See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/7297569

The Impact of Foster Care on Development

Article  in  Development and Psychopathology · February 2006

DOI: 10.1017/S0954579406060044 · Source: PubMed

CITATIONS READS

253 5,356

3 authors, including:

Elizabeth A. Carlson Byron Egeland

University of Minnesota Twin Cities University of Minnesota Twin Cities
73 PUBLICATIONS   9,258 CITATIONS    151 PUBLICATIONS   20,363 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Steep effectiveness View project

All content following this page was uploaded by Byron Egeland on 28 April 2015.

The user has requested enhancement of the downloaded file.

Development and Psychopathology 18 ~2006!, 57–76
Copyright © 2006 Cambridge University Press
Printed in the United States of America
DOI: 10.10170S0954579406060044

The impact of foster care on development

CATHERINE R. LAWRENCE, ELIZABETH A. CARLSON,

and BYRON EGELAND
University of Minnesota

Abstract
Foster care is a protective intervention designed to provide out of home placement to children living in at-risk home
environments. This study employs prospective longitudinal data ~N ⫽ 189! to investigate the effects of foster care
on the development of child behavior and psychological functioning taking into account baseline adaptation prior to
placement and socioeconomic status at the time of placement. Comparisons were made among three groups:
children who experienced foster care, those who were maltreated but remained in the home, and children who had
not experienced foster care or maltreatment despite their similarly at-risk demographic characteristics. In the current
sample, children placed in out of home care exhibited significant behavior problems in comparison to children who
received adequate care, and using the same pre- and postplacement measure of adaptation, foster care children
showed elevated levels of behavior problems following release from care. Similarly, children placed into unfamiliar
foster care showed higher levels of internalizing problems compared with children reared by maltreating caregivers,
children in familiar care, and children who received adequate caregiving. Findings suggest that outcomes related to
foster care may vary with type of care and beyond the effects associated with maltreatment history, baseline
adaptation, and socioeconomic status.

The foster care social service system is de- Rockefeller, 1999!, although rigorous em-
signed to ameliorate adverse family and envi- pirical investigations of the sequelae of foster
ronmental conditions that may interfere with placement are limited.
typical child development. Currently, the sys- The current study addresses limitations in
tem provides short- and long-term out of the existing literature by evaluating the direct
home placement to children whose parents are effects of foster care on behavior problems
deemed unable to care adequately for them. within a prospective, longitudinal sample of
The effectiveness of foster care as an interven- high-risk children and their families. The study
tion, however, is the subject of controversy. examines the impact of foster care on behav-
Throughout the current foster care literature, ior problems and psychological functioning
removing children from their families of ori- over and above risks associated with baseline
gin and placing them in out of home care has developmental adaptation ~prior to place-
been associated with negative developmental ment! and socioeconomic status ~SES!. We
consequences that place children at risk for begin by reviewing the history and general
behavioral, psychological, developmental, and characteristics of the foster care system, the
academic problems ~Curtis, Dale, Kendall, & potential developmental risks posed for chil-
dren entering the system, and results of empir-
ical studies related to foster care experience.
Preparation of the work and the research described herein During 2001 it was estimated that 542,000
were supported by a National Institute of Mental Health children resided in foster care ~US Depart-
grant ~MN 40864! to Byron Egeland.
ment of Health and Human Services, 2003!.
Address correspondence and reprint requests to: By-
ron Egeland, Institute of Child Development, University The need for foster placements presently ex-
of Minnesota, 51 East River Road, Minneapolis, MN ceeds available homes by 30– 40% ~US De-
55455; E-mail: egela001@umn.edu. partment of Health and Human Services, 2003!.

57
58 C. R. Lawrence, E. A. Carlson, and B. Egeland

Criteria for placing children into foster care hood experience, further challenging already
are not well delineated. Entry is often associ- vulnerable children. Thus, while out of home
ated with a history of child maltreatment per- care is intended to ameliorate adverse caregiv-
petrated by the primary caregiver, failure of ing conditions, the accumulation of experi-
primary caregivers to protect children from ences necessitating placement often render
maltreatment by others, parental chemical ad- children even more vulnerable to emotional
diction, psychological or physical illness of and behavioral difficulties.
the primary caregiver, homelessness, children’s Maltreatment is a common preplacement
behavior problems, poor quality of the parent– experience that poses significant risk for poor
child relationship, and parental abandonment developmental outcomes ~Cicchetti & Toth,
of children ~Arad, 2001; Curtis, 1999; Jones, 2000!. Multifaceted negative sequelae in-
1985; United States General Accounting Of- clude domains of school performance ~Ege-
fice, 1995!. land, 1997!, overall functioning ~i.e., increased
Proponents of foster care note that 70– behavior problems; Egeland, 1991a; Egeland
80% of children in out of home placements & Sroufe, 1981!, and psychopathology ~Ege-
have been maltreated in the home of origin, land, 1997!. Children entering foster care with
and that prevention of further maltreatment is a history of maltreatment may be at increased
achieved in the majority of cases ~Arad, 2001; risk. Research suggests that these children have
Landsverk, 1996; Landsverk & Garland, 1999!. endured more severe abuse and markedly in-
The likelihood of parental recidivism in the adequate care in comparison to maltreated chil-
areas of physical abuse and neglect following dren who do not come to the attention of
reunification is reduced ~Landsverk, 1996; social service providers. A related preplace-
Landsverk & Garland, 1999; Zuravin & ment risk factor is problematic attachment for-
DePanfilis, 1997!, a noteworthy finding be- mation ~Carlson, 1998; Carlson, Cicchetti,
cause 60% of children exiting foster care in Barnett, & Braunwald, 1989; Egeland &
recent years reunify with the preplacement par- Sroufe, 1981; McCrone, Egeland, Kalkoske,
ent and family ~US Department of Health and & Carlson, 1994!. Differences in the quality
Human Services, 2003!. of infant attachment organization are related
to variations in caregiving experience during
the first year of life ~Ainsworth, Blehar, Wa-
Developmental Context of Foster Care ters, & Wall, 1978; Bowlby, 196901982!. Chil-
dren who experience rejecting or insensitive
Developmental researchers recognize that it is care frequently come to expect parental un-
often the association of multiple risk factors availability, and are likely to develop behav-
that derails the potential for positive develop- ioral patterns that have long-term negative con-
mental outcomes ~Rutter, 1987!. The resil- sequences for socioemotional development
ience literature and cumulative risk models ~Egeland & Carlson, 2004; Sroufe, 1996!.
regard chronic poverty, disrupted and dysfunc- Placement of children into foster care is of-
tional family situations, child maltreatment as ten precipitated by failures within the care-
well as foster placement as risks that heighten giving environment. Moreover, foster care, by
vulnerability to maladaptation and psychiatric design, challenges caregiving relationships
disorder ~Egeland, Carlson, & Sroufe, 1993; through extended caregiver–child separations
Garmezy, 1993; Garmezy & Masten, 1994; during infancy and toddlerhood. For some chil-
Masten & Garmezy, 1985; Masten & Wright, dren, separations may be experienced as sig-
1998; Rutter, 1987!. Out of home care may be nificant rejection or loss, compounding a
associated with a single or multitude of risks history of parental unavailability ~Bowlby,
and0or chronic exposure to adverse circum- 1960; Dozier, Stovall, Albus, & Bates, 2001!
stances within the context of the home envi- and potentially distorting the child’s adjust-
ronment ~Masten & Wright, 1998!. In addition ment to surrogate caregivers and the foster
to these baseline risks, entry into foster care home environment ~Cummings & Cicchetti,
itself lies outside of the range of typical child- 1990!.
Foster care and development 59

