Content uploaded by Leo Loman
Author content
All content in this area was uploaded by Leo Loman on Feb 04, 2021
Content may be subject to copyright.
0
Financial Hardship and Child
Maltreatment: Six Studies in Five States
Summary Paper 1
by L. Anthony Loman PhD & Gary L. Siegel PhD
IAR Associates (January, 2021)
1
Abstract
This is the first IARA summary papers on family financial hardship and child welfare.
It is a summary of six multi-year, multi-method evaluation studies conducted by the authors
of programs directed toward families reported for child maltreatment. The programs each
involved intensive services to families, including increased financially related (material)
services. They also involved greater flexibility in worker-family decision making regarding
service needs and any assistance to be provided. Together the studies involved upwards of
20,000 families.
Five of the six were field experiments. In those studies, experimental families were
provided with intensive and flexible services while control families were treated in the
traditional manner by Child Protection Services (CPS) workers. The first two studies
summarized were experiments in which workers were permitted to expend Title IV-E funds
to address a broad array of family needs rather than simply payment for foster care. Those
programs were focused on high-risk child protection cases. Two other experiments
concerned lower-risk cases. In those, foundations provided extra funds that could be
expended on experimental children and their families. In the fifth experiment,
experimental families were directed into a state program designed to assist families
experiencing financial hardships. The sixth study provided intensive services on a voluntary
basis to families whose child maltreatment reports had not been accepted by CPS for
further action and other families referred from other state agencies.
Families were impoverished and, in most cases, suffering financially. In each study
material services increased for experimental families, and outcome measures demonstrated
improvements in the long-term safety and welfare of the children. The present summary
concludes by advocating reuniting income maintenance and child protection in the mold of
family and child welfare work before state CPS agencies were established in the early
1970’s.
This summary study is highly focused on particular subset of families suffering
financial hardship. Summary Paper 2 (available on our website: www.iarstl.org) approaches
the problem more broadly, focusing on poor and working class families generally.
Subsequent summaries turn to solutions.
2
Introduction
This paper is a summary of six large-scale multiyear, multimethod studies of the
effects of material assistance on families accused of child maltreatment (abuse or neglect).
Four employed prospective experimental designs, and three of those were Random Control
Trial (RCT) studies. One involved a retrospectively selected control group. The unique part
of the designs in each of the experimental studies was the provision of additional financially
related services to experimental families over and above those provided to control families.
Thus, they all addressed the question:
Does relief of financial hardship in families alleged to have abused and neglected
their children result in improved child safety and welfare?
The concepts underlying this question are embodied in what we refer to as
the Economic Hardship-Child Outcome (EHCO) Model. This model combines two others:
The Family Resource model and the Family Stress model, explained below.
The populations of families studied together covered the full spectrum of types of
maltreatment reports received by state and local Child Protection Services (CPS) agencies.
Two studies were focused on children either removed from their families or at risk of being
removed. Their cases represented the most extreme and dangerous end of the CPS
spectrum. Three of the five experiments concerned families served under a new approach
in CPS that excluded child maltreatment reports alleging severe physical abuse, sexual
abuse, and the most dangerous types of child neglect. Finally, a sixth non-experimental
study is reviewed that involved increased material services to families with reports of child
maltreatment that were not accepted by the CPS agency for further action.
The diagram in Figure 1, illustrates the process involved in the five experimental
studies. Each of the studies involved large populations (see 1 in Figure 1) of families from
which study groups were selected. The process of assigning families (2) to experimental or
control status involved random assignment or other quasi-experimental methods. The
assignment process produced similar groups, as we demonstrate in the following material.
The experimental treatment (3 and 4) in each case involved both intensive and flexible
financial assistance and material support. Enhanced and intensive assistance in two of the
studies arose from utilizing Title IV-E foster care money for other family- and child-related
needs beyond simply paying providers for residential care. In the two RCT DR studies
intensive assistance arose from funds provided by large private foundations. In the fifth
experimental study the enhancement was accomplished by redirecting experimental
families into an intensive assistance program. Flexibility in each study was achieved by
permitting social workers and families to determine jointly what the family and children
needed.
utilities,
counseling.
years
during th
present summaries will be on final long
reports are available to readers who want to know how families, workers and agencies may
have been af
summary studies have considered the existing literature linking poverty and child
maltreatment. Aislinn Conrad and associates were concerned with
about income maintenance and child welfare, including how cash and in
impact families in the CPS system.
improved family wellbeing. A study by Ashley Landers and
publications for the ten
that the studies show a link between poverty and maltreatment.
72 of the journal
related to this topic:
15 research articles on this topic written from a variety of theoretical perspectives.
population of
families with
needed.
This process revealed family needs
utilities,
and the like,
counseling.
Experimental and control families were tracked for
years
)
in each of the studies (5
during th
is follow
present summaries will be on final long
reports are available to readers who want to know how families, workers and agencies may
have been af
fected over time.
Poverty, Financial
Example
Studies of Poverty, Financial Hardship and Child Welfare.
summary studies have considered the existing literature linking poverty and child
maltreatment. Aislinn Conrad and associates were concerned with
about income maintenance and child welfare, including how cash and in
impact families in the CPS system.
improved family wellbeing. A study by Ashley Landers and
publications for the ten
that the studies show a link between poverty and maltreatment.
72 of the journal
related to this topic:
15 research articles on this topic written from a variety of theoretical perspectives.
1
. Large
population of
families with
children
2
.
Random assignment,
Prospective matching,
Retrospective matching,
etc.
Families selected
This process revealed family needs
and the like,
beyond the trad
Experimental and control families were tracked for
in each of the studies (5
is follow
-up
period and at the conclusions of the studies (6).
present summaries will be on final long
reports are available to readers who want to know how families, workers and agencies may
fected over time.
Poverty, Financial
Studies of Poverty, Financial Hardship and Child Welfare.
summary studies have considered the existing literature linking poverty and child
maltreatment. Aislinn Conrad and associates were concerned with
about income maintenance and child welfare, including how cash and in
impact families in the CPS system.
improved family wellbeing. A study by Ashley Landers and
publications for the ten
-
year period between 2008 and 2017.
that the studies show a link between poverty and maltreatment.
72 of the journal
Children
and Youth Services Rev
related to this topic:
Economic Causes and Consequences of Child Maltreatment
15 research articles on this topic written from a variety of theoretical perspectives.
Random assignment,
Prospective matching,
Retrospective matching,
Families selected
for study
This process revealed family needs
beyond the trad
itional
Experimental and control families were tracked for
in each of the studies (5
in Figure 1
period and at the conclusions of the studies (6).
present summaries will be on final long
-
term comparisons, but in each case full study
reports are available to readers who want to know how families, workers and agencies may
fected over time.
Poverty, Financial
Hardship
Studies of Poverty, Financial Hardship and Child Welfare.
summary studies have considered the existing literature linking poverty and child
maltreatment. Aislinn Conrad and associates were concerned with
about income maintenance and child welfare, including how cash and in
impact families in the CPS system.
1
They reviewed eight studies showing that such transfers
improved family wellbeing. A study by Ashley Landers and
year period between 2008 and 2017.
that the studies show a link between poverty and maltreatment.
and Youth Services Rev
Economic Causes and Consequences of Child Maltreatment
15 research articles on this topic written from a variety of theoretical perspectives.
Random assignment,
Prospective matching,
Retrospective matching,
Assignment
Process
Figure 1. Experimental Design.
3
This process revealed family needs
,
such as clothing, housing, transportation,
itional
social work
Experimental and control families were tracked for
in Figure 1
). Outcomes were compared
period and at the conclusions of the studies (6).
term comparisons, but in each case full study
reports are available to readers who want to know how families, workers and agencies may
Hardship
and Child Maltreatment
Studies of Poverty, Financial Hardship and Child Welfare.
summary studies have considered the existing literature linking poverty and child
maltreatment. Aislinn Conrad and associates were concerned with
about income maintenance and child welfare, including how cash and in
They reviewed eight studies showing that such transfers
improved family wellbeing. A study by Ashley Landers and
year period between 2008 and 2017.
that the studies show a link between poverty and maltreatment.
and Youth Services Rev
iew
Economic Causes and Consequences of Child Maltreatment
15 research articles on this topic written from a variety of theoretical perspectives.
Experimental
3
.
Enhanced and flexible financial
assistance and
4
.
Traditional financial assistance
and material resources
Assignment
Figure 1. Experimental Design.
such as clothing, housing, transportation,
social work
categories of education and
Experimental and control families were tracked for
extensive periods (measured in
). Outcomes were compared
period and at the conclusions of the studies (6).
term comparisons, but in each case full study
reports are available to readers who want to know how families, workers and agencies may
and Child Maltreatment
Studies of Poverty, Financial Hardship and Child Welfare.
summary studies have considered the existing literature linking poverty and child
maltreatment. Aislinn Conrad and associates were concerned with
about income maintenance and child welfare, including how cash and in
They reviewed eight studies showing that such transfers
improved family wellbeing. A study by Ashley Landers and
associates reviewed a list 86
year period between 2008 and 2017.
2
One of their conclusions is
that the studies show a link between poverty and maltreatment.
iew
was devoted completely to issues
Economic Causes and Consequences of Child Maltreatment
15 research articles on this topic written from a variety of theoretical perspectives.
Experimental
Families
Control
Families
Enhanced and flexible financial
assistance and
material resources
Traditional financial assistance
and material resources
Figure 1. Experimental Design.
such as clothing, housing, transportation,
categories of education and
extensive periods (measured in
). Outcomes were compared
at various points
period and at the conclusions of the studies (6).
The focus of the
term comparisons, but in each case full study
reports are available to readers who want to know how families, workers and agencies may
and Child Maltreatment
Studies of Poverty, Financial Hardship and Child Welfare.
summary studies have considered the existing literature linking poverty and child
maltreatment. Aislinn Conrad and associates were concerned with
multiple questions
about income maintenance and child welfare, including how cash and in
-
kind transfers
They reviewed eight studies showing that such transfers
associates reviewed a list 86
One of their conclusions is
that the studies show a link between poverty and maltreatment.
A supplement to v
was devoted completely to issues
Economic Causes and Consequences of Child Maltreatment
15 research articles on this topic written from a variety of theoretical perspectives.
Observe
Observe
Enhanced and flexible financial
material resources
Traditional financial assistance
and material resources
6
5
.Time= years
Figure 1. Experimental Design.
such as clothing, housing, transportation,
categories of education and
extensive periods (measured in
at various points
The focus of the
term comparisons, but in each case full study
reports are available to readers who want to know how families, workers and agencies may
and Child Maltreatment
Two recent
summary studies have considered the existing literature linking poverty and child
multiple questions
kind transfers
They reviewed eight studies showing that such transfers
associates reviewed a list 86
One of their conclusions is
A supplement to v
ol
ume
was devoted completely to issues
Economic Causes and Consequences of Child Maltreatment
. It includes
15 research articles on this topic written from a variety of theoretical perspectives.
3
Experimental
Families
Control
Families
6
.Compare
extensive periods (measured in
at various points
reports are available to readers who want to know how families, workers and agencies may
Two recent
kind transfers
They reviewed eight studies showing that such transfers
associates reviewed a list 86
One of their conclusions is
ume
was devoted completely to issues
. It includes
4
The Correlation between Child Maltreatment Reporting and Financial Hardship.
The relationship of financial hardship and poverty to child maltreatment, and more
generally, child welfare is well known. For example, Sedlak and her associates studied the
relationship in the 2010 Fourth National Incidence Study (NIS) of Child Abuse and Neglect.
They measured low socioeconomic status (SES) by combining measures of levels of income
and education and participation in poverty programs. Low SES children were approximately
five times more likely to experience maltreatment than children who were not in low SES
families.4 We found this again and again in our studies. For instance, in the Ohio study
considered below, we found that nearly seven in ten surveyed families reported for child
maltreatment had 2008 incomes of $15,000 or less compared to approximately 8% of the
entire population of Ohio families (Figure 2).5
Similar results were observed in the sixth study examined in this paper, the Parent
Support Outreach Program (PSOP).6 When workers assessed the financial condition of the
families who accepted services, they found that 60% had inadequate incomes or were in
poverty. In about 14%, the poverty was extreme and severe (Table 1).
34.2%
21.3%
12.6%
9.0%
10.2%
4.9%
7.8%
0% 5% 10% 15% 20% 25% 30% 35% 40%
Less than $4,999
$5,000 to $9,999
$10,000 to $14,999
$15,000 to $19,999
$20,000 to $29,999
$30,000 to $39,999
$40,000 or more
Figure
2
. Family income during the previous 12 month
s (Ohio Alternative
Response Evaluation, 2008 Incomes)
5
Table
1
. Household income during the past
12 months (Minnesota PSOP Study)
Percent
Cumulative
Percent
Less than $5,000 20.0% 20.0%
$5,000-$9,999 22.5% 42.5%
$10000>$14,999 18.9% 61.4%
$15000>$19,999 11.4% 72.8%
$20,000>$29,999 13.4% 86.2%
$30,000>$39,999 7.5% 93.7%
$40,000>$49,999 3.4% 97.1%
$50,000 or more 2.9% 100.0%
Thus, most families within the population reported for child maltreatment are
experiencing financial hardship. Looking at data from the other direction, we can ask
whether more reports of child maltreatment occur in financially distressed populations.
Community and neighborhood comparisons have shown higher incidences of child
maltreatment in areas of high and moderate child poverty compared to areas with low child
poverty. Several studies have shown that maternal age and residence in neighborhoods of
poverty strongly predicts substantiated reports of child maltreatment. For instance, Claudia
Coulton and associates demonstrated the relationship in a Midwestern urban area. Brett
Drake and Shanta Pandey did the same for an entire Midwestern state demonstrating a
relationship by neighborhood. Another study based on family surveys showed that child
abuse potential was predicted by neighborhood impoverishment and childcare burden.7
How is financial hardship related to potential child maltreatment or more generally
threats to the safety, health, and general welfare of children? There are two ways. First,
financial hardship and poverty are often misinterpreted as child maltreatment. Second,
financial hardship may be a moderating cause of child maltreatment. The distinction hinges
on the intentions of the parents.
Misinterpreting Financial Hardship as Child Maltreatment. This is nowhere better
explained than in an article by Leroy Pelton on the role of material factors in child abuse and
neglect.8 Pelton explains that injuries due to neglect are largely unintentional, that is,
accidental. Because of the U.S. nationwide system of reporting child maltreatment, many
incidents of severe injury or severe threats to child safety that might have once been
attributed to mere accident are now labeled as parental neglect. This is a confusion of
financial hardship and poverty with child neglect. Case examples rather than statistics are
the best way to come to understand this. The following example is drawn from descriptions
provided by a worker during our evaluation of the Minnesota project, discussed below.
