| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Initiatives on Aging, Boston College, Newton, Massachusetts
Rachel A. Pruchno, Initiatives on Aging, Boston College, 885 Centre Street, Newton, MA 02459 E-mail: pruchno{at}bc.edu.
Decision Editor: Margie E. Lachman, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
In 2000, 6.3% of children in the United States, or 4,533,016 children, were living with a grandparent (Bryson 2001
). These figures reflect a growing trend, as the percentage of grandchildren living with a grandparent increased from 3.2% in 1970 to 5.4% in 1997 (Lugaila 1998; Saluter 1996). Although the majority of households that include both a grandparent and a grandchild also include at least one of the grandchild's parents, the fastest growing type of household since 1990 is that which does not include either of the grandchild's parents (Bryson and Casper 1999; Casper and Bryson 1998). By 2000, there were 2,354,121 grandparent caregivers in the United States, representing 8.6% of all U.S. households.
When grandmothers share a household with grandchildren in the absence of the grandchild's parents, they play a pivotal role in the lives of the children (Pruchno 1999). Although the number of grandmothers raising grandchildren has grown, there remains a dearth of information about the ways in which these responsibilities affect the psychological well-being of these women. The analyses that follow examine the ways in which the tasks with which grandmothers are helping their grandchildren, the extent to which maladaptive behavior characterizes the grandchild, the grandmother's health, the centrality of the grandparent role, and the relationship between the grandmother and the grandchild's parents affect the caregiving satisfactions and burdens as well as the positive and negative affect experienced by grandmothers raising grandchildren.
Evidence suggesting that some grandmothers raising grandchildren experience threats to their psychological well-being comes from two types of data. The first includes small, homogeneous samples (e.g., Burton 1992; Giarrusso, Feng, Wang, and Silverstein 1996; Jendrek 1993a,Jendrek 1993b, Jendrek 1994; Kelley 1993; Minkler and Roe 1993; Minkler, Roe, and Price 1992; Shore and Hayslip 1994; Strawbridge, Wallhagen, Shema, and Kaplan 1997). The second is based on data from the National Survey of Families and Households (e.g., Fuller-Thomson, Minkler, and Driver 1997; Minkler, Fuller-Thomson, Miller, and Driver 1997; Szinovacz, DeViney, and Atkinson 1999). These studies highlight the complexities involved in the relationship between assuming care for a grandchild and psychological well-being. Minkler and colleagues 1997, for example, found that grandmothers caring for a grandchild were nearly twice as likely as grandmothers not providing such care to experience depressive symptoms. Similarly Szinovacz and colleagues 1999 found that when grandchildren moved into the grandparents' household, grandmothers experienced an increase in depressive symptoms.
The ways in which race affects the experiences of grandmothers raising grandchildren have received only limited attention. Yet differences in the role of grandparent within Black and White cultures suggest the importance of examining the ways in which race affects both the caregiving experience and its effects on psychological well-being. Historically, Black and White grandmothers have played different roles within families, with Black grandmothers playing a more central role in holding kin networks together (Burton and Dilworth-Anderson 1991; Hagestad and Burton 1986). Black households are more likely than White households to include both a grandparent and a grandchild (9.2% vs. 2.3%; Bryson and Casper 1999), and Black grandmothers are less likely than White grandmothers to embrace norms of noninterference (Cherlin and Furstenberg 1986; Kornhaber and Woodward 1981).
We developed the conceptual framework for predicting the psychological well-being of caregiving grandmothers by integrating the risk factors identified in the existing literature regarding grandparent caregivers, the broader caregiving literature, and empirical and theoretical knowledge about family stress. The model builds on Lawton's two-factor model of caregiving appraisal and psychological well-being (Lawton, Kleban, Dean, Rajagopal, and Parmelee 1992; Lawton, Moss, Kleban, Glicksman, and Rovine 1991), and empirical work by Pruchno and her colleagues (Pruchno, Patrick, and Burant 1996; Pruchno, Peters, and Burant 1995).
Positive and negative affect, the central outcomes in the model, are subjective states resulting from both long-term personality dispositions, general psychopathology, and situation-specific stressors. Research has clearly established that positive and negative affect are different from one another and that both are necessary for a comprehensive characterization of psychological well-being (Diener and Emmons 1984; Warr, Barter, and Brownbridge 1983; Watson and Tellegen 1985). Positive affect has been consistently associated with the quality of external events, such as activities or social behavior, whereas negative affect has typically been found to be predicted by health and other internal attributes (Bradburn 1969; Lawton 1983).