Sequelae of Foster Care 2000!. However, because these studies were

based on foster parent report with varying de-
A broad review of foster care research sug- grees of contact with the participants ~3 months
gests that foster children deviate from typical to several years!, it is difficult to evaluate
development in all domains and are at signif- whether findings were related to severity of
icant risk for unusually high rates ~30–80%! symptoms, transient behavioral responses, or
of psychological and behavioral problems and degree of familiarity between foster mother
special needs ~Arad, 2001; Hochstadt, Jaudes, and child.
Zimo, & Schachter, 1987; McIntyre & Keesler, Controlling for degree of familiarity with
1986; Rutter, 2000; Zima et al., 2000!. Dispa- the children, Clausen, Landsverk, Ganger,
rate methodologies employed by researchers Chadwick, and Litrownik ~1998! examined
and differing sample sizes, however, yield con- licensed foster parent CBCL reports during
flicting specific conclusions. the first 2– 4 months of foster care ~N ⫽ 140!.
Early investigations of foster care em- Fifty percent of the sample scored at or above
ployed a wide range of methodologies but com- the borderline clinical range, and over 40%
monly found that poor social functioning and had one subscale score in the clinical range.
emotional difficulties were the result of mul- Although preplacement adaptation was not as-
tiple placements during longer term stays in sessed, behavior problems and symptom lev-
care ~Theis, 1924; Weinstein, 1960!. The neg- els may have been related to experience prior
ative impact of child behavior problems on to entry into foster care.
foster parent–child relationships was thought Taking into account the timing of preplace-
to contribute to multiple placement changes ment assessment, Armsden, Pecora, Payne, and
~Maas & Engler, 1959!. Szatkiewicz ~2000! documented the psycho-
The first large-scale studies of the mental logical functioning of 362 children ages 4–18
health issues of foster children in the United years just prior to placement in the foster care
States reported frequent diagnoses of anxiety system. At intake, an adult deemed most fa-
and0or depression among those in foster care miliar with the child ~i.e., a recruited foster
~Shah, 1974; Swire & Kavaler, 1977!. Fanshel parent related to the child, caseworker, or rel-
and colleagues ~Fanshel, Finch, & Grundy, ative other than the participant’s parent! served
1989; Fanshel & Shin, 1978! conducted the as informant. For this sample, 30% of CBCL
first longitudinal investigation of behavior Total scores fell within the clinical range; how-
problems in foster care children. In this sam- ever, a desire to influence the placement deci-
ple ~N ⫽ 585! behavior problems were present sion and differing levels of familiarity with
in 46% of children discharged from foster care the child may have contributed to response
at 1 year, and among 54% remaining in foster biases. Assessments of children “at intake”
care for 5 years or longer. into foster care may not capture preplacement
In response to the varying prevalence rates characteristics, instead measuring transient be-
cited in past studies and general methodolog- haviors related to anticipation of entry into
ical problems related to the use of nonstan- foster care.
dardized instruments, investigators have sought Milan and Pinderhughes ~2000! examined
to assess foster children with norm-referenced the incidence of internalizing and externaliz-
measures of behavior problems completed by ing behavior of 32 children ~ages 9–13 years!
a parent or parental figure. Rates of behavior reported by foster mothers 1 month after entry
problems and clinically significant symptoms into care. In this study, children who per-
measured by the Child Behavior Checklist ceived their new relationships as being more
~CBCL; Achenbach & Edelbrock, 1986! are affectively positive tended to be viewed by
reported as up to 2.5 times higher for children foster mothers as showing more relational and
in foster care than for those of samples thought less internalizing behavior. Foster mothers who
to share demographic characteristics but not reported more symptomatic externalizing be-
entering protective custody ~Hulsey and White, havior reported less relational behavior in chil-
1989; McIntyre and Keesler, 1986; Zima et al., dren in their care. Maltreatment severity was
60 C. R. Lawrence, E. A. Carlson, and B. Egeland

associated with decreased relational behavior foster care experience. The study examines
and increased internalizing symptoms, but not the following:
to externalizing behavior. Milan and Pinder-
hughes suggest that such early placement 1. the impact of length of foster care place-
symptom reports may be influenced by a va- ment as longer term placements that have
riety of factors including habituated symptom been related to placement instability, num-
patterns, developing foster parent–child rela- ber of placements to determine the relation
tionship quality, as well as informant bias. between stability of placement and devel-
In summary, studies of children in foster opment of behavior problems, and age of
care suggest that this population is at signifi- first placement to determine a possible re-
cantly heightened risk for behavior problems. lation between age of placement and
The severity and frequency of behavior prob- the subsequent development of behavior
lems far exceed the norm for children reared problems;
at home with similarly adverse backgrounds. 2. the direct impact of foster care on behavior
Moreover, children with significant behavior problems, controlling for baseline adap-
problems and clinical diagnoses are likely to tation and SES including a comparative
remain in foster care for longer periods and evaluation of the behavior problems of
are at significant risk for multiple placements maltreated children reared at home and chil-
due to the level of care required to adequately dren who received adequate parental care;
treat them ~Fanshel & Shin, 1978; Simms & 3. change in pre- and postplacement adap-
Halfon, 1994!. Foster care studies also high- tation through repeated measure design
light a number of methodological consider- among the participants who experienced
ations that restrict the interpretation of research foster care and an examination of change
findings and our understanding of the impact in behavior problems over time among the
of the foster care system on development. maltreated participants and those who re-
These include the limited use of preplacement ceived adequate care;
adaptation assessment, the range of infor- 4. difference in outcome for children placed
mants ~e.g., foster parent, relatives, social in child protective service prescribed fos-
workers! with varying degrees of familiarity ter care versus care with an adult familiar
and interest in the child ~Halfon, Mendonca, to the child; and
& Berkowitz, 1995!, and the lack of differen- 5. long-term consequences of foster care
tiation between kinship and unrelated care- on behavior problems, overall emotional
giver placements. health, and psychopathology in adolescence.

Method
Current Research Design
Participants
The present study employs prospective longi-
tudinal data to address these research limita- The participants in this investigation included
tions. The study examines the relation between 189 children and families from the Minnesota
foster care placement and the development of Longitudinal Study of Parents and Children, a
behavior problems controlling for child adap- prospective study of families at risk because
tation prior to placement ~baseline!. The high- of poverty and associated factors such as low
risk sample includes participants who entered educational status, young age of mothers at
the foster care system, those who remained the birth of the first child, and chronically
with and were reared by caregivers who unstable home environments ~Egeland &
showed a continuous propensity for maltreat- Brunquell, 1979!. From the total sample of
ing their children, and at-risk participants from 189 children, three subgroups were identified:
the same sample who received adequate care 46 children who entered the foster care sys-
from parental figures in terms of parenting tem; 46 children who were maltreated but re-
style with no history of maltreatment or mained at home with the maltreating caregiver;
Foster care and development 61

Table 1. Foster, maltreated, and control group demographic characteristics

Foster Maltreated Control

Total sample n ⫽ 46 n ⫽ 46 n ⫽ 97
Gender ~% male! 56 65 47
Race ~% Caucasian! 37 54 53
SES 44.24a,b 47.53a,c 52.73b.c
Mother’s age ~mean years at child’s birth! 19.0 20.5 21.4
Mother’s education ~mean grade at child’s birth! 10.76 11.53 12.30
Elementary school subgroup n ⫽ 15 n ⫽ 15 n ⫽ 15
Gender ~% male! 60 60 60
Race ~% Caucasian! 30 56 40
SES 44.34d 46.88 49.79d
Mother’s age 18.73 19.60 19.00
Mother’s education 11.00 11.40 11.61
Familiar Unfamiliar
Foster care group n ⫽ 23 n ⫽ 23
Gender ~% male! 53 52
Race ~% Caucasian! 60 52
SES 43.47 45.02
Mother’s age 18.69 18.95
Mother’s education 10.81 10.68

Note: Means with the same subscript differ significantly at p , .05.

and 97 children who did not experience foster responsive caregiving; 3%, n ⫽ 1!, and mal-
placement or maltreatment ~see Table 1!. treatment history of two or more categories
~72%, n ⫽ 23, one of whom also experienced
Foster care participants. The 46 foster care sexual abuse!. Maltreatment severity ratings
children were identified by reviewing inter- were not available for developmental periods
view data, life events inventories, and case examined in this study. However, the foster
summaries obtained during infancy, toddler- care and the maltreated groups did not differ
hood, preschool, kindergarten, and Grades 1, on measures of adaptation prior to placement
2, 3, and 6. Foster care group assignment re- suggesting the possibility that maltreatment
quired a minimum of 4 consecutive weeks of experienced was equivalent for both groups.
out of home placement. Overall, age of first Of the 46 children in out of home care, 23
placement ranged from birth to 9 years ~M ⫽ were placed in unfamiliar foster care with
4 years, 5 months!; length of placement ranged adults unrelated and unknown to the child prior
from 1 to 45 months ~M ⫽ 13 months!; and to placement ~mean length of care ⫽16 months,
number of placements ranged from 1 to 10 range ⫽ 1– 45 months!. Children were also
~M ⫽ 3!. For the group placed into foster care placed with familiar caregivers including: ma-
during early elementary school ~postkinder- ternal and paternal grandparents, aunts and
garten!, mean length of placement was 25.76 uncles, maternal significant others, and friends
months ~range ⫽ 17–38 months!. of the family ~familiar foster care: n ⫽ 23,
Entry into foster care was precipitated by mean length of care ⫽ 9 months, range ⫽1–32
maltreatment, death of a parent, parental in- months!. Although our evidence of Child Pro-
carceration, parental chemical addiction, or tective Services ~CPS! intervention in these
homelessness. Maltreatment accounted for 69% cases was inconclusive, it appeared that ap-
~n ⫽ 32! of foster care placement in this sam- proximately 40% of these placements were
ple. Maltreatment history included neglect initiated by CPS involvement. All 46 children
~19%, n ⫽ 6!, physical abuse ~6%, n ⫽ 2!, who entered foster care were intermittently or
psychological unavailability ~emotionally un- permanently reunited with biological care-
62 C. R. Lawrence, E. A. Carlson, and B. Egeland

