Children and Youth Services Review 33 (2011) 2138–2145

Contents lists available at ScienceDirect

Children and Youth Services Review

j ourna l homepage: www.e lsev ie r.com/ locate /ch i ldyouth

Behavior problems in children raised by grandmothers: The role of caregiver distress,family resources, and the home environment

Susan J. Kelley a,⁎, Deborah M. Whitley b, Peter E. Campos c

a Byrdine F. Lewis School of Nursing, Georgia State University, 140 Decatur Street, Atlanta, GA 30303, USAb School of Social Work, Andrew Young School of Policy Studies, Georgia State University, 14 Marietta St., Atlanta, GA 30303, USAc Project Healthy Grandparents, Georgia State University, 140 Decatur Street, Atlanta, GA 30303, USA

⁎ Corresponding author at: Georgia State University, P.USA. Tel.: +1 404 413 1101; fax: +1 404 413 1090.

E-mail address: skelley@gsu.edu (S.J. Kelley).

0190-7409/$ – see front matter © 2011 Elsevier Ltd. Aldoi:10.1016/j.childyouth.2011.06.021

a b s t r a c t

a r t i c l e i n f o

Article history:Received 26 April 2011Received in revised form 23 June 2011Accepted 24 June 2011Available online 2 July 2011

Keywords:Children raised by grandparentsKinship careChild behavior problems

Over the past few decades there has been a dramatic increase in children being raised by grandparents due toa variety of factors including child maltreatment, parental abandonment, substance abuse, and incarceration.The purpose of this study is to examine the extent of behavior problems in children being raised bygrandmothers and to determine factors in their current environment that are related to child behaviorproblems. The study draws on McCubbin's resiliency model of family stress, adjustment and adaptation.The sample comprised of 230 children ages 2 to 16 years who were being raised by grandmothers in parent-absent homes. The families were predominantly low income and African American. The grandmothers rangedin age from 37 to 80 years. Grandmothers responded to several self-report measures pertaining to childbehavior problems, family resources, social support, and caregiver psychological distress. Trained researchassistants completed a standardized observational measure of the home environment.Results indicated that 31.3% of child participants scored in the clinically elevated range for total behaviorproblems, with 21.3% and 32.6% scoring in the elevated range for internalizing and externalizing behaviors,respectively. Hierarchical regression analyses revealed that of the variables examined, increased psychologicaldistress in grandmothers was most predictive of child behavior problems, followed by less supportive homeenvironments and fewer family resources.Results of the present study underscore the need for interventions that focus on reducing child behaviorproblems, as well as enhancing the parenting skills of grandmothers raising grandchildren. Findings alsosupport the need for strategies to reduce stress in grandmother caregivers and to improve access to resourcesneeded to provide supportive home environments for their grandchildren.

O. Box 3995, Atlanta, GA 30302,

l rights reserved.

© 2011 Elsevier Ltd. All rights reserved.

1. Introduction

An increasing number of children are being raised by grandparents,yet limited information is available about their well-being. Between1990 and2000 therewas a 30% increase in the number of grandparentsproviding primary care for their grandchildren (Hardy, 2005). Thisincrease has been attributed primarily to rises in parental substanceabuse, incarceration, and violence, as well as the trend to placechildren with relatives over non-relatives when removed by childprotective services (CPS). According to the 2010 Current PopulationSurvey, over 1.6 million children are being raised by grandparents inparent-absent households (U. S. Bureau of the Census, 2010). For thepurposes of this paper, the term “children raised by grandparents”refers to children in the primary care of grandparents, whetherthrough informal arrangements or through placement by the publicchild welfare system.

Being raised by grandparents can occur abruptly or after a long anddifficult period with the biological parents. Oftentimes, there aremultiple and interrelated reasons why children are raised bygrandparents. The most commonly reported reasons are parentalsubstance abuse, child maltreatment, abandonment, mental illness,incarceration, and homicide (Kelley, Whitley, Sipe, & Yorker, 2000;Smith, Palmieri, Hancock, & Richardson, 2008; Weber & Waldrop,2000). Thus, many children raised by grandparents have experiencedmultiple adverse events that place them at increased risk foremotional and behavioral problems. Furthermore, when birth parentsare no longer able to provide care, children can experience significantdisruptions in key attachment relationships which are critical to theirlater social and emotional development (Poehlmann, 2003).

Despite a body of research indicating that children in the foster caresystem experience increased emotional and behavioral problems(Holtan, Ronning, Handegard, & Sourander, 2004; Leslie, Hurlburt,Landsverk, Barth, & Slymen, 2004; McMillen et al., 2005), studiesfocusing on children raised by grandparents are less available. Of thelimited studies available, findings suggest that these children also haveincreased behavior problems (Campbell, Hu, & Oberle, 2006; Edwards,

2139S.J. Kelley et al. / Children and Youth Services Review 33 (2011) 2138–2145

2006, 2009; Ghuman,Weist, & Shafer, 1999). It is important to note thatmany of these studies are limited by relatively small sample sizes, use ofretrospective chart reviews, or use of non-standardized measures.

Researchers comparing children raised by grandparents to thoseraised by a single biologic parent, as well as to those raised by twobiological parents, found that children in grandparent-headed homeswere perceived to be academically less successful than children intwo-parent families, but equally well behaved at school (Solomon &Marx, 1995). When children raised by grandparents were comparedto children from single parent families, they were perceived to bebetter behaved and comparable on academic performance. Otherresearchers, however, found significantly higher levels of behavioraland emotional problems with children in the care of grandparents incontrast to the general population (Smith & Palmieri, 2007).

1.1. Caregiver characteristics and home environment

Previous studies have found that the family environments ofchildren in foster care can influence their current well-being. Fosterchildren in better quality and more nurturing home environmentsmanifest fewer behavioral problems than those in lower quality andless nurturing home environments (Dolan, Casanueva, Smith, &Bradley, 2009; Lindhiem & Dozier, 2007; Orme & Buehler, 2001). Todate, there is very limited information available on the influence offamily characteristics when grandmothers are the primary careproviders. This current study helps to fill this void by examiningfactors related to the caregiving environment, including caregiverattributes, and the well-being of children raised by grandmothers.

