he 1

1::

:35

E

Jinnx
‘ :u‘ c h :.
.q.u_nn:

. n)

 

 

 

 

. armg 3

‘ 14::1:

3% ,
awfifl

A???

, a;

k? ”.4.”

.1...“ .y
m,

3% W a»

. an

a“.
$7.}...
no:

. .2

c
‘ . .5511 .
the.
MP 1.

 

THESIS

2" ‘ LIBRARY

Michigan State
University

 

 

 

 

This is to certify that the

dissertation entitled

Examining Physical Child Abuse Potential as a
Function of Risk load, Maternal Affective
Risk, and Self-rated Parenting Skill

presented by

Lucy H. Seabrook

has been accepted towards fulfillment
of the requirements for

Ph.D . degree in Psychology

 

 

/4..V p 57' ”Mé/
Ol/ Majorpyafidér /
Date 2.7 NW 2000

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

8n] 2 -nonn’h

 

“r EAYE‘BY'J?

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 chIHC/DateDuepfiS—p.“

M_._“

EXAMINING PHYSICAL CHILD ABUSE POTENTIAL AS A FUNCTION OF RISK
LOAD, MATERNAL AFFECTIVE RISK, AND SELF-RATED PARENTING SKILL

By

Lucy H. Seabrook

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Psychology

2000

ABSTRACT
EXAMINING PHYSICAL CHILD ABUSE POTENTIAL AS A FUNCTION OF RISK
LOAD, MATERNAL AF FECTIVE RISK, AND SELF-RATED PARENTING SKILL

By

Lucy H. Seabrook

Despite recognition of well-established risk factors for child maltreatment, and
evidence that more risk factors increases the likelihood of poor outcomes, the utility of
risk load theory as a framework with which to investigate the structural relationships
among such risk factors relative to child abuse potential remains unknown. This study
examined the roles of family risk load, maternal afl‘ect variables (depression and
parenting stress), and parenting skill in regard to increased potential to physically aggress
against a child. Results indicate that higher risk load, increased maternal depression and
increased parenting stress are cumulatively related to higher physical child abuse
potential. Self-rated parenting skill, while negatively affected by parenting stress, did not
account for significant variance in abuse potential. Environmental contexts of mothers
who scored highest on child abuse potential are described. Future child maltreatment

research utilizing risk load theory is encouraged.

Copyright by

Lucy H. Seabrook
2000

This document is dedicated to Mr. Yellow Bunny, who continuously reminds me about
the important things in life: shared warmth, unconditional love and t-u-n-a. Thanks also
to my dear girlfi-iends, without whom my life would be much less meaningful and fun:
you are the greatest and I love you dearly.

ACKNOWLEDGMENTS

Many thanks to my advisor, Hiram E. Fitzgerald, for his years of guidance. I
appreciate his mentorship, his humor and frankness, and the opportunities he has

provided to me.

TABLE OF CONTENTS

List of Tables .................................................................................................................... vii
List of Figures .................................................................................................................. viii
Introduction .......................................................................................................................... 1
Risk Load ............................................................................................................................. 4
Affective Risk: The Distressed Mother ............................................................................. 10
Depression ..................................................................................................................... 10
Parenting Stress ............................................................................................................. 11
Knowledge of Infant Development .................................................................................... l3
Self-rated Parenting Skill ................................................................................................... 15
Maternal Variables as Links between Risk Load and Abuse Potential ............................. l7
Purpose and Goals .............................................................................................................. 20
Hypotheses ......................................................................................................................... 20
Methods .............................................................................................................................. 22
Participants and Description of the Sample .................................................................. 22
Recruitment ................................................................................................................... 24
Data Collection ............................................................................................................. 24
Measures ....................................................................................................................... 26
Results ................................................................................................................................ 3 1
Data Estimation ............................................................................................................. 31
Descriptive Analyses .................................................................................................... 32
Correlations ................................................................................................................... 39
Analyses ........................................................................................................................ 40
Additional Analyses ...................................................................................................... 46
Discussion .......................................................................................................................... 52
Strengths and Limitations .................................................................................................. 66
References .............................................................................................. . ........................... 74
Appendix ........................................................................................................................... 97

vi

LIST OF TABLES

Table 1: Demographic Data

Table 2: Correlations among depression, parenting stress, &
child abuse potential subscales

Table 3: Missing Data/Valid Cases

Table 4: Descriptive Statistics of Scaled Scores

Table 5: Summary of Individual Risk Factors of the Risk load Index

Table 6: CAPI Subscale Scores & Comparison Norms

Table 7: Correlations among Scaled Scores

Table 8: Correlations among all Variables

Table 9: Descriptive Data on Child Abuse Potential Clusters

Table 10: One-way AN OVAs individual risk factors and child
abuse potential scores
Table 10a: Descriptives
Table 10b: ANOVAs

Table 11: Model 1 Predictors = abuse & neglect, single parenthood,
teenage parent, difficult temperament, unstable residence.
Dependent Variable = Child Abuse Potential

Table 12: Differences among Child Abuse Potential Clusters
(One-way AN OVAs comparing Clusters formed by Child
Abuse Potential Scores)

Table 13: Output for Model 1

vii

25

98
99
33
35
38
40
100
47

51
51

52

105
113

LIST OF FIGURES

Figure 1: Proposed Conceptual Model
Figure 2: Model 1 (Full Model)

Figure 3: Model 2 (Indirect Effects Model)
Figure 4: Model 3 (Direct Effects Model)

viii

43
45
46

INTRODUCTION

Across a range of theoretical positions and populations, research has consistently
demonstrated that environmental characteristics affect parent-child interactions
(Garbarino, 1976, 1977; Holden & Ritchie, 1989). In particular, negative life
circumstances, such as poverty and social isolation, have been shown to have a
detrimental impact on parenting and to be associated with increased parent-child
interactive stress (Conger, McCarty, Yang, Lahey & Kropp, 1984; Mash & Johnston,
1990). Risk exposure is quite diverse, however, because “at risk” families may vary
widely in personal, parental and family characteristics and resources (Coll et al., 1996).
Researchers (e. g., Webster-Stratton, 1990) have pointed to the need to better understand
the cumulative effects of various factors in combination, or the relative effects of
different risk factors, on parenting behaviors.

The term “risk factor” refers to any influence that increases the probability of
onset, digression to a more serious state, or the maintenance of a problem condition (Coie
et al., 1993). Of the many variables that are considered risk factors for child
maltreatment, the most salient are those that arise from parental characteristics and
adverse environments (Ammerman, 1990; Mash & Johnston, 1990; Reid, Barbera-Stein
& Bennett, 1986). As such, risk for abusive parenting can be viewed as being comprised
of both parent-level variables and environmental variables that are external to the parent.

The family stress literature has described an association between the number of
stressors experienced by families and level of family functioning (Egeland, Breitenbucher
& Rosenberg, 1980). That is, as the number of concurrent risk factors increases, the

likelihood for poor outcomes also increases (Sameroff, Bartko et al., 1998; Samerofi &

Seifer, 1983; Spicer & Franklin, 1994). This notion is related to models of risk
cumulation, risk load and stress pile-up (Cicchetti & Toth, 1995; Liaw & Brooks-Gunn,
1994; McCubbin & McCubbin, 1993; Sameroff, 1975). Risk load theory posits that the
effects of multiple, non-specific risk factors may be cumulative in that the presence of
more risk factors is related to a higher certainty of negative outcome (Samerofl‘ & Seifer,
1983; Seifer, Sarneroff, Baldwin & Baldwin, 1992).

Questions remain regarding whether a family’s risk load also serves as an
important determinant of parenting practices, and if so, how heightened risk load
translates into tendencies 'for particular parental behaviors. In particular, is there a link
between a family’s risk load and a mother’s increased potential for physically abusive
parenting? What are the conditions—over and above a set of generic risk factors for all
family functioning problems—that “produce” abuse potential, and how do maternal
characteristics interact with such conditions?

To give order to the growing body of research on risk, a conceptual framework
that incorporates both individual and contextual conditions affecting the probability of a
problem is needed (Fraser, 1997). In the current study, maternal affective variables of
depression and parenting stress, self-rated parenting skill and knowledge of infant
development were examined for their contribution to the proposed relationship between
risk load and physical child abuse potential. Physical child abuse potential—related to a
restricted range of options in disciplining, lack of flexibility in child rearing, and
particular attitudes about parenting (Trickett & Kuczynski, 1986)——rather than abusive
behaviors per se, is the outcome of interest: Other studies (Jackson, Thompson,

Christiansen, Colman et al., 1999) have referred to this as “abuse proneness.”

Conceptually, increased risk load is seen as the driving force toward physical
child abuse potential, mediated by maternal depression and parenting stress. These
relationships, however, are hypothetically affected by an interaction between two
additional parent-level factors: knowledge of infant development and self-rated parenting
skill. Depression and parenting stress may be understood as indicators of a mother’s
emotional well-being, likely to influence the way she relates to her children. There are
other possible psychological mechanisms that might account for the association between
risk load and maternal behavior. However, depression and parenting stress seem to have
particular promise as links in the causal chain leading to abuse potential because of their
ties to previous research and theory (Conger et al., 1984). In addition, knowledge of a
child’s normal developmental milestones and a perception of oneself as a skillfirl parent

are also important variables in the probabilistic framework of maltreatment etiology.

 

Figure 1: Proposed Conceptual Model

 

 

Parenting Stress‘]

 

 

 

l Dep’°55i°“l Child Abuse

\ Potential

Child Abuse
Moderator
Parenting /
“KnOWohow”

 

 

 

 
 
 
    
 
   
   

   

  
 

Contextual
Risk

  

 
 

Affective Risk

‘
\
\
\
‘

   
  

M ediator “a

 

# of
Risk Factors

 

 

 

 

 

Knowledge of
Self-rated Infant
Parenting Skill Development

 

 

 

 

 

 

 

 

 

Theoretical support for the proposed model (Figure l) is gleaned from several
sources, including an ecological model of etiology (Bronfenbrenner, 1979), which directs
attention to a range of environmental and individual influences, and stress models, which
postulate that life stress may accelerate child maltreatment, given the presence of
predisposing factors (Hillson & Kuiper, 1994; Howze & Kotch, 1984; Kotch et al., 1995).
While employing the concepts of stress and multiple systems of influence, however, the
model does not fully invoke the typology of either the ecological or stress theories.

Risk Load

Samerotf, Seifer and colleagues (Sameroff & Chandler, 1975; Samerofl & Seifer,
1995; Samerofi‘, Seifer, Zax & Barocas, 1987) conducted several large studies (both
longitudinal and cross-sectional) attending to the analysis of multiple risk factors in the
family environment. Initially, these studies were focused on socioeconomic status (SES)
as an overarching risk factor for families. With increasing evidence, however, SES was
broken down into individual risk factors—“correlates but not equivalents” of
socioeconomic status (p.23 8)—including social status, parental mental illness and
perspectives, family stress and family size. It was discovered that high risk load
predicted poorer child outcomes for both high and low socioeconomic status groups. In
these studies, risk load generally ranged from 0 to 9 or 13 factors (Sameroff, Bartko,
Baldwin, Baldwin & Seifer, 1998), while in other studies, Sameroff and colleagues
divided risk load into levels of low (0 — 1 risk factors), medium (2 — 3 risk factors) and
high risk (4+ risk factors). Findings consistently demonstrated that risk load scores were

better predictors of child-related outcome (viz., IQ, competence and mental health, or

other indicators of social-emotional outcome) than were single risk factors. Other
investigators found similar results.

Rutter (1979) found that psychiatric risk for 10-year-olds rose from 2% in
families with zero or one risk factor to 20% in families with four or more risk factors.
The six risk factors considered were marital distress, low SES, family size, paternal
criminality, maternal mental health and child admission to foster care. However, it was
the total number of risk factors and not the type of factor that affected the outcome.
Moreover, experiencing more than four risk factors did not affect the predicted outcome.
Two studies examining child behavior problems with a risk load index (Williams,
Anderson, McGee & Silva, 1990, and Ackerman, Schoff, Levinson, Youngstrom, &
Izard, 1999) found that poor outcome was far more likely as families experienced more
than six risk factors.

Within each of these studies there were significant effects for single risk factors,
yet no single variable was the determinant of outcome: only in families with multiple risk
factors was poor outcome manifest. Sameroff and colleagues (1998) maintain that
particular risk factors—such as SES and marital status—seem to make major
contributions to outcome not in that they are overarching variables in themselves, but that
they are strongly associated with combinations of other risk factors. These findings
suggest the usefulness of representations that aggregate risk factors. The concept of
effectively using a risk load index has also been demonstrated relative to parents’
perceptions of family environment and attitudes about parenting (Peterson & Hawley,

1998) and family functioning (Dickstein et al., 1998). However, the usefulness of risk

load as a theoretical framework with which to examine physical child abuse potential has
not been tested.

The maltreatment literature, however, is replete with individual risk factors that
are clearly associated with abuse. For example, McCurdy (1995) described factors
associated with parents who scored highest on abuse potential: an adult household
member with mental illness or developmental delay, living in a shelter or temporary
housing, having children in placement, and the experience of childhood abuse.1 Other
factors that were associated with high child abuse potential scores (although not the
highest risk) were the presence of violence in the household, living in substandard
housing, parental inability to solicit social support, and social isolation. Finally, factors
that were related to above-average abuse potential were “adult functioning problems,”
such as lack of interest in the child, poor personal hygiene, neglect of routine household
responsibilities, inability to manage anger, excessive need for the child to comply with
parental demands, and low self-esteem (McCurdy, 1995). Such findings indicate that
there are many factors that place a family at risk for dysfunctional parenting.

These many risk factors—each with its own significant literature base validating
its potential negative impact on parenting—may be categorized into biological,
economic, parental, family-structural and extra-familial domains. For example, child-
oriented risk factors for maltreatment include low birth weight and nutritional deficit
(Browne & Saqi, 1988; Herrenkohl & Herrenkohl, 1979; McGrath & Meyer, 1992), as
well as language deficiency or developmental immaturity (Larsen & McCreary Juhasz,

1985; McCurdy, 1995; Steele & Pollock, 1974). Economic hardship and maltreatment

 

' Together, these factors accounted for 30% of the variance in abuse potential scores
(McCurdy, 1995).

are linked via unemployment and low family income (Cantrell, Carrico, Franklin, &
Grubb, 1990; Cicchetti & Lynch, 1993; Gelardo & Sanford, 1987; Gelles, 1973; Gil,
1970; Glachan, 1991; Herrenkohl, Herrenkohl, Rupert, Egolf, & Lutz, 1995; Kowal et al.,
1989)

Parent-level risk factors include a wide range of variables, such as physical
disability, history of abuse or neglect (Belsky, 1984, 1993; Caliso & Milner, 1992;
Fennell & Fishel, 1998; Haynes, Cutler, Gray & Kempe, 1984); or chronic
mental/emotional illness (Mash & Johnston, 1990; McCurdy, 1995). Also important
relative to child maltreatment are a difficult temperament, history of delinquency,
violence or incarceration (Browne & Saqi, 1988; Connelly & Straus, 1992; Kotch et al.,
1995); substance abuse or addiction (Dore, Doris & Wright; 1995; McCurdy, 1995);
teenage parenting (Conger et al., 1984; Connelly & Straus, 1992; Fulton, Murphy &
Anderson, 1991; Jones & McCurdy, 1992); and low educational achievement (Brayden et
al., 1992; Kotch et al., 1995; Straus, 1983).

