You are on page 1of 17

Child Abuse & Neglect, Vol. 19, No. 2, pp.

233-249, 1995
Copyright 1995 ElsevierScience Ltd
Printed in the USA. All rights reserved
0145-2134/95 $950 + .00

Pergamon

0145-2134(94)00119-7

THE MEASUREMENT OF MALTREATMENT:


A COMPARISON OF APPROACHES
ROBIN A.

MCGEE, D A V I D

A. WOLFE,

SANDRA A. YUEN, SUSAN K. WILSON,

AND JEAN C A R N O C H A N
The University of Western Ontario, London. Ontario, Canada

A b s t r a c t - - T h i s study examined the comparability and predictive validity of three approaches to the measurement of
child maltreatment. Adolescents (N = 160, aged I I-17) were randomly selected from the open caseload of a child
protection agency. Global ratings of maltreatment severity were made by three reporting sources: researchers on the
basis of protection agency case files, protection agency social workers, and the adolescents themselves. Ratings were
made of five types of maltreatment: physical, sexual, emotional, neglect, and exposure to family violence. Selfreported (YSR) and caretaker-reported (CBCL) adjustment measures were also obtained for each subject. Results
indicated that over 90% of the sample had experienced more than one type of maltreatment. Comparison of ratings
across sources indicated considerable disagreement with respect to judgments of maltreatment occurrence and severity.
Relative to professional ratings, adolescent ratings were better predictors of externalizing and internalizing symptomatology in both univariate and multivariate analyses.

Key Words--Maltreatment, Measurement, Adolescents.

INTRODUCTION
THE DOCUMENTATION AND measurement of maltreatment is a critical issue in child
abuse and neglect research. Several authors have stressed the crucial importance of adequate
conceptual and operational definitions of maltreatment (Aber & Zigler, 1981; Besharov, 1981;
Cicchetti & Barnett, 1991; McGee & Wolfe, 1991b; Plotkin, Azar, Twentyman, & Perri,
1981). The purpose of the present study was to compare and contrast three approaches to
the measurement of maltreatment: ratings obtained through child protection social workers,
researchers of protection agency case history files, and adolescent victims. Each reporting
source was scrutinized with respect to the information it provided regarding maltreatment
occurrence, maltreatment severity, and potency in the prediction of externalizing and internalizing adjustment problems in adolescence.
In 1981, Aber and Zigler outlined three perspectives that have influenced the research
definition of maltreatment: the medical-diagnostic approach, the legal approach, and the sociological approach (Cicchetti & Barnett, 1991). In the medical approach, maltreatment is defined
by physical injury to the child (e.g., Kavanagh, 1982). In the legal approach, maltreatment is
The data for this study were part of the doctoral dissertation of the first author, who was supported by a Social
Sciences and Humanities Research Council of Canada Doctoral Fellowship. Data collection was supported by a grant
from the Medical Research Council of Canada to the second author.
Received for publication December 8, 1992; final revision received November 22, 1993, accepted November 29,
1993.
Requests for reprints should be addressed to Robin McGee, Ph.D., Child & Adolescent Mental Health Services, Valley
Regional Hospital, 150 Exibition St., Kentville, Nova Scotia, Canada B4N 5E3,
233

234

R. A. McGee,D. A. Wolfe, S. A. Yuen, S. K. Wilson, and J. Camochan

that which results in measurable serious harm to the child (although not necessarily physical
harm). In the sociological approach, maltreating acts are defined by society or the state.
The sociological perspective is the conceptual approach to definition employed most commonly in previous maltreatment research. Using this perspective, maltreatment is operationally
defined by a family's involvement with a child protection agency. Most commonly, the specific
type of maltreatment is defined by the official reason for service identified by the protection
agency (e.g., Wolfe & Mosk, 1983). Recent studies have extended this approach to include
severity ratings made by child protection workers (e.g., Strainer & Thieman, 1991; Wolfe,
Gentile, & Wolfe, 1989).
Although the sociological approach has the advantage of ensuring social relevance, operational definitions based on this approach will vary across nations and cultures. In an effort to
improve reliability, some researchers have sought to quantify maltreatment on the basis of
protection agency records (e.g., Barnett, Manly, & Cicchetti, 1991). Researchers may not rely
solely on state-derived definitions; rather, they seek to develop their own criteria for the
operational definition of maltreatment using previous theory and research.
Recently, a new approach to measuring maltreatment has emerged. In this "subjective"
tradition, maltreatment is defined by the victim. Maltreatment is operationally defined by
respondents' self-report of occurrence and severity, rather than "objective" criteria dictated
by legal or statutory requirements. Self-report methods of quantifying maltreatment are most
evident in retrospective studies of the maltreatment history of adults (e.g, Briere & Runtz,
1988; Rausch & Knutson, 1991) and adolescents (Powers, Eckenrode, & Jaklitsch, 1990;
Stiffman, 1989). Typically, this approach has involved global severity ratings (e.g., Ney,
Moore, McPhee, & Trought, 1986) or frequency ratings of specific items (e.g., Berger, Knutson,
Mehm, & Perkins, 1988).
Regardless of the conceptual approach employed, much of the previous research has operationally defined maltreatment using broad labels (e.g., "physically abused") to categorize
children. An occurrence/nonoccurrence judgment is made on the basis of information from a
reporting source. This practice suffers from two important limitations. First, it disregards the
heterogeneity and co-occurrence of maltreatment--most maltreated children experience more
than one type of abuse. Second, it ignores the variance in outcome attributable to maltreatment
severity. These problems have led several authors (e.g., Barnett et al., 1991; McGee & Wolfe,
1991a, 1991b) to argue for the continuous measurement of maltreatment using designs that
acknowledge the reality of multiple maltreatment experiences.
Another methodological consideration in maltreatment impact research concerns the operational definition of "outcome." The legal approach to maltreatment definition suggests that
psychological harm to the child should be quantified using standardized, norm-referenced
measures of adjustment. Unfortunately, much maltreatment research has employed arbitrary
indices of behavior and emotional problems (e.g., Vissing, Straus, Gelles, & Harrop, 1991).
The source that reports on victims' adjustment has also varied across studies. Operational
definitions of "adjustment problems" have included opinions of social workers (e.g., Conte &
Schuerman, 1987), impressions of trained observers (e.g., Claussen & Crittenden, 1991), parentreport measures (e.g., Jaffe, Wolfe, Wilson, & Zak, 1986), and self-report measures (e.g.,
Briere & Runtz, 1988). As meta-analytic research has revealed considerable discrepancies
between self- and third-party reports of adjustment (Achenbach, McConaughy, & Howell,
1987), equivocal findings in the previous literature may be due to this plethora of measurement
approaches.
Despite the increasing sophistication of our measurement practices, there is no information
on the basic concordance or equivalence of different approaches to maltreatment measurement.
Multimethod quantification is vital to the establishment of validity in nonexperimental research

Measurement of maltreatment

235

(Cook & Campbell, 1979). The present investigation sought to represent the major conceptual
traditions while simultaneously refining the operational definitions of maltreatment and adjustment. The study compared two operational definitions from the sociological tradition with one
from the "subjective" self-report tradition. In consideration of the sociological tradition,
subjects were adolescents from a child protection agency. To assess the state-derived sociological approach, global ratings of the severity of adolescents' maltreatment experiences were
obtained from the adolescents' protection workers. To assess a more research-based sociological perspective, global ratings were also obtained by researchers who rated the protection
agency files. In consideration of the "subjective" tradition, global ratings were also obtained
from the adolescent him or herself. To ensure comprehensive measurement, each of five
types of maltreatment were measured distinctly and continuously: physical, sexual, emotional,
neglect, and exposure to family violence. Standardized norm-referenced measures of symptomatology were employed to ensure reliability, validity, and clinical relevance. To avoid monosource bias, adolescent adjustment was evaluated using two sources: the adolescent's caretaker
and the adolescent him or herself.

