks of the elderly.

The preceding survey indicates that existing and anticipated data from surveys of the U.S. elderly offer a variety of detail on respondents' kin networks. While some offer an extraordinary degree of detail, none are

5  

If the respondent has more than four living siblings (siblings-in-law), interviewers select a random subset of four, for which the additional items are recorded.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

complete. For example, none of the data sources reviewed provide information on linkages across kin groups, e.g. information on the in-laws of the respondents' married children.

In addition to the sources discussed above, there are several other existing public-use data sources containing at least some information with which to study kin patterns. There are also several sources of microdata from other countries, particularly in Europe and Asia. Many of these additional data sources are used in applied studies cited later in this chapter; readers interested in details of the data sources should consult the references cited.

Selected Results from Microdata

Information about kin groups containing elderly can be provided by elderly survey respondents who enumerate their living kin, or it can come from nonelderly survey respondents who report the presence of elderly (e.g., parents) in their kin networks. The way in which the data are collected influences the way in which kin networks can be portrayed. Descriptive data on kin networks, drawn from data sources such as those enumerated above, have appeared in several places. The following discussion concentrates on findings from the most recent sources.

A time series of information on kinship patterns for the United States can be assembled by combining data from several surveys. Crimmins and Ingegneri (1990) combined Shanas's 1962 and 1975 National Survey of the Aged (NSA) data with data from the 1984 SOA. Some of their findings, further augmented by data from the 1987-1988 NSFH, are displayed in Table 5-1. These data indicate that between 1962 and 1975, the proportion of elderly with no living children grew, while the proportion with many living children fell. From 1975 to 1984, the percentage childless fell, while the percentage with large numbers of surviving offspring rose slightly.6 As noted before, these figures tell a story of the demographic histories of successive cohorts of elderly.

Table 5-1 presents data provided by elderly respondents. Such data can tell us, for example, the distribution by number of children (including zero) of kin groups containing elderly parents. What these data cannot reveal is the extent to which elderly parents appear in the kin groups of the nonelderly.

6  

Comparisons between figures for 1984 and 1987 must be made with great caution because of limits on their comparability. As noted in the table, the 1987 figures pertain only to older women, whereas the earlier figures are for older men and women. Furthermore, the NSFH used a very detailed sequence of questions about children, obtaining separate counts of biological, step- and adopted children; this questionnaire detail may have produced higher (and, presumably, more accurate) levels of reported counts of children than did previously administered surveys.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

TABLE 5-1 Percent Distribution of Population by Number of Surviving Children, Persons 65 and Older, United States (in percent)

Number of Children

1962

1975

1984

1987a

0

18

21

14

21

1

15

20

18

16

2

17

21

25

22

3-5

27

28

31

}41

6+

14

10

9

 

Total

100

100

100

100

a Women 65 and older only.

SOURCE: Derived from data presented in Crimmins and Ingegneri (1990) and Wolf et al. (1991b).

Table 5-2 considers the issue of kin groups containing elderly parents and their children, but uses data provided by adult respondents in the NSFH, of all ages. Table 5-2 should be read by column; in each column we find first the percentage of an age group with a living mother aged 65-84, classified by number of siblings, then the corresponding percentages with a mother 85 or older, followed by those whose mother is either under 65 or dead. The latter group constitutes a majority of all three age groups shown in the table. Of the three age groups shown, people 40-64 years old have the highest percentage with a living elderly mother, 44.6 percent. It is interesting to note that among those in this age group with a living mother, a substantial proportion has no siblings (11.7 percent). Children without siblings are likely to bear a larger burden of parental-care than those with siblings.

Note that whereas Table 5-1 tells us that 16 percent of women 65 and older have exactly one living child, Table 5-2 tells us that only about 4 percent of the population simultaneously has no living sibling and a living mother 65 or older. That is, the sibships of size one that are attached to 16 percent of older women represent only about 4 percent of all extant only-child sibships. These distinctions, which illustrate the importance of the ''perspective" from which kin relationships are examined (Freedman et al., 1991), must be borne in mind when considering the distribution of familial links within the population, and must be made clear when presenting data on the size and composition of kin groups.

As mentioned earlier, changes in cohabitation, divorce, remarriage, and childbearing lead to changes in the nature of kin networks, lending prominence to distinctions involving half- and stepsiblings/children/parents. Over

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

TABLE 5-2 Distribution (in percent) of Adult Population by Age, Existence and Age of Mother, and Number of Siblings, United States, 1987

 

Age Group

Age of Mother

19-39

40-64

65+

Total

65-84

0 siblings

2.7

10.5

0.1

5.0

1 sibling

1.9

7.1

0.3

3.5

2 siblings

2.5

7.5

0.1

3.9

3 siblings

1.5

4.8

0.1

2.4

4+ siblings

3.4

10.0

0.2

5.3

85+

0 siblings

0.0

1.2

1.4

0.7

1 sibling

0.0

0.9

0.6

0.4

2 siblings

0.0

0.8

0.6

0.4

3 siblings

0.0

0.4

0.3

0.2

4+ siblings

0.0

1.3

1.3

0.7

Other (mother under 65, or dead)

87.9

55.4

95.0

77.7

Total

100.0

100.0

100.0

100.0

 

SOURCE: Wolf et al. (1991b).

time, we can expect kin networks to become more extensive, more diffuse, and characterized by possibly weaker ties to a broader array of people. The behavioral consequences of such patterns, with respect to coresidence, other resource flows, and other types of interactions, represent an issue largely unexplored so far, due in part to data shortcomings. There is a need to develop new survey instruments that are sensitive to a broad set of kin and kin like distinctions, as well as a need for research that will uncover the substantive importance of such distinctions with respect to their behavioral consequences. An additional challenge, already present given available data, but likely to grow in salience with better data on more complex kinship networks, is to develop new ways to summarize and display information about the composition of kin groups.

Models for Analyzing Kin Distributions

In view of the deficiencies of available data on kinship patterns and the need to project future kin patterns, the development of models of kin networks has been an area of considerable activity among demographers in recent years. The modeling efforts can be grouped under three headings: analytic models, macrosimulation models, and microsimulation models.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Analytic Models

Analytic models employ functional transformations of demographic parameters (e.g., age schedules of fertility and mortality) to derive quantities of interest such as indicators of the presence or number of kin. Brass (1983), for example, relates mean household size to patterns of mortality, fertility, divorces, and ages of household formulation and dissolution. Other models have addressed a broader range of kinship indicators, the best-known approach being that of Goodman et al. ("GKP"; 1974, 1975). GKP derive expressions for the expected number of female kin for several ascending and descending generations and degrees of removal from ego, by age of ego, and do so using a minimum of inputs, namely, age schedules of fertility and female mortality. A limitation of this model is that the kin are not themselves classified by age, nor does the model produce the frequency distribution underlying the expectations. Moreover, since information on the variance of number of children is not used in the model, the frequency distribution of number of siblings (and other lateral kin) is distorted (Keyfitz, 1985). Goldman (1986) has used the GKP equations to investigate the consequences for kin counts of recent mortality declines in Korea.

Another class of analytic kin models is based on branching processes (Pullum, 1982). These models can be used to produce full frequency distributions of kin numbers, but generally dispense with the age dimension altogether, which limits their usefulness.

Macrosimulation Models

Macrosimulation is not, strictly speaking, a type of model but rather a way of performing calculations in order to derive results from models too complex to manipulate analytically (Keilman and Keyfitz, 1988). The distinctive features of this approach—of which the ordinary life table is a simple example—are (1) the representation of a population at the group rather than the individual level, (2) a deterministic application of probabilities or rates of transition between groups, and (3) an assumption that individuals are homogeneous within groups. Numerous studies have used the "multistate" life-table methodology to produce marital status life tables, which can be used to study variations in marital status distributions (i.e., the presence of a spouse) among the older population (for examples of this approach, see Bongaarts, 1989; Espenshade, 1983, 1987; Schoen and Nelson, 1974). Bongaarts (1987) has further developed the multistate life-table approach, introducing additional complexity by explicitly representing parity, the number, age, and sex of surviving children, and the coresidence status of children. Bongaarts's model thus represents a substantial step toward a full representation of the nuclear family kin group. Applications

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

of this approach can be found in Watkins et al. (1987), Wijewickrema (1987), and Yi (1986). Lee and Palloni (1992) use the family status life-table approach to investigate the relative contributions of changing patterns of fertility, mortality, age at marriage, and age differences between spouses to the prevalence of widowhood, average number of surviving children, and proportion of widows without surviving sons, for Korean women born in a broad range of cohorts (1890-1894 through 1970-1974).

Microsimulation Models

Analyses employing microsimulation techniques, by contrast to macrosimulation, represent individual members of a population while modeling population change as a result of individual-level stochastic processes. The basic approach used in microsimulation models is as follows: individuals from a real, or hypothetical, population are represented in a microdata file, and relevant events in the lives of individuals—their own birth, the birth of their children, and in some models, marital status changes—are determined to occur by use of Monte Carlo techniques. The input parameters used in these stochastic assignments include birth rates, death rates, marriage rates, age differences between spouses at marriage, divorce rates, and so on. By keeping track of the linkages between individuals in the micropopulation, it is possible to depict the kin network of sample members at any point in simulated time. Since a uniform list of descriptors can be attached to every such individual, kin networks can be described in considerable detail.

Kin models based on microsimulation have been used in a wide assortment of applications. For example, Howell and Lehotay (1978) used their AMBUSH model to analyze the effects of changes in demographic processes on kinship patterns of hunting and gathering societies. The SOCSIM model developed by Hammel, Wachter, and others (Hammel et al., 1976) has been used to investigate historical living arrangements found in preindustrial England (Wachter et al., 1978), to project kin patterns of U.S. elderly in the year 2000 (Hammel et al., 1981), and to study the consequences of fertility change for old-age support in China (Hammel et al., 1991; Lin, 1993), among other applications. Wolf's KINSIM model, which simulates a sample of "family trees," has been used in several studies of family support for the elderly (Wolf, 1988, 1990b; Tu et al., 1993). Other efforts of this type include the work of Ruggles (1987), whose MOMSIM model has been used in historical studies of household composition in the United States, and Smith (1987). For a survey of these and related efforts, see De Vos and Palloni (1989).