givers during the course of this study. The garten, first, second, third, and sixth grades,
overall foster care group included three sub- n ⫽ 15!. Maltreatment time periods corre-
groups: children first placed in foster care in sponded approximately to the developmental
the infancy0toddler period ~0–24 months; n ⫽ periods associated with foster care placement.
14!, in the preschool period ~25– 64 months; Maltreatment history included physical abuse
n ⫽17!, and during the elementary years ~kin- ~37%, n ⫽ 17!, neglect ~13%, n ⫽ 6!, sexual
dergarten, Grades 1, 2, 3, or 6; n ⫽ 15!. abuse ~7%, n ⫽ 3!, and psychologically un-
available ~emotionally unresponsive caregiv-
Maltreatment participants. For the purpose of ing 13%, n ⫽ 6!. Thirty percent ~n ⫽ 14!
the current study, a subsample of 46 children experienced two or more types of maltreat-
who had experienced maltreatment between ment ~cf. Egeland & Sroufe, 1981!.
birth and sixth grade, but had not experienced
foster care, was identified. These children were Control participants. Within the high risk sam-
reared within their families of origin. Maltreat- ple, 97 control participants were identified.
ment was identified on the basis of laboratory According to longitudinal study records ~i.e.,
and home observations and home interviews interviews, observations!, these children had
including the Child Care Rating Scale ~Ege- no history of maltreatment or foster care place-
land & Deinard, 1975! and questions regard- ment during the designated time periods.
ing caretaking skills, feelings toward the child,
and disciplinary practices.
Measures
Behaviors considered to be physically abu-
sive ranged from frequent and intense spank- Preplacement (baseline) measures. Baseline
ing to unprovoked angry outbursts resulting in measures were selected to represent child func-
serious injuries ~e.g., severe cigarette burns!. tioning and developmental adaptation during
In all instances, the abuse was seen as poten- infancy, toddlerhood, the preschool years, and
tially physically damaging to the child. Moth- kindergarten. For the foster care participants,
ers identified as hostile0verbally abusive these measures represented preplacement ad-
chronically found fault with their children and aptation prior to entry into care. Assessments
engaged in constant berating and harassment. included ~a! attachment quality ~12–18
Caregivers considered to be psychologically months!, ~b! toddler–caregiver experience rat-
unavailable were emotionally unresponsive to ing ~problem-solving task, 24 months!, ~c! per-
their children and, in many cases, passively sistence and ego control ratings ~teach and
rejecting of them. These mothers appeared de- barrier box tasks, respectively, 42 months!,
tached and uninvolved ~i.e., withdrawn, dis- and ~d! emotional health rank ~teacher rank-
playing flat affect, depressed!, interacting with ing, kindergarten!. Baseline composite scores
their children only when necessary. Caregiv- were derived from raw scores were converted
ing neglect referred to the lack of responsible to standard scores ~z scores! and averaged.
or competent management of day to day child For example, in infancy, standardized attach-
care activities ~e.g., health or physical care!. ment ratings were used as the baseline mea-
Despite an expressed interest in their children’s sure. For children placed or maltreated from
well-being, these caregivers seemed to lack 25 to 41 months of age, baseline scores con-
the skill, knowledge, or understanding to pro- sisted of standardized and averaged attach-
vide consistent, adequate care. The validity of ment and tool task ratings. For the control
group placement was supported by Child Pro- group the composite consisted of all available
tection and Public Health records ~Egeland, baseline assessment scores standardized and
1991b; Egeland & Sroufe, 1981; Egeland, averaged.
Sroufe, & Erickson, 1983!.
Maltreatment status was evaluated at three Attachment assessment (12–18 months). At-
time points: infancy ~birth to 24 months, n ⫽ tachment was assessed at 12 and 18 months in
14!, the preschool years ~25– 64 months, n ⫽ the Strange Situation, a standardized labora-
17!, and the elementary school years ~kinder- tory procedure designed to assess infant pat-
Foster care and development 63

terns of attachment and exploration in relation the context of a parent–child teaching task. A
to the primary caregiver ~Ainsworth et al., series of four tasks including: block building,
1978!. Critical observations include infant ex- naming objects, matching colors and shapes,
ploration of the laboratory room, response to and tracing were presented, and mothers were
the mother’s departure, response to the entry instructed to offer any level of guidance or
of a stranger, and reunion behavior with the assistance deemed necessary. Child persis-
mother. Infant–mother dyads were classified tence ~i.e., a critical quality in adverse circum-
into four major categories: anxious0avoidant stances! was rated using a 7-point Likert-style
~A!, secure ~B!, anxious0resistant ~C!, and dis- scale. At the high end of the scale ~7!, chil-
organized ~D!. The validity and reliability of dren demonstrated competent, goal-oriented
attachment classifications have been well doc- problem-solving behavior regardless of the
umented ~Ainsworth et al., 1978; Weinfield, mother’s emotional support. Low ratings ~1!
Sroufe, Egeland, & Carlson, 1999!. Interrater were assigned for active avoidance of the task
agreements for ABC classifications for this as a result of attentional or behavioral difficul-
sample were 89 and 93% ~12 and 18 months, ties, and0or interactional difficulties with the
respectively!. Interrater agreement for D clas- mother. Assessments were rated by two inde-
sification was 86% based on 35 cases selected pendent observers. The Persistence rating has
at random across 12 and 18 months ~k ⫽ .72; been related positively to child response to
see Carlson, 1998!. The baseline adaptation challenge and negatively to distractibility and
score used in the current study represented the behavioral difficulties that interfere with so-
number of assessments infants rated as secure cial and academic functioning ~Pianta, Erick-
~0, 1, 2!. son, Wagner, Kreutzer, & Egeland, 1990!.
Interrater reliability ~intraclass correlation! was
Problem-solving assessment (24 months). .88 ~n ⫽ 87; Egeland et al., 1983!.
Child adaptation during the toddler period was
assessed in a laboratory problem-solving situ- Barrier box (42 months). Child behavior in-
ation consisting of four tasks designed to chal- dependent of caregiver presence and assis-
lenge toddler skill and emerging autonomy tance in the early childhood period was
~Matas, Arend, & Sroufe, 1978!. Mothers were examined in response to a barrier box chal-
present and instructed to offer assistance only lenge task ~Harrington, Block, & Block, 1978!.
when necessary during the tasks. Based on Children were presented with a latched, Plexi-
videotaped observations, toddler experience glas box containing attractive toys following
with the caregiver in the session was rated on the potentially frustrating removal of the iden-
a 5-point scale. At the high end of the scale tical, desirable toys from the assessment room.
~5!, children were judged to have had positive Ego control, the ability to control impulses or
experiences ~i.e., child responded to task in affect-laden emotion in the face of frustration,
confident positive manner, drawing upon care- was rated on a 7-point Likert-type scale based
giver resources when necessary!. At the low on videotaped observations. At the high end
end of the scale ~1!, children were judged to of the scale ~7!, there was no evidence of
have had markedly poor experiences, result- overt frustration nor was there evidence that
ing from behavioral difficulties, failure to com- children were attempting to suppress frustra-
plete the tasks, and0or significant lack of tion. A rating of 1, at the low end of the scale,
support or conflict with the mother. This vari- suggested that children became frustrated, an-
able has been validated as a measure of child gry, or panicked in response their emotion.
adaptation within the context of the relation- The ego control rating was selected to repre-
ship at this developmental period ~Erickson, sent young children’s capacities to regulate
Sroufe, & Egeland, 1985!. Interrater reliabil- emotion and impulses in challenging situa-
ity ~intraclass correlation! was .87 ~N ⫽ 185!. tions. This variable has been validated as an
index of coping in response to frustration, and
Teaching task assessment (42 months). At age it discriminates maltreatment and control group
42 months, child functioning was assessed in functioning ~Egeland et al., 1983; Pianta,
64 C. R. Lawrence, E. A. Carlson, and B. Egeland