METHOD

Subjects
Two hundred and fifty-nine adolescents were invited to participate in the study. All adolescents were randomly selected from the open caseload of a Canadian child protection agency.
A control group could not be employed in the study design, as nonagency youth do not have
records describing their maltreatment histories. Apart from age, there were no other selection
criteria for potential participants; that is, they were not selected on the basis of maltreatment
history or any other characteristic. One hundred and sixty-two adolescents elected to participate
in the study. Of these, two teens returned unusable data. Thus, 160 adolescents (70 males, 90
females) were retained in the study (62% participation). Subjects were between the ages of
11 and 17 years, with a mean age of 13.8 years. The data presented were part of a larger study
of the impact of maltreatment on adolescent adjustment.
Approximately half the sample (56%) resided with their biological mother at the time of
recruitment, a quarter (23%) were in agency care, and the remainder lived with other relatives.
The adolescents and/or their families had been involved with the protection agency for an
average of 6 years. Most of the respondents (69.4%) had entered agency care at least once.
The average age at which subjects first entered care was 8.8 years for an average duration of
12.5 months. The adolescents and/or their families were involved with the agency for a wide
variety of family problems. For each family, the child protection agency codes an "official
reason for service." Based on this code, most subjects were serviced for parent-child conflict
(47%), followed by neglect (17%), physical abuse (15%) and personal counselling (12%).
Smaller proportions were coded as sexual abuse (6%), family violence (2%), or emotional
abuse (1%).

MEASURES AND PROCEDURES

Maltreatment Ratings
In the protection agency, each youth is assigned a primary social worker who is responsible
for protection as well as other service issues. The worker is also required to maintain a case

236

R. A. McGee, D. A. Wolfe, S. A. Yuen, S. K. Wilson. and J. Carnochan

file on the adolescent. Global severity ratings were obtained from each reporting source (social
workers, file researchers, and adolescent victims) using the same rating scale. Each source was
asked to rate "the extent to which the adolescent had experienced five types of maltreatment":
physical, sexual, emotional, neglect, and exposure to family violence. Maltreatment experiences
were rated along a four-point scale: 0 ("not at all"), 1 ("mildly"), 2 ("moderately"), and 3
("severely"). Separate ratings were made for each of three possible perpetrators: biological
mother, biological father, or "other" (step, adoptive, and foster parents were considered to
be "other"). A score for each type of maltreatment was obtained by taking the maximum
rating across perpetrators.
File researcher ratings. Two file researchers were employed in the current study: the first
author and a trained research assistant with 3 years experience in child protection. The criteria
employed by the researchers in determining severity ratings were based on a system developed
for a more complex and detailed measurement instrument. On the basis of conceptual arguments
made elsewhere (McGee & Wolfe, 1991a, 1991b), researchers' ratings were based upon the
parental behavior described in the case file, not its impact on the child. This fundamental
principle was employed to avoid the conceptual tautology inherent in defining maltreatment
by adjustment. Each maltreatment type was defined by gradations along a specified dimension.
Thus, severity of sexual abuse was defined by gradations in penetration and force (Russell,
1983). Family violence and physical maltreatment were quantified by the degree of violence
employed (Straus, 1979). Neglect was quantified by the pervasiveness and extent of the violation of basic needs (Zurvain, 1991). Emotional maltreatment was defined as verbal abuse and
aggression (McGee & Wolfe, 1991b; Vissing et al., 1991). Severity of emotional maltreatment
was defined through minor insensitivity through to direct assaults on the child's sense of self
or safety. For all maltreatment types, chronicity and frequency of the caretaker behaviors were
often taken into account when making the rating. The specific criteria were determined through
a priori consultation and consensus with experts, and established through practice using actual
and hypothetical case information. The coding scheme employed by the file researchers is
described in the Appendix.
The names of potential participants were selected at random from the open caseload of the
protection agency. Once a name was selected, the research coordinator contacted the youth's
primary social worker. The worker could elect to approach the family personally regarding
participation. Alternatively, the worker could authorize sending the family a letter that described
the project and indicated that research personnel would telephone them shortly regarding their
willingness to participate. The research was explained to parents and teens as a study on the
relationship between background life events in childhood and current adjustment in adolescence. If family members were willing to participate, a research assistant visited them at home.
During this visit, the assistant obtained informed consent from both the adolescent and the
parent.
Once an adolescent and the caretaker agreed to participate, one of the two researchers
examined the target adolescent's protection agency case history file. These files included
caseworker notes, police reports, assessments compiled by intervening professionals (e.g.,
physicians, psychologists, public health nurses), foster or residential care progress reports,
juvenile and family court summaries, school records, documents pertaining to legal issues
(e.g., wardship status), all the agency's correspondence regarding the adolescent, and any other
relevant material. Using this wealth of information, the researchers completed a detailed measure summarizing the adolescent's maltreatment history. They then made global ratings of
each type of maltreatment, using the format described previously. The first author and the
assistant reviewed 40% and 60% of the cases, respectively. For 20% of the sample, both

Measurement of maltreatment

237

researchers rated the same cases: interrater reliability (Pearson r) on the global ratings was
very good. They were as follows: .79 for both neglect and emotional maltreatment, .87 for
family violence, .89 for physical maltreatment, and .96 for sexual abuse.

Social worker ratings. After completing the file ratings, a researcher met with the primary
protection worker of the target adolescent. The protection worker was asked to complete the
global ratings of the adolescent's maltreatment experiences. Although the worker made his or
her ratings in the presence of a file researcher, the worker was blind to the ratings made by
the researcher. Workers were not given the severity definitions developed by the researchers;
rather, they were asked to complete the ratings on the basis of their professional judgment
and training (and, of course, their experience with the target family). This was done to allow
for an uncontaminated estimate of the state-determined sociological approach to the definition
of maltreatment severity.
Approximately 40 protection workers were involved in the study. On average, workers had
worked with the target adolescent for 18 months (SD = 14.02). Workers also rated how well
they knew the case, on a scale of 1 ( " d o not know at all") to 5 ("know extremely well").
For the majority of the 160 subjects (58.8%), the workers indicated that they knew the case
very well or extremely well. For an additional 33.1%, workers knew the case somewhat well.
The remainder (8.1%) were rated as cases of which the worker knew very little. In no instance
did a worker indicate that he or she did not know a case at all.
Unfortunately, test-retest and interrater reliability estimates could not be obtained for agency
social workers. Because the amount of time agency personnel could spend on the project was
limited, and because they were already extensively involved in other project tasks, it was not
feasible to have them rate each other's cases or to rate several cases twice. However, previous
psychometric research has suggested that protection workers are sensitive to differences between cases and substantially agree on maltreatment when examining the same case (Alter,
1985; McGee & Wolfe, 1990).