Structural elements of the family also play a role in risk for poor outcomes. These
may include single parenthood (Bird, 1997; Webster-Stratton, 1990; Zuravin, 1988), high
family density (Bell, Johnson, McGillicuddy-Delisi & Sigel, 1980; Biller & Solomon,
1986; Bredehoft, 1990; Egeland & Brunnquell, 1979; Straus, Gelles & Steinmetz, 1980),
and divorce (Hetherington, Cox & Cox, 1982; Webster-Stratton, 1990). There are also
extra-familial factors that have a negative impact on parenting, including a lack of a
stable residence (Garbarino & Sherman, 1980; Perry, Wells & Doran, 1983; Wolfe, 1987)
and housing in a rural or segregated area (Garbarino & Sherman, 1980; Gil, 1975;

Sharma, Singh & Sidhu, 1985).

While this list of individual risk factors is extensive, it is not inclusive of the
many variables that may place a family at risk for poor parenting outcomes. In a
multiple-risk factor approach, risk is conceptualized as an accumulated set of adverse
experiences or conditions that aflect families (Dickstein et al., 1998). Thus, risk load is
operationalized as the sum of variables characterizing a family situation relative to family
and environmental conditions that may increase the potential for abusive parenting. Any
one specific risk factor does not define risk; rather it is the additive effect of a number of
risk factors that is considered detrimental to parent functioning.

The risk factors listed above are often conceptualized as family stressors. The
stress literature is traditionally classified into two theoretically separate approaches,
related to either [1] chronic strains or conditions that exact a toll over time because of the
persistence of difficult environmental factors or [2] new demands or important changes in
‘ people’s life situations. Researchers have used a variety of labels to describe these two
stress types: chronic vs. situational stressors (Linsky, Straus & Colby, 1985), everyday
vs. event stressors (Cmic & Acevedo, 1995), and enduring vs. transitional stressors
(Cicchetti & Rizley, 1981). Importantly, the majority of risk factors examined in the
current study are considered chronic, rather than acute, stressors.

As argued by Quittner, Glueckauf and Jackson (1990), the effects of chronic
stressors are likely to be more pervasive, leading to alterations in several life domains,
including parental and work roles, expectations for the future, and relationships within
the larger social milieu. Chronic stress in this context also implies its continuation into
the future. Thus, in terms of both its sc0pe and enduring impact, risk load as assessed in

this study reflects chronic stress that may be differentiated from both major life events

and daily hassles (Quittner, Glueckauf & Jackson, 1990). Such chronically stressful
conditions may affect the parent-child relationship through parents’ psychological well
being.

Indeed, besides their association with parenting attitudes and behavior, the risk
factors described above have been shown to have a detrimental impact on parental
psychological well-being in general (see Conger et al., 1984). That is, a family context
characterized by multiple risk factors is also more likely to positively affect individual-
level parent variables, such as depression and stress levels. Simons and Johnson (1996)
maintain that the disruptive effect of stress on parenting is indirect through its impact on
the parent’s emotional state, a claim supported by several other studies (Conger, Elder,
Lorenz, Simons & eritbeck, 1992; Simons, Beaman, Conger & Chao, 1993; Simons,
Whitbeck, Melby & Wu, 1994). This notion is based on a portion of Belsky’s (1984)
process model of parenting, which asserts that parental functioning is multiply
determined and that sources of contextual stress can affect parenting directly or indirectly
by first impacting individual psychological well-being.

Thus, while the importance of contextual risk is recognized, special emphasis is
placed on parents’ affective state as a primary determinant of how a family’s risk load
may influence parental behavior (Abidin, 1990, 1992; Mash & Johnston, 1990; Webster-
Stratton, 1990). In sum, the importance of examining the contribution of parental
affective variables is based on the assumption that high risk alone does not directly
predict poor outcomes: the relationship between risk load and abuse potential may be
mediated by factors that pertain to parental fimctioning and coping (Dukewich,

Borkowski & Whitman, 1996; Mash & Johnston, 1990). This notion is in accord with

Belsky’s (1984) assertion that the personal resources of the parent are more important
than contextual variables in determining parenting behavior.
Affective Risk: The Distressed Mother

A large body of literature documents ineffective parenting among distressed
parents (Cicchetti & Howes, 1991; Gaines, Sandgrund, Green & Power, 1978; Gelfand,
et al., 1992; Haskett, Smith Scott, & Farm, 1995; Lahey, Conger, Atkeson & Treiber,
1984; Rodriguez & Green, 1997; Spicer & Franklin, 1994; Whissell, Lewko, Carriere &
Radford, 1990). Parents who physically abuse their children have been found to report
both higher stress and depression levels (Gelfand, Teti & Fox, 1992; Mash, Johnston, &
Kovitz, 1983; Weissman Wind & Silverrr, 1994; Whipple & Webster-Stratton, 1991;
Wolfe, 1985) and stress-related symptoms (Wolfe, 1985).

In addition to affecting parenting behavior, depression and parenting stress also
positively affect each other (Conger et al., 1984; Cummings & Davies, 1994; Pianta &
Egeland, 1990; Reis, Barbera-Stein & Bennett, 1986). Gelfand et al. (1992) affirmed that
the greater the severity of maternal depression, the more likely were mothers to
experience stress related specifically to the parenting role. Both depression and parenting
stress have been consistently linked with physical child abuse, as described in the
following sections.

Depression

Depression is the most widely studied disturbance in parental psychological
frmctioning as a well-known risk factor for child maltreatment (Belsky, 1984; Gelardo &
Sanford, 1987; Gelfand et al., 1992; Kotch et al., 1995; Pianta & Egeland, 1990; Whipple

& Wilson, 1994). Depressed mothers tend to make more negative statements about their

10

children, to use guilt- and anxiety-provoking methods of controlling their children, and to
have unrealistic expectations regarding their children’s ability to self-regulate
(Ammerman, 1990; Cummings & Davies, 1994; Jameson et al., 1995; Lahey et al.,

1984). Depressed parents also have a much lower tolerance threshold for child
misbehavior and resort more quickly to high authoritarian, overcontrolling parenting
(Dore, 1993; Evans, 1980; Gelfand, Teti & Fox, 1992; Herrenkohl & Herrenkohl, 1979;
Kotch et al., 1995; Whipple & Webster-Stratton, 1991, 1994). Moreover, an examination
of mothers who were able to break the cycle of intergenerational abuse revealed that
depression significantly differentiated the abusers from the nonabusing mothers (Egeland
et al., 1988). One reason behind this connection may be that depressed mothers have
relatively little confidence in their caregiving abilities and low levels of parental self-
efficacy, which, in turn, have been associated with parenting impairments (see Cummings
& Davies, 1994; Frankel & Harmon, 1996).

Depression may be especially sensitive to a distinctive type of stress experience—-
namely, undesired experiences that are both enduring and resistant to efforts of change
(Pearlin, Menaghan, Lieberman, & Mullan, 1981). Certainly, both these features
characterize the individual risk factors that comprise risk load in this study.

Parenting Stress

Parenting stress is another variable that is clearly implicated in riSk for physical
child abuse (Abidin, 1990; Crnic & Greenberg, 1990; Rodriguez & Green, 1997;
Webster-Stratton, 198 8). Parenting stress has been associated with inconsistent parental
discipline, coercive parent-child interactions, and child maltreatment (Pianta & Egeland,

1990; Rogers, 1998). In Crnic and Greenberg’s (1990) research, high-stress mothers

11

interacted in a less positive manner with their infants than did low-stress mothers. Others
(Egeland & Brunnquell, 1979; Justice & Duncan, 1976) have also found that abusing
parents experience greater stress levels than non-abusing parents. While stress arising °
from many sources can burden mothers and affect both their perceptions of parenthood
and their parenting behavior (Gelfand, Teti & Fox, 1992), DeLongis et al. (1988) have
argued that ongoing stressors specifically related to being a parent (such as increased
economic hardship or responsibility for child care) play a more central role in parenting
than do major life events. Indeed, Gaudin and Pollane (1983) found that stress that was
particular to the role of parent was the largest contributor to variance in the prediction of
child abuse.

Whipple and Webster-Stratton (1991) found that abusive parents experienced
stress significantly more often from low income, maternal depression and anxiety,
negative life events, frequent child behavior problems and less marital support. Caring
for an infant—a difficult task in itself—can be especially difficult for parents who have
other pressing problems, as highly stressed parents lose their ability to care for their
children in a warm, sensitive and competent manner (Crnic, Greenberg, Ragozin,
Robinson & Basham, 1983; Gelfand, Teti & Fox, 1992). Indeed, at-risk parents report
higher levels of total parenting stress than other types of parents (Holden, Willis & Foltz,
1989). However, some studies report rather modest relationships between parents’ stress
levels and authoritarian, power assertive discipline strategies (Deater-Deckard & Scarr,
1996)

Moreover, despite the strong links among parenting stress, depression and abusive

parenting, not all emotionally distressed parents perpetuate abuse, suggesting that some

12

additional factors may exacerbate or pacify the pathways fi'om these parental risk factors
to physical child abuse. For example, impaired parenting skills and low levels of
awareness and understanding of the difficulties and demands involved in parenting are
also associated with abusive parenting (Dukewich, Borkowski & Whitman, 1996;
Egeland, Deinard, Brunnquell, Phipps-Yonas & Chricton, 1979; Fennell & Fishel, 1998;
Garbarino, 1976, 1977; Houck & King, 1989; Reis, Barbera—Stein & Bennett, 1986).
How parenting skill- and knowledge-based factors fimction in the relationship between
maternal affective risk and abuse potential, however, is unclear.
Knowledge of Infant Development

As a construct related to cognitive readiness to parent, knowledge of infant
development pertains to factors that are directly related to the knowledge and skills
needed to perform the functions of parental care for children. Parents must have a
minimum knowledge of the normal range of child development in order to adequately
meet their children’s needs (Burke et al., 1998); it is well recognized that a lack of
knowledge about children’s development is a risk factor for poor parenting (see Spinetta
& Rigler’s 1972 review of the literature; Benasich & Brooks-Gunn, 1996; Haskett,
Johnson & Miller, 1994). The opposite relationship holds as well: more knowledge about
normal child development—including realistic expectations for behavior and
development—is related to increased parenting skills,2 quality of the home environment
and positive child outcomes (Fulton, Murphy & Anderson, 1991; Reid, Barbera-Stein &
Bennett, 1986). Stevens (1984) found that, after controlling for income and education,

mothers who scored highest on measures of child development knowledge showed higher

 

2 Stevens (1984) has estimated that knowledge of development accounts for 20% of the
variance in parenting skill.

13

levels of parenting skill. Similarly, Benasich and Brooks-Gunn (1996) reported that
maternal knowledge at 12 months was positively related to quality of home environment
provided by the mother simultaneously and at 36 months.

Multiple studies have demonstrated that abusive parents have rnisperceptions
about the capabilities of their child. Many of these provide evidence that the parent
usually overestimates the child’s abilities and shortens the time required to reach
developmental milestones (Bavolek et a1, 1990; Hefler, 1973; Justice & Justice, 1976;
Larsen & McCready Juhasz, 1985; Steele & Pollack, 1974). However, more recent
studies have found that inaccurate parental expectations Operated in the opposite
direction: abusive parents expected their children to develop certain skills at older ages
than normal (Perry, Wells & Doran, 1983; Twentyman & Plotkin, 1982). Meta-analytic
studies (Black, Heyman, & Smith Slep, in press) conclude that physically abusive
mothers have higher than normal expectations of their children, as well as less
understanding of appropriate developmental norms.

Such rnisperceptions of child development are related to maltreating parents’
negative perceptions of their children and attributions of their children’s misbehavior as
deliberate efforts to challenge their authority (Azar, et al., 1984; Cicchetti & Howes,
1991; Cicchetti & Lynch, 1993; Dore, 1993; Egeland & Brunnquell, 1979; Trickett,
Aber, Carlson, & Cicchetti, 1991; Creighton, 1985). For example, Whipple and Webster-
Stratton (1991) found that abusive mothers were significantly more critical in their
interactions with their children and reported more child behavior problems, despite the
fact that independent observations did not concur with such reports. Moreover, families

with unrealistic expectations and cognitive distortions regarding children’s intentions at

14

entry are significantly less likely to improve over the course of a parenting intervention
(Kowal et al., 1989).

Parents who have more accurate views of child development may be more likely
to correctly perceive their child’s behaviors, be better prepared to establish healthy
relationships with their children, and be less likely to misplace their stress-related
aggression onto the child via abusive parenting (Egeland & Brunnquell, 1979; Mercer &
Ferketich, 1994; Shaner, Peterson & Roscoe, 1985). Thus, an important component of
skillful parenting would appear to be the accumulation of a sound knowledge base about
basic mechanisms underlying developmental processes as well as descriptive information
about the course of development (Stevens, 1984). Indeed, many parent-education
programs have been developed on the belief that knowledge of developmental norms will
enable parents to raise their children more effectively and prevent child maltreatment
(Fennell & Fishel, 1998; Olds, Kitzrnan, Cole & Robinson, 1997; Wasik & Roberts,
1994; US Dept. of Health & Human Services, 1990; Weiss, 1993).

Based on its link to parenting skill, knowledge of infant development may serve
either as a buffering or reinforcing variable relative to a mother’s affective risk and abuse
potential. A mother’s confidence in and perceptions of her own parenting skill may also
serve to strengthen or weaken the link between affective risk and abuse potential.
Self—Rated Parenting Skill

Wolfe (1985) has argued for a “better understanding and assessment of
psychological variables that exert an influence on parental competence, as opposed to
psychopathology” (p.479). A sense of competence and effectiveness in the parenting role

is essential to parental well-being and child adjustment (Esdaile & Greenwood, 1995;

15

Goodnow, 1988). In relation to parenting, a mother’s appraisal of her ability to perform
competently in a designated situation would be enhanced by positive earlier experiences
in parenting or successfully caring for children. Conversely, parenting self-efficacy
would be lowered if previous parenting experiences had been unrewarding or difficult
(Bandura, 1982, 1986).

Similar to self-efficacy, self-perceived parenting skill is a parent’s view of her
ability to meet the situational demands of parenting. A mother’s perception of her skill in
the parenting role reflects her maternal confidence, which is a basic determinant of her
capacity as a mother (Bullock & Pridharn, 1988; Mercer & Ferketich, 1994). Self-
evaluations of parental skill influence a range of weighty factors, including the way in
which the parent processes and attends to information, the amount of effort she devotes to
child-rearing, and her emotional reactions to child behavior and the parenting role
(Esdaile & Greenwood, 1995; Goodnow, 1988).