Sroufe & Egeland, 1989!. Interrater reliability Duncan Socioeconomic Index in the prenatal
~intraclass correlation! was .86 ~n ⫽ 60!. period, at 16, 42, and 54 months, and in Grades
1, 2, 3, and 6. The SES at time of placement
Emotional health ranking (kindergarten). Par- for foster care participants was used to cap-
ticipant teachers were asked to rank order stu- ture the most current SES for the family prior
dents in accordance with a written description to placement. For the comparison group mal-
of an “emotionally healthy child” without treatment time period was used as the basis
awareness of the research target child. The for the socioeconomic index score. Grade 1
emotional health ranking assessed child con- SES was used for all control participants.
fidence, curiosity, self-assuredness, enjoy-
ment, and involvement. Scores were computed
Outcome measures
as ratios of the child’s rank order standing
~Connolly & Doyle, 1981!. This measure has Specific outcome measures were selected to
been related in expected directions to concur- represent foster care child functioning imme-
rent ratings of behavior problems and peer diately following ~or close to! release from
competence ~Heister, Carlson & Sroufe, 1993!. care. For maltreated participants, timing of
Reliability data were not available for the as- outcome measures was selected based on time
sessment ~due to individual teacher rankings!; of release of the corresponding foster care
however, rankings were found to be stable group. Sixth grade measures were used as the
across elementary years, and interrater relia- outcome assessment for the control group.
bility coefficients ranged from .63 to .81 on
similar child rank orders completed by multi- CBCL-TRF (Grades 1, 2, 3, 6, 10). The CBCL-
ple project counselors. TRF was completed by participant teachers at
the end of the school year ~Achenbach & Edel-
CBCL—Teacher’s Report Form (CBCL-TRF; brock, 1986; see description above!. Assess-
kindergarten). The CBCL-TRF was com- ments at Grades 1, 2, 3, 6, and 10 served as
pleted by participants’ teachers at the end of outcome measures.
the school year ~Achenbach & Edelbrock,
1986!. The kindergarten assessment served as Kiddie Schedule for Affective Disorders and
baseline measure for a subsample of partici- Schizophrenia Rating (17.5 years; KSADS).
pants. The TRF consists of 113 items rated by In adolescence, participants were adminis-
the teacher on a 3-point scale reflecting pres- tered the KSADS-III-R ~Ambrosini, Metz,
ence, frequency, and severity of the problem Prabucki, & Lee, 1989; Puig-Antich & Cham-
~2 ⫽ often0very true, 1 ⫽ sometimes or some- bers, 1978!, a semistructured interview de-
what true, 0 ⫽ not true!. TRF items were signed to evaluate psychopathology. The
constructed to evaluate a broad range of be- interview yields diagnostic information as well
havior problems and symptoms associated with as present and past symptoms of psychopathol-
psychopathology. Externalizing ~i.e., aggres- ogy. The test–retest reliability within a 3-day
sive and delinquent behavior, a ⫽ .82!, Inter- period varied by diagnosis, but it remained in
nalizing ~i.e., somatic complaints, withdrawn, the moderate range ~Chambers et al., 1985!.
and anxious0depressed behavior, a ⫽ .63! and Various forms of the measure have been ex-
Total ~a ⫽ .79! T scores were used in the tensively validated through reassessment of
current study. The TRF was norm referenced clinical patients with known diagnoses and by
on a sizeable representative national sample. examining treatment and biological correlates
The clinical range is defined as 1.5 SD above of specific diagnoses ~Costello, 1991!. Inter-
the mean. For the current sample, T scores rater reliability ~mean k ⫽ .79 for child-
above 60 were regarded as within the clinical derived diagnoses! of the KSADS-III-R was
range. reported by Ambrosini and colleagues ~1989!
based on present symptom information. For
SES (time of placement). The SES of partici- the current study, a 7-point Likert-style rating
pant families of origin was assessed using the scale was employed to quantify the number
Foster care and development 65

and severity of present and past diagnoses. behavior problems! in comparison to the con-
Participants with five or more distinct severe trol group. The foster care and maltreated
diagnoses were rated 7. At the low end of the groups did not differ significantly with re-
scale, a rating of 2 signified one minor diag- spect to baseline developmental adaptation.
nosis such as simple phobia, and a rating of 1 A series of partial correlations was con-
indicated no past or present diagnosis. Inter- ducted to determine whether the age of first
rater reliability ~intraclass correlation! for the placement into foster care, length of time in
current sample was .94 ~n ⫽ 35!. placement, and number of placements were
related to behavior problems ~TRF! at release
from foster care controlling for baseline adap-
Data analysis
tation and SES at time of placement. Length
Five sets of analyses were conducted: pre- and stability of care, and age of entry into care
liminary correlations, analysis of covariance were not significantly correlated with the TRF
~ANCOVA! to examine group differences in indicator of behavior problems ~r ⫽ .13, .16,
behavior problems with baseline adaptation and .09; ns, respectively!. Participants whose
and SES as covariates, repeated measures AN- care was arranged by county social services
COVA to examine changes in behavior prob- did not differ from those whose care was ar-
lems over time, analysis of variance ~ANOVA! ranged without documented evidence of county
examining the effects of unfamiliar and famil- services involvement in placement.
iar care, and ANCOVA to examine the long-
term effects of foster care on emotional health
Principal analyses
and psychopathology diagnoses.
To examine the direct impact of foster care on
the development of behavior problems, a se-
Results
ries of one-way ANCOVAs was conducted.
Analyses were designed to test whether the
Preliminary analyses
three groups ~foster care, maltreated, and con-
Preliminary analyses examined the correla- trol! differed with respect to TRF score at
tions between the baseline adaptation score release from care controlling for baseline ad-
~adaptation prior to placement for foster care aptation and SES at time of placement. The
group!, SES, and outcome measures ~TRF at analyses investigated whether foster care di-
release from care and Grade 6!. Because both rectly impacted the development of behavior
baseline adaptation scores ~r ⫽ ⫺.25, p , .01; problems above the risks associated with base-
r ⫽ ⫺.19, p , .05! and SES measures ~r ⫽ line adaptation, SES, and child maltreatment.
⫺.17, p , .05; r ⫽ ⫺.20, p , .01! were sig- ANCOVA results indicated significant overall
nificantly correlated with outcome measures, group mean differences in TRF total and ex-
baseline adaptation scores and SES at time of ternalizing scores at release from foster care
placement served as covariates in subsequent controlling for baseline adaptation and SES
analyses. Length of time in care, age of place- ~see Table 2!. Post hoc contrasts revealed sig-
ment into care, and child protective service nificant differences between the foster care
involvement were not correlated with out- and the control groups on both TRF total and
come measures. An ANOVA revealed statisti- externalizing scores ~t ⫽ 2.36, p , .05; t ⫽
cally significant group differences in baseline 3.16; p , .01, respectively!. Post hoc tests
adaptation score, F ~2, 181! ⫽ 5.75, p , .01. also indicated significant differences in TRF
Foster care and maltreatment groups each dif- Externalizing ~but not Total! scores between
fered significantly from the control group in the maltreated and control groups ~t ⫽ 2.60;
post hoc analyses ~t ⫽ ⫺2.45, p , .01; t ⫽ p , .01!. Foster care and maltreated group
⫺2.22, p , .05, respectively!. The findings TRF total and externalizing scores did not dif-
suggest that prior to placement children enter- fer significantly. The results of the Internaliz-
ing foster care and a comparable maltreated ing Scale analyses were not significant ~see
group exhibited poor adaptation ~i.e., more Table 2!.
66 C. R. Lawrence, E. A. Carlson, and B. Egeland

Table 2. Adjusted means, standard deviations, and analysis of covariance results for TRF
total, externalizing, and internalizing at release from foster care as a function
of baseline adaptation and SES

Foster Care Maltreated Control

~n ⫽ 39! ~n ⫽ 42! ~n ⫽ 95!
ANCOVA
Variable M SD M SD M SD F ~2, 173!