Although grandparents raising grandchildren demonstrate re-markable commitment to their grandchildren, the considerablechallenges they encounter may hinder their ability to provide asupportive home environment. These challenges often includeinadequate financial resources, the physical demands of parentinglater in life, and increased psychological distress. A number of studiesindicate that these caregivers tend to be economically disadvantaged,often living in poverty, with minority grandparents experiencing thegreatest economic vulnerability (Hayslip & Kaminski, 2005; Minkler &Fuller-Thomson, 2005; Simmons & Dye, 2003). Because the majorityof children raised by grandparents are there through informalarrangements, their access to essential resources may be diminished.In one study researchers found that grandmothers raising grand-children under the auspices of the state foster care system had betteraccess to essential resources than those raising grandchildren throughinformal arrangements (Goodman, Potts, & Pasztor, 2007). Comparedto other substitute caregivers, grandmother caregivers tend to beolder (Dolan et al., 2009), averaging around 57 years of age, butfrequently reaching into their 60s and 70s (Kelley,Whitley, & Campos,2010; Smith et al., 2008).

A substantial body of literature indicates a propensity forsignificant psychological distress among caregiving grandmothers(Fuller-Thomson & Minkler, 2000; Kelley et al., 2000; Scarcella, Ehrle,& Geen, 2003). For example, researchers using a nationally represen-tative sample determined that grandmothers raising grandchildrenwere more likely to have significantly higher levels of depressivesymptomatology than non-caregiving grandmothers (Fuller-Thomson& Minkler, 2000). In another population-based study, researchersfound that nearly one-third of grandchildren were raised bygrandparents experiencing poor mental health (Scarcella et al.,2003). Risk factors for increasedpsychological distress in grandmothercaregivers include lack of economic resources, physical healthchallenges, and, to a lesser extent, lack of social support (Kelley etal., 2000). Other factors that may contribute to their psychologicaldistress include parenting children with emotional or developmentalchallenges, coping with the negative circumstances surrounding thegrandchild's parents (e.g. substance abuse, incarceration, mentalhealth issues), and managing health problems associated with aging.

Parenting later in life, when one would not normally expect to havefull-time child care responsibilities, may add to their distress. Inaddition to grandchildren, these grandmothers may also be caring forill spouses or aging parents. Considering these comparative disadvan-tages and challenges, research on the caregiving environment ofchildren raised by grandparents is imperative.

Studies are beginning to emerge that focus specifically on thecaregiver characteristics and home environment of caregivinggrandmothers (Dolan et al., 2009; Smith et al., 2008). Using anationally representative sample of children reported to CPS formaltreatment, researchers compared the parenting and homeenvironments provided by grandmothers to non-relative fostercaregivers (Dolan et al., 2009). Findings indicated that grandmothercaregiverswere older, lesswell educated, less likely to bemarried, andmore likely to be low income than non-relative foster care providers.However, after controlling for the race, education, and poverty level ofthe caregivers, as well as the age of the children, grandmothersdemonstrated significantly better parenting than non-relative fostercaregivers. The researchers concluded that the home environmentprovided by grandmothers was generally comparable to that providedby foster caregivers.

To date, only one study has identified specific factors in thegrandmother caregiving environment that influence child outcomes(Smith et al., 2008). Employing a large national sample of grandmothersraising grandchildren, researchers found that the influence of grand-mothers' psychological distress on their grandchildren's adjustmentwas mediated by dysfunctional parenting, especially with regards toexternalizing behavior problems (Smith et al., 2008). The effects ofcontextual factors, such as social support and family dysfunction, ongrandchildren's adjustment were also indirect. The researchers con-cluded that caregiving grandmothers' distress levels do not bear anincreased risk for grandchildren's psychological difficulties independentof the family's psychosocial circumstances as a whole.

In summary, the limited available literature indicates an elevatedrisk for behavioral problems among children raised by grandmothers.The increased levels of psychological distress found in grandmothercaregivers raise questions regarding their ability to parent mosteffectively. Furthermore, the lack of resources experienced by many ofthese caregiversmaynegatively impact the caregiving environment andsubsequently affect the emotional well-being of the grandchildren.

1.2. Theoretical framework

The current study draws on the resiliency model of family stress,adjustment and adaptation (McCubbin, Thompson, & McCubbin, 1996)to conceptualize how the stressors placed on the family system mightinfluence the well-being of grandchildren raised by grandmothers. Theresiliency model was developed to provide an understanding of howsome families are able to cope, endure, and survive when faced withadversity (DeMarco, Ford-Gilboe, Friedmann, McCubbin, & McCubbin,2000). The model considers the challenges to the family systemimposed by family demands and how resources can affect individualor family adaptation and well-being (e.g. child behavior problems).According to this model, a failure to alleviate or resolve family demandsthrough resources may increase the likelihood of negative outcomes,including compromised mental health.

1.3. Purpose and hypothesis

The purpose of this study is to examine the extent of behaviorproblems in children raised by grandmothers and to describe therelationship among select background variables, as well as length oftime in care, family resources, social support, the home environment,grandmother psychological distress and child behavior problems.Based on available literature and the theoretical framework, wehypothesized that less optimal family resources, social support, home

2140 S.J. Kelley et al. / Children and Youth Services Review 33 (2011) 2138–2145

environments, and grandmother psychological distress would predictincreased behavior problems in children raised by grandmothers,after controlling for select background variables (grandmother age,level of education, number of grandchildren under care, length of timewith grandmother, and grandchild age).