Several studies have demonstrated that a parent’s perception of her skills is highly
related to parenting competence (McClennan & Harkless, 1998; Teti & Gelfand, 1991).
Maternal confidence has also been related to mothers’ self esteem, mental health,
adaptation to parenting and perception of infant temperaments (see Conrad, Gross, Fogg
& Ruchala, 1992; Holden & Banez, 1996; Teti & Gelfand, 1991). Conversely, a low
sense of parenting competence is related to the use of coercive and intense disciplinary
tactics in efforts to control child behavior mugenthal, Blue & Lewis, 1989; Bugenthal,
Blue & Cruzcosa, 1989; Bugenthal & Cortez, 1988; Mash, Johnston & Kovitz, 1983).

However, this relationship is not as clear as it may seem. Some doubt the

systematic relationship of self-perceived parenting skill (i.e., subjective maternal

16

experiences) into actual quality of parenting (i.e., mothering behaviors). While several
studies have noted positive correlations between maternal confidence and quality of
mother-child interactions (Bohlin & Hagekull, 1987; Gelfand & Teti, 1991), other
researchers (Julian, 1983; Walker, Crain & Thompson, 1986; Williams et al., 1990; Zahr,
1991) have found nonsignificant relationships, and one investigator (Fuller, 1989) found
an inverse relationship between maternal confidence and quality of mother-child
interactions.3 Such discrepant findings direct our attention to the complexity in how a
mother perceives her parenting skills and how that subjective experience relates to actual
behavior.
Maternal Variables as Links between Risk load and Child Abuse Potential

Relationships among depression, parenting stress, self-ratings of parenting skill
and knowledge of infant development may present a complex pattern of parental
strengths and limitations that affect a mother’s potential to aggress against the child.
Parenting stress, depression and behavior are related phenomena, and various structural
relationships among them have been investigated (Rodgers, 1998). Results indicate that
both stress and symptomology independently impact parenting behavior, but more
strongly in a direct manner, rather than through the other. Thus, in the current study,
depression and parenting stress were assessed as separate constructs that operate in a
parallel fashion to impact child abuse potential. Additional variables associated with
maternal parenting abilities—viz, knowledge of developmental milestones and a parent’s
perspective of her own skills—may strengthen or weaken existing relationships among

depression, parenting stress and child abuse potential.

 

3 With a sample of teenage mothers (Fuller, 1989).

17

To illustrate, it is conceivable that the mother who is depressed and highly
stressed may be prone to abuse, but if she feels competent as a parent, and also has an
understanding of her child’s needs and limitations, her potential to abuse may be
dampened. Conversely, a highly stressed and depressed parent who also rates her
parental skills poorly and knows little about her child’s development may be even more
likely to aggress against her child.

Conrad and colleagues (1992) found that the interaction between maternal
confidence and knowledge of early child development predicted quality of mother-child
interactions, but that neither was predictive of mother-child interactions when considered
alone: maternal confidence was related to quality of mother—child interaction only when
knowledge of child development was taken into account. Mothers who were least
knowledgeable about development but who maintained confidence about their parenting
abilities demonstrated the least positive interactions. The researchers refer to these
parents as “naively confidentz” those who are unaware of the complexities of parenting.
These mothers may be the same type as described by Sameroff and Feil (1985) as
viewing development from a concrete and simplistic perspective with little reflection on
their own parenting abilities or behaviors. Conrad et al. (1992) suggest that these
mothers may be the ones who have traditionally been most difficult to retain in parenting
intervention programs. It is important to note, however, that Conrad’s sample was
middle class and married. How self-rated parenting skill, knowledge and behavior play
out in a sample of low SES, high risk mothers is not known.

Given these ambiguous findings, what is needed is a framework that considers the

mutual influences of mother’s perceptions of her parenting skill and her understanding of

18

child development in the context of risk load and maternal affect. Behavior is influenced
by the combined effects of individuals’ knowledge about a task and their confidence in
their ability to perform that task successfully (Bandura, 1977, 1986, 1989); that is, both
knowledge and confidence operate together to predict performance. Thus, maternal
confidence is a necessary but not sufficient condition for successful task performance-—
mothers must also possess knowledge about child development and parenting (Conrad et
al., 1992). One may find that mothers who are operating within a high risk, stressful
environment need such maternal confidence and knowledge to safeguard them from
heightened abuse inclinations. As such, the current study offered that these variables be
considered both separately and jointly within the conceptual model (a) as a statistical
interaction for which parents’ standing on the two indicators created an interaction score
and (b) as separate variables. In addition, both mediating and moderating roles were of
interest.

In the case of a moderating role for self-rated parenting skill or knowledge of
infant development, it may be possible to explain when depression and parenting stress
lead to heightened abuse potential. To illustrate, it may be possible that some depressed
and stressed mothers hold adequate feelings of self-efficacy and knowledge in the
parenting role to reduce the potential to abuse. In contrast, a mediator specifies the how
or why of causal effects (Baron & Kenny, 1986). Thus, a mediating role for either self-
rated parenting skill or knowledge of infant development would indicate that the variable
of interest was significantly associated with maternal affective variables (i.e., depression
and parenting stress) and abuse potential, and that after removing the effects of the

mediating variable on abuse potential, depression and parenting stress should no longer

19

relate to abuse potential. In this case, a mother’s standing on skill and knowledge should
play a central role in determining her physical child abuse potential, mediating any
influences of depression and parenting stress.

P_urpgse & Goals

Considerable effort has been directed toward discovering the etiology of
maltreatment, which is critical in screening for high risk groups so that appropriate
treatments can be implemented to remediate associated deleterious effects (Ammerman,
1991; Dukewich, Borkowski & Whitman, 1996). Utilizing risk load as a theoretical
framework may provide clarity to the concept of risk for maltreatment by addressing the
cumulation of both contextual and individual elements of risk.

By focusing on the structural relationships among risk load, maternal affective
variables, and parental knowledge and skill variables, the goal of the study was to
elucidate the nature of maternal abuse potential in a high risk context. That is, is there a
relationship between chronic contextual stress (as indicated by risk load) and maternal
abuse potential, and if so, is there a set of maternal variables that is responsible for this
relationship? Such knowledge may provide guidance on the design and delivery of social
and health programs, in that they may better address the individual, family, and broader
contextual conditions that are salient to a mother’s heightened abuse potential.
Hypotheses

The conceptual model of interest in the current study is articulated in Figure 1.
As discussed in the literature review, associations among parenting stress and depression
and heightened risk for poor parenting outcomes are established. In addition, the concept

of risk load leading to poorer child outcomes and family environment has its own

20

literature base. However, these combined relationships have not been examined in the
context of child abuse potential. Moreover, the contribution of knowledge of infant
development and self-rated parenting skill relative to risk load and maternal affective
variables (i.e., depression and parenting stress) has not yet been fully examined.

Thus, preliminary hypotheses regarding the relationships among risk load and
parenting stress and depression were examined (Hypothesis 1: Higher risk load is
associated with greater depressive symptomology; Hypothesis 2: Higher risk load is
associated with increased parenting stress). Links between parenting stress and child
abuse potential and between depression and child abuse potential were also assessed
(Hypothesis 3: Greater depressive symptomology is associated with higher abuse
potential; Hypothesis 4: Increased parenting stress is associated with higher abuse
potential).

Since the roles of self-rated parenting skill and knowledge of infant development
are not clear in the literature, both moderating and mediating models were considered for
an interaction variable and separately (Hypothesis 5: The relationship between depression
and abuse potential will be moderated by the interaction between self-rated parenting
skill and knowledge of infant development; Hypothesis 6: The relationship between
parenting stress and abuse potential will be moderated by the interaction between self-
rated parenting skill and knowledge of infant development; Hypothesis 7: The
relationship between depression and abuse potential will be mediated by the interaction
between self-rated parenting skill and knowledge of infant development; Hypothesis 8:
The relationship between parenting stress and abuse potential will be mediated by the

interaction between self-rated parenting skill and knowledge of infant development.

21

Hypothesis 9: The relationship between depression and abuse potential will be moderated
by self-rated parenting skill; Hypothesis 10: The relationship between parenting stress
and abuse potential will be moderated by knowledge of infant development; Hypothesis
11: The relationship between depression and abuse potential will be mediated by the self-
rated parenting skill; Hypothesis 12: The relationship between parenting stress and abuse
potential will be mediated by knowledge of infant development. Hypothesis 13: The
relationship between parenting stress and abuse potential will be mediated by self-rated
parenting skill; Hypothesis 14: The relationship between parenting stress and abuse
potential will be moderated by self-rated parenting skill. Hypothesis 15: The relationship
between depression and abuse potential will be mediated by knowledge of infant
development; Hypothesis 16: The relationship between depression and abuse potential
will be moderated by knowledge of infant development).
Methods

Participants and Description of the Sample

Participants in the current study were a subset of families participating in a larger
study, an evaluation of a four-county Early Head Start (EHS) program located in
Michigan. EHS is a child development program offered to pregnant women, infants and
toddlers under age 3. This particular EHS program is a home—based program, with
weekly visits to the home or day care and monthly group meetings for theparent(s) and
child. By definition, most families who enroll for EHS services meet Federal guidelines
for poverty.

Criteria for selection of the sample from the larger pool of EHS children included:

a) in cases with siblings, the eldest child was retained; b) only cases where mother was

22

identified as the primary caregiver were retained;4 and c) cases without complete data on
child abuse potential were deleted. These selection criteria resulted in a, sample of 95
cases, described below.

Thirty-seven percent of children had a sibling who was also enrolled in the
program. Only a few children in the sample (4.3%) had enrolled prenatally. Mothers’
mean age at time of childbirth was 23.4 years ($1 = 5.6), ranging fiom 12.7 to 42 years.
Children averaged 23 months in age at the time of data collection (_sd = 8.0), ranging fiom
9 to 36 months. Slightly more than 62% of children in the sample were male. The
majority (89.5%) of children was identified as white, with 4.2% each identified as black
and Hispanic and 2.1% identified as Native American. Mothers’ ethnicity was described
as follows: 94.5% white, 3.3% Hispanic and 2.2% Native American. About 13% of
children had a health condition requiring an emergency care plan. Several children were
described as having delays in development (7.4%), hearing (2.1%), language (2.1%),
sociability (2.1%) or emotional health (1.1%).

About 94% of children were Medicaid-eligible cases. Almost 89% of cases
received WIC (Aid to Women with Dependent Infants and Children) and 48.9% of
households received food stamps. Over half (52.7%) rented their home, although 23.1%
owned their home and 11% resided in subsidized housing. In addition, 13.2% of cases
described their living arrangement as “other”—usually indicating that they lived with
their parents. Most families in the sample (43.3%) lived in a single family dwelling,

followed by mobile homes (24.4%), multifamily homes of 5 or more families (18.9%)

 

4 Data collection was aimed at mothers (regardless of coupled status), as she was
typically the parent targeted by the program. Where fathers were considered primary
caregivers, these cases were removed from the sample.

23

and multifamily homes of 2 — 4 families (12.2%) (1.1% were “other” housing type).
Regarding transportation, most (79.8%) owned a car, followed by “other” transportation
(15.7%) and “none” (4.5%). See Table 1 for the complete respondent demographic
profile. i

W

Parents were recruited for the Early Head Start program via a collaborative
recmitment effort serving the four county area This effort was controlled by the EHS.
program’s host agency. Recruitment efforts included posters advertising free preschool
with an “800” number for information and referrals through agencies working with the
target population (e.g., Early On, WIC, the intermediate school district, physicians,
domestic violence agencies, family day care homes, etc.). Applications were collected
and registered based on income eligibility. Families that did not qualify for the EHS
program (due to income ineligibility) were referred to other programs.

_D_a_t2_C_011_ec_tiQ_I;

Data for this study were derived from two sources: (a) program enrollment
documents and (b) surveys completed by parents/ guardians as part of the evaluation
project. The minimum child age for survey completion was 9 months. However, as
families could enroll in the program at different child ages (from prenatal to just under 36
months), the surveys were collected across a range of child ages. EHS staff conducted
data collection: a trained Home Visitor aided the parent in completing the survey in the
parent’s home, while an Intake Coordinator collected data regarding risk load at the
program enrollment office. These positions require a minimum of an Associates Degree

in early childhood education or child development.

24

 

Table 1: Demographic Data (N=95)

 

 

 

 

 

 

Valid _M_ _sc_i Min Max
%
Yearly income $9793.34 6175.91 $100.00 $27,526.00
Yearly rent + heat $3739.94 2460.19 $.00 $10,700.00
Head of household
Male 7.7
Female 92.3
Ethnicity
White 94.5
Hispanic 3.3
Native Am. 2.2
Maternal age 23.4 5.6 12.7 42
Marital status
married 39.3
divorced 10.1
never married 43.8
separated 5.6
widowed 1.1
Education 11.8 yrs 1.6 6 yrs 16 yrs
Employment status
employed FT 17.9
employed PT 17.9
unemployed 34.7
not seeking 21.1
Household
commsition
# people 3.7 1.3 l 7
Member w/disability 13.3
Spouse present 38.2
Living w/partner 18.2
# children in home 2.2 1.04 0 prenatal 5
Member rec.
unemployment 2.1
social security 1.1
pension 1.1
$81 7.4
2 members earn $ 25.3
3 members earn $ 22.1
Rec TANF support 41.1
ley TANF $ $4351.98 1647.13 $680.00 $6916.00
Rec child support 14.7
ley support $ $2056.60 1747.09 $100.00 $4940.00

 

 

25

arm-Tr“?

Measures

Risk Load. Risk load was operationalized as the total number of risk factors
experienced by a family. A multiple risk index was constructed from individual items
assessing risk in child, parent and family contexts (e.g., low birth weight, teenage parent,
unstable residency/fiequent moves). These individual items were scored dichotomously
as either “Present” or “Absent,” determined via interview with the parent.

The use of a single composite score of risk factors was based on the
nonspecificity hypothesis: all risk factors were assumed to have equal potential with
regard to total risk load. Thus, each risk factor was given the same weight and summed,
with a higher risk load score indicating a greater number of risk factors experienced by
the family. Alpha (KR-20) for this 24-item measure was .67.

Depression. Depression was assessed with the short form of the Beck Depression
Inventory (BDI). The BDI (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) is a
widely used phenomenological measure of depressive state that has received extensive
validational study. The short form of the BDI is a 21-item self-report instrument that
assesses cognitive, emotional, motivational, and physical manifestations of depression. It
is a better measure of psychological depression than of the biological aspects, and is more
sensitive to moderate levels of depression than more profound states.

Beck reports a split-half reliability of .93. Scores on the long and short forms of
the BDI correlate between .89 and .97. A meta-analysis of 25 years of data on the BDI
yielded an internal consistency mean coefficient alpha of .86 for psychiatric patients and
.81 for non-psychiatric patients. For the current study, this measure had an alpha of .84

(with an equivalent standardized item alpha).