Caregiving is an activity of mixed valence for the caregiver. On the one hand, caregiving activities are positively affirming; on the other hand, caregiving is a demand that competes with other roles and responsibilities. As such, caregiving activities have the potential to simultaneously enhance caregiving satisfaction while increasing caregiving burden. Furthermore, greater satisfaction from the caregiver role should be associated with positive affect but be less effective in mitigating negative affect, whereas caregiving burden should increase negative affect, but be less effective in diminishing positive affect.
Consistent with findings from research focused on predictors of positive and negative affect, it is expected that caregiving satisfaction is predicted by the quality of external events related to the caregiving experience, whereas caregiving burden is predicted by internal attributes of both the caregiver and care receiver. Following this, greater centrality of the grandparent role (Lawton et al. 1991; Pruchno et al. 1995) and better relationships between the caregiver and other family members (Pruchno, Peters, Kleban, and Burant 1994) should lead to greater caregiving satisfaction, whereas poorer health on the part of the grandmother and more aberrant behaviors on the part of the grandchild are expected to increase caregiving burden. In addition, because research consistently has found that the lack of caregiving satisfaction promotes caregiving burden (Lawton et al. 1991; Pruchno et al. 1995, Pruchno et al. 1996), this relationship is also modeled.
The hypothesized relationships among model variables are depicted in Fig. 1:
|
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
The analyses that follow are based on a subsample of 867 grandmothers (488 White and 379 Black) older than the age of 50 years who were living with a grandchild in households that did not include either the grandchild's mother or father. According to the grandmothers' reports, there were often complex reasons, involving multiple family problems, that resulted in their ultimately assuming primary responsibility for their grandchild. Although the most common reason given for this shared living situation was drug addiction on the part of the child's mother (42.7%) or father (27.3%), grandmothers also reported that they were living with the grandchild because the child's mother was: (a) dead (9.0%), (b) in prison (5.3%), (c) mentally ill (5.8%), (d) addicted to alcohol (20.2%), (e) physically abusive (9.8%), (f) emotionally abusive (18.9%), (g) physically neglectful (32.2%), and (h) emotionally neglectful (33.6%). Grandmothers reported that the child's father was not raising his child because he was: (a) dead (4.3%), (b) in prison (8.0%), (c) mentally ill (1.7%), (d) addicted to alcohol (20.6%), (e) physically abusive (6.2%), (f) emotionally abusive (11.0%), (g) physically neglectful (15.2%), and (h) emotionally neglectful (14.3%).
Individuals learned about the study primarily through national media press releases (69.9%). Additional referral sources included (a) paid advertisements (2.4%), (b) contact with social agencies (5.9%), (c) schools (2.9%), (d) word of mouth (5.0%), (e) support groups (7.2%), and (f) referral from others who had participated in the study (6.1%).
Grandmothers participating in the study resided in 44 states, with most living in urban areas (90.2%). Respondents were distributed across the country as follows: (a) 11.3% in the Northeast, (b) 35.3% in the Midwest, (c) 35.6% in the South, and (d) 17.8% in the West. Blacks were more likely to be living in the South, whereas Whites were more likely to be living in the Midwest and West (
2(3, N = 867) = 15.82, p < .001).
Table 1 details demographic differences between the Black and White respondents. Grandmothers participating in the study ranged in age from 50 to 83 years (M = 57.9, SD = 6.1). Black respondents were more likely than White to be divorced or widowed, whereas White respondents were more likely to be married. The majority of the sample (66.1%) was Protestant, with 17.3% Catholic, 14.3% other, 0.8% Jewish, and 1.5% reporting no religion. Approximately half of the sample was currently working. Among those respondents who were currently working, the mean number of work hours per week was 35.02. Respondents reported that they worked in a variety of positions, with Black respondents more likely than White respondents to report working in service and laborer positions. A minority of the sample indicated that they had been housewives for most of their lives, with this being more characteristic of White than Black respondents. The women had an average of 13.4 years of education. Average annual income reported by respondents was lower for Black respondents than for White respondents.
|
Measures
We assessed grandchild behaviors by using a revision of the Achenbach Child Behavior Checklist (Achenbach 1991). We included in these analyses those items identified by Achenbach as discriminating between children referred for clinical treatment and those who were not referred, and those that assess hyperactivity and temper. Three items measured hyperactivity: (a) can't concentrate or pay attention for long, (b) can't sit still, is restless, or hyperactive, and (c) is impulsive or acts without thinking. Temper was measured with seven items (e.g., argues a lot; has temper tantrums). Grandmothers rated each behavior as either not true, sometimes true, or often true of the target grandchild during the past 6 months. Scores on both the hyperactivity and temper scales ranged from 0 to 6. Coefficient alpha was .77 for the hyperactivity scale and .84 for the temper scale. Table 2 includes information about the means of all model variables for the complete sample and for the Black and White subsamples.
