Content uploaded by Emily Putnam-Hornstein
Author content
All content in this area was uploaded by Emily Putnam-Hornstein
Content may be subject to copyright.
http://cmx.sagepub.com/
Child Maltreatment
http://cmx.sagepub.com/content/18/4/232
The online version of this article can be found at:
DOI: 10.1177/1077559513508001
2013 18: 232 originally published online 11 October 2013Child Maltreat
Bryn King, Jennifer Lawson and Emily Putnam-Hornstein
Predictors of Maltreatment Substantiation
Examining the Evidence: Reporter Identity, Allegation Type, and Sociodemographic Characteristics as
Published by:
http://www.sagepublications.com
On behalf of:
American Professional Society on the Abuse of Children
can be found at:Child MaltreatmentAdditional services and information for
http://cmx.sagepub.com/cgi/alertsEmail Alerts:
http://cmx.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://cmx.sagepub.com/content/18/4/232.refs.htmlCitations:
What is This?
- Oct 11, 2013OnlineFirst Version of Record
- Oct 17, 2013OnlineFirst Version of Record
- Oct 25, 2013Version of Record >>
at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from at UNIV OF SOUTHERN CALIFORNIA on November 4, 2013cmx.sagepub.comDownloaded from
Regular Article
Examining the Evidence: Reporter Identity,
Allegation Type, and Sociodemographic
Characteristics as Predictors of
Maltreatment Substantiation
Bryn King
1
, Jennifer Lawson
1
, and Emily Putnam-Hornstein
1,2
Abstract
Using linked administrative data from child protection and birth records in California, this study examined whether the mandated
status and type of reporter are independent predictors of substantiation among infants and young children across maltreatment
types and after adjusting for characteristics of the child and family. Of the 59,413 children born in 2002 who were reported and
investigated for maltreatment before the age of 5 years, 26% were substantiated. Reports originating from mandated sources
were 2.5 times as likely (95% confidence interval, CI [2.40, 2.60]) to be substantiated as those from nonmandated reporters.
Findings demonstrated that children whose allegations were reported by law enforcement, medical professionals, and workers
in public agencies were consistently substantiated at higher rates than allegations from other mandated reporters. Results also
indicated that the relationship between reporter type and the likelihood of substantiation varied by maltreatment type.
Children reported by law enforcement for physical abuse were 6.3 times as likely (95% CI [4.86, 8.04]) to be substantiated as
those reported by nonmandated sources.
Keywords
child maltreatment, mandated reporters, child protective services, substantiation, mandatory reporting laws
The adoption of mandatory child abuse reporting requirements
beginning in the 1960s has expanded the reach of child welfare
agencies, resulting in an increased number and rate of children
reported to child protective service (CPS) agencies (Besharov &
Laumann, 1996; Kesner, 2008). In 2011, child welfare agen-
cies in the United States investigated more than 2 million
reports of abuse and neglect, and 18.5%of those were sub-
stantiated for maltreatment (U.S. Department of Health and
Human Services [USDHHS], 2012). Mandated professional
reporters are the largest reporting source, constituting 58%
of all reports nationally (USDHHS, 2012). Previous research
has demonstrated that a higher proportion of reports initiated
by mandated professionals are substantiated than those initi-
ated by nonmandated reporters (English, Marshall, Coghlan,
Brummel, & Orme, 2002; McDaniel, 2006).
Given the potential implications of a substantiated report of
maltreatment for both children and families, an exploration of
factors that may influence the CPS system’s decision to sub-
stantiate a child as a victim of maltreatment is warranted.
Research has indicated that children who have been maltreated,
particularly younger children, may experience substantial
adverse developmental (Shonkoff, Richter, van der Gaag, &
Bhutta, 2012), social (Currie & Widom, 2010), emotional
(Kaplow & Widom, 2007), and physical health (Lanier,
Jonson-Reid, Stahlschmidt, Drake, & Constantino, 2010)
consequences, including death (Jonson-Reid, Chance, & Drake,
2007; Putnam-Hornstein, 2011). In this study, we used linked
administrative data to prospectively follow a population-based
cohort of children reported and investigated for maltreatment
by the age of 5 years. We focused our analysis on infants and
young children because this age group is at greatest risk of mal-
treatment victimization (Wulczyn, Barth, Yuan, Harden, & Land-
sverk, 2005), comprising 41%of all children substantiated for
maltreatment and the vast majority (86%) of child fatalities due
to maltreatment (USDHHS, 2012). Moreover, the provision of
formal child welfare services is often dependent upon substantia-
tion (National Survey of Child and Adolescent Well-Being,
2007), and there is perhaps no more important developmental
window for CPS and other systems to intervene than during the
1
California Child Welfare Indicators Project, School of Social Welfare, Uni-
versity of California, Berkeley, CA, USA
2
School of Social Work, University of Southern California, Los Angeles, CA,
USA
Corresponding Author:
Bryn King, California Child Welfare Indicators Project, School of Social
Welfare, University of California, Berkeley, 16 Haviland Hall, Berkeley, CA
94720, USA.
Email: brynking@berkeley.edu
Child Maltreatment
18(4) 232-244
ªThe Author(s) 2013
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1077559513508001
cmx.sagepub.com
first 5 years of life. Our objective was to assess the independent
association of reporter identity with the likelihood of substantia-
tion in the context of a number of factors associated with CPS
involvement among young children reported for maltreatment.
Specifically, we explored and modeled the relationship between
the CPS substantiation decision and (1) the identity of the mal-
treatment reporter; (2) the type of maltreatment reported; and
(3) the sociodemographic characteristics and risk factors at birth
of the reported child and his or her family.
Background
Reporter Identity
Studies based on administrative CPS data have consistently
found that substantiation rates vary based on reporting source,
with allegations from certain types of reporters more likely to
result in substantiation than others (Drake, 1995; English
et al., 2002; Giovannoni, 1995; Kesner, 2008). In the aggregate,
reports from mandated sources are more likely to be substan-
tiated than reports from nonmandated sources (English et al.,
2002; McDaniel, 2006). McDaniel’s (2006) study produced the
strongest findings regarding the role of reporter status. That
study, which examined a number of sociodemographic charac-
teristics (e.g. income, welfare receipt, and race) and family-
level risk factors for maltreatment (e.g. substance abuse,
domestic violence, and mental illness), found that mandated
reporter status was the only variable that significantly predicted
substantiation, with reports from mandated professional report-
ers having 5 times greater odds of substantiation than those
from nonmandated reporters. Despite this significant finding,
the study did not examine whether and how the odds of a sub-
stantiated report varied by maltreatment type, which may mod-
ify the effect between reporter status and case disposition.
Although mandated reporters are more likely to report alle-
gations that are ultimately substantiated, there are notable dif-
ferences in substantiation rates among different types of
mandated professionals (Drake, 1995; English et al., 2002;
Giovannoni, 1995; Kesner, 2008; Kesner & Robinson,
2002). Legal reporters (i.e., those associated with law
enforcement) generally have the highest rate of substantiation
(Drake, 1995; English et al., 2002; Kesner, 2008; King,
Trocme´, & Thatte, 2003), while school personnel have lower
overall rates of substantiation among mandated reporters
(English et al., 2002; Kesner & Robinson, 2002). For exam-
ple, one study found that 68%of the reports from legal
sources were substantiated in a multistate sample of cases,
while 50%of the reports initiated by educational profession-
als were substantiated (Kesner, 2008).
Variations in substantiation rates may be explained by dif-
ferences in the nature of contact professionals have with chil-
dren and families, which provide unique opportunities for not
only observing specific forms of maltreatment but also supply-
ing corroborating evidence for the substantiation of a report.
Giovannoni (1995), who originally advanced this ‘‘vantage
point’’ hypothesis, stated, ‘‘What individuals report can be seen
to be influenced by what they have the opportunity to observe,
not simply those things about which they have more or less
concern’’ (p. 494). As such, an examination of how maltreat-
ment allegation type is related to reporter identity and the like-
lihood of substantiation could prove useful in examining this
hypothesis.