26

 

Due to changes in data collection protocol, the Center for Epiderrriologic Studies
Depression Scale, Revised (CES-D-R) was also used in a small number of cases (i.e., 11).
CES-D items may be mapped onto the diagnostic criteria for major depression. When .
employing the CES-D-R as a scale, each item is coded from zero (“not at all or less than
one day”) to 3 (“five to seven days” or “nearly every day for two weeks”). While the
national EHS evaluation project reports a Cronbach alpha of .89 for white mothers, the
alpha for this sample was .40, with a standardized item alpha of .61.

While each instrument has particular strengths and limitations with different
populations (e. g., in general, the BDI has greater specificity with clinical populations)
both instruments discriminate individual differences in depressive severity. With
nonclininal (college student and adolescent) populations, the CBS-D may be more
appropriate for assessing dysthymia or depressive severity, while the BDI is better for
identifying cases of depression (Roberts, Lewinsohn, & Seeley, 1991; Wilcox, Field,
Prodromidis, & Scafidi, 1998). Both the EDI and the CES-D have been shown to detect
depressive symptoms fairly accurately in primary care settings, with sensitivities and
specificities that vary depending on the cut-off score selected.5 It is important to note,
however, that for the current study neither instrument is utilized for diagnostic purposes;
assessing the degree of depressive symptomology was the goal.

Parenting Stress. Parenting stress was measured with two subscales of the
Parenting Stress Index (PSI) short form, focusing on parental distress and dysfunctional

parent-child interactions. The original PSI (Abidin, 1983) is a 126-item questionnaire

 

5 For example, when compared to the diagnosis of major depression in primary care
patients, the BDI had a sensitivity of 100% and a specificity of 89% at a cut-off score of
16; the CES-D had a sensitivity of 89% and a specificity of 70% at a cut-off score of 27
(Zich, Attkisson, & Greenfield, 1990).

27

reflecting areas of potential stress in the parent-child relationship. The Parent Distress
subscale measures parent’s dej ectedness in their parenting role, with items such as “I feel
trapped by my responsibilities as a parent.” The Parent-Child Dysfirnctional Interaction
subscale measures a parent’s perceptions of the emotional quality of her relationship with

the child, relative to her expectations about the parent-child relationship. A

 

representative item is “When I do things for my child I get the feeling that my efforts are
not appreciated very much.” Each item is rated on a 5-point scale ranging from 1 h

(strongly agree) to 5 (strongly disagree) indicating increasing levels of perceived

 

parenting stress by parents.

The PSI has been the subject of considerable empirical scrutiny and demonstrates
good psychometric properties (see Abidin, 1983), with a 6 month test-retest reliability of
.7 to .8 and Cronbach’s alpha at .8 (Deater-Deckard & Scarr, 1996). Several studies on
the temporal stability of the PSI report that test-retest reliabilities range between .55 and
.96, depending on the specific population tested and the length of the inter-test interval
(Abidin, 1990). In this study, this 23-item measure had an alpha of .87, with a
standardized item alpha of .88.

Knowledge of Infant Development. Parent’s knowledge about child development
was measured with the short form of the Knowledge of Infant Development Inventory
(KIDI) (Macphee, 1981), a subset of 14 questions from the original KlDI. This measure
includes statements that reflect how babies behave, how they develop, and how to best
care for them. The use of the KIDI short form in this project is supported by its prior use

/’ in the national Early Head Start evaluation (Benasich & Brooks-Gunn, 1996).

28

The long form of the KIDI, a 75-item questionnaire, samples fiom three areas of
parental knowledge: child—rearing practices, developmental processes and infant
developmental norms and milestones. The KIDI is written to reflect a sixth to seventh
grade reading level. Each of the items is a statement about infant behavioral norms, such
as “A one-year—old knows right from wrong.” Items are dichotomized into Knowledge
(i.e., combination of strongly agree and agree) and No knowledge (i.e., combination of
strongly disagree, disagree and not sure). This scoring procedure resulted in a reliability
of .54 (standardized item alpha = .56).

Self-rated Parenting Skill. Global perceived parenting skill was assessed with one
item from the Parent Interview. This item asks: “Do you feel that you are: Not very good
at being a parent (1), A person who has some trouble being a parent (2), An average
parent (3), A better than average parent (4), or A very good parent (5)?” This item is
included in a national longitudinal study on the impact of the EHS program on child
development and farme health.

Child Abu_se Potentia_l. Abuse potential was operationalized as a mother’s score
on a measure of potential to physically abuse her child. Child abuse potential was
measured by the Child Abuse Potential Inventory (CAPI), a self-report questionnaire
designed to aid in the screening of suspected physical child abusers (Milner, Gold, &
Wimberley, 1986). The conceptualization of the scale is described in earlier studies
(Milner & Wimberley, 1979, 1980) and in a technical manual (Milner, 1989a). The full
questionnaire contains 160 items that are answered in a forced-choice, agree-disagree
format. The CAPI contains a physical child abuse scale (77 items) comprised of six

descriptive factor scales, including Distress, Rigidity, Unhappiness, Problems with Child

29

and Self, Problems with Family, and Problems from Others. Respondents are asked to
agree/disagree with statements relating to each subscale, such as “Everything in a home
should always be in its place” or “I have several close friends in my neighborhood.” A
total score is obtained by summing the items indicative of greater potential for physical
abuse; individual items are weighted according to results from multiple validation
studies. Elevated scores indicate that the examinee has characteristics similar to known,
active physical child abusers (Milner, 1989c).

Data on the classification rates of the CAPI abuse scale are available (Milner,
1989, 1991; Milner, Gold, & Wimberley, 1986; Milner & Robertson, 1989).
Collectively, the data indicate that overall abuse scale classification rates are in the mid-
80% to low-90% range. However, in most studies using equal numbers of matched
abusers and nonabusers, more false negative than false positive classifications have been
reported.

The CAPI abuse scale reliabilities are adequate across a variety of groups (Milner,
1986). For example, internal consistency (KR-20) reliabilities for the abuse scale range
from .92 to .96 for abusive (N = 152) and general population participants (N = 2,062).
For the current study, the 77-item CAPI had an alpha of .93, with a standardized item
alpha of .95. Alphas for the six subscales ranged from .91 to .44.6

Temporal stability estimates (test-retest) for l-day, 1-week, l-month, and 3-month
intervals for the abuse scale are .91, .90, .83, and .75, respectively. Predictive validity

data indicate a significant relationship between elevated abuse scores and later physical

 

6 Reliability analyses resulted in the following alphas for the CAPI subscales
(standardized item alpha values in parentheses): Distress .91 (.95), Rigidity .73 (.80),
Unhappiness .59 (.72), Problems with Child and Self .44 (.57), Problems with Family .49
(.59), Problems from Others .60 (.69).

30

child abuse (Milner, Gould, Ayoub, & J acevvitz, 1984). Many other validity studies have
been conducted (Ayoub, J acewitz, Gold & Milner, 1983; Burrell, Thompson & Sexton,
1992; Milner, 1989; Milner & Ayoub, 1980; Milner & Robertson, 1983, 1985; Monroe &
Schellenbach, 1989; Pruitt & Erickson, 1985; Robertson & Milner, 1985).

Given that associations among the variables of interest in the current study have
been clearly described in the child maltreatment literature, one would expect
correspondence among measures. Correlations among parenting stress, depression, and
the subscales of the CAPI (presented in the Appendix, Table 2), reveal some degree of
relationship while also indicating that these measures are not overlapping in content.
While tapping similar constructs, the measures also assess unique variance (see Holden,
Willis, & Foltz, 1989).

Results.
Data Estimation

Missing data points and number of valid cases are presented in the Appendix
(Table 2). For all scales except depression, missing data estimation was completed with
SPSS using regular linear regression with several correlated and theoretically related
predictor variables. If an item was missing, it was typically only for one or two items
across each of the instruments.

While changing the measurement instrument for depressive symptomology during
data collection was an unanticipated event, it is not likely that this had any meaningful
effect on the results. Analytic techniques were used to ameliorate any complications.
Most of mothers in the sample (88.4%) had complete BDI scores; 11.6% had completed

the CBS-D. For the 11 cases where BDI scores were missing (but CES-D scores were

31

available), these scores were estimated using percentile ranking. The Blom method of
percentile ranking was used to approximate the normal distribution for the BDI and CBS-
D scores. Values were assigned to cases based upon parallel percentile ranks between the
two instruments within the sample. This strategy maintains a continuous distribution and
removes the need to use cut-off scores, which can vary widely by sample and purpose.
Results of this method of missing data estimation were compared to the results based on
estimating data with linear regression, with more acceptable results from the percentile
ranking strategy.
Descriptive Analysis

Means and standard deviations were calculated for all scaled scores (Table 4). A
higher score indicates a greater degree of the variable of interest, i.e., more knowledge of
infant development, higher risk load, “better” self-rating of parenting skill, higher child
abuse potential, greater degree of depressive symptomology, or higher degree of

parenting stress.

32

 

 

 

 

 

 

 

 

 

Table 4: Descriptive Statistics of Scaled Scores
N Min Max M sd Skewness Kurtosis

std error = (std error =
.247) .490)

Risk load 95 1 16 6.07 3.19 1.133 .816

Depression 95 0 35 11.94 8.30 .680 .11?

Stress 95 24 87 45.49 15.58 .752 -.191

Knowledge 95 6 14 10.89 1.94 -.673 -.089

of Infant

Devel.

Self-rated 95 1 5 3.59 1.01 .070 -.844

Parenting

Skill

Child 95 2 422 152.88 113.71 .558 -.870

Abuse.

Potential

 

 

 

 

 

 

 

 

 

For this sample, a KIDI score was created as a sum of the items (reverse scoring
where appropriate), resulting in a sample mean of 10.89 (s_d = 1.94), with scores ranging
from 6 to 14. Preliminary analyses revealed that knowledge of infant development did
not contribute to the model in a meaningful way: this variable was not correlated with
any other scaled variables, nor did partial multiple regression analyses indicate any

significant relationship to other variables in the model. Knowledge of infant

33

development was significantly correlated with only one demographic variable, maternal
education (I = .23, p <.05). In addition, examination of 3-D scatterplots revealed no
relationship between knowledge of infant development and other variables. As an
additional step, the KIDI was rescored using the scoring protocol of the national EHS
evaluation (i.e., using a mean rather than a summed score. This strategy averages across
missing data points). With this strategy, this sample’s average knowledge score M =
4.06, _s_d = .46) can be compared to the national EHS sample (M = 3.05, sd = .41). While
this change in scoring protocol created a few new correlations among variables (i.e.,
knowledge of infant development became significantly correlated with a history of family
problems [r = .24, p < .05] and issues with illiteracy [r = -.27, p < .01]), knowledge of
infant development still did not correlate with any other predictor variables in the model.
The variable was therefore dropped from further analyses. Therefore, group differences
based on the interaction between self-rated parenting skill and knowledge of infant
development were also not examined.

The remainder of variables was tested for normality with the Kolmogorov-
Smirnov (Lilliefors) test. All distributions were normal. However, outliers were noted:
six cases were high-scoring extremes on risk load and two cases were high-scoring
extremes on depressive symptomology.

The average risk load score was 6.07 (sd = 3.19), indicating that families
experienced an average of slightly more than six risk factors. Families’ risk load ranged
from 1 to 16 factors. The most common individual risk factors—those affecting over 20%
of the sample—include low family income and unemployment; single- and teen-

parenthood; school drOpout and issues with illiteracy; rural housing; experiencing loss

related to death or divorce; physical, mental or emotional illness; history of diagnosed
family problems; and difficult or violent temperament. The summary of individual risk

factors is presented in Table 5.

 

 

Table 5: Summary of Individual Risk factors of the Risk load Valid %
Index

Low birth weight 16
Developmental immaturity 11. 7
Nutritional deficit 17
Language deficiency or immaturity 19.1
Physical and/or sexual abuse & neglect 16
Long term or chronic illness 9.6
Diagnosed disability 5.3
Difficult or violent temperament 20.2
Substance abuse or addiction 14.9
Issues with illiteracy 26.6
Single parent 53.2
Death/divorce 26.6
Teenage parent 40.4
Physical, mental, or emotional illness 25.5
Incarcerated parent 7.4
Low school achievement or drop out 45.7
History of delinquency 19.1
Unstable residence 9.6
Non-English or limited English speaking household 4.3
History of diagnosed family problems 22.3
High family density 13.8
Housing in rural or segregated area 30.9
Unemployment 58.5
Low family income 90.4

 

 

 

 

Mothers’ average depression score was 11.94 (sd = 8.3), ranging from 0 to 35. If

assigning a diagnosis, cut-off scores have been suggested.7 In reference to published

 

7 Mothers averaged 11.71 on the BDI, (s_d = 8.11); ranging from O to 35. Beck has
suggested multiple cut-offs: minimal (< 10), minimal to moderate (10 - 18), moderate to
severe (19 - 29), and severe (over 29). Others have suggested using a cut-off score of 10
with nonclinical samples. For the 11 mothers who completed the CES-D: M = 13.64 ($1
= 10.46), ranging from 1 to 31. This value can be compared with mothers participating in
the national EHS evaluation (N=2242), where the average CES-D score was 13.4 (_s__ =
9.9). With the CES-D and clinical samples, the literature suggests a cut-off score of 16,

35

norms, mothers in this sample scored slightly higher than average nonclinical samples in
depressive symptomology, but slightly lower than comparison groups (see footnote 6).
However, as noted previously, a continuous distribution of depressive symptomology was
the desired structure of the variable, and suggested cut-off scores are provided for
comparison only. Moreover, other studies using intensity of depressive symptoms have
found that nonspecific indicators of maternal psychopathology—rather than diagnostic
category—are better predictors of family firnctioning (Dickstein et al., 1998).

On self-rated parenting skill, mothers’ mean score was 3.59 (sd = 1.01), indicating
that most mothers rated their parenting skill as above average. Scores on this variable
ranged from 1 to 5.

Mothers averaged 45.49 (s_d = 15.58) for Parenting Stress, with scores ranging
from 24 - 87. While the national EHS sample’s PSI scores are divided by subscale (that
is, PSI distress M = 27.28, _s_c_1 = 9.45; PSI dysfunctional interaction M = 17.54, sd = 5.90),
a rough comparison value can be formed by adding the two subscale scores, resulting in a
mean of 44.82, close to the sample in the current study.

Mothers’ average CAPI score of 152.88 (_sp = 113.71; range = 2 — 422) can also
be compared to norms provided by the instrument’s author. Milner offers score
guidelines as a function of education, gender and ethnic background. For a female
sample with 12 or fewer years of education, the average score is 120 (_sp = 88), which
drops to 80 (sd = 64) with 13+ years of education. For samples of white females and

males, 12 or fewer years of education, the average score is 105 ($1, = 83), which drops to

 

whereas studies using nonclinical samples (teenagers) report that such a cut-off
dramatically increases the number of "depressed" cases.