6
Case Study 1. A report was received on a mother-only family living in a trailer home.
The reporter alleged that the trailer was unsafe. A family assessment worker (rather
than a traditional CPS investigator) was dispatched to determine whether the report
was correct and what was needed. She called the mother and then went to visit her
and the children at the house trailer. She observed that there were several rotten
boards in the floor of the trailer. This was dangerous for the children, all of whom
were preschool ages. In cases we have observed under the traditional child
protection system this kind of problem sometimes led to removal of children from
the home until the family corrected the problem. The mother acknowledged the
problem but said she did not have enough money to fix it. She told the worker that
she was handy and could fix the boards herself if only she could purchase them.
The worker returned to her office and explained the situation to her supervisor. So,
the alternatives were to remove the children at a cost of hundreds of dollars in
worker time, court personnel time, transportation costs, foster care home costs, etc.
or fix the trailer. Fortunately, this state had the extra funds (as part of the
Minnesota AR (differential response) project that we were evaluating) set aside for
experimental families that could be legally used to remedy these kinds of situations.
The worker suggested that a purchase order to the local hardware store might be
provided to the mother. The supervisor agreed and it was done. The worker than
returned to the home and the mother agreed to do the work. She went to the
hardware store and purchased the materials she needed, returned, and replaced the
rotten floorboards. Problem solved and case closed.
Here are two other examples drawn from the Mississippi IV-E study, outlined below.
The first also considers inadequate housing. The second involved homelessness.
Case Study 2. A five-year-old boy was at risk of being removed from his parents who
lived in a mobile home. The parents, both described as mentally challenged, had
extremely limited financial resources. While the family had no prior case history,
there were significant safety concerns for the child due to the uninhabitable
condition of the mobile home. Part of the trailer had collapsed, and the hot water
heater and commode had fallen out and into the back yard. Electric wires coming
into the home ran through water beneath the trailer. Waiver funds were added to
county funds to purchase construction material, a new hot water heater, windows,
and a door. Through a family-team conference the grandparents became involved,
taking temporary care of the boy, while members of the family’s church provided
volunteer labor to rehab the trailer.
7
Case Study 3. This case involved a two-parent family with an 11-year-old son and a
14-year-old daughter. The father was disabled and unable to work and the family
had a history of instability and frequent moves that led to serious behavioral and
emotional problems for the children. The family became homeless when they were
unable to pay their rent and were forced out of their home. The children were taken
into custody and placed in foster care, but the placement situations were unreliable.
Through the waiver, the family received short-term assistance to pay rent and the
parents found new living arrangements that allowed the son to live with them once
again. Waiver funds were also used to pay for tutoring services for the boy to help
him catch up for missed time at school. A placement with relatives was secured for
the daughter through waiver funds which paid for a bed and medication prescribed
to address her bouts of severe anxiety. Finally, the mother was helped to find a job
and the family attained a level of stability it had lacked for several years.
Poverty and Financial Hardship as a (Moderating) Cause of Child Maltreatment.
The term cause is a dangerous one to use when discussing social phenomena. If someone
says economic hardship causes child abuse or child neglect the implication is that families
experiencing economic hardship will always engage in child maltreatment. This is obviously
incorrect. Most families suffering financial hardship do not intentionally abuse or neglect
their children. However, there is another more complex sense in which something can be a
cause. We term it as moderating. As in the following diagram (Figure 3), some
characteristic of a child or parent—for example, the practice of using physical punishment
to discipline their child—is worsened because of the stress of financial hardship.
Rand Conger and Katherine Conger have outlined a causal sequence that illustrates
the moderating cause process more fully, which they name The Family Stress Model. 9 The
model includes the following:
1. Economic hardship (low income, high debt, low assets, and negative financial events)
leads to:
2. Economic pressure (unmet material needs, unpaid debts and painful cutbacks).
These in turn lead to:
DIRECT CAUSES: Caregiver/Child/Context Variables
Child Maltreatment
cause
MODERATING CAUSE: Financial Hardship
moderates = exacerbates/worsens the
direct causes through stress, etc.
Figure
3. Diagram of direct and moderating causes
8
3. Parent distress (emotional and behavioral problems), which produces:
4. Disrupted family relations (inter-parental conflict or withdrawal and harsh or
inconsistent parenting). Finally, this produces:
5. Child and adolescent adjustments (emotional problems, behavioral problems and
impaired competence).
This implies that if economic hardships were relieved outcomes would be improved
for families and children. The Rands also present the obverse of the stress model: The
Family Resource Model. In that model family resources and the lack thereof are predicted
to affect child outcomes. There are resources that are only indirectly related to finances
but the critical ones such as health, education, housing, living arrangements, transportation,
leisure activities, neighborhoods, safety, etc. are clearly produced or directly affected by
income and wealth. This model may be regarded as describing the misinterpretation
understanding outlined in the previous sub-section. The combination of these models
might be called the Economic Hardship-Child Outcome Model, the EHCO Model.
The EHCO model implies that lack money, that is, economic hardship, may directly
harm children. However, it also suggests that such hardship may exacerbate negative traits,
destructive behaviors and harmful human relationships that directly cause child
maltreatment. On the other hand, the model predicts that sufficient money makes the
expression of positive traits, supportive behaviors and beneficial relationships more likely.
This might be called the Aggravation from Deprivation-Alleviation through Relief, the ADAR
understanding of child welfare. Financial deprivation aggravates other problems in families
and makes positive outcomes less likely while financial relief alleviates problems and makes
positive outcomes more likely.
A recent study by Mi-Youn Yang shows the effects of relationship between material
hardship, parent psychological states on investigated reports of child maltreatment among
families receiving cash welfare.10 Material hardship was measured in four areas: sufficient
food, adequate housing and homelessness, problems with utilities, and medical
deficiencies. Yang demonstrated the correlation between future investigations of child
maltreatment and problems in any one of these areas. The correlation was stronger when
hardship was experienced in two or more areas. The study demonstrates the relationship
of parental stress and depressive symptoms arising from material hardship on later reports,
as well. Discussions of food insecurity, inadequate housing, and homelessness are
discussed in greater detail in the second summary paper in this series (at www.iarstl.org).11
Another recent paper focused on family income and child welfare by Greg Duncan
and associates considers how economic disadvantage leads to stress in families and in turn
9
affects the development of children. For those who wish to immerse themselves further in
the literature on this topic this study is abundantly referenced.12
Here are several case examples illustrating moderating causes. As you will see,
some of these tend to shade over into the previously discussed category of
misinterpretation. This might be argued in the following case, which is drawn from an
earlier observational study we conducted. The mother’s psychological state, her
depression, may have been a causative factor in what occurred but and her unemployment
and lack of income might have worsened her mental condition.
Case Study 4. This case arose from a report that two younger children (ages 3 and 5
years) were being cared for by a preteen child (age 10). After an investigation, it was
substantiated as lack of proper supervision. The mother had recently lost her job
and was suffering from depression. The family was currently living on the social
security check that came to the older child, whose father had died the year before.
When one of us visited the family with a service worker, the mother expressed guilt
about using her daughter’s check for food and rent. She had no alternative for
childcare when she had to go out. Limited resources were a major part of the
problem leading to the finding of lack of proper supervision.
The following case illustrates how depression, stress and poverty may interact in a
family. This was voluntary case that we followed in the previously mentioned PSOP study.
Case Study 5. A referral was made by another agency for a young woman, who was
requesting services. She had worked with the county previously and requested a
specific worker, as she felt comfortable with her. She had just had a baby and was
feeling overwhelmed financially and emotionally. She had recently begun working at
a local grocery store but was still relying on welfare (TANF). In addition, she had
been diagnosed with post-traumatic stress disorder, depression, and anxiety. She
wanted help with parenting initially, but the PSOP worker primarily intervened to
help manage crises that arose. The young woman had trouble maintaining
employment. Indeed, she lost two jobs during the case period. She had several
barriers to becoming financially independent: no high school diploma, no license, no
vehicle, no child support, and trouble with childcare. The program was able to assist
with rent, utilities, and other basic needs for a month. Her infant son had also been
hospitalized twice in the previous month for a medical condition. The worker helped
her think about how to manage the child’s condition. She also encouraged her to
work on developing her parenting skills. A monthly budget was developed. She was
referred to a mental health agency for long term help. Chemical use was also a
possible issue, and she was provided some education on substance abuse. A home-
10
based worker was recruited for the case to work more directly on mental health and
parenting, and this allowed the worker to focus more on coordination and case
management. During the case, a report was received regarding the mother’s
behavior and parenting, but nothing new or substantial was reported. Later, she
became pregnant again and began having domestic incidents with her new
boyfriend. However, she did secure employment and began to attend individual
counseling. The workers continued to address the goals of safety, budgeting, and
home management. The case was closed by mutual agreement after the mother felt
she had adequate access to other services and was satisfied with her current
situation.
The following is a case of a drug-addicted mother. It was part of an earlier
observational study. The report was made by a policeman. In this instance a formal CA/N
investigation was conducted resulting in the removal and placement of her children into
relative care.
Case Study 6. An anonymous report was received that a mother had left her five
children alone from 7:00 a.m. to 7:00 p.m. This was a mother-only family, as she
was separated from the father of the children. The mother told the children that
she was going to the grocery store, but when she returned, she had no groceries,
because she had sold her food stamps in order to buy drugs. The reporter said that
the children were left alone often and that the two youngest (ages one and two)
usually played alone outside. The electricity had been turned off in the house the
day before.
Following the report, a maternal aunt of the children took them in, but two more
hotline reports were received two days later. The first may have been from the
aunt, who reported that the children’s maternal grandmother had come that
morning and picked them up. The reporter claimed that the grandmother was not
physically capable of caring for the children. The second hotline was from a
policeman who was at the grandmother’s home, saying that immediate action was
needed because the mother was at the grandmother’s home. She wanted to take
the children, although she had no money, no food, and no home. The local shelters
were not a resource, because the mother was on drugs and was extremely high at
the time. The policeman confirmed that the mother had used all her welfare and
food stamps to buy drugs. The policeman arrested the mother and took her to the
county jail. He also took the children into protective custody and transported them
to the family court. The investigator went to the family court, met the children, and
interviewed them with a juvenile officer. They found the children highly active but
11
unresponsive to adults. All the children were aware that their mother had spent all
the family’s money on drugs and showed detailed knowledge of buying and using
dope. Together the investigator and juvenile officer decided to place the three
youngest children with the mother’s sister and the two oldest children with their
father. Whether the placement with the father was wise is debatable since we later
heard that he was a drug dealer who had supplied the drugs to mother as she
became addicted. However, this was hearsay. This arrangement was approved by
the juvenile court, and at the detention hearing, custody and control was given to
the child welfare agency. The investigator completed her report with a formal
finding of child neglect.
About three months later, one of us accompanied the foster care worker and the
children on a visit to the mother at a drug rehabilitation center. The mother had
voluntarily agreed to undergo drug treatment and was doing well. She was well
dressed and cheerful. The children obviously loved their mother and were taking
turns being held. Later she was released and regained custody of her children.
This case illustrates how poverty and other mediating variables may interact.
Whether and how poverty might have been involved in creating her addiction is unknown.
That problem may have been more attributable to the influence of the children’s father, but
we did not confirm that. After the mother was addicted, her need for drugs contributed to
her financial distress. Had she not been addicted she would probably have been simply
another poor single mother. In this case, unlike many we have observed, she had the
advantage of a supportive extended family that helped during her rehab.
Here is another case in which funds were used to prevent removal and placement of
children. It was part of the Indiana IV-E project, discussed below.13
Case Study 7. A report came from the police regarding lack of supervision. A
maintenance man who came to fix the stove reported finding a two-year-old child
alone in the apartment. The mother arrived an hour later and stated that she had
gone to the Trustee’s office for some financial help and did not want to wake her
child. She had left him alone with safety gates on the bedroom door and admitted to
previously having left the child alone on several occasions. The report was
substantiated, and a short-term Informal Adjustment case was opened.
The family consisted of a single mother and her two-year old son who was
developmentally disabled. It appeared that the family was struggling financially, was
socially isolated and that the mother was depressed. The mother had been
unemployed for the last two months, in part due to attending to the disabilities of
12
her child. During those two months, the child had undergone two surgeries for
clubfoot. The mother had been employed full-time prior to the operations, but she
was forced to quit her job when she could not secure extended leave. The lack of
regular income threatened imminent homelessness for the family. They did not have
any family support in the area. The family had no prior history with the Department
of Children’s Services.
The mother agreed to cooperate with a parent aide and an in-home therapist,
accepted home visits from First Steps, properly supervise her son, obtain adequate
medical care, and administer medication he needed. Although the mother had two
job prospects, she was waiting for appropriate daycare for her son. Past rent in the
amount of $677 was owed and the family was facing eviction. Despite support from
Section 8, she had accrued a large past due balance on her rent, including late fees
and eviction filing fees. The woman was provided with financial help for rent
assistance, vouchers for daycare, counseling, and an in-home parent aide. The
parent aide helped the mother with budgeting, looking for work and locating the
nearest food pantry. The mother was compliant with DCS and all service providers.
The waiver utilization included $5,912 for rent assistance, home- based counseling
and parent aide.
The case was assigned to waiver status to assist the family maintain its housing and
avoid eviction. Without this assistance, the family may have lost their apartment and
the case could have easily risen to a removal and placement due to the child’s
special needs. The mother successfully completed 90-days of in-home therapy and
parent aide sessions. Great progress was made, and the assigned therapist
concluded that the mother did not need continued services. All of the child’s
doctor’s appointments were successfully attended and the mother also obtained
consistent daycare. At case closure, the mother was employed and she and her child
were living in a stable environment.
Here is another case of homelessness, something we observed in large numbers in
the impoverished Mississippi counties in that IV-E study. Note here the reference to
disabilities and frequent moves as direct causes.
Case Study 8. This case involved a two-parent family with an 11-year-old son and a
14-year-old daughter. The father was disabled and unable to work and the family
had a history of instability and frequent moves that led to serious behavioral and
emotional problems for the children. The family became homeless when they were
unable to pay their rent and were forced out of their home. The children were taken
13
into custody and placed in foster care, but the placement situations were unreliable.
