Family Dysfunction and Social Isolation as Moderators Between Stress and Child Physical Abuse Risk

Abstract

Stress is a significant contributor to child physical maltreatment risk. Family and social supports are expected, but less studied, risk factors. Little empirical support clarifies the interactive influence on abuse risk for non-abusive parents. This study examined whether the stress-abuse risk relation was moderated by family dysfunction and social isolation. Subjective appraisals of these factors were administered to 95 community mothers. After creating composite scores using factor loadings from a CFA, multiple regression analyses were conducted to predict abuse risk. As expected, stress predicted abuse risk, with social isolation and, to a lesser extent, family dysfunction serving as moderators. Perceived stress and dysfunctional supports are important, interactive predictors of abuse risk. Future directions consider interactions within other ecological levels.

Full text

RELATIONSHIP, FAMILY AND INDIVIDUAL CHARACTERISTICS THAT PREDICT BEHAVIOR
Family Dysfunction and Social Isolation as Moderators
Between Stress and Child Physical Abuse Risk
Meagan C. Tucker &Christina M. Rodriguez
Published online: 29 January 2014
#Springer Science+Business Media New York 2014
Abstract Stress is a significant contributor to child physical
maltreatment risk. Family and social supports are expected, but
less studied, risk factors. Little empirical support clarifies the
interactive influence on abuse risk for non-abusive parents. This
study examined whether the stress-abuse risk relation was mod-
erated by family dysfunction and social isolation. Subjective
appraisals of these factors were administered to 95 community
mothers. After creating composite scores using factor loadings
from a CFA, multiple regression analyses were conducted to
predict abuse risk. As expected, stress predicted abuse risk, with
social isolation and, to a lesser extent, family dysfunction serv-
ing as moderators. Perceived stress and dysfunctional supports
are important, interactive predictors of abuse risk. Future direc-
tions consider interactions within other ecological levels.
Keywords Child maltreatment risk .Physical child abuse
potential .Social support .Family dysfunction .Perceived
stress
Introduction
In 2010, the U.S. Department of Health & Human Services
(DHHS) confirmed over 754,000 cases of child maltreatment
(U.S. Department of Health and Human Services 2011). Of
these, nearly 18 % were victims of physical maltreatment or
child physical abuse (U.S. Department of Health and Human
Services 2011). However, underreporting and biased reporting
suggest that national statistics underestimate the prevalence of
physical abuse, with other estimates suggesting that millions of
children are victims of abuse annually (Sedlak et al. 2010;
Straus et al. 1998). Instances of more severe physical abuse
are overrepresented in agencies; consequently, research studies
that rely upon sampling from those sites raise questions about
generalizing to families engaging in less severe abuse or sub-
abusive parent-child aggression (Wissow and Wilson 1992).
As an alternative, community sampling has been used to obtain
a more representative sample of the potentially varying degrees
of parent-child aggression, including sub-abusive discipline,
which is physical discipline behavior that is socially acceptable
in frequency and severity (Graziano 1994). The inclusion of
parents engaging in sub-abusive behavior in research provides
an important opportunity to explore the circumstances that
may surround a parent becoming abusive (Graziano 1994).
Parent-child aggression can be represented along a contin-
uum, where physical discipline transitions at some point to
become physically abusive (Graziano 1994; Greenwald et al.
1997; Whipple and Richey 1997), contributing to the difficul-
ty in distinguishing physical discipline from abuse (Brooker
et al. 2001). Physical abuse often occurs within the context of
a parent administering physical discipline (Herrenkohl et al.
1983; Whipple and Richey 1997) because of the unintentional
escalation of sub-abusive, more normative physical discipline
(Graziano 1994). Thus, physically abusive parents initially
implement sub-abusive behavior (Graziano and Namaste
1990). The concept of child abuse potential includes beliefs,
attitudes, and behaviors identified in abusive parents to esti-
mate the likelihood that a parent will engage in physical
discipline that becomes abusive (Milner 1986,1994).
Although a considerable amount of research has been devoted
to identifying variables that predict physical abuse potential
(see Black et al. 2001 or Stith et al. 2009 for reviews), many of
M. C. Tucker (*):C. M. Rodriguez
Psychology Department, University of North Carolina at
Greensboro, P.O. Box 26170, Greensboro, NC 27402-6170, USA
e-mail: mctucker@uncg.edu
Present Address:
C. M. Rodriguez
Department of Psychology, University of Alabama at Birmingham,
Birmingham, AL 35294, USA
J Fam Viol (2014) 29:175–186
DOI 10.1007/s10896-013-9567-0

the conditions that foster abuse risk remain unclear.To prevent
abuse, identifying risk factors, particularly in non-identified
populations, is critical in order to better predict what brings a
parent closer to the point of engaging in physical abuse.
Theoretical Framework
A constellation of diverse risk factors can contribute to phys-
ical abuse, wherein factors at multiple levels of influence
surround the parent-child dyad (Belsky 1980,1993).
Ecological perspectives support that physical abuse is best
understood, and better predicted, by models that include mul-
tiple risk factors simultaneously impinging upon the parent-
child relation (Belsky 1980,1993; Sidebotham 2001). For
abusive parents, such models have highlighted the role of
personal and interpersonal vulnerabilities at multiple levels
including ontological factors (e.g., psychopathology, sub-
stance use, history of childhood abuse, etc.), microsystem
aspects (e.g., poor family relationships, intimate partner vio-
lence), and exosystem characteristics (e.g., social relation-
ships, neighborhood characteristics, etc.).
Ecological models thus propose that abuse risk is influ-
enced by the converging effects of multiple levels of influence
interacting with a parent’s unique personal, ontogenic quali-
ties. However, previous research has been limited in examin-
ing one level of interest at a time (e.g., see Hillson and Kuiper
1994 for discussion). As a consequence, current findings fall
short in demonstrating the interactive nature of risk factors and
their cumulative effect on abuse risk, rendering ecological
models largely theoretical (Hillson and Kuiper 1994).
However, more recent investigations exploring how risk fac-
tors singularly and in combination affect abuse risk have
demonstrated a cumulative effect on abuse potential such that
a greater number of psychosocial factors is associated with
elevated abuse risk (Masten and Wright 1998; Wekerle et al.
2007). Studies like these that examine individual and cumu-
lative main effects of such factors suggest the next step is to
also consider interactive effects of risk factors on abuse po-
tential to better capture the complexity of the true nature of
ecologically interactive levels.
Another issue apparent within this area of research is rooted
in the process by which risk factors are identified. To describe
those at greatest risk for engaging in physical abuse, research
has largely focused on identifying risk factors within four
domains: demographics, family relationships, and parental
and child characteristics (Belsky and Vondra 1989;Hillson
and Kuiper 1994). Beyond concerns that sampling and
reporting biases overestimate risk for certain individuals or
families, previous findings have often focused on risk factors
that are not clinically modifiable (e.g., ethnicity, marital status,
family composition, etc.); such an approach cannot effectively
contribute to prevention and intervention efforts. Although
recognition of particular risk groups is important, expending
resources on clinically relevant risk factors seems prudent.
Therefore, the current study focused on both ecologically
valid and clinically modifiable factors to examine how stress,
family dysfunction, and inadequate social support may col-
lectively relate to child abuse risk in non-identified parents.
Stress and Child Abuse Potential
The relation between stress and child physical abuse risk has
been well documented (Crouch and Behl 2001;Kolko1992;
Milner 1994;Rodriguez2010;RodriguezandGreen1997;
Taylor et al. 2009; Wekerle et al. 2007; Williamson et al.
1991). Yet the literature examining the effect of stress lacks
a consistentoperational definition, often failing to differentiate
the experience of stress from a stressor, making it difficult to
identify which underlying construct (life event, parenting, or
perception of stress) contributes to elevated abuse risk
(Whipple and Webster-Stratton 1991). Stress is conceptual-
ized in the current study as the subjective emotional experi-
ence that ensues when one does not feel capable of coping
with the presence of stressors. Such an emphasis on personal
perceptions and experiences of stress thus represents an onto-
logical risk factor.
Although abusive parents often report experiencing a num-
ber of significant life events (or stressors), such experiences
affect individuals differently because each person’s perception
of an event is unique. Merely totaling these experiences does
not account for interacting influences (e.g., coping, perception
of significance) unique to the individual, making a life-events
conceptualization of stress less relevant to predicting abuse
risk. Similarly, parenting stress has been strongly linked with
abuse potential (Rodriguez and Green 1997; Rodriguez and
Richardson 2007) and abusive parents describe parenting as
more distressing than their non-abusive counterparts
(Benedict et al. 1992; Burrell et al. 1994; Curenton et al.
2009); however, personal stress moderates the relation be-
tween child-related stress and abuse risk such that at-risk
parents are more easily overwhelmed by the stress of raising
a child they perceive to be difficult (Holden and Banez 1996).
In contrast to the number of stressors or parenting stress,
perceived stress incorporates the cognitive attributes that drive
one’s experience of stress; the extent to which a parent feels
overwhelmed by stressors and the subjective sense of effec-
tiveness in their coping abilities (Cohen et al. 1983)implies
perceived stress may be particularly relevant to exacerbating
abuse risk. Given that research continues to support that stress
is conceptually and practically relevant in predicting child
abuse potential (Black et al. 2001; Stith et al. 2009), this study
expected parents’perceptions of stress to be strongly associ-
ated with abuse potential (thus subsequent use of the term
“stress”refers to perceived stress, unless otherwise noted).
However, only some overwhelmed parents make the decision
to engage in abuse, indicating stress alone is clearly not
176 J Fam Viol (2014) 29:175–186

