Full text of DOI:10.1093/swr/24.2.119

NOTE ON RESEARCH
can be analyzed by subgroup (for example, age
and ethnicity), placement type, placement reason,
placement region, and various other factors
METHODOLOGY
(
Needell et al., 1998; Wulczyn et al., 1994). Sur-
veys and focus groups with social workers, inter-
views with children’s care providers, and case
record extraction also have been used as method-
ological tools to help explain the phenomenon of
foster care. Yet relatively little research has in-
cluded children as research participants. In fact,
some of the seminal works in child welfare re-
search have excluded children from participating
directly in the research enterprise (Fein, Maluccio,
& Kluger, 1990), unless their voices were cap-
tured through interviews conducted primarily for
clinical purposes (Fanshel, & Shinn, 1978). This
limitation in foster care research has been wide-
spread, despite the fact that children’s perspectives
on their experiences in care could inform the ser-
vice delivery system.
Despite the importance of including children’s
voices in child welfare research, their relative ab-
sence from the literature is not surprising. Admin-
istrative, political, legal, and pragmatic barriers all
conspire to limit researchers’ access to and contact
with foster children.
This article discusses some of the methodologi-
cal issues raised in one study conducted in Califor-
nia. The study sample included 100 children ages
six to 13 residing in kinship or nonkinship care for
a minimum of six months. The study used face-to-
face interviews with the children in the homes of
their caregivers and was built on previous work by
the investigator (Berrick, Needell, Shlonsky, &
Simmel, 1998), also involving interviews with the
children’s kin and nonkin foster parents.
Children’s interviews lasted approximately one
hour. The interviews were designed to assess
children’s experiences of care in four fundamental
domains: their experience of safety, their under-
standing of and contact with family, their experi-
ence of permanency, and their experience of
caregiver support for their overall well-being.
On the basis of our experience, we review in
this article three of the most challenging issues
that may be faced in conducting research with
children in foster care: (1) recruitment of the
study sample, (2) development of the study instru-
ment, and (3) selection and training of interviewers.
We highlight potentially problematic method-
ological concerns and suggest a variety of strate-
gies to minimize their effect. It should be recog-
nized, however, that the political or administrative
Assessing children’s experiences of
out-of-home care: Methodological
challenges and opportunities
Jill Duerr Berrick, Karie Frasch, and
Adair Fox
he U.S. foster care system has undergone pro-
found changes during the past decade. Case-
T
load growth, increases in the number of very
young children entering care, and especially prob-
lematic behaviors among some children character-
ize the shifting foster care population (Barth,
Courtney, Berrick, & Albert, 1994; U.S. House of
Representatives, 1998; Wulczyn, Harden, &
Goerge, 1998). Changes among out-of-home-care
clients have been accompanied by a rapid transfor-
mation in the services delivery system designed for
children. Kinship care has absorbed much of the
growth in foster care (General Accounting Office,
1
999; Hegar & Scannapieco, 1999). Specialized
or treatment foster care has found increasing favor
in some states (Needell, Webster, CuccaroAlamin,
&
Armijo, 1998), and new paradigms of service
delivery that include alterations in public finance
for foster care, privatization, and managed care
have been developed (Peter & Johnson, 1999,
Wulczyn, Zeidman, & Svirsky, 1997). When sys-
tems of care undergo fundamental changes such as
these, it is important to understand outcomes for
the clients the systems are designed to serve. Al-
though child welfare researchers are making sig-
nificant contributions toward developing an un-
derstanding of foster care outcomes, the primary
clients of this system—children—have been given
few opportunities to contribute to the literature.
Researchers now have increased opportunities
to understand the case characteristics and case
outcomes of children in the foster care system.
Administrative data systems in several large states
allow extensive analysis of factors such as caseload
dynamics, entries and exits from care, reunifica-
tion, and adoption. Each of these outcomes now
Assessing children’s experiencCeCsC oCofdoe:u1t0-7o0f--5h3o0m9/e00ca$r3e.:0M0 e©th2o0d0o0l,oNgaictiaonlaclhAasslolceiantgioensofaSnodciaol pWpoorkretrus,nInitci.es / Berrick, Frasch, and Fox
119

context of other settings might pose different
challenges for other researchers.
the targeted characteristics. Our staff provided the
necessary contact materials to child welfare staff
who then conducted a mailing through their of-
fices, using agency envelopes and agency mailing
labels. This strategy allowed letters to be mailed
without the researchers’ knowledge of potential
respondents’ names or addresses. The caregivers
who consented to participate contacted the re-
searchers and, in doing so, revealed their identi-
ties. Once we had the sample, we returned to the
caregivers a year later with a solicitation letter re-
questing their child’s participation. The letter ex-
plicitly stated that the juvenile court judge in the
county had approved of the study and had pro-
vided blanket consent for children to participate.