Through the waiver, the family received short-term assistance to pay rent and the
parents found new living arrangements that allowed the son to live with them once
again. Waiver funds were also used to pay for tutoring services for the boy to help
him catch up for missed time at school. A placement with relatives was secured for
the daughter through waiver funds which paid for a bed and medication prescribed
to address her bouts of severe anxiety. Finally, the mother was helped to find a job
and the family attained a level of stability it had lacked for several years.
As noted, cases of extreme poverty were common. The approach was to try to meet
basic and home-related needs so that children did not have to be removed and placed in
foster care. Here is another example that would seem to fall into the previous category of
poverty misinterpreted as child neglect.
Case Study 9. One such case involved a woman with four children between the ages
3 and 9. A boyfriend of the woman lived with her from time to time and was the
father of the children, but he did not provide a steady source of support. A child
protection case was opened on the family after the utilities had been turned off and
there was little food in the house. To forestall the removal of the children until the
situation could be more permanently addressed, waiver funds were used to
purchase food and pay the utility bills so that power, heat, and water could be
restored to the home.
The stop-gap nature of this case was replicated in many other cases we followed.
The following is an example of the use of IV-E waiver funds and the hard work of a
wonderful worker to assist in family reunification. Direct causes in this case means ‘low
functioning’ parents.
Case Study 10. This example involved the case of a young family with low
functioning parents and three children ages 2, 7 and 9. The family had been
homeless until a brother gave them a small mobile home to use. But the trailer was
unsafe and did not have functioning sanitary or electrical systems. The children were
taken into custody, removed, and placed in foster care for neglect. The foster home
was 100 miles from the parents; no closer foster placement could be found that
would accept all three children. Waiver and county funds were used to purchase
needed materials to repair the trailer. The parents took an active hand in the
process and repaired the windows and fixed the leaking roof and had a septic tank
installed. The county social worker, showing extraordinary commitment to her case
family, laid the plumbing and did all the needed electrical work herself during her off
14
hours. Once repairs were completed, the Juvenile Court judge allowed the children
to be released from paid foster care and reunited with their parents.
Increased Material Services Coupled with Flexible Decision-
Making
Two changes were represented in the experimental approaches of each of the
studies. The first involved increased funding for various services including material services,
such as housing, rent, food, clothing, household items, transportation, and the like.
The approaches taken with the families in the following studies as illustrated in the
previous case examples were not representative of what has happened traditionally in CPS.
Urban CPS offices, for instance, receive thousands of reports of child maltreatment each
year. Each requires a response. Only a minority are substantiated and opened as
protection and service cases. In most instances, cash-strapped agencies are unable to do
much more than simply monitor the family for a period. The bulk of agency funds in CPS is
expended on foster care and expensive residential treatment services.14
An author who has written extensively on poverty and child maltreatment is Duncan
Lindsey. He offers evidence in an older book, The Welfare of Children, that the
fundamental problem that underlies most cases encountered by child protection agencies is
child and family poverty.15 Lindsey also presents evidence in that book showing the
ineffectiveness of social casework in producing changes in families. The reason is not so
much that social work with families is a waste of time. Rather the problem is the restrictive
ideology that traditionally surrounded social work with families accused of child
maltreatment, which he called the residual approach. Under that way of thinking, only the
most severe cases are given services while less severe cases are provided minimal or no
assistance other than monitoring. In addition, the severe cases that do receive services are
provided with only enough assistance to bring them back to a level of minimally acceptable
functioning. Then the case is closed.
The six programs reviewed and summarized in this paper were the opposite of the
residual approach. Those programs emphasized more intensive services aimed at
remediating various family problems and preventing future problems, especially those
related to financial hardship.
The second change in the programs reviewed here involved greater flexibility and
democracy in the process of selecting and applying those services. In each of the projects,
workers were directed to engage experimental families in assessing family needs and in the
decisions surrounding how to address those needs. The previous example of the joint
decision of the mother and worker to pay for boards to repair a dangerous floor in the
15
family house trailer rather than removing the children illustrates this process. This occurred
not simply because of increased funds to pay for material services but because workers
were instructed to take the opinions and desires of parents and children into account in
making decisions, that is, joint worker-family decision-making.
In the following summaries of the four prospective experimental studies we first
show the similarity between experimental and control groups. Then we demonstrate that
services of various kinds increased for experimental families. For this we depend mainly on
worker reports and agency databases. Study families were only contacted several weeks
after case closure and survey response rates were low because these impoverished families
are typically very residentially unstable. Nonetheless, responses of families about the
nature and level of services were similar to those of the workers who served them. Finally,
we illustrate outcome differences reflecting the safety and welfare of children that we were
able to measure.
16
Two Title IV-E Waiver Experiments
Title IV-E of the Social Security Act provides states and tribes with funding to assist
with the costs of foster care maintenance for eligible children; administrative expenses to
manage the program; and training for staff, foster parents, and certain private agency staff.
Eligibility varies somewhat from state to state but generally the families and children served
are in or near official poverty levels. State’s must provide a percentage match to the federal
dollars. In the present studies the federal contribution amounted to around two-thirds of
the total foster care cost. Thus, for every $100 spent to provide foster care for a qualifying
child the states received $60-70 in compensation. In the projects considered here,
Mississippi and Indiana were granted waivers from the standard IV-E spending rules. The
states were permitted to spend monies more broadly and flexibly. These are examples of
what we earlier referred to as intensive services. Funds that traditionally could only be
spent to pay for foster care could now be used to address family needs of various kinds.
There was also greater flexibility in utilizing available funds, including addressing the
material needs of involved families. In both states, the waiver programs included children
who had been removed from their families and placed in out-of-home care as well as
children who were in danger of being removed but were permitted to remain in their
homes. For the former group of children, a primary goal was shortening the length of time
that the child remained in foster care. For the latter, it was preventing removal.
As noted, the children in the studies represented the most extreme side of the child
protection/child welfare spectrum, that is, children considered to be at high risk to their
ongoing safety and welfare.
Study 1: The Mississippi Title IV-E Waiver Experiment
This study took place during a 42-month period from April 2001 through September
2004 in six Mississippi counties.16 To get a sense of the population of families included in
the study, Mississippi ranked first in child poverty among the 50 states in 2000, first in the
percent of families in poverty, first in the number of households headed by single women,
and 47th in median household income.
Initially families and children in existing Title IV-E eligible cases were identified and
submitted to a random assignment program on portable computers we supplied to the local
offices. New cases continued to be assigned during subsequent months. Experimental
children and their families were eligible for intensive services paid for with IV-E funds;
control cases were not. By the time of the final analysis, 777 experimental children in 346
families and 772 control children in 321 families were available for analysis. Experimental
17
and Control groups were similar. No statistically significant differences were found in the
following demographic or case characteristics (Table 2).
Table
2
.
Characteristics of
experimental and control
children and families
Children
Experimental
(n=777)
Control
(n=772)
Risk Level
High 61.1% 60.0%
Medium 35.6% 37.6%
Low 2.4.% 2.5%
Gender
Male 47.2% 50.3%
Female 52.8% 49.7%
Race
White (Caucasian) 48.5% 42.6%
Black (African American) 38.5% 44.0%
Asian American .2% .1%
Unindicated 12.8% 13.1%
Families
Experimental
(n=346)
Control
(n=321)
Children
One 35.1% 37.6%
Two 21.8% 25.9%
Three or more 43.1% 36.5%
Race
White (Caucasian) 51.9% 45.0%
Black (African American) 34.5% 44.1%
Mixed 9.3% 8.4%
Household
Female Present 92.2% 91.9%
Male Present 44.9% 42.2%
Female and Male Present 39.1% 37.2%
Initial Maltreatment Finding
Physical abuse 19.3% 17.3%
Emotional abuse 6.2% 9.5%
Neglect 58.3% 59.0%
Sexual Abuse 12.1% 9.2%
Nearly all the children were rated as high or medium risk at the time of entry into
the random assignment program. This is a confirmation that this study concerned CPS cases
from the more dangerous end of the child safety spectrum. Differences in racial
designations between children and families illustrate inaccuracies and inconsistencies in
such designations in the state’s data system, something we have found many times in other
states. A mother was present in the large majority of families but only abo
were two
those cases the neglect was actually family poverty along the lines the case examples
presented above.
Services.
and their children
The service increases were modest
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
Adult Substance Abuse Services
Counseling/Therapy/Treatment
Prescription Glasses/Hearning Aids
Prescription Drugs/Medication
Other Unmet Personal Needs
states. A mother was present in the large majority of families but only abo
were two
-
parent. About six in every ten families involved child neglect, and in many of
those cases the neglect was actually family poverty along the lines the case examples
presented above.
Services.
Figure
and their children
were on the waiver
The service increases were modest
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
no funded services
Child care expenses
Independent living
Foster home/facility
Temporary Boarding
Emergency Shelther
Adult Substance Abuse Services
Counseling/Therapy/Treatment
Psychological Evaluation
Prescription Glasses/Hearning Aids
Prescription Drugs/Medication
Unmet Medical Dentist
Other Unmet Personal Needs
Transportation
Personal Hygiene
Home Improvement
Housing,Rent, Utilities
School Supplies
Figure
states. A mother was present in the large majority of families but only abo
parent. About six in every ten families involved child neglect, and in many of
those cases the neglect was actually family poverty along the lines the case examples
Figure
3
shows the differences in services
were on the waiver
The service increases were modest
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
0%
no funded services
any services
Child care expenses
Independent living
Foster home/facility
Temporary Boarding
Emergency Shelther
Adult Substance Abuse Services
Counseling/Therapy/Treatment
Psychological Evaluation
Prescription Glasses/Hearning Aids
Prescription Drugs/Medication
Unmet Medical Dentist
Medical
Other Unmet Personal Needs
Transportation
Personal Hygiene
Food Needs
Home Improvement
Housing,Rent, Utilities
School Supplies
Clothing
Figure
3
.
Services provided to the experimenta
states. A mother was present in the large majority of families but only abo
parent. About six in every ten families involved child neglect, and in many of
those cases the neglect was actually family poverty along the lines the case examples
shows the differences in services
were on the waiver
.
The service increases were modest
overall
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
2.9%
8.4%
10.7%
1.4%
15.6%
7.5%
3.2%
13.9%
4.9%
14.5%
6.4%
4.9%
15.0%
5.5%
16.8%
1.6%
6.2%
11.2%
0.6%
13.4%
6.9%
2.2%
11.2%
3.4%
14.6%
5.0%
5.3%
10.9%
4.0%
15.3%
11.8%
0%
10%
20%
Services provided to the experimenta
during the Mississippi waiver
18
states. A mother was present in the large majority of families but only abo
parent. About six in every ten families involved child neglect, and in many of
those cases the neglect was actually family poverty along the lines the case examples
shows the differences in services
overall
. Total funded services increased for
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
25.7%
21.4%
33.2%
10.7%
15.6%
13.9%
14.5%
15.0%
23.4%
16.8%
33.0%
20.2%
34.3%
11.2%
13.4%
11.2%
14.6%
10.9%
15.3%
11.8%
20%
30%
40%
Control
Experimental
Services provided to the experimenta
during the Mississippi waiver
states. A mother was present in the large majority of families but only abo
parent. About six in every ten families involved child neglect, and in many of
those cases the neglect was actually family poverty along the lines the case examples
shows the differences in services
during the months the families
. Total funded services increased for
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
33.2%
48.3%
33.0%
34.3%
55.8%
49.2%
40%
50% 60%
Control
Experimental
Services provided to the experimenta
l and control groups
during the Mississippi waiver
states. A mother was present in the large majority of families but only abo
ut four in ten
parent. About six in every ten families involved child neglect, and in many of
those cases the neglect was actually family poverty along the lines the case examples
during the months the families
. Total funded services increased for
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
74.6%
61.6%
67.0%
55.8%
70% 80%
l and control groups
ut four in ten
parent. About six in every ten families involved child neglect, and in many of
those cases the neglect was actually family poverty along the lines the case examples
during the months the families
. Total funded services increased for
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Money for housing, rent and utilities (E: 23%; C: 15%); Food needs (E: 15%; C: 11%);
74.6%
l and control groups
experimental families (Experimental (E): 75%; Control (C): 67%). Here are some examples:
Prescription drugs and medications (E: 14
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
families.
any positive effects.
19.7% of control cases compared to 14.5% of
was substantial considering the short time frames involved (less than two years for most
cases) and was statistically significant
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
percent of co
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
compared
new reports of sexual maltreatment compared to 3.0 percent of control children. The
pattern seen in
0%
5%
10%
15%
20%
25%
Prescription drugs and medications (E: 14
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
families.
Outcomes.
any positive effects.
19.7% of control cases compared to 14.5% of
was substantial considering the short time frames involved (less than two years for most
cases) and was statistically significant
Whe
n considering specific types of maltreatment (see Figure 11), a statistically
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
percent of co
ntrol children (p=.02). Differences between the two groups in new reports of
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
compared
to 14.6 percent of control children; and 2.4 percent of
new reports of sexual maltreatment compared to 3.0 percent of control children. The
pattern seen in
Figure
14.6%
12.4%
Neglect
Figure
4
.
Prescription drugs and medications (E: 14
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
Outcomes.
We examined whether these increases in mainly material services had
any positive effects.
W
e found that subsequent new
19.7% of control cases compared to 14.5% of
was substantial considering the short time frames involved (less than two years for most
cases) and was statistically significant
n considering specific types of maltreatment (see Figure 11), a statistically
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
ntrol children (p=.02). Differences between the two groups in new reports of
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
to 14.6 percent of control children; and 2.4 percent of
new reports of sexual maltreatment compared to 3.0 percent of control children. The
Figure
4
of more reports for control children was found for
12.4%
.