Adolescent ratings. Each target adolescent was individually interviewed by the two researchers.
This interview occurred approximately 2 weeks after the researchers and social workers had
made their ratings. During the interview, each subject completed the global ratings of his or
her maltreatment experiences. Adolescents were not specifically instructed in the definitions
developed by the researchers to ensure that the "subjective" measurement tradition was
represented without undue influence. Nevertheless, in the interests of ensuring comprehension
and reliable interpretation across adolescents, each type of maltreatment included brief and
broad behavioral examples. The examples were as follows: physical maltreatment was
prompted as "hit or slapped;" sexual maltreatment was "touched in a sexual way that made
you uncomfortable;" exposure to family violence was "witnessing physical fighting between
parents and/or their partners;" emotional maltreatment was "being criticized, yelled at, or
treated unfairly;" and neglect was "not being cared for properly (e.g., lack of food, medicine),
ignored, or not paid attention to." To avoid the minimization that can occur when experiences
are labelled " a b u s e " (Berger et al., 1988), prompts referred to the experiences as "maltreatment."
Test-retest reliability of adolescent ratings was established on a subsample of 33 subjects
(20% of the sample). The retest interval ranged from 10 days to 113 days, with a mean of 31
days. Retest reliability coefficients (Pearson r) were very good, and were as follows: .70
(emotional), .89 (neglect), .90 (physical), .92 (family violence), and .93 (sexual).

238

R.A. McGee, D. A. Wolfe, S. A. Yuen, S. K. Wilson, and J. Carnochan

Adjustment Measures
Two measures were used to quantify internalizing and externalizing symptomatology. These
measures were completed by the adolescent and the caretaker during a home visit that took
place 1 week prior to the personal interview with the adolescent.

The Child Behavior Checklist. The CBCL (Achenbach & Edelbrock, 1983; Achenbach, 1991a)
is a well-standardized measure of the social competence and behavior problems of children
aged 4 to 18. It is completed by the child's primary caretaker. The CBCL yields an Externalizing
and Internalizing scale. The present study employed the 1991 norms for this instrument. The
CBCL has excellent test-retest reliability, with an mean test-retest reliability for behavior
problems of .89 over a 1-week interval. The mean stability coefficient for behavior problem
scores was .75 at 1 year and .71 at 2 years. Mean interparent agreement ranged from .65 to
.75. The CBCL has also demonstrated excellent construct validity, correlating .82 with the
Conner's Parent Questionnaire and .81 with the Quay-Peterson Revised Behavior Problem
Checklist (Achenbach, 1991a).
The Youth Self-Report. The YSR (Achenbach & Edelbrock, 1987; Achenbach, 1991b) is a
self-report measure of social competence and behavioral difficulties appropriate for adolescents
aged 11 to 18. It is virtually identical to the CBCL in format and content, but is completed
by the adolescent. Like the CBCL, the YSR was reformed in 1991 (Achenbach, 1991b). On
the new revision, the subscales that comprise the Internalizing and Externalizing factors are
entirely consistent with those on the CBCL. Reliability estimates at 1-week retest for the YSR
total scores are very good, ranging from .79 to .81. Longer-term stability is moderate, with a
mean of .49 at 7-month retest (Achenbach, 1991b). In the present study, CBCL and YSR
internalizing and externalizing scales correlated moderately with their counterparts on the other
measure, at .36 and .41 respectively; these findings are consistent with those reported by
Achenbach (1991 b).

RESULTS
Analyses are reported in two sections. The first section concerns the degree of concordance
among reporting sources; that is, the degree of agreement regarding maltreatment occurrence,
and the agreement regarding maltreatment severity. The second section concerns the predictive
validity of the measurement approaches. The multivariate and univariate correlation of maltreatment ratings from each reporting source with each adjustment measure are reported. Also,
the increment in prediction of adjustment due to one reporting source relative to the others
was explored using hierarchical regression analyses.

Concordance Between Reporting Sources


Occurrence. For each reporting source, for each type of maltreatment, ratings were coded into
occurrence/nonoccurrence judgments. A nonoccurrence judgment was coded when the source
rated the maltreatment dimension as "not at all" for all possible perpetrators. Table 1 reports
the percentage of the sample experiencing each type of maltreatment, according to each source.
It illustrates the extent of multiple maltreatment in a child protection population. According
to the "official" sources (file researchers and social workers), more than one-third of the
sample had been sexually abused, two-thirds had been physically maltreated and/or witnesses
to family violence, nearly 80% had experienced neglect, and virtually all (approximately 90%)

Measurement of maltreatment

239

Table 1. Percentage of the Sample Experiencing Maltreatment, According to Each


Reporting Source
Percentage Maltreated

Maltreatment
Type
Social Workers
Physical
Family violence
Sexual
Emotional
Neglect
File Researchers
Physical
Family violence
Sexual
Emotional
Neglect
Adolescents
Physical
Family violence
Sexual
Emotional
Neglect

By
Biological
Mother

By
Biological
Father

By
Other

Overall

47.5
58.7
0.0
76.9
64.4

31.2
40.0
8.1
51.9
45.6

31.9
37.5
32.5
45.0
25.6

73.1
61.3
37.5
94.4
80.6

45.0
60.6
0.6
76.9
77.5

25.6
40.6
10.0
52.5
49.5

30.0
34.4
34.4
45.0
32.5

68.8
61.3
39.4
91.9
87.5

49.4
46.9
0.6
53.7
35.6

37.5
32.5
7.5
41.3
36.2

41.3
30.0
28.7
42.5
27.5

83.8
53. l
34.4
81.9
62.5

N = 160.

had been emotionally maltreated. Moreover, according to official sources, 36% of the protection
agency youth had experienced four of the five types of maltreatment, and 20% had experienced
them all. Only 6.3% of the sample had encountered one type of victimization alone. It is worth
noting that the "official reason for service code" supplied by the agency greatly underestimated
the co-occurrence of maltreatment in these cases.
Percentage agreement regarding the occurrence of each maltreatment type is presented in
Table 2. The percentages reported include agreement regarding both presence or absence of
maltreatment. Concordance levels are greatest between social workers and file researchers,
who were in agreement on an average of 87% of the cases. Approximately three-quarters of the

Table 2. Percentage Agreement among Reporting Sources on the Occurrence/


Nonoccurrence of Maltreatment
Reporting Source
Maltreatment Type
Physical
Family Violence
Sexual
Emotional
Neglect
N = 160.

Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher

File Researcher

Adolescent

81.9

73.1
72.5
72.5
77.5
88.1
90.0
81.3
81.3
59.4
65.0

88.7
93. l
92.5
81.9

240

R.A. McGee, D. A. Wolfe, S. A. Yuen, S. K. Wilson, and J. Carnochan

time, adolescents agreed with the occurrence judgment of social workers and/or file researchers.
Concordance between the adolescent and official sources is greatest for sexual abuse (over
90%), and poorest for neglect (approximately 60%). Thus, reporting sources are in agreement
whether maltreatment has occurred in a background of a given adolescent the majority of
the time.
However, there is also considerable disagreement about the occurrence of maltreatment
between the reporting sources. Table 3 illustrates the percentage of disagreements where the
adolescent denied documented abuse or indicated the presence of a type of maltreatment
unknown to the agency. As can be seen, patterns of disagreement vary with the type of
maltreatment. For sexual abuse, the disagreement was fairly low, with adolescent denials
(7.5%) slightly outnumbering new disclosures (2.5%). For physical maltreatment, 18.8% of the
adolescents reported being physically maltreated, even though the agency had no information of
this kind. However, for all other types of maltreatment, adolescents tended to deny documented
occurrences. Over 16.3% denied family violence, 14.4% emotional maltreatment, and 29.4%
neglect--all in instances where social workers and agency records indicated that these forms
of maltreatment had occurred. These findings suggest considerable discrepancies in the perceptions of victims and official sources regarding the definition of maltreatment occurrence.

Severity. Multiple measures of the same construct (e.g., physical maltreatment) existed for
each subject. Therefore, repeated measures MANOVA was employed to compare overall mean
differences in maltreatment severity ratings. The results indicated mean differences in severity
ratings for physical maltreatment (F(2, 158) = 19.9, p < .0001), emotional maltreatment (F(2,
158) = 6.9, p < .001), and neglect (F(2, 158) = 11.7, p < .0001). Planned contrasts were
conducted to compare adolescent reports to those Of "official" sources (the mean ratings of
social workers and file researchers are averaged by the contrast procedure). The results indicated
that adolescents' ratings of physical maltreatment (M = 1.61) were greater than those of
officials (M = 1.20), p < .0001. Adolescents' ratings of emotional maltreatment (M = 1.61)
and neglect (M = 1.29), however, were lower than those of official sources (M = 1.95 and
M = 1.61, respectively, ps < .002). There were no mean differences between sources for
either family violence or sexual abuse. On average, therefore, adolescents appeared to regard

Table 3. Percentage Disagreement between Adolescents and Official Sources Regarding the OccurrencefNonoccurrence of Maltreatment
Type of Disagreement with
Adolescent

Maltreatment Type
Physical
Family Violence
Sexual
Emotional
Neglect
N = 160.

Official Source

Official Source
Yes/
Adolescent No

Official Source
No/
Adolescent Yes

Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher

8.1
6.3
18.8
16.3
7.5
7.5
15.6
14.4
29.4
30.0

18.8
21.3
8.8
6.3
4.4
2.5
3.1
4.4
11.3
5.0

241

Measurement of maltreatment
Table 4. Correlations Among Reporting Sources' Maltreatment Severity Ratings (All
Sources Agree that Maltreatment Occurred)

Reporting Source
File
Researcher

Maltreatment Type
Physical (N = 91)
Family Violence (N = 67)
Sexual (N = 47)
Emotional (N = 121)
Neglect (N = 79)

Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher
Social Worker
File Researcher

.63
.47
.80
.49
.52

Adolescent
.30
.31
.01
.06
.42
.37
.13
.16
.32
.02

their physical maltreatment experiences to be more severe than did official sources, but characterized their experiences of emotional maltreatment and neglect as milder.
Adolescents and official sources also disagreed substantially regarding the severity of the
adolescents' experiences, even when they agreed on occurrence. Table 4 shows the correlations
between adolescents and official sources regarding severity ratings when the sample is restricted
to those instances in which all sources agreed the maltreatment had occurred. The correlations
are all remarkably low. Particularly poor agreement was obtained on the severity of exposure
to family violence (r = .06). The best agreement was still modest (r = .42) for sexual abuse.
Indeed, it would almost appear that maltreated youth and official report sources are rating
different experiences.

Patterns among maltreatment types. To determine the dimensions along which the various
sources conceptualized maltreatment, maltreatment ratings within each source were subjected
to a principal components analysis with varimax rotation. A three-factor solution emerged for
each source. For file researchers and social workers, the first factor was comprised of physical
maltreatment, emotional maltreatment, and family violence while neglect and sexual abuse
emerged as the second and third factors, respectively. Among adolescents, the first factor was
comprised of physical maltreatment, emotional maltreatment, and neglect, with family violence
and sexual abuse emerging independently. It would appear that professionals conceptualized
maltreatment along the dimensions of aggression and violence, whereas adolescents conceptualized it along the dimension of parenting. Factor structures are based on the intercorrelations
of ratings within each source. For simplicity and ease of interpretation, these intercorrelations
are reported in Table 5. As can be seen, adolescents do not report a relationship between
emotional maltreatment and family violence, whereas social workers and file researchers do.
Emotional maltreatment was associated with family violence (r =.43) in the opinions of
professionals. However, among adolescents, emotional maltreatment correlated most highly
with physical maltreatment (r = .59) and neglect (r = .59), and had a poor relationship with
family violence (r = .18). Nevertheless, all sources perceived similar association between
physical and emotional maltreatment (r = .49 to .59), reflecting the co-occurrence of these
maltreatment forms. Thus, relative to adolescents, official sources appeared to have a different
perception of how maltreatment types cluster or covary.

242

R. A. McGee, D. A. Wolfe, S. A. Yuen, S. K. Wilson, and J. Carnochan


Table 5. Intercorrelations of Maltreatment Ratings within each Reporting Source

Maltreatment Type

Social Workers
Physical
Family Violence
Sexual
Emotional
File Researchers
Physical
Family Violence
Sexual
Emotional
Adolescents
Physical
Family Violence
Sexual
Emotional

Family Violence

Sexual

Emotional

Neglect

.38*

.29*
.17

.55*
.43*
.18

.29*
.23*
.14
.44*

.30*

.27*
.06

.49*
.42*
.28*

.30*
.27*
18
.50*

.27*

.18

.59*

.41"

.06

.18
.14

.23*
.11
.60*

N = 160.
* p < .005.