|
We assessed grandmother's health by using self-report measures examining both objective and subjective aspects of physical well-being. To assess the number of physical illnesses, grandmothers completed the Health Conditions Checklist (National Center for Health Statistics 1982). They reported on whether during the past year they had experienced 29 physical health problems. Objective health was a sum of the number of health conditions. Scores ranged from 0 to 17. We measured subjective health by using four items from the Multiphasic Assessment Inventory (Lawton, Moss, Fulcomer, and Kleban 1982). The four items formed a scale ranging from 4 to 13; coefficient alpha was .76. Finally, we measured the functional ability of grandmothers by asking respondents how much difficulty (from none, 0, to unable, 3) they have doing 10 tasks, which included walking for a quarter of a mile without resting, reaching up as if to get something from a shelf, and lifting or carrying something as heavy as 10 pounds. Scores ranged from 0 to 25. Coefficient alpha for the scale was .86.
We used the following three items to measure the extent to which grandmothers viewed their grandparent role as central: (a) "I am one of those people whose life revolves around my grandchildren," (b) "My grandchildren are my main reason for living," and (c) "I measure out the rest of my life in terms of milestones in the grandchildren's lives." We used a 7-point Likert scale ranging from very strongly agree (6) to very strongly disagree (0). Scores on the scale ranged from 0 to 18. Coefficient alpha was .79.
We measured quality of the relationship between the grandmother and the grandchild's mother and father by using questions developed by Gronvold 1988. We assessed the quality of the current relationships by asking a set of questions that focused on the relationship between grandmother and grandchild's mother and a set that focused on the relationship between grandmother and grandchild's father. Grandmothers were asked about the extent to which they (a) feel close to the grandchild's mother and father, (b) get along with the grandchild's mother and father, and (c) can talk about things that concern them with the grandchild's mother and father. We measured each question on a 4-point Likert scale. Scores on each scale ranged from 4 to 16, with a higher score indicating a better relationship. Coefficient alpha for relationship with grandchild's mother was .92. Coefficient alpha for relationship with grandchild's father was .92. Only 3.3% of the grandmothers had no contact with either of the grandchild's parents. For purposes of the analyses that follow, they received relationship quality scores of "0."
Indicators of a grandmother's perception about providing care for her grandchild included both positive (caregiving satisfaction) and negative (caregiving burden) appraisals. We measured caregiving satisfaction with a six-item Caregiving Satisfaction scale developed by Lawton and colleagues 1982. Extensive information regarding scale development has been reported (Lawton, Kleban, Moss, Rovine, and Glicksman 1989), and the scale has been used in several caregiving studies (Lawton et al. 1991; Pruchno, Burant, and Peters 1994). We rated each of the six caregiving satisfaction questions on a 5-point Likert scale ranging from never (0) to nearly always (5). Scores ranged from 12 to 30. Coefficient alpha for the scale was .79.
We measured caregiver burden by using a nine-item scale developed by Lawton and colleagues 1989. Scores ranged from 9 to 39. Coefficient alpha for the scale was .86.
We assessed negative affect by using the Center for Epidemiologic Studies–Depression scale (CES-D; Radloff 1977) and the Negative Affect scale developed by Lawton and colleagues 1992. We scored the CES-D following the procedures described by Radloff 1977. Scores ranged from 0 to 49, with 21.5% of grandmothers scoring at or above a score of 16, suggesting that like other samples of caregivers (McCallum, Mackinnon, Simons, and Simons 1995; Pruchno et al. 1996), this group was not at significant risk for depressive symptomatology (Radloff 1977). Coefficient alpha for the scale was .89. The Negative Affect scale developed by Lawton and colleagues 1992 included self-ratings of whether, during the past week, respondents were: sad, annoyed, worried, irritated, or depressed. We used a 5-point Likert scale (never, 1, to very frequently, 5). The range was 5 to 25. Coefficient alpha was .82.
We measured positive affect with the Life Satisfaction Index A (LSIA; Neugarten, Havighurst, and Tobin 1961), and the Positive Affect scale developed by Lawton and colleagues 1992. Scores on the LSIA ranged from 2 to 20. Coefficient alpha was .78. The Positive Affect scale developed by Lawton and colleagues 1992 included five items, including whether the grandmothers felt: happy, interested, energetic, content, and warmhearted during the past week. We used a 5-point Likert scale (never, 1, to very frequently, 5). Responses ranged from 7 to 25. Coefficient alpha was .80.