sufficient in estimating abuse risk. Rather, stress may predis-
pose a parent to be more vulnerable to the effects of other risk
factors within their environment that may buffer or exacerbate
abuse risk. Specifically, factors that provide the opportunity to
diffuse stress, such as high quality family and social support,
may be of value in further understanding how and for whom
elevated stress predicts greater abuse risk.
Family Dysfunction and Child Abuse Potential
Given the nature of child abuse, parents in maltreating fami-
lies are assumed to evidence not only personal but familial
dysfunction, which would operate at the microsystem level
given the interaction between parents and family members.
However, given that dysfunctional family patterns can be
evident in families with non-abusive and low abuse-risk par-
ents, family dysfunction alone has long been recognized as
insufficient to lead to abuse (Burrell et al. 1994;Mollerstrom
et al. 1992). Therefore, researchers have sought to identify
family dynamic characteristics of not only dysfunctional fam-
ilies but, specifically, families at greater risk for abuse as well.
Consistent with ecological and transactional perspectives,
and in response to work demonstrating the importance of
relationship quality over quantity, research has shifted toward
focusing upon how the quality of relationships among family
members may impact individual functioning that leads to
abuse risk (Paavilainen and Astedt-Kurki 2003; Stith et al.
2009; Stockhammer et al. 2001). Abusive families are more
likely to display an unstable and unpredictable family struc-
ture and to rigidly adhere to their roles as distinct individuals;
such family members do not encourage communication but
engage in relationship patterns that are coercive, unsupportive,
and conflictual (Kolko et al. 1993;Mapp2006;Mollerstrom
et al. 1992; Paavilainen and Astedt-Kurki 2003). Although
research cannot support family dysfunction as a single, inde-
pendent predictor, clearly this factor serves some role in
predicting abuse risk. Though the over-reliance upon identi-
fied abusive samples, as well as more simplistic statistical
analyses (e.g., bivariate analyses), have resulted in a limited
understanding of how family dysfunction relates to abuse risk
as an actual risk factor predictive of abuse potential when
other risk factors are also considered. However, dysfunctional
family relationships possibly not only represent an additional
stressor but also do not serve as an outlet for stress, thus
intensifying parents’experience ofstress. Conceivably, family
dysfunction may moderate the effect of stress on abuse risk,
such that the quality of family relationships may contribute to
or diffuse parents’perceptions of stress.
From an ecological perspective, sources of support can
include those within the microsystem level (e.g., family mem-
bers, spouses) and exosystem level (e.g., friends, school or
work associates) that affect both personal and interpersonal
functioning (Belsky 1980,1993; Corse et al. 1990;Moncher
1995). Comparable to the role of family level supports, rela-
tionships beyond the family can also impact individual level
functioning.
Social Isolation and Child Abuse Potential
Abusive parents often report feeling disconnected from sup-
portive resources at most ecological levels (Gracia and Musitu
1994,2003), and social isolation—when sources of emotional
or material resources are absent or of insufficient or
unsatisfying quality—differentiates abusive from non-
abusive parents (Corse et al. 1990; Gracia and Musitu 1994,
2003), even when samples are matched for demographics
(Milner and Dopke 1997). Perceptions of social support qual-
ity may be particularly important for families at risk for abuse
as external connections provide the opportunity to receive
advice or feedback about child-rearing practices that would
most likely be supported by family members who share sim-
ilar beliefs (Belsky 1980,1993). Social isolation may be a
consequence of abuse risk, rather than a cause, as ontogenic
vulnerabilities (e.g., cognitive biases) contribute to at-risk
parents’perceptions of support as unavailable or critical and
highly conflictual (Corse et al. 1990; Crouch et al. 2001;
Moncher 1995; Rodriguez 2008; Wekerle et al. 2007),
prompting many to actively withdraw (Black et al. 2001;
Milner 2000). Social network size has also been shown to
moderate the relation between number of stressful life events
and subsequent abuse reports in a sample of mothers substan-
tiated for abuse (Annerbäck et al. 2010; Kotch et al. 1997;
Ortega 2002). Thus, given that social support is generally
considered to buffer stress (Cohen and Wills 1985), similar
to dysfunctional family relationships, perceptions of poor
quality social supports (i.e., social isolation) may interact with
individual level factors, such as a parent’s perception of stress,
to impact abuse risk (Crouch et al. 2001).
The Current Study
The purpose of the present study was to extend current re-
search examining the relationship between ecological factors
and physical abuse risk in a non-identified sample of mothers.
Previous research has identified stress as a strong correlate of
child abuse potential (Black et al. 2001;Milner1993,1994,
2000;Rodriguez2010; Stith et al. 2009)aswellasestablished
associations with other identified risk factors, such as family
dysfunction or social isolation (Annerbäck et al. 2010;
Benedict et al. 1992;Crouchetal.2001; Mollerstrom et al.
1992; Moncher 1995; Wekerle et al. 2007). However, existing
research has often focused on samples identified through
protective services that cannot adequately clarify how these
might instead predict abuse risk, or often do not nest such
variables within a larger predictive model to examine potential
interactive effects on physical abuse risk. The current study
J Fam Viol (2014) 29:175–186 177

