American Grandparents Providing Extensive Child Care to Their Grandchildren: Prevalence and Profile

Abstract

Purpose: This study sought to determine the prevalence and profile of grandparents providing extensive care for a grandchild (grandparents who provide 30+ hours per week or 90+ nights per year of child care, yet are not the primary caregiver of the grandchild). Design and methods: Secondary analysis of the 3,260 grandparent respondents in the 1992–94 National Survey of Families and Households (NSFH). Extensively caregiving grandparents were compared with custodial grandparents (those with primary responsibility for raising a grandchild for 6+ months), noncaregivers, occasional caregivers (<10 hours per week), and intermediate caregivers using chi-square tests, one-way analysis of variance tests, and logistic regression analyses. Results: Close to 7% of all grandparents provided extensive caregiving, as did 14.9% of those who had provided any grandchild care in the last month. Extensive caregivers most closely resembled custodial caregivers and had least in common with those grandparents who never provided child care. Implications: Areas for future research, policy, and practice are highlighted, including the potential impact of welfare reform legislation on extensively caregiving grandparents.

Decision Editor: Laurence G. Branch, PhD

The last decade has seen increasing awareness among health and social service providers, researchers, policy makers, and the general public of the substantial number of grandparents who provide child care on a regular basis to their grandchildren. As Hirshorn 1998 has pointed out, such activities as routinely getting young children ready for school in the mornings, providing full-time child care for preschoolers, or serving as a custodial parent to children whose parents are unwilling or unable to provide care "move the grandparent out of the more traditional older-generation-in-the-family roles … and into roles typically assumed by a child's parents" (p. 200). As she further noted:

In so doing, grandparent caregiving stretches, reorganizes, and redefines the relationships between family members; redraws the boundaries of family and, often, of household units; and redirects the transfer of resources within the family (Hirshorn 1998, p. 200).

Since 1970, two notable and diverging trends in grandparent care for grandchildren have been observed. On the one hand, this period saw a dramatic 76% increase in the number of children under 18 living in households maintained by a grandparent. The number of children in such households grew from 2.2 million (3.2%) in 1970 to 3.9 million (5.5%) in 1997, with particularly rapid growth since 1980 (Lugaila 1998). In an estimated one third of these homes neither parent was present and, since 1990, by far the fastest growth has been in these "skipped generation" households (Casper and Bryson 1998) in which grandparents are typically the sole or primary caregiver. A number of factors help explain the increase in grandparents raising grandchildren, including parental substance abuse, teen pregnancy, AIDS, unemployment, and incarceration, as well as legal and policy changes favoring kinship care over other placement arrangements (Burnette 1997; Minkler 1999).

Although custodial care among grandparents has increased substantially, the proportion of young children for whom noncustodial grandparents and other relatives are the main source of child care has declined dramatically, corresponding to the rise in more formal center-based child care arrangements. A fivefold increase in center-based child care (from 6% to 30% of all child care arrangements) occurred between 1965 and 1993, whereas family day care use remained constant at about 17% and nonparental relative care declined from 33% to 25% (Hofferth 1996). When the discussion is confined to preschool children whose mothers work full time, 1993 census data indicate that more than one million preschoolers are cared for by their grandparents (Casper 1996). This figure represents 16.6% of preschoolers whose mothers work full time (Casper 1996) and 5.4% of all preschoolers (U.S. Census Bureau 2000).

Increased female labor force participation (affecting both young mothers and midlife grandmothers), the rise in single-headed households (leaving mothers with fewer relatives to call upon for assistance), and a growing parental belief in the value of nonfamily child care in socializing young children for school appear largely to have spurred these changes (Hofferth 1996). Although noncustodial grandparents no longer constitute the primary source of nonparental child care for children, they remain frequently called-upon sources of child care on a regular basis. Research by Baydar and Brooks-Gunn 1998 using 1987–88 data from the National Survey of Families and Households, for example, suggested that 43% of grandparents provided child care for their grandchildren at least once monthly. Similarly, Bass and Caro 1996 research, using national field poll data, found that 32% of individuals with grandchildren or great grandchildren had provided care for at least one hour during the previous week. Of those providing care, 13.6 hours constituted the average time devoted to this activity. Data like these are helpful in highlighting the high proportion of grandparents who provide at least some care to their grandchildren on a weekly or monthly basis. Yet, they tell us little about those grandparents who, while not defining themselves as the primary caregiver of their grandchildren, nevertheless provide extensive child care on a regular basis.