Procedures
We used the Analysis of Moment Structures (AMOS) structural modeling program (Arbuckle 1995) in all analyses. We examined multiple indexes of fit to evaluate the fit of the data to the model. We used an overall chi-square index to assess the degree of fit between the estimated and observed covariance matrices. Lower values indicate better fitting models. Problems can arise with the chi-square index when the sample size is large (Bentler and Bonett 1980), so we used additional indexes to assess model fit. We included the following indexes: (a) comparisons of incremental changes in chi-squares among nested models, (b) Bollen and Stine 1992 incremental fit index (IFI), (c) root mean square error of approximation (RMSEA; Browne and Cudeck 1989), and Hoelter 1983 critical N (CN).
We tested the theoretical model first by using data generated from the White subsample. We made modifications to the model until an acceptable fit was found, and then we tested the goodness of fit of the modified model on a sample of Black grandmothers raising a grandchild. Next, we tested the similarity in the magnitude of the regression paths, covariances, and variances across White and Black grandmothers raising a grandchild. Finally, we tested the model on the complete sample.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
2(72, N = 488) = 208.15, p < .01; IFI = .99, RMSEA = .06, Hoelter CN = 222, suggest a good fit of the data to the model, with all hypothesized paths significant, the modification indexes suggested the addition of paths between grandchild behavior and caregiving satisfaction and between grandchild behavior and negative affect. We added these two paths to the model one at a time. Each additional path was significant and improved the fit of the model. Addition of these paths resulted in a final model with 2(70, N = 488) = 144.82, p < .01, IFI = .99, RMSEA = .05, Hoelter CN = 311. Maximum likelihood estimates documenting the relationships among latent variables are found in Table 4 . The critical ratios listed in the table represent significance tests, with paths having critical ratios greater than 1.96 interpreted as being significant.
|
|
2(70, N = 379) = 114.98, p = .001; IFI = .99, RMSEA = .04, Hoelter CN = 306). All hypothesized paths, with the exception of that from relationship between grandchild's parents and grandmother to satisfaction were significant; no additional unhypothesized paths were significant. Maximum likelihood estimates for the sample of Black grandmothers are found in Table 4 .
Stability of the Model Across Groups of White and Black Grandmothers
To more carefully assess the stability of the latent paths across the groups of White and Black grandmothers, we tested the final model simultaneously. We used multisample AMOS analysis to test a model in which the same parameter pattern was freely estimated within each group. This chi-square value was the starting point for each nested sequential analysis that follows. We tested the equality of the factor loadings for grandchild behaviors, grandmother health, grandmother–parent relationship, positive affect, and negative affect by simultaneously equating each loading. We did not find any differences in the magnitude of the factor loadings between the Black and White grandmothers. We contrasted the equality of the magnitude of the regression weights between latent variables simultaneously. This omnibus test indicated that there were no significant differences between the groups. Next, we contrasted the magnitude of the 11 covariances across the two groups simultaneously, with results indicating no significant differences between the groups. Finally, we contrasted the magnitudes of the 19 variances across the two groups simultaneously. Follow-up analyses indicated that there were significant differences in the magnitudes of two of these variances (error associated with the caregiving satisfaction construct, and error associated with positive affect).
Because of the parameter stability and equivalency across the two samples, it was possible to test the model on the complete sample of 867 grandmothers. The increase in sample size acted to further stabilize the parameters. Results from that analysis yielded 2(70, N = 867) = 202.871, p < .01, IFI = .99, RMSEA = .047, Hoelter CN = 429. The final model, depicted in Fig. 1 with parameter estimates for the complete sample, includes all hypothesized paths as well as the two additional significant paths (from grandchild behavior to caregiving satisfaction and from grandchild behavior to negative affect) that are presented in bold face.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
These analyses reveal the robust nature of the theoretical model. Developed to explain the psychological well-being of caregivers to older people, the model was previously tested and found to be efficacious for explaining the well-being of mothers of adult children with chronic disabilities (Pruchno et al. 1996). Explaining the affect of yet a different group of caregivers increases the validity of the model.
These data confirm the importance of conceptualizing affect as having both positive and negative components that are related to one another, yet also independent of one another. They also confirm that the principles guiding a two-factor model of psychological well-being operate at the level of subjective role appraisals. The data indicate that although caregiving satisfaction and burden have some predictors in common (help from grandmother and grandchild behavior) they also have unique predictors, with the quality of the relationship between grandmother and grandchild's parents and centrality of the grandparent role predicting caregiving satisfaction and grandmother's health predicting caregiving burden. In addition, the predicted relationship between caregiver satisfaction and caregiver burden is supported by these data.