sought to examine factors previously identified as relevant
predictors of abuse risk, focusing on potential interactive
effects and utilizing multiple indicators of each construct,
within the context of an ecological model. Multiple indicators
were utilized to assess abuse risk, perceived stress, and per-
ceived quality of social supports (except for family dysfunc-
tion for which equivalent measures were not readily available)
to achieve greater depth in capturing the underlying construct
and to balance the potential strengths or weaknesses of any
given measure. For example, we decided it was critical to
include two measures of abuse potential to provide broader
coverage of the parenting beliefs and attitudes in abuse risk
than could be accomplished by relying on a single abuse
potential instrument wherein questions of item overlap across
constructs could become problematic.
The first goal of the study was to extend previous empirical
support in demonstrating that perceived stress, one’ssense
that cumulative stressors cannot be managed, is associated
with child abuse risk, such that greater levels of stress were
expected to relate to elevated abuse risk. The second goal was
to examine the potential moderating role of both family dys-
function and social isolation, such that the quality of familial
andextra-familialsocialrelationshipscanexacerbateordif-
fuse the expected relation between stress and abuse risk.
Community sampling was employed to help identify whether
the clinically modifiable factors considered in the present
study would be useful targets for prevention efforts.
Method
Participants
As part of a larger parenting study conducted in the Southeast,
mother-child dyads were recruited from various sites in the
community, including day care centers and after school pro-
grams. The current sample included 95 mothers of 6 to 9 years
old children. Maternal age ranged from 23 to 51 years (M=
37.89, SD=6.85), whereas children’smeanagewas7.46years
(SD=1.13). The majority of the sample was the child’sbio-
logical mother (93.7 %) and thus all maternal caregivers will
be referred to as “mothers”. Mothers primarily self-identified
as Caucasian (56.2 %) or African-American (39.3 %), with
6.3 % also selecting Hispanic/Latino. Given the limited rep-
resentation of other ethnic groups, mothers’race/ethnicity was
collapsed into a dichotomous variable to assess for potential
differences between Caucasian versus all other racial/ethnic
minority mothers (e.g., African-American, Hispanic/Latino,
Asian, etc.). The majority (73.7 %) reported currently residing
with the child’s biological father, whereas 22.1 % identified as
single parents. On average, mothers held a 4-year university
degree with an annual family income ranging from $40,000 to
$49,000 for an average family size of four.
Measures
Parental Abuse-Risk Measures The Child Abuse Potential
Inventory (CAPI; Milner 1986,1994) is a frequently used
screening instrument of child abuse risk involving character-
istics identified in abuse perpetrators. The CAPI is a 160-item
self-report questionnaire presented in an Agree/Disagree
forced choice format. Only 77 items contribute to the Abuse
Scale score, with higher scores indicative of greater physical
abuse potential. The CAPI has been found to be reliable across
age, gender, education level, and ethnic groups (Milner 1994).
Internal consistency (KR-20) for the Abuse Scale ranges from
0.92 to 0.96 for abusive and non-abusive populations (Milner
1986). Studies also indicate that CAPI scores demonstrate
predictive validity, with a correct classification rate of
89.2 % of confirmed child abusers and 99 % of controls
(Milner 1994).
The Adult-Adolescent Parenting Inventory-2 (AAPI-2;
Bavolek and Keene 2001) is a 40-item abuse potential mea-
sure that utilizes a 5-point Likert scale, ranging from 1 (strong-
ly agree)to5(strongly disagree), to determine degree of
agreement with beliefs and behaviors regarding child-rearing
characteristics of abusive parenting. For this study, the scale
was reversed so that higher AAPI-2 Total scores were associ-
ated with more dysfunctional parenting attitudes and beliefs
(i.e., higher abuse potential), consistent with the CAPI direc-
tion. Internal consistency for the AAPI-2 Total score has been
reported at 0.85, with support for concurrent validity (Conners
et al. 2006). In the present study, the AAPI-2 also demonstrat-
ed good internal consistency, with α=.84.
Measures of Predictors: Perceived Stress The Perceived
Stress Scale (PSS; Cohen et al. 1983) is comprised of ten
items used to assess the extent to which participants felt their
lives were overwhelming, uncontrollable, or unpredictable in
the past month. Items are rated on a 4-point Likert scale
ranging from 0 (never)to3(very often). Total scores are
generated by summing individual items, with higher scores
indicative of greater perceived stress. Coefficient alpha has
been reported to range from 0.84 to 0.86 across samples for
the PSS total score (Cohen et al. 1983). Internal consistency
observed in the current sample was also high, with α=.89.
The Daily Hassles and Uplifts Scale (DHUS; De Longis
et al. 1988) is a revised form of a longer measure of the same
name and consists of the 53 most frequently endorsed items.
The DHUS is assessed using a split scale, allowing for each
item to be rated in terms of the degree to which it represents
both a hassle and an uplift. For the purposes of the current
study, only the Hassles scale was used to determine the degree
to which participants perceived these daily events as bother-
some, annoying, or irritating and participants were asked to
consider these items for “the past week”instead of the standard
“the past 24 h”to obtain a broader representation of
178 J Fam Viol (2014) 29:175–186