Surprisingly few studies have focused primarily on grandparents who play a substantial role in caring for their grandchildren without crossing the line to become primary caregivers (Caputo 1999). In one such study, Presser 1989 examined the economic complexities involved in child care provision by grandparents who were providing substantial caregiving to assist a parent and concluded that the juggling of work and family roles in these situations may be substantial. Folk and Beller 1993 research also examined the provision of child care by relatives within the context of maternal employment, noting a continued preference for such care, particularly for very young (<age 3) children. Finally, several researchers have examined the often extensive child care and related roles of grandmothers in assisting their teenage daughters who have become mothers (Burton and Bengtson 1985; Flaherty 1988; Cherlin and Furstenberg 1986).

The present study was designed to provide a profile of American grandparents who provide extensive secondary child care to their grandchildren, and to compare and contrast this population group with those who are not providing such care along a variety of social and demographic dimensions. To increase our understanding of the continuum of child care provision for grandchildren, a typology was created that enabled us to conduct five-way comparisons. We compared (1) noncustodial extensive child care providers, (2) grandparents who self-defined as having been the primary caregiver for their grandchildren (custodial grandparents), (3) grandparents providing no child care (noncaregiver), (4) grandparents providing <10 hours a week of child care for their grandchildren (occasional caregivers), and (5) grandparents providing 10–29 hours of care (intermediate caregivers).

Methods

Sample

The grandparents analyzed in this study were a subsample drawn from the larger two-wave nationally representative panel study, the National Survey of Families and Households (NSFH). The surveys were conducted by the Center for Demography and Ecology at the University of Wisconsin–Madison. In 1992, 1993, or 1994, 10,008 respondents were interviewed. They comprise 77% of the original 13,008 respondents interviewed during the first wave of data collection in 1987 and 1988. Questions on grandparent caregiving only occurred in the second wave of data collection; therefore, all analyses reported here are restricted to the 1992–1994 data. In the second wave, 3,477 reported having one or more grandchildren. Once we deleted from the sample those who had provided custodial care during the 1980s or earlier but who were no longer doing so during the 1990s and those who had missing information on care provision, 3,260 subjects remained. Of these remaining respondents, 223 provided extensive caregiving during the second wave of interviews, 173 were custodial grandparents during the 1990s, 1,319 provided no care, and 757 provided occasional care. An additional 788 grandparents provided intermediate care.

In order to adjust for nonresponse and for oversampling of ethnic minorities, nontraditional families, and recently married people, a weighting variable was constructed by the NSFH. This weighting represents a sample that is demographically representative of the coterminous United States. All analyses reported in this study are weighted. (Further details on the study design and questions can be found in Sweet, Bumpass, and Call 1988).

Measures

As noted above, we developed five categories: one self-defined primary caregiver category (custodial grandparents) and four groups who did not define themselves as the primary caregiver (extensive, intermediate, occasional, and noncaregivers).

"Extensive" caregivers were providing a significant proportion of their grandchild's care although they did not report themselves as the primary caregiver. We operationalized this variable in two ways: Grandparents were included if they (1) provided at least 30 hours of child care in an average week and/or (2) cared for grandchildren for at least 90 nights per year.

Thirty hours a week was decided upon as the cutpoint in the first instance because it reflects a substantial portion of the grandparents' week—approximately 6 hours a day for 5 days. To determine grandparents in the NSFH who provided 30 or more hours a week babysitting, a proxy measure was constructed. All respondents were asked how many hours in an average week they spent helping their child and/or children age 19 or over, including all the following kinds of help: shopping and/or transportation; work around the house; advice or support; and child care. If the grandparents did not report having primary responsibility for raising a grandchild, yet they did state that they helped with child care and that they provided help for 30 or more hours a week, they were included in our subsample of 30+ hour-a-week child care providers. In the cases where the respondent provided child care and one or more of the other types of assistance to their adult children, it was assumed that child care represented the largest proportion of this time. The nonspecificity of the duration question leaves some room for false positives (e.g., inclusion of some people who provided child care for 10 hours and housework for 20 hours). However, the authors believe that this error will be relatively rare.