In terms of affect, results indicate that although both positive and negative affect are predicted by grandmother's health and caregiver burden, caregiving satisfaction is a unique predictor of positive affect, and grandchild behavior is a unique predictor of negative affect. The salience of grandmother's health and caregiving burden as predictors of both positive and negative affect must not go unnoticed. Although positive and negative affect are two distinct constructs, the powerful effects of physical health and caregiver burden are demonstrated by these findings. Poor physical health and feelings of caregiving burden are so powerful that not only do they increase negative affect, they also decrease feelings of positive affect.
Although the overall model is stable across race, it is important to acknowledge the mean differences as well as the similarities between the Black and White grandmothers. The caregiving role has greater centrality for Black grandmothers, suggesting that this role is more important in the lives of Black grandmothers. Consistent with research reported by Lawton and colleagues 1992, Mui 1992, and Pruchno, Patrick, and Burant 1997, caregiving burden and negative affect are greater for the White grandmothers than for the Black grandmothers.
It is interesting to note that whereas the quality of relationship with grandchild's parents is significantly related to caregiving satisfaction for the White grandmothers, this relationship is not significant for the Black grandmothers. These findings suggest that the family dynamics underlying the relationship between the grandchild's parents and the grandmother may be important for understanding the caregiving experiences of White grandmothers. Future work focused on unraveling the reasons for the different family dynamics existing within Black and White families would provide important new information.
On the other hand, the similarities in the experiences for Black and White grandmothers must not be overlooked. Black and White grandmothers report similar levels of behavior problems on the part of their grandchild and indicate that they provide similar levels of help to these grandchildren. The health of Black and White grandmothers is similar to one another on all three dimensions examined and grandmothers report similar quality of relationships with the grandchild's mother. Finally, many indicators of positive and negative affect of Black and White grandmothers are similar to one another, including levels of caregiver satisfaction, life satisfaction, positive affect, and depression. Together these data point to the need to examine the stresses of caregiving across race and ethnicity, highlighting the similarities of the experience, while acknowledging the differences.
These findings, although intriguing, have important limitations that must be acknowledged. Although the sample is large, diverse, and heterogeneous, with respondents from across the United States, it is composed of individuals who volunteered to participate. As such, it is unclear how these people are similar to or different from the grandmothers who find themselves in the role of custodial grandparent but who did not come forward to participate in the study. It is also important to note that all data represent subjective reports from the grandmothers and as such, care must be taken in interpreting the findings.
Future research examining the extent to which results are generalizable to grandmothers living in households that include the grandchild's parents is needed. In addition, the extent to which findings can be replicated with longitudinal data would increase the ability to understand the intricate relationships among variables. Furthermore, the stability of the paths between grandchild behavior and caregiving satisfaction and between grandchild behavior and negative affect should be examined in future studies.
Nonetheless, these findings provide important new information which, used carefully, should be useful to social policy makers, legislators, and educators. These data suggest that in light of limited resources, special emphasis should be put on grandmothers who are in the poorest health and those whose grandchildren exhibit the most maladaptive behaviors. Grandmothers who live with and are responsible for the care of their grandchildren provide society with a vital service. Were it not for these women, the majority of their grandchildren would most likely become wards of the state. However, as the number of grandmothers raising grandchildren grows, it is crucial that attention be paid to the effects that this experience can have on the grandmothers' psychological well-being. Should they become unable to meet the needs of their grandchildren, society would be faced with a significant challenge.
|
Acknowledgments |
|---|
We are grateful to Sheri Kunovich for comments on an earlier draft of this article.
Received for publication July 19, 2000. Accepted for publication December 3, 2001.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. E. Hughes, L. J. Waite, T. A. LaPierre, and Y. Luo All in the Family: The Impact of Caring for Grandchildren on Grandparents' Health J. Gerontol. B. Psychol. Sci. Soc. Sci., March 1, 2007; 62(2): S108 - S119. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. C. Goodman and M. Silverstein Grandmothers Raising Grandchildren: Ethnic and Racial Differences in Well-Being Among Custodial and Coparenting Families Journal of Family Issues, November 1, 2006; 27(11): 1605 - 1626. [Abstract] [PDF]
|
|
|
|
|
|
C. C. Goodman Intergenerational Triads in Grandparent-Headed Families J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2003; 58(5): S281 - 289. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