respondents’perceived daily hassles. The Hassles scale alpha
reliability coefficient ranges from 0.57 to 0.83 (Holm and
Holroyd 1992). In the current study, reliability was high, with
α=.94.
Measures of Predictors: Family Dysfunction The Family
Relationship Index (FRI; Moos and Moos 1986) is one of three
dimensions that comprise the original Family Environment
Scale (FES; Moos and Moos 1976). The FRI is used to focus
specifically on the relationships between family members. The
FRI is a 27-item true/false measure which assesses parents’
experiences of family relationships across three subscales: co-
hesion, expressiveness, and conflict. A Total Score is obtained
by summing scores from both the Cohesive and Expressiveness
scales and then subtracting the score from the Conflict scale
(e.g., Total score= [(Coh. + Exp.)—Conflict]). Lower Total
scores represent more dysfunctional family relationships.
The FRI has previously been shown to have high inter-
nal consistency (Varni and Setoguchi 1993) and good
construct validity (Hoge et al. 1989). Cronbach’s alpha
for the FRI has been reported as 0.91 (Howell et al.
2007); within the current study αvalues ranged from
0.61 to 0.69 across subscales.
Measures of Predictors: Social Isolation The De Jong-
Gierveld Loneliness Scale (Loneliness Scale; De Jong-
Gierveld and Van Tilburg 1999) is an 11-item Yes/No self-
report questionnaire, including five positive and six negative
items. A total score was computed by adding across individual
items, with high scores suggesting greater perceived loneli-
ness. Although the majority of prior samples have been Dutch,
this measure has been shown to be valid and reliable across
cultural contexts and ethnic groups (De Jong-Gierveld and
Kamphuis 1985). Coefficient alpha ranges from 0.84 to 0.88
across samples (De Jong-Gierveld and Kamphuis 1985), with
α=.85 in the present study.
The Social Support Resources Index (SSRI; Vaux and
Harrison 1985) was used to assess perceived extra-familial
social support. The measure’s design allows respondents to
list up to 10 individuals from their support network and then
rank those individuals in terms of the quality of emotional
and/or socializing support, practical and/or financial assis-
tance, and advice/guidance they provide. For the current
study, the quality of social support was of particular interest,
and, as such, the instructions were adjusted to reflect the
quality of the relationships between the participant and their
closest supporters. Respondents were asked to rate relation-
ship satisfaction on five items using a 5-point Likert scale,
ranging from 1 (not at all satisfied)to5(extremely satisfied),
for only the two closest individuals from their social network.
Items were reverse scored prior to computing a Total score
(sum of items) such that higher ratings represent greater
dissatisfaction with their social network members (e.g.,
greater perceived social isolation). The SSRI has been report-
ed as having adequate reliability (0.76; Vaux and Wood 1987),
with high internal consistency in the current study (α=.91).
Procedures
The Institutional Review Board at the University of North
Carolina at Greensboro granted approval for the larger study
designed to examine novel methodologies in studying parent-
ing challenges. Participants were recruited using flyers and
advertisements. Over the course of 20 months, interested
mothers called the university research lab and scheduled a
time to participate in a 1 h session at the lab. Upon their
arrival, parents provided informed consent and were escorted
to a private room to complete self-report questionnaires on a
computer. Within the context of the larger study, our measures
were interspersed, generally ordered with abuse risk measures
followed by social isolation, stress, and family dysfunction.
As part of the programming, responses were automatically
stored into a database identified only by a randomly assigned
ID number. Thus, participants were informed that their re-
sponses would be anonymous and candid responding was
strongly encouraged. Mothers were compensated for their
time with a $20 gift card.
Results
Analyses
Basic analyses were conducted using SPSS 18 for Windows.
After the potential need for covariates and simple bivariate
relationships were examined, the predictor and dependent
measures were standardized. To test for moderation with a
conventional regression approach, factors were weighted
using standardized regression weights derived from their load-
ings on a confirmatory factor analysis (CFA). AMOS-18 was
utilized to evaluate the fit of three proposed underlying fac-
tors: Abuse Risk, Social Support, and Stress. Fit of the CFA
was evaluated using comparative fit index (CFI), normed fit
index (NFI), goodness-of-fit index (GFI), and root mean
square error of approximation (RMSEA; Byrne 2001;
Tabachnick and Fidell 1996). With respect to the fit indices,
CFI, NFI, and GFI values greater than 0.90 are ideal with
RMSEA values below 0.08 suggesting reasonable fit (Byrne
2001; Tabachnick and Fidell 1996). In the current CFA, the
CFI= .97, NFI=.95, GFI= .96, and RMSEA=.10. An exami-
nation of the factor loadings confirmed that all the variables
loaded significantly onto the expected underlying latent fac-
tors. Thus, factor loadings for each underlying construct were
as follows: Abuse Risk (CAPI Abuse Scale = .99; AAPI-2
Total= .36); Stress (PSS Total= 1.00; DHUS=.53); Social
Isolation (Loneliness Total=.83; SSR-Satisfaction Total=.70).
J Fam Viol (2014) 29:175–186 179

The weighted indicators were summed to create composite
variables for Abuse Risk, Stress, and Social Isolation for the
multivariate analyses that would include moderators using
standardized multiplicative terms. A series of hierarchical mul-
tiple regressions were performed to examine whether family
dysfunction and social isolation reliably contributed to the
prediction of abuse risk, beyond stress and their independent
main effects, including examining the potential for interactive
effects of family dysfunction and social isolation as moderators.
Demographic Comparisons
To assess the need for demographic statistical controls, pre-
liminary analyses were conducted to determine whether pre-
dictor or outcome variables differed across demographic char-
acteristics. No significant differences in outcome variables
were found for participating child age. Parental age was
significantly correlated to Abuse Risk scores (r=−.27,
p≤.01), such that younger maternal age was associated with
elevated abuse risk. Family size was associated with Abuse
Risk scores (r=−.26, p≤.05), such that larger family size was
associated with greater abuse risk.
A significant correlation was identified for both income and
education level on the outcome variable, such that mothers of
low income and those who reported less formal education
evidenced greater abuse risk (r=−.51, p≤.001; r=−.31,
p≤.01, respectively). Analyses also revealed significant t-test
findings regarding group mean differences based on relation-
ship status and ethnicity. There were significant mean differ-
ences for relationship status and Abuse Risk scores (t(93) =
5.69, p<.001) such that single mothers evidenced greater abuse
potential (M=1.11, SD=1.11) than mothers living with a part-
ner (M=−.31, SD=.98). The Abuse Risk score means were
significantly higher (t(93)=−2.84, p<.05) for ethnic minority
mothers (M=.35, SD=1.18) than Caucasian mothers (M=−.31,
SD=1.08).
In sum, abuse risk was elevated for mothers who represented
an ethnic minority group, obtained less education, earned less
income, and/or were single parents. Furthermore, minority
status was also associated with greater perceived exposure to
daily stressors as well as both indicators of social isolation.
Consequently, these demographic characteristics were consid-
ered as potential covariates in the multivariate analyses.
Preliminary Correlation Analyses
The initial correlations among the measures of interest were
examined (see Table 1). As expected, indicators of abuse risk
were significantly correlated with family dysfunction (low
FRI scores) and positively associated with indicators of
perceived stress and social isolation. In other words, great-
er abuse potential was observed in mothers with a greater
degree of perceived stress, family dysfunction, and
perceived social isolation (e.g., less satisfying extra-
familial supports).
Multiple Regression Analyses
Hierarchical multiple regression analyses were performed to
determine the unique role of family dysfunction and social
isolation, beyond perceived stress, in predicting abuse risk
and, specifically, to examine for potential interactive effects.
Initial analyses of the regression models were structured as
follows: potential demographic controls (income, parent age,
education, partner status, and ethnic status) were entered at
block one, each main effect was entered at step two
(Stress) and step three (e.g., FRI, Social Isolation), follow-
ed by the interaction term, using the standardized multi-
plicative term computed from the two composite indicators
(e.g., Stress x FRI or Stress x Social Isolation). Consideration
of multicollinearity diagnostics confirmed that all variables
across regressions demonstrated robust tolerance, with no
evident multicollinearity.
In predicting Abuse Risk scores, initial regressions with
variables entered as described yielded an R
2
=.63, F(8, 86)=
18.46, p<.001, for family dysfunction alone in block 3 and an
R
2
=.68, F(8, 86)= 22.97, p<.001, for Social Isolation alone in
block 3. However, income contributed significant unique
variance in each model and was thus retained as a significant
covariate in predicting both, and partner status was a signifi-
cant covariate in the model with family dysfunction. As would
be expected conceptually, the family level demographic fac-
tors of partner status and total family size covaried (r=.37,
p<.000), but both were significantly associated with abuse
risk. However, partner status was considered to be of greater
clinical relevance as the categorical nature of this factor ex-
plicitly identifies a target group of families evidencing greater
risk (e.g., single versus non-single parents). Thus, partner
status was retained as an additional covariate in subsequent
analyses. Notably, the contribution of partner status was re-
duced to trend level when the main effect and interaction of
social interaction was included in the model and, thus, was not
retained as a significant contributor. The final regression re-
sults reported below thus reflect all of the theorized predictors
and the statistically significant demographic covariates.
The Role of Stress
The first regression was performed to examine the role of
perceived stress in predicting Abuse Risk scores and, specif-
ically, to determine whether a significant main effect would be
confirmed. Stress strongly contributed a significant amount of
unique variance in Abuse Risk scores, beyond income and
partner status, resulting in a final model of R
2
=.59, F(3, 91)=
44.21, p≤.001, and contributed an additional 24 % of the
variance explained beyond the covariates. Thus, as expected,
180 J Fam Viol (2014) 29:175–186