Percentages of new reports of maltreatment in experimental and
control groups
Prescription drugs and medications (E: 14
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
We examined whether these increases in mainly material services had
e found that subsequent new
19.7% of control cases compared to 14.5% of
was substantial considering the short time frames involved (less than two years for most
cases) and was statistically significant
(p = .004)
n considering specific types of maltreatment (see Figure 11), a statistically
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
ntrol children (p=.02). Differences between the two groups in new reports of
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
to 14.6 percent of control children; and 2.4 percent of
new reports of sexual maltreatment compared to 3.0 percent of control children. The
of more reports for control children was found for
6.0%
3.7%
Physical Abuse
Control
Percentages of new reports of maltreatment in experimental and
control groups
19
Prescription drugs and medications (E: 14
%; C: 11%); Other personal needs (E: 62%; C: 56%);
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
We examined whether these increases in mainly material services had
e found that subsequent new
19.7% of control cases compared to 14.5% of
Experimental
was substantial considering the short time frames involved (less than two years for most
(p = .004)
.
n considering specific types of maltreatment (see Figure 11), a statistically
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
ntrol children (p=.02). Differences between the two groups in new reports of
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
to 14.6 percent of control children; and 2.4 percent of
new reports of sexual maltreatment compared to 3.0 percent of control children. The
of more reports for control children was found for
3.0%
Physical Abuse
Sexual Abuse
Control
Exp
Percentages of new reports of maltreatment in experimental and
control groups
(Mississippi evaluation)
%; C: 11%); Other personal needs (E: 62%; C: 56%);
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
We examined whether these increases in mainly material services had
e found that subsequent new
child maltreatment
Experimental
cases (
was substantial considering the short time frames involved (less than two years for most
n considering specific types of maltreatment (see Figure 11), a statistically
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
ntrol children (p=.02). Differences between the two groups in new reports of
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
to 14.6 percent of control children; and 2.4 percent of
new reports of sexual maltreatment compared to 3.0 percent of control children. The
of more reports for control children was found for
3.0%
2.4%
Sexual Abuse
Percentages of new reports of maltreatment in experimental and
(Mississippi evaluation)
%; C: 11%); Other personal needs (E: 62%; C: 56%);
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
We examined whether these increases in mainly material services had
child maltreatment
rep
cases (
Figure 4
). This difference
was substantial considering the short time frames involved (less than two years for most
n considering specific types of maltreatment (see Figure 11), a statistically
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
ntrol children (p=.02). Differences between the two groups in new reports of
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
to 14.6 percent of control children; and 2.4 percent of
experimental children had
new reports of sexual maltreatment compared to 3.0 percent of control children. The
of more reports for control children was found for
both
19.7%
New Report
Percentages of new reports of maltreatment in experimental and
(Mississippi evaluation)
%; C: 11%); Other personal needs (E: 62%; C: 56%);
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
and children received an average of 3.2 different services compared to 2.8 for control
We examined whether these increases in mainly material services had
rep
orts occurred in
). This difference
was substantial considering the short time frames involved (less than two years for most
n considering specific types of maltreatment (see Figure 11), a statistically
significant difference was found in new reports of physical abuse: 3.7 percent of
experimental children had new incident reports of physical abuse compared with 6.0
ntrol children (p=.02). Differences between the two groups in new reports of
neglect or sexual abuse were not significant, although the differences were in the
hypothesized direction: 12.4 percent of experimental children had new neglect reports
experimental children had
new reports of sexual maltreatment compared to 3.0 percent of control children. The
both
pre-
existing
19.7%
14.5%
New Report
Percentages of new reports of maltreatment in experimental and
%; C: 11%); Other personal needs (E: 62%; C: 56%);
School supplies (E: 17%; C: 12%). Looking across all services received, experimental families
We examined whether these increases in mainly material services had
orts occurred in
). This difference
was substantial considering the short time frames involved (less than two years for most
ntrol children (p=.02). Differences between the two groups in new reports of
hypothesized direction: 12.4 percent of experimental children had new neglect reports
experimental children had
new reports of sexual maltreatment compared to 3.0 percent of control children. The
existing
20
cases and new cases. It was also found for cases that had closed prior to the end of data
collection and those that remained open.
A survival analysis was conducted that considered variations in time after each
target case had closed. This method considers both whether new reports occur and how
long it was before they occurred. This time period is referred to as survival time, that is,
how long the family “survives” until a terminal event occurs. In this case the event was a
new accepted report of child maltreatment.17 Control children experienced new reports
sooner. Thus, more reports were recorded for this group during the follow-up period. The
experimental-control differences are shown in Figure 5 and were statistically significant (p =
.03).
Perhaps most
importantly, we were
able to show that the
greatest differences
occurred for children who
received services. The
reduction in new reports
of child maltreatment in
such cases amounted to
5.6% (E: 15.8%; C: 21.4%;
p = .04). The increase in
material services had
positive effects on
families.
In addition, under
the waiver program
children who remained
with their parents were
less likely to be removed (9.1%) than similar control children (14.1%). Again, and most
importantly, the effects were powerful in families that received services. Among those
families 33.1% of children were later removed compared to 57.2% in control families. This
also was shown to be a real difference when variations in time were considered. A survival
analysis confirmed that control children experienced out-of-home placement more often
and sooner that experimental children (p = .025)
The Mississippi IV-E experiment demonstrated that spending money on poor
families improved the subsequent welfare and safety of the children. Intensive services to
Survival Function
Days to a New Child Maltreatment Report
120010008006004002000
Cumulative Survival
1.1
1.0
.9
.8
.7
.6
.5
.4
.3
The higher the
line the better the
survival
c
Experimental
Control
E
o
Figure
5
.
Survival of experimental and control families until
a new child maltreatment report (20-day intervals)
21
such families enhanced their stability and reduced the likelihood that children would later
be taken into state custody.
Study 2: The Indiana Title IV-E Waiver Experiment
We conducted a similar intensive and flexible services experimental study in Indiana
over a five-year period (2005-2010).18 Because the state would not permit random
assignment to experimental and control groups, a child-matching design was developed.
This is a weaker assignment method than random assignment but has been utilized in many
past experimental studies.19 The object was not to produce matched pairs for analytic
purposes but to produce matched group for final group comparisons. The numbers
eventually assigned were exceptionally large and by the conclusion of the study there 9,475
experimental (waiver) children in closed cases and 9,358 similar control children available
for comparison.
As a part of the waiver program design, Indiana was permitted to assign up to half of
children to the program who were not eligible for Title IV-E services. This was permitted so
long as the state maintained cost-neutrality, that is, so long as no greater expenditures took
place than would have taken place under the traditional IV-E program.20 Characteristics of
experimental (waiver) and control children can be seen in Table 3.
The two groups were highly similar on a variety of characteristics suggesting that the
matching procedures were accurate and effective. Differences can be seen in some initial
case types but these types shifted and changed in the period following case opening. As can
be seen this study included delinquent youths as well as children in CPS cases. Delinquents
were, of course, older as a group. The placement statistics illustrate the high-risk nature of
most of the cases.
Services. Services to families and children expanded significantly and substantially
for many types (Figure 6).
The information in this figure came from workers.21 They had had direct and
ongoing contact with both the children and their families. Workers in experimental cases
were more likely to report that families had received services to prevent placement (89.3%)
than workers in control cases (75.0%) (p < .0001). They were also more likely to report that
families had received services in the home after a reunification took place (76.5% versus
50.7%, p < .0001).
22
Table 3. Characteristics of
Indiana
experimental and control children
Experi
-
mental Control
Experi
-
mental Control
Title IV-E Eligible 47.9% 49.3% Case Type at Assignment
Title IV-E Not Eligible 52.1% 50.7% CHINS (Child in Need of Services)
54.7% 61.5%
Male 53.9% 53.7% Informal Adjustment 21.5% 16.2%
Female 46.1% 46.3% Service Referral 3.1% 2.2%
Mean Age (all children) 8.6 yrs.
8.8 yrs.
Service 8.5% 6.8%
Mean Age CPS children 7.7 yrs.
7.7 yrs.
Juvenile Delinquent 12.2% 13.1%
Mean Age of Delinquents 15.48 15.57 Placement History
Household Characteristics Placed at waiver assignment 36.6% 40.4%
Mean household size 5.2 5.1 Placed prior to waiver assign. 54.3% 66.8%
Two or more adult caregivers 64.4% 63.7% 2 or more prior removals 10.3% 9.4%
One adult caregiver 35.5% 36.1% Placed in:
Single mother households 30.6% 30.1% Foster care 65.6% 70.4%
Special Needs Relative care 29.4% 20.0%
Psychological Problems 10.5% 11.5% Institutional care 25.3% 32.0%
Medical Conditions 0.3% 0.3% Correctional facilities 4.2% 6.4%
MR/DD 4.8% 5.2% Other facilities/settings 17.8% 13.4%
Physical Disabilities 2.9% 2.8% One type of facility 59.8% 61.1%
Any Special Need 14.7% 15.7% Two or more types of facilities 40.1% 38.8%
Multiple Conditions 3.3% 3.4%
As can be seen in Figure 6, worker reports of services received were higher for
waiver cases in most categories (25 of 34). Service categories with the strongest differences
were those that addressed financial insecurity or family integrity. Help with basic needs,
housing, homemaking, and family preservation were all provided to waiver families at least
12 percent more often than to matched control families. Assistance with household needs,
such as small purchases for the home or cleaning supplies, was provided nearly 30 percent
more often for waiver families.
Workers also wrote in additional services that were provided to waiver families.
Other services that were mentioned included adoptive and custody related assistance;
specific financial purchases, such as a washer/dryer and car repair; specific therapeutic
services, such as home-based counseling and drug screens; and support services for
children, such as mentoring. The waiver also led to the provision of a greater number of
different services in individual cases. On average, waiver families received 5.4 services
each, while control families received 4.4 services (p < .001).
23
Referrals to community-based services outside the CPS agency jumped as well.
These included many financially-related categories such as: transportation (E: 16.1%, C:
5.2%
3.8%
3.5%
2.3%
6.5%
4.2%
3.2%
8.2%
6.4%
6.5%
3.0%
3.0%
11.2%
12.2%
12.5%
7.7%
13.9%
16.2%
13.7%
14.7%
10.4%
28.1%
17.7%
21.9%
20.7%
34.6%
16.2%
34.9%
30.8%
26.6%
40.3%
14.0%
32.3%
62.7%
2.2%
4.4%
4.5%
5.2%
5.4%
5.5%
5.9%
6.7%
6.9%
8.0%
8.4%
9.1%
9.9%
10.7%
13.6%
15.8%
15.8%
18.6%
20.8%
21.8%
23.1%
24.8%
25.9%
27.1%
28.7%
28.9%
30.6%
32.8%
34.6%
38.8%
41.7%
42.7%
48.1%
59.8%
0% 10% 20% 30% 40% 50% 60% 70%
child transitional living *
respite/crisis nursery
vocational training
family member with disability*
emergency shelter
special education
child support*
recreational
legal assistance
domestic violence
emergency services*
childcare*
independent living skills for child
alcohol abuse treatment
education
other financial assistance*
support groups
medical or dental care
job services*
TANF, food stamps*
housing*
mental health services
money management*
Medicaid or other medical…
transportation*
basic needs to child in placement
basic needs*
drug abuse treatment
marital/group counseling
homemaker*
parenting classes
household needs*
family preservation*
individual counseling
Waiver
Control
Figure
6
.
Types of services or referrals received by experimental and control
families and children
24
10.7%; p = .004), household needs (E: 15.2%, C: 9.0%; p = .001), housing (E: 13.5%, C: 6.7%;
p < .001), other financial assistance (E: 9.0%, C: 3.5%; p < .001), and many others.
Families were also surveyed in this study. While response rates were substantially
lower among families in comparison to workers, family responses were consistent with
those of workers indicating substantial increases in services of various kinds, including
material services and community-based referrals. Experimental families were more likely to
indicate that services received were the kind they really needed and were enough to help.22
Within the experimental group, we compared reports of services of CPS families
receiving food stamps (as a proxy for low income) with families that did not. The latter
amounted to only 30% of responding families (Table 4).
Table 4. Service
levels reporte
d by experimental families in
CPS
cases
receiving and not receiving food stamps
Food stamp
recipients
Other CPS
families
Help in obtaining housing
16.4%
4.0%
p
<
.00
1
Food or clothing
29.3%
10.3%
p
<
.00
1
Money to pay rent
21.4%
5.6%
p
<
.00
1
Appliances/furniture/home repair
18.4%
7.1%
p = .003
Help paying utilities
24.0%
11.9%
p = .005
Help in getting TANF or food stamps
25.7%
6.4%
p
<
.00
1
Help in managing your money
14.1%
7.1%
p = .043
Childcare
or day care
14.8%
6.3%
p = .015
Help with
employment /changing jobs
17.8%
4.8%
p
<
.00
1
Car repair or transportation assistance
17.1%
4.8%
p = .001
Job/ vocational training
6.3%
0.8
%
p = .014
Alcohol or drug treatment
28.3
%
15.1
%
p = .004
Assistance at home
12.5
%
3.2
%
p = .003
Within the experimental group various financially related services were provided
significantly more often to poorer families. This difference was expected and was found
repeatedly in intensive services and flexible spending programs that we evaluated.
Looking at state financial records, we found that overall spending on services averaged
$2,472 per experimental child compared to $708 per control child. Among control children
26.7% received a service purchased by the state compared to 64.3% of experimental children.
Based on this evidence there can be no doubt that experimental children and their
families received more services than corresponding control children.
Outcomes. Outcomes were also tracked. The major goals of the demonstration
were: 1) preventing/reducing out-of-home placements; 2) reducing lengths of stay in out-of-
home care; 3) decreasing the incidence and recurrence of child maltreatment; and 4) enhancing
child and family well-being.
25
We examined removals and out-of-home placements of children after the original
child maltreatment investigation and case opening. Beginning at the time of assignment
and tracking forward until the end of the case (or the end of current data collection), 15.7%
of experimental children had subsequently been removed and placed in out-of-home care
compared to 18.0% of control children, a difference that was statistically significant (p =
.003). Time in out-of-home placement was reduced for waiver children (E: 314 days, C: 427
days; p < .001).
Among closed CPS cases substantially more experimental children (n = 4,076) were
reunified with their families (E: 63.5%, C: 46.9%; p < .001) whereas more control children (n
= 4,177) were not reunified but were adopted into other families (E: 14.2%, C: 30.1%; p <
.001). In addition, more experimental children were placed in guardianship (E: 10.6%, C:
8.2%; p < .001).23 Regarding reunification with family, similar results occurred among
delinquents (315 experimental cases with 75.2% reunification, 439 control cases with 68.3%
reunification (p = .02). A more sophisticated analysis showed that various kinds of family
support, including increased material services, led to higher reunification of children with
their families.24
We found a small but statistically significant difference in new substantiated child
abuse and neglect reports (E: 23.2%, C: 24.3%; p = .045). Because children in the five-year
tracking period had varying opportunity for new reports to occur, a survival analysis was
conducted. In this case, a more powerful method, proportional hazards analysis, was
utilized. It permitted the introduction of risk level and services as covariates in the analysis.
The results showed statistical significance for risk level (p < .001) and the introduction of
any services in the case (p < .001) and experimental-control group membership (p < .001).