Total score TRF 58.58a 10.79 56.75 10.39 53.91a 9.09 3.07*
Externalizing TRF 59.40b 10.26 57.95c 9.52 53.74bc 9.49 5.06**
Internalizing TRF 54.64 10.76 52.92 10.69 54.41 8.91 .46

Note: Means with the same subscript differ significantly.

*p , .05. **p , .01.

To further examine group differences, a sub- For 13 of the 15 children, behavior prob-
sample of participants placed into foster care lems were also assessed while in care. After
during the early elementary school years controlling for baseline adaptation and SES,
~Grades 1–3! was identified ~n ⫽ 15!. For this “in care” TRF total scores differed signifi-
subset, TRF total scores served as both base- cantly, F ~2, 37! ⫽ 3.39, p , .05. Post hoc
line ~kindergarten! and postplacement mea- analyses indicated that the foster care group
sures ~immediately following release from differed significantly from both the maltreated
foster care!, permitting pre- and postplace- and control groups ~t ⫽ 2.10, p , .05; t ⫽
ment comparisons on the same measure of 2.91; p , .05, respectively!. Descriptive sta-
behavior problems. Corresponding maltreated tistics were as follows: foster care: M ⫽ 63.84,
and control groups were constructed for the SD ⫽ 11.96; maltreated: M ⫽ 54.92, SD ⫽
analyses. The 3 ⫻ 2 ~Group ⫻ Change! 11.62; control: M ⫽ 52.76, SD ⫽ 7.45.
repeated-measure ANOVA examining change The effects of unfamiliar versus familiar
in TRF total scores was significant ~see foster care on the development of behavior
Table 3!. The Group ⫻ Change interaction problems were examined controlling for base-
was also significant. Post hoc comparison of line adaptation and SES at the time of place-
change scores indicated a significantly greater ment. ANCOVA results indicated significant
rise in scores within the foster care group than overall group mean differences on TRF Total,
within the maltreated group ~t ⫽ 1.74, p , Externalizing, and Internalizing scores at re-
.05!. Repeated-measure ANOVA examining lease from care ~see Table 4!. Post hoc analy-
change in TRF externalizing and internalizing ses indicated that unfamiliar and familiar foster
scores were also significant ~see Table 3!. care groups differed with respect to Internal-
Group ⫻ Change interactions were signifi- izing ~t ⫽ 3.11, p , .001!, but not Externaliz-
cant as well. Post hoc analyses indicated that ing or Total score. Children in unfamiliar foster
the rise in the foster care externalizing and care exhibited higher internalizing, but not ex-
internalizing scores differed significantly from ternalizing or total TRF scores following re-
the change in the maltreatment group scores lease from care. Unfamiliar foster care and
~t ⫽ 1.93, p , .05; t ⫽ 3.62, p , .001, respec- control groups differed significantly with re-
tively!. The foster care group rise in external- spect to Total ~t ⫽ 2.75, p , .01! and Exter-
izing scores also significantly exceeded that nalizing ~t ⫽ 2.92, p , .001! scores.
of the control group ~t ⫽ 1.96, p , .05!. The To examine the long-term effects of foster
findings suggest that the externalizing and in- care on behavior, a 3 ⫻ 2 repeated-measure
ternalizing behavior problems of children in ANCOVA ~Group ⫻ Assessment Period! was
foster care increased significantly between conducted on the entire sample with baseline
baseline assessment and subsequent measure- adaptation and SES at time of placement con-
ment immediately following release from care. trolled ~see Table 5!. Changes in TRF scores
Foster care and development 67

Table 3. Means, standard deviations, and repeated measure analysis of variance

for change in TRF scores among participants placed into foster care during
the early elementary school years

Foster Care Maltreated Control

~n ⫽ 15! ~n ⫽ 15! ~n ⫽ 15!
ANCOVA
Variable M SD M SD M SD F ~2, 173!

TRF total score

Preplacement 52.80 11.34 55.86 8.22 43.33 9.19
Postplacement 61.00 7.69 57.66 10.46 47.66 7.94
Change 8.20a 9.69 1.80a 10.40 4.33 9.37
Between group 9.34**
Group ⫻ Change 2.29*
TRF externalizing
Preplacement 52.73 10.08 56.53 10.35 45.66 9.29
Postplacement 61.60 8.55 58.06 9.47 48.26 9.83
Change 8.87bc 10.78 1.53b 10.13 2.60c 8.99
Between group 8.26**
Group ⫻ Change 2.64*
TRF internalizing
Preplacement 47.80 10.34 54.53 8.77 44.00 7.66
Postplacement 58.13 7.76 54.33 8.64 49.40 10.60
Change 10.33d 6.73 ⫺.20d 9.03 5.40 8.44
Between group 6.30**
Group ⫻ Change 3.34*

Note: Means with the same subscript differ significantly.

*p , .05. **p , .01.

Table 4. Adjusted means, standard deviations, and analysis of covariance results for TRF
total, externalizing, and internalizing scores at release from care of unfamiliar and familiar
foster care, abuse, and control as a function of baseline adaptation and SES

Unfamiliar Familiar Maltreated Control

~n ⫽ 19! ~n ⫽ 20! ~n ⫽ 42! ~n ⫽ 95!
ANCOVA
Variable M SD M SD M SD M SD F ~2, 173!

Total score TRF 61.29a 12.31 56.04 8.77 56.75 10.39 53.91a 9.09 3.04*
Externalizing TRF 60.42b 11.03 58.37 10.21 57.95 9.52 53.74b 9.49 3.31*
Internalizing TRF 59.47c 12.38 50.12c 9.64 52.92 10.69 54.41 8.91 3.87*

Note: Means with the same subscript differ significantly.

*p , .05.

at release from foster care, sixth grade, and Additional analyses of covariance exam-
during high school at age 16 were analyzed ined the long-term effects of foster care on an
for within subject and between group dif- overall Index of psychopathology at age 17.5
ferences and for interactions. There was no years ~see Table 6!. Controlling for baseline
significant main effect across time period; how- adaptation and SES, the ANCOVA examining
ever, between group differences were statisti- psychopathology diagnoses was significant
cally significant for Total and Externalizing with the foster care and maltreated groups dif-
scores. Repeated Measure ⫻ Group inter- fering from the control group ~t ⫽ 1.96, p ⫽
actions were not significant. .05; t ⫽ 2.95, p , .01!.
68 C. R. Lawrence, E. A. Carlson, and B. Egeland

Table 5. Adjusted means, standard deviations, and results of repeated measure analysis of
covariance for TRF total, externalizing, and internalizing scores at release from foster care,
Grade 6, and age 16 as a function of baseline adaptation and SES

Foster Care Maltreated Control

~n ⫽ 30! ~n ⫽ 39! ~n ⫽ 88!
ANCOVA
Variable M SD M SD M SD F ~2, 154!

TRF total score

Release from care 58.53a 11.11 57.43 10.77 53.47a 8.71
Grade 6 58.33b 9.40 58.25 8.61 53.47b 8.71
Age 16 59.03c 9.72 56.30 9.20 54.44c 8.05
Between group 3.28*
Repeated Measure ⫻ Change 0.95
TRF externalizing
Release from care 58.86d 10.83 58.38 9.82 53.64d 9.16
Grade 6 57.83e 9.49 58.56 9.99 53.64e 9.16
Age 16 58.13f 10.44 56.89 9.64 54.85f 8.20
Between group 3.91*
Repeated Measure ⫻ Change 0.88
TRF internalizing
Release from care 54.33 10.56 53.41 11.09 53.57 8.07
Grade 6 55.76 10.49 57.28 8.49 53.57 8.07
Age 16 56.53 9.87 54.61 8.36 52.84 7.76
Between group 0.18
Repeated Measure ⫻ Change 1.64

Note: Means with the same subscript differ significantly.

*p , .05.

Table 6. Adjusted means, standard deviations, and analysis of covariance results for
psychopathology diagnostic ratings (age 17.5 years) as a function of baseline
adaptation and SES

Foster Care Maltreated Control

~n ⫽ 32! ~n ⫽ 37! ~n ⫽ 88! ANCOVA

Variable M SD M SD M SD F df

Psychopathology 3.53a 1.98 3.83b 1.90 2.80ab 1.73 3.41* 2, 154

Note: Means with the same subscript differ significantly.

*p , .05.