Special challenges associated with this method
include the fact that some child welfare agencies
do not have databases that list their active foster
parents; some databases do not distinguish among
kin, nonkin, or specialized foster caregivers; and
many of these databases are not linked to the chil-
dren in care. In such instances it is difficult to de-
termine whether caregivers have any children re-
siding in their homes—let alone, children in the
age range of interest.
As the method for contacting potential sample
respondents is being developed, researchers need
to determine whether they want respondents to
perceive the research as completely separate from
the child welfare agency or whether the collabora-
tive relationship developed between the researcher
and the agency should be highlighted. Some
caregivers may see the research as more valid if it is
viewed as emanating from the child welfare
agency; others may look with skepticism on re-
search associated with their child’s social worker.
Determine Who Is the “Parent.” Foster
RECRUITMENT
Although parents normally have authority over
their children’s participation in research, the legal
status of children changes considerably when they
receive out-of-home care. Under these circum-
stances, biological parents, foster parents, social
workers, social services agency administrators, law-
yers, and judges all may exert influence on
children’s activities. Access to children may be
tightly controlled, foster parents may be suspicious
of researcher motives, and incentives may be re-
quired for busy caregivers as much as for the chil-
dren in their care.
Gaining Access to Children
Determine Sample Recruitment Methods
through Collaboration with the Social Services
Agency. Unlike research with children in schools,
hospitals, recreation centers, or other settings, ac-
cess to children in foster care is hampered by their
dispersion in homes throughout a county or across
counties or states. Contact must be made indi-
vidually, and many barriers exist to keep research-
ers and other interested parties from contacting
children in care. Public child welfare officials
closely guard children’s names and other identify-
ing characteristics. It is with great difficulty that
researchers can gain the confidence of agency ad-
ministrators and the consent from legal officials to
release this information for sampling purposes. In
addition to children’s identities being protected,
foster parents’ names and addresses also are tightly
controlled. Unless they have given the child wel-
fare agency advance permission, public officials are
wary of providing researchers with caregivers’
names and contact information.
One approach to gaining access to a foster child
sample is to work through a county’s foster parent
population to select and recruit study participants.
With assistance from a community advisory board,
we determined that the best approach to gaining
access to children was through our established re-
lationship with children’s caregivers. In the pre-
ceding year, we had conducted a series of inhome
interviews with kin and nonkin caregivers in our
identified counties. (For a review of that study’s
methods, findings, and instrumentation, see
Berrick et al., 1998). To contact our original
sample, the local child welfare agency examined its
database of caregivers and developed a list with
children’s biological parents may or may not be
notified about their child’s participation in a re-
search study, and they may or may not be asked
for consent. Because the legal authority for chil-
dren in foster care is transferred to the court, par-
ents are no longer obliged legally—or in some
cases, allowed—to consent to their children’s par-
ticipation. Instead, researchers must gain informed
consent from social services agencies and judges.
In some cases they also may need to gain informed
consent from children’s lawyers, biological par-
ents, and caregivers.
Obtaining Consent
The social services agency must be an active
partner in developing the research study and
1
20
Social Work Research / Volume 24, Number 2 / June 2000

therefore, must give consent for children’s partici-
pation. Agency staff must be ready to describe the
study if social workers, foster parents, or other
concerned parties raise questions about the study’s
purpose. Although one or more agency adminis-
trators should be involved in developing the pro-
cedures for the study and all social workers should
be notified that the study is underway, it may not
be necessary or advisable to inform social workers
about the participation of individual children on
their caseload. Informing social workers would
breach the confidentiality of children and might
prompt social workers to question children in a
way that could be inappropriate and upsetting.
Beyond social workers and agency administra-
tors, judges must provide informed consent for
children’s participation. In our research we met
individually with the juvenile court judges; intro-
duced the purpose, methods, and potential impli-
cations of the study; and reviewed the interview
protocol in some detail. The judges approved of
the study for all children in the county who fit the
sample criteria and signed a consent form to grant
their participation. To highlight one of the most
daunting challenges in conducting research in this
area, one need only examine the role of judges in
gaining consent. Judges, acting in loco parentis,
may or may not consent to children’s participa-
tion. In one county our staff repeatedly attempted
to gain consent from the local judge. Despite our
efforts and those of the agency administrator, the
judge would not be swayed, and children in that
county were systematically excluded from the
study.
Beyond administrators and judges, some locali-
ties also may require consent from other promi-
nent individuals in foster children’s lives, such as
their lawyers. Lawyers, in turn, may require con-
sent from children’s biological parents. In this
study judges required caregivers’ consent, not be-
cause they had legal authority for the children in
their care, but simply because their homes were
used as the host setting for data collection.