The differences in the survival analysis are illustrated graphically in Figure 7. The
first thing to be noticed is that as the risk level increased the number of later
substantiations increased as well (the bars grow in length). Another thing that is obvious
from the table is that when services were delivered the overall levels of substantiations
were greater. This is a common finding within service delivery systems with limited-service
funding. Services are a proxy measure of family need and risk of child maltreatment. The
services in the traditional CPS system tend to go to the most difficult cases, as discussed
earlier regarding the residual approach in child welfare. The most important aspect of the
figure, however, is the reversal that takes place in waiver-control percentages as the eye
moves from no services down to some services. Under the no service category,
experimental children actually had increased proportions of later substantiations in several
risk categories. Under the “some service” category, however, this pattern reverses and
waiver children had fewer investigations.
26
As noted, experimental families were more likely to report that the kinds of
assistance received were what they needed and enough to really help. Relief of these
conditions in families generally has positive effects on the relationships of family members
and in turn on child development. We found that reported school performance improved
among experimental children (p = .037). Waiver families more often reported that their
children were much or somewhat better off because of the experience (E: 75.8%, C: 68.4%;
Figure
7
.
Percent of waiver and
control children with a subsequent substantiated
investigation by risk level and delivery of preventative or remedial services
59.3%
27.3%
35.6%
30.8%
33.9%
29.2%
27.0%
22.4%
18.5%
40.0%
18.2%
39.5%
18.3%
31.0%
26.0%
21.1%
19.6%
12.9%
65.5%
43.1%
46.3%
40.3%
36.4%
35.2%
32.6%
27.1%
23.0%
25.7%
25.2%
32.2%
24.6%
22.5%
24.3%
19.6%
18.8%
15.5%
0% 10% 20% 30% 40% 50% 60% 70%
Highest Risk-8
7
6
5
4
3
2
1
Lowest Risk-0
Highest Risk-8
7
6
5
4
3
2
1
Lowest Risk-0
Some Services No Services
Control
Waiver
27
p = .017) and similarly that their families were much or somewhat better off (E: 74.5%, C:
66.1%; p = .027).
Conclusions Regarding Title IV-E Waiver
The Indiana and Mississippi experiments concerned the highest risk families and
children in the CPS system. They each illustrate the benefits of increased spending for
financially related services and addressing material needs. The welfare and safety of
children were enhanced in each study. Subsequent reports of child maltreatment
decreased significantly in frequency in both projects. The need for removal and placement
of children dropped significantly. Time that children had to remain in foster or residential
care decreased as well. There were indications of greater family satisfaction with services,
improvements in family life, and improvement in school performance of children. Now we
can turn to two other studies within the CPS system.
Studies 3 and 4: The Minnesota and Ohio Differential
Response Experiments
We conducted two other prospective experimental studies that addressed financial
hardship of families. In each family and child welfare and child safety improved. The Ohio
study was essentially a replication of the Minnesota study utilizing a similar design and data
collection methods. For this reason, they will be considered together in this section.
Results in both states were subsequently published in professional journals.25
The Minnesota study began in 2001 and ran through 2005.26 It involved over 5,000
families reported to Child Protection Services (CPS) in 20 Minnesota counties. However, the
experiment was limited to 14 counties that agreed to random assignment of control cases.
For this analysis, 3,861 families are considered: 2,605 experimental families and 1,256
control families. Families were assigned randomly but a weighting was used to keep control
numbers to approximately half the size of experimental.27 Because we were doing other
studies in Minnesota, we continued to receive data on study families. The analysis in this
paper is based on family data from 1999 through 2010.28
The Ohio study began in 2008 and ran through 2013 in 10 Ohio counties.29 Over
4,500 families were involved and tracked. Again, random assignment was utilized with
2,383 families assigned to the experimental group and 2,247 in the control group.
Both studies involved a new approach in CPS that has come to be called Differential
Response (DR). Briefly, the approach excluded families accused of severe physical abuse
and neglect or any allegations of sexual abuse. Thus, compared to the two IV-E studies
28
reviewed, these experiments were focused on lower risk cases, usually involving families
accused of child neglect of various kinds and less severe physical abuse, such as oversevere
discipline.
Under the traditional CPS approach, all families with accepted reports alleging child
maltreatment were subjected to a forensic and adversarial investigation. The term
adversarial refers to investigatory behavior similar to that of law enforcement with the
objective of determining whether the allegations of the report (or any other maltreatment)
actually occurred. The families assigned to the control group received this kind of traditional
investigation. Investigations were concluded by either substantiating maltreatment or not.
Substantiated investigations led to opening and maintaining CPS cases on these families. By
contrast, the experimental families were not investigated. Instead, they received a non-
adversarial family assessment (FA), which included a full child-safety assessment but then
focused on broader family needs. In some instances, ongoing cases were opened on these
families as well, especially if initial assessments of child safety and family needs led to
concerns about the welfare and safety of the children. Please refer to the online reports if
you want greater details about DR. However, there was another dimension to the
experimental treatment in these studies. In each state a private foundation provided extra
funding for services to experimental cases only. The differential response approach
provided flexibility and family involvement in decision-making regarding service needs. The
additional funding permitted more intensive services. In this way the approach in these two
experiments resembled the approach utilized in the IV-E studies.
The random assignment process was successful in both studies. Table 5 illustrates a
select few statistics showing the similarity of the groups. More detailed tables are available
in the two journal articles previously referenced.30
Services. The additional funding led to higher levels of services for experimental
families. The differences are illustrated in Table 6, which shows workers reports of services
provided to experimental and control families. The data for these table came from case-
specific sample surveys of workers responsible for cases. Like surveys described for the IV-E
studies, response rates of workers were high (greater than 85%), with lack of response
attributable to worker turnover and worker leave.
29
As can be seen in Table 6, experimental families received more help in nearly every
service area but much more in material services of various kinds, such as food and basic
household needs (furniture, appliance, clothing, etc.). Also, families were provided cash for
rent, childcare, respite care as well as other kinds of financial assistance. The differences in
some categories were substantial and in many were statistically significant.
Many of these were services directed toward problems of financial hardship. The FA
approach with its emphasis on the underlying needs of families helped FA workers to
discover such problems more often than traditional investigators.
FA workers also had more flexibility to address needs that were much more rarely
addressed under the traditional residual approach. This is not at all surprising. Workers
were confronted with families generally ranging in income from extreme poverty to near
official poverty levels. They had the means to help, so they helped. Analyses showed that
the more the need of the family, the more likely services were provided.
Table 5. Select Demographic and Cases Characteristics of Study
Groups
Minnesota
Ohio
Experimental (E)
–
Control (C)
E
C
E
C
Number
2605
1256
2383
2247
Family Characteristics
Race
Caucasian
71.4%
69.8%
62.2%
63.5%
African American
17.3%
16.2%
24.9%
24.7%
American Indian
3.3%
3.9%
Other or Unknown
8.0%
11.1%
12.4%
11.7%
Persons
Mean number of
adults 2.2 2.3 1.72 1.72
Mean number of
children 2.5 2.5 2.01 2.04
Initial
Allegations
Neglect
58.1%
58.1%
53.0%
53.9%
Physical Abuse
42.2%
42.3%
44.1%
44.0%
30
Table 6. Worker reports of services provided to families in the Minnesota and Oh
io Alternative
Response studies (Case review samples)
Service Categories
Minnesota
Ohio
Control
Exp
e
r
i
-
mental p Control
Experi
-
mental p
Help with rent/house payments
2.4%
11.0%
<= .001
3.5%
9.3%
< .05
Housing services
3.9%
9.5%
< .05
3.9%
7.2%
< .10
Help with basic home needs
2.9%
16.1%
<= .001
6.1%
20.3%
<= .001
Emergency food
0.0%
9.9%
<= .001
1.7%
6.8%
< .05
Assistance with transportation
1.9%
8.8%
<= .001
3.9%
7.6%
TANF, SSI or food stamps
2.9%
7.3%
< .05
2.6%
6.8%
< .05
Medical or dental
care
8.2%
9.5%
2.2%
5.9%
< .05
Assistance with employment
1.9%
5.5%
< .05
2.2%
3.8%
Vocational/skill training
0.5%
4.8%
< .05
0.9%
1.3%
Educational services
5.3%
7.7%
0.9%
2.5%
Legal services
4.8%
6.6%
5.2%
4.7%
Childcare/daycare services
5.3%
12.8%
< .05
2.6%
8.9%
<= .001
Homemaker/home management
1.4%
5.8%
< .05
1.7%
2.1%
Respite care/crisis nursery
3.9%
7.3%
1.3%
1.7%
Emergency shelter
1.9%
2.6%
2.2%
2.1%
Parenting classes
11.6%
18.3%
< .05
4.8%
6.4%
Marital/family/group
counseling
14.5%
18.6%
4.4%
9.3%
< .05
Individual counseling
15.5%
24.2%
< .05
16.2%
18.6%
Mental health/psychiatric services
10.1%
15.3%
9.2%
10.6%
Drug abuse treatment
6.3%
2.9%
5.2%
1.7%
< .05
Alcohol abuse treatment
4.8%
4.0%
2.2%
3.0%
Domestic violence services
9.7%
9.5%
4.8%
2.5%
Assistance from support groups
4.3%
6.6%
1.3%
3.0%
Disability services
1.9%
2.2%
0.4%
0.8%
Recreational services
1.4%
5.9%
< .05
0.9%
2.1%
Family preservation services
3.9%
4.8%
2.6%
1.7%
Total
Families
227
220
207
273
Outcomes. Did these differences in financial and material aid make a difference in
long-term outcomes for families? One important outcome was the percentage of families
in the experimental group that were the subject of later reports of child maltreatment.
Random assignment made for great similarity between the two groups, and other things
being equal, we would expect to see families re-reported at about the same rate. However,
this is not what happened. Subsequent child maltreatment reports and investigations
occurred significantly less often for experimental families. Analysis showed that the risk of
new child abuse and neglect investigations in Minnesota was 28% greater for control
31
Figure 8. Ohio proportional hazards analysis of new accepted
reports of child maltreatment
2. C
ontrol
1. Experimental
families.31 We also looked at subsequent removals and out-of-home placements of children,
which was also significantly lower among experimental families. Subsequent analysis of the
data through 2010, some five years after the original follow-up study showed similar results
for the full samples of families.32 However, in that analysis we were able to show that
material services to both low-risk and high-risk families were effective in reducing later child
maltreatment reports and subsequent child removals.33
Here is an example chart from the Ohio study (Figure 8). The lines in the chart show
the cumulative survival rates, where survival indicates the length of time until a new child
maltreatment report was received (the hazard referenced in the chart title). The higher the
line the better the survival rate.34
Another set of outcomes were indicative. In Ohio, new assessments of child safety and
risk were conducted over a five-year period following the original intervention. In analyzing
subsequent child safety and risk assessments associated with new child maltreatment
reports, we found the following:
Children in experimental families were judged to have received serious inflicted harm less
often.
Children in experimental families were judged to be less often in danger from an adult who
was mentally or physically ill.
Children in experimental families were judged to be less often in danger of neglect, including
lack of supervision, food, clothing or shelter.
Children in experimental families were less often in families that refused access to the child or
were likely to flee.
32
Children in experimental families were less often found in situations of failure to meet their
serious physical or mental health needs.35
Conclusions Regarding the Two DR Studies.
These RCT studies used the strongest experimental design, random assignment,
creating virtually identical groups of families. Significantly more material supports were
provided to families in experimental groups. Outcome analysis indicated that the risk and
actual occurrence of subsequent child abuse and neglect reports and child removals were
substantially reduced over a period of years. The later safety of children, as measured by
child safety assessments, was also demonstrated. They provide strong evidence that
financial assistance to families experiencing financial hardship can significantly improve the
welfare of such families and the safety and welfare of children.
Study 5: The Nevada Differential Response (DR) Evaluation
This study tracked families in a special statewide differential response (DR) program
in Nevada.36 While no extra money was provided for families that entered the program, the
structure of the program ensured that they were provided with substantially more material
(financially related) services than standard cases. Families that entered the program had
been reported to Child Abuse and Neglect county hotlines. Only those in which there were
no children younger than five years could be referred to DR. The types of reports included
educational neglect, environmental neglect, physical or medical neglect, improper
supervision, or inappropriate discipline with non-severe physical harm. Reports of these
kinds, if selected, did not receive a traditional CPS investigation (as described previously)
but in this project were instead referred to a Family Resource Center (FRC) office.
The FRCs were in the business of assisting families in neighborhoods that lacked ‘the
basic necessities of life’ and in which various services were unavailable. FRCs operated as
Nevada’s child welfare agency. The law establishing FRCs stated: “Nevada's most
vulnerable families and children live in these neighborhoods [and] many such families not
only live in poverty, but also experience divorce or are headed by a single parent....
[Furthermore,] children who are raised in such neighborhoods frequently experience
physical and mental abuse.”
Services. The Nevada study did not involve a prospectively selected control group.
Thus, comparisons of services provided to highly similar groups of experimental and control
families, like those in the four previously discussed experiments were not possible.
However, we conducted surveys of workers involved with samples of DR families and other
workers involved with samples of families that were traditionally investigated. These
33
surveys comparing services typically provided by FRCs in comparison to CPS can be found in
the final evaluation report. They show that the FRCs provided substantially more services of
all kinds, including material, than CPS. These included emergency food supplies, assistance
with transportation, help with household needs (such as utilities, furniture, appliances,
etc.), housing, and help with rent or house payments.37 This is shown in Figure 9.
Outcomes. It was not possible to create a contemporaneous control group in the
Nevada study. However, at the end of the study, we were able to create a retrospective
control group. This occurred because several thousand cases that could have been referred
to FRCs were not referred. The selection method matched families that were appropriate
for referral to FRC but were not referred on a variety of characteristics recorded in the state
Figure 9.
Percent of families provided specific services according to DR and CPS workers
0% 5% 10% 15% 20% 25%
Other
Substance abuse treatment
Domestic violence services
Emergency shelter
Help with rent or house payments
Housing services
Vocational/skill training
Legal services
TANF, SSI or food stamps
Disability services
child/daycare services
Assistance with employment
Assistance from support groups
Recreational services
Parenting classes
mental health/psychiatric services
Help with household needs (utilities, repair,…
marital/famiy/group counseling
medical or dental care
Assistance with transportation
Emergency food
Educational services
DR Workers
CPS Workers
34
child welfare data system. There were 1,861 families referred to FRCs during the study for
which full data were available for comparison.38 This constituted the experimental group
for subsequent analysis. Only families with a disposition of investigation were included in
the pool of potential control cases. Initially, two matches were selected for each DR family
from among the pool of investigated cases based on paired similarity. These paired cases
had child maltreatment reports within 60 days (plus or minus) of the matching experimental
family. This group of 3,722 families was submitted to a computer program designed to
produce greater group similarity between experimental and control families. This was
weighted procedure that set families aside in successive steps while comparing
experimental and control groups as wholes at each step.