Maltreatment Type
Prior research has indicated that, independent of reporter iden-
tity, some maltreatment allegation types are more likely to be
substantiated than others (Cross & Casanueva, 2009; Drake,
1995; English, et al., 2002). One study indicated that medical
neglect was the most likely to be substantiated (English,
et al., 2002), while two other studies identified sexual abuse
as the most likely to be substantiated (Cross & Casanueva,
2009; Drake, 1995). Cross and Casanueva (2009) found that
physical abuse was the second most frequently substantiated
allegation after sexual abuse, whereas Drake (1995) found
physical abuse was the second least likely allegation to be sub-
stantiated. The vantage point hypothesis may help explain the
differential confirmation of maltreatment allegations across
both reporter and allegation types (Drake, 1995; English
et al., 2002; Kesner & Robinson, 2002). For instance, Drake
(1995) found that law enforcement reporters in Missouri had
substantiation rates ranging from 31%(for physical abuse) to
58%(when reporting multiple maltreatment types, including
sexual abuse), whereas substantiation rates of reports by school
personnel ranged from 20%to 50%across different allegation
types.
Inconsistent findings regarding substantiation by maltreat-
ment type may be related to the identity of the reporting source,
but these differences may also be accounted for by variation
across studies in terms of data sources (multistate vs. local or
state level) and analysis (descriptive vs. multivariate). The
majority of the published studies on factors related to reporter
identity, allegation type, and maltreatment substantiation also
used data collected during the 1990s and prior to 2001, a time
period during which rates of substantiated sexual and physical
abuse were declining (Jones, Finkelhor, & Halter, 2006). Both
nationally (U.S. Department of Health and Human Services,
1999, 2012) and in California (Needell, et al., 2013), substan-
tiation rates overall have declined since the late 1990s, suggest-
ing that a contemporary examination of these dynamics using
more recent data is warranted. The addition of other risk fac-
tors, including sociodemographic characteristics, could further
explicate these relationships.
Sociodemographic Characteristics
Race and Socioeconomic Status. The sociodemographic charac-
teristics of the child and family may also be related to the like-
lihood of maltreatment substantiation, although findings are
mixed to date. Some studies find that income and race are asso-
ciated with the decision to substantiate (King et al., 2003),
while others find no evidence of associations between race or
King et al. 233
socioeconomic status and substantiation (Drake & Zuravin,
1998; English, et al., 2002; McDaniel, 2006). The most
recently conducted study on this issue examined administrative
data from Texas and found that when controlling for family
income, race was not a significant predictor of substantiation
until caseworker assessment of risk was added to the analysis,
suggesting an interactive relationship between race, income,
risk, and disposition (Dettlaff, et al., 2011). Compared to White
children, the odds of substantiation were 15%greater for Black
children and 12%greater for Latino children. However, it is
important to note that even in the fully adjusted model account-
ing for sociodemographic characteristics, the most significant
predictor of substantiation was reporter identity, with allega-
tions from legal, medical, and child welfare agencies leading
to nearly three times greater odds of substantiation than rela-
tives and family friends.
Age. Few studies have examined the effect of child age on the
likelihood of substantiation, and those studies that have mod-
eled child age show mixed findings. Some studies find no sig-
nificant differences in the odds of substantiation with respect to
child age (King, et al., 2003; McDaniel, 2006). Other studies
have indicated that the age of children may be a factor in sub-
stantiation when combined with reporter identity (Kesner,
2008) or maltreatment type (Cross and Casanueva, 2009). For
example, Cross and Casanueva (2009) found that school-age
children (6–10 years old) had greater odds of substantiation
than infants and very young children (from infancy to 2 years
old), particularly when reported for physical neglect. Among
younger children, sexual abuse is also less likely to be substan-
tiated (Eckenrode, Munsch, Powers, & Doris, 1988; Ecken-
rode, Powers, Doris, Munsch, & Bolger, 1988).
Risk Factors at Birth. Population-based data derived from vital
birth records and linked to CPS records have been used to
assess the relationship between characteristics present at birth
and the likelihood of CPS contact. However, these studies have
not specifically examined how such factors affect decision
making once a report has been made and an investigation initi-
ated. Family- and child-level risk factors present at birth and
noted in birth records are related to the likelihood of a report
to child protective services (Putnam-Hornstein & Needell,
2011) and a substantiated allegation of maltreatment (Putnam-
Hornstein, Needell, King, & Johnson-Motoyama, 2013; Wu
et al., 2004; Zhou, Hallisey, & Freymann, 2006). Compared to
the full cohort of children born, children whose mothers were
younger, had less education, received inadequate prenatal care,
used public insurance (Medicaid), and were unmarried or did not
have the father present at birth to establish paternity were more
likely to be reported or substantiated for maltreatment (Putnam-
Hornstein & Needell, 2011; Wu et al., 2004; Zhou et al., 2006).
On balance, the effect of sociodemographic characteristics
on report substantiation is somewhat unclear, with inconsistent
and sometimes contradictory findings across studies. Race and
socioeconomic status appear to be only weakly related to the
substantiation decision of a particular report. However, given
the documented interplay of race and income in the risk of CPS
involvement and the persistence of racial disparities (Dettlaff,
et al., 2011; Putnam-Hornstein, et al., 2013), accounting for
these variables is essential to quantifying the likelihood of sub-
stantiation in the context of other factors. The age of the child
reported for maltreatment may be meaningful, but evidence
regarding the age at which children are most likely to be sub-
stantiated and under what conditions is limited. Some research
suggests that infants and very young children have a specific
profile with respect to substantiation rates and maltreatment
type, but characteristics most likely related to the likelihood
of substantiation have never been studied among this popula-
tion. Risk factors present at the birth of the child, including
family characteristics and pregnancy- or birth-related circum-
stances, hold promise as potential explanatory variables. Previ-
ous work only compared reported or substantiated children
with children in the population who were not reported or sub-
stantiated for maltreatment. However, such work has not
assessed the association of these risks in children with investi-
gated reports. Additionally, the simultaneous associations of
birth-related variables, maltreatment type, and reporter identity
on substantiation are yet to be examined.
Substantiation as an Outcome
Given that this study examines substantiation as an outcome,
critiques of substantiation as a meaningful indicator of mal-
treatment warrant a brief discussion. A number of scholars
have cautioned that the lack of substantiation should not be
regarded as an indicator that the child was not maltreated, since
a number of factors independent of risk or harm may influence
this decision (Cross and Casanueva, 2009: Drake, 1996; Eng-
lish, et al., 2002). Additionally, among children reported for
maltreatment, substantiation did not predict long-term beha-
vioral and developmental outcomes (Hussey, et al., 2005).
Despite these critiques regarding the utility of substantiation
as an indicator of maltreatment, for the purposes of this study,
substantiation is not intended to serve as a measure of whether
or not maltreatment actually occurred. Rather, it is used as an
indicator of a critical CPS decision point that determines
whether an investigated maltreatment report is deemed credible
and therefore may serve as a gateway to family court involve-
ment and the family’s receipt of formal CPS services and
supervision. Because this study is concerned with the differen-
tial validation of investigated reports based on reporter identity,
substantiation is considered an important outcome.
The Current Study
BirthrecordslinkedtoCPSrecordsfromCaliforniaprovide
a unique opportunity to better understand the dynamics of
report substantiation by simultaneously examining reporter
identity and maltreatment type with the addition of a rich
set of sociodemographic and health-related characteristics
at birth that are unavailable in most CPS data systems.
These characteristics include risk factors present at the birth
234 Child Maltreatment 18(4)
of the child, which have not yet been studied among chil-
dren reported and investigated for maltreatment. Inconsis-
tencies in previous work with respect to the associations
of maltreatment type or sociodemographic characteristics
with substantiation may be due to the descriptive nature
of a number of these studies, which did not adjust for the
presence of multiple factors that may have been related to
substantiation. Prior studies that have accounted for socio-
demographic and child- and family-level risk factors
included limited combinations of these covariates and often
neglected to examine the simultaneous associations of mal-
treatment type. The present study extends previous research
by using longitudinal, population-level data to prospectively
examine whether the status and identity of the maltreatment
reporter are independent predictors of substantiation across
maltreatment types and in the presence of multiple sociode-
mographic characteristics and risk factors present at birth.