2. Methods

2.1. Participants

Data on child behavior problems and the characteristics of thecaregiving environment came from a larger study examining theefficacy of an intervention to improve the well-being of grandparent-headed families. Analyses for this study are based on data collectedpre-intervention. Families were eligible for participation in this studyif a grandmother was raising one or more grandchildren, ages 2 to16 years, in a parent-absent household. Children under the age of twowere excluded because of age related measurement issues. Because ofthe small number of grandfathers raising grandchildren, they wereexcluded from the present study. Participants were eligible forinclusion if they resided in one of two counties in a majormetropolitan area in the Southeast. There were no inclusion criteriabased on income level, race, or ethnicity. Grandmothers raisinggrandchildren through either the state foster care system or informalarrangements among family members were eligible for participation.

On the basis of the inclusion and exclusion criteria described above,data on 511 children from 230 families were originally available foranalysis. As some grandmothers hadmore than one grandchild in theircare, the following procedure was used to obtain data on onegrandchild for each grandparent with two or more children. Thiswas to ensure that analyses would not include duplicated ratings fromany grandparent measure used to predict child behavior scores.Ninety-two of the 230 families had only one grandchild; all of thesegrandchildren were retained. Among the 138 remaining families, onechildwas selected at randomusing their unique IDnumber to compareagainst a table of random numbers (Excel generated). In order topreserve the demographics of the original sample, grandchildren's ageand genderwere compared between the new random samplewith thesingle children, and the original sample. If a significant difference inage or distribution of gender was found, the randomization sequencewas repeated and a new sample generated. After three iterations, nosignificant differences were found in age or gender distribution,indicating that the new sample (N=230) and the original sample(N=511) of grandchildren were demographically similar. The finalsample for the present study comprised 230 children with a mean ageof 8.2 years (SD=3.6) and a range of 2 to 16 years. The majority ofchildren were African American (97.8%); slightly over half (53.9%)were male.

2.2. Procedures

A variety of community agencies, including health care clinics, childcare centers, public schools, churches, and child welfare agencies,referred participants to the study. As part of the research protocolapproved by the university's institutional review board, all grandparentparticipants signed letters of informed consent. Children aged 11 to16 years provided written assent and those aged 6 to 10 years gaveverbal assent. Prior to implementation of an intervention, trainedgraduate student research assistants collected data in participants'homes. Data were obtained during both structured interviews andobservations, as well as through administration of standardizedmeasures. Because of the overall low educational attainment of thesample, research assistants read questionnaires to participants. Partic-ipants were compensated $40 for data collection.

2.3. Measures

2.3.1. Child behavior checklist (CBCL)The CBCL is a widely used, empirically derived measure of

children's behavior problems (Achenbach, 1991, 1992). Whilemultiple versions are available based on the informant, we used theParent Report Form. Achenbach created two versions based on agerange. The CBCL-4-18 has 113 items pertaining to behavioralproblems, while the CBCL-2-3 consists of 100 items. Respondentsindicate on a 3-point Likert scale which problems their child displays.Items are scored according to internalizing, externalizing, and totalbehavior problems. T scores and clinical cut-off scores are determinedfrom corresponding raw scores, based on the appropriate norm groupfor age. We used the standardized T scores as they provide anadvantage over raw scores in that they are adjusted for child age andgender. The reliability and validity of the CBCL are well established(Achenbach, 1991). For the current sample, the reliability coefficientswere .75 for internalizing behaviors, .83 for externalizing behaviors,and .88 for total behavior problems. Achenbach (1991) identified thecut point T≥60 as providing efficient discrimination betweenchildren referred for mental health services and those who are notreferred. Thus, T scores of≥60, which is 1 standard deviation abovethe mean, serves as the clinical cutoff.

2.3.2. Home observation measurement of the environment (HOME)The HOME is a widely used screening inventory that was designed

to assess the quality of the home environment in terms of the type andamount of support made available to the child that would fosteremotional, social, and cognitive development (Caldwell & Bradley,1984, 2003). Four versions (ranging from 45 to 60 items) have beencreated based on various age groups: the infant and toddler version(ages 0 to 2), the preschool version (ages 3 to 6), the elementaryversion (ages 6 to 10), and the early-adolescence version (ages 10 to14). All four versions were used for this study. The assessment(approximately one hour) is administered through direct in-homeobservation of interactions between child and caregiver, as well asthrough a brief semi-structured interview between observer andcaregiver. Specific psychometric data are available for each HOMEversion, including internal consistency scores, test–retest correlations,and well established concurrent and predictive validity (Caldwell &Bradley, 1984, 2003). For the current sample, the reliability co-efficients were .94 for infant toddler scores, .94 for early childhoodscores, .88 for middle childhood scores, and .86 for adolescent scores.

2.3.3. Brief symptom inventory (BSI)The BSI is a 53-item, self-report symptom inventory designed to

measure the psychological symptom patterns of psychiatric andmedical patients as well as community nonpatient respondents(Derogatis, 1993). The instrument is a shorter version of the SymptomChecklist-90-R (Derogatis, 1983). Participants rate each item of themeasurement on a 5-point scale of distress ranging from 0 (“not atall”) to 4 (“extremely”). The BSI has shown high levels of both internalconsistency and test–retest reliability and has well-establishedvalidity. The BSI is scored and evaluated in terms of three globalindices of distress, including the Global Severity Index (GSI), a globalmeasure of stress. By combining information about numbers ofsymptoms and intensity of distress, the GSI is considered the bestsummary score for psychological distress. Thus, the GSI was used tomeasure psychological distress in this study. For the current sample,the GSI reliability coefficient was .90.

2.3.4. Family resource scale (FRS)We chose the FRS (Dunst & Leet, 1987; Dunst, Trivette, & Deal,

1988) to measure caregivers' perceptions of adequacy of familyresources. The FRS is a 31-item, self-report inventory derived from aconceptual framework that predicts an inadequacy of resources will

Table 1Descriptive statistics on independent and dependent variables.