Our second type of extensively caregiving grandparents reported that a grandchild and/or grandchildren stayed overnight at their house, without their parents present, for at least 90 nights per year. Ninety nights was chosen because this exceeds the average length of a child's summer holidays. Although taking care of a grandchild for 2 months in the summer reflects a significant contribution of energy and a life-style change, we felt that the 3-month cutpoint would limit the analysis to only the most heavily involved caregivers. If grandparents met the criteria for either form of caregiving, they were defined as "extensive caregivers" (n = 223). Note that we could not translate the nights per year into an average hours per week value and sum the combined value of the two types because in some cases nights may reflect only 8–12 hours of caregiving (e.g., while the grandchild's parent is working a night shift), whereas in other cases "nights" reflect 24 hour-a-day care (e.g., while the child visits for a month in the summer).

Intermediate caregivers provided child care for a grandchild between 10 and 29 hours a week and/or had their grandchild stay overnight without his or her parent for 7 to 89 nights.

Occasional caregivers provided between 1 and 9 hours of child care in an average week and/or had a grandchild stay overnight without his or her parents 1–6 days in the past year.

Noncaregiving grandparents provided no babysitting during an average week and, in the past year, had not had a grandchild stay overnight without his or her parents.

Custodial caregivers were grandparents who replied affirmatively to the question, "For various reasons, grandparents sometimes take on the primary responsibility for raising a grandchild. Have you ever had the primary responsibility for any of your grandchildren for 6 months or more?" and who reported beginning or ending caregiving during the 1990s. Caregivers who reported they had been primary caregivers but were no longer caregiving during the 1990s were removed from the sample so as not to confound the analysis. (For a more detailed discussion and analysis of this group, see Fuller-Thomson, Minkler, and Driver 1997.)

A total of 10 independent variables were included in the analyses. Demographic variables included age, gender, race (African American vs other), marital status (married vs other), education (high school completion vs not), geographic region (South vs other), and poverty status (above or below the poverty line). Familial variables included total number of grandchildren and location of nearest child (coresident, within 20 miles, more than 20 miles away). Finally, self-reported closeness to grandparents was measured using the question, "On a scale from 0 to 10, where 0 is ‘not at all close’ and 10 is ‘extremely close,’ how would you describe your relationship with your grandchild(ren)?"

Statistical Analyses

Overall prevalence of caregiving type was determined by generating a frequency of the five categories of caregivers for all grandparents. We also were interested in the percentage of extensive caregivers among three distinct subgroups of grandparents: (1) those who had at least one grandchild under age 5 (n = 1,950), (2) those who had provided some child care to their grandchildren during the previous month (n = 1,305), and (3) those who had provided at least some child care in the previous month for their grandchildren while their adult child was working (n = 813). These percentages were obtained through frequencies of caregiving type for each identified subsample.

To determine differences among the caregiving type for demographic variables, chi-square tests were used for categorical variables, and one-way analysis of variance (ANOVA) tests were used for interval and ratio level variables. For those variables that reached the level of significance, it was necessary to conduct a further test to determine which of the grandparent types differed significantly from the extensive caregiving category. For the ratio and interval level variables, Tukey's honestly significant difference post-hoc tests were conducted. For the categorical variables, subanalyses were conducted doing pair by pair chi-square comparisons. To keep the overall significance level at .05, the pairwise comparisons' significance level was reduced to .013 (from .05 for overall significance/4 sets of comparisons).

In order to assess the independent contribution of each predictor variable while controlling for the other variables in the equation, it was necessary to conduct four logistic regression analyses with extensive caregiving coded 1 and the other four caregiving categories used in succession as the reference category. In the first case, the sample was restricted to extensive caregivers (n = 204) and noncaregivers (n = 1,174) who had complete data for all the independent variables in the analysis. In the second case, the sample included only extensive caregivers (n = 204) and occasional caregivers (n = 713) with complete data. The third analysis was limited to extensive caregivers (n = 204) and intermediate caregivers (n = 736) with complete data. The final analysis included only extensively caregiving grandparents (n = 204) and custodial caregivers (n = 151) with complete data. This method provided odds ratios for extensive caregivers in comparison to noncaregivers, occasional caregivers, intermediate caregivers, and custodial caregivers. In each analysis, all 10 independent variables were entered in a block.

In addition to the odds ratios, two other statistics are reported: the R statistic and the Nagelkerke R-squared. The R-statistic measures the individual partial correlation between the dependent variable and each independent variable in logistic regression analyses (Norusis 1997). The Nagelkerke R-squared value is analogous to the R-square in a linear regression model (Norusis 1997). The larger the R-squared value, the greater the explanatory power of the combination of the independent variables.