Based on the vantage point hypothesis and findings of
related studies, we predicted that even after adjusting for other
risk factors (1) children reported by mandated reporters would
have a higher rate of substantiation than those reported by non-
mandated sources; (2) among children reported by mandated
reporters, those reported by law enforcement and medical pro-
fessionals would be the most likely to be substantiated; and (3)
among children reported by mandated reporters, the likelihood
of substantiation would vary by maltreatment type. For the
third hypothesis, we expected that children reported for types
of maltreatment that are more difficult to substantiate, such
as sexual and physical abuse, substantiation would be more
likely if allegations were reported by mandated sources.
Because existing literature on the relationship of sociodemo-
graphic characteristics with substantiation is less clear, we did
not make explicit predictions regarding the associations of
these variables, However, we were interested in whether the
inclusion of such characteristics would be related to the likeli-
hood of substantiation when included in multivariable models.
Method
Data Set
CPS records for all children born in 2002 and referred for mal-
treatment for the first time prior to age 5 years were extracted
from California’s Child Welfare Services/Case Management
System (CWS/CMS), a statewide administrative data system.
All records fall under the authority of the California Depart-
ment of Social Services and are shared with the University of
California at Berkeley under a long-standing interagency colla-
borative agreement. CPS records were probabilistically
matched to birth records from 2002 obtained from the Califor-
nia Department of Public Health. This linkage methodology
has been described in other publications (Putnam-Hornstein &
Needell, 2011; Putnam-Hornstein, Webster, Needell, &
Magruder, 2011). This project received both state and univer-
sity institutional review board approvals and the endorsement
of the state’s Vital Statistics Advisory Board.
The linkage process resulted in successfully matching 84%
of children referred to CPS by 5 years of age to a birth record.
Children not matched included those with missing information
in one or both files such that we could not confidently identify a
corresponding record as well as children who were referred for
maltreatment in the state but were born outside of California.
Of the 531,035 children born in California in 2002, 14%
(74,182) were referred for possible maltreatment before 5 years
of age. In this population of children referred to CPS and linked
to a birth record, complete data were available for 94%.
Because the focus of the study was on substantiation, we
excluded children whose first referrals were ‘‘evaluated out’’
and therefore did not receive an investigation based on their
first contact with CPS and could not have had a first report that
was substantiated. The final data set included 59,413 children
born in 2002, who were successfully matched to a birth record,
had complete data available in the birth record, and whose first
report for maltreatment was received and investigated by CPS
prior to 5 years of age.
Variables
Substantiation. The dependent variable for this study was
whether or not the first investigated report of maltreatment was
substantiated. Although some children in our data set had mul-
tiple allegations of maltreatment, as well as multiple reports
between birth and 5 years of age, we conducted a child-level
analysis rather than a report- or allegation-level analysis. Based
on their first investigated report, children were categorized
according to the disposition hierarchy used by the California
Department of Social Services: (1) substantiated (evidence was
sufficient to determine that maltreatment was more likely to
have occurred than not); (2) inconclusive (evidence was insuf-
ficient to determine whether or not maltreatment had
occurred); and (3) unfounded (it was determined that maltreat-
ment had not occurred). If a child had multiple allegations in
the first report, the child was first classified according to the
most serious disposition assigned to any of the allegations and
then by the severity of the type of the allegation (see descrip-
tion of maltreatment type variable for details). Substantiation
was coded as a dichotomous variable (substantiation¼1, all
other dispositions ¼0).
Reporter Type. To determine the influence of reporter type on
the substantiation process, we examined the mandated status
and professional identity of the reporter. When entering referral
information into CWS/CMS, child welfare workers select from
a specified list of possible reporter types. Mandated reporter
status was coded dichotomously (mandated vs. nonmandated)
and encompassed all reporter types required by California law
to report incidents of maltreatment (Child Abuse and Neglect
Reporting Act, 1974). Mandated reporters were then coded into
six professional fields: (1) legal/law enforcement (law enforce-
ment, probation/parole officers, and guardians ad litem); (2)
medical professionals (medical and dental professionals); (3)
public agencies (CPS and other government agency staff); (4)
King et al. 235
other professionals; (5) helping professionals (counselors,
therapists, clergy, and court-appointed special advocates); and
(6) school/child care (teachers, school personnel, day care, and
substitute care providers). Sorting each reporter type from
CWS/CMS into a professional field was relatively straightfor-
ward in most cases (e.g., legal/law enforcement); in those
instances that were less clear, efforts were made to categorize
reporters based on the nature of their relationship to the chil-
dren or families involved (e.g., the school/child care category)
and as guided by reporter categories adopted in prior research
(Giovannoni, 1995; Kesner, 2008; McDaniel, 2006).
The ‘‘other professionals’’ category, an available option for
CPS staff entering data, was the third largest reporter type at
15%. To better understand the composition of these reporters,
a small sample (n¼50) was randomly selected to qualitatively
assess the status and identity of these reporters in individual
case files. All were described as mandated reporters in the case
record. As such, other professionals were ascribed mandated
status. In terms of reporter type, a substantial proportion
(32%) of the 50-case sample included social workers in medi-
cal settings (e.g., hospitals, community clinics, etc.). The next
largest group included counselors or advocates from a range of
social service agencies (30%), followed by child welfare and
other county agency staff members, such as public assistance
workers (22%). Given the variation in this small sample and
the inability to manually examine the entirety of records for
the study population, it was not possible to reliably reallocate
reports from other professionals to more specific professional
categories.
Reporters coded as nonmandated were those not required by
law to report incidents of child maltreatment. In this population,
we classified reporters into three types: (1) family members (sib-
lings, grandparents, aunts and uncles, cousins, godparents, and
relatives), (2) community members (friends, neighbors, land-
lords, and unrelated individuals), and (3) unidentified reporters.
Unidentified reporters presented a challenge similar to that
posed by other professional reporters. For those observations
without an identified reporter, the reporter type field in the
administrative data was either missing or coded as 0. Because
California law requires that mandated reporters provide identify-
ing information (Child Abuse and Neglect Reporting Act, 1974),
we assumed these reports likely originated from nonmandated
sources but wanted support for this determination. A review of
cases with unidentified reporters was conducted. Again, a small
random sample (n¼50) was drawn from the subgroup of uni-
dentified reporters and, using the unique referral identifier for
each child, individual case records were examined. Of the 50
cases, 96%involved nonmandated reporters, and of those,
77%involved anonymous sources and 23%involved various
nonmandated sources (self-reports, family members, and unre-
lated community members). As such, unidentified/anonymous
reporters were classified as nonmandated.
Maltreatment Type. The form of alleged maltreatment was cate-
gorized into five types. Children with a first report that
included multiple types of maltreatment were first selected
by the maltreatment disposition hierarchy (e.g., substantiated,
unfounded) and then according to the maltreatment allegation
hierarchy (see below) established by the California Department
of Social Services, with the most severe maltreatment type used
for analysis. This allegation hierarchy is as follows: (1) sexual
abuse (sexual abuse or exploitation); (2) physical abuse; (3)
neglect (severe neglect, general neglect, or caretaker absence
or incapacity); (4) emotional abuse (may include witnessing
domestic violence); and (5) substantial risk (at risk due to sib-
ling being maltreated or substantial risk of abuse). For exam-
ple, if a first report included a substantiated allegation of
neglect and an unfounded allegation of physical abuse, the case
would be coded as neglect (because the disposition hierarchy
would supersede the allegation hierarchy).
Sociodemographic Characteristics. Adjustments for sociodemo-
graphic and pregnancy-related risk factors included child age
at first report (<1 year, 1–4 years); birth weight (<2,500 g,
2,500 g); birth abnormality (one or more abnormalities,
none); commencement of prenatal care (first trimester, second
trimester, third trimester, or no care); birth order (firstborn or
second born, third born or higher); maternal race/ethnicity
(Asian/Pacific Islander, Black, Latino, Native American, or
White); maternal birthplace (U.S. born or foreign born); mater-
nal age (<25 years or 25years); maternal education (less than
high school, completed high school, some college or more);
health insurance (public insurance/Medi-Cal or private insur-
ance); and paternity (established at birth or missing). It should
be noted that we tested a continuous measure of child age and
found that children were more likely to be substantiated during
the first year of life than at any other age. As there were no sub-
stantial differences by age after the child’s first birthday, we
elected to use an indicator variable.