Predictive Validity of Each Reporting Source


Correlations with behavior problems. Professionals and adolescent victims may disagree about
the occurrence or nature of the maltreatment experience. Which reporting source best predicts
current adolescent adjustment problems?
Table 6 displays the univariate correlations obtained between ratings of each maltreatment
type and the four adjustment measures, for each of the three reporting sources. It also displays
the multivariate correlation obtained by simple regression in which all five maltreatment ratings
are used to predict each adjustment measure.
Some robust relationships were detected. Sexual abuse severity was significantly associated
with caretaker-reported (CBCL) internalizing across sources, and physical maltreatment was
associated with self-reported (YSR) internalizing by both researchers and adolescents. Generally, however, few significant correlations between maltreatment and adjustment resulted from
either social workers' or file researchers' ratings. In the multivariate context, the severity
ratings made by social workers were not associated with any measure of adjustment. Ratings
by file researchers fared somewhat better: in a multiple regression, they were able to predict
a significant amount of variance in caretaker-reported (CBCL) internalizing (R -- .29, p <
.05). On a univariate level, researchers' ratings of emotional maltreatment correlated .22 (p
< .01) with caretaker-reported (CBCL) internalizing.
In contrast, adolescent maltreatment ratings predicted a significant amount of variance in
three of the four adjustment measures: self-reported (YSR) internalizing (R = .38, p < .001)
and externalizing (R = .30, p < .01), as well as caretaker-reported (CBCL) internalizing (R
= .27, p < .05). At the univariate level, the largest correlations were obtained between selfreported (YSR) internalizing and severity ratings of emotional maltreatment (r = .37, p <
.001) and neglect (r = .25, p < .01),
Incremental predictive value. Hierarchical regressions were computed to determine if addition
of information from one reporting source improved prediction of adjustment beyond the information afforded by another reporting source. Controlling for the contribution of one reporting

Measurement of maltreatment

243

Table 6. Uulvariate and Multivariate Correlations of Maltreatment Ratings with Adjustment Measures

Adjustment Measures
Source
Social Workers

Researchers

Adolescents

Maltreatment
Type

CBCL
Internala

CBCL
Externalb

YSR
Internalc

Physical
Family violence
Sexual
Emotional
Neglect

.12
.10
.21"*
.11
.11

.12
.07
.15
.l 5
.00

.12
-.05
.11
.12
.11

-- .03
--. 10
.00
.01
--.03

Multiple R

.23

.22

.21

.13

Physical
Family violence
Sexual
Emotional
Neglect

.14
.04
.23**
.22**
.06

.15
.00
.18"*
.12
-.01

Multiple R

.29*

.25

Physical
Family violence
Sexual
Emotional
Neglect

.18**
.08
.18"*
.16
.04

Multiple R

.27*

.19"*
.05
.11
.12
.05
.20

YSR
Externald

.01
--.10
.00
--.02
--.09
.14

.05
.07
.13
.01
-.03

.22**
.11
.10
.37***
.25'*

--.02
.03
.05
.20**
.03

.17

.38***

.30**

*p < .05; **p < .01; ***p < .001.


N = 160.
a Child Behavior Checklist Internalizing scale.
b Child Behavior Checklist Externalizing scale.
Youth Self Report Internalizing scale.
d Youth Self Report Externalizing scale.

source, how much more variance in adjustment can be predicted if ratings from another source
are added to the regression equation? The results indicated that ratings made by adolescent
victims significantly improved prediction of self-reported behavior problems above and beyond
the ratings o f "official" sources. In contrast, ratings made by other sources never added unique
variance above that explained by adolescent ratings. Above and beyond social worker ratings,
adolescent ratings added 8% (p < .05) and 10% (p < .01) to the prediction o f self-reported
(YSR) externalizing and internalizing, respectively. Above and beyond file researcher ratings,
adolescent ratings added 9% (p < ,01) and 11% (.001) to self-reported (YSR) externalizing
and internalizing. Ratings by file researchers or social workers did not add significant variance
to the prediction of adjustment above that contributed by adolescent ratings, nor did they add
significant variance when controlling for each other.

DISCUSSION
The results underscore an important p o i n t - - a d o l e s c e n t s from a protection agency population
have typically experienced multiple maltreatment. In the present sample, over 90% experienced
more than one type. " P u r e " maltreatment types do not exist in reality. These findings highlight
the importance o f recognizing the complexity of maltreatment. Future researchers may need
to critically examine the utility of the " c a t e g o r y " approach to the measurement of mal-

244

R. A. McGee, D. A. Wolfe, S. A. Yuen, S. K. Wilson, and J. Carnochan

treatment. The "official" agency label for service, used so frequently in previous research,
can often be an extremely inaccurate and simplistic representation of the actual maltreatment
status of many children and youth. It behooves researchers to thoroughly document the maltreatment histories of potential subjects, and to use statistical or design controls (e.g., protection
agency control groups) to assess the unique effects of any given maltreatment type.
The present study conceptually defined maltreatment using the sociological and the subjective approaches. The results suggest that there are important conceptual and methodological
differences in the definitions employed by each reporting source and that these differences
have important implications when investigating the impact of maltreatment on adjustment.
Overall, there was more agreement than disagreement regarding the presence of various
types of maltreatment in the backgrounds of protection agency youth. Occurrence judgments
made by social workers and file researchers were often highly concordant. The extent of
concurrence of official sources with adolescents varied with the type of abuse under examination. Official sources and adolescents concurred the most regarding the presence of sexual
abuse, and the least regarding neglect.
There were also considerable discrepancies between reporting sources. Two kinds of discrepancy were examined: disagreement regarding maltreatment occurrence, and disagreement regarding maltreatment severity. With respect to occurrence, adolescents reported more physical
maltreatment but less family violence, emotional maltreatment, and neglect relative to the
official sources. When the entire sample was considered, disagreement regarding severity was
evident for physical maltreatment, emotional maltreatment, and neglect. When the sample
was restricted to those individuals for whom all sources agreed the maltreatment occurred,
disagreement regarding severity was evident for all maltreatment types in that the correlations
among severity ratings were remarkably low.
Disagreement among reporting sources is a longstanding finding in child psychopathology
research. Several studies document disagreement between children, parents, clinicians, and
public officials with respect to stressful life events (Johnson, 1986) and behavioral disturbance
(Achenbach et al., 1987). There are several reasons why official and subjective estimates of
maltreatment might disagree. Each reporting source may possess unique information. For
example, official sources often have access to information regarding experiences in the youth's
infancy and toddlerhood that the adolescent cannot remember, and youth will be privy to
family events that are not recorded by the agency. Also, each reporting source may employ
overlapping conceptual definitions that nevertheless have unique elements. The different pattern
of intercorrelations among maltreatment types found between official sources and adolescents
suggests that adolescents construct or conceptualize their experience differently than professionals. Adolescents appeared to view maltreatment along the dimension of inadequate parenting: their ratings of physical maltreatment covaried with neglect and emotional maltreatment.
However, social workers and file researchers appeared to view maltreatment along a dimension
of aggression: their ratings of physical, verbal, and spousal aggression covaried. Both professionals and adolescents regarded sexual abuse as unlike other maltreatment types.
Reporting sources may also employ different heuristics, standards, or thresholds for evaluating what is considered maltreatment. Mandated by law to focus on child protection, agency
personnel and their records will reflect statutory requirements when determining whether
maltreatment occurred. Moreover, they will define severity in relative terms. That is, they will
compare a given case to others they have seen. Lacking this basis for comparison, victims
will evaluate whether an event qualifies as maltreatment by the distinctiveness and painfulness
of the experience. For example, adolescents concurred with professionals most often regarding
the occurrence of sexual abuse, perhaps because all sources consider this maltreatment type
distinctive. However, agreement regarding sexual abuse severity was low. Examination of the