Mean SD Median Skewness

Total behavior problemsa 53.31 12.02 54 0.147Internalizing behavior problemsa 50.41 11.94 51 0.293Externalizing behavior problemsa 53.69 11.48 53 0.307Child age 8.21 3.59 8 0.003Grandmother age 56.07 8.75 55 0.381Number children in household 2.73 1.64 2 0.905Length of time with grandparent (months) 62.01 47.24 53 0.644Psychological distressa 55.96 10.60 57 −0.147HOME totalb 42.10 8.68 43 −0.073FRS totalb 99.42 18.21 99 −0.075FSS totalb 24.67 11.06 23 0.067HS graduate 51%bHS graduate 49%

HOME, Home Observation Measurement of the Environment; FRS, Family ResourceScale; FSS, Family Support Scale. HS, High school.

a Low score optimal.b High score optimal.

Table 2Child behavior checklist clinical scores.

Total Internalizing Externalizing

Nonclinical (T≤59) 68.7% 78.7% 67.4%Clinically elevated (T≥60) 31.3 21.3 32.6

2141S.J. Kelley et al. / Children and Youth Services Review 33 (2011) 2138–2145

negatively affect personal well-being and parental commitment.Items refer to specific resources and are rated on a 5-point Likertscale. Total scores are obtained by adding the score for each item, withhigher scores indicating more resources. The measure has goodinternal consistency (.92) and test–retest reliability (.52). For thecurrent sample, the reliability coefficient was .88.

2.3.5. Family support scale (FSS)The FSS (Dunst & Trivette, 1989) measures the helpfulness of

various sources of support to families raising children. The FSS includes18 items that are rated on a 5-point scale from1 (“not at all helpful”) to5 (“extremely helpful”). Total scores are obtained by adding the scorefor each item, with higher scores indicating a perception of increasedsocial support. The measure has strong validity and reliability. For thecurrent sample, the reliability coefficient was .70.

2.4. Analyses

Prior to analysis, we examined the variables through various SPSSprograms for assumptions of multivariate analysis. Initial examinationof the data revealed no violations of multiple regression assumptionsamong the variables (normal distribution, multicollinearity, homosce-dasticity). Descriptive statistics were used to summarize demographicandbackgrounddata, aswell as todetermineproportionsof participantswith clinically elevated behavior problem scores. Correlations wereused to determine relationships among continuous independent anddependent variables.

Hierarchical regression was employed to determine if familyresources, family support, home environment, and grandmotherpsychological distress improved prediction of child behavior scoresbeyond that attributed to differences in grandmother age and level ofeducation, number of children in her care, and length of time caringfor grandchild. Regression analysis was conducted using the entermethod in SPSS for the dependent variables of externalizing andinternalizing child behavior problems.

3. Results

As a group, the grandmother caregivers (N=230) had loweducational attainment, with approximately half (49%) having lessthan a high school education. While the majority were grandmothers(96%), some were great-grandmothers (4%). For the purposes of thispaper, the great-grandmothers are referred to as grandmothers. Thirtypercent of participants were currently employed; slightly over half(51.7%)were unemployed for various reasons, and 18.3%were retired.Themean age of the grandmother caregiverswas 56.1 years (SD=8.7)(range=37 to 80 years), with almost a third (32.6%) over the age of60. The marital status of the grandmothers was as follows: divorced/separated, 41.3%; widowed, 24.3%; married, 19.2%; and single, 15.2%.On average, the grandmothers were raising 2.7 children (SD=1.6)(range=1 to 8). The vast majority (98.7%) of grandmothers wereAfrican American.

Participants reported the following as themain reasons for assumingfull-time care of their grandchildren: child abuse or neglect, 78.7%;parental alcohol or drug abuse, 67%; child abandonment, 37%; childremoved by CPS, 18%; one or both parents deceased, 17%; one or bothparents incarcerated, 16%; and parent with HIV disease, 3%. It isimportant to note that many of these factors are interrelated and, thus,exceed 100%when added together.With regard to the type of past childmaltreatment, neglectwas themost common form (76.5%), followed byemotional abuse, 34%; physical abuse, 15%; and sexual abuse, 3%.Despite the finding that most children had experienced one or moreforms of maltreatment, only a small proportion (6.8%) of children werein state foster care; the remainder (93.2%) were in informal kinshiparrangements. Descriptive statistics for all independent and dependentvariables, as well as skewness values are found in Table 1.

3.1. Child behavior problems

The first objective was to determine the proportion of childrenwith total, internalizing, and externalizing behavior problem scores inthe clinically elevated range (see Table 2). Results indicate that 31.3%of participants scored in the clinically elevated range for totalproblems, with 21.3% and 32.6% scoring in the clinically elevatedrange for internalizing and externalizing problems, respectively.

3.2. Bivariate analyses

Correlations for the continuous variables of child's age, number ofgrandchildren in care, grandmother's age, length of time living withgrandmother, family support total score, family resource total score,home environment total score, grandmother psychological distress,and total, externalizing and internalizing child behavior scores appearin Table 3. Using Cohen's (1977) recommendations, correlations inthis study are termed weak (R=.1–.3), moderate (R=.3–.5), orstrong (R≥ .5). Participants reporting more family resources and lesspsychological distress tended to report fewer internalizing andexternalizing behavior problems. These relationships were moderatein strength. It can also be noted that grandmothers with fewergrandchildren in their care tended to report more internalizingbehavior problems, although this relationship was weak. A moderaterelationship was found among family resources and family socialsupport. Grandmothers with more children, as well as those withyounger children, tended to report fewer resources, although theserelationships were weak. Those with less supportive home environ-ments tended to have more grandchildren in the home and reportedmore internalizing and externalizing behavior problems, as well asfewer resources, less social support, and increased psychologicaldistress. These relationships were also weak.

3.3. Multivariate analyses

The second objective was to describe the relationship among selectbackground variables, family resources, social support, the homeenvironment, grandmother psychological distress and behavior prob-lems in children raised by grandmothers. We conducted two separatemultiple regressions using standardized scores for internalizing and

Table 3Correlations of independent and dependent variables.