The most important factor in this type of selection process is risk of future child
maltreatment.39 The control group selected was on average equivalent on these measures:
previous reports leading to an investigation over about eight years before the initial report
in the study (mean per group: E: 1.21, C: 1.26) and previous child removals (mean per
group: E: .33, C: .26). Neither of these differences were statistically significant.
No significant differences were found for most areas of allegations associated with
previous reports: sexual abuse, severe physical abuse, physical abuse, drug-exposed infant,
severe neglect, emotional abuse, medical neglect, unmet medical needs, and lack of
supervision. Experimental families were more likely to have been previously reported for
neglect of basic needs, such as food, clothing and housing (mean reports, E: mean of .75, C:
mean of .57; p = .002) and educational neglect (E: .12, C: .07; p = .005).
Control families on average had slightly more children (E: mean of 2.8 children; C:
mean of 3.2 children; p < .001) but slightly more two-parent households (E: mean of 1.74
parents, C: mean of 1.8 parents; p = .003). The former makes comparison families at slightly
higher risk for new reports while the latter reduces their risk slightly. In addition, control
parents were slightly older. For example, the primary parents/caregivers for experimental
families averaged 35.7 years while the same for control families averaged 36.6 years (p <
.001). While these differences were statistically significant, the sizes of the differences were
small.
There were positive outcomes. Experimental and control families had identical rates
of previous investigations (E: 42.5%, C: 42.4%), but experimental families experienced 8.6%
fewer subsequent investigations (E: 15.7%, C: 24.3%; p < .001). A survival analysis was
conducted in which subsequent investigations of experimental and control families were
compared while controlling the level of past investigations (that is, setting the groups equal
in terms of risk). This showed a statistically significant difference (p < .001). The same kind
of analysis was conducted for any report, whether investigated or not. This analysis also
showed improvement for experimental families (p < .001). The analysis showed a reduced
35
relative risk of new reports of any kind for experimental families of .73. This means that for
every 100 reports on control families 73 were received on experimental families.40
Subsequent removals and placements of children because of new child
maltreatment reports after experimental and control case closings occurred at very low
rates. There were only 10 children removed from 1,861 experimental families (.05%)
compared to 12 children from 1,105 control families (1.1%). The difference in a survival
analysis controlling for past removals was statistically significant (p < .001), but because of
the small numbers, the results should be regarded with caution.
Conclusions Regarding Nevada. This was the weakest of the studies from an
experimental design perspective. Nonetheless, substantial reductions in future reports and
investigations of child maltreatment were found for families provided with FRC services.
Study 6: The Minnesota Parent Support Outreach Program
(PSOP)
The final study we consider was not a controlled experiment. It was an evaluation
study we conducted in Minnesota of the Parent Support Outreach Program (PSOP).41 The
PSOP operated from 2005 to 2008.
In family-oriented field experiments, like the five considered above, control groups
are possible when comparable data can be collected on characteristics of families and
family members for all potential experimental and control cases. In addition, it must be
possible to collect information on members of both groups about program participation and
subsequent changes (outcomes) that take place in families. As will be evident in the
following, collecting these kinds of data on potential control families was virtually
impossible in PSOP study.
On the other hand, the data collected on families that did participate in PSOP was
more consistent and detailed than in the previous five. Furthermore, various correlational
analyses indicated that providing material resources to families experiencing financial
hardship had positive effects on family life and on the safety and welfare of the children.
We hope to show that the PSOP program addressed the problems arising in the Family-
Stress Model and the Economic-Resource Model of the Congers presented above or as we
have combined them in the EHCO model and the ADAR understanding of child welfare.
The PSOP targeted families who had been reported for child maltreatment.
However, in these cases the reports were not accepted by CPS for further action. When
child maltreatment telephone hotline reports are received, intake workers question the
reporters about what they have observed or heard to determine whether the problems
36
being reported correspond to the state statures governing child abuse and neglect.42 For
these families, workers determined that the report did not qualify for further action.
The basic idea underlying PSOP was to contact these families to determine whether
they needed help and solicit their voluntary participation in services. After the first year of
the program the referral process was expanded, permitting families to be referred to the
program from other programs, such as the Minnesota Family Investment Program (MFIP),
Minnesota’s welfare to work program.
The same Minnesota foundation (McKnight) that funded the original Minnesota DR
project referenced above, provided extra money for PSOP. Thirty-eight counties
participated. In each, a target number of families to be served was set. Each county
received $1,000 per target family that could be spent in various ways.
By the conclusion of the study PSOP services had been offered to 7,752
(unduplicated) families of which 3,841 (49.5%) decided to participate. Families with past
reports to CPS and/or open CPS cases were more likely to accept PSOP services (34.7% of
accepters versus 27.7% of decliners). More generally, families that had received services
from other agencies such as the Minnesota Child Welfare program, childcare services,
services related to developmental disabilities of children and various other adult services
tended to accept PSOP services more often.
County program procedures utilizing the funds varied greatly. The analyses of
approaches in the 38 participating counties were instructive.43 One of the most interesting
findings is that contacted families accepted PSOP services substantially more often (64.3%)
in counties that relied on workers from private agencies compared to about half (49.6%) in
counties relying on public CPS agency workers. Rates of acceptance among families with
previous CPS reports and cases was higher in counties utilizing private agency workers.
Nonetheless, the majority of families who participated received services that they would
not have received without the outreach program.
The income status of participating families was shown above in Table 1 of the
present report. The underlying data came from reports of 608 families responding to
surveys after their cases were closed.44 There were other indicators of financial hardship.
Nearly two-thirds of families (65.3%) were receiving food stamps. More than half had
participated in WIC (52.6%). Families who responded to surveys indicated stress about their
financial outlook (A lot: 45.7%, Some: 36.7%).
Workers provided assessment data on families. As noted above (see text before
Table 1), they reported that 59.7% had inadequate incomes or were in poverty. In 13.6%,
workers judged that the poverty was extreme and severe. Workers indicated problems
associated with employment in 48.6% of families served (underemployment: 13.8%,
37
unemployment: 34.8%). They also noted chronic emotional health to be a problem for over
a quarter (28.1%) of an adult in families.
Services. The average (mean) number of contacts made with or on behalf of
families was 16. Four or more face-to-face contacts were made with 54 percent of families
and 11 or more with 18 percent of families. Workers indicated that services were provided
directly in many areas related for financial hardship and poverty (for example: basic
household needs: 28.3%, emergency food: 15.1%, transportation assistance: 18.8%,
medical/dental: 5.3%). In addition, the large majority of families (87.3%) were referred to
at least one community service provider (for example: childcare/Head Start: 30.3%,
emergency food: 28.9%, mental health: 32.8%, community action agency: 14.2%, legal
services: 12.5%, domestic violence shelter: 9.8%, and many others). Families indicated many
areas of such service reception (for example: food or clothing 30.2%, money for rent: 19.1%,
help paying utilities: 15.3%, childcare: 15.0%, housing: 12.7%, appliance/furniture/home
repair: 8.8%, and many others).45
Outcomes. Both workers and families generally agreed that the assistance provided
fit the needs of families and was effective. Among families responding to surveys, 42.5%
indicated that they were much better off because of the PSOP experience and another
36.9% said they were somewhat better off. A minority indicated that they were somewhat
worse off (17.8%). Concerning needs that they were able to address workers responded
that they observed marked improvement in numerous areas. These can be seen in Figure
10. Workers indicated that they observed improvements in family income (25.3%) and
employment (20.9%), in parent-child relationships (14.8%) and parenting skills (15.3%) and
in numerous other areas. Overall, workers noted that marked improvement had occurred
in 62.1% of families for at least one if the issues listed in Figure 10.
We developed an outcome assessment approach that we called the dosage model
(see Figure 11). The model was based on the assumption that in a program of the
magnitude of PSOP, families with similar sets of needs could be found that for different
reasons received variable levels of services addressing those needs. If the services were
effective, then families that received and utilized more (higher doses) might be expected to
have better outcomes than families with less or none. The model was utilized because
service and outcome information were collected in sufficient detail along with views of
workers and family caregivers concerning the benefits and detriments of the PSOP service
approach.
38
The diagram in Figure 11
shows the approach to analysis in
the dosage model. This may be
compared to the approach
encapsulated in the experimental
design as shown in Figure 1.
While three levels of service
participation are shown in the
diagram, the number is arbitrary
and as few as two could be
analyzed as well as many more
than three levels and even
continuous measures across a
range of participation. Needs and services were considered and matched in three critical
areas,: 1) serious basic needs deficiencies and poverty-related services, 2) under-
7.7%
7.0%
4.7%
8.4%
10.1%
5.1%
6.6%
7.4%
6.5%
4.5%
5.6%
9.3%
9.6%
10.7%
10.1%
12.7%
14.8%
10.3%
9.1%
10.6%
15.3%
16.1%
20.9%
25.3%
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70%
Structural condition/safety of home
School attendance of children
Extended family financial support
Cleanliness/order of home
Progress of children in school
Emotional maturity of parent/caregiver
Physical health of children
Control of children
Physical health of parent/caregiver
Substance abuse
Developmental level of children
Mental health of children
Approach to child discipline
Household management skills
Stability/integrity of family as a unit
Financial planning
Parent-child relationship/communication
Quality/stability of adult relationships
Extended family emotional support
Support system of friends and neighbors
Parenting skills of adults
Mental health of parent/caregiver
Employment of any adult
Family income
Addressed & Marked Improvement
Addressed
Figure 10. Family Functioning Issues Addressed while the PSOP Case was Open and
the Proportion in which the Worker Felt there was Marked Improvement
Compare
Service
Participation
Families with a particular need
Full service
participation
Partial service
participation
Little or no service
participation
Compare
R
a
n
g
e
Outcomes
Outcomes
Outcomes
Services designed
to address the
need
Figure 11. The Dosage Model
39
employment/unemployment and assistance with welfare and employment and training, 3)
substance abuse and substance abuse treatment.
Services addressing financial hardship are included in the following list. We
considered these to be the strongest contributors to the measure. The needs associated
with these services are highly interrelated and arise in part from low income and
unemployment.
Rent/house payments
Emergency food
Housing
Basic household needs
Transportation
Employment
Emergency shelter TANF/SSI/FS
We also decided to use these because the component services were offered to more
PSOP clients than any other kinds of services. Families with high scores on this measure
participated in several of these services. Families with lower scores participated in fewer or
none of these services. The emphasis was primarily on participation in services, which, of
course, can only occur after provision of services. There were families that were offered
many services in this list but that had lower scores on this measure because they used the
services at low levels or not at all. This is an important distinction: only families that utilized
services at high levels had high scores on the services measures. The focus was on
something actually delivered to and utilized by a family that might produce a change.
As we have noted, the majority of families accepting PSOP services were in or near
poverty and a subset of these were deeply in poverty. There are, however, various
mitigating factors. For example, families may have low incomes but some financial support
from their extended families or families may receive support from various non-cash
programs. The important consideration for this analysis is not poverty per se but need for
the kinds of services in the preceding list of services. Thus, we decided to use the measure
of needs for basic services that workers completed for each family. For this analysis, the
first two and the last two were each collapsed into a single category. There were 1,541
families (86.0 percent) in the adequate or some problems category and 250 families (14.0
percent) in the serious/chronic basic needs deficiency category.
Families entering PSOP during the first year had two to three years for tracing and
follow-up compared to only a few months for most families entering during the third year.
As we noted in regard to the experimental studies, a survival analysis is the stronger
approach in studies of this kind in which varying time exists for families to experience a
negative outcome—in this case one or more subsequent reports to CPS of child
maltreatment. The following chart (Figure 12) illustrates this analysis. The lines represent
the survival patterns (cumulative survival for each of the four groups over the entire follow-
40
up period (a maximum of approximately 1,150 days). Survival in this case indicates the
proportion of families remaining without a new CPS report. The higher the line the better
the outcome. The difference of interest in this diagram is that between the bottom line
representing the group of families with serious/chronic basic needs with no (or low) basic
services and the other three groups. The difference was statistically significant, as is shown
by the variable listing near the bottom of the figure in which the difference between the last
variable (the bottom line in the graph) and the other variables is statically significant (p =
.05). The analysis indicates that serious or chronic basic needs families did as well as
families with fewer needs when services addressing those needs were utilized and did
significantly worse when such services were not made available or were not utilized. The
latter had significantly more subsequent CA/N reports screened-in to CPS and the reports
were received sooner.
This was a correlational analysis, and alternative explanations of the results are
possible. Dividing the PSOP families into those with adequate or some basic needs versus
those with serious or chronic basic needs certainly has an empirical basis in the judgments
of workers that visited and assisted the families. However, within each of these groups the
division into service groups of low or none versus moderate or high was based in part on
120010008006004002000
Day to a New CPS Report
1.0
0.8
0.6
0.4
0.2
0.0
Cumulative Survival
Figure 12. Survival Analysis (Proportional Hazards) Low and High Basic Needs by
Low and High Poverty
-
Related Services
Serious/chronic needs Basic
services low or none
Serious/chronic needs
Basic services moderate or high
Adequate/Some needs Basic
services low or none
Adequate/Some needs
Basic services moderate or high
41
the choice of families to accept and utilize such services and in part on barriers to services
beyond their control. It is possible that their willingness and their capacities to overcome
barriers reflected other characteristics of families (e.g., attitudes, skills, social support,
environmental) that explain to some extent the differences observed.
A similar analysis was conducted based on the much smaller sample of families that
responded to the family survey. We divided the families into four groups based on
combining those with incomes above and below the $10,000 threshold who received or did
not receive poverty related services. We conducted a proportional hazards analysis for
these four groups. We controlled for the social isolation of the family and the quality of the
neighborhood in which they lived but found no effects on either of these variables on report
recurrence. Satisfaction with their PSOP worker was also analyzed and this was associated
with a statistical trend (p = .08): families more satisfied with their worker returned less
often. Controlling for these three variables a significant relationship was found between
poverty-related services and later reductions in subsequent child maltreatment reports (p =
.04). The most impoverished families with the least services fared the worst. They were
more likely to have new reports and to have them more quickly. Those with higher
incomes, who participated in no services fared best. The other two other groups fell in
between.