Measurement of maltreatment

245

distribution of self-reported sexual abuse shows that virtually all sexual abuse victims regarded
their experience as severe. Thus, even when the criterion for judging occurrence is similar
between sources, the heuristic for judging severity may differ.
Adolescents reported greater and more severe experiences with physical maltreatment than
did official sources. This finding may result from the more stringent standards for physical
maltreatment employed by professionals. Some protection-related laws emphasize physical
maltreatment that leaves marks or injuries; lesser types of maltreatment may not be recorded
by workers. Similarly, if a family is involved with the protection agency for reasons other
than physical abuse, this kind of information may be superseded in favor of more immediate
casework concerns. The possibility that adolescents were exaggerating their maltreatment
experiences was not supported by the relative minimization they exhibited with most other
types of abuse. Moreover, previous research has found minimization of physical abuse to be
common among self-reports of agency subjects (Femina, Yeager, & Lewis, 1990). Adolescents
may include a broader range of parental behaviors (e.g., slapping, spanking) in their definition
of physical maltreatment.
Relative to professionals, adolescents underreported the occurrence of emotional maltreatment and neglect. Verbal hostility has been found to be chronic among troubled families
(Burgess & Conger, 1978; Patterson, 1982). Some protection agency youth may fail to see
such interaction as abusive or even distinctive. Similarly, conditions of neglect and deprivation
may be so pervasive that some victims are unaware that better parenting exists. Agreement
regarding emotional maltreatment severity may be low because of the relatively greater sophistication of professionals' concepts of emotional maltreatment. Professionals may include such
subtle but damaging dynamics as parent-child role reversal and exposure to criminal influences
in their definition of emotional maltreatment--patterns that some victims may not recognize.
That is, whereas adolescents may reflect on hostile verbal exchange, the social worker and
file researcher may concentrate on dysfunctional and chronic family patterns. Similarly, the
lack of agreement regarding neglect severity may be due to differing breadth in definition.
Professionals may use specific statutory definitions to rate neglect, whereas adolescents appeared to have included emotional neglect and rejection in their definition. Neglect may be
regarded as more of a context, and not a set of behaviors or omissions. The fact that file
researchers and social workers disagreed between themselves most often regarding the occurrence of neglect illustrates the nebulous nature of this maltreatment type.
Approximately 17% of this sample denied the presence of documented domestic violence
in their backgrounds. Even when teens and professionals agree that family violence occurred,
their concordance regarding severity was virtually nonexistent. Some subjects may have denied
or minimized the family violence for fear that their parent would be criminally charged. Others
may have regarded such marital behavior as normative. Professionals quantified family violence
in terms of injury or severity of the violent act; adolescents may have quantified it in terms
of chronicity. Further multimethod research on the perceptions of teens regarding family
violence is warranted.
Another hypothesis for the discrepancies in occurrence for sexual and physical abuse bears
examination. Some memory theorists (Rubin & Kozin, 1984) have proposed that life events
involving physically painful stimuli are more deeply encoded and better recalled than less
physically threatening experiences. Noncontact maltreatment experiences such as neglect may
be more difficult to identify or remember. Because physical maltreatment and sexual abuse
involve physical contact, these forms of maltreatment may be better identified or recalled in
memory. Thus, adolescents may report such experiences more accurately.
Overall, subjective estimates of victimization appear to be more predictive of behavior
problems than objective estimates. Relative to those of professionals, victims' assessments of

246

R. A. McGee, D. A. Wolfe, S. A. Yuen, S. K. Wilson, and J. Carnochan

their own maltreatment severity were more predictive of current behavioral adjustment. Knowledge of "official" maltreatment added little to the prediction of adjustment once the victims'
perceptions were accounted for. Why were adolescents' perceptions of their maltreatment
histories more predictive of adjustment than those of professionals? Some of the predictive
potency of self-reported maltreatment may be due to shared method variance with self-reported
adjustment measures. However, that is not the whole story: Adolescents' reports of maltreatment were also associated with caretaker-reported behavior problems. Previous research
has emphasized that subjective appraisals of life events are critical to adjustment generally
(Lazarus & Folkman, 1984). In the stress literature, it has been demonstrated that individuals'
appraisals of stressors are more strongly associated with symptomatology than are estimates
made by " e x p e r t " third parties (Lazarus, 1974; Tennen & Affleck, 1990). Victims are in a
better position to evaluate the impact of a stressful experience and may incorporate impact
judgments into their definition of maltreatment. These findings underscore the importance of
exploring the phenomenology of child maltreatment, and the mechanisms by which children
and youth appraise and interpret their experiences.

Methodological considerations. Divergence between the reporting sources may also be due to
error within each source. A variety of factors can influence the information available to each
reporting source and hence the extent to which they will agree. Social worker ratings can be
undermined by the rapid job turnover among caseworkers, such that current workers are not
familiar with the early history of a case. Also, workers may be most concerned with service
provision to families on their caseload and less preoccupied with previously documented abuse.
As a worker's caseload increases, it becomes more difficult to know each case intimately.
Also, ongoing involvement with families may also influence how "abusive" a family is
regarded by the worker. That is, current receptivity or resistance of families to service may
influence a worker's perception of a child's background.
Ratings by trained researchers also have their pitfalls. Case files often do not contain detailed
information on the nature or degree of certain abuse types. This is particularly true for more
subtle kinds of maltreatment such as emotional maltreatment. Because exposure to wife assault
is not itself legally actionable as a protection issue, case files often do not describe the specific
acts of family violence. Global severity ratings can therefore become subjective and prone to
error and halo effects from other documented abuses. Often, files contain scant information
on those families who are relatively new to the agency. Information on siblings can be rare
when the investigation and casework have focused on only one child in the family. Retrospective reports provided by adolescents are influenced by the limitations of memory, denial, and
other biases. Adolescents' perceptions of maltreatment may vary over the course of their
development. Indeed, the relatively lower test-retest coefficient for emotional maltreatment (r
= .70) suggests that this maltreatment type is most subject to variability over time, and may
be more likely to be influenced by recent events.
On the basis of the present investigation, it appears that ratings provided by file researchers
possess a slight edge over social worker ratings in the prediction of adolescent behavioral
adjustment. This may occur because file judgments are likely to be more reliable, and hence
have a smaller margin of measurement error.
The present study sought to compare and contrast the "real life" perceptions of researchers,
social workers, and adolesce~ats, respectively. For that reason, each source was free to employ
its own definition of abuse occurrence and severity. Of course, if all parties had been provided
with explicit definitions and inculcated regarding their application, agreement would have
increased. If maximizing cross-source agreement is important to future research questions,
researchers ought to fumish specific and consistent criteria to all raters.