Determine Whether Children Are Old Enough
to Consent. Children generally do not give consent
toward their participation in research. Neverthe-
less, children should be given as much information
as possible about the research before the interview
begins. The purpose for the study can be de-
scribed in child-friendly terms, and their assent
can be obtained verbally or in written form. Chil-
dren should be empowered to understand their
right not to participate at all, to terminate the in-
terview at any time, and to “pass” on any ques-
tions they wish.
Expect and Manage Caregiver Suspicion. Fos-
ter children’s personal and family histories may
have left painful emotional scars. Caregivers who
are responsible for protecting these vulnerable
children therefore are appropriately suspicious of
strangers who wish to query children about foster
care. On the one hand, it is rather remarkable that
any caregivers consent to their child’s participa-
tion in these face-to-face interviews, whereas on
the other, their suspicions can lead to consider-
able selection bias in the final sample. When re-
searchers assume that suspicion is appropriate and
healthy, their interactions with caregivers are
likely to be more fruitful and may lead to less de-
fensive responses and, in some cases, very positive
responses. Caregivers are a critical link in gaining
access to children, and their role in either sup-
porting or undermining the research cannot be
underestimated.
Use Appropriate Incentives. As a courtesy to
study participants and as acknowledgment for
their time, modest honoraria should be provided.
In the case of foster care, it also may be appropri-
ate to offer incentives to caregivers for arranging
children’s interviews. In our study it became clear
that foster parents often had complicated schedules
to negotiate to accommodate in-home interviews.
As an incentive as well as recognition for their ef-
forts, we offered children a gift certificate to Toys
‘R Us, along with a “Certificate of Achievement.”
We also offered caregivers cash payments.
MEASUREMENT
Once a sampling strategy has been devised, in-
strument development begins. Conducting re-
search with children requires designing an instru-
ment package that is developmentally appropriate
and previously validated in the field. Instrumenta-
tion for studies with children considered vulner-
able, such as those in out-of-home care, create
additional methodological challenges. Issues in
instrument development can be grouped into
three main areas: (1) selecting scales and designing
questions, (2) creating the instrument package,
and (3) including special considerations for vulner-
able populations. (See Fox, Frasch, & Berrick,
2000, for a copy of the final instrument.)
Selecting Scales and Designing Questions
Review Existing Literature Critically. Very
few researchers have examined out-of-home care
Assessing children’s experiences of out-of-home care: Methodological challenges and opportunities / Berrick, Frasch, and Fox
121

from children’s perspectives, and of those who
have, few used standardized measures (Johnson,
Yoken, & Voss, 1994; Wilson, 1996). We re-
viewed the literature on our major domains of in-
terest to identify previously developed measures,
including ones that could be used in their original
form or adapted to accommodate the unique
needs of our population. Alternatively, we could
formulate entirely new scales and items. Trade-offs
clearly exist. Using a measure with a population
for which it is not intended undermines the
measure’s validity. And changing the wording of a
measure may mean a loss of psychometric rigor, as
well as the ability to compare results found in
other studies. Inclusion of largely untested scales
and items is sometimes necessary, but not always
meaningful in the long run.
they are in stable, permanent living situations. We
adapted other questions from a published study by
Johnson et al. (1994) and a child survey developed
by Westat, Inc. (1998). Some of the questions in-
clude: “Do you feel like you’re part of this family?”
“Do you think you’ll be living with [caregiver]
next year?” “Can you keep living here until you
grow up?” To understand children’s experiences
more thoroughly, we also included open-ended
questions in each section of the interview. This
qualitative approach significantly strengthened our
overall understanding of the topic and provided an
opportunity for children to express their concrete
and emotional experiences in their own words. A
number of themes that emerged from children’s
responses to these openended questions would
have been impossible to elicit using traditional,
quantitative survey methods. One of the questions
was “If you met a person about your age who
could not live at home with [his or her] birth par-
ents, you might be able to give him [her] some
advice. What advice would you give?”
Select Available Measures Where Possible.
Some of the questions we used to assess existing
measures included Is the measure developmentally
appropriate for the age range represented in the
study? Is each question short, and does it build on
children’s concrete approach to reasoning tasks? Is
it an appropriate length? Does it engage children?
Is it sensitive to the lives of children in out-of-
home care, such as their experience living in an
alternative family structure? Does it make biased
assumptions about out-of-home care?
In the domain of children’s understanding of
and contact with family, we were able to use the
Relatedness Scale developed by Wellborn and
Connell (1987) to assess children’s relatedness to
their current caregiver and one of their biological
parents. Earlier studies have used the measure with
maltreated and nonmaltreated children to explore
self-reported patterns of relatedness (Lynch &
Cicchetti, 1991). To assess children’s experiences
of their family environment, we adapted the lan-
guage and tone of the Revised Social Climate
Scale, used by Colton (1989) with a sample of
older British children in foster and group homes.