Similar analyses were conducted for employment status and employment related
services as well as for substance abuse in the family and substance abuse treatment
services. Both these analyses were concluded with similar results.46
Summary and Conclusions
We have examined six studies. Five were experimental. Of these three involved
Randomized Control Trials (RCT) and the other two involved matching, one prospective in
nature and the other retrospective. The sixth was a correlational study included because it
involved detailed information from workers about characteristics of families and the types
of services delivered to them.
The programs being evaluated in each case involved substantial increased in
material services for experimental families and greater flexibility in the application of those
services, particularly greater participation of families in the decision-making process.
Material services refer to financially related services, that is, more money was available in
each case to make sure that effects of financial hardship and poverty were addressed.
These were the major difference in the experimental studies between experimental and
control groups. Thus, any differences observed during years of follow-up on families was
almost certainly due to this experimental treatment.
42
The studies were large, each involving thousands of families. The follow-up periods
were long consisting of years for the majority of cases. These two factors add to the
strength and validity of study findings.
In each study the levels and types of services that constituted the experimental
treatment were measured in various ways. In some cases, state data systems were used to
view services delivered. In others the reports of workers and families concerning level and
types of services were collected.
Positive outcomes occurred in each study.
Concerning the safety and welfare of children, each of the studies demonstrated
statistically significant reductions in child maltreatment reports that must be
attributed to the financial assistance that was offered.
o In several cases, the level of financial assistance was shown to produce
stronger effects, that is, reductions in subsequent reports were greater
among families receiving higher levels of material services.
In two studies in which the safety of children was such that imminent removal
from their homes was possible significantly reduced out-of-home placement was
subsequently observed.
o In addition, length of stays of children in out-of-home placement were
reduced among children in placement.
Statistically significant reductions in subsequent removal and out-of-home
placement of children were also observed in four of the studies.
In one study, evidence of improved family relations and general child welfare
was found, including parents reports of improved child development and school
performance along with a sense that their children were better off.
One of the RCT studies considered subsequent safety assessments of children
and found:
o Children in experimental families were judged to have received serious
inflicted harm less often.
o Children in experimental families were judged to be less often in danger from
an adult who was mentally or physically ill.
o Children in experimental families were judged to be less often in danger of
neglect, including lack of supervision, food, clothing or shelter.
o Children in experimental families were less often in families that refused
access to the child or were likely to flee.
o Children in experimental families were less often found in situations of
failure to meet their serious physical or mental health needs.
43
These studies validate the Economic Hardship-Child Outcome (EHCO) model of child
welfare. The EHCO model shows that financial hardship and poverty are implicated in child
abuse and neglect and more generally in child welfare problems and are manifested in two
ways:
1. Reports of child maltreatment that are in reality reports of the effects of poverty,
and
2. Reports of child maltreatment in which financial hardship is a moderating cause
that exacerbates neglectful and abusive behavior of caregivers toward children.
In each of these types of reports relieving financial deprivation led to improvements
in the welfare and safety of children.
Implications. The implication of this research is that income maintenance and child
protection services should be recombined. This was the approach some 60 years ago
before the introduction of the Child Abuse Prevention and Treatment Act (CAPTA) in 1972,
when these functions were separated. Workers before those changes were responsible for
both functions. When we first began our differential response studies in Missouri in the
1990s, there were still older workers around that remembered the pre-1972 period. They
often commented the DR felt like a return to those days when they were concerned with
family welfare generally as well as protecting children.
In addition, these studies suggest that financial resources controllable by CPS
workers should be expanded to enable financial hardship and poverty to be addressed in
families in the less risky end of the current CPS spectrum. Services should definitely
continue to address particular child safety issues (for example, reducing the use of corporal
punishment, improvement parenting skills generally, and the like) and removal of children
when their lives and health are seriously threatened. However, they should also focus on
more basic areas of family needs that are found in large proportions of CPS caseloads, such
as housing safety and cleanliness, rent assistance, adequate food and clothing, household
items, appliances, transportation assistance, childcare, respite care, and many similar
needs and services.
A just rejoinder to this is that CPS and income maintenance are targeted programs.
A much broader set of families have similar needs who were never reported to CPS and will
never be reported. Yet they also would likely benefit from similar assistance. We agree and
we support such general initiatives as a Federal Jobs Guarantee and Universal Basic Income.
(See our second summary paper on financial hardship and child welfare more generally.)47
We would only say the following regarding CPS:
1. CPS agencies are one of the most broadly distributed agencies in the U.S. with
workers in virtually every city and county in the country.
44
2. Hundreds of thousands of families are reported to CPS each year in the United
States. Most reports are not accepted for further action and of those that are
only the most extreme receive any assistance, and often that assistance does not
address the kinds of needs we have examined in this paper.
3. Expanding the resources available to workers and permitting greater flexibility in
decision-making (with families) and in the kinds of needs that can be legitimately
addressed could have immense benefits, significantly improving the welfare of
U.S. children.
References
Conrad A., Gamboni C., Johnson V., Wojciak A. S. & Ronnenbery M. (2020). Has the US Child
Welfare System Become an Informal Income Maintenance Programme? A Literature
Review. Child Abuse Review, (wileyonlinelibrary.com) DOI: 10.1002/car.2607.
Conger R. & Conger K. (2008). Understanding the Processes through which Economic
Hardship Influences Families and Children. Chapter 5 in Handbook of Families and
Poverty, ed. Crane, D., & Heaton, T. Los Angeles: Sage Publications.
Cooper K. & Stewart K. (2017). Does Money Affect Children’s Outcomes? An Update.
London: Centre for Analysis of Social Exclusion, London School of Economics.
Available at: https://www.jrf.org.uk/report/does-money-affect-
children%E2%80%99s-outcomes
Coulton C. J., Korbin J. E., Su M. & Chow J. (1995). Community level factors and child
maltreatment rates. Child Development, 66, 1262-1276.
Coulton C. J., Korbin J. E. & Su M. (1999). Neighborhoods and child maltreatment: A multi-
level study. Child Abuse and Neglect, 23(11), 1019-1040.
Drake B. & Pandey S. (1996). Understanding the relationship between neighborhood
poverty and specific types of child maltreatment. Child Abuse and Neglect, 20(11),
1003-1018.
Duncan G.J., Magnuson K., Votruba-Drzal E. 2014. Boosting family income to promote child
development. The Future of Children, 24(1): 99–120. Available at:
https://escholarship.org/uc/item/5rv2k936
Landers A. L., Carrese D. H. & Spath R. (2019). A Decade in Review in Social Work Literature:
The Link between Poverty and Child Maltreatment in the United States. Child
Welfare, 97, 4, pp 67-96.
Lindsey D. (1994). The Welfare of Children. New York: Oxford University Press.
Loman L. A. & Siegel G. L. (2004). Minnesota Alternative Response Evaluation: Final Report.
Institute of Applied Research. Available at:
http://www.iarstl.org/papers/ARFinalEvaluationReport.pdf
45
Loman L. A. (2006). Families Frequently Encountered in Child Protection Services: A Report
on Chronic Child Abuse and Neglect. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/FEfamiliesChronicCAN.pdf
Loman L. A., Filonow C. S. Sapakaite L. & Siegel G. L. (2009). Minnesota Parent Support
Outreach Program Evaluation. Institute of Applied Research. Available at
http://www.iarstl.org/papers/PSOPFinalReport.pdf.
Loman L. A., Filonow C. S. & Siegel G. L. (2010). Ohio Alternative Response Evaluation: Final
Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/OhioAREvaluation.pdf
Loman L. A., Filonow C. S. & Siegel G. L. (2011). Indiana IV-E Child Welfare Waiver Extension:
Final Evaluation Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/Indiana%20IV-E%20Final%20Evaluation%20Report-
2011.pdf.
Loman L. A. & Siegel G. L. (2012). Effects of anti-poverty services under the differential
response approach to child welfare. Children and Youth Services Review, 34, 1659–
1666.
Loman L. A. & Siegel G. L. (2014). Ohio Alternative Response Evaluation Extension: Final
Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/OhioARFinalExtensionReportFINAL.pdf
Loman L. A. & Siegel G. L. (2015). Effects of approach and services under differential
response on long term child safety and welfare. Child Abuse and Neglect, 39, 86-97.
Available at:
https://www.sciencedirect.com/science/article/pii/S0145213414002099
Pelton L. H. (2015). The continuing role of material factors in child maltreatment and
placement. Child Abuse and Neglect, 41, 30-39.
Saar-Heiman Y. & Krumer-Nevo, M. (2020). Redistribution and Recognition in Social Work
Practice: Lessons Learned from Providing Material Assistance in Child Protection
Settings. American Journal of Orthopsychiatry, Online 12-7-2020.
http://dx.doi.org/10.1037/ort0000525
Sapokaite L., Filonow C. S. & Siegel G. L. (2010). Case Studies, The Use of Flexible Funding in
Indiana’s Child Welfare Demonstration, Institute of Applied Research. Available at:
http://www.iarstl.org/papers/IN-
IVE%20Special%20Project%20Case%20Studies%20Report-final.pdf
Sedlak A. J., Mettenburg J., Basena M., Petta I., McPherson K., Greene A. & Li S. (2010).
Fourth national incidence study of child abuse and neglect (NIS-4): Report to
Congress. Washington, DC: U.S. Department of Health and Human Services,
Administration for Children and Families. Available at:
46
https://www.acf.hhs.gov/sites/default/files/opre/nis4_report_exec_summ_pdf_jan
2010.pdf
Siegel G. L. & Loman L. A. (2005). State of Mississippi Title IV-E Child Welfare Waiver
Demonstration Project. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/MSIVEFinalReport.pdf
Siegel G. L. & Loman L. A. (2006). Extended Follow-up Study of Minnesota’s Family
Assessment Response: Final Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/FinalMNFARReport.pdf
Siegel G. L., Filonow C. S. & Loman L. A. (2010). Differential Response in Nevada: Final
Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/NevadaDRFinalReport.pdf
Slack K. S., Berger L. M. & Noyes J. L. (2017). Economic Causes and Consequences of Child
Maltreatment. Children and Youth Services Review, 72, 1-150.
Yang M. (2015). The effect of material hardship on child protective service involvement.
Child Abuse & Neglect, 41, pages 113-125.
1 Aislinn Conrad, Casey Gamboni, Victoria Johnson, Armeda Stevenson Wojciak & Megan Ronnenbery. (2020).
Has the US Child Welfare System Become an Informal Income Maintenance Programme? A Literature Review.
Child Abuse Review, (wileyonlinelibrary.com) DOI: 10.1002/car.2607.
2 Ashley L. Landers, Domenica H. Carrese & Robin Spath. (2019). A Decade in Review in Social Work Literature:
The Link between Poverty and Child Maltreatment in the United States. Child Welfare, 97, 4, pp 67-96.
Another very recent summary study in this area is Yuval Saar-Heiman & Michal Krumer-Nevo. (2020).
Redistribution and Recognition in Social Work Practice: Lessons Learned from Providing Material Assistance in
Child Protection Settings. American Journal of Orthopsychiatry, Online 12-7-2020.
http://dx.doi.org/10.1037/ort0000525
3 Here is the web access address: https://www.sciencedirect.com/journal/children-and-youth-services-
review/vol/72/suppl/C. Unfortunately the articles are not open access. Universities and many public libraries
offer access. Otherwise readers have to purchase them. Many of us would like to see academic and
professional publishing taken out of the hands of for-profit businesses. Authors are paid no money for these
sales, only the publishers. They should be free to the public. Here are the titles of the articles in this volume:
Child welfare involvement and contexts of poverty: 1. The role of parental adversities, social networks, and
social services. 2. Economic predictors of child maltreatment in an Australian population-based birth cohort.
3. The influence of concrete support on child welfare program engagement, progress, and recurrence. 4. Out-
of-home placement and regional variations in poverty and health and social services spending: A multilevel
analysis. 5. Intersections of individual and neighborhood disadvantage: Implications for child maltreatment.
6. Pathways of risk and resilience between neighborhood socioeconomic conditions and parenting. 7. Money
matters: Does the minimum wage affect child maltreatment rates? 8. The Great Recession and risk for child
abuse and neglect. 9. The effect of lowering welfare payment ceilings on children's risk of out-of-home
placement. 10. The impact of income on reunification among families with children in out-of-home care. 11.
Making parents pay: The unintended consequences of charging parents for foster care. 12. The effect of
47
monthly stipend on the placement instability of youths in out-of-home care. 13. The potential educational
benefits of extending foster care to young adults: Findings from a natural experiment. 14. Employment
outcomes of young parents who age out of foster care. 15. Adverse childhood experiences and life
opportunities: Shifting the narrative.
4 Andrea J. Sedlak, Jane Mettenburg, Monica Basena, Ian Petta, Karla McPherson, Angela Greene & Spencer Li.
(2010). Fourth national incidence study of child abuse and neglect (NIS-4): Report to Congress. Washington,
DC: U.S. Department of Health and Human Services, Administration for Children and Families. Available at:
https://www.acf.hhs.gov/sites/default/files/opre/nis4_report_exec_summ_pdf_jan2010.pdf
5 L. Anthony Loman, Christine Shannon Filonow & Gary L. Siegel. (2010). Ohio Alternative Response
Evaluation: Final Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/OhioAREvaluation.pdf and the extended follow-up: L. Anthony Loman & Gary L.
Siegel. (2014). Ohio Alternative Response Evaluation Extension: Final Report. Institute of Applied Research.
Available at: http://www.iarstl.org/papers/OhioARFinalExtensionReportFINAL.pdf
6 The full PSOP report is available at http://www.iarstl.org/papers/PSOPFinalReport.pdf. The third chapter
(pages 25-40) contains a variety of charts showing characteristics of families from state data systems, worker
assessments and family feedback. Those interested might also examine Figure 3.6 (page 32), which show
participation in financial assistance programs.
7 Claudia J. Coulton, Jill E. Korbin, Marilyn Su & Julian Chow. (1995). Community level factors and child
maltreatment rates. Child Development, 66, 1262-1276. Brett Drake & Shanta Pandey. (1996). Understanding
the relationship between neighborhood poverty and specific types of child maltreatment. Child Abuse and
Neglect, 20(11), 1003-1018. Claudia J. Coulton, Jill E. Korbin & Marilyn Su. (1999). Neighborhoods and child
maltreatment: A multi-level study. Child Abuse and Neglect, 23(11), 1019-1040.
8 Leroy H. Pelton. (2015). The continuing role of material factors in child maltreatment and placement. Child
Abuse and Neglect, 41, 30-39.
9 Rand Conger & Katherine Conger. (2008). Understanding the Processes through which Economic Hardship
Influences Families and Children. Chapter 5 in Handbook of Families and Poverty, ed. Crane, D., & Heaton, T.