36

72 (_351, = 58) with 13+ years of education. Cut-off scores are also suggested.8 Since
mothers’ average years of education was 11.78 (s_d = 1.64) and about 95% were identified
as white, both of these normed samples serve some utility as references, and both indicate
that this study’s sample averaged higher on child abuse potential.

Mothers’ scores on the CAPI subscales can also be compared to norms published
by Milner (1989b). On five of the six subscales embedded in the CAPI (i.e., Distress,
Unhappiness, Problems with Family, Problems with Others, Problems with Self and
Child), this sample scored significantly higher than the norm (as determined with one-
sarnple t-tests, using the normed indices as the specified constant)9. Rigidity is the only
subscale for which no significant differences emerged between the sample and the
normed index (Table 6). As a whole, this sample’s mean scores were not above the
suggested cut-off score for each CAPI subscale or the entire instrument. However,
examining which abuse factor scores are elevated may provide suggestions about which
dimensions in the scale are contributing to the total abuse score.

While only 3.2% of the sample scored above the cut-off value across all
subscales, 36.8% of mothers scored above the cut-off for Problems with Family. Over
28% of mothers scored above the cut-off value for Unhappiness, and about 27.4% of the

sample scored above the Distress cut-off score. About 19% scored above the cut-off

 

8 Cut-ofi' scores are suggested if using the CAPI as a diagnostic tool (i.e., 215 with
control groups, 166 with controls/abusers); however, this variable was left as a
continuous variable in the current study.

9 It is important to note the following demographic details about the comparison group
(N=836). This group was comprised of parents from parent-teacher organizations,
developmental evaluation centers, departments of social services, and other community
organizations. The average age was 32.3 ($1 = 8.1), the average education was 14 years
(s_d = 2.6), and 23% were males. Ethnicity included 13% black, 83% white and 4% other.
Averaging 2 children (m = 1), 8% of these parents were single, 76% were married, and
16% were divorced/separated/widowed.

37

values for both Problems with Child and Problems from Others, while only 8.4% scored

above the Rigidity subscale cut-off value.

 

Table 6: Child Abuse Potential Inventory Factor subscale scores and Comparison Norms

* denotes significant difference between sample factor scores and means published in
scoring manual (p < .05; ** p < .01; *** p < .001)

 

 

 

 

 

 

 

 

Factor Comparison Comparison Suggested M (N =95) _sg Range

M (N =836) M cut-off

scores10

Distress 58 56.8 152 94.44 *** 78.91 0 — 261
Rigidity 10.1 12.5 30 11.89 12.76 0 - 59
Unhappiness 8.1 9.2 23 17.54 *** 14.72 0 - 55
Problems 2.8 5.2 11 5.06 ** 6.88 O - 29
with self &
child
Problems 6 7 18 13.76 *** 14.108 0 - 38
with family
Problems 6.4 8.6 20 10.19 *** 8.32 0 - 24
with others

 

 

 

 

 

 

 

 

 

’0 From CAPI scoring manual: The factor scores should not be employed for
classification or predictive purposes. Normal range and elevated factor scores should only
be employed for descriptive purposes and for the formulation of tentative clinical
hypotheses about the examinee.

38

Correlations

In order to examine relationships among variables, Pearson product-moment
coefficients were calculated for all measures. For the most part, the variables’
correlations supported the hypothesized relationships. Significant positive relationships
were found among risk load, child abuse potential, depression and parenting stress. Self-
rated parenting skill Was significantly negatively related to abuse potential, depression
and parenting stress. Correlations among scaled scores are listed in Table 7.

Significant biserial correlations among individual risk factors were also found.
Both chronic illness and a history of abuse or neglect were related to a host of other
individual risk factors. Chronic illness, found in 9.5% of families in the sample, was
associated with language deficiency, developmental immaturity and nutritional deficit;
handicap; difficult temperament, delinquency and substance abuse; and physical or
mental illness. A history of abuse or neglect, notably high in this sample at 16%, was
significantly associated with unstable residency, difficult temperament, substance abuse,
school dropout, a history of delinquency and family problems, and a family member with
a physical or mental illness.

Unstable residency and difficult temperament also emerge as key factors.
Unstable residency, reported for 9.5% of families, was related to difficult temperament
and substance abuse, language deficiency, school dropout and delinquency, family
problems and death/divorce. Difficult temperament was associated with substance abuse,
language deficiency, delinquency and family problems, single parent, and physical,
mental, or emotional illness. Other correlations of interest, including those among

demographic variables, are presented in the Appendix (Table 8).

39

 

Table J7?Correlations among Scaled Scores

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Risk load Knowledge Self-rated Child Depression Parenting

of Infant Parenting Abuse Stress
Devel. Skill Potential

Risk load 1

Knowledge -.073 1

of Infant

Devel.

Self-rated -.017 .120 1

Parenting

Skill

Child 357*“ -.088 -.235* 1

Abuse

Potential

Depression 226* -093 -.330*** 740*“ 1

Parenting .221* -.148 -.352*** 662*“ 565*” 1

Stress

Analyses

The hypotheses that a) risk load, maternal affective variables and self-rated

parenting skill are associated with physical child abuse potential; b) self-rated parenting

skill mediated the relationships between parenting stress and child abuse potential, and

between depression and child abuse potential; and c) self-rated parenting skill moderated

the relationships between parenting stress and child abuse potential, and between

depression and child abuse potential were tested by comparing several structural equation

40

 

models. This strategy allows objective evaluation of the adequacy of fit of the theoretical
model to data, as indicated by the degree to which the matrix of interrelations among
observed variables constructed by the model differs from its sample complement .
(Raykov, Tomer, & Nesselroade, 1991). As only observed variables were examined,
these analyses were equivalent to path analysis.

Goodness-of-fit indices apply to both the overall model and individual specified
paths. The goodness-of-fit indices used to determine adequacy in these analyses include
GFI, CFI, RMSEA and Chi Square. Chi-square is a statistical test of lack of fit, with a
desired value of a nonsignficant result. The Goodness-of-Fit Index (GFI) indicates the
relative amount of observed variances and covariances accounted for by a model. The
Comparative Fit Index (CPI) indexes the relative reduction in lack of fit as estimated by
the noncentral Chi-square of a target model versus a baseline model. For both GF I and
CFI, values of greater than .90 are desired. The Root Mean Square Error of
Approximation (RMSEA) is a degree of parsimony fit index, with a desired value of
<05.

Amos 4 (Arbuckle & Wothke, 1999) was used to obtain generalized least squares
estimates of the model coefficients. A generalized least squares estimator minimizes the
sum of squared residuals, weighted by the inverse of the sample covariance matrix.
Several structural models were examined and compared for goodness of fit to the data.
As a first step, an overall inclusive model was run. This model included direct and
indirect effects for the following paths 1) parenting stress to self-rated parenting skill to

child abuse potential; 2) depression to self-rated parenting skill to child abuse potential;

41

3) parenting stress to child abuse potential; 4) depression to child abuse potential; and 5)
self-rated parenting skill to child abuse potential.

Results of these initial analyses suggested two modification indices. The first
modification index was to include a direct path from risk load to child abuse potential;
the second was to allow the errors between depression and parenting stress to correlate.
These suggested modifications make intuitive sense. Regarding the direct path from risk
load to child abuse potential: this added path is reasonable despite not being included in
the proposed hypotheses, given that it is a plausible derivative of risk load theory. The
correlation between errors for parenting stress and depression is also an acceptable
modification since the respondents and the assessment situation are the same, and the
systematic variation between parenting stress and depression is not fully explained by
risk load. Thus, these two suggested modifications were included in the remainder of the
analyses.

Using these two modifications to respecify the model, another firll model was run
(Figure 2). This model is described as Model 1, in which both indirect and direct effects
of parenting stress and depression onto child abuse potential were examined.

This model represents a good fit to the data based on multiple goodness-of-fit
indices. The overall 21’ is nonsignificant [11’ (df= 1, N = 95) = .759, 12 = .384], and the
GFI and CFI are acceptable values (.997 and 1.0, respectively). RMSEA (.000) is well
below cut—off range. The parsimony ratio (df/[.5k (k+1)]), which compares degrees of
freedom to number of nonredundant elements of the analyzed moment matrix, is .067 (k

= number of observed variables).

42

 

Figure 2: Model 1 (Full Model) X2 = .759 (P = 334)

; Depression
.23"
E . 16"

 

df= l
GFI = .997
CFI = 1.0

   
   
 

 

RMSEA = .000

 

 

 

 

 

 

 

 

Risk load \. 23:11::
self'mcd Potential

 

 

     

Parenting

0

 

 

 

 

.54
22*

Relationships in the model are all in the expected direction, and six of eight

Stress

 

 

 

 

 

 

standardized path coefficients were found significant (indicated with * in the figure).
Risk load directly affected both depression (standardized path coefficient = .23) and
parenting stress (standardized path coefficient = .22), as well as child abuse potential
(standardized path cOeffrcient = .16). In turn, parenting stress affected child abuse
potential (standardized path coefficient = .35) and self-rated parenting skill (standardized
path coefficient = -.24). Depression is also associated with child abuse potential
(standardized path coefficient = .53). However, the paths fi'om depression to self-rated
parenting skill, and from self-rated parenting skill to child abuse potential were not
significant.

These results indicate that both risk load and maternal affective variables of

depression and parenting stress impact child abuse potential. In addition, parenting stress

43

has a negative effect on mother’s rating of her parenting skill. However, self-rated
parenting skill does not serve as a mediator through which depression and parenting
stress lead to increased child abuse potential, and self-rated parenting skill itself does not
affect child abuse potential. Thus, once the variance between depression and abuse
potential and between parenting stress and abuse potential is accounted for, self-rated
parenting skill is not important as examined in this model.

To further examine the role of self-rated parenting skill, two additional models
were tested. Model 2 was run to ascertain the nature of the indirect paths from
depression and parenting stress to abuse potential via self-rated parenting skill. As can be
seen in Figure 3, the main impact of removing the direct paths from depression to abuse
potential and from parenting stress to abuse potential is a much larger standardized path
coefficient from risk load to abuse potential (from .16 in Model 1 to .35 in Model 2) and
a newly significant indirect relationship fiom the maternal affective variables to abuse
potential via self-rated parenting skill. These results indicate that risk load significantly
affects child abuse potential, as does self-rated parenting skill (standardized path
coefficient = -.23). While depression does not affect abuse potential through self-rated
parenting skill, parenting stress does (standardized path coefficient = -.24). The
parsimony ratio for this model is 0.2. However, the goodness-of-fit indices fail to
support the model. The overall 11’ is significant [21’ (df= 3, N = 95) = 85.02, p = 000)],
GFI (.809) is slightly low, CFI (.457) is very low, and RMSEA (.539) is far above the

desired cut-off.

 

 

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3: Model 2 x, = 85.02 (p =.000)
(Indirect Effects Model) (if: 3
GFI = .809
CFI = .457
Depression RMSEA = '539
' - (1D
.54 _,, ?
. 35* Child abuse
RlSk load Self:‘rated / Potential
Parenting .23‘
.22. Q) Skill
Parenting A:
Stress

 

 

 

 

 

 

Thus, an additional comparison model, in which only direct paths to child abuse
potential from risk load, depression, parenting stress, and self-rated parenting skill were
included, was examined (Figure 4). In this model (Model 3), there were significant paths
from risk load to abuse potential (standardized path coefficient = .16), from depression to
abuse potential (standardized path coefficient = .52), and from parenting stress to abuse
potential (standardized path coefficient = .34). However, the path from self-rated
parenting skill to abuse potential was not significant. While both GFI and CF I were of
respectable values (.941 and .915, respectively), Chi-square was significant [x2 (df = 3, N
= 95) = 15.91, p = .001] and RMSEA value (.214) overshoots the desired cut-off. This
model’s parsimony ratio is 0.2. These results indicate that risk load, depression and

parenting stress are significantly associated with child abuse potential, but that self-rated

45

parenting skill is not a useful element in understanding increased potential for abusive

parenting.

 

Figure 4: Model 3 x, = 15.91 (p = .001)
(Direct Effects Model) df= 3
GFI = .941

CFI = .915
RMSEA = .214

 

    

Depression

 

 

 

.52'
9s,
.16“
S lf (1 / Child abuse
1 e -rate Potential
.‘Parenting 06
®\

 

 

 

Riskload

 

 

 

 

 

 

 

 

 

/ .34“
Parenting

Stress

.22“

 

 

 

 

 

 

 

 

In Model 1, self-rated parenting skill emerged as a significant variable in the
model, despite not significantly affecting child abuse potential. In addition, the multiple
goodness-of-fit indices collectively suggested that Model 1 fit the data well. For these
reasons, Model 1 was selected as the best fitting model. Squared multiple correlations for
the variables are as follows: riskload (.00), depression (.05), parenting stress (.05), self-
rated parenting skill (.15), and child abuse potential (.67). All output pertaining to this
model is included in the Appendix.

Additional analyses

Bylcomparing differences among groups of mothers based on varying degrees of

child abuse potential, one can answer the question: What are the conditions—over and

above a set of generic risk factors for all family flmctioning problems—that characterize

46

a family context that is marked by high abuse potential? While maintaining family risk
load as a conceptual framework, it is possible to determine which individual risk factors
accounted for the most variance in this sample’s child abuse potential scores. It is
important to note that the set of individual risk factors accounting for the greatest
pr0portion of the variance would be expected to vary across samples and contexts.

In order to make such comparisons, K-means cluster analyses were conducted.
This procedure attempts to identify relatively homogeneous groups of cases based on
selected characteristics and determines group membership indicating the Euclidean
distance between each case and its classification center. Two series of K—means cluster
analysis were run, created based on mothers’ child abuse potential score.

In the first run, a three-cluster option was specified. This number of clusters was
specified based on preliminary examination of the distribution of child abuse potential
scores, in which three groups of score ranges appeared to emerge. However, after the
maximum number of iterations (i.e., 10), a solution did not converge. In the next run, a
four-cluster option was specified. After 5 iterations, a solution was reached. The final
cluster centers were 381 (Cluster 1), 156 (Cluster 2), 287 (Cluster 3) and 46 (Cluster 4).

Descriptive data on these clusters are listed in Table 9.

 

 

 

 

 

 

 

 

Table 9: Descriptive Data on Child Abuse Potential Clusters
Overall Cluster 1 Cluster 2 Cluster 3 Cluster 4
sample (highest (average (heightened (low risk)
CAPI risk) risk) risk)

N 95 6 28 21 40

.M 152.88 380.50 155.89 287.14 46.15

pd 113.71 30.12 31.71 28.26 23.95

Min. value 2 345 107 235 2

Max. value 422 422 21 8 328 87

 

 

 

 

 

 

 

47

The highest-scoring group of mothers (Cluster 1) contains only six cases, with an
average abuse potential score that places them close to two standard deviations above the
mean for the entire sample. The next highest-scoring group of mothers (Cluster 3) falls
roughly between one and two standard deviations above the sample mean. Cluster 2
holds 28 mothers essentially of “average risk” for abuse potential (relative to this
sample). Finally, Cluster 4 is comprised of 40 mothers at the lowest risk for abuse.
Using means published by Milner (1989a),ll it is clear that even Cluster 2 mothers—
those of “average risk” in this sample—score well above the means for these comparison
samples. In contrast, the Cluster 4 mothers fall far below comparison groups’ mean
abuse potential scores.