1 2 3 4 5 6 7 8 9 10 11

1. Total problems –

2. Externalizing .90⁎⁎⁎ –

3. Internalizing .87⁎⁎⁎ .69⁎⁎⁎ –

4. Child age .05 .06 .03 –

5. No. children in care −.15⁎ −.03 −.18⁎⁎⁎ −.01 –

6. Grandparent age −.11 −.10 −.09 .22⁎⁎ −.06 –

7. Family support −.07 −.01 −.09 −.03 −.06 −.11 –

8. Family resources −.26⁎⁎⁎ −.23⁎⁎⁎ − .25⁎⁎⁎ .13⁎ −.16⁎ .15⁎ .47⁎⁎⁎ –

9. Psychological distress .51⁎⁎⁎ .47⁎⁎⁎ .48⁎⁎⁎ −.01 .01 − .19⁎⁎ −.19⁎⁎ −.45⁎⁎⁎ –

10. HOME environment − .21⁎⁎ −.19⁎⁎ −.17⁎ .12 −.21⁎⁎ .03 .16⁎ .17⁎⁎ −.17⁎⁎ –

11. Length of time in care .06 .09 .02 .48⁎⁎⁎ .05 .22⁎⁎ .07 −.03 .04 −.03 –

HOME, Home Observation Measurement of the Environment.⁎ pb .05.

⁎⁎ pb .01.⁎⁎⁎ pb .001.

2142 S.J. Kelley et al. / Children and Youth Services Review 33 (2011) 2138–2145

externalizing behavior problems as the respective dependent variables.Potential predictors consistedof nine independent variables: child's age,grandmother's age, level of education (high school graduate vs. non-graduate), number of children in grandmother's care, length of timewith grandmother (months), grandmother psychological distress asmeasured by the GSI score, home environment as measured by totalscore on theHOME inventory, family resources asmeasured by the totalFRS score, and family social support as measured by the total FSS score.Independent variableswere entered into hierarchical regressionmodelsfor each analysis.

Because of the likelihood that certain extraneous variables(grandmother age and level of education, number of grandchildrenin her care, and length of time caring for grandchild), would influencethe dependent variable, they were entered first as a block to controltheir effect in Step 1 for each regression analysis. Family resources,family social support, and home environment scores were enteredsimultaneously in Step 2 for two reasons. First, family resources,family social support, and the home environment can be viewed asclosely related in terms of supporting the child's well-being. Second,the theoretical model does not indicate whether family resources,family social support, or home environment is more likely tomoderate child behavior problems. Grandmother psychologicaldistress was entered in Step 3.

3.3.1. Internalizing behaviorsHierarchical multiple regression was first conducted with inter-

nalizing behavior scores as the dependent variable. Table 4 contains

Table 4Regression findings: internalizing behaviors.

Variables entered Step 1 B (Β) Step 2 B (Β)

Education .283 (.012) .231 (.094Child age .099 (.030) .355 (.107Number children in care −1.367 (−.188)⁎ −1.836 (−.25Grandmother age −.152 (−.111) −.081 (−.06Length of time in care .011 (.043) .001 (.004Family support scale .048 (.045Family resource scale −.185 (−.28HOME environment −.297 (−.21Psychological distress

ANOVA tests at each step F (5, 224)=2.152 F (8, 221)=5

R at each step .214 .401R2 (adjusted R2) at each step .046 (.025) .161 (.130Intercept 60.732 83.413

Note: Initial R2 at step 1 is given, followed by subsequent incremental R2 changes at steps⁎ pb .05.⁎⁎ pb .001.

the unstandardized regression coefficients (B) and intercept, thestandardized coefficients (β), and the semipartial correlations (sr2)for each step, and the multiple correlation coefficient (R), the squaredmultiple correlation coefficient (R2), and the adjusted squaredmultiple correlation (adjusted R2) after entry of all nine independentvariables. R was significantly different from zero at the end of eachstep. After Step 3, with all nine independent variables in the equation,R=.55, F (9, 220)=10.33, and p=b.001.

After Step 1, with number of grandchildren in care, grandmotherage, education level (bHS graduate), grandchild age, and length oftime with grandmother in the equation, R2=.05, F (5, 224)=2.15,and pb .05. After Step 2, with family social support, family resources,and home environment scores added, R2=.16, F (8, 221)=5.29, andpb .001. After Step 3, with psychological distress in the equation,R2=.30, F (9, 220)=10.33, and pb .001.

3.3.2. Externalizing behaviorsNext, hierarchical multiple regression was conducted with

externalizing behaviors as the dependent variable. Table 5 containsthe unstandardized regression coefficients (B) and intercept, thestandardized coefficients (β), and the semipartial correlations (sr2)for each step, and the multiple correlation coefficient (R), the squaredmultiple correlation coefficient (R2), and the adjusted squaredmultiple correlation (adjusted R2) after entry of all nine independentvariables. R was significantly different from zero at the end of thesecond and third steps. After Step 3, with all nine independentvariables in the equation, R=.51, F (9, 220)=8.45, and p=b.001.

Step 3 B (Β) R square change (see note)

) 2.352 (.099) Step 1: .046) .261 (.078)2)⁎⁎ −1.528 (−.210)⁎⁎

0) .013 (.009)) −.005 (−.021)) .046 (.042) Step 2: .115⁎⁎

3)⁎⁎ −.065 (−.099)6)⁎⁎ −.227 (−.165)⁎

.479 (.426)⁎⁎ Step 3: .136⁎⁎

.290⁎⁎ F (9, 220)=10.332⁎⁎

.545) .297 (.268)

36.582

2 and 3.

Table 5Regression findings: externalizing behaviors.