the main effect for stress was significant in predicting Abuse
Risk scores.
The Role of Family Dysfunction
The second regression was performed to examine whether
family dysfunction interacted with stress to reliably explain
additional variance in Abuse Risk, beyond their main effects
alone. There was a significant main effect for family dysfunc-
tion, as the FRI uniquely contributed to the prediction of
Abuse Risk scores beyond the contribution of Stress alone.
However, the inclusion of the interaction term, although in the
expected direction, did not reliably account for significant
unique variance in Abuse Risk scores (see Table 2for a
summary of regression equation results for this full model).
Thus, the most parsimonious model retains income and part-
ner status but not the interaction term, resulting in an R
2
=.62,
F(4, 90)= 36.52, p≤.001.
The Role of Social Isolation
The third regression examined the potential role of per-
ceived extra-familial social isolation as a moderator of the
Stress-Abuse Risk relation and found that both the main
effect of Social Isolation and the interaction term were
significant, contributing beyond covariates (see Table 2for
this full model). The final, most parsimonious model,
variables entered as described above reliably contributed
to the prediction of Abuse Risk scores and resulted in an
R
2
=.67, F(4, 90) = 45.67, p≤.001 (See Fig. 1for a visual
representation of the interaction).
Although, when controlling for all previously noted demo-
graphics and Stress, the main effects for both the FRI and
Social Isolation independently predicted Abuse Risk, with the
main effects of social isolation and family dysfunction entered
simultaneously followed by the interaction terms R
2
=.69,
F(10, 84)=18.56, p≤.001. Thus, Stress and Social Isolation
contributed significant unique variance in predicting Abuse
Risk (β=4.78, p≤.001 and β=3.02, p≤.01, respectively),
where income was the only significant demographic covari-
ate. However, family dysfunction did not contribute signifi-
cant unique variance as a main effect (β=-1.16, p>.05) or
interaction with Stress (β=−.57, p>.05), although the interac-
tion between Stress and Social Isolation approached signifi-
cance in this full model (β=1.75, p= .08). Therefore, the
quality of interpersonal relationships is important in predicting
abuse potential overall, wherein extra-familial social isolation
is “favored”over, or more salient than, family dysfunction.
Discussion
The current study sought to examine the relation between
psychosocial risk factors and physical maltreatment risk.
Specifically, an ecological approach was adopted to evaluate
whether parental vulnerabilities (e.g., stress) are further exac-
erbated by more distal stressors, such as family dysfunction
and social isolation in a non-identified sample of mothers.
These distal factors were targeted primarily due to the litera-
ture suggesting that the quality of familial and extra-familial
social supports might contribute to or diffuse parents’stress.
Overall, the findings of the present study, using 95 mothers
recruited from the community, provided support for the hy-
potheses. Perceived stress consistently contributed to greater
overall abuse risk as did family dysfunction and social isola-
tion, independently. Moreover, greater physical abuse risk was
evidenced whereby stress interacted with these interpersonal
Tabl e 1 Means, standard deviations, and correlations for measures
M(SD)123456
1CAPI Abuse 95.91 (81.80)
2AAPI-2 Total 88.52 (16.54) .36***
3PSS 25.78 (8.04) .72*** .20
4 DHUS 88.43 (23.11) .38*** .34*** .54***
5FRI 9.82 (4.28) -.48*** -.29** -.54*** -.40***
6 SSRI 19.16 (7.93) .52*** .24* .40*** .22* -.48***
7 Loneliness 2.83 (2.93) .61*** .31** .45*** .24* -.40*** .58***
Grey boxes highlight the indicators that will contribute to the composite scores for multivariate analyses
CAPI Abuse child abuse potential inventory abuse scale; AAPI adult-adolescent parenting inventory-2; PSS perceived stress scale; DHUS daily hassles
and uplifts scale; FRI family relationship index; SSRI social support resources index; Loneliness De Jong-Gierveld loneliness scale
*p≤.05; **p≤.01; ***p≤.001
J Fam Viol (2014) 29:175–186 181

factors, predominantly for social isolation relative to family
dysfunction.
In addition to substantiating the critical role of stress in
predicting abuse risk, the current results indicate that
additional background factors strongly improve the esti-
mation of abuse risk. Specifically, income and single
parenthood predicted abuse risk. Lower income level
and being a single parent conceivably represent stressors
to mothers, as indicated by the negative association
between these demographic factors and stress. Thus, this
may explain why low-income and single parents are
considered abuse risk groups (Berger 2005; Stith et al.
2009). Notably, these background characteristics were
more salient than educational level or ethnic minority
status when considered simultaneously.
The Role of Stress
As predicted, stress significantly contributed to elevated
Abuse Risk. Specifically, consistent with the other conceptu-
alizations of stress, the extent to which mothers considered
themselves overwhelmed by various daily stressors was pre-
dictive of elevated abuse risk. Thus, consistent with previous
findings (Crouch and Behl 2001; Curenton et al. 2009; Taylor
et al. 2009; Wekerle et al. 2007; Whipple and Webster-Stratton
1991), parents who experience themselves as overwhelmed
by multiple stressors are more likely to resort to physical
violence in an attempt to regain control over their environ-
ment. This strong relationship extends findings that life stress
(e.g., Williamson et al. 1991) and parenting stress relate to
abuse potential (e.g., Rodriguez and Green 1997;Rodriguez
2010) by incorporating parents’perceptions of their manage-
ment and coping with stressors. Thus, the results corroborate
claims that stress is one of the strongest correlates of abuse
risk (Burrell et al. 1994;Casanovaetal.1992;Milner1993)
and underscores the importance of understanding how per-
ceptions of stress serve to lead parents toward the more
abusive end of the physical discipline continuum.
The current findings do not disentangle the nuances of why
these parents may be reporting greater experiences of stress,
particularly given its strong contribution even after major
demographic covariates (and likely stressors) were controlled.
For example, increased perceived stress reported by the pres-
ent sample may partly result from a greater number of actual
stressors or, alternatively, relate to a hyperresponsiveness to
stressors (c.f., Bauer and Twentyman 1985). The degree to
which stressors are perceived as overwhelming is likely driven
by other personal, ontogenic factors, which may also con-
tribute to increased abuse risk either independently or
interactively in the presence of stressors. Future investi-
gations should examine potential cognitive processes or
biases that may play a crucial role in understanding what
contributes to the heterogeneity of response to a given
stressor, potentially further elucidating factors that may
increase one’s vulnerability to stressors.
The Role of Interpersonal Relations in Predicting Physical
Abuse Risk
In terms of the role of family dysfunction, a significant main
effect indicates the quality of family relationships contributes,
beyond the influence of stress, to predict abuse risk.
Specifically, abuse risk increases as family relationships be-
come more dysfunctional. Although in the expected direction,
the current findings indicated that stress was exacerbated by
family dysfunction but only at trend levels.
Tabl e 2 Multiple regression results
bβtsr
2
Full multiple regression for abuse risk with family dysfunction
Block 1 0.36***
Age −0.01 −0.03 −0.35
Ethnicity 0.11 0.05 0.62
Education 0.01 0.02 0.17
Annual income −0.07 −0.23 −2.21*
Partner status −0.58 −0.21 −2.66**
Block 2 0.24***
Stress 0.38 0.44 5.24***
Block 3 0.02***
Family dysfunction −0.19 −0.16 −2.00*
Block 4 0.01
Stress × Family dysfunction
interaction
−0.10 −0.12 −1.65
Intercept = .96
R=.80, F(8, 86) = 18.46, p≤.001
R
2
=.63 (Adjusted R
2
=.60)
Full multiple regression for abuse risk with social isolation
Block 1 0.36***
Age 0.01 0.04 0.54
Ethnicity 0.05 0.02 0.30
Education −0.01 −0.01 −0.15
Annual income −0.07 −0.23 −2.42*
Partner status −0.34 −0.12 −1.60
Block 2 0.24***
Stress 0.36 0.42 5.98***
Block 3 0.07***
Social isolation 0.24 0.28 3.58***
Block 4 0.02*
Stress × Social isolation
interaction
0.08 0.15 2.23*
Intercept = .46
R=.83, F(8, 86) = 22.97, p≤.001
R
2
=.68 (Adjusted R
2
=.65)
*p≤.05; **p≤.01; ***p≤.001
182 J Fam Viol (2014) 29:175–186