Table 5-3 provides some information that hints at the value of microsimulation in studies of the elderly. The upper panel, from Wolf (1988), is a tabulation

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

TABLE 5-3 Cross-Tabulation of Older Women and Working-Age People in Simulated Populations

 

Population Group

 

Working Age People Age 20-64

Elderly with Living Mothers, Age 65+

Older Women

Country

Number

Proportion

of Total

Number

Proportion of

20-64 Group

Number

Proportion

of Total

Netherlands, 1970-1980

 

 

 

 

 

 

No living mother 65+

21,442

0.698

 

 

 

 

Mother 65+;

 

 

 

 

 

 

Sibship = 5+

1,107

0.036

13

0.012

300

0.075

Sibship = 4

2,206

0.072

36

0.016

570

0.143

Sibship = 3

3,032

0.099

58

0.019

1,042

0.262

Sibship = 2

2,173

0.071

53

0.024

1,113

0.280

Sibship = 1

737

0.024

20

0.027

775

0.195

No living children

 

 

 

 

176

0.044

Total

30,699

1.000

180

 

3,976

1.000

Taiwan, 1985

 

 

 

 

 

 

No living elderly mother

15,483

0.696

 

 

 

 

With living elderly mother

6,763

0.304

247

 

 

 

Sibship = 5+

550

0.025

9

0.016

104

0.025

Sibship = 4

845

0.038

31

0.037

219

0.053

Sibship = 3

2,168

0.097

82

0.038

750

0.182

Sibship = 2

2,633

0.118

93

0.035

1,363

0.330

Sibship = 1

567

0.025

32

0.056

599

0.145

No living children

 

 

 

 

1,096

0.265

Total

22,246

1.000

247

0.011

4,131

1.000

 

SOURCES: Wolf (1988); Tu et al. (1993).

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

of individuals in a simulated population corresponding to the hypothetical stable population implied by birth and death rates observed in the Netherlands around 1970-1980. The simulation that produced these numbers employed the assumption that a single set of fertility rates pertains to women of all parities and thus, like the GKP model, may distort the frequency distribution of siblings. The lower panel, from Tu et al. (1993), depicts the hypothetical population implied by mortality and parity-specific fertility rates observed in Taiwan in 1985. In each part of the table three groups are represented. Those aged 20-64 ("working-age people") are tabulated according to whether they have a living older mother and the size of their sibship. Elderly with living mothers (in both examples, a small group) are also tabulated according to the size of their sibship. Finally, women 65 and older are tabulated according to the number of their living children.

Table 5-3 depicts two very different demographic regimes: in the Taiwanese example, more than a fourth of older women have no living children, whereas in the Dutch example less than 5 percent of older women are in this situation. In comparing the under-65 population that does have a living elderly mother, in the Dutch case the modal situation is to be part of a sibship of three (i.e., to share the potential burden of parent care with two siblings), whereas in the Taiwanese example the modal case is to have just one sibling with whom to share the potential tasks of parent care. These differences may be due, in part, to the fact that more realistic assumptions regarding fertility can be maintained by using the Taiwanese data.

It must be admitted that the data shown in Table 5-3 depict kin networks in simulated populations with about the same degree of detail found in Tables 5-1 and 5-2, which use survey data sampled from real populations. On the other hand, the NSFH data used earlier (and other data of similar completeness) are rare. Microsimulation offers an opportunity to fill in gaps in existing data—including, as an important special case, data for historical populations that have vanished far in the past (e.g., Ruggles, 1987, or Wachter et al., 1978)—although at the cost of accepting a considerable degree of abstraction in the model and of dealing with pervasive shortcomings of information about the parameters used as model inputs.

Issues in Modeling Kin Distributions

An assumption made in most models of kinship is that the various demographic events underlying kin patterns are independent. In other words, the models assume that the fertility of mothers and daughters is uncorrelated, that mortality is independent within and across generations, that divorce "proneness" is not correlated within or across generations, that mating is random with respect to kin patterns across potential mates, and so on. In

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

the remainder of this section we focus on the issue of independence, or lack thereof, of some of these demographic outcomes.

If demographic events are correlated along family lines, we would expect to see correlations of various counts of kin by type: for example, if women whose mothers had many children also have many children of their own, then there should be a positive correlation between numbers of siblings and numbers of aunts and uncles, and between numbers of siblings and numbers of children. Similar associations could be anticipated if longevity were passed from generation to generation. The converse, however, is not necessarily true: Pullum and Wolf (1991) demonstrate that there are "built-in" correlations among certain kin counts, such as number of children and number of grandchildren, even when all demographic events are independent. For example, Pullum and Wolf show that in a stationary population with independent fertility, the correlation between number of daughters and number of granddaughters is .71. Yet in such a population, the correlation between counts of kin in different lineages (e.g., sisters and daughters, or aunts and granddaughters) must be zero.

There exists ample evidence of correlations between kin counts that are higher than would be expected if fertility were independent within and across generations. Some of this evidence is presented in Table 5-4, which shows correlations between counts of living kin of selected types for women surveyed in the 1984 Hungarian microcensus, for U.S. women in 1984, and for Canadian women in 1985. In the Hungarian data, the correlations between counts of children and grandchildren are, in almost all age groups, well above the .71 that would be expected if fertility across generations were independent. In both the U.S. and the Canadian data, correlations between counts of siblings and offspring are distinctly nonzero; they are also, for unknown reasons, much higher in Canada than in the United States. Further evidence of this sort is presented in Table 5-5, which again uses the NSFH data, to illustrate the magnitude of correlations between selected indicators of kin network composition. As in Table 5-4, the correlations shown pertain to living kin and therefore reflect the combined effects of fertility and mortality. There is a fairly consistent pattern of significant positive correlation between numbers of siblings and numbers of children.

Correlated Fertility Across Generations Particularly for the younger age groups—19-59—the most likely explanation for the correlations shown in Table 5-5 is that there does exist a positive correlation between the fertility of mothers and that of their daughters. A number of papers provide evidence of such a correlation, including Anderton et al. (1987), Hodge and Ogawa (1986), and Danziger and Newman (1989). One unexpected aspect of Table 5-5 is the negative correlation between the number of living parents (which can only equal 0, 1, or 2) and the number of children, for those

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

TABLE 5-4 Selected Empirical Correlations Between Counts of Living Kin, Various Countries Results from Hungarian Microcensus of 1984: Older Women; Children and Grandchildren

Results from Hungarian Microcensus of 1984: Older Women; Children and Grandchildren

Age Group

Mean Number of Children

Mean Number Grandchildren

Correlation Coefficient

N

55-59

1.9307

2.5572

.7431

7,444

60-64

1.9870

2.9568

.8070

7,239

65-69

1.9442

3.0685

.8255

3,725

70-74

1.9007

3.0863

.8337

5,259

75-79

1.8434

3.0874

.8349

3,467

80-84

1.8339

3.2100

.8039

1,957

85-89

1.6773

2.9902

.7859

815

90+

1.6449

3.2757

.7043

214

Results from U.S. Supplement on Aging File of 1984: Older Women; Sisters and Daughters

Age Group

Mean Number of Sisters

Mean Number of Daughters

Correlation Coefficient

N

55-59

1.7015

1.5103

.0916

1,262

60-64

1.6394

1.3991

.0246

1,207

65-69

1.5717

1.2398

.0686

2,139

70-74

1.4638

1.1811

.0738

1,805

75-79

1.2357

1.1485

.0825

1,396

80-84

1.0265

1.0609

.0710

839

85-89

0.8375

1.1615

.0886

420

90+

0.5460

1.3664

-.1346

150

Results from the 1985 Canadian General Social Survey: Older Women; Siblings and Children

Age Group

Mean Number of Siblings

Mean Number of Children

Correlation Coefficient

N

55-59

4.1469

3.2030

.3138

315

60-64

3.6726

3.2887

.3266

311

65-69

3.5634

3.2593

.2407

401

70-74

3.4654

2.8560

.1874

507

75-79

3.3790

2.3742

-.0578

434

80-84

2.6147

2.6742

.0943

362

85+

1.8872

2.5524

.1663

357

 

SOURCE: Data from Pullum and Wolf (1991).

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

TABLE 5-5 Average Numbers of Living Kin and Correlations Between Kin Numbers, by Age, United States, 1987

 

 

Average Number of Living Kin

Correlation Coefficients

 

 

 

 

 

Parent, Sibling

Parent, Child

Sibling, Child

Age Group

N

Parents

Siblings

Children

 

19-29

1,795

1.79

2.86

0.87

-.07a

-.07a

.18a

30-39

1,889

1.56

3.00

2.08

-.07a

-.08a

.16a

40-49

1,025

1.14

2.58

3.11

-.08a

-.09a

.07b

50-59

759

0.57

2.49

3.69

-.05

-.00

.14a

60-69

738

0.21

1.94

3.05

.04

.02

.01

70-84

705

0.02

1.96

2.20

-.02

.03

.12a

85+

86

0.00

1.07

1.78

.00

.00

-.01

a p < .01.

b .01 < p <.05.

SOURCE: Wolf et al. (1991b).

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

under age 50. A young person with fewer than two living parents is likely to have had a parent who died relatively young. It may be that people who experience parental death while young tend to marry early, hastening their exposure to the risk of childbearing.

Familial Patterns of Longevity Negative correlations between numbers of parents and siblings, such as those shown in Table 5-5, may reflect intergenerational transmission of longevity. There is more direct evidence of such intergenerational correlations of mortality in other research. A number of empirical papers have addressed the "heritability of longevity" (e.g., Abbott et al., 1978). From these studies it has been concluded that any index of such heritability must be very small, although positive. Recent models of mortality incorporating unobserved "frailty" parameters fixed over the lifetime suggest that the upper bound on correlations between observed ages at death must necessarily be quite small, even with perfect inheritance of frailty (Vaupel, 1988).

Some research on familial correlations of age at death has attempted to identify the genetic component of survival by limiting environmental variation and/or by controlling genetic variation. For example, using a sample from an isolated Quebec population, Phillipe (1978) examined the correlations between age at death of parents and offspring, comparing them to correlations between spouses' ages at death. Finding a small difference between the two sets of correlations, the author concluded that the genetic portion of survival is probably near zero, and that correlations between ages of death of parents and children could be due to environmental factors. A recent study of Danish twins (Hougaard et al., 1992) finds evidence for dependence across twin pairs in survivorship, but concludes that the magnitude of the dependence is extremely small.

The sensitivity of measures of the composition of kin networks to variations in the degree of intrafamilial interdependence of fertility, mortality, or any other possible demographic events, has not been suitably explored in models of kin, through either analytic or simulation techniques. This remains as an area for considerable further research.

LIVING ARRANGEMENTS

A major thrust of recent research in the demography of aging has been to document and explain dramatic trends in the size and composition of households containing elderly people. The following sections survey major theoretical and methodological topics, and highlight the large empirical literature on this question. As pointed out by De Vos and Holden (1988), there are several alternative indicators of living arrangements that could be used in applied research. In the present discussion we limit our attention to

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

measures of the sort labeled ''family household type" by De Vos and Holden. These measures can vary in complexity, but tend to highlight differences among elderly living "alone" (including, among married elders, those living with spouse only), those living "with child(ren)," and "others"—a residual category that, in some instances, can be further subdivided. Thus the categories of the outcome of interest are defined so as to reflect kin availability. Furthermore, by taking this approach, we ignore related dimensions such as type of dwelling unit (e.g., private dwelling, congregate housing for the elderly, institutions) and household headship.