Clusters were compared across a range of variables (see descriptive data in Table
9). One-way AN OVAs with Student-Newman-Keuls post hoc comparisons were
conducted; only significant differences are described in the paragraphs that follow.

Clusters significantly differed relative to risk load (p = .005), depression (p =
.000) and parenting stress (p = .000), as well as individual risk factors of nutritional
deficit (p = .006), unstable residency (p = .045), single parenthood (p = .007) and teen
parenthood (p = .007). In addition, a trend is noted for chronic illness (p = .052). Each
of six CAPI subscales also emerged as significantly differentiating among clusters, each
at p = .000.

Post hoc tests revealed the following difi‘erences. Clusters 2 and 4 experienced
significantly lower risk loads than did Cluster 1 (Cluster 3 could fit in either subset).

Each cluster demonstrated significant differences in terms of depression, with Cluster 1

 

11 That is, a mean of 120 for females with less than 12 years education and 105 for white
males and females with 12 or fewer years education.

48

rated highest and Cluster 4 lowest on depressive symptomology. In terms of parenting
stress, three subsets emerge, with both Clusters 3 and 1 falling into a subset of high-stress
parents.

While family risk load was clearly associated with child abuse potential scores,
several individual risk factors were also useful in distinguishing among clusters. For
example, Cluster 1 has a significantly higher report of nutritional deficit. Teenage
parenthood also emerged as an important variable in terms of increased child abuse
potential. Mothers in Cluster 1 were significantly more likely to be teen mothers, relative
to mothers in Clusters 4 and 2 (Cluster 3 belonged in either subset). While the AN OVA
analyses revealed significant differences for single parenthood and unstable residence,
and a trend for chronic illness, post hoc comparisons did not identify subsets based on
these individual risk factors.

In sum, several of the proposed hypotheses were supported, viz., hypotheses 1 and
2, that higher risk load is associated with greater depressive symptomology and increased
parenting stress. Hypotheses 3 and 4 were also supported by the data: greater depressive
symptomology and increased parenting stress were both associated with higher child
abuse potential. Hypotheses 9 and 11, pertaining to moderating and mediating roles for
self-rated parenting skill in the relationship between depression and abuse potential, were
not supported. Self-rated parenting skill did not interact meaningfully with depression.

Hypotheses 13 and 14, pertaining to moderating and mediating roles for self-rated
parenting skill in the relationship between parenting stress and child abuse potential, were
partially supported. Self-rated parenting skill served as a mediator between parenting

stress and child abuse potential, but only when direct effects of parenting stress onto

49

abuse potential were controlled. In the best-fitting model, higher parenting stress was
associated with poorer self-rated parenting skill, but self-rated parenting skill does not
then serve as a mediator between parenting stress and abuse potential. In addition,
increased risk load was associated with higher child abuse potential, a finding not
originally proposed. Hypotheses that included knowledge of infant development, singly
or as part of an interaction variable, were not tested (i.e., Hypotheses 5, 6, 7, 8, 10, 12, 15
and 16).

A final analytic procedure was conducted to assess which individual risk factors
comprising the risk load index were contributing most to mother’s abuse potential scores.
A series of one-way AN OVAs was conducted relating each risk factor to child abuse
potential (Table 10). For these analyses, the sample was treated as a whole, without
division into clusters based on child abuse potential scores. Only the significantly
contributing risk factors are described, in order of importance. A significant effect was
found for teen parent status [F (l, 92) = 13.72, p = .000], difficult temperament [F (l, 92)
= 7.06, p = .009], single parent status [F (1, 92) = 6.61, p = .012], unstable residency [F
(1, 92) = 4.97, p = .028], and history of physical/emotional abuse or neglect [F (1, 92) =

3.80, p = .054].

50

 

Table 10: One-way AN OVAs individual risk factors and child abuse potential scores

 

Table 10a: Descriptives

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

95% confidence
interval
N M _s_d_ Std Lower Upper Min Ma
error bound bound x
Abuse/ no 79 141.4 107.8 12.1 117.3 165.6 2 397
neglect yes 15 202.6 129.9 33.5 130.7 274.5 9 422
Unstable no 85 142.9 108.2 11.7 119.6 166.2 2 422
residence yes 9 229.4 134.4 44.8 126.1 332.8 9 399
Difficult no 75 136.1 107.3 12.4 111.4 160.7 2 422
temperament yes 19 210.8 118.7 27.2 153.7 268.0 24 399
Single no 44 120.1 99.5 15.0 89.9 150.4 9 422
parenthood yes 50 178.5 118.1 16.7 144.9 212.1 2 399
Teen no 56 117.8 93.5 12.5 92.8 142.8 2 345
parenthood yes 38 200.4 122.4 19.9 160.1 240.6 10 422
Table 10b: ANOVAs Sum of df Mean F Sig.
Squares Square
Abuse/ B/w Grps 47189.1 1 47189.1 3.80] .054
neglect W/in Grps 1 1422598 92 12415.9
Total 1189447.9 93
Unstable B/w Grps 60963.7 1 60963.7 4.970 .028*
residence W/in Grps 11284843 92 12266.1
Total 1189447.9 93
Difficult B/w Grps 84762.7 1 847E7 7.059 009*
temperament W/in Grps 11046852 92 12007.4
Total 1189447.9 93 -
Single B/w Grps 79722.2 1 79722.2 6.609 .012*
parenthood W/in Grps 11097257 92 12062.2
Total 1189447.9 93
Teen B/w Grps 1543242 1 1543242 13.716 .000*
parenthood W/in Grps 1035123.7 92 1 1251.3
Total 1189447.9 93

 

 

 

 

 

 

 

When child abuse potential was regressed onto these five risk factors as a group,

51

about 21% (p = .001) of the variance is accounted for. However, of all the variables in

 

 

the equation, only teenage parenting remained significant (beta = .291 , p = .004) (Table

11).

 

Table 11: Model 1
Predictors = Abuse/neglect, Single parenthood, Teenage parent, Difficult temperament,

 

 

 

 

 

 

 

 

 

 

 

 

 

Unstable residence.
Dependent Variable = Child abuse potential
Sums of (If Mean F Sig.
Squares Square
Regression 2486016 5 497203 4.650 .001
Residual 9408463 88 10691 .4
Total 1 189447.9 93
Unstandardized Standardized L
Coefficients Coefficients

Model B Std error beta t Sig.
Constant 93.070 17.306 5.378 .000
Teen 66.675 22.474 .291 2.967 .004*
parenting
Difficult 36.404 30.427 .130 1.196 .235
temperament
Single 34.551 22.425 .153 1.541 .127
parenthood '
Unstable ‘ 38.564 48.522 .101 .795 .429
residence
Abuse/ 10.829 39.938 .035 .271 .787
neglect

Discussion

The goal of this study was to elucidate the nature of physical child abuse potential
in a high risk context by focusing on the associations among risk load, maternal affective
variables of depression and parenting stress, and self-rated parenting skill. Results
revealed that there is a relationship between risk load and a mother’s child abuse potential
that is both direct and indirect by means of increased parenting stress and depression. In
accord with Belsky’s (1984) theory and findings by Simons and colleagues (Simons &
Johnson, 1996; Simons et al., 1993; Sirnons etal., 1994), the impact of chronically

stressful contextual conditions (i.e., risk load) is less consequential for parenting than is

52

 

mother’s well-being. Thus, increased risk for physically abusive parenting can be viewed
as being comprised of both parent-level variables and environmental variables that are
external to the parent. These results harmonize with recent theories that view child
maltreatment as a product of parent characteristics that exist within the context of a multi-
problem family (Black, Heyman, & Smith Slep, in press).

When the direct effects of depression and parenting stress onto abuse potential are
not considered, a mother’s self-rated parenting skill-—negatively impacted by her
increased parenting stress—also influences her abuse potential. However, a clearer
picture of relationships among these variables emerges when direct effects of these
maternal affective variables onto abuse potential are accounted for. In the optimized
model, standardized regression weights for the paths from risk load to depression, and
from risk load to parenting stress, are essentially of equal value, suggesting that risk load
accounts for the same portion of variance in each of these affective variables. In
addition, depression and parenting stress are positively correlated. Higher risk load is
associated with increased depression and increased parenting stress. These affective
variables are then associated with increased potential to abuse. Thus, the impact of
heightened risk load onto a mother’s abuse potential depends in part on her psychological
well being. '

Maternal depression accounts for a large portion of the variance in child abuse
potential, a relationship that has been well established in the child maltreatment literature.
However, depression does not appear to affect a mother’s rating of her own parenting
skill, and as such, does not affect child abuse potential indirectly via self-rated skill.

Parenting stress also maintains a direct relationship with child abuse potential: higher

53

parenting stress is associated with increased abuse potential. In addition, higher
parenting stress is associated with lower ratings of parenting skill. This lowered sense of
parenting skill, however, does not then affect mother’s risk for child abuse, in contrast to
evidence provided by other studies (McClennan & Harkless, 1998).

Overall, the model that best fit the data indicates that heightened risk load,
depression and parenting stress operate in a cumulative fashion to influence abuse
potential. A mother who experiences increased risk and heightened depression and
parenting stress is at greater risk to abuse her child. Relationships among such variables
depict a parenting environment that is characterized by heightened chronic stress, a
connection described in the maltreatment literature (Egeland & Brunnquell, 1979; Gaudin
& Pollane, 1983; Justice & Duncan, 1976).

A common feature of parents prone to abuse may be from where this increased
stress originates. Whipple and Webster-Stratton (1991) found that abusive parents
experienced stress more often from low income, maternal depression and anxiety,
negative life events, frequent child behavior problems and less marital support. In the
current study, many of the same sources of stress were found: higher parenting stress
levels were significantly associated with higher risk load and depression, lower self-rated
parenting skill and increased child abuse potential. Higher parenting stress was also
significantly correlated with low family income, substantiating many studies’ findings
that the experience of stress is much greater in poor or lower-class families than for
middle-class families. In addition, developmental immaturity and teen parenthood were

significantly correlated with parenting stress. Unlike results from other studies (Deater-

54

Deckard & Scarr, 1996), child age was not a correlate of parenting stress for the mothers
in this sample.

Comparable relationships among similar constructs were found by Conger et a1“.
(1984), who noted that chronic environmental stress (as measured by financial strain,
family structure, and maternal education and age) accounted for 54% of the variance in
mothers’ psychological characteristics (i.e., emotional distress, child-rearing values and
perception of children). These characteristics then accounted for 15.1% of the variance
in maternal affective responses to children. However, chronic environmental stress
accounted for a larger portion of the variance in maternal behavior (36.6%) than did
psychological characteristics. The most consistent demographic predictors of maternal
behavior in that study were mothers’ age and education. However, the Conger et al.
study focused on maternal affective responses to children, not abuse potential per se.

While not using risk load as a theoretical framework, Brown and colleagues
(1998) discovered that the prevalence of child abuse or neglect increased from 3% when
no risk factors were present to 24% when four or more risk factors were present.
Similarly, Harman and Luster (1991) created a family risk index afier determining which
risk factors significantly affected the quality of the home environment. There was a clear
relationship to the six-factor risk index (comprised of teen parenthood, lower intelligence
scores, lower income, single parenthood, family density and difficult temperament), such
that as a mother’s score on the risk index increased, the probability that her family
provided a lower quality home environment group also increased. In the current study,
risk load correlated with several demographic variables: it was positively correlated with

child age and receiving public assistance, while negatively associated with the number of

55

people in the household and having a live-in spouse. Thus, risk may be buffered if
mother is experiencing increased support from a husband or additional household
members.

A greater understanding of child abuse potential—and the variables most closely
associated with it——-—may be found by examining relationships among individual risk
factors and demographic characteristics in the current study, although it is recognized that
demographic factors, which may be risk indicators, are not necessarily causal factors
(Black, Heyman, & Smith Slep, in press). Many of the most prevalent individual risk
factors were—unsurprisingly—related to social class. This finding is expected given the
nature of the program from which families were selected, and the fact that the average
yearly income was below $10,000. Some of these risk factors revealed notably high
prevalence rates.

“Common” risk factors of low income and single parenthood notwithstanding,
this sample had a high pr0portion of difficult or violent temperament and history of abuse
or neglect. Although it is difficult to find adequate comparison values, the US. 1998 rate
of maltreatment was 12.9 per 1,000 children12 UDHHS, 1998). With 16% of mothers in
the current sample reporting a history of some form of maltreatment, it is worth
considering that victimization (especially by mothers) increases mothers’ risk for
perpetrating physical abuse (Black, Heyman, & Smith Slep, in press). Others (Haskett,
Johnson, & Miller, 1994) have found that a history of maltreatment was related to self-

reported problems with family members but not to mothers’ overall risk of abuse. It is

 

’2 53.5% of these are cases of neglect, 22.7% cases of physical abuse, 11.5% cases of
sexual abuse, and 6% or fewer cases each of emotional abuse and medical neglect. In
addition, 25.3% were reported to be victims of more than one type of maltreatment

56

thus noteworthy that 36.8% of mothers in this sample scored above the suggested cut-off
value for a child abuse potential factor that indicates problems with family members.

The prevalence of difficult or violent temperament in this sample is also notable
at over 20%. This variable played an important role in elevated child abuse potential, as
did unstable residency, low educational attainment and issues with illiteracy. These
variables were then associated with a host of other risk factors (e.g., history of
abuse/neglect is associated with depression and substance abuse; difficult temperament
and unstable residency are associated with delinquency and diagnosed family problems;
unstable residency is also related to receipt of public assistance).

It is also informative to examine demographic variables that do not share an
association to physical abuse potential. Of all the demographic variables examined in
this study—matemal age and education, child age and sex, ethnicity, income and receipt
of TANF and Medicaid, household density, and housing and transportation needs—only
younger maternal age was significantly correlated with abuse potential, as previously
noted. Despite the fact that the demographic variables examined here are among the
most commonly studied potential risk factors for parent-to-child physical aggression,
other studies (see Black, Heyman, & Smith Slep, in press) have found conflicting
evidence regarding their role in child maltreatment etiology.

Depression is known to covary with other factors that adversely affect family
functioning. In this sample, higher depression was significantly associated with several
individual risk factors, including history of physical or emotional abuse/neglect, single

parenthood and teen parenthood. Lower depression was associated with older child age.

 

(DHHS 1998': Reports from the States to the National Child Abuse and Neglect Data
System).

57

Interestingly, increased depressive symptomology was not associated with low income or
other related variables, such as low educational attainment and receiving public
assistance. It appears that the individual factors associated with increased depression in
this sample—such as single— and teenage-parenthood—operate through increased stress.