In examining the connection between quality of social
relationships and abuse risk, the current findings supported
the hypothesis. Social isolation moderated the stress-abuse
risk relation, such that mothers who reported higher levels of
perceived stress and social isolation in combination evidenced
greater abuse risk than those reporting only higher levels of
stress or social isolation. Consistent with proponents empha-
sizing the importance of relationship quality (Corse et al.
1990; Moncher 1995; Paavilainen and Astedt-Kurki 2003;
Stith et al. 2009; Stockhammer et al. 2001), the current find-
ings support that the extent to which social supports are
satisfying thus influences their utility as a buffer against stress
in predicting abuse risk.
The simultaneous influence of both family dysfunction and
social isolation in the prediction of abuse risk was also exam-
ined to determine which might emerge as a stronger predictor,
beyond stress. The resulting findings supported that, when
competing, social isolation accounted for the majority of
unique variance in abuse risk, eliminating the variance attrib-
utable to family dysfunction. Although, the two measures that
contributed to the abuse risk composite score (CAPI and
AAPI-2) do include some broad aspects of relationship qual-
ity, the family relationship is comparatively less evident in
those measures relative to overall social functioning, which
may account for why family dysfunction interacted only mar-
ginally in predicting abuse risk. Thus, although family dys-
function was predictive of abuse risk, the relation was not
strong enough to combat the more salient contribution of
social isolation which may be more readily represented in
standard measures of abuse potential. Given the trend toward
significance shown in the present study, further investigations
should continue to examine how the quality of family relations
may interact with stress, potentially with supplemental mea-
sures of family dysfunction and abuse potential.
The need for supplemental measures of family dysfunction
was evident in the present study that relied on a single indicator
of the quality of family relationships. Given that other studies
have questioned the strength and reliability of the Family
Relationship Index, with reliabilities ranging from 0.40 to
0.74 (Roosa and Beals 1990), additional measures of family
dysfunction are needed. If comparable measures of these dys-
functional family relations were available, using multiple indi-
cators could compensate for weaknesses of a single measure,
allowing for a more comprehensive assessment of the underly-
ing construct. Additionally, although cohesion, expressiveness,
and degree of conflict are considered important markers of
family dysfunction, perhaps a more explicit assessment of
degree of actual support between family members would im-
prove the prediction of abuse risk. Alternatively, future studies
could consider the possibility that the buffering/exacerbating
effect of family dysfunction may be stronger in certain contexts
over others (e.g., exacerbating in the presence of domestic
violence) or in specific family structure or ethnic/racial cultural
groups (e.g., buffering for families of specific cultural affilia-
tions). Furthermore, other dimensions of family relationships
were not considered in the present study that could be
Fig. 1 Simple slopes analysis.
Analysis was conducted to
determine the effect of Social
Isolation (moderator) on the
relation between Stress and
Abuse Risk. The effect of Stress
on Abuse Risk varied at each
level of Social Isolation,
indicating a significant
moderator effect
J Fam Viol (2014) 29:175–186 183

influential in predicting abuse risk. As previously discussed,
functional families evidence stable, structured, yet flexible roles
for each member which dictate boundary setting, communica-
tion between members, and ultimately inform how interactions
between individual units contribute to the family unit as a
whole (Mollerstrom et al. 1992;MilnerandCrouch1993;
Paavilainen and Astedt-Kurki 2003). Future research should
continue to examine variations in family dynamics that may
contribute to elevated abuse potential.
Summary Physical Abuse Risk Interpretations
Consistent with prior research (e.g., Rodriguez 2010; Taylor
et al. 2009), the ontological factor of interest (perceived stress)
consistently demonstrated a strong relation to abuse risk, with
additional support for the role of interpersonal relations in
exacerbating stress. Perceptions of not being able to handle
stress may serve as a catalyst that brings to prominence other
risk factors which individually and in combination influence
the decision to engage in harsh physical discipline. Thus,
perhaps these findings suggest a transactional process by
which stress affects abuse risk on two fronts. First, increased
feelings of being overwhelmed might interfere with the ability
to adaptively cope with and respond to a challenging parent-
child interaction; second, responding to feeling overwhelmed
can then shape their beliefs and attitudes such that inappropri-
ate discipline reactions become viewed as more acceptable or
justified to diffuse their personal experience of stress. The
degree of harsh discipline employed, driven by a desire to
control or quickly resolve a stressful parent-child interaction,
may be initially experienced as inconsistent with more func-
tional beliefs about physical discipline held by the parent.
However, as the frequency of aggressive parent-child encoun-
ters increases, a parent may begin to alter his/her discipline
beliefs to regain consistency with his/her behavior, thus re-
solving the uncomfortable dissonance experienced. A longi-
tudinal examination of caretaker attitudes and beliefs over
time would better inform the explanatory, and potentially
evolving, role of stress in justifying harsher discipline that
could inadvertently lead to physical abuse.
CAPI and AAPI-2 as Physical Abuse Risk Indicators
The AAPI-2, assessing maladaptive or dysfunctional parent-
ing beliefs and attitudes, represents only one facet of abuse
risk. Yet the AAPI-2 complements the CAPI Abuse Scale
which generally does not explicitly assess parenting beliefs;
their combination thus provided a more complete picture of
physical abuse risk in the current study. The CAPI, given its
broader scope and well-established psychometric properties,
was more heavily weighted in the factor analysis, suggesting
this indicator contributed more to abuse risk than the AAPI-2.
Although the CAPI is often considered a gold standard for
abuse risk assessment, the items often correspond closely to
predictors like stress, which is why we elected to balance the
strength of the CAPI with an additional measure like the
AAPI-2 that does not overlap with predictors. Despite this,
the inclusion of both permitted us to more reliably capture
overall abuse risk, and similar multi-indicator approaches in
future investigations are strongly encouraged. Future research
should consider supplementing such self-reports with other
methods (e.g., behavioral observations) to further strengthen
the measurement of abuse potential as a dependent variable.
Additional Limitations
As with any research endeavor, limitations to the present
study, other than those already noted, warrant mentioning.
The present findings were correlational not causal and, given
the potential transactional nature of stress, highlight the need
for evaluating how changing interpersonal interactions mod-
erate stress across time. Given that the present study focused
on a sub-abusive community sample, the current findings may
not generalize to at-risk mothers, to families wherein abuse
has been substantiated, or to mothers of lower educational or
income levels. Additionally, as the present study utilized
mothers as the primary caregiver, future investigations should
examine whether these findings generalize to paternal care-
givers. For example, given that fathers often represent a sec-
ondary caregiver, it is unclear whether paternal abuse risk or
attitudes regarding the caretaker role would be more or less
susceptible to the stress effects found in the present study. To
address method bias from using a single source, including
other informants rather than solely self-report when examin-
ing these variables could provide a more comprehensive, and
thus meaningful, examination of the relations among the
constructs of interest.
In sum, the current study extended previous findings em-
phasizing the importance of perceived stress in estimating
physical maltreatment risk in which quality of family and
social relationships may contribute in a non-identified sample
of mothers. Given the importance of perceptions of stress
beyond demographic factors, these findings underscore the
need for community based interventions aimed at increasing
adaptive coping in response to contextual or demographic
stressors. Moreover, parents may benefit from adopting cop-
ing techniques aimed at increasing positive beliefs about
parenting. Additionally, future investigations that identify
specific cognitive factors (e.g., biases and distorted cogni-
tions) which contribute to perceptions of stressors as over-
whelming have clinical utility as these findings represent
relevant targets for intervention. Considering the present find-
ings, prevention efforts should focus on increasing the number
of high quality relationships available to the parent, outside
the family but potentially improving family relations as well.
The main goal of the present study was to synthesize and
184 J Fam Viol (2014) 29:175–186