Conceptual and Theoretical Issues

Most recent research on living arrangements of the elderly is consistent with a conceptual framework of rational choice, in which the living arrangement actually observed is assumed to be chosen from a set of discrete alternatives and is assumed to be the one alternative valued most highly by the relevant decision makers. This framework has been developed by several writers, including Beresford and Rivlin (1966), Michael et al. (1980), and Schwartz et al. (1984).

In the abstract, we may suppose that at any point in time a given elderly person (or couple) is faced with an array of specific kin, each of whom represents a distinct opportunity for shared living arrangements. The full array of potential living arrangements includes, in addition, unrelated persons with whom the elder might coreside and a "null" option associated with living alone. For simplicity, we can suppose that each child, and possibly each sibling, in the elderly person's kin group constitutes a separate element in this set of opportunities. Attached to any child or sibling, there may of course be additional people such as that child's spouse, own children, and so on.

Suppose, for the moment, that only children and the null option appear in the opportunity set, and that the relevant characteristics of children are captured in the array Sk, for k = 1, ... , K (K being the number of children). In addition, the elderly person/couple is described by the array X. The choice framework asserts that decision makers are able to attach a value to each option, representing the level of well-being they attach to the option. There are two important aspects to this evaluation. First, we can suppose that each potential household in the choice set will produce optimal quantities of each of several household goods, with the array of outputs depending on time inputs of each household member and goods inputs. This "productivity" aspect of the determinants of living arrangement choices depends on the technology of household production, summarized in a household production function. The second important aspect concerns the distribution among household members of the output of household production (i.e., the

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

share of each household good produced that is actually consumed by each household member). Some household goods are like pure public goods (i.e., one person's consumption of such goods does not reduce the consumption of others). Other (perhaps most) household goods are, however, partly or fully private in nature; that is, one person's consumption of the good diminishes or rules out another's consumption of that good. Here, we subsume both the productive and the distributive functions in a combined value function that assigns a value to the live-alone option, V0 = f0(X), as well as to each coresident-child option, Vk = fk(XSk). In each case, the results represent the value of the indicated living arrangement as determined by the elderly person. In the case of the kth child, there is an analogous function that returns the value to that child of the particular living arrangement (i.e., of coresidence with the elderly parent). The manner in which a division of well-being between parent and child in a coresident household is determined is not addressed here. However, there is no reason in principle why a competitive, bargaining framework, of the sort applied to household decisions by Manser and Brown (1980) or McElroy and Horney (1981), cannot be used here as well. Thus, variations in children's willingness to "supply" (and to demand) coresidence are fully incorporated into the framework and can be viewed as operating through household production and/or the division of household output.

The question of what it is about different living situations that causes them to be valued more or less highly has been addressed by several authors, most comprehensively by Burch and Matthews (1987). Burch and Matthews note that each potential household living situation available to an individual conveys a distinct array of "component" household goods, including physical shelter; storage of property; domestic services (meals, laundry, cleaning); personal care (including, of special relevance to the elderly, assistance with everyday tasks including hygiene, locomotion, and so on); companionship (both social and sexual); recreation and entertainment; privacy; independence/autonomy; power/authority; and the benefits of economies of scale in consumption of any household goods (Burch and Matthews, 1987:499). The benefits of economies of scale can take the form of a larger share of personal money income left for discretionary uses, after paying for market inputs to the production of household goods.

It is clear that the size and characteristics of one's kin network represent constraints on the set of living arrangement choices facing an elderly person. If attributes of children systematically differ with respect to their productivity in household goods and/or the distribution of these goods across individuals in the household, then having more children makes it more likely that the option set contains a highly valued option. However, if there is substantial variation in household output according to the traits of individual household members, then it may be that the composition of the kin

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

network matters more than its size, as a determinant of parent-child coresidence. Monetary resources are an additional constraint on the choice of living arrangements.

Health and disability status can be thought of as a further type of constraint. However, it is perhaps more straightforward to think of health/ disability status as factors that operate through the technology of household production, represented here in implicit form by the value functions f0; ..., fK. Thus, an unmarried older woman who develops a severe functional limitation is expected to reduce her relative valuation of the living-alone option, since in this living arrangement, basic needs can be met only poorly or not at all, whereas coresidence with a child willing to provide a substantial amount of assistance with personal care will become more highly valued relative to living alone.

The preceding discussion is highly simplified but contains the elements from which a more comprehensive model could be developed. Siblings and other relatives can be added to the set of living arrangement choices. It is impossible to enumerate (or, at least, to represent in an empirical analysis) the set of nonrelatives (friends, paid companions or helpers, paying roomers, etc.) with whom an elderly person might potentially coreside. The residual "other" category of living arrangements can, nonetheless, appear in an empirical analysis; one approach is to assume that the elderly person considers only a restricted set of feasible and/or acceptable alternatives, and that the value function by which the "best" of such alternatives is ranked is adequately represented as a function solely of the older person's traits, Vr= fr(X).

Empirical Research on the Living Arrangements of the Elderly

Data and Their Methodological Implications

The previous section suggests a framework for empirical analysis in which some measure of living arrangements is related to the full array of attributes of available kin and to other factors thought to influence choices of living situations, through their effects on the production and distribution of household goods. A great deal of the existing literature is, however, based on a much narrower representation of the conceptually relevant factors.

Aggregate Data As discussed earlier, microdata containing information on the availability of kin are not widely available. One solution to the problem of data availability is to construct indices of both living arrangements and kin availability by using aggregate data, and several papers have taken this approach.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Michael et al. (1980) use U.S. state-level data for 1970, regressing a measure of the proportion of widows 65 and older who live alone on variables representing economic and demographic factors, including as a measure of kin availability a "mother-daughter" ratio constructed as the number of women 65 and older relative to the number of women aged 35-44. The estimated regression coefficient for "mothers per daughter" in this analysis is 0.07 (t = 1.92).7

Wolf (1989a), using very different data (country-level data for 16 countries, taken at a variety of years during the 1960s to 1980s) obtains exactly the same regression coefficient for mother-daughter ratio as Michael et al. (1980), 0.07 (t = 0.77), but in this case the regression coefficient cannot be judged to be significantly different from zero.8

A recent paper by Macunovich et al. (1992) further extends the aggregate data approach, using U.S. data for the period 1965-1990 for five age groups, 65-69 through 80-84, and 85 and older. Measures of the average number of living never-married and unmarried children for these age group/year combinations were developed by using estimates of cumulative fertility and projected survivorship within the respective cohorts. In a regression in which the dependent variable is the logistic transformation of the percentage of widows living alone, the regression coefficient on the variable "never-married children" is -3.441 (t = -5.04). Note that this variable (implicitly, never-married children per older widow) is the reciprocal of a variable such as the mother-daughter ratio used in the aforementioned studies; therefore, all three studies cited obtain consistent results, albeit without uniformly obtaining findings that would be judged statistically significant.

An intermediate use of data is found in Pampel (1992), who employed individual-level survey data as the unit of analysis but an aggregate measure of kin availability (a parent-child ratio analogous to the sex-specific measures discussed above) as a covariate. Pampel is able to use such a measure by exploiting variability across countries (10 western European countries) and time (twice yearly surveys over a 15-year period). Some of Pampel's findings indicate that as the ratio of older to younger people grows, the odds of living alone are significantly increased; however, this result is not robust to alternative specifications.

7  

The other variables in the model are (1) Social Security Income (SSI), measuring average Social Security payments to survivors; (2) mobility, a measure of the residential stability of a state's population; (3) the percentage of women 65-69 who have a college education; and (4) the percentage of a state's population that is black.

8  

The additional variables used in the regression reported are (1) the ratio of elderly females to elderly males, (2) per capita income, and (3) per capita housing stock—the latter two variables referring to the entire population of the country.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Microdata: Issues of Model Specification and Estimation As discussed previously, available data sources present a range of detail for the measurement of kin availability. At the minimum level of acceptable detail are those sources that record whether an elderly person/couple has living children or, more satisfactorily, their number. Similarly, there is a range of potential detail with which to construct measures of living arrangements, with most research being based on binary or polychotomous categorizations. Specification and estimation of multivariate models are constrained not only by the degree of detail contained in the data, but also by the empirical frequencies of categories of potential interest. Thus, for example, the category "living with sister(s) but not others" may be of theoretical interest but observed so rarely in even a large sample that it cannot be distinguished in a model.

Since logit models have been most widely used to study living arrangements, the following discussion makes explicit use of the logit approach. Many (but not all) of the suggestions made would pertain equally to other approaches, such as probit. As is amply documented elsewhere, there are technical problems associated with the logistic model that might limit its applicability to the study of elderly living arrangements; 9 these technical issues are not, however, addressed here.

The most basic approach relates the type of living arrangement to an array of variables including measures of kin availability, possibly classified by type. In almost all cases the explanatory variables used pertain to the elderly person (or couple); measures of the number of children (or sons, daughters, married sons, unmarried sons, and so on) can readily be viewed as attributes of the elderly reference person. Such an approach has been called a "structural" analysis (see Soldo et al., 1990) since it relates family structure to household structure. For a binary representation of living arrangements (e.g., "alone" versus "with others"), unknown parameters consist of a single vector of coefficients on the measured attributes. For polychotomous coding of living arrangements (e.g., alone versus with child(ren) versus other) the model becomes multinomial logit (MNL), with unknowns consisting of vectors of coefficients on the measured attributes for all but one category of the living arrangements variable.

One variant on the structural approach, applicable to polychotomous outcomes, consists of imposing restrictions on the "choice" probabilities that reflect variations in kin availability in the sample (Wolf, 1984). For example, if the outcome is the trinomial variable with categories living

9  

In particular, the multinomial logit approach entails an "independence from irrelevant alternatives" assumption that might be viewed as overly restrictive; see Ben-Akiva and Lerman (1985).

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

alone, living with children, and other, but the sample contains elderly both with and without children, then constraints can be imposed such that the probability that a childless elder lives with children is identically zero. This restricted-probabilities approach has also been used in Martin (1989), Soldo et al. (1990), and Tsuya and Martin (1992). The alternatives to the approach include (1) performing separate analyses for subsamples defined according to the structure of the kin network; and (2) using all observations in a single model, without imposing the zero-probability restrictions mentioned above. The first alternative will often be infeasible since with the sample sizes commonly available subsets defined according to criteria such as, for example, "no children, one or more siblings," "no children, no siblings," and so on, will be too small to permit separate analysis. The second alternative leads to inefficient estimates (i.e., coefficients with standard errors that are too large) and, in addition, to inappropriate predictions (i.e., an elder with no living children is assigned a positive probability of living with children).