Statistical Analysis
The relative risk of substantiation was calculated using general-
ized linear models (McCullagh & Nelder, 1989) and a modified
Poisson regression technique proposed as an alternative
method for use with prospective cohort studies in which the
outcome of interest is binary (Zou, 2004). This technique spe-
cifies a Poisson distribution and log link, using a robust stan-
dard error adjustment (sandwich estimator) to correct for
estimated confidence intervals (CIs) that would otherwise be
too wide. Results were compared with those obtained from
logistic regression models. As expected, the same variable
associations emerged, although logistic regression produced
more extreme point estimates.
We examined the unadjusted risk of substantiation across
reporter identity, maltreatment type, and sociodemographic
characteristics. To test our first hypothesis regarding the inde-
pendent association of the status of the reporter on the likeli-
hood of substantiation, we specified a multivariable model
that compared mandated to nonmandated reporters, adjusting
for maltreatment type and sociodemographic covariates. For
our second hypothesis assessing the differences between
236 Child Maltreatment 18(4)
mandated reporter types on the likelihood of substantiation, we
generated a second adjusted model that compared professional
reporter types to nonmandated reporters. To test our third
hypothesis regarding how these relationships vary by maltreat-
ment type, we created stratified models across maltreatment
type, which meant that both the reporter status and reporter
types models were specified for each allegation. Stratified
models also were adjusted for sociodemographic covariates.
It should be noted that statistical significance is deempha-
sized in the results section, although the results of such testing
are reported. Instead, our discussion focuses on effect size
because this study relies on the full census of children reported
and investigated for maltreatment in California. Because this is
not a sample, the use of inferential statistics to generalize to a
broader population is unnecessary. Additionally, the size of our
study population is so large that even modest differences will
emerge as statistically significant.
Results
Descriptive Statistics
The left side of Table 1 presents the distribution of children in
the study with a first report of maltreatment before 5 years of
age, the percentage of those reports that were unsubstantiated
and substantiated, and the crude risk ratios (RRs) of substantia-
tion for each variable. Of 59,413 children experiencing their
first investigated report of maltreatment, the majority were not
substantiated. Notable differences were observed in the unad-
justed rates of report substantiation for all variables of interest
except maternal age. Overall, children reported by mandated
reporters were more likely to be substantiated than those
reported by nonmandated reporters. Among mandated profes-
sionals, reports from legal sources and law enforcement consti-
tuted the largest reporter type group, followed by other
professionals and medical professionals. More than a third of
children reported by law enforcement, medical professionals,
and public agencies were substantiated as victims, whereas less
than a fifth of children reported by school and day care person-
nel, family, community members, and unidentified reporters
were substantiated. Crude RRs indicated that children reported
by law enforcement were far more than likely to be substantiated
as victims than children from unidentified sources.
As with the reporter type, there were significant variations
in the distribution and rate of substantiation across maltreat-
ment types. By far, the most common maltreatment type was
neglect which was also substantiated more frequently than all
other forms of maltreatment. Allegations of sexual abuse were
comparably rare and were substantiated less often than other
maltreatment types. Among sociodemographic covariates,
there was both expected and unexpected variation in the rate
of substantiation across child and family characteristics. Infants
were substantiated as victims more frequently than children
between the ages of 1 and 4 years. Children with low birth
weight or birth abnormalities were also more likely to be sub-
stantiated as victims, as were children born into larger families
(in which there were at least two other children born to a
mother). Children whose mothers received prenatal care later
in their pregnancy or not at all were more likely to be substan-
tiated as victims than children whose mothers received prenatal
care beginning in the first trimester. Gradations in risk or sub-
stantiation were observed across the trimester in which care
began, although the rate among children whose mothers
received no prenatal care was substantially higher. Unadjusted
differences in the rate of substantiation across maternal nativ-
ity, race, and ethnicity were modest. Children whose mothers
were born in the United States were more likely to be substan-
tiated than children whose mothers were not, although this dif-
ference was relatively low in magnitude. A slightly greater rate
of substantiation was observed among children with Native
American mothers relative to those of White mothers; chil-
dren born to Black, Latino, and Asian/Pacific Islander moth-
ers had rates of substantiation that were less than or
statistically equivalent to those born to White mothers. Simi-
lar to race and ethnicity, the rate of case substantiation across
other sociodemographic characteristics followed expected
patterns of risk, although the magnitude of these differences
was generally low. Overall, unadjusted risk ratios for substan-
tiation suggested that the most salient predictors of maltreat-
ment are reporter status, reporter type, and allegation type,
whereas child and family characteristics demonstrated more
modest relations.
Adjusted Models
Reporter Status. As indicated on the right side of Table 1, Model
1 estimated the substantiation risk for children reported by
mandated versus nonmandated reporters after adjusting for
allegation type and sociodemographic characteristics of the
child and family. Confirming our first hypothesis and consis-
tent with bivariate findings, children reported by mandated
reporters were more likely to be substantiated than those
reported by nonmandated reporters. Variation in the likelihood
of substantiation across maltreatment allegation types was
largely unaffected by covariate adjustments. Children reported
for neglect were still more likely to be substantiated than those
reported for physical abuse and the lower rate of substantiation
observed in the unadjusted model for children reported for sex-
ual abuse was also sustained.
Differences in a child’s risk of substantiation across socio-
demographic and birth record variables were largely consistent
between Model 1 and crude estimates. For all variables, RRs
were attenuated in the adjusted model, including child age,
pregnancy-related variables, maternal birthplace and educa-
tion, insurance coverage, and paternity. This attenuation was
particularly evident when examining the effect of prenatal care.
The unadjusted risk of substantiation among mothers who
received no prenatal care was 2.6 times that of those who
received timely care, whereas in the adjusted model this risk
decreased to slightly more than 1.5. This shift is likely attribu-
table to the effect of public insurance coverage (Medi-Cal), as
described in previous research (Putnam-Hornstein & Needell,
King et al. 237
Table 1. Children First Reported: Variable Distribution, Percentage Substantiated, Unadjusted and Adjusted Risk of Substantiation.