Measurement of maltreatment

247

The use of a 4-point maltreatment scale may have resulted in some restriction of range, thereby
reducing the size of the correlations in this study. The fact that significant relationships were
obtained in this study, even with such restricted variables, speaks to the potency of child maltreatment. To ensure greater variability, future studies might employ a global scale with greater
potential range.
Which reporting source provides the most valuable approach to the measurement of maltreatment? The answer depends on the research question, particularly the criterion one seeks to
predict. Each reporting source may be properly regarded as providing a "window" into the
experience and impact of maltreatment. If a different criterion was employed (e.g., parenting
competence), different results might obtain. However, because previous research has demonstrated
that self-reported adjustment often differs from other-reported adjustment (Achenbach et al., 1987),
and because both approaches are important to the understanding of mental health, using multiple
sources to describe maltreatment and adjustment is highly recommended for future research.
Taken together, the results suggest that the conclusions one draws regarding maltreatment
occurrence and severity depends upon whether one employs a sociological or subjective approach
to the measurement of maltreatment. In summary, the study demonstrated that youth experience
a high degree of multiple maltreatment, regardless of reporting source. "Subjective" adolescent
reports differed from the "sociological" state-derived or research-derived reports with respect to
maltreatment occurrence and severity. Moreover, adolescent reports had greater predictive validity
relative to official sources in the prediction of standardized self- and caretaker-reported adjustment
measures. The findings underscore that researchers should be aware of the limitations of the
methods they use and recognize that the identification of abuse occurrence and severity will vary
with the reporting source and the type of maltreatment under consideration.
Acknowledgement Thanks to Douglas Barnett for his comments on an earlier draft of this paper.

REFERENCES
Aber, J. L., & Zigler, E. (1981). Developmental considerations in the definition of child maltreatment. In R. Rizley &
D. Cicchetti (Eds.), Developmentalperspectives in child maltreatment (pp. 1-29). San Francisco, CA: Jossey-Bass.
Acbenbach, T. M. (1991a). Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT:
University of Vermont.
Achenbach, T. M. (1991b). Manual for the Youth Self Report and 1991 Profile. Burlington, VT: University of Vermont.
Achenbach, T. M., & Edelbrock, C. S. (1983). Manual for the Child Behavior Checklist and Child Behavior Profile.
Burlington, VT: University of Vermont.
Acbenbach, T. M., & Edelbrock, C. S. (1987). Manual for the Youth Self-report and Profile. Burlington, VT: University
of Vermont.
Achenbach, T. M., McConaughy, S. H., & Howell, C. T. (1987). Child/adolescent behavioral and emotional problems:
Implications of cross-informant correlations for situational specificity. Psychological Bulletin, 101, 213-232.
Alter, C. F. (1985). Decision-making factors in cases of child neglect. Child Welfare, 64, 99-111.
Barnett, D., Manly, J. T., & Cicchetti, D. (1991). Continuing toward an operational definition of psychological
maltreatment: A commentary. Development and Psychopathology, 3, 19-29.
Berger, A. M., Knutson, J. F., Mehm;J. G., & Perkins, K. A. (1988). The self-report of punitive childhood experiences
of young adults and adolescents. Child Abuse & Neglect, 12, 251-262.
Besharov, D. J. (1981). Toward better research on child abuse and neglect: Making definitional issues an explicit
methodological concern. Child Abuse & Neglect, 5, 383-390.
Briere, J., & Runtz, M. (1988). Multivariate correlates of childhood psychological and physical maltreatment among
university women. Child Abuse & Neglect, 12, 331-341.
Burgess, R., & Conger, R. (1978). Family interaction in abusive, neglectful, and normal families. Child Development,
19, 1163-1173.
Cicchetti, D., & Barnett, D. (1991). Toward the development of a scientific nosology of child maltreatment. In D.
Cicchetti & W. Grove (Eds.), Thinking clearly about psychology: Essays in honor of Paul E. Meehl (pp. 346-377).
Minneapolis, MN: University of Minnesota Press.

248

R.A. McGee, D. A. Wolfe, S. A. Yuen, S. K. Wilson, and J. Carnochan

Claussen, A. H., & Crittenden, P. M. (1991). Physical and psychological maltreatment: Relations among types of
maltreatment. Child Abuse & Neglect, 15, 5-18.
Conte, J. R., & Schuerman, J. R. (1987). Factors associated with an increased impact of child sexual abuse. Child
Abuse & Neglect, 11, 201-211.
Cook, T. D., & Campbell, D. T, (19;79). Quasi-experimentation: Design and analysis issues forfield settings. New
York: Rand-McNally.
Femina, D. D., Yeager, C. A., & Lewis, D. O. (1990). Child abuse: Adolescent records versus adult recall. Child
Abuse & Neglect, 14, 227-231.
Jaffe, P., Wolfe, D. A., & Wilson, S., & Zak, L. (1986). Family violence and child adjustment: A comparative analysis
of girls' and boys' behavioral symptoms. American Journal of Psychiatry, 143, 74-77.
Johnson, J. H. (1986). Life events as stressors in childhood and adolescence. Beverly Hills, CA: Sage.
Kavanagh, C. (1982). Emotional abuse and mental injury: A critique of the concepts and a recommendation for
practice. Journal of the American Academy for Child Psychiatry, 21, 171-177.
Lazarus, R. S. (1974). Psychological stress and coping in adaptation and illness. International Journal of Psychiatry
in Medicine, 5, 321-333.
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer.
McGee, R. A., & Wolfe, D. A. (1990). Development of a record of maltreatment experiences. Toronto, Canada: The
Institute for the Prevention of Child Abuse.
McGee, R. A., & Wolfe, D. A. (1991a). Between a rock and a hard place: Where do we go from here in defining
psychological maltreatment? Development and Psychopathology, 3, 119-124.
McGee, R. A., & Wolfe, D. A. (1991b). Psychological maltreatment: Towards an operational definition. Development
and Psychopathology, 3, 3-18.
Ney, P. G., Moore, C., McPhee, J., & Trought, P. (1986). Child abuse: A study of the child's perspective. Child
Abuse & Neglect, 10, 511-518.
Patterson, G. R. (1982). Coercive family process. Eugene, OR: Castalia.
Plotkin, R. C., Azar, S., Twentyman, C. T., & Pen-i, M. G. (1981). A critical evaluation of the research methodology
employed in the investigation of causative factors of child abuse and neglect. Child Abuse & Neglect, 5, 449-455.
Powers, J. L., Eckenrode, J., & Jaklitsch, B. (1990). Maltreatment among runaway and homeless youth. ChildAbuse &
Neglect, 14, 87-98.
Rausch, K., & Knutson, J. F. (1991). The self-report of personal punitive childhood experiences and those of siblings.
Child Abuse & Neglect, 15, 29-36.
Rubin, D. C., & Kozin, M. (1984). Vivid memories. Cognition, 16, 81-95.
Russell, D. E. H. (1983). The incidence and prevalence of intrafamilial and extrafamilial sexual abuse of female
children. Child Abuse & Neglect, 7, 133-146.
Stiffman, A. R. (1989). Physical and sexual abuse in runaway youths. ChildAbuse & Neglect, 13, 417-426.
Strainer, K. E., & Thieman, A. (April, 1991). The relation of child, parent, and family characteristics to the severity
of child maltreatment. Paper 15resented at the biennial meeting of the Society for Research in Child Development,
Seattle, WA.
Straus, M. A. (1979). Measuring intrafamily conflict or violence: The Conflict Tactics (CT) Scales. Journal of Marriage
and the Family, 41, 75-88.
Tennen, H., & Affleck, G. (1990). Blaming others for threatening life events. Psychological Bulletin, 105, 209-232.
Vissing, Y. M., Straus, M. A., Gelles, R. J., & Harrop, J. W. (1991). Verbal aggression by parents and psychosocial
problems of children. Child Abuse & Neglect, 15, 223-238.
Wolfe, D. A., & Mosk, M. D. (1983). Behavioral comparisons of children from abusive and distressed families.
Journal of Consulting and Clinical Psychology, 51, 702-708.
Wolfe, V. V., Gentile, C., & Wolfe, D. A. (1989). The impact of sexual abuse on children: A PTSD formulation.
Behavior Therapy, 20, 215-228.
Zurvaln, S. J. (1991). Research definitions of child physical abuse and neglect: Current problems. In R. Starr & D,
Wolfe (Eds.), The effects of child abuse and neglect: Issues and research (pp. 100-128). New York: Guilford
Press.