If available data contain some information about each living child (or about each person in any specified category of kin) a more complex approach can be used, one that utilizes all available information in the data even if the number of children (or other kin) differs across observations in the sample. The way in which the data might be used will depend, in part, on whether (and how much) multiple coresidence—that is, coresidence of the elder and two or more children at the same time—occurs in the data.

Case 1: No Multiple Coresidence If multiple coresidence does not occur in the sample, then a straightforward MNL specification can be used. Let the "dependent" variable Y be an indicator of which child the older parent coresides with, with Y = 0 denoting all "other" living arrangements; the notation for other variables is as defined above. Then the probability that the parent coresides with child k is given by the expression

Notice that this expression can accommodate families with differing numbers of living children (K). Note also that while parental characteristics, X, are constant over alternatives, the child characteristics, Sk, vary over alternatives. The "other" category can be further divided, for example into the categories "alone" and "other." The latter approach is found in Wolf and Soldo (1988), who use equation (1) as the basis for their analysis of the living arrangements of older unmarried women. The results consist of an array of coefficients on the older women's characteristics representing their effects on the log-odds of living with "others" (relative to living alone), an array of coefficients on the older women's characteristics representing their

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

effects on the log-odds of living "with children" (relative to living alone), and an array of coefficients on child characteristics representing their effects on the log-odds that the parent and that particular child will coreside (relative to living alone). Interaction variables representing combinations of parental and child attributes can be added to this model; since they are child specific, they should be considered part of the array of child characteristics. Note also that this approach could be extended further so that individual children and individual siblings appeared as distinct potential coresidents, providing that it remained the case that there were no instances of multiple coresidence. The latter extension does not, however, appear to have been tried.

Case 2: Multiple Coresidence If simultaneous coresidence of elderly parent(s) with two or more children is sufficiently prevalent in a sample, then it may be possible to model the joint distribution of each child's coresidence behavior, and to identify parameters representing interaction effects between and among children's traits in the observed coresidence patterns. Suppose, for example, that there are just two children, to each of whom there corresponds a binary indicator of coresidence with the parent(s). If each coresidence indicator were independently determined, then we could derive a joint probability expression for any combination of the probabilities that Y1 = 0(1) and Y2 = 0(1) by multiplying together the respective binary logit expressions. The logit functional form has the advantageous property that the resulting product is conveniently in the MNL form. However, to accommodate the possibility that the two children's coresidence indicators are not independent, we add another term to the joint probability expression representing potential interaction effects, for example,

in which the interaction effect is represented by the term B2D12, where B2 is an array of coefficients and D12 is an array of variables representing interactions between the two children's characteristics or other variables describing the "duet" containing child 1 and child 2. There are four terms in the denominator of equation (2), one corresponding to each of the four possible coresidence patterns (neither child, child 1 only, child 2 only, both children). If the coefficient vector B2 is found to be no different from zero, then the children's coresidence probabilities can be judged to be independent.

In practice, samples will contain older parents with one, two, three, and more children. Coresidence with none, one, two, three, or more children may also be observed. The approach outlined above can be extended to the

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

more general case. It is necessary to enumerate all possible duets, all possible "trios," and so on; it is also necessary to derive variables capturing three-way interactions, four-way interactions, and so on. The number of conceptually permissible combinations becomes extremely large, so parsimony in the parameterization of the MNL probabilities becomes crucial. In order to keep analysis practicable, it may be necessary to limit the sample with respect to maximum size of kin groups and/or maximum number of recognized multiple coresidents.

The simultaneous logit approach just described has been applied by Ofstedal and Chi (1992), who analyze data from a recent survey of Taiwanese elderly. Ofstedal and Chi limit their analysis to a sample of 2,853 elders who coreside with zero, one, or two children. In this sample, multiple coresidence is fairly common: 20 percent of the elderly respondents are residing with two children simultaneously (the average number of living children in the sample analyzed is 4.61). The findings include three highly significant coefficients on "duet" variables, indicating that there are substantial interactions between individual children's coresidence behavior.

The preceding discussion pertains to "complete" kin sets, that is, to data containing at least some information about each member of a specified class of kin (e.g., children). Another type of data commonly used in research on elderly living arrangements comes from surveys in which the children of the elderly are respondents, providing information on the circumstances of their parents including coresidence status. Models developed for such data focus on the coresidence of a given parent-child pair, rather than whether or not coresidence with any child occurs. As an inevitable consequence of the way in which the data are collected, models of this sort tend to be richer with respect to their inclusion of relevant attributes of the child, but poor with respect to measures of the parent's attributes. Since the outcome is usually binary (child and parent coreside; other), binary logit or probit models are most often used in this research.

A Selective Survey of Findings Based on Microdata

The empirical literature on living arrangements of the elderly has grown rapidly in the last several years, so much so that it is impossible to present a comprehensive survey of the literature in these pages. The papers mentioned are all ones in which nationally representative samples were used, multivariate models were estimated, and at least some measure of current kin availability was included. These selection criteria exclude numerous papers of considerable interest, including several mainly descriptive studies of living arrangements among the elderly (see, for example, Link, 1987; Keilman, 1988; Schwartz, 1988; Wall, 1989; Grundy, 1992) and others using tabular approaches (e.g., Grundy and Harrop, 1992). Several otherwise

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

informative multivariate analyses using the variable "children ever born" as a measure of kin availability (e.g., Avery et al., 1989; Burr and Mutchler, 1992; Wister and Burch, 1983) were also excluded by the above criteria. Even allowing for these admittedly arbitrary devices by which the field has been narrowed, the following survey does not claim to be complete.

Cross-Sectional Results Most of the available studies using microdata are cross-sectional in nature. We first consider several examples of what earlier were classified as "structural" analyses, that is, models in which the living arrangement outcome distinguished among categories according to the presence or absence of relatives by type, and some measures of the type of kin available for coresidence are included as covariates.

Selected features of a sample of such papers are summarized in Table 5-6. These papers represent analyses of several countries (although the United States is most heavily represented) and a narrow range of time periods, mostly in the 1980s. In the columns labeled "category of dependent variable" are numbers indicating the coding scheme for the dependent variable; thus a study represented by a 0 and a 1 in these columns uses a binary dependent variable, whereas others use 3- or 4-category outcomes as indicated.10 All the studies shown happen to use logit (or MNL) as an estimation technique. The 0 category always represents the excluded category (i.e., the category for which logit coefficients are normalized to zero). The columns labeled "estimated effects of kin" summarize the estimated partial effects of kin availability in these studies, with an "x" indicating that the corresponding variable appears in the analysis without a statistically significant coefficient, while "-" and "+'' entries indicate that statistically significant coefficients with the corresponding signs were found in the analysis. In reading these entries, an entry such as "+;-" means that the indicated variable had a significant positive coefficient in the outcome category coded 1 and a significant negative coefficient in the outcome category coded 2. For example, the first study mentioned, Bishop's 1986 article, uses a binary dependent variable with "alone" coded as 1 and "with others" coded as 0. The estimated effect of "number of children" on the log-odds of living alone (relative to with others) is negative and significant. A more complicated example is Wolf's 1984 study of Hungarian women. Here, a 4-category variable is used, with categories "alone" (coded 0), "with child" (coded 1), "with other relative" (coded 2), and "with others" (coded 3). The

10  

Some of the studies have living arrangement coding schemes more complex than indicated in Table 5-6; the table does, however, attempt to fairly represent the selected aspects of the studies that are presented.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

TABLE 5-6 Summary of Findings from Structural Analyses of Living Arrangements of Elderly

 

 

Category of Dependent Variablea

 

Estimated Effects of Kina

Study

Country; Year

Groupb

Alone

With Spouse

With Child (only)

With Other Relative

With Others

Children

Sons

Daughters

Siblings

Bishop, 1986

United States;

UmW

1

 

 

 

0

 

 

 

 

1986

UmM

1

 

 

 

0

 

 

 

Soldo et al., 1990

United States; 1982

UmW

0

 

1

 

2

 

x;x

+;

 

Wolf, 1990a

United States; 1984

UmW

1

 

 

 

0

 

-c

Stinner et al., 1990

United States; 1981

M

0

 

 

1

 

+d

 

 

 

Wolf et al., 1990

Canada; 1985

UmW

0

 

1

2

3

+;-;x

 

 

x;x;+

Wolf, 1989b

Hungary; 1984

NmW

0

 

1

2

3

x;x;x

 

 

x;x;x

 

 

WidW

0

 

1

2

3

+;-;

 

 

-;-;x

 

 

DivW

0

 

1

2

3

x;x;x

 

 

x;x;x

Martin, 1989

Korea; 1984

All

0

1

2

 

3

-;xe;

 

 

 

 

Malaysia; 1984

All

0

1

2

 

3

x;+e;

 

 

 

 

Philippines; 1984

All

0

1

2

 

3

-;+e;

 

 

 

 

Fiji; 1984

All

0

1

2

 

3

-;+e;

 

 

 

Chan and

Malaysia;

M

0

 

1

 

 

+

 

 

 

DaVanzo, 1991

1988-1989

W

0

 

1

 

 

+

 

 

 

a For explanation of table entries see text.

b Div = divorced; Nm = never-married; Um = unmarried; Wid = widowed; M = men; W = women.

c Sisters only.

d Includes some cases with coresident nonrelatives.

e From separate binary analysis of "lives with child" outcome.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

estimated effect of "number of children" on the log-odds of living "with child" (relative to "alone") is positive and significant, while the estimated effect of "number of children" on the log-odds of living ''with other relative" (again, relative to "alone") is negative and significant.

Some of the studies shown have focused on the distinction between living alone and not alone (Bishop, 1986; Wolf, 1990a) whereas others have specifically identified living with child(ren) as a distinct type of living arrangement. However, in nearly all cases these research findings support the general conclusion that having more living children reduces the probability of living alone and increases the probability of living with children. This appears to be true in countries where coresidence of elderly with children is common (as in several Asian countries) and in countries such as the United States where coresidence is a minority phenomenon. Not shown in Table 5-6 (because it cannot be summarized easily in the format used there) is the study of Aquilino (1990), who uses NSFH data to analyze whether any children coreside with older parents, given that they have at least one unmarried adult child. In his results, it is not the number of children, but rather the characteristics of the pool of available children—especially age and marital status—that emerge as significant predictors of parent-child coresidence.