First Reports Risk of Substantiation
All Unsub. Sub. Unadjusted Model 1 Model 2
N% % % RR 95% CI Adj. RR 95% CI Adj. RR 95% CI
Total 59,431 73.9 26.1 — — — — — —
Reporter status
Mandated 41,280 69.5 68.2 31.8 2.44*** [2.34, 2.54] 2.49*** [2.40, 2.60] — —
Nonmandated 18,151 30.5 86.9 13.1 ref ref ref ref — —
Reporter type
Legal/law enforcement 10,510 17.7 59.2 40.8 3.30*** [3.13, 3.48] — — 3.35*** [3.21, 3.50]
Medical professionals 7,878 13.3 61.6 38.4 3.11*** [2.94, 3.29] — — 2.55*** [2.43, 2.67]
Public agencies 3,190 5.4 64.6 35.4 2.86*** [2.68, 3.06] — — 2.59*** [2.44, 2.74]
Other professionals 8,980 15.1 67.7 32.3 2.62*** [2.47, 2.77] — — 2.32*** [2.21, 2.43]
Helping professionals 3,850 6.5 78.6 21.4 1.73*** [1.60, 1.87] — — 1.82*** [1.70, 1.96]
School/child care 6,872 11.6 85.9 14.1 1.14*** [1.05, 1.23] — — 1.29*** [1.20, 1.39]
Family 3,937 6.6 84.8 15.2 1.23*** [1.13, 1.35] — — ref ref
Community members 2,329 3.9 87.0 13.0 1.05 [0.93, 1.18] — — ref ref
Unidentified 11,885 20.0 87.7 12.3 ref ref — — ref ref
Allegation type
Sexual abuse 1,678 2.8 89.9 10.1 0.83* [0.70, 0.97] 0.84* [0.71, 0.98] 0.78** [0.66, 0.91]
Physical abuse 5,288 8.9 87.8 12.2 ref ref ref ref ref ref
Neglect 29,625 49.8 67.7 32.3 2.65*** [2.46, 2.85] 2.36*** [2.20, 2.54] 2.13*** [1.98, 2.29]
Emotional abuse 7,841 13.2 77.3 22.7 1.86*** [1.71, 2.02] 1.80*** [1.66, 1.96] 1.52*** [1.40, 1.65]
Substantial risk 14,999 25.2 77.7 22.3 1.83*** [1.69, 1.98] 1.62*** [1.50, 1.75] 1.65*** [1.53, 1.78]
Child’s age
Infants (<1 year) 20,761 34.9 64.2 35.8 1.71*** [1.67, 1.76] 1.32*** [1.28, 1.36] 1.30*** [1.27, 1.34]
1–4 years 38,670 65.1 79.1 20.9 ref ref ref ref ref ref
Birthweight
Low (<2,500 g) 5,015 8.4 66.7 33.3 1.31*** [1.26, 1.37] 1.07*** [1.03, 1.12] 1.08*** [1.04, 1.13]
Normal (2,500 g) 54,416 91.6 74.6 25.4 ref ref ref ref ref ref
Birth abnormality
One or more 4,448 7.5 66.4 33.6 1.32*** [1.26, 1.38] 1.05 [1.00, 1.10] 1.06* [1.01, 1.10]
None 54,983 92.5 74.6 25.4 ref ref ref ref ref ref
Prenatal care
No prenatal care 1,023 1.7 37.5 62.5 2.64*** [2.51, 2.78] 1.57*** [1.50, 1.65] 1.59*** [1.52, 1.68]
Third trimester 2,302 3.9 63.4 36.6 1.55*** [1.46, 1.63] 1.23*** [1.16, 1.30] 1.23*** [1.17, 1.30]
Second trimester 10,368 17.4 69.3 30.7 1.30*** [1.26, 1.34] 1.13*** [1.10,1.17] 1.13*** [1.09, 1.17]
First trimester 45,738 77.0 76.3 23.7 ref ref ref ref ref ref
Birth order
Third born or higher 26,577 44.7 71.9 28.1 1.15*** [1.12, 1.18] 1.06*** [1.03, 1.09] 1.10*** [1.07, 1.13]
Firstborn or second born 32,854 55.3 75.5 24.5 ref ref ref ref ref ref
Mother’s race
Black 7,502 12.6 72.9 27.1 1.00 [0.96, 1.04] 0.92 [0.89, 0.96] 0.93*** [0.89, 0.97]
White 17,190 28.9 72.8 27.2 ref ref ref ref ref ref
Latino 31,453 52.9 75.2 24.8 0.91*** [0.86, 0.94] 0.97 [0.94, 1.01] 0.97 [0.94, 1.01]
Asian/Pacific Islander 2,665 4.5 71.4 28.6 1.05 [0.99, 1.12] 1.20*** [1.13, 1.29] 1.19*** [1.12, 1.27]
Native American 621 1.0 65.4 34.6 1.27*** [1.14, 1.42] 1.10 [0.99, 1.22] 1.09 (0.99, 1.21)
Mother’s birthplace
U.S. born 41,166 69.3 72.4 27.6 1.21*** [1.18, 1.26] 1.17*** [1.13, 1.20] 1.14*** [1.10, 1.16]
Foreign born 18,265 30.7 77.4 22.6 ref ref ref ref ref ref
Mother’s age
<25 years 28,886 48.6 74.1 25.9 0.99 [0.96, 1.02] 1.00 [0.97, 1.03] 0.98 [0.95, 1.01]
25 years 30,545 51.4 73.8 26.2 ref ref ref ref ref ref
Mother’s education
Less than high school 25,137 42.3 72.2 27.8 1.25*** [1.20, 1.30] 1.12*** [1.08, 1.17] 1.14*** [1.09, 1.19]
Completed high school 21,779 36.6 73.7 26.3 1.18*** [1.14, 1.23] 1.10*** [1.06, 1.15] 1.11*** [1.06, 1.15]
Some college plus 12,515 21.1 77.8 22.2 ref ref ref ref ref ref
Health insurance
Public (Medi-Cal) 38,796 65.3 72.5 27.5 1.18*** [1.14, 1.21] 1.08*** [1.05, 1.12] 1.08*** [1.05, 1.11]
Private insurance 20,635 34.7 76.7 23.3 ref ref ref ref ref ref
(continued)
238 Child Maltreatment 18(4)
2011). Adjusting for all covariates shifted some of the differ-
ences regarding race and ethnicity, although again, those asso-
ciations were modest.
Mandated Reporter Type. As described in the last column of
Table 1, Model 2 estimated the adjusted risk of substantiation
but separated mandated reporters by reporter type to test our sec-
ond hypothesis. The differences that emerged in terms of unad-
justed RRs for reporter type remained after adjusting for
allegation type and potential sociodemographic risk confoun-
ders. As hypothesized, children reported by law enforcement and
legal sources, medical professionals, public agencies, and other
professionals were considerably more likely to be substantiated
than children reported by nonmandated reporters. Children with
reports originating from helping professionals and school/child
care personnel were moderately more likely to be substantiated
than children with reports from family, community members,
and unidentified reporters. The risk of substantiation across mal-
treatment types when adjusting for mandated reporter type was
largely consistent with both unadjusted and Model 1 estimates.
The differences in the risk of substantiation across sociodemo-
graphic characteristics were also consistent with the estimates
for Model 1, with one exception. After controlling for reporter
type, maltreatment type, and other sociodemographic risks in
Model 2, children of Black mothers were less likely to experi-
ence substantiation than children of White mothers; however,
this difference was relatively modest.
Adjusted Models Stratified by Maltreatment Type
We hypothesized that the association between reporter identity
and the likelihood of substantiation would vary by maltreatment
type, but Models 1 and 2 were not able to fully explain this rela-
tionship. As such, separate models were specified to estimate the
likelihood of substantiation for each allegation, first by reporter
status (mandated vs. nonmandated) and then by reporter type, as
presented in Table 2. For allegations of sexual abuse, children
reported by mandated reporters were more likely than those
reported by nonmandated reporters to be substantiated. This was
particularly true for sexual abuse reports originating from law
enforcement sources, which were substantiated far more often
than reports originating from nonmandated sources. We con-
ducted a follow-up analysis to test whether or not child age could
be a factor in sexual abuse substantiation by recoding the child’s
age from infants versus aged 1–4 to children aged 0–2 versus
3–4 years. Children who were 3 years and older comprised
only 29%of the study population, but they constituted 63%
of the children reported for sexual abuse. Among those
reported for sexual abuse, children aged 3 and older were
79%(RR: 1.79, p¼.001, 95%CI [1.28, 2.51]) more likely
to be substantiated thanthoseaged2andunder.
Children reported for physical abuse by mandated reporters
were also more likely to be substantiated than those reported by
nonmandated sources. When stratifying by reporter type, chil-
dren reported for physical abuse by law enforcement and med-
ical professionals were substantially more likely to be
substantiated than children reported for physical abuse by non-
mandated sources; indeed, children with reports initiated by
law enforcement were more than 6 times as likely to be sub-
stantiated. Children reported for physical abuse by teachers,
school personnel, and child care staff were more likely to be
substantiated than children with reports received from nonman-
dated reporters. Across the other maltreatment types, children
with reports originating from mandated sources, particularly
law enforcement, medical professionals, public agencies, and
other professionals, were more likely to be substantiated than
children with reports originating from family, community
members, or unidentified reporters.
Discussion
Consistent with our hypotheses and with previous research,
findings from this analysis indicated that reporter identity is
a strong and independent predictor of maltreatment substan-
tiation among young children. This finding remained even
after adjusting for both maltreatment type and a range of
sociodemographic characteristics and birth-related risk fac-
tors that have not been included in previous studies. Children
with reports originating from mandated sources, including
and especially law enforcement, medical professionals, and
government agency workers, were more likely to be substan-
tiated as victims of abuse or neglect than those reported by all
three types of nonmandated sources. Type of reported mal-
treatment was also a significant predictor of substantiation,
both before and after adjusting for reporter identity and the
characteristics of the child. When models were specified
Table 1. (continued)
First Reports Risk of Substantiation
All Unsub. Sub. Unadjusted Model 1 Model 2
N% % % RR 95% CI Adj. RR 95% CI Adj. RR 95% CI
Paternity established at birth
Missing 13,314 22.4 67.0 33.0 1.37*** [1.33, 1.41] 1.14*** [1.10, 1.17] 1.13*** [1.10, 1.17]
Established 46,117 77.6 75.9 24.1 ref ref ref ref ref ref
Note. Unsub. ¼unsubstantiated; Sub. ¼substantiated; RR ¼risk ratio; Adj. ¼adjusted; CI ¼confidence interval.