Trajectories of foster care, maltreated, and Discussion

control group adaptation ~z scores! from pre-
placement, or baseline, to release from care This study examined the impact of foster care
and at ages 16 and 17 are depicted in Figure 1. placement on the development of behavior
For subsets of the three groups, trajectories of problems within the context of a prospective
TRF Total scores from baseline to age 16 are multimethod, multiple informant longitudinal
depicted in Figure 2. Assessment periods in- design. The consequences of foster care place-
clude baseline, in care, release of from, age 11 ment were evaluated immediately following
and age 16. release from care and at several points later in
69

Figure 1. Adaptation z scores for preplacement, release from care, 16- and 17-year assessments for foster care ~n ⫽ 46!, maltreatment ~n ⫽ 46!, and control ~n ⫽
97! groups.
70

Figure 2. Teacher Report Form ~TRF! total scores from preplacement to 16 years of foster care, maltreatment, and control groups ~n ⫽ 150group!.
Foster care and development 71

development. Controlling for developmental elevated levels during their time in care. In-
adaptation and SES prior to placement, the care behavior problems significantly exceeded
results support a general view that foster care those of the maltreated and control compari-
may lead to an increase in behavior problems son groups assessed during the same develop-
that continues after exiting the system. An at- mental periods. These findings are especially
tempt to determine the extent to which foster compelling considering that outcomes were
care influences development is an important obtained from measures that eliminate biases
process as foster care necessitates a signifi- in previous studies relying on foster parents or
cant disruption in the caretaking environment social workers as informants.
of children who have likely experienced ad- An additional subgroup analysis examined
verse circumstances prior to placement. the extent to which unfamiliar and familiar
Initial analyses suggest that for the current foster care experiences influenced the devel-
sample, length of time in care, age of first opment of subsequent behavior. Unfamiliar
placement, and the risks associated with mul- foster care arranged by CPS was provided by
tiple placements were found to be unrelated to an unfamiliar and unrelated caregiver. Famil-
the development of behavior problems. County iar care was arranged with a known caregiver
protective service involvement also did not ~often a relative or family friend!. Based on
relate to behavior problems. With baseline ad- available project records, a minimum of 40%
aptation and SES at time of placement re- of familiar care cases were precipitated by
moved to control for preexisting influences, social service intervention. Although children
findings suggest that children who experi- in unfamiliar foster care may have been ex-
enced foster care displayed higher levels of pected to show higher levels of maladaptation
behavior problems immediately following prior to entry into care, assuming that they
release from care compared to children who had come to the attention of social services
received adequate parental care within disad- due to markedly adverse family contexts, un-
vantaged home environments. The behavior familiar and familiar foster children did not
problems of children who had left foster care differ with respect to baseline adjustment in
continued to be elevated during adolescence this sample. Type of foster care experience,
above levels exhibited by children who had however, did differentiate behavior problem
experienced adequate parental care. Analyses status immediately following release from
of the entire sample did not differentiate the foster care. Internalizing behavior problems
risks associated with foster care placement among children exiting unfamiliar foster care
from those remaining with the family of ori- significantly exceeded those of children exit-
gin and a maltreating caregiver. ing familiar care, also exceeding those of the
Analysis of children placed into care after maltreated home-reared group and adequately
kindergarten permitted the examination of pre- cared for children.
and postplacement change in behavior prob- Analyses to examine the extent to which
lems assessed with the same measure ~TRF!. foster care experience impacts psychopathol-
In these analyses, the foster and maltreated ogy diagnoses during adolescence ~with base-
groups did not differ prior to placement. How- line adaptation and SES controlled! also
ever, immediately following placement, chil- revealed significant differences between both
dren in foster care exhibited an increase in the foster care and maltreated groups and the
behavior problems. The increase in problem- control group ~with no significant distinction
atic behavior following departure from foster between foster care and maltreatment psycho-
care significantly exceeded change in behav- pathology indices!.
ior problems among those reared by maltreat- Despite the expectations that length of time
ing parental figures, suggesting an exacerbation in foster care and placement instability would
of problem behavior in the context of out of be associated with the development of behav-
home care. The assessment of behavior prob- ior problems, no statistically significant rela-
lems of a small sample of participants ~n ⫽ tions were found in the present study. Length
13! while in care also suggested significantly of time in care, age of first placement, the
72 C. R. Lawrence, E. A. Carlson, and B. Egeland

risks associated with multiple placements and Unfamiliar versus familiar foster care out-
the involvement of county protective services come differences ~e.g., higher level internal-
were found to be unrelated to the develop- izing problems! further suggest that facets of
ment of behavior problems. These findings the foster care experience may be challenging
deviate from past research suggesting that mul- to children. Although it is difficult to general-
tiple placements in longer term care pose a ize from this relatively small sample, it is pos-
risk for poor outcomes ~McDonald, Allen, sible that familiar care with relatives or familiar
Westerfelt, & Piliavin, 1996!. It is possible figures presents an environment that necessi-
that the small sample size of the current study tates fewer changes in caregiving routine, so-
did not generate significantly meaningful re- cial milieu, and school environment. It may
lations between these qualitative aspects of be that the availability and suitability of rela-
the foster care experience and subsequent be- tives to provide familiar foster care influence
havior. Length of time in care and the influ- the type of care a child receives to a greater
ence of multiple placements are important extent than mitigating factors that precipitated
qualitative measures of the foster care experi- placement. It should also be acknowledged
ence. Within a larger sample, these salient fac- that familiar care arrangements may permit
tors may prove to be more powerful predictors ongoing contact between biological mother and
of behavior problems and related outcomes. child changing the dynamic of the separation
In the current study, the heightened behav- experience. This appears especially relevant
ior problem levels manifested in children in care in the context of the current study because the
and immediately following release from foster unfamiliar and the familiar care recipients did
care raise cautious concern regarding the im- not differ on a measure of baseline develop-
pact of foster care on development. Although mental adaptation. The availability of suitable
based on small samples, this study controlled caregivers within a family network may serve
for baseline developmental adaptation ~prior to as a protective factor in that children have
placement! and the risks associated with low familiar adult resources to draw upon for sup-
SES, ruling out the possibility that differences port possibly reducing experiences of anxiety,
between foster care and control participants depression, and withdrawal.
were due to differences in initial adaptation. Long-term effects of foster care were also
Several factors may account for the in- investigated in the current study. One interpre-
crease in problematic behavior associated with tation of these findings is that foster care itself
out of home care. First, foster care as an inter- may have a direct impact on the long-term
vention may expose its recipients to difficult increase in behavior problems. However, be-
developmental challenges. For example, sep- cause maltreated and foster care children did
aration from primary caregivers in the context not differ with respect to behavior problems,
of placement with unfamiliar adults may pose it is difficult to interpret whether the foster
an especially difficult challenge for very young care experience itself or aspects of the inter-
children. Second, although not a reflection on vening years ~e.g., intermittent or permanent
specific foster caregivers, the relations may reunion with caregivers! influenced this find-
reflect weaknesses within the foster care sys- ing. It is unclear whether further exposure to
tem: the school, social, and familial changes risk factors within the home environment or
that foster care placement often entails; the the lingering impact of foster care placement
lack of comprehensive psychological services influenced long-term outcomes.
offered to foster children; as well as often
inadequate training and support services for
Methodological Considerations
foster parents. It is also possible that the am-
biguity of the placement experience with no As previously noted, the small sample size of
delineated endpoint or stated outcome may the current study is a clear limitation, as it is
contribute to children’s emotional difficulties difficult to generalize findings to larger popu-
in processing the experience as well as care- lations and to find statistical associations
giver commitment to the children. among variables that may in fact represent
Foster care and development 73