Analyses based on the restricted-probability variant of MNL described earlier also do not fit well in the format used in Table 5-6 since they do not necessarily produce coefficients directly representing the effects of kin availability. In Wolf (1984), for example, there are no coefficients representing the effect of having living children on the probability of living with a child (which is one of the outcome categories) since these effects are implicit in the probability expressions used in the estimation. It can be shown by using the results presented in Wolf (1984), however, that the differences in probabilities of each living arrangement type differ substantially according to whether or not there are living children. The same approach was used by Tsuya and Martin (1992), who employed data from a 1988 household survey conducted in Japan. Tsuya and Martin present a model based on five categories of living arrangements (alone; with spouse only; with unmarried children; with married children; with others), including observations with and without spouses, with and without unmarried children, and with and without married children, with appropriately constrained probabilities. Tsuya and Martin find several statistically significant "cross" effects; for example, the existence of unmarried children has a large negative effect on the probability of living with married children (-0.229), whereas the existence of married children has a smaller negative effect on the probability of living with unmarried children (-0.153).

Similarly, the "microanalytic" approach to modeling living arrangements, embodied in equation (1), does not necessarily produce coefficients

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

indicating the effect of kin numbers on type of living arrangement. Rather, this model produces a set of probabilities of coresidence with each individual living child (which can be added to produce a predicted probability of living with a child under the maintained assumption of no multiple coresidence). The microanalytic approach does produce coefficients representing the effects on coresidence probabilities of individual attributes that may vary across children. This approach has, apparently, been employed only in Wolf and Soldo (1988), who use Shanas' 1975 data (Shanas, 1982). The results in Wolf and Soldo (1988) indicate that the child's sex and marital status matter (unmarried children are more likely coresidents than married children; daughters are more likely coresidents than sons; yet unmarried sons are the most likely type of coresident child among all sex/marital status combinations), as does the number of younger sisters (which reduces a given child's likelihood of coresidence, other factors being the same). Wolf and Soldo conclude that it is the composition, more than the sheer size, of the available kin network that helps explain variations in living arrangements.

So far we have considered only research in which the entire kin network (however poorly or incompletely measured) of an elderly person/couple is the unit of analysis. Such analyses have used data provided by elderly respondents and have tended to focus on consequences of variation in the availability of relatives, by type, for living arrangement outcomes. A different type of analysis uses data provided by the child of an elderly person, in which (in all such analyses to date) the dependent variable indicates whether or not that child and the parent coreside. Here, in other words, we examine the marginal probability of coresidence for a given parent-child pair rather than the probability of coresidence with each of (possibly) several children (or with any child). An important distinction to bear in mind is that in an analysis of marginal probabilities, the existence of the kin category in question (here, children) is built into the data analyzed, although variables indicating the existence of other kin can appear as covariates. The natural focus of the analysis tends to be on the importance of particular attributes of children and their parents to the probability of parent-child coresidence; the inherent weakness of such analyses is that they usually control inadequately, if at all, for the characteristics of siblings, who represent a pool of alternative parent-child coresidence options.

Among the extant examples of models of marginal parent-child coresidence probabilities are several articles using Japanese data (Bumpass, 1990; Kojima, 1992a,b; Martin and Tsuya, 1991), and an analysis of the NSFH data presented in Wolf et al. (1991a). Both of Kojima's papers, as well as Wolf et al. (1991a), focus on married couples, and distinguish between coresidence with husband's and wife's parents in the dependent variables analyzed; also, in both papers, covariates are included that represent the existence of

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

siblings (and siblings-in-law), thus capturing some of the effects of variations in configurations of available kin. The model estimated in Wolf et al. (1991a) indicates that in most cases, the wife's elderly mother is a more likely coresident than the husband's. An exception to this, judged by predicted probabilities based on estimated coefficients, is the case in which the husband has no siblings but the wife does, and the husband's mother is in poor health but the wife's mother is in good health; in such circumstances, the husband's rather than the wife's mother is the more likely coresident. It should be noted, however, that in all cases, coresidence of either elderly mother with a married couple is a low-probability outcome.

A final model of this type is presented in Kotlikoff and Morris (1988), whose work is unique in that it uses data from paired interviews with elderly parents and one of their children. The drawback of the data is that they come from a small area sample of doubtful generalizability. Nonetheless, their study includes a better representation of parent and, simultaneously, child characteristics than is generally found in the literature. Of particular interest is the finding that the child's income has a positive effect on the log-odds that the parent lives alone, net of the (insignificant) effects of the parent's income. In virtually all the other studies cited here, in which income was included, only the parent's income was used in the analysis. In most such studies, it has been found that parental income is positively associated with living alone.

Models of Living Arrangements Transitions The availability of longitudinal data from several sources has led to a growing representation in the literature of papers analyzing transitions in the living arrangements of older persons. Some of the important multipurpose longitudinal data sources, such as the Survey of Income and Program Participation (SIPP), fail to include data on the existence of kin. Therefore, in keeping with the selection criteria used earlier, research on living arrangements based on the SIPP (e.g., Speare and Rendall, 1989; Mutchler, 1990) is not be discussed in detail. All of the studies cited use data from the United States.

An early analysis of living arrangement transitions is presented in Tissue and McCoy (1981), who used data from a sample of elderly welfare recipients interviewed in 1973 and 1974. Their analysis distinguishes between elderly living alone and those living with others, at baseline. Between baseline and follow-up, transitions due to arrival/departures of others, or of the respondent, can all be distinguished. A somewhat puzzling result (for which no interpretation is offered by Tissue and McCoy) is that among those living with others at baseline, the number of living children is positively associated with the respondents' moves out of the home occupied at baseline. It should be noted that the sample consisted of welfare recipients who experienced substantial income increases between the two inter-

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

views, as a consequence of the implementation of the Supplemental Security Income program.

Recently several studies of living arrangement transitions have appeared that use data from the LSOA. Of interest here are the findings concerning the effects of kin availability. Worobey and Angel (1990) use the 19841986 LSOA, which permits analysis of a single living arrangement transition for community residents at baseline, over a 2-year period. Their analysis is limited to unmarried elderly and consists of an ordered-logit model of a trichotomous variable measured in 1986 (0 = alone; 1 = with others: 2 = in institution). They find no effect of number of children on either of the two implied transition probabilities. Speare et al. (1991) also use the 19841986 LSOA, but include both married and unmarried elderly. In separate analyses of transitions from community to institutional arrangements, and of the probability of living with others in 1986 given the baseline living arrangement (which, in their model, implies four separate transition probabilities), they find no significant effects of number or sex composition of living children on transition probabilities. Spitze et al. (1992), who also use the 1984-1986 LSOA and consider only respondents unmarried at baseline, distinguish the following categories: alone, with children, with others, and in institution. Among those living alone in 1984, they find a significant positive effect of number of living children on the probability of a transition to the category of living with children, but no such effects for transitions to other categories. Among those living with children in 1984, no effects of number of children are found.

Recently, results have appeared based on the 1984-1986-1988 release of the LSOA. Angel et al. (1992) include married and unmarried respondents, and develop a model of living with others versus living alone in 1988. In this model, there is a significant effect of number of children (in 1984) on the probability of living with others in 1988. Although the model is not presented as a model of living arrangement transitions, it does (in one variant reported) include a control for living alone in 1986, from which transition probabilities can be inferred.

Assessment and Critique of Empirical Literature on Living Arrangements

The foregoing review of empirical literature is admittedly selective in its inclusion of papers and makes no attempt to provide a thorough survey of the findings reported in the selected papers. Instead, the intent has been to identify findings that relate to the central issues of this chapter, namely, the consequences of kin availability patterns for choices of living arrangements. In this narrow context, the following conclusions can be reached: First, observed living arrangements among the elderly are influenced by the

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

size and composition of their available kin networks. Having more children makes coresidence with children more likely. Second, however, it seems clear that the characteristics of individual children strongly influence the observed coresidence outcome. More attention could, and should, be devoted to discovering the influence on living arrangement choices of particular traits of children (as well as of other kin) and, equally important, of the net effects of a given child trait in the specific context of a kin network containing other children, each having particular traits. Methods that permit such analyses are more complex than those generally encountered in the literature, but they are available.

In connection with this need for increased use of what has been called the microanalytic approach to modeling living arrangements, it can be noted that existing data have been underutilized. Aquilino (1990), for example, reduces the child-specific information found in the NSFH into several summary indicators of the set of children as a whole (e.g., age of youngest child, number of sons, number of never-married children, and so on).

It is also possible, in some cases, to construct child-specific variables even if the underlying data are not collected in this way. In the 1984 SOA, for example, the variables "number of living sons" and "number of living daughters" can be found. A household roster, containing a ''relationship to respondent" code for every household member, is also part of these data. Therefore it can be determined whether the respondent lives with a daughter or a son. However, even with this limited information it is a simple task to create K arrays of child characteristic variables (the Sk arrays defined above), each of which contains variables indicating the sex of that child, whether (and how many) brothers that child has, and whether (and how many) sisters that child has. Finally, if there is a coresident son (daughter), one of the Sk arrays coded as son (daughter) can arbitrarily have associated with it an indicator of coresidence with the parent. Now, the MNL approach shown as equation (1) can be used with the rearranged data. This approach will work as long as there are not two or more coresident children, and it makes much fuller use of the data than the simpler structural approach.

Importance of Controlling for Income and Disability/Health Most of the papers cited above include some measure of health and/or functional limitation status, as well as income, for the elderly parent. Such variables are clearly theoretically relevant as suggested by the discussion of conceptual issues. Two points need to be made about these variables, however. First, if an analysis fails to control for health/disability and/or income, then biased estimates of the effects of kin availability variables may be obtained, because there is likely to be a correlation between family composition and income, and between family composition and health status. Higher-income people, for example, may have fewer children, and have higher incomes

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

both during childbearing years and in old age; also people with more children may also save less when young, leading to lower retirement incomes. Such considerations imply a need for a concern with omitted-variable bias in the estimated effects of kin structures on living arrangements.

A second point with respect to controls for health/disability status can also be made. Most studies have included measures of functional limitations in the form of simple counts of Activities of Daily Living (ADL) and/ or Instrumental Activities of Daily Living (IADL) that are difficult or impossible for the elders being studied. Yet there is some evidence that different specific types of functional impairment have differential impacts on living arrangement choices. In a study of the dichotomous outcome of living alone versus living with others, Wolf (1990a) addressed this question using Canadian and U.S. data. In the Canadian sample it was possible to control separately for whether respondents were unable to perform any of the following four tasks: meal preparation, money management, light housework, and grocery shopping. In the U.S. sample (from the 1984 SOA) an array of dummy variables was used to represent individually each of six ADL and six IADL limitations.

The results of these more detailed regressions are striking. In the equation based on the SOA, none of the indicators of ADL limitations was associated with living alone. On the other hand, difficulties with three of the IADL tasks—meal preparation, money management, and the use of the telephone—were found to significantly reduce the odds of living alone. In the Canadian equation, variables indicating the inability to prepare meals and to do grocery shopping were significantly associated with a reduced likelihood of living alone. These findings suggest the fruitfulness of a closer examination of associations between specific functional limitations and decisions regarding living arrangements.