Rtsumf--Cette 6tude a voulu d&erminer si trois mtthodes pour mesurer la maltraitance 6taient comparables et valides
en tam qu'instruments pour prtdire les mauvais traitements. Dans une agence de protection de l'enfance, on a choisi
160 adolescents figts de I 1 h 17 ans, de fa~on altatoire, comme sujets pour l'ttude. On a demand6 h trois groupes
de personnes de coter la stvtrit6 des mauvais traitements: les chercheurs qui ont 6tudi6 les dossiers, les agents
responsables de la protection des jeunes et les adolescents eux-mSmes. On a class6 les mauvais traitements selon cing
cattgories: physiques, sexuels, affectifs, ntgligence et t6moins de violence dans la famille. On a ajust6 les donntes
selon que les situations 6talent dtvc~ltes par les jeunes ou par ceux qui avaient les jeunes h leur charge. Les rtsultats
ont dtmontr6 que plus de 90 p.c. des jeunes darts l'tchantillon avaient vtcu plus d'une forme de mauvais traitements.
En comparant les cotes venant des trois groupes, on a not6 un manque notable de consensus par rapport h la dtcision
de considtrer un incident comme de la maltraitance et par rapport/~ sa gravitt. Les cotes accordtes par les jeunes
eux-m~mes, compartes ~tcelles des professionnels, se sont av6rtes d'une plus grande valeur pour prtdire des sympt6mes
exttriorists et inttriorists, tant dans les analyses h variables simples que les analyses h variables multiples.

Measurement of maltreatment

249

Resumen--Este estudio examina la comparatividad y validez predictiva de tres enfoques a la medici6n del maltrato
a los nifios. Se seleccionaron al azar adolescentes (N = 160, 11-17 de edad) de los casos tratados en una agencia de
protecci6n infantil. Evaluaciones globales de la severidad del maltrato se realiz6 por tres fuentes de reporte: de
investigadores en base a los archivos de la agencia de protecci6n, de los trabajadores sociales de la agencia de
protecci6n infantil, y de los mismos adolescentes. Las evaluaciones se hicieron sobre cinco tipos de maltrato: fisico,
sexual, emocional, negligencia, y la exposici6n a violencia familiar. Tambirn se obtuvieron para cada sujeto ajustes
de medida con auto-reportes (YSR) y reportes del cuidador. Los resultados indicaron que m~is del 90% de la muestra
habia sufrido la experiencia de mas de un tipo de maltrato. Las comparaciones de las evaluaciones de las diferentes
fuentes indicaron considerable desacuerdo respecto a los juicios sobre la ocurrencia y la severidad del maltrato. En
relaci6n a las evaluaciones de los profesionales, las evaluaciones de los adolescentes eran mejores predictores de
sintomatologias externas e internas tanto en los an~lisis univariados y los multivariados.

APPENDIX:

CODING

SCHEME

EMPLOYED

BY FILE RESEARCHERS

Neglect was defined as acts of omission that deprive the child of fundamental needs (Zurvain, 1991).
1. Mild neglect was conceptualized as lapses in parenting that could be painful to the child. Examples included failure
to: assist the child with important tasks, compliment the child, show respect for the child's opinions, encourage
peer activity, be generally attentive, offer comfort, and/or spend time with the child.
2. Moderate items referred to acts of omission that put the child at risk for developmental deviation, and concerned
parental consistency and availability. Examples included failure to: provide regular routines, provide stimulation,
follow through on opportunities to enhance the child's development (e.g., therapy), provide consistent discipline,
and ensure attendance at school.
3. "Severe" neglect referred to parental acts of omission that put the child at physical risk for harm. Examples
included failure to: feed the child appropriately, protect from dangerous situations, provide proper medical attention,
provide proper supervision, clothe the child appropriately, protect from abusive adults, and keep the home environment safe.
Emotional maltreatment was defined as parental communications that could be damaging to the child's development
(McGee & Wolfe, 1991b).
1. Mild psychological maltreatment referred to indirect communications that represent lapses in adequate parenting.
Examples included: belittling the child's feelings, giving the child silent treatment, comparing the child to disliked
others, refusing to discuss issues of concern to the child, and denigrating others the child cares for.
2. Moderate psychological maltreatment referred to indirect communications regarding the child's worth: blaming
the child, denying the child's reality, placing the child in role reversal, punishing the child for failure to meet
excessive expectations, ridiculing the child, scapegoating the child for family problems, speaking to the child in
hostile or sarcastic manner, having emotional outbursts in front of the child, exposing the child to criminal
influences, being unpredictable in discipline, and threatening other family members.
3. Severe psychological maltreatment was considered those parental acts that represent direct attacks on the child's
sense of self or safety. Examples included: telling the child s/he is unwanted, destroying something the child
values, denigrating the child, threatening with extreme punishment, threatening to kill or abandon, using extreme
and humiliating punishment, and deliberate attempts to terrify.
Exposure to Family Violence was defined as exposure to physical violence between parents and/or parents and their
partners. It was quantified in terms of the degree of violence employed.
1. Noncontact experiences (e.g., smashing, hitting, or kicking something).
2. Contact experiences that are typically nonlethal (e.g., pushing partner, slapping partner, and throwing something
at partner).
3. Potentially damaging or lethal violence (e.g., beating partner, threatening, and/or use of weapons).
Sexual abuse was defined in terms of the degree of force and penetration involved (Russell, 1983).
1. Noncontact experiences (e.g., exposure, inviting the child to engage in sexual activities).
2. Contact experiences that did not involve penetration or force (e.g., fondling).
3. Contact involving penetration or force. Examples included: digital penetration, oral sex, anal sex, sexual intercourse,
and all bizarre sex acts (e.g., bestiality, group sex).
Physical maltreatment was quantified in terms of degree of violence employed and the risk of potential physical harm.
1. Typical discipline situations: (e.g., spanking, grabbing).
2. Contact experiences that are typically nonlethal (e.g., shaking a grown child, throwing something at the child).
3. Potentially damaging or lethal behavior (e.g., beating the child, throwing the child, shaking an infant, strangulation,
burning, and use of weapons).