For example, we changed a true/false statement
from “The staff/foster parents here allow you to
be cheeky,” to “[Caregiver] lets you be disrespect-
ful.”
Creating the Instrument Package
Assemble a Community-Based Advisory Com-
mittee. We included former foster youths, current
and former foster parents (kin and nonkin), social
workers, social work administrators, and research-
ers not working directly on the project in a work-
shop to examine the draft instrument item by
item. Our advisory committee offered critical in-
formation on neglected areas of research, sensitive
ordering of questions, and instrument language.
For example, our advisory committee developed a
protocol for using the terms “real” or “birth” in
reference to biological parents. Although the feed-
back of the committee informed our decision
making usefully, the feedback was not prescriptive,
and we attempted to weigh the feedback with
unique methodological considerations.
Pilot Test the Instrument with Children from
the Same Population. We pilot tested our instru-
ment with current foster children, although access
to the children was very difficult. We also tested
the instrument with children who were not in fos-
ter care. It was important to pilot test the instru-
ment with children of various ages to understand
whether the instrument’s distinct components
were developmentally appropriate. Pilot partici-
pants, particularly those who were older, provided
important feedback on these components.
Develop New Questions Where No Instrumen-
tation Exists. In child welfare little instrumenta-
tion exists for assessing certain aspects of
children’s experiences in care. Children’s experi-
ence of permanency, for example, is a critical do-
main of interest but has not been examined ad-
equately in the past. In this domain we developed
many questions to assess whether children believe
Create an Instrument that Engages Children’s
Attention. Young children, particularly those who
1
22
Social Work Research / Volume 24, Number 2 / June 2000

have experienced maltreatment, may have a diffi-
cult time maintaining focus during an interview.
We created an interview schedule that was in-
tended to last no more than one hour, including
the introduction and conclusion. The actual time
range was wide; some interviews lasted up to two
hours, and others were as short as 45 minutes.
Significantly, our population of interest in-
(1) the verbal introduction to the study clearly
identified the importance of the interview, and (2)
interviewers were trained to help keep children on
track and take breaks when necessary.
Including Special Considerations for
Vulnerable Populations
Order the Instrument Scales and Questions to
Match the Unique Needs of Children in Care
Thoughtfully. We developed an “ice breaker” sec-
tion in the beginning of the instrument to learn
the terms that each child used for his or her cur-
rent caregivers and biological parents. This strat-
egy allowed us to tailor the rest of the interview to
the child’s unique family situation. For example,
some children addressed their foster mother as
“mom” and their biological mother by her first
name, or vice versa. Similarly, in our sample of 100
children, 19 percent had a deceased biological par-
ent. For this sizable proportion of children, we
skipped all questions in the interview that inquired
about a current relationship with that biological
parent. In addition, some children had spent their
entire life with a relative, such as a grandmother,
and did not know that they lived in out-of-home
care. We assessed these situations and then skipped
questions such as “Has (caretaker) ever talked to
you about having this be your permanent home—
a home where you’ll live until you’re grown?”
Allow Children to “Pass” on Sensitive Ques-
tions. We frequently reminded children that if a
question raised uncomfortable feelings, they could
say “pass” or honk a horn that meant “pass.” Al-
though we were concerned that some children
might overuse the pass option, that was not our
experience. If a child seemed distracted or
cluded children in the concrete as well as the for-
mal operational stage of cognitive development
(
Piaget & Inhelder, 1969). For this reason our
questions needed to be sufficiently descriptive. We
used a standardized measure called Things I Have
Seen or Heard (Richters & Martinez, 1993) to
examine children’s exposure to community vio-
lence. After our review of existing instruments, we
concluded that the Richters and Martinez scale
was by far the most appropriate instrument for
young children. Nevertheless, some of the ques-
tions are potentially upsetting and include: “I have
heard guns being shot,” “Grownups in my home
yell at each other,” and “Somebody threatened to
kill me.” Because we anticipated that some young
children might have difficulty differentiating what
they have fantasized or seen on television and
what they have actually seen or heard, we were
careful to administer the scale in such a way that
children understood the questions posed to them.
Make the Interview Experience Enjoyable. We
created a sheet of paper with squares for children
to stamp after completing each colorcoded section
of the instrument. In this way, children were al-
ways aware of how much of the interview was left
to complete. To engage children’s attention and
make the interview experience enjoyable, inter-
viewers invited children to use props to accom-
pany or serve in lieu of verbal responses. For ex-
ample, for yes/no questions children could place
rubber fruit into a balance beam, ring a bell, or
shake a maraca. For Likert-scale questions children
could toss a Koosh ball onto a mat or use a stamp
to mark their response. Although a few older chil-
dren did not elect to use props, the majority of
children appreciated them. It is important to note
that for some children props provided a safer way
of responding to sensitive items by allowing them
to give nonverbal, rather than verbal responses.