important associations within the general pop- could also be used as outcome measures to
ulation. The small sample size, differing lengths directly examine change following the foster
of placement, and varying ages of entry into care experience. It is also important for future
care also made it unfeasible to assess enough research to focus on the effects of unfamiliar
children during their stays in foster care to versus familiar foster care as well as the rela-
render meaningful conclusions. Similarly, the tive impact of the age or developmental status
differing ages of entry into care and lengths of of the children when placed in out of home
time spent in foster care limit speculation re- care. This research could examine the extent
garding points in development and placement to which children in differing types of care
lengths that could potentially benefit those in have ongoing contact with their biological care-
care. Differing ages of entry into care in the givers and the impact of this contact on chil-
current study also restricted the range of ap- dren in care. Additional factors that may
propriate and available baseline adaptation contribute to developmental risk for children
measures that could be selected for the chil- in foster care require investigation. These in-
dren placed earliest in development. Despite clude ~a! the child’s experience of separation
these limitations, the current study illustrated and ambiguity of placement ~lack of clarity
some methodological improvements over pre- with respect to parental reunification or
vious investigations. Most previous studies custody termination!; ~b! the child’s expecta-
have not included measures of baseline devel- tions, attitudes, and feelings regarding rela-
opmental adaptation to evaluate the effects of tionships derived from relationship history;
the foster care experience. Similarly, the op- ~c! the instability, quality and type of care
portunity to evaluate differences among groups provided; and ~d! the relationship history and
of children, including those who had experi- experience of the caregiver.
enced foster care, maltreated nonfoster chil- Ongoing research with foster care provid-
dren, and children who received adequate ers may shed light on some of these issues and
caregiving within their families of origin, was provide direction for policy and intervention
a clear advantage of the present study. A final ~Dozier, 2002; Dozier, Stovall, & Albus, 1998!.
benefit of this study was the use of teachers as Studies based on attachment theory and re-
informants, eliminating differing degrees of search suggest that routinely sensitive care may
familiarity and any related unfamiliar and be insufficient for children with histories of
familiar foster parent biases regarding the grossly inadequate care. Severe histories of
children. maltreatment and loss are associated with a
range of maladaptive behaviors as well as dis-
tortions in representations of self and others
Future Research and Policy Implications
in relationships ~Cicchetti & Toth, 2000; Lynch
Evaluation of the strengths and limitations of & Cicchetti, 1991; McCrone et al., 1994!. Fos-
the current study suggests a variety of improve- ter caregivers may require training to recog-
ments for future investigations. There is need nize and respond therapeutically to the signals
for a large-scale systematic examination of and special needs of foster children.
children at risk for entering foster care ~within Policy implications of developmental re-
a county or specific geographic region! with search on the impact of foster care services
assessments conducted in advance of place- are numerous. Research investigating the ef-
ment with ongoing, frequent follow-up. The fects of timing of foster care entry may inform
current study illustrates the importance of ex- policy regarding types of services, transi-
amining adaptation as well as risk status prior tional allowances, and care offered to meet
to placement to more clearly define the im- the developmental needs of children entering
pact of the foster care experience. A large- foster placement at different points from in-
scale investigation would be enhanced by the fancy to adolescence. Research may guide fos-
development of a consistent battery of base- ter parent training and inform policy regarding
line measures to be administered to all poten- long-term placement planning ~e.g., develop-
tial participants. Several of these measures mental timing, duration!. Targeted research
74 C. R. Lawrence, E. A. Carlson, and B. Egeland

may also indicate whether therapeutic foster dren in care is regarded as notably variable
care settings with specific treatment goals and at times nonexistent in systems that are
would serve the complex purposes of protecting overwhelmed by large populations of foster
children and ameliorating past circumstances. children. The intervention services necessary
Policymakers must continue to address the to assist children throughout their stay in
issues of family reunification service and sup- foster care must continue to be defined and
port for children receiving care and work to- expanded. Collaborative efforts, including spe-
ward increased funding for foster care services. cialized medical, psychological, and aca-
It is recognized anecdotally that attempts to demic services in coordination with ongoing
assist and treat parents while their children are school and agency programs, are required to
in foster care are often ineffectively executed, significantly reduce the development of be-
increasing the likelihood of recidivism and havior problems and related symptoms while
future placements. Similarly, support for chil- children reside in foster care and beyond.

References
Achenbach, T., & Edelbrock, C. ~1986!. The Child Behav- Costello, A. ~1991!. Structured interviewing. In M. Lewis
ior Checklist–Teacher’s Report Form. Burlington, VT: ~Ed.!, Child and adolescent psychiatry: A comprehen-
Department of Psychiatry, University of Vermont. sive textbook ~pp. 463– 472!. Baltimore, MD: Wil-
Ainsworth, M., Blehar, M., Waters, E., & Wall, S. ~1978!. liams & Wilkins.
Patterns of attachment. Hillsdale, NJ: Erlbaum. Cummings, E. M., & Cicchetti, D. ~1990!. Toward a trans-
Ambrosini, P. J., Metz, C., Prabucki, K., & Lee, J. ~1989!. actional model of relations between attachment and
Videotape reliability of the third revised edition of the depression. In M. T. Greenberg, D. Cicchetti, & E. M.
K-SADS. Journal of the American Academy of Child Cummings ~Eds.!, Attachment in the preschool years
and Adolescent Psychiatry, 28, 723–728. ~pp. 339–374!. Chicago: University of Chicago.
Arad, B. D. ~2001!. Prenatal features and quality of life in Curtis, P. A. ~1999!. In P. A. Curtis, G. Dale, Jr., J. C.
the decision to remove children at risk from home. Kendall, & J. D. Rockefeller, IV ~Eds.!, The foster
Child Abuse and Neglect, 25, 47– 64. care crisis: Translating research into policy and prac-
Armsden, G., Pecora, P. J., Payne, V. H., & Szatkiewicz, tice ~pp. 1–17!. Lincoln, NE: University of Nebraska
J. P. ~2000!. Children placed in long-term foster care: Press.
An intake profile using the Child Behavior Checklist Curtis, P. A., Dale, G., Jr., Kendall, J. C., & Rockefeller,
4–18. Journal of Emotional and Behavioral Disor- J. D. IV. ~1999!. The foster care crisis: Translating
ders, 8, 49– 64. research into policy and practice. Lincoln, NE: Uni-
Bowlby, J. ~1960!. Grief and mourning in infancy and versity of Nebraska Press.
early childhood. The Psychoanalytic Study of the Child, Dozier, M. ~2002, May!. Intervening with young chil-
15, 9–52. dren: Targeting three critical needs. Paper presented
Bowlby, J. ~196901982!. Attachment and loss: Vol. 1. At- at the Irving B. Harris Training Center Forum, St.
tachment. New York: Basic Books. Paul, MN.
Carlson, E. A. ~1998!. A prospective longitudinal study of Dozier, M., Stovall, K. C., & Albus, K. E. ~1998!. A
attachment dissorganization0disorientation. Child De- transactional intervention for foster infants’ caregiv-
velopment, 69, 1107–1128. ers. In D. Cicchetti & S. L. Toth ~Eds.!, Rochester
Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. Symposium on Developmental Psychopathology: Vol.
~1989!. Disorganized0disoriented attachment relation- 9. Developmental approaches to prevention and inter-
ships in maltreated infants. Developmental Psychol- vention ~pp. 195–219!. Rochester, NY: University of
ogy, 25, 525–531. Rochester Press.
Chambers, W., Puig-Antich, J., Hirsch, M., Paez, P., Dozier, M., Stovall, K. C., Albus, K. E., & Bates, B.
Ambrosini, P., Tabrizi, M. A., et al. ~1985!. The as- ~2001!. Attachment for infants in foster care: The role
sessment of prepubertal major depression with the of caregiver state of mind. Child Development, 72,
Kiddie-SADS-E. Journal of the American Academy 1467–1477.
of Child Psychiatry, 21, 695–707. Egeland, B. ~1991a!. A longitudinal study of high risk
Cicchetti, D., & Toth, S. ~2000!. Developmental pro- families: Issues and findings. In R. Starr & D. Wolfe
cesses in maltreated children. In D. J. Hansen ~Ed.!, ~Eds.!, The effects of child abuse and neglect: Issues
Nebraska Symposium on Motivation: Vol. 46. Motiva- and research ~pp. 33–56!. New York: Guilford Press.
tion and child maltreatment ~pp. 85–160!. Lincoln, Egeland, B. ~1991b!. From data to definition. Develop-
NE: University of Nebraska Press. ment and Psychopathology, 3, 37– 43.
Clausen, J. M., Landsverk, J., Ganger, W., Chadwick, D., Egeland, B. ~1997!. Mediators of the effects of child
& Litrownik, A. ~1998!. Mental health problems of maltreatment on developmental adaptation in adoles-
children in foster care. Journal of Child and Family cence. In D. Cicchetti & S. L. Toth ~Eds.!, Rochester
Studies, 7, 283–296. Symposium on Developmental Psychopathology: Vol.
Connolly, J., & Doyle, A. ~1981!. Assessment of social 8. The effects of trauma on the developmental process
competence in preschoolers: Teachers versus peers. ~pp. 403– 434!. Rochester, NY: University Press of
Developmental Psychology, 17, 454– 462. Rochester.
Foster care and development 75