Indeed, both heightened child abuse potential and parenting stress were associated
with maternal age variables (viz., teen parenthood status as well as age in years).
Although mothers across the entire sample averaged slightly over 23 years old, about
23% of the sample were younger than 19 years old, with a mean age of 17.6 years (s_d =
1.3). However, teen parenthood was indicated in 40% of cases, indicating that some
proportion of mothers in this sample have other children born at an earlier time (i.e.,
when they were teenagers). These young mothers account for a greater proportion of
high depression-high parenting stress cases. In addition, higher maternal age is related to
higher education, income and rent, more people and children in the household, and
having a live-in spouse. These findings mirror those of other studies that link early
parenthood with attenuated educational achievement and related problems of
underemployment and restricted life—time earnings (Fulton, Murphy & Anderson, 1991;
Luster & Brophy-Herb, 2000; Miller, Miceli, Whitman, & Borkowski, 1996).13

Thus, for this sample of mothers, younger parenting age is part of a high risk
context that may present a greater likelihood for physically abusing one’s children.
Indeed, a range of problematic parenting behaviors has been found in samples of

adolescent mothers, including deficiencies in maternal sensitivity, contingent

 

'3 However, others report that early childbearing is associated with larger family size (see
Luster & Brophy-Herb, in press). In the current sample, it is not know whether more
people and children in the household refers to cases where mother lives with the family
of origin.

58

responsivity, physical involvement, and verbal stimulation (Miller et al., 1996; Ragozin,
Basham, Crnic, Greenberg & Robinson, 1982), and noteworthy similarities between
characteristics of the abusive parent and those of the adolescent parent have been noted
(see Fitzgerald, Strommen, & McKinney, 1982). Several other studies have
demonstrated that teen mothers’ lower quality child rearing environment is related in part
to the accompanying stressors of lower education, income, and socio-emotional maturity,
as well as reduced likelihood of paired parenting (Luster & Dubow, 1990; Luster &
Rhoades, 1989; Miller et al., 1996). Others (Goerge & Lee, 1997; Haskett, Johnson, &
Miller, 1994; Passino et al., 1993) have reported that younger teen mothers were at
greater risk of abuse, advocated the use of physical punishment more strongly, and
reported greater unhappiness and more parenting stress than older adolescents. However,
Zuravin and DiBlasio (1996) found that teen mothers’ low educational attainment was
related to neglect, but not to abuse. It is also important to note that some researchers
believe that the link between teen parents (and maternal age in general) and child
maltreatment is tentative (see Black, Heyman, & Smith Slep, in press; Miller, 1984). Yet
for this sample, teenage parenting emerged as the most notable contributor to the
relationship between risk load and child abuse potential.

The importance of teenage parenting as a risk factor for child abuse is illustrated
by comparisons of group membership based on child abuse potential. The mothers at
highest risk for abuse presented a significantly higher risk load than did other mothers.
Significantly more likely to be teen parents, these mothers were more depressed and
experienced more parenting stress than other mothers. Children of mothers in the highest

abuse potential group were more likely to have some sort of nutritional deficit. A

59

connection among teen mothers, maltreatment, and inadequate child nutrition has been
noted (Miller, 1984). On the variables that comprise abuse potential—distress, rigidity,
unhappiness, and problems with child, family and others—-—these high risk mothers
outscored every other cluster of mothers. 1" These results build upon and expand other
studies (McCurdy, 1995) that have described mothers with high potential for child abuse.

Families’ average risk load of six risk factors seems parallel to the results of other
studies that have used such an index to characterize the degree of risk in families. For
example, Sameroff et al. (1998) utilized an index with 13 factors and found a mode of 5;
other Sameroff studies defined high risk as 4+ risk factors. Rutter (1979) examined
changes in risk for children’s psychiatric illness with fewer than four factors. Williams et
al. (1990) examined increased likelihood of child behavior problems for children with 8+
risk factors. However, Ackerman and colleagues (1999) found a mean of only 3.3 risk
indicators (of an index of 11) per family based on a sample of Head Start participants. It
is important to note that the measurement of risk load is particular to each study and
cross-study comparison is difficult at best.

Relative to other studies, the current study’s measurement of risk load seems
more thorough and broad. Assignment of whether a particular risk factor was present in
the family context could possibly incorporate both a mother’s developmental history as
well as her life choices. To illustrate, loss due to death/divorce could indicate that a
mother’s parents had divorced, or that she herself had been divorced. Similarly, low
school achievement could refer to a mother, her husband, or both. However, the

theoretical underpinnings remain the same across studies, and provide additional support

 

’4 For rigidity, Cluster 3 mothers converge with Cluster 4 mothers. For distress,
unhappiness, and problems with child, family and others, Cluster 4 mothers are distinct

60

to Belsky’s (1984) speculation that contextual sources of stress not only affect parenting
behavior directly but also indirectly as well. Thus, the concept of risk load, previously
used primarily with child-related outcomes such as IQ, competence and mental health,
also serves as a usefirl framework for examining child abuse potential in high risk
families. It is therefore possible to utilize a contextual risk index—rather than any
individual risk factor—as in indicator of the child rearing context and maternal potential
for physical child abuse. However, examination of the individual risk factors that
comprise the risk load index reveals some of the multiple and varied aspects of
environmental adversity that families in early intervention programs may experience.

For this particular sample, the more important individual risk factors relative to
abuse potential were teen parenting, difficult temperament, single parenthood, unstable
residency and history of abuse or neglect, with teen parenthood emerging as the most
significant contributor. Relationships among these factors are described in the literature
addressing some of the implications associated with teen parenting (see Luster & Brophy-
Herb, 2000). This particular combination of risk factors contributed the most to maternal
child abuse potential.

Several other findings are worth exploration. In the current study, depression and
parenting stress were positively correlated . Both mothers’ depression and parenting
stress were directly and significantly associated with child abuse potential, albeit not
through self-rated parenting skill, as predicted. In addition, a mother’s feelings of
depression did not significantly alter her perception of her parenting skill. This finding is
in opposition to those found by Cummings and Davies (1994) and Frankel and Harmon

(1996) that depressed mothers have relatively little confidence in their caregiving abilities

 

from the other three groups of mothers.

61

and low levels of parental self-efficacy, which, in turn, have been associated with
parenting impairments. Unlike depression, parenting stress does significantly affect a
mother’s rating of her parenting skills. However, self-rated parenting skill does not then
affect child abuse potential.

The fact that the path fiom self-rating parenting skill to physical child abuse
potential was not significant is interesting to note, especially as it is incongruous with
findings from other studies documenting the relationship between a low sense of
parenting efficacy and coercive discipline (Bugenthal, Blue, & Cruzcosa, 1989a;
Bugenthal & Cortez, 1988; Mash et al., 1983). This finding seems to indicate a schism
between a mother’s subjective experience of self in the parenting role and her potential
behavior. However, while self-rated parenting skill did not affect child abuse potential, it
was significantly negatively correlated with abuse potential, similar to other studies
linking maternal confidence and quality of mother-child interactions (Bohlin & Hagekull,
1987; Gelfand & Teti, 1991).

One explanation for these findings may relate to the position that self-rated
parenting skill holds in the model. That is, some (Esdaile & Greenwood, 1995) have
found that mothers’ stress levels are predicted by their low self-concept as educators of
their children. Perhaps switching the places of stress and self-rated parenting skill (i.e.,
with parenting stress as the mediator through which self-rated parenting skill impacts
child abuse potential) in the conceptual model would provide a meaningful exploration of
these variables regarding abuse etiology. It is also likely that the manner in which self-
rated parenting skill was measured—as a single-item indicator—may explain the

variable’s lack of influence in the study. Given that the sample’s mean score on this

62

variable indicated that most mothers rated their skill as above those of “the average
parent,” these mothers as a whole may have a heightened sense of parental well being,
without sufficient variability to add meaningfully to the model.

Similarly, other studies (Benasich & Brooks-Gunn, 1996; Egeland & Brunnquell,
1979; Mercer & Ferketich, 1994; Shaner, Peterson & Roscoe, 1985; Stevens, 1984) have
found knowledge of children’s developmental milestones to be an important contributor
to child abuse etiology, parental competence and quality of home environment, although
this was not the case in the current study. Scores on this variable demonstrated notable
relationships only with maternal education variables (including a significant positive
correlation with years in school and significant negative correlations with school dropout
and. problems with literacy). The association between education variables and knowledge
of child development is bolstered by studies linking adolescence, social-emotional
maturity and knowledge of child deve10pment to attitudes toward parenting (Larsen &
Juhasz, 1985), quality of home environment (Benasich & Brooks-Gunn, 1996; Harman &
Luster, 1991), and child abuse potential (Dukewich, Borkowski, & Whitman, 1997).15
This variable did not contribute in any meaningful way to the model(s) examined in the
current study. Given the role that knowledge of developmental milestones has shown in
the literature, this lack of importance in the model is most likely explained by the
instrument with which knowledge of infant development was measured (i.e., low

reliability for this sample).

 

’5 Dukewich et al. (1997) found that preparation for parenting, a construct that included
knowledge and attitudes about children’s development, was the strongest predictor of
abuse potential; however, its effects were partially mediated by the mother’s
psychological predisposition for aggressive coping.

63

In conclusion, results of this study help to explain a mother’s potential to
physically abuse based in part on presentations of contextual diversity and affective risk.
Robust links among risk load, depression and parenting stress are related to mothers who
are more likely to endorse attitudes and characteristics that past research has linked to
physical child abuse. Therefore, assessing a family’s risk load—as a broad construct
incorporating multiple individual risk factors—as well as a mother’s levels of depression
and parenting stress, could significantly increase detection of abuse risk. Moreover,
professionals may find interventions more effective if they incorporate the notion of
parental sense of competence as it relates to these other important variables.

An important cautionary note when considering these findings is that child abuse
potential does not translate directly into behavior. For example, Kolko and colleagues
(1993) found that mothers scoring high on abuse potential reported greater child
externalizing and depressive symptoms, child rejection, personal psychological
dysfirnction, stressful life events, and family problems, but that these mothers did not
differ from moderate- and low-abuse potential mothers in terms of parent management
practices or family violence. In study of parents at risk for problems in parenting, Milner,
Gold, Ayoub and Jacewitz (1984) reported that although all abusive parents earned scores
above the CAPI cutoff score for abuse, the majority of subjects earning elevated scores
did not abuse their children.

As a construct, child abuse potential refers to a particular constellation of factors
that is strongly correlated with physical child abuse (viz., intra- and interpersonal
difficulties, inflexible views about children’s behavior, and attitudes about child-rearing

and discipline). These factors were assessed in the current study within the framework of

high family risk load and maternal distress and were not offered as a means with which to
predict parental behavior. As noted by Wolfe (1987) and illustrated by-findings from the
current study, results from the CAPI may serve as an indicator of general functioning
since the measure is so highly correlated with measures of emotional well-being.

In addition, previous research has demonstrated that labeling behaviors or people
as “at ris ” may have harmful efieets on the individuals or families (Palmer, 1983), and
using the results of this study to determine that a particular mother is likely to harm a
child would not be constructive. When used to improve intervention and prevention
programs by directing content after identifying participant needs, however, risk
assessment tools can provide meaningfirl information (McCurdy, 1995). It is strongly
encouraged that a multimethod approach to the assessment of risk be taken so that
corroborating information may be gathered in the process of decision-making regarding
risk (Haskett, Smith Scott, & F arm, 1995).

Relative to these findings and their association with child abuse etiology, what are
the implications? Given the preponderance of evidence that positive child outcomes are
associated with maternal sensitivity, attentiveness, stimulation, responsivity and
nonrestrietiveness (Belsky, Lerner, & Spanier, 1984; Harman & Luster, 1991), examining
child abuse potential and its contributing variables has important implications for early
intervention programs. Due to the variety of risk factors that comprise risk load, it would
seem that a multifaceted and comprehensive approach is necessary (Christmas, Wodarski,
& Smokowski, 1996). Interpersonal psychotherapy, cognitive/behavioral therapy, and
cognitive therapy could be utilized for treatment of depression. Sex education is related

to child maltreatment prevention through its impact on delayed pregnancy; and

65

employment training and referrals would help decrease the impact of low SES on child
maltreatment. Parent—training services, including stress- and anger-management, would
benefit parents as well. Reviews of intervention studies report that an individualized
approach is most successful in promoting maternal adjustment and childrearing skills,
especially with parents considered to be at greater risk of maltreatment (W ekerle &
Wolfe, 1993). Identifying a parent’s risk load, and which individual risk factors are most
salient to a particular family, may aid in creating individualized treatment plans.

Given increasing social and scientific interest in early intervention programs like
Early Head Start, research findings regarding the challenges faced by the families served
by these programs are likely to receive much-warranted attention. This growing social
and scientific interest in early intervention programs translates to a large degree into
attention to child. outcomes, such as physical and. cognitive development. The findings 1
from the current study emphasize the need to attend to a family’s contextual and parental
characteristics for their role in creating the child-rearing environment. These variables
are highly likely to impact the child’s experience and thus, child development outcomes.
As such, it is essential that early intervention programs like EHS collect data beyond
programmatic activities like treatment plans and program dosage. Knowmg the
implications of a family’s risk load and the affective/emotional state of the mother on
child development may be an imperative component of drawing conclusions about
program effectiveness. Results of the current study are a first step in that direction.

Strengths and Limitations
One limitation of this study is that the timing of data collection varied for

participants in two related manners: across child age and across program duration. For

66

collecting the majority of data (i.e., depression and parenting stress, child abuse potential,
knowledge of child development and self-rated parenting skill), a minimum child age of
nine months was established. However, the program served families with children
between prenatal stages and 36 months of age. Thus, data were collected across ages
where important differences in the variables of interest may have taken place. In
addition, families were accepted into the program across various child ages and therefore
may have received some amount of intervention (i.e., “program dosage”) that differed
depending on duration of treatment. It is possible that these differences in length of
participation affected the variables of interest.

A second limitation relates to the generalizability of results. Since the
participants were families that had been accepted into the Early Head Start program, they
are by definition at increased risk for a range of negative outcomes. Therefore the results
of this study may not be generalizable to the general population. In addition, the
economic and ethnic backgrounds of the families in this sample is not representative of
the general population, and thus hypotheses need to be tested within a more culturally
and economically diverse sample. Moreover, the lack of fathers in this sample is
unfortunate. Comparing the findings of the current study of mothers to fathers may
provide meaningful information regarding potential gender differences and similarities,
and many have begun to call for greater attention to the male in social/psychological
research.

Research by Mash and Johnston (1990) on parenting efficacy has suggested that
the relative influence of this parental cognition changes with the situation. In the current

study, self-rated parenting skill was assessed as a global indicator and at only one time

67

point, not related to a given family context or situation. It may be helpfirl to assess
context salience when examining the role of perceived parenting skill. In addition, this
variable was assessed via a single item. There are obvious methodological weaknesses
associated with single-item measures.