Los Angeles: Sage Publications.
10 Mi-Youn Yang. (2015). The effect of material hardship on child protective service involvement. Child Abuse
& Neglect, 41, pages 113-125.
11 L. Anthony Loman. (2021). Relieving Financial Hardship Improves Child Welfare. St. Louis, MO: IAR
Associates. Available at: http://www.iarstl.org
12 Duncan G.J., Magnuson K., Votruba-Drzal E. 2014. Boosting family income to promote child development.
The Future of Children, 24(1): 99–120. Available at: https://escholarship.org/uc/item/5rv2k936
13 The full research report can be found at: http://www.iarstl.org/papers/Indiana%20IV-
E%20Final%20Evaluation%20Report-2011.pdf. This case is taken from a second final report on the project:
Lina Sapokaite, Christine Shannon Filonow & Gary L. Siegel. (2010). Case Studies, The Use of Flexible Funding
in Indiana’s Child Welfare Demonstration. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/IN-IVE%20Special%20Project%20Case%20Studies%20Report-final.pdf The
primary work on this report was conducted by two talented researchers: Lina Sapokaite (now Millett) and
Christine Filonow and finally edited by Gary Siegel. Many of the cases in the report are more complex with
multiple family problems, often including drug and alcohol abuse and emotional and other psychological
problems.
48
14 Our analysis of chronic abuse and neglect shows this. See: L. Anthony Loman. (2006). Families Frequently
Encountered in Child Protection Services: A Report on Chronic Child Abuse and Neglect. Institute of Applied
Research. Pages 43ff. Available at: http://www.iarstl.org/papers/FEfamiliesChronicCAN.pdf.
15 Duncan Lindsey. (1994). The Welfare of Children. New York: Oxford University Press.
16 Gary L. Siegel & L. Anthony Loman. (2005). State of Mississippi Title IV-E Child Welfare Waiver
Demonstration Project. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/MSIVEFinalReport.pdf
17 This was a simple life-table analysis. It involved no statistical controls. The analysis is described and shown
graphically in the original report (Siegel and Loman, ibid., pages 64-65). Survival analysis as applied to the
recurrence of reports of child abuse and neglect is concerned with the period of time until a new report
occurs. The main problem that arises in most evaluations of time-to-a-terminal-event is that tracking of cases
is cutoff at the end of the study. Cases that do not experience the event before the study ends are called
censored cases in survival analysis. We assume that some of the censored cases would have been observed to
experience a report had data collection continued, but we cannot know which. Nonetheless, all cases, both
censored and uncensored are used in computing life tables. Life tables use a particular technique to
determine the number of cases exposed to risk of the terminal event while at the same time considering the
censored cases. In this study, we determined how many families were exposed to the risk of new CA/N
reports while also considering families tracked to the very end of data collection with no new reports. In
constructing a life table, the tracking time is divided into fixed intervals. For example, in this analysis 20-day
intervals were chosen. The maximum tracking time was 45 months, which means there were 67 such intervals
to consider. The rate of new reports in each interval can be thought of as a probability, with a value ranging
from 0 (no terminal events) to 1 (every case experienced a terminal event). The probabilities can be
accumulated until at the end of the last time interval to give a total probability. Finally, in an experimental
study, separate life tables can be constructed for the experimental and control groups. Then the survival
times of cases in the experimental group can be compared to those in the control group to see if, as a whole,
they are different. If the overall difference is great enough to be unlikely to have occurred by chance, we can
assert that the experiment was a success.
18 L. Anthony Loman, Christine Shannon Filonow & Gary L. Siegel. (2011). Indiana IV-E Child Welfare Waiver
Extension: Final Evaluation Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/Indiana%20IV-E%20Final%20Evaluation%20Report-2011.pdf
19 Children assigned to the waiver were pair-matched on a monthly basis with children not assigned. Care was
taken that control-group children were never a part of past or present families assigned to the waiver. The
following 12 characteristics were utilized. Weights were assigned to each of the 12 variables (shown in
parentheses). Summated scores were developed for each newly assigned waiver child and for all non-waiver
children. For each waiver child, the child with the highest matching weighted score from the potential match
pool was then assigned as a match. Weighting: 1. IV-E eligibility (100); 2. Case Type, including service
(voluntary, court request, court ordered), service (adoption, AG, DOC, IL), SRA, IA, CHINS, and Delinquent (64);
3. Special Needs: psychological, medical, developmental disabilities and disabilities (64); 4. Case Begin Data
(50); 5. Placement / Removal status (36); 6. Case County (32); 7. Age (16); 8. Physical Abuse (12); 9. Sexual
Abuse (12); 10. Neglect (8); 11. Gender (8); 12. Number of Caregivers (8).
20 The evaluators were responsible for the cost-neutrality calculations during the entire demonstration period
and regularly demonstrated to Federal authorities that indeed no excess spending occurred during the
demonstration.
49
21 Random samples of cases were selected as they closed and workers in the cases were surveyed. The survey
instrument was completed online. Response rates were high (> 85%). Exceptions were workers who had left
the agency or were on leave.
22 The full table of community-based services can be found on page 99 of the previously cited final report: L.
Anthony Loman, Christine Shannon Filonow & Gary L. Siegel. (2011). Indiana IV-E Child Welfare Waiver
Extension: Final Evaluation Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/Indiana%20IV-E%20Final%20Evaluation%20Report-2011.pdf
The full listings of reports from workers and families concerning direct and indirect services from community-
based providers can be found pages 94-113 The full outcome/impact analysis is on pages 114-140.
23 For those interested in experimental design, the substantially higher rates of adoption might indicate a bias
in the matching method, since the child and family characteristics underlying this alternative could not be
taken into account. However virtually no delinquent child (n = 1) was adopted, which argues against this
interpretation.
24 This factorial analysis can be found on page 125 to 127 of the previously cited Loman, Filonow & Siegel 2011
report. Pay particular attention to the graphic representation of the interaction effect between study group
and family support on reunification (p 127).
25 Long-term analysis of the Minnesota DR project: L. Anthony Loman & Gary L. Siegel. (2012). Effects of anti-
poverty services under the differential response approach to child welfare. Children and Youth Services
Review, 34, 1659–1666. Five-year follow-up of the Ohio DR project: L. Anthony Loman & Gary L. Siegel.
(2015). Effects of approach and services under differential response on long term child safety and welfare.
Child Abuse and Neglect, 39, 86-97. Available at:
https://www.sciencedirect.com/science/article/pii/S0145213414002099
26 The original final report was : L. Anthony Loman and Gary L. Siegel. (2004). Minnesota Alternative Response
Evaluation: Final Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/ARFinalEvaluationReport.pdf . There was also a two-year follow-up report: Gary
L. Siegel and L. Anthony Loman. (2006). Extended Follow-up Study of Minnesota’s Family Assessment
Response: Final Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/FinalMNFARReport.pdf
27 This was necessary to ensure that the 14 participating counties assigned sufficient numbers of children to
the waiver so as to collect all foundation monies available. Counties received a set amount for each waiver
child assigned. However, the received funds could be spent on any currently active waiver case.
28 This is the analysis recounted in the above cited article: L. Anthony Loman & Gary L. Siegel. (2012). Effects of
anti-poverty services under the differential response approach to child welfare. Children and Youth Services
Review, 34, 1659–1666.
29 The original final report: L. Anthony Loman, Christine Shannon Filonow & Gary L. Siegel. (2010). Ohio
Alternative Response Evaluation: Final Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/OhioAREvaluation.pdf . The extended follow-up report: L. Anthony Loman &
Gary L. Siegel. (2014). Ohio Alternative Response Evaluation Extension: Final Report. Institute of Applied
Research. Available at: http://www.iarstl.org/papers/OhioARFinalExtensionReportFINAL.pdf
30 Minnesota: L. Anthony Loman & Gary L. Siegel. (2012). Effects of anti-poverty services under the differential
response approach to child welfare. Children and Youth Services Review, 34, 1659–1666. Ohio: L. Anthony
Loman & Gary L. Siegel. (2015). Effects of approach and services under differential response on long term child
50
safety and welfare. Child Abuse and Neglect, 39, 86-97. Available at:
https://www.sciencedirect.com/science/article/pii/S0145213414002099
31 This analysis comes from the MN Follow-Up report. Gary L. Siegel and L. Anthony Loman. (2006). Extended
Follow-up Study of Minnesota’s Family Assessment Response: Final Report. Institute of Applied Research,
pages 30-1. Available at: http://www.iarstl.org/papers/FinalMNFARReport.pdf
32 L. Anthony Loman and Gary L. Siegel. (2004). Minnesota Alternative Response Evaluation: Final Report.
Institute of Applied Research, page 143. Available at:
http://www.iarstl.org/papers/ARFinalEvaluationReport.pdf ‘Hazard’ refers to the relative risk of an undesired
outcome, which in this case refers to a child removal.
33 Minnesota: L. Anthony Loman & Gary L. Siegel. (2012). Effects of anti-poverty services under the differential
response approach to child welfare. Children and Youth Services Review, 34, 1659–1666. For this study, the
Cox Proportional Hazards analysis, discussed in reference to Table 3 demonstrated the effect of such services.
34 L. Anthony Loman, Christine Shannon Filonow & Gary L. Siegel. (2010). Ohio Alternative Response
Evaluation: Final Report. Institute of Applied Research, page 138. Available at:
http://www.iarstl.org/papers/OhioAREvaluation.pdf
35 L. Anthony Loman & Gary L. Siegel. (2014). Ohio Alternative Response Evaluation Extension: Final Report.
Institute of Applied Research, page 34. Available at:
http://www.iarstl.org/papers/OhioARFinalExtensionReportFINAL.pdf These findings are also discussed in a
subsequent journal article: L. Anthony Loman & Gary L. Siegel. (2015). Effects of approach and services under
differential response on long term child safety and welfare. Child Abuse and Neglect, 39, 86-97. Available at:
https://www.sciencedirect.com/science/article/pii/S0145213414002099
36 Gary L. Siegel, Christine Shannon Filonow & L. Anthony Loman. (2010). Differential Response in Nevada:
Final Report. Institute of Applied Research. Available at:
http://www.iarstl.org/papers/NevadaDRFinalReport.pdf
37 Ibid. pages 78-84.
38 A total of 1,903 unduplicated families were referred. Of these, incomplete data were available for 52
families from the Nevada SACWIS system (UNITY) as provided to evaluators.
39 In this study we were not dealing with the general population but with families already identified in the
state CPS system. So, the question was whether we could identify families that were likely to reappear in the
system at a later time. In this sense, risk refers to the probability of being reported again for child
maltreatment. What characteristics of families are associated with high or low risk? There are certain
demographic characteristics such as the caregiver’s age, the number of children in the home, single-parent
status that are associated with repeat child maltreatment. Note that each of these characteristics is
associated with financial stress. However, it turns out that key variables in determining risk of being re-
reported are those showing past involvement with the CPS system. Thus, a family in the system currently who
has had several previous child maltreatment reports, after leaving the system this time, is more likely to
reappear than a family who had no previous contacts. Thus, the number of past child maltreatment reports,
past investigations and previous removals of children to be placed out-of-home are risk measures.
We examined standard risk measures extensively in another Minnesota study: L. Anthony Loman & Gary L.
Siegel. (2004). An Evaluation of the Minnesota SDM Family Risk Assessment: Final Report. Institute of Applied
Research. Available at: http://www.iarstl.org/papers/FinalFRAReport.pdf Relevant analyses can be found on
pages 11-24. We recommend reading the discussion on page 20 of that report. In the risk scale studied, the
strongest predictor item of future child maltreatment reports was the number of past reports. The discussion
51
of services on pages 60ff considers the effects of services. Traditionally, the most services were offered to the
highest risk families. But the analysis in this report shows how services to low-risk families reduce
maltreatment report recurrence.
40 This analysis is highly detailed in the full report. Ibid., pages 121-126. While experimental and control
families had virtually identical rates of past investigations, experimental families had significantly more past
reports that included those that were not investigated. Looking at reports of neglect of basic needs (food,
clothing, shelter), experimental families had received many more over about eight years in the past (E: 75 of
every 100 families, C: 57 of 100) but these kinds of reports subsequently received during the follow-up period
were reduced to equivalent levels (E: 15 per 100, C: 13 per 100; p = ns).
41 L. Anthony Loman, Christine Shannon, Lina Sapakaite & Gary L. Siegel. (2009). Minnesota Parent Support
Outreach Program Evaluation. Institute of Applied Research. Available at
http://www.iarstl.org/papers/PSOPFinalReport.pdf.
42 Examples of reports that do not are: accidents that clearly did not involve parental negligence or malice;
family situations that are not dangerous enough to pose safety hazards for children; second-hand knowledge
of abuse or neglect incidents; crimes that are real (such as rape) that do not involve neglect or are not
perpetrated by relative or family members. There are many others. The acceptance of reports varies
significantly in the states we have studied. In some states very few reports are rejected, leaving decisions
about child safety to the judgment of CPS investigators and assessment workers. In others, like Minnesota,
intake workers are more likely to make that decision and larger numbers of reports are not accepted.
43 Pages 9-24 of the previously cited PSOP report (note 38).
44 Like many of our surveys of families (both telephone and mail) the response rate was low (15.8%). Worker
response rates were much higher, however, since they completed family strengths and needs (FSN)
assessment instruments (along with child safety scales and an evaluator created extension of the FSN
instrument) on most accepting families (2,624 cases by the conclusion of the study). That information was
available for the study. In the appendix of the PSOP report (cited above in note 38, pages 91ff) we conducted
a comparative analysis considering any biases in the family survey sample. The analysis indicated relatively
small differences between descriptions provided by responding families compared to those given by workers
for the entire population of PSOP accepting families. A difference that was found indicated a slight bias
among responders toward families that were slightly better off in terms of basic needs and social support.
45 See the above cited final report (note 38). Extensive discussions can be found on pages 45-51, followed by
several case examples.
46 These analyses can be found on pages 78-83 in the above cited report.
47 In case you cannot read the second summary report, a paper we found very useful in that summary is: Kerris
Cooper & Kitty Stewart. (2017). Does Money Affect Children’s Outcomes? An Update. London: Centre for
Analysis of Social Exclusion, London School of Economics. We highly recommend the paper. It cannot be read
without coming away convinced of the effects of financial hardship on child welfare generally and the power
of relieving those conditions. Available at: https://www.jrf.org.uk/report/does-money-affect-
children%E2%80%99s-outcomes