Variables entered Step 1 B (Β) Step 2 B (Β) Step 3 B (Β) R square change (see note)

Education −1.353 (−.059) .125 (.005) .239 (.010) Step 1: .031Child age .123 (.038) .374 (.117) .285 (.089)Number children in care −.398 (−.057) −.805 (−.115) −.516 (−.074)Grandmother age −.193 (−.147)⁎ −.114 (−.087) −.026 (−.019)Length of time in care .025 (.103) .013 (.054) .007 (.030)Family support scale .136 (.131) .134 (.129) Step 2: .096⁎⁎

Family resource scale −.177 (−.281)⁎⁎ −.064 (−.101)HOME environment −.267 (−.202)⁎⁎ −.201 (−.152)⁎

Psychological distress .450 (.416)⁎⁎ Step 3: .130⁎⁎

ANOVA tests at each step F (5, 224)=1.419 F (8, 221)=4.006⁎⁎ F (9, 220)=8.457⁎⁎

R at each step .175 .356 .507R2 (adjusted R2) at each step .031 (.009) .127 (.095) .257 (.227)Intercept 65.073 83.686 39.679

Note: Initial R2 at step 1 is given, followed by subsequent incremental R2 changes at steps 2 and 3.⁎ pb .05.⁎⁎ pb .001.

2143S.J. Kelley et al. / Children and Youth Services Review 33 (2011) 2138–2145

After Step 1with number of grandchildren in care, grandmother age,education level, number of grandchildren in care, grandchild age, andlength of timewith grandmother in the equation, R2=.03, F (5, 224)=1.42, and pb .05. After Step 2 with family social support, familyresources, and home environment scores added, R2=.13, F (8, 221)=4.01, and pb .001. After Step 3 with psychological distress in theequation, R2=.26, F (9, 220)=8.45, and pb .001.

For both regression analyses, psychological distress was the mostpredictive variable, suggesting that higher psychological distress ingrandmothers raising grandchildren is associated with higher internal-izing and externalizing child behavior problems. The next mostpredictive variable was home environment. Children with lesssupportivehomeenvironments displayed increasedbehavior problems,both internalizing and externalizing. Fewer family resources, to a lesserextent, were predictive of increased behavior problems in bothmodels.

4. Discussion

The finding that almost one-third of children in the present studyhad clinically elevated behavior problem scores is not unexpectedconsidering the vast majority had experienced multiple adverseevents prior to living with grandmothers, such as child maltreatment,abandonment, or the incarceration of one or both parents. Further-more, children raised by grandmothers typically experience adisruption in attachment relationships with their birth parents,placing them at increased risk for emotional distress.

The clinically elevated child behavior problems found in this studyfall within the range of those found in numerous studies of children instate foster care, whether residing with relative or non-relativecaregivers (Dubowitz et al., 1994; James, Landsverk, & Slymen, 2004;Leslie et al., 2004, 2005; Lindhiem & Dozier, 2007). These behaviorproblems raise concerns for grandmother caregivers as they undoubt-edly contribute to the myriad of challenges already faced bygrandmothers who are trying to provide a safe and stable environment.Coping with these child behaviors can contribute to, or exacerbate, theelevated psychological distress levels found in grandparents raisinggrandchildren (Fuller-Thomson & Minkler, 2000; Kelley et al., 2000;Minkler & Fuller-Thomson, 2001; Ross & Aday, 2006; Scarcella et al.,2003). If grandmother caregivers reach a point inwhich they are unableto effectively copewith problematic behaviors, their grandchildrenmaybe at risk for placement in the state foster care system, a residentialsetting, or even in the juvenile justice system. Research has shown thatdisruptions in foster care placements lead to further emotional trauma,thereby worsening behavior problems (Leathers, 2006; Newton,Litrownik, & Landsverk, 2000). It is imperative, therefore, that children

in the care of grandmothers receive mental health services, whenwarranted. Additionally, the grandmothers may benefit from assistancein developing effective strategies for coping with problematic childbehaviors.

Children of grandmother caregivers who had increased psycho-logical distress, fewer family resources, less social support, and lesssupportive home environments had increased behavior problems,both internalizing and externalizing. These findings are consistentwith previous research on the effects of foster family characteristicsand the home environment on children's behavioral functioning(Lindhiem & Dozier, 2007). The results also support a major premiseof McCubbin's family resiliencymodel, which served as the theoreticalbasis of the current study, in that inadequate resources were found tobe associated was less favorable outcomes. The association foundbetween child outcomes and custodial grandmothers' psychologicaldistress, as well as other contextual factors in the present study,supports the findings of other researchers (Smith et al., 2008). Thepresent study expands on prior research (Dolan et al., 2009) byexamining the role of the home environment on behavior outcomes inchildren raised by grandmothers primarily through informal arrange-ments among family members.

It remains unknown whether the elevated behavior problemsreported in this study are the result of conditions precipitatingplacement with grandparents (e.g. child maltreatment, parentalincarceration), disruption in attachment relationships with birthparents, or circumstances in the current caregiving environment. Adirect causal relationship between grandparent caregiving and grand-child behaviors should, therefore, not be assumed. The findings do,however, suggest that factors related to the caregiving environment(e.g. lack of resources) and attributes related to grandparent caregivers(e.g. psychological distress) are closely related to the child's well-being,with caregiver psychological distress having the strongest association.

From the bivariate analyses, the finding that less supportive homeenvironments are associated with increased behavior problems isconsistent with other study findings of children in foster care (seereview by Orme & Buehler, 2001), as well as in the general population(Bradley & Corwyn, 2007; Votruba-Drzal, 2006). At a time in theirlives when income is typically limited, grandparents often do not havethe economic resources necessary to adequately provide for grand-children. Many grandparents are subsisting on “working poor” wagesor fixed incomes from retirement or social security benefits. Whenpublic assistance is available through programs such as TemporaryAssistance to Needy Families (TANF) or kinship care stipends, it isoften inadequate to properly house, clothe, and feed children. Theincreased health problems experienced by grandmothers raising

2144 S.J. Kelley et al. / Children and Youth Services Review 33 (2011) 2138–2145

grandchildren may further drain limited economic resources (Kelleyet al., 2010; Whitley, Kelley, & Sipe, 2001).

In this study, grandmothers with fewer children in their caretended to report more internalizing behavior problems. One possiblereason is that grandmothers with fewer children under their care maybe in a better position to notice the more subtle signs of internalizingbehaviors such as withdrawal, anxiety and depression. Anotherplausible explanation is that children with siblings may transitionmore successfully from a negative caregiving environment to livingwith grandmothers than children with no siblings. In other words,siblings may buffer some of the stress associated with difficult familysituations.