Results

Using the four categories of caregivers defined above, 6.8% of all grandparents were serving as extensive caregivers at the time of the interview. In contrast, 40.4% of the grandparents in the total sample were noncaregivers, 23.3% were occasional caregivers, 24.2% were intermediate caregivers, and 5.3% had provided custodial care at some time during the 1990s. Grandparents who were providing both extensive and custodial care (n = 67) were classified as custodial caregivers—a fact which may have resulted in an underestimation of the proportion of extensive caregivers.

Age of grandchildren clearly played an important role in influencing the likelihood of extensive care provision, with 9% of all grandparents with a grandchild under age 5 being extensive caregivers.

Of grandparents interviewed during the 1990s, 1,305, or 40.4%, had provided some child care to their grandchildren during the previous month. Of these, 14.9% were extensive caregivers. When the sample was further restricted to the 813 respondents who had provided at least some child care in the previous month for their grandchildren while their adult child was working, the percentage of extensive caregivers rose to 20.2%.

As indicated in Table 1 , significant differences within the five categories of grandparents were evident for all the characteristics investigated. Post-hoc tests provided further information on which pairs within each analysis were significantly different.

In comparison to noncaregivers, bivariate analyses indicated that extensive caregivers were significantly younger (56.4 vs 66.8 years), more likely to be married (73% vs 59%), African American (17% vs 9%), female (65% vs 52%), to have coresident children (51% vs 24%), and to have reported greater closeness to their grandchildren (9.6 vs 7.8 on a 10-point scale); see Table 1 . The groups were statistically similar with respect to educational level, geographic region, income, proportion below the poverty line, and total number of grandchildren. In order to determine the individual contribution of each independent variable while other variables are held constant, a logistic regression analysis of extensive caregivers versus noncaregivers was performed (see Table 2 ). This analysis showed that, in comparison to noncaregivers, the odds of being an extensively caregiving grandparent were 87% higher for married grandparents, 53% higher for grandmothers, and 77% lower for those who live more than 20 miles from all their children. The odds of being an extensive caregiver were higher for younger respondents (odds ratio [OR] = 0.92 for age), for those with more grandchildren (OR = 1.08), and for the grandparents who reported greater closeness to their grandchildren (OR = 1.73). The R statistic indicates that age and closeness to grandchildren are the variables in the model that provide the most explanatory power.

In contrast with occasional caregivers, extensively caregiving grandparents were significantly younger (56.4 years vs 59.5 years) and more likely to be African American (17% vs 8%), female (65% vs 54%), to live in the South (43% vs 30%), and to have coresident children (51% vs 34%). Extensive caregivers were poorer than occasional caregivers; they had a lower mean income ($34,286 vs $47,389) and were more than twice as likely to live below the poverty line (18% vs 8%). Extensive caregivers were less likely to have graduated from high school (67% vs 81%) and reported greater closeness with their grandchildren (mean scores of 9.6 vs 9 on a 10-point scale); see Table 1 . The logistic regression analysis comparing extensive and occasional caregivers, reported in Table 2 , indicates that respondents who lived below the poverty line had 89% higher odds and high school graduates had 37% lower odds of being extensive caregivers. In comparison to those who had coresident children, the odds of extensive caregiving were low for those whose nearest child lived nearby and even lower for those whose nearest child was more than 20 miles away (OR = 0.63 and 0.50, respectively). Those who reported greater closeness with their grandchildren also had higher odds of providing extensive care (OR = 1.32).

In comparison to intermediate caregivers, extensive caregivers were significantly younger (56.4 years vs 58.9 years), more likely to be African American (17% vs 11%) and to have coresident children (51% vs 32%), and less likely to be a high school graduate (67% vs 77%); see Table 1 . Extensive caregivers earned, on average, approximately $7500 less per year and were 6% more likely to be living below the poverty line than intermediate caregivers. The logistic regression analysis (see Table 2 ) indicates that extensive caregivers had much lower odds than intermediate caregivers of all their children living outside the grandparent's home. The odds were 47% lower that the children of extensive grandparents lived between 1–20 miles from the home as opposed to being coresident. The odds of children living more than 20 miles from the grandparent's home were 57% lower for extensive caregivers.

Custodial and extensive caregivers were similar with respect to most characteristics (see Table 1 ). However, extensive caregivers were more likely to be married (73% vs 54%) and less likely to be African American (17% vs 27%) or female (65% vs 77%). Extensive caregiving grandparents also had fewer grandchildren than did their custodial caregiving peers (5.2 vs 7.3 grandchildren). In the logistic regression analyses, married respondents had twice the odds of being extensive caregivers (see Table 2 ). With each additional grandchild, the probability of providing extensive as opposed to custodial care decreased by 5%.