*p< .05; **p< .01; ***p< .001.
King et al. 239
separately by maltreatment type, children reported for allega-
tions of physical abuse had a greater likelihood of substantia-
tionwhenreportedbymandatedsourcescomparedtofamily,
community members, and unidentified reporters. Although
allegations of sexual abuse were consistently least likely to
be substantiated, we found that among children reported for
sexual abuse, those reported by law enforcement were more
likely to be substantiated than those reported by nonmandated
sources, after adjusting for other child characteristics.
It is notable that sexual abuse was generally the least likely
maltreatment type to be substantiated, especially because it
stands in contradiction to previous research, which has found
that sexual abuse was the most likely maltreatment type to be
substantiated (Cross & Casanueva, 2009; Drake, 1995). These
reports are relatively rare in this age group—the proportion of
children younger than 5 years reported for sexual abuse in the
study population was less than 3%. The major reason for this
discrepancy is likely related to the age of the children in our
study, since previous research has indicated that younger chil-
dren are less likely to be substantiated for sexual abuse. Fur-
ther, it can be more difficult to reach the level of evidence
necessary to substantiate such an allegation among very young
children, as they often lack the capacity to reliably recount and
articulate their experiences (Haskett, Wayland, Hutcheson, &
Tavana, 1995). This suggests that allegations of sexual abuse
are more likely to be substantiated as children age. Our
follow-up analysis supports this idea since children who were
between 3 and 4 years old were more likely to be reported and
substantiated for sexual abuse than their younger counterparts.
Child and family characteristics, including pregnancy-
related risk factors and sociodemographic variables, did not
modify the relationship between reporter identity and substan-
tiation and did not figure prominently in the likelihood of report
substantiation. The modest association of race and other socio-
demographic characteristics with the substantiation decision
may reflect that, in general, children investigated for maltreat-
ment already comprise a fairly homogenous high-risk subset of
children. As such, risk factors that are predictive of CPS invol-
vement and substantiation among infants and young children in
the general population (Putnam-Hornstein et al., 2013; Putnam-
Hornstein & Needell, 2011; Wu et al., 2004) may be less salient
in predicting which reported children will be substantiated for
maltreatment.
Proximity to Maltreatment and Evidence of
Maltreatment
Given previous research, we expected variations in the likeli-
hood of substantiation across reporter types. For most mandated
reporters, access and proximity to either the maltreatment event
or its effects may be a direct result of their vantage point (Gio-
vannoni, 1995; McDaniel, 2006). When considering potential
reasons why children reported by law enforcement may have
higher rates of substantiation than children reported by other
sources, Drake (1995) noted that this may be due to factors
related to law enforcement’s investigative and evidence-
gathering skills, rather than the nature of the maltreatment
reported. This perspective may help to explain the disposition
of reports in this study; children who were reported for sexual
abuse, physical abuse, and neglect by law enforcement were far
more likely to be substantiated than those reported by family and
community members. Given that these data are restricted to
reports involving infants and young children who have less
capacity to articulate maltreatment events, direct evidence of
maltreatment—which law enforcement is well positioned to
observe and collect—may play a more significant role in the
substantiation determination.
During the course of their work, police, probation, and par-
ole officers are more likely than other reporters (e.g. teachers,
therapists, community members, etc.) to observe the direct
effects of maltreatment or the maltreatment itself, because they
have contact with families for other reasons (such as a parole
visit or a domestic violence call) or because they are specifi-
cally called out to investigate situations in which maltreatment
Table 2. Adjusted Risk of Substantiation, by Allegation Type.
Sexual Abuse Physical Abuse Neglect Emotional Abuse Substantial Risk
n¼1,678 n¼5,288 n¼29,625 n¼7,841 n¼14,999
Adj. RR 95% CI Adj. RR 95% CI Adj. RR 95% CI Adj. RR 95% CI Adj. RR 95% CI
Reporter status (vs. nonmandated)
Mandated 1.65* [1.10, 2.47] 3.29*** [2.60, 4.16] 2.78*** [2.65, 2.92] 2.04*** [1.82, 2.30] 1.72*** [1.56, 1.88]
Reporter type (vs. nonmandated)
Legal/law enforcement 3.36*** [2.18, 5.18] 6.25*** [4.86, 8.04] 3.92*** [3.40, 3.98] 2.08*** [1.83, 2.37] 2.47*** [2.22, 2.75]
Medical professionals 1.19 [0.71, 2.00] 3.91*** [3.02, 5.06] 2.57*** [2.15, 2.54] 2.16*** [1.73, 2.70] 2.23*** [2.00, 2.49]
Public agencies 1.53 [0.77, 3.04] 2.79*** [1.44, 4.34] 2.83*** [2.46, 2.95] 1.84*** [1.42, 2.38] 2.01*** [1.76, 2.29]
Other professionals 0.86 [0.45, 1.65] 2.18** [1.30, 3.12] 2.43*** [2.04, 2.40] 2.12*** [1.84, 2.43] 1.82*** [1.63, 2.03]
Helping professionals 1.36 [0.73, 2.53] 1.59* [1.03, 2.46] 2.06*** [1.60, 2.00] 2.12*** [1.79, 2.51] 1.16 [1.00, 1.34]
School/child care 1.02 [0.53, 1.98] 2.16*** [1.44, 3.45] 1.34*** [1.20,1.50] 1.55*** [1.25, 1.91] 0.98 [0.86, 1.10]
Note. All models adjusted for child age; birthweight, birth abnormalities, prenatal care, and birth order; maternal race, birthplace, age, and education; health insur-
ance coverage; and paternity.
*p< .05; **p< .01; ***p< .001.
240 Child Maltreatment 18(4)
is suspected. By virtue of their roles as investigators, law
enforcement officers are also able to collect additional infor-
mation from collateral contacts and witnesses. Their documen-
ted observations may provide critical evidence for CPS because
they determine whether or not an allegation of maltreatment is
credible (although it is also possible that the authority of their
professional role provides credibility rather than the acuity and
documentation of their observations).
Drake’s (1995) proposed interpretation regarding the
level of evidence that law enforcement is able to provide
is related to his subsequent model of substantiation (Drake,
1996). He maintained that the level of available evidence is
as important as the level of assessed harm in the substantia-
tion decision and that a deficiency in either of these dimen-
sions may influence the ability of CPS officials to
substantiate maltreatment. In Drake’s model, evidence plays
a crucial role in substantiation. Cross and Casanueva (2009)
tested and validated his model and found that the level of
evidence superseded the effect of assessed levels of harm
and risk in predicting substantiation. Findings from the pres-
ent study may lend further support to Drake’s hypothesis.
For example, medical professionals, who also have a high
relative rate of substantiation in the current study, have the
opportunity to document the observable effects of maltreat-
ment, such as bruising, malnutrition, or the lack of appropri-
ate medical care, because they attend to children and
families on both routine medical visits and in emergency
rooms. This firsthand knowledge and documentation may
contribute to a greater likelihood of substantiation.
As with medical and law enforcement, other mandated
reporters who have routine contact with children and families
may also be in the position to both observe and offer evidence
of maltreatment that ultimately has an influence on substantia-
tion. Public agency employees, including child welfare workers,
have a consistently higher rate of substantiation compared to
nonmandated reporters, especially in cases involving physical
abuse or neglect. Likewise, school and child care personnel have
a substantially higher likelihood of substantiation when report-
ing physical abuse (which may present clearer visible evidence)
than other allegation types.