4.1. Methodological limitations

Several limitations may qualify the current findings. First, thesample is limited primarily to African American children whosegrandmothers had predominantly low educational attainment andwere residing in an urban setting in the Southeast. Second, thegrandchildren were primarily in the care of grandmothers throughinformal arrangements. Third, with the exception of the homeenvironment assessment, caregiver reports served as the primarymeasurements of both caregiving environment and child outcomes,raising concerns about shared method variance. Interpretation ofthese findings must consider the potential of the caregivers' ownlevels of psychological distress to influence their assessments of thegrandchildren's distress levels.

4.2. Implications for research, policy, and practice

In order to identify critical intervention points, longitudinalresearch is needed to investigate the emotional well-being of childrenraised by grandmothers and their adjustment over time. Additionalstudieswith participantswho aremore diversewith regard to race andethnicity, socioeconomic status, geographical settings, and types ofplacements (state foster care vs. informal arrangements) will furtherour understanding of the influence of caregiving characteristics andhome environment on child outcomes.

Study findings have several implications for policy and practice.Interventions that focus on reducing child behavior problems, as wellas enhancing the parenting skills of grandmothers raising grand-children are warranted. Investigators have identified an increasingnumber of evidence-based interventions that improve outcomes forchildren with behavioral problems and facilitate caregivers' ability tocope (i.e., McMahon & Forehand, 2003; Timmer, Urquiza, Zebell, &McGrath, 2005). Very few, however, have focused on children infoster care, whether through formal or informal arrangements(Chamberlain et al., 2006). Children with behavioral problems requirespecialized support services and resources that may not be easilyaccessible to grandmothers because of limited financial resources orlack of knowledge about the types of services required. Theassociation found between the amount of support in the homeenvironment and child outcomes speaks to the need for practicestrategies and public policy that address the community-based needsof grandmothers raising grandchildren, especially as it relates toeconomic resources for raising children in a supportive environment.

References

Achenbach, T. M. (1991). Manual for the child behavior checklist-4-18 and the 1991profile. Burlington, VT: University of Vermont Department of Psychiatry.

Achenbach, T. M. (1992).Manual for the child behavior checklist-2-3 and the 1992 profile.Burlington, VT: University of Vermont Department of Psychiatry.

Bradley, R. H., & Corwyn, R. F. (2007). Externalizing behavior problems in fifth grade:Relations with productive activity, maternal sensitivity, and harsh parenting frominfancy through middle childhood. Developmental Psychology, 43, 1390–1401.

Caldwell, B. M., & Bradley, R. H. (1984). Home observation for measurement of theenvironment. Little Rock, AR: University of Arkansas.

Caldwell, B. M., & Bradley, R. H. (2003). Home observation for measurement of theenvironment: Administration manual. Little Rock, AR: University of Arkansas.

Campbell, L. R., Hu, J., & Oberle, S. (2006). Factors associated with future offending:Comparing youth in grandparent-headed homes with those in parent-headedhomes. Archives of Psychiatric Nursing, 20, 258–267.

Chamberlain, P., Price, J. M., Reid, J. B., Landsverk, J., Fisher, P. A., & Stoolmiller, M.(2006). Who disrupts from placement in foster and kinship care? Child Abuse &Neglect, 30, 409–424.

Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New York:Academic Press.

DeMarco, R., Ford-Gilboe, M., Friedmann, M. L., McCubbin, H. I., & McCubbin, M. (2000).Stress, coping, and family health. In V. H. Rice (Ed.), Handbook of stress, copingand health: Implications for nursing research, theory, and practice (pp. 295–332).Thousand Oaks, CA: Sage.

Derogatis, L. R. (1983). SCL-90 administration, scoring, and procedures manual-II. MD:Clinical Psychometric Research.

Derogatis, L. R. (1993). Brief symptom inventory manual. Towson, MD: ClinicalPsychometric Research.

Dolan, M. M., Casanueva, C., Smith, K. R., & Bradley, R. H. (2009). Parenting and thehome environment provided by grandmothers of children in the child welfaresystem. Children and Youth Services Review, 31, 784–796.

Dubowitz, H., Feigelman, S., Harrington, D., Starr, R., Zuvarin, S., & Sawyer, R. (1994).Children in kinship care: How do they fare? Children and Youth Services Review, 16,85–106.

Dunst, C. J., & Leet, H. E. (1987). Measuring the adequacy of resources in householdswith young children. Child: Care, Health and Development, 13, 111–125.

Dunst, C. J., & Trivette, C. M. (1989). Toward experimental evaluation of the family,infant, and pre-school program. In H. Weiss, & F. Jacobs (Eds.), Evaluating familyprograms (pp. 315–346). New York: Aldine Publishing.

Dunst, C. J., Trivette, C. M., & Deal, A. G. (1988). Enabling and empowering families:Principles and guidelines for practice. Cambridge, MA: Brookline Books.

Edwards, O.W. (2006). Teachers' perceptions of the emotional and behavioral functioningof children raised by grandparents. Psychology in the Schools, 43, 565–572.

Edwards, O. W. (2009). Empirical investigation of the psychosocial functioning ofchildren raised by grandparents. Journal of Applied School Psychology, 25, 128–145.

Fuller-Thomson, E., & Minkler, M. (2000). The mental and physical health ofgrandmothers who are raising their grandchildren. Journal of Mental Health andAging, 6, 311–323.

Ghuman, H. S., Weist, M. D., & Shafer, M. E. (1999). Demographic and clinicalcharacteristics of emotionally disturbed children being raised by grandparents.Psychiatric Services, 50, 1496–1498.

Goodman, C. C., Potts, M. K., & Pasztor, E. M. (2007). Caregiving grandmothers with vs.without child welfare system involvement: Effects of expressed need, formalservices, and informal social support on caregiver burden. Children and YouthServices Review, 29, 428–441.