Although the use of props was a positive addition
to the overall study instrument, the props also
could have contributed to the unintended conse-
quence of encouraging children to treat the inter-
view as a game rather than an important interview.
Two strategies helped to reduce this possibility:
unengaged, the interviewer was trained to ask the
child if he or she wanted to take a break. In this
way we prevented situations in which children
might say “pass” only to get through the inter-
view more quickly.
Selection of Interviewers
Children are notoriously difficult to interview
in a way that yields reliable data across respon-
dents (Bierman, 1983; Sattler, 1992; Yarrow,
1960). The literature, coupled with our awareness
that foster children often evoke particularly strong
feelings in clinicians, urged us to be thoughtful
about our selection and training of interviewers.
To minimize interviewer bias and build interview-
ers’ skills in conducting semistructured interviews
with this population, we implemented a rigorous
Assessing children’s experiences of out-of-home care: Methodological challenges and opportunities / Berrick, Frasch, and Fox
123

application process, required participation in two
weeks of training, and provided ongoing supervi-
sion throughout the data collection process.
Seek the Ideal Interviewer, but Be Prepared to
Set Priorities. Strong interviewers will possess a
variety of characteristics, all of which are very im-
portant to the study’s success. These qualities in-
clude the following: experience and previously de-
veloped skills working with children in the age
range of interest; a high level of maturity for com-
municating effectively with caregivers; experience
working with families whose lives can be disorga-
nized; interest in spending time in poor communi-
ties; a capacity to focus in noisy, distracting envi-
ronments; a racial or ethnic background that
matches one of the population’s primary racial
groups; proficiency in the population’s spoken lan-
guage; a flexible schedule; a car; familiarity with
the child welfare system; research experience; and
a generally engaging personality. Although all
qualities may be important, some compromises are
inevitable, and it is important to anticipate those
qualities most important to gaining caregivers’
trust and children’s honest responses.
Use Mock Interviews to Assess Potential Inter-
viewer Applicants. As a research team, we re-
viewed applicants’ letters of interest, resumés, ref-
erences, and responses to standard interview
questions. In addition, we asked each applicant to
conduct a mock interview with a staff member
playing the role of an eightyearold foster child.
This component of the interview process provided
the team with direct insights into how prospective
interviewers might build rapport with children and
engage their attention.
example, situations when time pressures required
compromising certain sections, or situations when
the child’s clear desire to “pass” on a sensitive
question required compromising the concurrent
need to maximize a forced choice response). Our
training provided the interviewers with ample
background information for making informed,
spontaneous judgments.
Develop a Plan for Achieving Interrater Reli-
ability in Assessments of the Home and Neighbor-
hood. The instrument included an
interviewerbased measure for assessing children’s
homes and neighborhoods. To enhance interrater
reliability we followed several important steps. Be-
fore study initiation the interviewers conducted
independent assessments of nine representative
sites. Reliability was poor on seemingly objective
dichotomous questions (for example, Does the
child live in a public housing project?), as well as
more subjective questions requiring Likert-scale
assessments (for example, On a scale of 1 to 10,
how much green space exists in the neighbor-
hood?). Given initial poor interrater reliability, the
team spent substantial time creating operational
definitions for certain words, as well as a detailed
rubric to accompany the instrument. Subsequent
neighborhood assessment rounds achieved high
interrater reliability.
Adopt Strategies to Maximize Interviewers’
Safety in the Field. Before study initiation we es-
tablished certain precautionary measures, given
our experience in the field during the previous
study (Berrick et al., 1998) and given that a large
percentage of foster children reside in lowincome,
high unemployment neighborhoods. First, for in-
terviews occurring in high-crime neighborhoods,
we automatically paired interviewers with a
“shadow,” an accompanying peer or supervisor
who also served as a scribe during the interview.
Second, no interviews were scheduled for the first
weekday of the month. Experience from our previ-
ous year’s interviews with caregivers suggested
that interviewers were uncomfortable with the
level of street activitiy in these low-income commu-
nities on those days. Third, most interviews were
scheduled on weekdays between the hours of 9:00
Training the Interviewers
Provide Interviewers with Ample Information
about the Research. To ensure that interviewers
felt entirely comfortable explaining the study’s
purpose, they read selected literature about the
child welfare system and met with supervisors to
discuss the study’s relevance. We then conducted
an item-by-item review of the instrument, provid-
ing theoretical and psychometric reasons underly-
ing inclusion of each question. We were particu-
larly concerned about interviewers’ comfort
regarding the delivery of sensitive questions. In
the event that caregivers raised concerns about
these questions, interviewers needed to provide
appropriate justifications. Finally, interviewers
needed a context for making educated decisions in
situations involving potentially missing data (for
A
.M
. and 4:00
P.M
. and on Fridays before 2:00
P
.M.