Discussion

The fact that 6.8% of the grandparents in this national sample were currently caring for a grandchild on an extensive basis suggests that one out of every 15 American grandparents may be doing the equivalent of almost a full-time job in terms of the hours devoted to this activity. It should be noted, moreover, that the denominator here includes grandparents who may be very elderly or ill, and those who are geographically at a considerable distance from their grandchildren. When we limited our analysis to those grandparents who had provided some child care in the last month (40.4% of the grandparent sample), almost 15% were providing extensive child care, as were one in five (20.2%) of those who reported providing child care while their adult child worked. The latter finding is in keeping with earlier national studies by Hofferth, Brayfield, Deich, and Holcomb 1991 and others (Brayfield, Deich, and Hofferth 1993), who found that 19% of preschoolers whose mothers worked in 1990 were receiving child care primarily from grandparents and other relatives, rising to 28% for low-income preschool children.

Our finding that 9% of all Americans with grandchildren under age 5 were providing extensive caregiving to a grandchild is considerable higher than the U.S. Census Bureau's estimate that 5.4% of all preschoolers in 1993 were being cared for by a grandparent while the mother was in full-time employment (Casper 1996). The fact that the census data are restricted solely to caregiving while the grandchild's mother is working full-time is the likely cause of the discrepancy between the two figures. The 30 or more hours grandparents provide care per week may, in fact, be a combination of hours spent babysitting for different families of grandchildren, each of which had a mother who worked. Similarly, the 30 hours may include time spent babysitting while the grandchild's parents are not at work. The inclusion of 90 or more nights in our extensive caregiving category widened the scope of caregiving to include extended visits and part-time night-shift work.

It should be recalled that, with the exception of custodial caregiving, the measures utilized in this study assessed only caregiving during the past year. If we were to consider the percentage of grandparents providing extensive care at some point in their lives, the estimates cited would undoubtedly be considerably higher.

Finally, as noted earlier, the rates of extensive grandchild care provision revealed in our study omit two small but important groups: grandparents who were providing such care to the offspring of teenage children, and those who at some point in the 1990s were primary caregivers to one or more grandchildren. Despite recent declines, the continuing high rates of teenage childbirth in the United States resulted in almost half a million live births in 1999 (Curtin and Martin 2000). Earlier research, suggesting that grandparents play a particularly important caregiving role for the children of teenage parents (Burton 1990; Cherlin and Furstenberg 1986), underscores the importance of further research that includes this category. Similarly, the omission from our sample of extensive caregivers of the 67 respondents who were providing both custodial and extensive care decreased the percentage of extensive caregivers from 9% to the just under 7% reported.

When we turned from prevalence to a sociodemographic profile of grandparents providing extensive care to their grandchildren, an interesting picture emerged. As suggested above, the majority of grandparents playing this role were married White women in their mid 50s who had five grandchildren and were as likely as not to have one of their own children residing with them. Most were high school graduates, and the mean 1993 income was over $34,000.

Our finding that extensive caregivers were significantly younger than noncaregivers is understandable, because younger grandparents are likely to be in better health and hence physically and mentally better able to provide extensive child care than their noncaregiving counterparts. The greater prevalence of extensive caregiving among grandmothers than grandfathers similarly is in keeping with much earlier sociological research (Abel 1991; Hochschild 1989) suggesting that women do the bulk of caregiving over the life course. However, our finding that over a third (35%) of extensive caregivers were men (compared to just 23% of custodial caregivers) suggests that a sizable proportion of grandfathers do in fact provide considerable care for grandchildren on a regular basis, and underscores the need for further study of this population.

Although, as noted earlier, the majority of extensive grandchild care providers in our study were White women, the fact that such grandparents were considerably more likely than noncaregivers, as well as occasional and intermediate care providers, to be African American also is in keeping with the literature (Hunter 1997). Such extensive care provision thus has been shown to be more normative among African Americans, for whom the role of "caregiver across the generations" has been a critical one historically. African American caregiving thus is rooted in West African culture and tradition (Sudarkasa 1981; Wilson 1989) and played a critical role in the United States during and after slavery and during the great migrations north after World War II (Wilson 1989).

Although most grandparents reported that they were quite close to their grandchildren, the extra contact experienced by extensive caregivers was associated with significantly higher reports of closeness than was experienced by noncaregivers and those providing only occasional child care. This association may reflect the fact that greater contact allows for a closer bonding with grandchildren—a phenomenon also observed among custodial and intermediate care providers. An alternate explanation is that closer attachment to grandchildren motivates a grandparent to undertake the extensive caregiving role. Further qualitative research into the nature and nuances of such closeness, from the perspective of both grandparents and grandchildren, is needed to help enrich our understanding of this phenomenon.