Additional Methodological Problems In addition to the problems already discussed, a few others can be mentioned in brief. First is the issue of possible endogeneity bias in models of living arrangements. Some analyses have included variables representing the employment status of adult children with whom an older parent might coreside (Wolf and Soldo, 1988). This variable is potentially endogenous, since a child (e.g., a married daughter) might simultaneously reduce employment and begin a period of coresidence with her elderly mother. In the absence of a formal test of exogeneity, another strategy is to control, instead, for the child's potential market wage (i.e., for the exogenous determinant of the child's employment status; see, for example, Wolf et al., 199 a).

Finally, most of the cross-sectional research available thus far is based on samples drawn exclusively from the noninstitutionalized population. At successively older ages a higher proportion of the elderly population is,

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

however, found in nursing homes. There is some evidence that transitions into and out of nursing homes are influenced, in part, by family structure (see, for example, Garber and McCurdy, 1989, or Freedman, 1993). Consequently, great care should be exercised when attempting to infer global patterns of association between kin patterns and living arrangements from research derived from samples likely to be selective with respect to the availability of familial resources.

PROXIMITY

We turn now to the issue of spatial proximity of the elderly and their kin. Proximity to kin is related to kin availability both in a trivial way—kin proximity is impossible without kin availability—and, less trivially, through any effects of the size and composition of kin networks on the spatial distribution of kin from the vantage point of an elderly person. Proximity is also related to living arrangements, in the sense that nearby kin can in many respects provide the same benefits as coresident kin. In fact, among the motivations thought to influence both migration behavior and living arrangement choices is a preference on the part of the elderly for living near (but not with) their children, for having (in the often quoted phrase of Rosenmayr and Köckeis, 1963) "... intimacy at a distance." The elderly's preferences for residential propinquity have been confirmed in many studies (e.g., Day, 1991; for several early citations, see Troll, 1971:265-266). Moreover, parents (and children) might hold norms of responsibility for mutual assistance among family members, suggesting that spatial proximity is maintained in order to facilitate such normative behavior (for a review, see Mancini and Bleiszner, 1989). There is ample empirical evidence of the importance of proximity in the actual mobilization of such resources. For example, the distance between a parent and a child is strongly (and negatively) related to whether the child performs household chores and provides other assistance, and to the frequency of contact between the two (Crimmins and Ingegneri, 1990).

Since parents and their children nearly always coreside at the time of the child's birth, it is plain that parent-child proximity late in the parents' lives must be the consequence of the subsequent migratory behavior of both the parents and their children. However, the spatial distribution of kin also influences individual migration behavior over the life course. Therefore, kin proximity is both cause and consequence of migration behavior among network members. The following review considers empirical findings both on the importance of proximity in migration outcomes and on cross-sectional patterns of kin proximity, as they relate to migration behavior and other factors. Nearly all the relevant research has addressed the proximity

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

of elderly and their children, rather than the wider kin network, and this limitation is reflected in the following discussion.

Elderly Migration as a Means of Achieving Proximity to Kin

Migration of the elderly has generally been divided into two broad classes. The first, "voluntary migration," consists of moves toward a more attractive location, such as moves associated with retirement. Those making such moves tend to be "young-old," married, well off, healthy, and well educated, and to have moved often (Wiseman and Roseman, 1979; Parker and Serow, 1983; Meyer and Speare, 1985; Litwak and Longino, 1987; Speare and Meyer, 1988). The second type of elderly migration is associated with an inability to care for oneself without help. These migrants tend to be older, less well off, in worse health, and widowed. Such moves may occur in stages. Speare and Meyer (1988) assert that some elderly move in anticipation of later care needs. Moves, possibly toward children, may also take place when an older person actually becomes disabled or widowed— which may mean losing a caretaker (Parker and Serow, 1983; Litwak and Longino, 1987; Longino, 1990). A final type of health-related move, which may follow the development of major chronic disability, is into an institution (Litwak and Longino, 1987; Longino, 1990).

Empirical Research on Parent-Child Proximity

As the preceding discussion shows, parent-child proximity is valued (by the parents, at least), and has important consequences with respect to helping behavior and other forms of contact. The obvious interrelationship between proximity and migration has been noted. There has, however, been relatively little research in which proximity itself has been the outcome of interest. Even more rarely has research attempted to examine directly the extent to which proximity to kin results from migratory behavior by parents and/or children.

Important contributions to the literature on parent-child proximity and migration have been made by Warnes and colleagues. Law and Warnes (1982) analyze retirement decisions, using samples of migrants and nonmigrants taken in two areas of England during 1976. Law and Warnes demonstrate the importance of kin proximity both in the decision whether to migrate and in destination choices among migrants. For example, 21 percent of migrants stated that the main reason for moving was to be nearer to children or friends while a majority of the nonmigrants gave as a reason for not moving the desire to remain in proximity to children, friends, or other relatives (Law and Warnes, 1982:74).

In a subsequent paper, Warnes (1986) examines data from a somewhat

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

specialized sample, one containing 432 English couples (at least one member of which is of pensionable age), each of which has exactly two living children. The data measure current proximity and also provide unusually complete migration histories. Several regression results are presented, in which the dependent variable is the (log) distance between parents and children; a consistent result is that moves made by the child increase the distance between parents and children.

The NSFH data, already described, include information on distance to nonresident parents and children. These data have been used in proximity models by Clark and Wolf (1992) and by Rogerson et al. (1993). Clark and Wolf use only those NSFH observations in which the respondent is 60 or older. In using this subsample, two types of logistic regression analysis are performed: the first uses as its dependent variable an indicator of whether the respondent has any nearby children (i.e., an indicator of whether the nearest child is "near"; defined as living within 10 miles of the respondent); the second uses the same underlying data but treats each parent-child pair as a separate observation. In each such parent-child pair, the dependent variable is an indicator of whether the child in question lives near the respondent. In this case, each child's proximity to his or her parent is treated as an independent outcome. The first model allows for simultaneity in children's proximity outcomes, without explicitly specifying the form of any such simultaneity, whereas the second model implicitly assumes away any simultaneity, and thus is restricted to providing inferences concerning the correlates of a child's marginal probability of living close to his or her parent. Finally, in Clark and Wolf (1992) the proximity measure studied is a binary indicator: coresident parent-child households are included in the near category.

The findings reported by Clark and Wolf (1992) can be summarized as follows: parents with more resources—whether these resources are relative youth, high levels of educational attainment, or a living spouse—are less likely to live near a child than parents with fewer such resources. The more children an older respondent has, the more likely it is that he or she will live near at least one child. Having a married child—who presumably has competing obligations—reduces the likelihood of living near at least one child. Parents with never-married children are more likely to live near a child than those without, but this appears to be the result of the relative youthfulness of unmarried children, rather than their marital status, per se. Among the young-old, migrants are less likely than nonmigrants to live near a child, but by age 77 those who have moved within the last 5 years are more likely to live near a child than those who have not migrated. Children's characteristics are important in determining which child an elderly person lives near. In general, as children age and marry, they are less likely to live near their parents. However, a reversal occurs when the children them-

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

selves have children: having two or more children has a positive effect on proximity to parents that outweighs the negative effect of being married.

In contrast to the Clark and Wolf study, Rogerson et al. (1993) use NSFH respondents who have living parents and thus analyze proximity of parent-child pairs using information provided by the child. Instead of a categorical indicator of proximity, Rogerson et al. use the natural logarithm of distance (in miles) from the respondent to the relevant parent, reporting separate analyses of those with (1) both parents alive and married to each other, (2) mother only alive, (3) father only alive, and (4) both parents alive, but not married to each other. For the latter group, separate regressions for distance to father and distance to mother are reported. Rogerson et al. also exclude from their analyses respondents who are coresiding with their parent. Although the same variables are not always significant in the five regressions reported, several patterns emerge from the findings. For example, respondents currently enrolled in college generally live farther from their parent(s) than others, and residents of the West consistently live farther from their parents. Migration history plays a role here as well: the greater is the distance of the child's most recent move, the greater is the distance to the parent (consistent with the result reported in Warnes, 1986). Offsetting this is a timing effect: the longer the time spent at the current address, the less is the distance to the parent or parents. In three of the five regressions, age (of respondent) has a significant effect: older children live farther from their parents than younger children, on average. The latter finding agrees with those in Clark and Wolf. Other points of comparability are more difficult to establish, due to differences in model specification and variable coding.

Taken together, the Clark and Wolf (1992) and Rogerson et al. (1993) findings suggest an image of children who make geographic moves that remove them from close proximity to their parents; later, the parents make moves that again bring them into closer proximity. Whether parents indeed chase their children from location to location (and the prevalence of such a pattern, if it does occur) is an issue that must be addressed by using linked migration histories of parents and children; it cannot be addressed satisfactorily by using current proximity data.

A final comment concerns the use of proximity as a covariate. Several studies have used proximity as one of several covariates in multivariate analyses of outcomes such as intergenerational resource flows (Eggebeen, 1992; Hoyert, 1991) and the provision of care by adult children to elderly parents (Lee et al., 1993). The preceding discussion suggests that proximity is itself an outcome, one resulting from the combined behavioral choices of parents and their children, and one motivated in part by the desire to facilitate intergenerational contact and/or resource flows. If so, then proximity

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

should not be treated as an exogenous covariate, but rather as an outcome jointly determined with the resource flow being examined.

CONCLUSION

At various points in this chapter, assorted deficiencies in the existing literature concerning kin availability, living arrangements of the elderly, and spatial proximity of elderly and their kin have been noted. These are not repeated here; rather, it is noted that they constitute a familiar litany: there is a need for more powerful theories, better data, improved methodologies, and more comprehensive empirical research.

The foregoing review does suggest a number of areas in which future research could make useful contributions. For example, despite the limitations of existing data, the data in hand have not been utilized fully. For purposes of description, the development of new indices or other ways to represent the composition of kin networks—including, perhaps, graphical methods—would be of great value. From a methodological standpoint, existing data could be used to support more complex and detailed representations of the linkages between kin availability (as determinants) and living or proximity outcomes (as consequences). Indeed, complexity is a recurring theme in the literature on kinship patterns and their consequences, suggesting that efforts to promote and enhance the use of methods appropriate for complexities such as these would be of great value.

The development of theoretical tools for analyzing kin networks would also be welcome. Of particular value would be the development of models of decision making in diffuse kin networks. Microeconomic theories have, for example, been extended beyond simple models of individual choice to more complex models placed in a household setting, including models in which individual preferences may be in conflict. Decisions involving potential migratory behavior, in combination with coresidence or close spatial proximity of an elderly parent and one or more of their children, are more complex still and require consideration of a broader array of preferences and constraints.