extend previous predictive models of physical abuse risk by
examining factors interactively. It was not limited to mothers
identified by protective services, who represent the potentially
extreme end of parental abusive behavior. Future investiga-
tions should consider the present findings as an example of
examining ecologically nested and clinically malleable risk
factors and their interactions to better predict abuse risk.
References
Annerbäck, E., Svedin, C., & Gustafsson, P. (2010). Characteristic fea-
tures of severe child physical abuse—A multi-informant approach.
Journal of Family Violence, 25,165–172.
Bauer, W. D., & Twentyman, C. T. (1985). Abusing, neglectful, and
comparison mothers’responses to child-related and non-child-
related stressors. Journal of Consulting and Clinical Psychology,
53,335–343.
Bavolek, S. J., & Keene, R. G. (2001). Adult–adolescent parenting
inventory-2. Asheville: Family Development Resources Inc.
Belsky, J. (1980). Child physical maltreatment: an ecological integration.
American Psychologist, 35,320–335.
Belsky, J. (1993). Etiology of child physical maltreatment: a
developmental-ecological analysis. Psychological Bulletin, 114,
413–434.
Belsky, J., & Vondra, J. (1989). Lessons from child abuse: The determi-
nants of parenting. In D. Cicchetti & V. Carlson (Eds.), Child
maltreatment: Theory and research on the causes and consequences
of child abuse and neglect (pp. 153–202). New York: Cambridge
University Press.
Benedict, M. I., Wulff, L. M., & White, R. B. (1992). Current parental
stress in maltreating and nonmaltreating families of children with
multiple disabilities. Child Abuse & Neglect, 16,155–163.
Berger, L. M. (2005). Income, family characteristics, and physical
violence toward children. Child Abuse & Neglect, 29,107–
133.
Black, D. A., Heyman, R. E., & Smith-Slep, A. M. (2001). Risk factors
for child abuse. Aggression and Violent Behavior, 6,121–188.
Brooker, S., Cawson, P., Kelly, G., & Wattam, C. (2001). The prevalence
of child abuse and neglect: a survey of young people. International
JournalofMarketResearch,43,249–289.
Burrell, B., Thompson, B., & Sexton, D. (1994). Predicting child
abuse potential across family types. Child Abuse & Neglect,
18,1039–1049.
Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic
concepts, applications, and programming. Mahwah: Lawrence
Erlbaum Associates.
Casanova, G. M., Domanic, J., McCanne, T. R., & Milner, J. S. (1992).
Physiological response to nonchild-related stressors in mothers at
risk for child abuse. Child Abuse & Neglect, 16,31–44.
Cohen, S., & Wills, T. (1985). Stress, social support, and the buffering
hypothesis. Psychological Bulletin, 98,310–357.
Cohen,S.,Kamarck,T.,&Mermelstein,R.(1983).Aglobalmeasure
of perceived stress. Journal of Health and Social Behavior, 24,
385–396.
Conners, N. A., Whiteside-Mansell, L., Deere, D., Ledet, T., & Edwards,
M. C. (2006). Measuring the potential for child maltreatment: the
reliability and validity of the Adult Adolescent Parenting Inventory-
2. Child Abuse & Neglect, 30,39–53.
Corse, S. J., Schmid, K., & Trickett, P. K. (1990). Social network
characteristics of mothers in abusing and nonabusing families and
their relationships to parenting beliefs. Journal of Community
Psychology, 18,44–59.
Crouch, J. L., & Behl, L. E. (2001). Relationships among parental beliefs
in corporal punishment, reported stress, and physical child abuse
potential. Child Abuse & Neglect, 25,413–419.
Crouch, J. L., Milner, J. S., & Thomsen, C. (2001). Childhood physical
abuse, early social support, and risk for maltreatment: current social
support as a mediator of risk for child physical abuse. Child Abuse &
Neglect, 25,93–107.
Curenton, S. M., McWey, L. M., & Bolen, M. G. (2009). Distinguishing
maltreating versus nonmaltreating at-risk families: implications for
foster care and early childhood education interventions. Families in
Society, 90,176–182.
De Jong-Gierveld, J., & Kamphuis, F. (1985). The development of a
Rasch-Type Loneliness Scale. Applied Psychological Measurement,
9,289–299.
De Jong-Gierveld, J., & Van Tilburg, T. (1999). Manual of the loneliness
scale. Amsterdam: Vrije Universiteit.
De Longis, A.,Folkman, S., & Lazarus, R. S. (1988). The impact of daily
stress on health and mood: psychological and social resources as
mediators. Journal of Personality and Social Psychology, 54,486–
495.
Gracia, E., & Musitu, G. (1994). Child physical maltreatment typology:
an empirical conceptualization. Revista de Psicología Social
Aplicada, 4,53–72.
Gracia, E., & Musitu, G. (2003). Social isolation from communities and
child maltreatment: a cross-cultural comparison. Child Abuse &
Neglect, 27,153–168.
Graziano, A. M. (1994). Why should we study subabusive violence
against children? Journal of Interpersonal Violence, 9,412–419.
Graziano, A. M., & Namaste, K. A. (1990). Parental use of physical force
in child discipline. Journal of Interpersonal Violence, 5,449–463.
Greenwald, R. L., Bank, L., Reid, J. B., & Knutson, J. F. (1997). A
discipline-mediated model of excessively punitive parenting.
Aggressive Behavior, 23,259–280.
Herrenkohl, R. C., Herrenkohl, E. C., & Egolf, B. P. (1983). Circumstances
surrounding the occurrence of child physical maltreatment. Journal of
Consulting and Clinical Psychology, 51,424–431.
Hillson, J. M. C., & Kuiper, N. A. (1994). A stress and coping model of
child maltreatment. Clinical Psychology Review, 14,261–285.
Hoge, R. D., Andrews, D. A., & Faulkner, P. (1989). The family rela-
tionship index: validity data. Journal of Clinical Psychology, 45,
897–903.
Holden, W. E., & Banez, G. A. (1996). Child abuse potential and
parenting stress within maltreating families. Journal of Family
Vio l e n ce, 11,1–12.
Holm, J. E., & Holroyd, K. A. (1992). The daily hassles scales (revised):
does it measure stress or symptoms? Behavioral Assessment, 14,
465–482.
Howell, A., Hauser-Cram, P., & Kersh, J. E. (2007). Setting the stage:
early childand family characteristics as predictors of later loneliness
in children with developmental disabilities. American Journal on
Mental Retardation, 112,18–30.
Kolko, D. J. (1992). Characteristics of child victims of physical violence:
research findings and clinical implications. Journal of Interpersonal
Vio l e n ce, 7,244–276.
Kolko, D. J., Kazdin, A. E., Thomas, A. M., & Day, B. (1993).
Heightened child physical abuse potential: child, parent, and family
dysfunction. Journal of Interpersonal Violence, 8,169–192.
Kotch, J. B., Browne, D. C., Ringwalt, C. L., Dufort, V., Ruina, E.,
Stewart, P. W., et al. (1997). Stress, social support, and substantiated
maltreatment in the second and third years of life. Child Abuse &
Neglect, 21,1025–1037.
Mapp, S. C. (2006). The effects of sexual abuse as a child on the risk of
mothers physically busing their children: a path analysis using
systems theory. Child Abuse & Neglect, 30,1293–1310.
J Fam Viol (2014) 29:175–186 185