Finally, during the preparatory training period and
throughout data collection, interviewers were
coached on ways to be mindful of their safety.
Train Interviewers on Ways to Create and
Maintain Effective Rapport with Caregivers. In
our study, interviewers needed to be skilled at
1
24
Social Work Research / Volume 24, Number 2 / June 2000

building rapport with a diverse group of
space, while maintaining their sensitivity to
caregivers’ concerns. In particular, they practiced
talking openly about the importance of identifying
a space with minimal distractions and a degree of
privacy so children would feel free to discuss their
experiences without worrying about family mem-
bers’ reactions.
Support Interviewers to Work with Children at
Varying Developmental Stages and with Varying
Temperaments. Guided role-playing exercises fo-
cused on building interviewers’ skills in the fol-
lowing areas: making quick, accurate assessments
of children (particularly their overall cognitive
functioning and ability to sustain attention);
modifying response formats to match children’s
developmental needs; using specific strategies to
engage children’s attention; and responding sensi-
tively when children appear distressed. In addi-
tion, during training the interviewers made posters
with 30 positive phrases that they might use over
the course of their interviews. The phrases were
used as additional training tools to help remind
interviewers about the many ways to influence
children’s interview experience positively.
Provide Interviewers with Specific Guidance
on Ways to Anticipate and Manage Children’s
Disclosures about Unsafe Situations. Interviewers
need to know their state’s mandated reporting
laws, as well as the procedures developed by their
research institution. In addition, interviewers need
ongoing support in managing their personal feel-
ings about their intimate knowledge that some
participating children live in nonnurturing and
sometimes unsafe environments. During our train-
ing the research team carefully reviewed legal and
procedural information. The team also facilitated
multiple roleplaying exercises. Using a script, in-
terviewers practiced explaining confidentiality and
its limits. They also practiced responding to
nonreportable, ambiguous, and clearly reportable
disclosures. Regardless of a response’s report-
ability, interviewers were trained to encourage
children to talk to adults about situations that
made them feel unsafe. If a child indicated hopes
regarding the interviewer’s future availability for
support, the interviewer gave the child her work
number, but helped the child brainstorm alterna-
tive sources of support.
caregivers. Caregivers’ ages spanned over 50 years.
Their personalities and life circumstances were
equally varied. Some were employed full-time,
whereas others were full-time caregivers. Some
suffered from substance dependencies and mental
illnesses. Caregivers also had various motivations
for participating in the study—most prominently,
interest in helping to improve the delivery of child
welfare services and interest in providing a mean-
ingful opportunity for their child.
Although interviewers generally spent the ma-
jority of their total interview time with children,
they also needed to be skillful in their various
communications with caregivers: confirming ap-
pointments, handling cancellations and “no show”
interviews, presenting thoughtful overviews of the
study, building trust, solidifying consent, respond-
ing to interruptions, responding to reservations
and occasional hostility, negotiating longer time
periods if needed, and providing closure at the
end of interviews. Scripts were created for phone
confirmations, greetings, and goodbyes. Inter-
viewers also provided caregivers with study bro-
chures and personal business cards. Interviewer
training involved extensive role playing, with staff
members enacting caregivers with varying person-
alities. Interviewers were trained to communicate
all of the information contained in the scripts,
while nurturing natural, responsive conversations.
Provide Guidance on Ways to Talk to
Caregivers about the Need for Privacy. To ensure
children’s willingness to talk openly about their
current living situation and elicit valid responses,
the interview needs to occur in a space that the
child perceives as private. Ideally, this space is a
separate room. More typically, because children
lived in cramped quarters (for example, small
apartments or large apartments with multiple fam-
ily members), interviews occurred in semiprivate
spaces (for example, a room with an open door or
one end of the living room). In anticipation of
possible difficulties surrounding the need for pri-
vacy, the phone scheduler provided advance notice
to each caregiver and requested that he or she se-
lect a space before the interview. Even with this
advance notice, interviewers almost always needed
to go through a similar process in person.
Many caregivers communicated a healthy skep-
ticism about the interviewer’s need for privacy. A
few were overtly hostile. As part of their training,
interviewers engaged in a number of roleplaying
exercises to practice negotiating an appropriate
Among 100 completed interviews, one led to
a Child Protective Services report. During this
interview two boys living in the same home inde-
pendently reported having personally experienced
and witnessed children being “beat up” several
Assessing children’s experiences of out-of-home care: Methodological challenges and opportunities / Berrick, Frasch, and Fox
125

times in their current home. With probing from
the interviewers, the boys did not provide any ad-
ditional information. In each interview the inter-
viewer reminded the boys that she might need to
break confidentiality and talk to a social worker.