A final finding to emerge from this study, which may also
highlight the potential role of the evidence available to report-
ers, involves children reported by helping professionals. Mal-
treatment reports received from helping professionals (mostly
counselors and therapists) were more likely to be substantiated
than reports from family and community members; however,
when stratifying by maltreatment type, it was apparent that this
was only true in cases involving allegations of neglect or emo-
tional abuse. These forms of maltreatment may be more likely
to feature evidentiary information (e.g., inadequate care or
supervision, verbal assaults) that can be obtained and documen-
ted in the context of a counseling session, particularly if parents
and caregivers are disclosing such information. Additionally,
because validation of emotional abuse largely depends on accu-
rately assessing its effects, counseling professionals are well
positioned to gather such information.
Limitations
Despite this study’s strengths in terms of size, scope, and design, it
suffered from many of the limitations common to studies involv-
ing large-scale administrative data (Drake& Jonson-Reid, 1999).
First, data entry errors were present in both CPS and birth records,
affectingourabilitytosuccessfullylinkrecords.Anexamination
of unmatched CPS records indicated several notable differences
between matched and unmatched records, including in the distri-
bution of maltreatment dispositions (Putnam-Hornstein & Nee-
dell, 2011). It is unknown how many children we could not link
to a birth record may have otherwise differed from those included
in our analysis. Second, we assumed independence of observa-
tions in our analysis, but it is unknown whether the likelihood
of substantiation differed by county or CPS investigator and
whether that affected substantiation for children grouped within
those counties or investigated by particular individuals. Third,
data entry errors in classifying reporters may have biased our
examination. Our qualitative review of reporters classified as
‘‘other professionals’’ and ‘‘anonymous’’ suggested that CPS
workers entering data could have coded many of these reporters
more accurately, but the nature of the large-scale administrative
data we were working with precluded any attempt to reclassify
these reporters. Nonetheless, the large size of our population
means that our findings should be only minimally compromised
by random data entry errors.
Fourth, previous research on the factors related to the deci-
sion to substantiate has included caseworker assessments of the
level of harm and risk (Cross & Casanueva, 2009; Dettlaff
et al., 2011). Our data did not include such assessments, which
limits our understanding of all of the factors relevant to report
substantiation. Fifth, we did not examine the dynamics of
reports in which there were multiple allegations. Sixth, reports
with dispositions of inconclusive and unfounded were col-
lapsed into a single category of children whose reports were
unsubstantiated. Investigations that result in a disposition of
inconclusive may be substantially different than those that are
unfounded. Finally, our analysis was confined to the state of
California, which limits generalizability to other states and jur-
isdictions. Future research should further assess the relation-
ship between reporter identity and case disposition in other
states and examine the relationship of multiple allegations in
a single report and caseworker assessments of risk and harm
as well as the factors related to report disposition among chil-
dren whose allegations were unfounded versus inconclusive.
Implications
Previous research has consistently demonstrated that the level
and quality of evidence available to CPS investigators is a pre-
dictor of whether or not allegations of maltreatment are substan-
tiated. The results of this study contributed to this knowledge
base by suggesting that the likelihood of substantiation may
depend on the level of evidence reporters can obtain and provide
to CPS based on their unique proximity and positioning in rela-
tion to the alleged maltreatment. At a policy level, future studies
King et al. 241
should determine whether thresholds for substantiating maltreat-
ment have increased over time. For example, among children
younger than 5 years in California, the reporting rate has
increased by 18%since 1998 (from 48.8 per 1,000 in 1998 to
57.7 per 1,000 in 2011), yet the proportion of maltreatment refer-
rals that are substantiated has decreased by 17%(from 29.2%in
1998 to 24.1%in 2011; Needell et al., 2013). These countervail-
ing trends in California suggest that more infants and very young
children are being reported for reasons other than verifiable mal-
treatment or that investigators are applying different, possibly
more rigorous, criteria to the cases they substantiate. Fakunmo-
ju’s (2009) study on substantiation and adverse appeal outcomes
suggested that legal scrutiny of investigations has increased and
as a result, evidentiary standards for determining whether a
report of maltreatment is credible may be more stringent. Actual
physical and sexual abuse victimization has also declined (Jones
et al., 2006), which could help explain the reductions in substan-
tiation. Previous research, however, has not found significant
differences between substantiated and unsubstantiated children
in terms of the likelihood of maltreatment recurrence or long-
term child behavioral and developmental outcomes (Hussey
et al., 2005; Kohl, Jonson-Reid, & Drake, 2009), which suggests
that many children in both groups are facing similar levels of risk
but only some are receiving services. Because evidence plays a
crucial role in report disposition, increasing evidentiary stan-
dards for substantiation could have the unintended effect of fail-
ing to protect children in need of intervention. Continued
research should investigate whether such policy and practice
shifts have occurred and whether this has affected the likelihood
that similarly situated cases are substantiated.
If, as the results of this and other studies have suggested, evi-
dence is a factor in substantiation, then improving the capacityof
professionals designated as mandated reporters to produce
higher quality reports that are more effective at identifying fam-
ilies in need of child welfare services is one possible practice
implication. In this study, substantiation rates for mandated
reporters were far higher than for nonmandated reporters, but
there are opportunities for improvement, especially among pro-
fessional reporters in therapeutic and educational settings. Given
that these reports are selected for investigation, it is possible that
many reporters in this study were alleging events that constituted
both reasonable and potentially verifiable maltreatment. Train-
ing to increase the ability of mandated reporters to collect and
report usable and detailed information that meaningfully contri-
butes to an investigation could improve the likelihood of
substantiation for reports in which the quality of the evidence
(rather than level of risk or harm) is the only actual barrier to
more intensive intervention. Additionally, inaccuracies and
missing information in CPS data suggest that increased training
and support for child protective hotline screening could also help
workers both elicit and record critical and accurate information
that ultimately aids in the investigation. Improving report quality
is not limited to mandated professionals. Public service cam-
paigns and other community education activities could be imple-
mented to improve the information and evidence provided in
reports from both nonmandated and mandated reporters.
Conclusion
Ultimately, because substantiation often determines service
eligibility and, more importantly, provides a pathway for inten-
sive court supervision in cases in which children are at greatest
risk, how and when reported children enter this pathway is crit-
ical for assessing our surveillance system’s capacity to identify
and protect maltreated children. Infants and young children not
only face a heightened risk of maltreatment relative to older
children (USDHHS, 2012) but the consequences of maltreat-
ment are also likely to be more severe. Indeed, previous
research has indicated that a report of maltreatment is an inde-
pendent risk factor for the most adverse outcome of injurious
and noninjurious death, particularly among infants and very
young children (Jonson-Reid et al., 2007; Putnam-Hornstein,
2011). Targeting resources to strengthen the capacity of the
CPS system to intervene when young children are reported is
critical. As a first step, reinforcing the capacity of reporters
to gather essential and usable evidence of maltreatment during
the course of their contact with high-risk populations may
ensure that children and families receive the protection and ser-
vices they need.
Acknowledgments
The authors would like to thank Mark Samples, Joseph Magruder, and
Barbara Needell for their contributions to preparing the analysis and
providing valuable feedback on the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to
the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for
the research, authorship, and/or publication of this article: Ongoing
support for the California Child Welfare Indicators Project is provided
by the California Department of Social Services and the Stuart Foun-
dation. Data linkages were funded through a grant received from the
Harry Frank Guggenheim Foundation.
References
Besharov, D. J., & Laumann, L. A. (1996). Child abuse reporting.
Society,33, 40–46. doi:10.1007/BF02700306
Child Abuseand Neglect Reporting Act, Cal. Penal Code § 11164 et seq.
(1974). Retrieved from Legislative Counsel of California website:
http://www.leginfo.ca.gov/cgi-bin/displaycode?section¼pen&grou
p¼11001-12000&file¼11164-11174.3
Cross, T. P., & Casanueva, C. (2009). Caseworker judgments and sub-
stantiation. Child Maltreatment,14, 38–52. doi:10.1177/1077559
508318400
Currie, J., & Widom, C. S. (2010). Long-term consequences of child
abuse and neglect on adult economic well-being. Child Maltreat-
ment,15, 111–120. doi:10.1177/1077559509355316
Dettlaff, A. J., Rivaux, S. L., Baumann, D. J., Fluke, J. D., Rycraft, J.