I will close with a question not previously addressed, one that is entirely speculative in nature. In particular, trends over time in living arrangements (and, for that matter, in kin proximity) are readily analyzed as the consequence of other, more fundamental, trends such as changing family size (i.e., changing fertility patterns in earlier times), rising income, and possibly improved health. The observed trends could easily be thought to be the result of choices that reflect fixed preferences (or, alternatively, behavioral "propensities") in combination with a changing composition of the population making those choices. A larger, and considerably more difficult, issue is the question of feedback effects: As the age structure of the popu-

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

lation (an aggregate phenomenon) changes, becoming more aged, will the decision parameters of the elderly (or of their children) change? Will a greater prevalence of older people translate into altered norms governing intergenerational relations at the individual-, kin network level? These are questions that lend themselves less readily to empirical research of the sort given most attention in the present paper, but are offered as a potentially fruitful area for future research.

REFERENCES

Abbott, M., H. Abbey, D. Bolling, and E. Murphy 1978 The familial component in longevity: A study of offspring of nonagenarians. III. Intrafamilial studies . American Journal of Medical Genetics 2:105-120.

Altonji, J., F. Hayashi, and L. Kotlikoff 1992 The Effects of Income and Wealth on Time and Money Transfers Between Parents and Children. Paper presented at the RAND Conference on Economic and Demographic Aspects of Intergenerational Relations, Santa Monica, Calif.

Anderton, D.L., N.O. Tsuya, L.L. Bean, and G.P. Mineau 1987 Intergenerational transmission of relative fertility and life course patterns. Demography 24(4):467-480.

Angel, R.J., J.L. Angel, and C.L. Himes 1992 Minority group status, health transitions, and community living arrangements among the elderly. Research on Aging 14(4):496-521.

Aquilino, W.S. 1990 The likelihood of parent-adult child coresidence: Effects of family structure and parental characteristics. Journal of Marriage and the Family 52:405-419.

Avery, R., A. Speare, and L. Lawton 1989 Social support, disability and independent living of elderly persons in the United States. Journal of Aging Studies 3:279-293.


Ben-Akiva, M., and S.R. Lerman 1985 Discrete Choice Analysis. Cambridge: MIT Press.

Beresford, J., and A. Rivlin 1966 Privacy, poverty and old age. Demography 3:247-258.

Bishop, C.E. 1986 Living arrangements choices of elderly singles: Effects of income and disability. Health Care Financing Review 7(3):65-73.

Bongaarts, J. 1987 The projection of family composition over the life course with family status life tables. Pp. 189-212 in J. Bongaarts, T. Burch, and K. Wachter, eds., Family Demography: Methods and Their Applications. Oxford: Clarendon Press.

1989 The demographic determinants of the duration and incidence of widowhood. Pp. 55-65 in E. Grebenik, C. Höhn, and R. Mackensen, eds., Later Phases of the Family Cycle. Oxford: Clarendon Press.

Bongaarts, J., T. Burch, and K. Wachter, eds. 1987 Family Demography: Methods and Their Application. Oxford: Clarendon Press.

Brass, W. 1983 The formal demography of the family: An overview of the proximate determinants. Pp. 37-49 in The Family. British Society for Population Studies (OCPS) Occasional Paper, No. 31. London: Office of Population Censuses and Surveys.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Bumpass, L.L. 1990 A Comparative Analysis of Coresidence and Contact with Parents in Japan and the United States. NSFH Working Paper No. 41. Madison: Center for Demography and Ecology, University of Wisconsin.

Burch, T.K., and B.J. Matthews 1987 Household formation in developed societies. Population and Development Review 13(3):495-511.

Burr, J.A., and J.E. Mutchler 1992 The living arrangements of unmarried elderly Hispanic females. Demography 29(1 ):93-112.

Chan, A., and J. DaVanzo 1991 Living Arrangements of Older Malaysians—Who Coresides with Their Adult Children? Presented at the 1991 annual meeting of the Population Association of America, Washington, D.C.

Clark, R.L., and D.A. Wolf 1992 Proximity of children and elderly migration. In A. Rogers, ed., Elderly Migration and Population Redistribution. London: Belhaven Press.

Crimmins, E.M., and D.G. Ingegneri 1990 Interaction and living arrangements of older parents and their children: Past trends, present determinants, future implications. Research on Aging 12(1):3-35.


Danziger, L., and S. Newman 1989 Intergenerational effects on fertility: Theory and evidence from Israel. Journal of Population Economics 2:25-38.

Day, A.T. 1991 Remarkable Survivors: Insights into Successful Aging Among Women. Washington, D.C.: Urban Institute Press.

De Vos, S., and K. Holden 1988 Measures comparing living arrangements of the elderly: An assessment. Population and Development Review 14(4):688-704.

De Vos, S., and A. Palloni 1989 Formal models and methods for the analysis of kinship and household organization. Population Index 55(2): 174-198.


Eggebeen, D.J. 1992 Family structure and intergenerational exchanges. Research on Aging 14(4):427-447.

Espenshade, T.J. 1983 Marriage, divorce, and remarriage from retrospective data: A multiregional approach. Environment and Planning 15:1633-1652.

1987 Marital careers of American women: A cohort life table analysis. Pp 150-167 in J. Bongaarts, T. Burch, and K. Wachter, eds., Family Demography: Methods and Their Application. Oxford: Clarendon Press.


Freedman, V.A. 1993 Kin and nursing home lengths of stay: A backward recurrence time approach. Journal of Health and Social Behavior. 34(June):138-152.

Freedman, V.A., D.A. Wolf, B.J. Soldo, and E.H. Stephen 1991 Intergenerational transfers: A question of perspective. The Gerontologist 31(5):640-647.

Furstenberg, F.F., S. Hoffman, and L. Shrestha 1993 The Effects of Divorce and Remarriage on Intergenerational Transfers: A Preliminary Analysis of the PSID. Paper presented at the annual meeting of the Population Association of America, Cincinnati, Ohio.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Garber, A., and T. McCurdy 1989 Predicting Nursing Home Utilization Among the High-Risk Elderly. NBER Working Paper No. 2843. Cambridge, Mass.: National Bureau of Economic Research.

Goldman, N. 1986 Effects of mortality levels on kinship. Consequences of Mortality Trends and Differentials. Population Studies, No. 95. New York: United Nations.

Goodman, L.A., N. Keyfitz, and T.W. Pullum 1974 Family formation and the frequency of various kinship relationships. Theoretical Population Biology 5:1-27.

1975 Addendum. Theoretical Population Biology 8:376-381.

Grundy, E. 1992 The living arrangements of elderly people. Reviews in Clinical Gerontology 4:355-363.

Grundy, E., and A. Harrop 1992 Co-residence between adult children and their elderly parents in England and Wales. Journal of Social Policy 21(3):325-348.


Hammel, E.A., D.W. Hutchinson, K.W. Wachter, R.T. Lundy, and R.Z. Deuel 1976 The SOCSIM Demographic-Sociological Microsimulation Program. Research series, No. 27. Berkeley, Calif.: Institute of International Studies, University of California.

Hammel, E.A., K.W. Wachter, and C.K. McDaniel 1981 The kin of the aged in A.D. 2000: The chickens come home to roost. In S. Keisler, J. Morgan, and V. Oppenheimer, eds., Aging: Social Change. New York: Academic Press.

Hammel, E.A., C. Mason, K.W. Wachter, F. Wang, and H. Yang 1991 Rapid population change and kinship. The effects of unstable demographic changes on Chinese kinship networks. Pp. 243-271 in Consequences of Rapid Population Growth in Developing Countries, 1750-2250. New York: Taylor and Francis.

Hill, M.S. 1992 The Panel Study of Income Dynamics: A User's Guide. Newbury Park, Calif.: Sage Publications.

Hill, M.S., J.N. Morgan, and R. Herzog 1993 Intergenerational Aspects of Family Help Patterns. Paper presented at the annual meeting of the Population Association of America, Cincinnati, Ohio.

Hodge, R.W., and N. Ogawa 1986 Siblings and Family Size from Generation to Generation. NUPRI Research Paper Series No. 29. Tokyo: Population Research Institute, Nihon University.

Hougaard, P., B. Harvald, and N.V. Holm 1992 Measuring the similarities between the lifetimes of adult Danish twins born between 1981-1930. Journal of the American Statistical Association 87:417.

Howell, N., and V. Lehotay 1978 AMBUSH: A computer program for stochastic microsimulation of small human populations. American Anthropologist 80(4):905-922.

Hoyert, D.L. 1991 Financial and household exchanges between generations. Research on Aging 13(2):205-225.


Juster, F.T. 1992 Overview: Response Rates, Survey Length, Non-Household Record Linkages. Paper presented at the annual meeting of the Gerontological Society of America Symposium on Early Results from the Health and Retirement Study, Washington, D.C.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Keilman, N. 1988 Recent trends in family and household composition in Europe. European Journal of Population 3:297-326.

Keilman, N., and N. Keyfitz 1988 Recurrent issues in dynamic household modelling. In N. Keilman, A. Kuijsten, and A. Vossen, ed., Modelling Household Formation and Dissolution. Oxford: Clarendon Press.

Keyfitz, N. 1985 Applied Mathematical Demography, 2nd ed. New York: Springer-Verlag.

Kobrin, F.E. 1976 The fall in household size and the rise of the primary individual in the United States. Demography 13:127-138.

Kojima, H. 1992a Determinants of Coresidence of Married Couples with an Older Mother in Japan. Working Paper Series, No. 11. Tokyo: Institute of Population Problems, Ministry of Health and Welfare.

1992b Sibling Configuration and Coresidence of Married Couples with an Older Mother in Japan. Working Paper Series, No. 12. Tokyo: Institute of Population Problems, Ministry of Health and Welfare.

Kotlikoff, L.J., and J. Morris 1988 Why don't the elderly live with their children? A new look. In D.A. Wise, ed., Issues in the Economics of Aging. Chicago: University of Chicago Press.

Kovar, M.G., J.E. Fitti, and M.M. Chyba 1992 The longitudinal study of aging: 1984-1990. Vital and Health Statistics 1(28).


Law, C.M., and A.M. Warnes 1982 The destination decision in retirement migration. In A.M. Warnes., ed., Geographic Perspectives on the Elderly. New York: John Wiley & Sons.

Lee, G.R., J.W. Dwyer, and R.T. Coward 1993 Gender differences in parent care: Demographic factors and same-gender preferences. Journal of Gerontology: Social Sciences 48(1):S9-16.

Lee, Y.-J., and A. Palloni 1992 Changes in the family status of elderly women in Korea. Demography 29(1):69-92.

Lin, J. 1993 Changing Kinship Structure and Its Implications for Old Age Security in Urban and Rural China. Paper presented at the 1993 annual meeting of the Population Association of America, Cincinnati, Ohio.

Link, K. 1987 Household Trends in Eastern Europe Since World War II. Working Paper No. 71. The Hague: Netherlands Interuniversity Demographic Institute.