Egeland, B., & Brunquell, D. ~1979!. An at-risk approach Landsverk, J., & Garland, A. F. ~1999!. Foster care and
to the study of child abuse: Some preliminary find- pathways to mental health services. In P. A. Curtis, G.
ings. Journal of the American Academy of Child Psy- Dale, Jr., J. C. Kendall, & J. D. Rockefeller, IV ~Eds.!,
chiatry, 8, 219–235. The foster care crisis: Translating research into
Egeland, B., Carlson, E., & Sroufe, L. A. ~1993!. Resil- policy and practice. Lincoln, NE: University of Ne-
ience as process. Development and Psychopathology, braska Press.
5, 517–528. Lynch, M., & Cicchetti, C. ~1991!. Patterns of relatedness
Egeland, B., & Carlson, E. A. ~2004!. Attachment and in maltreated and non maltreated children: Connec-
psychopathology. In L. Atkinson & S. Goldberg ~Eds.!, tions among multiple representational models. Devel-
Clinical Implications of Attachment ~pp. 27– 48!. Mah- opment and Psychopathology, 3, 207–226.
wah, NJ: Erlbaum. Maas, H., & Engler, R. ~1959!. Children in need of par-
Egeland, B., & Deinard, A. ~1975!. The Child Care ents. New York: Columbia University Press.
Rating Scale. Unpublished manuscript, University of Masten, A. S., & Garmezy, N. ~1985!. Risk, vulnerability
Minnesota, Minneapolis. and protective factors in developmental psychopathol-
Egeland, B., & Sroufe, L. A. ~1981!. Attachment and ogy. In B. Lahey & A. Kazdin ~Eds.!, Advances in
early maltreatment. Child Development, 52, 44–52. clinical child psychology ~Vol. 8, pp.1–52!. New York:
Egeland, B., Sroufe, L. A., & Erickson, M. F. ~1983!. Plenum Press.
Developmental consequence of different patterns of Masten, A. S., & Wright, M. ~1998!. Cumulative risk and
maltreatment. Child Abuse and Neglect, 7, 459– 469. protection models of child maltreatment. In B. B. R.
Erickson, M., Sroufe, L. A., & Egeland, B. ~1985!. The Rossman & M. S. Rosenberg ~Eds.!, Multiple vic-
relationship between quality of attachment and be- timization of children: Conceptual, developmental,
havior problems in preschool in a high risk sample. research and treatment issues. Binghamton, NY:
Monographs of the Society for Research in Child De- Haworth Press.
velopment, 50~Serial No. 209!, 147–186. Matas, L., Arend, R., & Sroufe, L. A. ~1978!. Continuity
Fanshel, D., Finch, S., & Grundy, J. ~1989!. Modes of exit of adaptation in the second year: The relationship
from foster family care and adjustment at time of between quality of attachment and later competent
departure of children with unstable life histories. Child functioning. Child Development, 49, 547–556.
Welfare, 68, 391– 402. McCrone, E. R., Egeland, B., Kalkoske, M., & Carlson,
Fanshel, D., & Shin, E. ~1978!. Child behavior character- E. A. ~1994!. Relations between early maltreatment
istics of foster children. In D. Fanshel & E. Shin and mental representations of relationships assessed
~Eds.!. Children in foster care: A longitudinal inves- with projective storytelling in middle childhood. De-
tigation ~pp. 325–372!. New York: Columbia Univer- velopment and Psychopathology, 6, 99–120.
sity Press. McDonald, T. P., Allen, R. I., Westerfelt, A., & Piliavin, I.
Garmezy, N. ~1993!. Children in poverty: Resilience de- ~1996!. Assessing the long-term effects of foster care.
spite risk. Psychiatry, 56, 127–152. A research synthesis. Washington, DC: CWLA Press.
Garmezy, N., & Masten, A. ~1994!. Chronic adversities. McIntyre, A., & Keesler, T. ~1986!. Psychological disor-
In M. Rutter, L. Hersov, & E. Taylor, ~Eds.!. Child ders among foster children. Journal of Clinical Child
and adolescent psychiatry ~3rd ed., pp. 191–208!. Ox- Psychology, 15, 297–303.
ford: Blackwell. Milan, S. E., & Pinderhughes, E. E. ~2000!. Factors influ-
Halfon, N., Mendonca, A., & Berkowitz, G. ~1995!. Health encing maltreated children’s early adjustment in foster
status of children in foster care: The experience of the care. Development and Psychopathology, 12, 63–81.
center for the vulnerable child. Archives of Pediatric Pianta, R. C., Erickson, M. F., Wagner, N., Kreutzer, T., &
and Adolescent Medicine, 149, 386–392. Egeland, B. ~1990!. Early predictors of referral for
Harrington, D. M., Block, J. H., & Block, J. ~1978!. Intol- special services: Child based measures versus mother–
erance of ambiguity in preschool children: Psycho- child interaction. School Psychology Review, 19,
metric considerations, behavioral manifestations, and 240–250.
parental correlates. Developmental Psychology, 14, Pianta, R. C., Sroufe, L. A., & Egeland, B. ~1989!. Con-
242–245. tinuity and discontinuity in maternal sensitivity at 6,
Heister, M., Carlson, E. A., & Sroufe, L. A. ~1993!. The 24 and 42 months in a high risk sample. Child Devel-
evolution of friendship in preschool, middle child- opment, 60, 481– 487.
hood, and adolescence: Origins in attachment his- Puig-Antich, J., & Chambers, W. ~1978!. The schedule for
tory. Paper presented at the biennial meeting of the affective disorders and schizophrenia for school-aged
Society for Research in Child Development, New Or- children. New York: New York Psychiatric Institute.
leans, LA. Rutter, M. ~1987!. Psychosocial resilience and protective
Hochstadt, N., Jaudes, P., Zimo, D., & Schachter, J. ~1987!. mechanisms. American Journal of Orthopsychiatry,
The medical and psychological needs of children en- 57, 316–331.
tering foster care. Child Abuse and Neglect, 2, 53– 62. Rutter, M. ~2000!. Children in substitute care: Some con-
Hulsey, T., & White, R. ~1989!. Family characteristics ceptual considerations and research implications. Chil-
and measures of behavior in foster and non-foster dren and Youth Services Review, 22, 685–703.
children. American Journal of Orthopsychiatry, 59, Shah, C. P. ~1974!. Psychiatric consultations in a child
502–509. welfare agency. Canadian Psychiatric Association
Jones, M. D. ~1985!. A second chance for families: Five Journal, 19, 393.
years later follow-up of a program to prevent foster Simms, M., & Halfon, N. ~1994!. School performance of
care. New York: Child Welfare League of America. children in kinship care. Child Abuse and Neglect, 18,
Landsverk, J. ~1996, April!. The mental health needs of 587–597.
children in foster care: A case of benign neglect? Sroufe, L. A. ~1996!. Emotional development: The orga-
Paper present at the University of Minnesota Foster nization of emotional life in the early years. New
Care Symposium. York: Cambridge University Press.
 76 C. R. Lawrence, E. A. Carlson, and B. Egeland

Swire, M., & Kavaler, F. ~1977!. The health status of ~Eds.!, Handbook of attachment: Theory, research and
foster children. Child Welfare, 58, 635– 653. clinical applications ~pp. 73–95!. New York: Guil-
Theis, S. ~1924!. How foster children turn out. New York: ford Press.
State Charities Aid Association. Weinstein, E. ~1960!. The role of placement experiences
US Department of Health and Human Services, Admin- in mediating the criminal consequences of early child-
istration for Children and Families, Administration on hood victimization. American Journal of Orthopsy-
Children, Youth and Families, Children’s Bureau. chiatry, 61, 195–209.
~2003!. The AFCARS report. Washington, DC: Gov- Zima, B. T., Bussing, R., Freeman, S., Yang, X., Belin,
ernment Printing Office. T. R., & Forness, S. R. ~2000!. Behavior problems,
United States General Accounting Office. ~1995!. Child academic skill delays and school failure among school-
welfare: Complex needs strain to capacity. GAO0 aged children in foster care: Their relationship to place-
HEHS-95-208. Washington, DC: Government Print- ment characteristics. Journal of Child and Family
ing Office. Studies, 9, 87–103.
Weinfield, N. S., Sroufe, L. A., Egeland, B., & Carlson, Zuravin, S. J., & DePanfilis, D. ~1997!. Factors affecting
E. A. ~1999!. The nature of individual differences in foster care placement of children receiving protective
infant–caregiver attachment. In J. Cassidy & P. Shaver services. Social Work Research, 21, 34– 42.

View publication stats