R., & James, J. (2011). Disentangling substantiation: The influence
of race, income, and risk on the substantiation decision in child
242 Child Maltreatment 18(4)
welfare. Children and Youth Services Review,33, 1630–1637. doi:
10.1016/j.childyouth.2011.04.005
Drake, B. (1995). Associations between reporter type and assessment
outcomes in child protective services referrals. Children and Youth
Services Review,17, 503–522. doi:10.1016/0190-7409(95)00036-C
Drake, B. (1996). Unraveling ‘‘unsubstantiated.’’ Child Maltreatment,
1, 261–271. doi:10.1177/1077559596001003008
Drake, B., & Jonson-Reid, M. (1999). Some thoughts on the increas-
ing use of administrative data in child maltreatment research. Child
Maltreatment,4, 308–315. doi:10.1177/1077559599004004004
Drake, B., & Zuravin, S. (1998). Bias in child maltreatment reporting:
Revisiting the myth of classlessness. American Journal of Orthop-
sychiatry,68, 295–304. doi:10.1037/h0080338
Eckenrode, J., Munsch, J., Powers, J., & Doris, J. (1988). The nature
and substantiation of official sexual abuse reports. Child Abuse &
Neglect,12, 311–319. doi:10.1016/0145-2134(88)90044-0
Eckenrode, J., Powers, J., Doris, J., Munsch, J., & Bolger, N. (1988).
Substantiation of child abuse and neglect reports. Journal of Con-
sulting and Clinical Psychology,56, 9–16. doi:10.1037/0022-
006X.56.1.9
English, D. J., Marshall, D. B., Coghlan, L., Brummel, S., & Orme, M.
(2002). Causes and consequences of the substantiation decision in
Washington state child protective services. Children and Youth Ser-
vices Review,24, 817–851. doi:10.1016/S0190-7409(02)00241-4
Fakunmoju, S. B. (2009). Substantiation and adverse appeal outcomes:
Content analysis and testing of Drake’s harm/evidence model. Child
Maltreatment,14, 53–68. doi:10.1177/1077559508320386
Giovannoni, J. M. (1995). Reports of child maltreatment from man-
dated and non-mandated reporters. Children and Youth Services
Review,17, 487–501. doi:10.1016/0190-7409(95)00035-B
Haskett, M. E., Wayland, K., Hutcheson, J. S., & Tavana, T. (1995).
Substantiation of sexual abuse allegations: Factors involved in the
decision-making process. Journal of Child Sexual Abuse,4, 19–47.
doi:10.1300/J070v04n02_02
Hussey,J.M.,Marshall,J.M.,English,D.J.,Knight,E.D.,Lau,
A. S., Dubowitz, H., & Kotch, J. B. (2005). Defining maltreat-
ment according to substantiation: Distinction without a differ-
ence. Child Abuse & Neglect,29, 479–492. doi:10.1016/j.
chiabu.2003.12.005
Jones, L. M., Finkelhor, D., & Halter, S. (2006). Child maltreatment
trends in the 1990s: Why does neglect differ from sexual and
physical abuse? Child Maltreatment,11, 107–120. doi:10.1177/
1077559505284375
Jonson-Reid, M., Chance, T., & Drake, B. (2007). Risk of death
among children reported for nonfatal maltreatment. Child Mal-
treatment,12, 86–95. doi:10.1177/1077559506296722
Kaplow, J. B., & Widom, C. S. (2007). Age of onset of child maltreat-
ment predicts long-term mental health outcomes. Journal of Abnor-
mal Psychology,116, 176–187. doi:10.1037/0021-843X.116.1.176
Kesner, J. (2008). Child protection in the United States: An examina-
tion of mandated reporting of child maltreatment. Child Indicators
Research,1, 397–410. doi:10.1007/s12187-008-9019-1
Kesner, J. E., & Robinson, M. (2002). Teachers as mandated reporters
of child maltreatment: Comparison with legal, medical, and social
services reporters. Children & Schools,24, 222–231. doi:10.1093/
cs/24.4.222
King, G., Trocme´, N., & Thatte, N. (2003). Substantiation as a multi-
tier process: The results of a NIS-3 analysis. Child Maltreatment,
8, 173–182. doi:10.1177/1077559503254143
Kohl, P. L., Jonson-Reid, M., & Drake, B. (2009). Time to leave sub-
stantiation behind: Findings from a national probability study.
Child Maltreatment,14, 17–26. doi:10.1177/1077559508326030
Lanier, P., Jonson-Reid, M., Stahlschmidt, M. J., Drake, B., &
Constantino, J. (2010). Child maltreatment and pediatric
health outcomes: A longitudinal study of low-income children.
Journal of Pediatric Psychology,35, 511–522. doi:10.1093/
jpepsy/jsp086
McDaniel, M. (2006). In the eye of the beholder: The role of reporters
in bringing families to the attention of child protective services.
Children and Youth Services Review,28, 306–324. doi:10.1016/j.
childyouth.2005.04.010
McCullagh, P., & Nelder, J. A. (1989). Generalized linear models
(2nd ed.). Boca Raton, FL: Chapman and Hall/CRC Press.
National Survey of Child and Adolescent Well-Being. (2007). Does
substantiation of child maltreatment relate to child well-being
and service receipt? (Research Brief; NSCAW, No. 9). Washing-
ton, DC: Administration for Children and Families.
Needell,B.,Webster,D.,Armijo,M.,Lee,S.,Dawson,W.,
Magruder, J., ... Nuttbrock, A. (2013). Child welfare service
reports for California. Retrieved from University at California
Berkeley Center for Social Services Research website: http://
cssr.berkeley.edu/ucb_childwelfare
Putnam-Hornstein, E. (2011). Report of maltreatment as a risk factor
for injury death: A prospective birth cohort study. Child Maltreat-
ment,16, 163–174. doi:10.1177/1077559511411179
Putnam-Hornstein, E., & Needell, B. (2011). Predictors of child pro-
tective service contact between birth and age five: An examination
of California’s 2002 birth cohort. Children and Youth Services
Review,33, 2400–2407. doi:10.1016/j.childyouth.2011.07.010
Putnam-Hornstein, E., Needell, B., King, B., & Johnson-Motoyama,
M. (2013). Racial and ethnic disparities: A population-based
examination of risk factors for involvement with child protective
services. Child Abuse & Neglect,37, 33–46. doi:10.1016/j.
chiabu.2012.08.005
Putnam-Hornstein, E., Webster, D., Needell, B., & Magruder, J.
(2011). A public health approach to child maltreatment surveil-
lance: Evidence from a data linkage project in the United States.
Child Abuse Review,20, 256–273. doi:10.1002/car.1191
Shonkoff, J. P., Richter, L., van der Gaag, J., & Bhutta, Z. A. (2012).
An integrated scientific framework for child survival and early
childhood development. Pediatrics,129, e460–e472. doi:10.
1542/peds.2011-0366
U.S. Department of Health and Human Services. (1999). Child mal-
treatment, 1998. Washington, DC: Children’s Bureau, Administra-
tion on Children, Youth, and Families.
U.S. Department of Health and Human Services. (2012). Child mal-
treatment: 2011. Washington, DC: Administration on Children,
Youth and Families, Children’s Bureau.
Wu, S. S., Ma, C.-X., Carter, R. L., Ariet, M., Feaver, E. A., Resnick,
M. B., & Roth, J. (2004). Risk factors for infant maltreatment: A
population-based study. Child Abuse & Neglect,28, 1253–1264.
doi:10.1016/j.chiabu.2004.07.005
King et al. 243
Wulczyn, F.,Barth, R. P., Yuan, Y.-Y. T., Harden, B. J., & Landsverk, J.
(2005). Beyond common sense: Child welfare, child well-being, & the
evidence for policy reform. New Brunswick, NJ: Aldine Transaction.
Zhou, Y., Hallisey, E. J., & Freymann, G. R. (2006). Identifying peri-
natal risk factors for infant maltreatment: An ecological approach.
International Journal of Health Geographics,5, 53. doi:10.1186/
1476-072X-5-53
Zou, G. (2004). A modified Poisson regression approach to prospec-
tive studies with binary data. American Journal of Epidemiology,
159, 702–706. doi:10.1093/aje/kwh090
244 Child Maltreatment 18(4)