Masten, A. S., & Wright, M. O. (1998). Cumulative risk and protection
models of child maltreatment. Journal of Aggression, Maltreatment
& Trauma, 2,7–30.
Milner,J.S.(1986).The child abuse potential inventory: Manual
(2nd ed.). Webster: Psytec.
Milner, J. S. (1993). Social information processing and physical child
abuse. Clinical Psychology Review, 13,275–294.
Milner, J. S. (1994). Assessing physical child abuse risk: the child abuse
potential inventory. Clinical Psychology Review, 14,547–583.
Milner, J. S. (2000). Social information processing and child physical
abuse: Theory and research. In D. J. Hansen (Ed.), Nebraska sym-
posium on motivation (Motivation and child physical maltreatment,
Vol. 46, pp. 39–84). Lincoln: University of Nebraska Press.
Milner, J. S., & Crouch, J. L. (1993). Physical child abuse. In R. L. Hampton
(Ed.), Family violence: Prevention and treatment (pp. 25–55).
Newbury Park: Sage Publications.
Milner, J. S., & Dopke, C. (1997). Child physical abuse: Review of
offender characteristics. In D. A. Wolfe, R. J. McMahon, & R. D.
Peters (Eds.), Child abuse: New directions in prevention and treat-
ment across the lifespan (pp. 27–54). Thousand Oaks: Sage.
Mollerstrom, W. W., Patchner, M. A., & Milner, J. S. (1992). Family
functioning and child abuse potential. JournalofClinical
Psychology, 48,445–454.
Moncher, F. (1995). Social isolation and child-abuse risk. Families
in Society: The Journal of Contemporary Human Services, 76,
421–433.
Moos, R. S., & Moos, B. S. (1976). A typology of family social environ-
ments. Family Process, 15,357–371.
Moos, R. S., & Moos, B. S. (1986). Family environment scale manual.
Palo Alto: Consulting Psychologist Press.
Ortega, D. M. (2002). How much support is too much? Parenting
efficacy and social support. ChildrenandYouthServices
Review, 24,853–876.
Paavilainen, E., & Astedt-Kurki, P. (2003). Functioning of child maltreat-
ment families: lack of resources of caring within the family.
Scandinavian Journal of Caring Sciences, 17,139–147.
Rodriguez, C. M. (2008). Ecological predictors of disciplinary style and
child abuse potential in a Hispanic and Anglo-American sample.
Journal of Child and Family Studies, 17,336–352.
Rodriguez, C. M. (2010). Personal contextual characteristics and cogni-
tions: predicting child abuse potential and disciplinary style. Journal
of Interpersonal Violence, 25,315–335.
Rodriguez, C. M., & Green, A. J. (1997). Parenting stress and anger
expression as predictors of child abuse potential. Child Abuse &
Neglect, 21,367–377.
Rodriguez, C. M., & Richardson, M. J. (2007). Stress and anger as
contextual factors and preexisting cognitive schemas: predicting
parental child maltreatment risk. Child Maltreatment, 12,325–337.
Roosa, M. W., & Beals, J. (1990). Measurement issues in family assess-
ment. The case of the family environment scale. Family Process, 29,
191–198.
Sedlak, A. J., Mettenburg, J., Basena, M., Petta, I., McPherson, K.,
Greene, A., et al. (2010). Fourth National Incidence Study of Child
Abuse and Neglect (NIS–4): Report to Congress, Executive Summary.
Washington, DC: U.S. Department of Health and Human Services,
Administration for Children and Families.
Sidebotham, P. (2001). An ecological approach to child abuse: a creative
use of scientific models in research and practice. Child Abuse
Review, 10,97–112.
Stith, S. M., Liu, T., Davies, L. C., Boykin, E. L., Alder, M. C., Harris, J.
M., et al. (2009). Risk factors in child maltreatment: a meta-analytic
review of the literature. Aggression and Violent Behavior, 14,13–29.
Stockhammer, T. F., Salzinger, S., Feldman, R. S., Mojica, E., &
Primavera, L. H. (2001). Assessment of the effect of physical child
abuse within an ecological framework: measurement issues. Journal
of Community Psychology, 29,319–344.
Straus, M. A., Hamby, S. L., Finkelhor, D., Moore, D. W., & Runyan, D.
(1998). Identification of child maltreatment with the Parent–child
Conflict Tactics Scales: development and psychometric data for a
national sample of American parents. Child Abuse & Neglect, 22,
249–270.
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics
(3rd ed.). New York: Harper Collins.
Taylor, C. A., Guterman, N. B., Lee, S. J., & Rathouz, P. J. (2009).
Intimate partner violence, maternal stress, nativity, and risk for
maternal maltreatment of young children. American Journal of
Public Health, 99,175–183.
U.S. Department of Health & Human Services (2011). Child physical
maltreatment 2010. Retrieved August 2012 from http://archive.acf.
hhs.gov/programs/cb/pubs/cm10/cm10.pdf.
Varni, J. W., & Setoguchi, Y. (1993). Effects of parental adjustment on the
adaptation of children with congenital or acquired limb deficiencies.
Journal of Developmental & Behavioral Pediatrics, 14,13–20.
Vaux, A., & Harrison, D. (1985). Support network characteristics associ-
ated with support satisfaction and perceived support. American
Journal of Community Psychology, 13,245–268.
Vaux, A., & Wood, J. (1987). Social support resources, behavior, and
appraisals: a path analysis. Social Behavior and Personality, 15,
105–109.
Wekerle, C., Wall, A. M., Leung, E., & Trocme, N. (2007). Cumulative
stress and substantiated maltreatment: the importance of caregiver
vulnerability and adult partner violence. Child Abuse & Neglect, 31,
427–443.
Whipple, E. E., & Richey, C. A. (1997). Crossing the line from physical
discipline to child abuse: how much is too much? Child Abuse &
Neglect, 21,431–444.
Whipple, E. E., & Webster-Stratton, C. (1991). The role of parental
stress in physically abusive families. Child Abuse & Neglect, 15,
279–291.
Williamson, J. M., Borduin, C. M., & Howe, B. A. (1991). The ecology
of adolescent physical maltreatment: a multilevel examination of
adolescent physical abuse, sexual abuse, and neglect. Journal of
Consulting and Clinical Psychology, 59,449–457.
Wissow, L. S., & Wilson, M. E. (1992). Use of epidemiological data in
the diagnosis of physical child abuse: variations in response to
hypothetical cases. Child Abuse & Neglect, 16,45–55.
186 J Fam Viol (2014) 29:175–186