One boy inquired whether the interviewer would
speak with his caregiver. He was relieved to hear
that she would not but that he could if he wanted.
Later that day the interviewer wrote detailed ac-
counts of the interviews and called the principal
investigator. The principal investigator called a
social worker in the county who had been identi-
fied previously as a contact person for any report-
able incidents. The social worker then contacted
the boys’ social workers. In turn, these social
workers replied that they were already aware of
possible abuse in the home and would not follow
up specifically on the two interviews but would
include the report in their files. In retrospect, the
legal and procedural training, coupled with ongo-
ing opportunities to discuss personal feelings
about children’s disclosures, critically influenced
the research team’s ability to respond effectively to
this particular incident.
identifying a sample, gaining consent, developing
population-specific research questions, managing
highly emotional topics, and preparing staff for
the work can hijack the research enterprise at ev-
ery turn. Indeed, in this study, one judge’s stroke
of a pen compromised our final sample size by
one-third. Our experience, however, highlights
the importance of working collaboratively with
social services agencies and the courts to over-
come some of these obstacles. Researchers who
have close working relationships with local agen-
cies probably are more successful than others, be-
cause the agency–researcher–court relationship
must be built on previously developed trust. Our
experience also suggests that this research will
continue to be implemented only sporadically and
will depend on individual judges and the unique
circumstances of particular jurisdictions.
The tenacious researcher may manage the
methodological challenges listed. Nevertheless,
these issues also have the potential to compromise
a study in several respects. Problems with recruit-
ment, for example, may lead to considerable selec-
tion bias that cannot be controlled by the investi-
gator. The resulting nonprobability sample has
implications for analysis, and severely limits the
investigator’s capacity to generalize to the full fos-
ter care population. Response rates might be in-
creased if judges, social workers, or lawyers
strongly encouraged potential respondents to par-
ticipate, but problems with coercion—perceived
or actual—would make this approach ethically un-
acceptable. Understanding and controlling for bias
in the analysis can be accomplished only if the re-
searcher has access to information concerning the
foster youth population from which the sample
was drawn. Depending on data systems and
agency collaboration, this information may not be
readily available to the researcher.
Provide Ongoing Training and Support
throughout the Study. During the data collection
phase of the study, at least one supervisor was
available at all times for consultation. Interviewers
were encouraged to call supervisors at home if
needed. Supervisors intermittently “shadowed”
interviewers in the field and provided constructive
feedback. When an interviewer anticipated feeling
uncomfortable in a particular home or neighbor-
hood, she could request a shadow. Weekly team
meetings provided opportunities to review inter-
views, examine recorded responses, give feedback,
brainstorm strategies for responding to unex-
pected events, and generally bolster interviewers’
sense of belonging in a productive team.
Even with extensive preparation, interviews with
children often exert an emotional toll on field
staff. Although many foster children live in safe,
loving homes, interviewers reported considerable
concern about some children’s neighborhood and
home conditions. Providing ample time to discuss
reflections about their field experience and time to
complete nonfield-related tasks helped balance the
emotional demands of the work so interviewers
could maintain their enthusiasm with children.
Instrumentation issues also may jeopardize the
study’s validity, despite the best efforts to create
an engaging, nonthreatening atmosphere for the
child. When the fundamental concepts in child
welfare, such as permanency, have little or no lin-
eage in the research literature, child welfare re-
searchers also must strive to develop standardized
scales with sufficient psychometric properties.
However, the first generation studies may suffer in
the interim.
Despite the considerable challenges posed by
research with children in foster care, much more
research is needed. Children are the primary cli-
ents of the child welfare system, yet their voices
CONCLUSION
The challenges of conducting research with
children in foster care are formidable. Barriers to
1
26
Social Work Research / Volume 24, Number 2 / June 2000

services in California: 1997. Berkeley: University of
California, Berkeley, Child Welfare Research Center.
Peter, C. G., & Johnson, L. C. (1999). Privatization of foster
care in Kansas: A cautionary tale. Social Work, 44,
263–268.
Piaget, J., & Inhelder, B. (1969). The psychology of the child.
New York: Basic Books.
Richters, J. E., & Martinez, P. (1993). The NIMH community
violence project: Children as victims of and witnesses
to violence. Psychiatry, 56, 7–21.
Sattler, J. M. (1992). Assessment of children. San Diego:
Jerome M. Sattler.
U.S. House of Representatives, Committee on Ways and
Means. (1998). 1998 green book. Washington, DC:
U.S. Government Printing Office.