Hardy, D. R. (2005). The booming dynamics of aging: From awareness to action. Reportof the 2005 White House Conference on Aging Retrieved June 21, 2011, from. http://www.genpolicy.com/articles/whcoa_report_2005.pdf

Hayslip, B., & Kaminski, P. L. (2005). Grandparents raising their grandchildren: A reviewof the literature and suggestions for practice. The Gerontologist, 45, 262–269.

Holtan, A., Ronning, J. A., Handegard, B. H., & Sourander, A. (2004). A comparison ofmental health problems in kinship and non-kinship foster care. European Child &Adolescent Psychiatry, 14, 200–207.

James, S., Landsverk, J., & Slymen, D. J. (2004). Placement movement in out-of-homecare: Patterns and predictors. Children and Youth Services Review, 26, 185–206.

Kelley, S. J., Whitley, D. M., & Campos, P. E. (2010). Grandmothers raising grand-children: Results of an intervention to improve health outcomes. Journal of NursingScholarship, 42, 379–386.

Kelley, S. J., Whitley, D. M., Sipe, T. A., & Yorker, B. C. (2000). Psychological distress ingrandmother kinship care providers: The role of resources, social support andphysical health. Child Abuse & Neglect, 24, 311–321.

Leathers, S. J. (2006). Placement disruption and negative placement outcomes amongadolescents in long-term foster care: The role of behavior problems. Child Abuse &Neglect, 30, 307–324.

Leslie, L. K., Gordon, J. N., Meneken, L., Premji, K., Michelmore, K., & Ganger, W. (2005).The physical, developmental, and mental health needs of young children in childwelfare by initial placement type. Journal of Developmental and BehavioralPediatrics, 26, 177–185.

Leslie, L. K., Hurlburt, M. S., Landsverk, J., Barth, R., & Slymen, D. J. (2004). Outpatientmental health services for children in foster care: A national perspective. ChildAbuse & Neglect, 28, 697–712.

Lindhiem, O., & Dozier, M. (2007). Caregiver commitment to foster children: The role ofchild behavior. Child Abuse & Neglect, 31, 361–374.

McCubbin, H. I., Thompson, A. I., & McCubbin, M. A. (Eds.). (1996). Family assessment:Resiliency, coping, and adaptation. Madison, WI: University of Wisconsin.

McMahon, R. J., & Forehand, R. L. (2003). Helping the noncompliant child: Family-basedtreatment for oppositional behavior (2nd ed.). New York, NY: Guilford Press.

McMillen, J. C., Zima, B. T., Scott, L. D., Auslander, W. F., Munson, M. R., Ollie, M. T., et al.(2005). Prevalence of psychiatric disorders among older youths in the foster caresystem. Journal of the American Academy of Child and Adolescent Psychiatry, 1, 88–95.

Minkler, M., & Fuller-Thomson, E. (2001). Physical and mental health status ofAmerican grandparents providing extensive child care to their grandchildren.Journal of the American Medical Women's Association, 56, 199–205.

Minkler, M., & Fuller-Thomson, E. (2005). African American grandparents raisinggrandchildren: A national study using the Census 2000 American CommunitySurvey. Gerontological Society of America, 60B, S82–S92.

2145S.J. Kelley et al. / Children and Youth Services Review 33 (2011) 2138–2145

Newton, R. R., Litrownik, A. J., & Landsverk, J. (2000). Children and youth in foster care:Disentangling the relationship between problem behaviors and number ofplacements. Child Abuse & Neglect, 24, 1363–1374.

Orme, J. G.,&Buehler,C. (2001). Foster family characteristics andbehavioral andemotionalproblems of foster children: A narrative review. Family Relations, 50, 3–15.

Poehlmann, J. (2003). An attachment perspective on grandparents raising their younggrandchildren: Implications for intervention and research. Infant Mental HealthJournal, 24, 149–173.

Ross, M. E., & Aday, L. A. (2006). Stress and coping in African American grandparentswho are raising their grandparents. Journal of Family Issues, 27, 912–932.

Scarcella, C. A., Ehrle, J., & Geen, R. (2003). Identifying and addressing the needs of childrenin grandparent care (No. B-55). Washington, DC: The Urban Institute.

Simmons, T., & Dye, J. L. (2003). Grandparents living with grandchildren: 2000 (CensusBureau Publication No. C2KBR-31). Washington, DC: U.S. Census Bureau.

Smith, G. C., & Palmieri, P. A. (2007). Risk of psychological difficulties among childrenraised by custodial grandparents. Psychiatric Services, 58, 1303–1310.

Smith, G. C., Palmieri, P. A., Hancock, G. R., & Richardson, R. A. (2008). Custodialgrandmothers' psychological distress, dysfunctional parenting, and grand-

child adjustment. International Journal of Aging & Human Development, 76,327–357.

Solomon, J. C., & Marx, J. (1995). “To grandmother's house we go”: Health andadjustment of children raised solely by grandparents. The Gerontologist, 35,386–395.

Timmer, S. G., Urquiza, A. J., Zebell, N. M., & McGrath, J. M. (2005). Parent–childinteraction therapy: Application to maltreating parent–child dyads. Child Abuse &Neglect, 29, 825–842.

U. S. Bureau of the Census. (2010). Table C4 children with grandparent by presence ofparents, sex, race, and Hispanic origins for selected characteristics, 2010: Currentpopulation survey, 2010 annual social and economic supplement. Retrieved 6-21-11,from. http://www.census.gov/population/www/socdemo/hh-fam/cps2010.html

Votruba-Drzal, E. (2006). Economic disparities in middle childhood development: Doesincome matter? Developmental Psychology, 42, 1154–1167.

Weber, J. A., & Waldrop, D. P. (2000). Grandparents raising grandchildren: Families intransition. Journal of Gerontological Social Work, 33, 27–46.

Whitley, D. M., Kelley, S. J., & Sipe, T. A. (2001). Grandmothers raising grandchildren:Are they at increased risk of health problems? Health & Social Work, 26, 105–114.