Litwak, E., and C.F. Longino, Jr. 1987 Migration patterns among the elderly: A developmental perspective. The Gerontologist 27(3):266-272.

Longino, C.F. 1990 Geographical distribution and migration. In Handbook of Aging and the Social Sciences, 3rd ed. New York: Academic Press.

Lotka, A.L. 1931 Orphanhood in relation to demographic factors. Metron 9:37-109.


Macunovich, D.J., R.A. Easterlin, E.M. Crimmins, and C. Macdonald 1992 Echoes of the Baby Boom and Bust: Recent and Prospective Changes in Intergenerational Living Arrangements of Elderly Widows in the United States. Paper presented at the annual meeting of the Population Association of America, Denver, Colo.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Mancini, J., and R. Bleiszner 1989 Aging parents and adult children. Journal of Marriage and the Family 51:275-290.

Manser, M., and M. Brown 1980 Marriage and household decision-making: A bargaining analysis. International Economic Review 21(1):31-44.

Martin, L.G. 1989 Living arrangements of the elderly in Fiji, Korea, Malaysia, and the Philippines. Demography 26(4):627-643.

Martin, L.G., and N.O. Tsuya 1991 Interactions of middle-aged Japanese with their parents. Population Studies 45:299-311.

McElroy, M., and M.J. Horney 1981 Nash-bargained household decisions: Towards a generalization of the theory of demand. International Economic Review 22(2):333-349.

Meyer, Judith W., and Alden Speare, Jr. 1985 Distinctly elderly mobility: Types and determinants. Economic Geography 61(1):79-88.

Michael, R., V. Fuchs, and S. Scott 1980 Changes in the propensity to live alone: 1950-1976. Demography 17:39-56.

Mutchler, J.E. 1990 Living Arrangements and Household Transitions Among Non-Married Older Women. Paper presented at the 1990 meeting of the Population Association of America, Toronto, Canada.

Ofstedal, M.B. and L. Chi 1992 Coresidence Choices of Elderly Parents and Adult Children in Taiwan. Paper presented at the annual meeting of the Gerontological Society of America, Washington D.C.


Pampel, F.D. 1992 Trends in living alone among the elderly in Europe. In A. Rogers, ed., Elderly Migration and Population Redistribution. London: Belhaven Press.

Parker, J.R., and W.J. Serow 1983 Migration of the elderly: A needed new dimension in research. In M.B. Kleiman, ed., Social Gerontology. New York: S. Karger.

Phillipe, P. 1978 Familial correlations of longevity: An isolate-based study. American Journal of Medical Genetics 2:121-129.

Pullum, T.W. 1982 The eventual frequencies of kin in a stable population. Demography 19:549-565.

Pullum, T.W., and D.A. Wolf 1991 Correlations between frequencies of kin. Demography 28:391-409.


Rogerson, P.A., R.H. Weng, and G. Lin 1993 The spatial separation of parents and their adult children. Annals of the Association of American Geographers 83:658-671.

Rosenmayr, L., and E. Köckeis 1963 Propositions for a sociological theory of aging and the family. International Social Science Journal 15(3):410-426.

Ruggles, S. 1987 Prolonged Connections: The Rise of the Extended Family in Nineteenth Century England and America. Madison: University of Wisconsin Press.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Schoen, R., and V.E. Nelson 1974 Marriage, divorce, and mortality: A life table analysis. Demography 11(2):267-290.

Schwartz, K. 1988 Household trends in Europe after World War II. In N. Keilman et al., eds., Modeling Household Formation and Dissolution. Oxford: Clarendon Press.

Schwartz, S., S. Danziger, and E. Smolensky 1984 The choice of living arrangements by the elderly. In H.J. Aaron and G. Burtless, eds., Retirement and Economic Behavior. Washington, D.C.: The Brookings Institution.

Shanas, E. 1982 National Survey of the Aged. DHHS Pub. No. (OHDS)83-20425. Washington, D.C.: U.S. Department of Health and Human Services.

Shanas, E., P. Townsend, D. Wedderburn, H. Fris, P. Milhoj, and J. Stehouwer 1968 Old People in Three Industrial Societies. New York: Atherton Press.

Smith, J. 1987 The computer simulation of kin sets and kin counts. Pp. 249-266 in J. Bongaarts, T. Burch, and K. Wachter, eds., Family Demography: Methods and Their Application. Oxford: Clarendon Press.

Soldo, B.J., D.A. Wolf, and E.M. Agree 1990 Family, households, and care arrangements of the frail elderly: A structural analysis. Journal of Gerontology: Social Sciences 45:S238-S249.

Speare, A., Jr., and J.W. Meyer 1988 Types of elderly residential mobility and their determinants. Journal of Gerontology 43(3):S74-S81.

Speare, A., and M. Rendall 1989 The Effects of Poverty and Disability on the Family Structures of Aging Individuals. Paper presented at the annual meeting of the American Sociological Association, San Francisco.

Speare, A., Jr., R. Avery, and L. Lawton 1991 Disability, residential mobility, and changes in living arrangements. Journals of Gerontology: Social Sciences 46(3):S133-S142.

Spitze, G., J.R. Logan, and J. Robinson 1992 Family structure and changes in living arrangements among elderly nonmarried persons. Journal of Gerontology: Social Sciences 47(6):S289-S296.

Stack, C. 1974 All Our Kin. New York: Harper.

Stinner, W.F., Y. Byun, and L. Paita 1990 Disability and living arrangements among elderly American men. Research on Aging 12:(3):339-363.

Sweet, J., L. Bumpass, and V. Call 1988 The Design and Content of the National Survey of Families and Households. Working Paper NSFH-1. Madison: Center for Demography and Ecology, University of Wisconsin.


Tissue, T., and J. McCoy 1981 Income and living arrangements among poor aged singles. Social Security Bulletin 44(4):3-13.

Troll, L. 1971 The family of later life: A decade review. Journal of Marriage and the Family 33:263-290.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Tsuya, N., and L.G. Martin 1992 Living arrangements of elderly Japanese and attitudes toward inheritance. Journal of Gerontology: Social Sciences 47(2):S45-S54.

Tu, E.J.C., V.A. Freedman, and D.A. Wolf 1993 Kinship and family support in Taiwan: A microsimulation approach. Research on Aging 15(4):465-486.

United Nations 1993 The Sex and Age Distribution of the World's Population: The 1992 Revision. New York: United Nations.


Vaupel, J. 1988 Inherited frailty and longevity. Demography 25(2):277-287.


Wachter, K., E. Hammel, and P. Laslett 1978 Statistical Studies of Historical Social Structure. New York: Academic Press

Wall, R. 1989 Living arrangements of the elderly in contemporary Europe. In E. Grebnik et al., eds., Later Phases of the Family life-cycle. Oxford: Clarendon Press.

Warnes, A.M. 1986 The residential mobility histories of parents and children, and relationships to present proximity and social integration. Environment and Planning 18:1581-1594.

Watkins, S.C., J. Menken, and J. Bongaarts 1987 Demographic foundations of family change. American Sociological Review 52:346-358.

Wijewickrema, S. 1987 Family Status Life Tables for Belgium: Construction and Ensuing Simulations. Interuniversity Programme in Demography Working Paper No 1987-5. Vrije Universiteit Brussel, Belgium.

Wiseman, R.F., and C.C. Roseman 1979 A typology of elderly migration based on the decision-making process. Economic Geography 55:324-337.

Wister, A.V., and T.K. Burch 1983 Fertility and household status of older women in Canada, 1971. Canadian Studies in Population 10:1-13.

Wolf, D.A. 1984 Kin availability and the living arrangements of older women. Social Science Research (13):72-89.

1988 Kinship and family support in aging societies. In Economic and Social Implications of Population Aging. Population Division, Department of International Economic and Social Affairs. New York: United Nations.

1989a Changes in the Households of Older Women: A Cross-National Study. Paper presented at the annual meeting of the Population Association of America, Baltimore, Md.

1989b Kinship patterns and household composition of the elderly: Hungarian women, 1984. European Journal of Population 4:315-337.

1990a Household patterns of older women: Some international comparisons. Research on Aging 12(4):463-486.

1990b Coresidence with an Aged Parent: Lifetime Patterns and Sensitivity to Demographic Change. Paper presented at the United Nations Conference on Population Aging in the Context of the Family, Kitakyushu, Japan.

Wolf, D.A., and B.J. Soldo 1988 Household composition choices of older unmarried women. Demography 25(3):387-403.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×

Wolf, D.A., T.K. Burch, and B. Matthews 1990 Kin availability and the living arrangements of older unmarried women: Canada, 1985. Canadian Studies in Population 17:49-70.

Wolf, D.A., V.A. Freedman, B.J. Soldo, and E.H. Stephen 1991a Making Room for Mom: Coresidence of Married Couples and Elderly Mothers. Paper presented at the annual meeting of the Population Association of America, Washington D.C.

Wolf, D.A., B.J. Soldo, and V.A. Freedman 1991b The Demography of Family Care for the Elderly. Paper presented at the US-USSR Population Symposium, The Urban Institute, Washington D.C.

Woodrow, K.A. 1990 Emigration from the United States Using Multiplicity Surveys. Paper presented at the annual meeting of the Population Association of America, Toronto.

Woodrow, K., and A. Peregoy 1991 Parents, Siblings, and Children: How Many Do Immigrants and Native-Born Persons Have? Paper presented at the annual meeting of the Population Association of America, Washington, D.C.

Worobey, J., and R. Angel 1990 Functional capacity and living arrangements of unmarried elderly persons. Journal of Gerontology 45:S95-101.

Yi, Z. 1986 Changes in family structure in China: A simulation study. Population and Development Review 12:675-703.

Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 146
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 147
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 148
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 149
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 150
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 151
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 152
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 153
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 154
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 155
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 156
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 157
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 158
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 159
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 160
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 161
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 162
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 163
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 164
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 165
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 166
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 167
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 168
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 169
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 170
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 171
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 172
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 173
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 174
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 175
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 176
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 177
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 178
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 179
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 180
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 181
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 182
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 183
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 184
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 185
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 186
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 187
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 188
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 189
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 190
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 191
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 192
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 193
Suggested Citation:"5 The Elderly and Their Kin: Patterns of Availability and Access." National Research Council. 1994. Demography of Aging. Washington, DC: The National Academies Press. doi: 10.17226/4553.
×
Page 194
Next: 6 Care of the Elderly: Division of Labor Among the Family, Market, and State »
Demography of Aging Get This Book
×
Buy Paperback | $107.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

As the United States and the rest of the world face the unprecedented challenge of aging populations, this volume draws together for the first time state-of-the-art work from the emerging field of the demography of aging. The nine chapters, written by experts from a variety of disciplines, highlight data sources and research approaches, results, and proposed strategies on a topic with major policy implications for labor forces, economic well-being, health care, and the need for social and family supports.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!