Wellborn, J. G., & Connell, J. P. (1987). Manual for the
Rochester Assessment Package for Schools. Rochester,
NY: University of Rochester.
Westat, Inc. (1998). Illinois Subsidized Guardianships
Questionnaire: Child Survey Scale. Rockville, MD:
Author.
are muffled by an array of difficult impediments.
These obstructions can be overcome with
thoughtful planning, but the process may be time
consuming and arduous. With a half million chil-
dren residing in out-of-home care in the United
States, researchers should continue to include
their perspective. Children can be important infor-
mants in our developing understanding of child
welfare outcomes, yet Festinger’s 1983 book title,
“
No One Ever Asked Us” continues to hold true
today. The current challenge to child welfare in-
vestigators is to devise and refine techniques for
gaining access to children in care and incorporat-
ing their perspectives into the research enterprise
with deliberation and utmost care. ■
Wilson, L. (1996). The 1995 annual client evaluation. Tulsa,
OK: Wilson Resources.
REFERENCES
Barth, R. P., Courtney, M., Berrick, J. D., & Albert, V. (1994).
From child abuse to permanency planning: Child
welfare services, pathways, and placements.
Hawthorne, NY: Aldine de Gruyter.
Berrick, J. D., Needell, B., Shlonsky, A., & Simmel, C.
Wulczyn, F. H., Harden, A.W., & Goerge, R. M. (1998).
Foster care dynamics, 1983–1997: An update from the
multistate foster care data archive. Chicago: University
of Chicago, Chapin Hall Center for Children.
Wulczyn, F. H., Zeidman, D., & Svirsky, A. (1997).
HomeRebuilders: A family reunification demonstra-
tion project. In J. D. Berrick, R. P. Barth, & N. Gilbert
(
1998). Assessment, support, and training for kin and
nonkin caregivers: An empirically based curriculum.
Berkeley: University of California, Berkeley, School
of Social Welfare.
(
Eds.), Child welfare research review (Vol II, pp. 252–
71). New York: Columbia University Press.
Yarrow, L. J. (1960). Interviewing children. In P. H. Mussen
Ed.), Handbook of research methods in child
2
Bierman, K. L. (1983). Cognitive development and clinical
interviews with children. In B. Lahey & A. Kazdin
(
(Eds.), Advances in clinical child psychology (pp.
217–250) New York: Plenum Press.
development (pp. 561–602). New York: John Wiley &
Sons.
Colton, M. (1989). Foster and residential children’s percep-
tions of their social environments. British Journal of
Social Work, 19, 217–233.
Fanshel, D., & Shinn, E. B. (1978). Children in foster care: A
longitudinal investigation. New York: Columbia
University Press.
Fein, E., Maluccio, A. N., & Kluger, M. P. (1990). No more
partings: An examination of longterm foster family
care. Washington, DC: Child Welfare League of
America.
Festinger, T. (1983). No one ever asked us—A postscript to
foster care. New York: Columbia University Press.
Fox, A., Frasch, K., & Berrick, J. D. (2000). Listening to
children in foster care: An empirically-based
curriculum. Berkeley: University of California,
Berkeley, Center for Social Services.
Jill Duerr Berrick, PhD
Associate Professor
School of Social Welfare
University of California, Berkeley
e-mail: dberrick@uclink4.berkeley.edu
Karie Frasch, MSW
Research Associate
School of Social Welfare
University of California, Berkeley
Adair Fox, MSW
Research Associate
School of Social Welfare
University of California, Berkeley
General Accounting Office. (1999). Foster care: Kinship care
quality and permanency issues (GAO/HEHS9932).
Washington, DC: U.S. Government Printing Office.
Hegar, R. L., & Scannapieco, M. (Eds.). (1999). Kinship
foster care: Policy, practice, and research. New York:
Oxford University Press.
Funding for the study was generously
provided by the California Social Work
Education Center and the David and
Lucile Packard Foundation. The authors
are indebted to Ellen Krouss, Barbara
Needell, and Amy Penkin for their
participation in the research study.
Johnson, P., Yoken, C., & Voss, R. (1994). Family foster care
placement: The child’s perspective. Child Welfare, 5,
9
59–974.
Lynch, M., & Cicchetti, D. (1991). Patterns of relatedness in
maltreated and nonmaltreated children: Connections
among multiple representation models. Development
and Psychopathology, 3, 207–226.
Original manuscript received July 3, 1999
Final revision received November 29, 1999
Accepted December 8, 1999
Needell, B., Webster, D., CuccaroAlamin, S., & Armijo, M.
(
1998). Performance indicators for child welfare
Assessing children’s experiences of out-of-home care: Methodological challenges and opportunities / Berrick, Frasch, and Fox
127