Extensive caregivers more closely resembled intermediate caregivers than either noncaregivers or occasional caregivers. However, extensive caregiving grandparents were from more vulnerable populations than their peers providing intermediate care: they were more likely to be poor, for example, and less likely to have completed high school. The presence of coresident children (which is also likely to increase the odds of coresident grandchildren) is a strong predictor of extensive as opposed to intermediate care provision. Such higher rates of coresidence may reflect fewer living arrangement options as a result of low income and related factors, but further research would be needed to substantiate this hypothesis.

A major finding of this study was that extensive caregivers appeared to resemble custodial caregivers to a much greater degree than did those who provided only occasional child care. For example, although just 8% of occasional child care providers lived below the poverty line, the figure was far higher (18%) for extensive caregivers and highest of all (23%) for custodial parents. Since grandparents who are poor are more likely to have grandchildren who also live in low-income households, it may well be that the former's heavy involvement in child care provision reflects, in part, the prohibitive costs of much organized child care (Casper 1996).

There were, of course, areas of significant difference between extensive and custodial caregivers. The fact that the latter were more likely to be unmarried, to be female, and to have more grandchildren, for example, is in keeping with earlier studies (Chalfie 1994; Fuller-Thomson et al. 1997; Harden, Clark, and Maguire 1997), which demonstrated major differences between custodial and other grandparents along these dimensions. Similarly, the fact that custodial grandparents were significantly more likely to be African American than were extensive caregivers in the bivariate analyses is in keeping with the far higher rates of custodial grandparenting among African American grandparents, one in four of whom occupies this role for at least 6 months (Szinovacz 1998). The earlier mentioned similarity between extensive and custodial grandparent caregivers along multiple dimensions, however, is of major consequence and may indeed reflect a blurring of the line between these two roles, as grandparents move from one to the other as need dictates. The impressive body of literature on custodial grandparents generated over the past decade (Burnette 1999; Burton 1992; Chalfie 1994; Emick and Hayslip 1996; Fuller-Thomson et al. 1997; Minkler and Roe 1993; Pruchno 1999; Sands and Goldberg-Glen 1998) needs to be supplemented by studies that explore in more detail the lives of both extensive and custodial grandparents, with particular attention to the trajectory of the caregiving experience over time.

In addition, the evolving nature of the continuum of child care provision by grandparents needs to be carefully explored within the context of the changing demographic, cultural, and sociopolitical landscape of the United States. The rapid growth of a variety of both Latino and Asian Pacific Islander populations, for example, underscores the need for far greater research attention to the grandparental child care provision roles of these largely neglected groups (see Burnette 1999, and Burnette in press for important exceptions). Changes over time in the extent and nature of child care provision within such groups as a result of acculturation and other factors also should be explored through both qualitative and quantitative studies.

Finally, sociopolitical factors, including, importantly, the implementation of welfare reform legislation, merit careful attention. A key informant study in California (Minkler, Berrick, and Needell 1999), for example, revealed considerable concern among policy analysts and providers that the Personal Responsibility and Work Opportunity Reconciliation Act's (1996) work requirements and the requirement that unmarried teenage mothers live with a parent or guardian and either attend school or hold a job as a condition of aid, could well result in an increasing number of middle-aged and older women being pressured to become extensive child care providers for their grandchildren. The fact that over half of extensive caregivers in this study had coresident children (as opposed to 29% of noncaregivers) suggests that such grandmothers may already be providing considerable family care, and additional responsibility for nonresident grandchildren may pose a substantial burden. As Mullen 2000(p. 129) has noted, most states "have largely neglected" the impact of welfare reform on grandparents, and efforts are needed to both monitor this issue and help ensure that grandparents are supported, rather than penalized, if they indeed become extensive child care providers to grandchildren as a consequence of the changing welfare environment.

Finally, as Casper and Bryson 1998(p. 14) have argued, "grandparents and their grandchildren would benefit greatly if policies and programs intended to help traditional parent–child families in times of need could be extended to grandchildren" and the grandparents who often play a substantial role in their care.

The authors gratefully acknowledge the Robert Wood Johnson Foundation and the Archstone Foundation for their support of this research. We thank the reviewers for their helpful comments on an earlier version of the manuscript.

References