The variables of interest in this study are highly related. Past research has shown
that child abuse, stress and depression can be closely linked phenomena This holds true
in the current study as well, as indicated by correlations among variables as well as the
modification index suggested by analyses (i.e., to allow errors between parenting stress
and depression to correlate). In particular, the measures for depressive symptomology
and child abuse potential most likely share particular elements of assessment, in that the
CAPI contains an “unhappiness” subscale. In addition, the PSI and CAPI share variance
in that parenting stress is highly related to abuse potential. In some cases, a factor
analytic procedure to determine degree of overlap may be an appropriate strategy.
However, there are several reasons why a factor analysis was not completed in this case.

Most important, factor analytic results can be very sample specific, and good
analyses require a large sample to get a stable correlation matrix. This sample of 95
cases is insufficient for this criterion (see MacCallum & Widaman, 1999, and Merenda,
1997). Secondly, the measures used in this study are validated scales with already-
defined conceptual structures. As such, any alteration of the instrument would be vastly
less defensible than keeping them intact and speculating about changes in validity due to
removal of subscales.

While possibly assessing some common element of affect, the BDI, PSI and CAPI

are not measuring the same things. Holden, Willis and Foltz (1989) compared the CAPI

68

and PSI with a sample of confirmed maltreating parents (abusive and neglectful parents)
and found that the correlation between the measures was .34 (p < .001) to .59 (p < .0001)
for the CAPI and 13 of PSI’s 16 subscales. The results suggest that both measures may
be more sensitive to abuse and high risk situations. Thus, it may be that this relationship
is inflated in the current study due to restricted variability in important sample
characteristics. Indeed, the correlation between parenting stress and abuse potential in
this sample is higher than in other studies utilizing these instruments (Rodriguez &
Green, 1997). Other studies (see Cummings & Davies, 1994) have also found
pronounced interrelations among risk factors within high risk families.

These measures also share method variance, in that they are paper-and-pencil self-
report. There is also the issue of reactivity: since the subjects completed these measures
at the same time in the same way, the likelihood of a relationship is increased, regardless
of the “true” relationship in the population. Related to this limitation is the possibility
that intercorrelations may reflect response sets like social desirability. As such, data are
as valid as the veracity of the informant. It is recommended that cross-validating studies
utilize multiple methods of data collection.

In sum, despite relationships and shared variance across the variables assessed in
the current study, each of the measures used is a valid indicator of the constructs it
purports to measure. Overall, it seems appropriate to conclude that the meaSures tap
similar constructs and have some overlap but are accounting for unique variance.

Another methodological limitation to note is related to the small sample size. An
N of 95 may test the limits of structural equation modeling, due to the sensitivity of

covariances, parameter estimates and x2 tests of fitness to sample size. Hoyle and Panter

69

(1995) warn against using post-hoe modifications, especially with small samples, citing
evidence that such modifications are likely to replicate when sample size is at least 800.
Obviously a sample this large is rare in social and behavioral research. It is
recommended that the models examined in this study be cross-validated against a larger
sample. For the purpose of this study, however, the plausibility of suggested post-hoe
modifications was cautiously considered before respecifying the model.

Given that risk load served as an appropriate fi'amework with which to assess
heightened risk for child abuse, it is recommended that the risk load concept be tested
with other family- and parent-related outcomes. Results of the current study also suggest
further examining relations between combinations of specific stressors and increased
child abuse potential, above and beyond examining the associations between child abuse
potential and number of stressors. That is, a risk load index aggregates sets of variables
that may be related to child abuse potential in different ways. Examples pertaining to
child outcomes are available: Liaw and Brooks-Gunn (1994) found that different sets of
risk factors were associated with children’s IQ scores and behavior problem scores.
Assessing risk within various sets may provide unique information about the relative
strengths of risk factor clusters, and it will be important for future work to elucidate more
fully differences among levels based on individual risk factors.

In addition, a multiple risk index that weights all factors equally does not
distinguish between persistent and transitory variables (such as current unemployment or
single parenthood), potentially privileging the transitory variables because chronic

adversity tends to relate more strongly to poor outcomes (Ackerman et al., 1999).

70

Longitudinal assessment of risk—beyond the score of the current study—would be able
to examine potential changes in families’ risk load.

Besides incorporating a longitudinal design, future research would benefit from
addressing child characteristics for their contributing role to abuse potential. Similarly,
measures of parent-child interaction would add breadth to this study and perhaps provide
information about the structural relationships among risk load, depression, parenting
stress, and abuse potential.

In addition, variation in data collection would provide important information
about these relationships. In this study, mother’s self-report was utilized for all variables
(based in part on the nature of the program’s intake process). Having observational
measures or alternative informants could strengthen, weaken, or change these results.

Future researchers need to pursue the direct and indirect mechanisms of a
multitude of variables to more firmly ascertain the pathways toward physical child abuse.
Utilizing different measures that assess knowledge of infant development and self-rated
parenting skill are also called for in order to more effectively examine the potential role
these variables play in mitigating the harmful psychological effects of chronic stress
associated with risk load.

Despite these limitations, there are many strengths associated with the current
study. One such strength is based on the nature of the sample. Examining risk load in a
sample of economically disadvantaged families is informative. Using heterogeneous
samples (relative to poverty) may be inopportune in this context because controlling for
economic resources in these samples also tends to control for the cofactors and obscures

the diversity among disadvantaged families (Ackerman et al., 1999; C011 et al., 1996).

71

More generally, children and families served by Early Head Start and other early
intervention programs are gaining increasing attention by policy makers and social
scientists alike; results from this study provide important information about this at risk
. population that may serve to provide feedback to program policy and practices. As
described by Olds and colleagues (1997), the early intervention programs emphasize the
development of the parent because the parent’s behavior constitutes the most powerful
and potentially alterable influence on the developing child (Brooks-Gunn, Klebanov,
Liaw, & Spiker, 1993; Liaw et al., 1995).

Another strength is the study’s focus on abuse potential, rather than abusive
behaviors. Defining and quantifying child maltreatment are complex challenges, because
maltreating behavior is not often reliably reported, and different forms of maltreatment
frequently coexist within a family (Howes, Cicchetti, Toth, & Rogosch, 2000).
Indicators of actual maltreatment (e.g., child abuse reports) are often unreliable (W ekerle
& Wolfe, 1993), and mothers are likely to feel resistance to disclosing abusive parenting
practices. In addition, to plan early intervention services, we need to rely on risk factors
that exist before the outcome. While abuse potential does not necessarily indicate current
or future abusive behavior, it may serve as an early warning for parents who are prone to
abuse.

The model examined has the inclusive characteristic of incorporating both
individual and contextual conditions affecting the probability of a problem. Such a
framework, using multiple dimensions of risk, allows for complexity in the relationships
among the variables of interest. Past studies have tended to investigate family variables

in isolation. This does not allow for the examination of how risk factors may operate

72

together (e. g., additively or interactively) in placing a mother at heightened risk for
abusive parenting. In addition, assessment of risk factors in this population would help -
determine whether specific factors influence program involvement and outcomes for

these families (Ayoub et al., 1992).

73

REFERENCES

74

References

Abidin, R. R. (1990). Introduction to the special issue: The stresses of parenting.
Journal of Clinical Child Psychology, 19(4), 298 - 301.

Abidin, R. R. (1983). The Parenting Stress Index. Charlottesville, VA: Pediatric
Psychology Press.

Abidin, R. R. (1990). The Parenting Stress Index Short Form. Charlottesville,
VA: Pediatric Psychology Press.

Abidin, R. R. (1992). The determinants of parenting behavior. Journal of Clinical
Child Psychology, 21(4), 407 - 412.

Ackerman, B.P., Schoff, K., Levinson, K., Youngstrom, E.A., & Izard, CE.
(1999). The relations between cluster indexes of risk and promotion at the problem
behaviors of 6— and 7-year-old children fi'om economically disadvantaged families.
Developmental Psychology, 35 (6), 1355 -— 1366.

Ammerman, R. T. (1990). Etiological models of child maltreatment: A
behavioral perspective. Behavior Modification, 14(3), 230 - 254.

Ammerman, R. T. (1991). The role of the child in physical abuse: A reappraisal.
Violence and Victims, 6(2), 87 - 101.

Arbuckle, J.L., & Wothke, W. (1999). Amos 4.0 ’s user ’s guide. USA: Small
Waters Corporation.

Ayoub, C., Jacewitz, M. M., Gold, R. G., & Milner, J. S. (1983). Assessment of a
program’s effectiveness in selecting individuals “at risk” for problems in parenting.

Journal of Clinical Psycholog, 39(3), 334 - 339.

75

Ayoub, C.C., Willett, J .B., & Robinson, BS. (1992). Families at risk of child
maltreatment: Entry-level characteristics and growth in family functioning during
treatment. Child Abuse & Neglect, 16, 495 - 511.

Azar, S. T., Robinson, D. R., Hekimian, E., & Twentyman, C. T. (1984).
Unrealistic expectations and problem-solving ability in maltreating and comparison
mothers. Journal of Consulting and Clinical Psychology, 52, 687 - 691.

Bandura, A. (1982). Self-efficacy mechanism in human agency. American
Psychologist, 37, 122 - 147.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive
theory. Englewood Cliffs: NJ: Prentice Hall.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic and statistical
considerations. Journal of Personality and Social Psychology, 51, 1173 - 1182.

Bavolek, S. J., & Henderson, H. L. (1990). Child maltreatment and alcohol
abuse: Comparisons and perspectives for treatment. Journal of Chemical Dependency
Treatment, 3(1), 165 - 184.

Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., Erbaugh, J. (1961). An
inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.

Bell, C. S., Johnson, J. E., McGillicuddy-Delisi, A. V., & Sigel, I. E. (1980).
Normative stress and young families: Adaptation and development. Family Relations, 29,
453 - 458. ’

Belsky, J. (1984). The determinants of parenting: A process model. Child

Development, 55, 83 - 96.

76

Belsky, J. (1993). Etiology of child maltreatment: A developmental-ecological
analysis. Psychological Bulletin, 114(3), 413 — 434.

Belsky, J ., Lerner, R. & Spanier, G. (1984). 772e child in the family. Reading,
MA: Addison-Wesley.

Benasich, A. A., & Brooks-Gunn, J. (1996). Maternal attitudes and knowledge of
child-rearing: Associations with family and child outcomes. Child Development, 67, 1186
- 1205.

Biller, H. B., & Solomon, R. S. (1986). Child maltreatment and paternal
deprivation: A manifesto for research, prevention and treatment. Lexington: MA: D.C.
Heath.

Bird, C. E. (1997). Gender differences in the social and economic burdens of
parenting and psychological distress. Journal of Marriage and the Family, 59, 809 - 823.

Black, D.A., Heyman, R.E., & Smith Slep, AM. (in press). Risk factors for child
physical abuse. In Aggression and Violence Behavior. State University of New York at
Stony Brook. ‘

Bohlin, G., & Hagekull, B. (1987). “Good mothering”: Maternal attitudes and
mother-infant interaction. Infant Mental Health Journal, 8(4), 352 - 363.

Brayden, R. M., Altemeier, W. A., Tucker, D. D., Dietrich, M. S., & Vietze, P.
(1992). Antecedents of child neglect in the first two years of life. Journal ofPediatrics,
120, 426 - 429.

Bredehoft, D. J. (1990). An evaluation study of the self esteem: A family affair

program with high-risk abusive parents. Transactional Analysis Journal, 20, 111 - 117.

77

Bronfenbrenner, U. (1979). Contexts of child rearing: Problems and prospects.
American Psychologist, 43(10), 844 - 850.

Brooks-Gunn, J ., Klebanov, P.K., Liaw, F-R., & Spiker, D. (1993). Enhancing
the development of low-birthweight, premature infants: Changes in cognition and
behavior over the first three years. Child Development, 64, 736 —- 753.

Brown, J. Cohen, R, Johnson, J .G., & Salzinger, S. (1998). A longitudinal
analysis of risk factors for child maltreatment: Findings of a 17-year prospective study of
officially recorded and self-reported child abuse and neglect. Child Abuse & Neglect, 22
(11),1065 — 1078.

Browne, K., & Saqi, S. (1988). Approaches to screening for child abuse and
neglect. In K. Browne & C. Davies (Eds), Early prediction and prevention of child abuse
(pp. 57 - 85). Chichester, England: John Wiley & Sons.

Bugenthal, D. B., Blue, J., & Cruzcosa, M. (1989a). Perceived control over
caregiving outcomes: Implications for child abuse. Developmental Psychology, 25, 532 -
539.

Bugenthal, D. 8., Blue, J ., & Lewis, J. (1989b). Caregiver cognitions as
moderators of aflective reactions to diflicult children: Implications for child abuse
(Unpublished manuscript). Santa Barbara: University of California, Department of
Psychology.

Bugenthal, D. B., & Cortez, V. L. (1988). Physiological reactivity to responsive
and unresponsive children as moderated by perceived control. Child Development, 5 9,

686 - 693.

78

/ Bullock, C. B., & Pridham, K. F. (1988). Sources of maternal confidence and
uncertainty and perceptions of problem-solving competence. Journal of Advanced
Nursing, 13, 321 - 329.

Burke, J., Chandy, J ., Dannerbeck, A., & Watt, J .W. (1998). The parental
environment cluster model of child neglect: An integrative conceptual model. Child
Welfare, 77(4), 389 — 405.

Burrell, B., Thompson, B., & Sexton, D. (1992). The measurement integrity of
data collected using the Child Abuse Potential Inventory. Educational and Psychological
Measurement, 52, 993 - 1001.

Caliso, J. A., & Milner, J. S. (1992). Childhood history of abuse and child abuse
screening. Child Abuse & Neglect, 16, 647 - 659.

Cantrell, P. J ., Carrico, M. E, Franklin, J. N., & Grubb, H. J. (1990). Violent
tactics in family conflict relative to familial and economic factors. Psychological Reports,
66, 823 - 828.

Christmas, A.L., Wodarski, J .S., & Smokowski, PR. (1996). Risk factors for
physical child abuse: A practice theoretical paradigm. Family Therapy, 23 (3), 233 —
248.

Cicchetti, D., & Howes, P.W. (1991). Developmental psychopathology in the
context of the family: Illustrations from the study of child maltreatment. Canadian
Journal of Behavioural Science, 23 (3), 257 - 281.

Cicchetti, D., & Lynch, M. (1993). Toward and ecological/transitional model of
community violence and child maltreatment: Consequences for children’s development.

Psychiatry, 56, 96 - 118.

79

Cicchetti, D., & Rizley, R. (1981). Developmental perspectives on the etiology,
intergenerational transmission and sequelae of child maltreatment. New Directions for
Child Development, 11, 31 - 55.

Cicchetti, D., & Toth, S. L. (1995). A developmental psychopathology
perspective on child abuse and neglect. Journal of the American Academy of Child and
Adolescent Psychiatry, 34(5), 541 - 565.

Coie, J. D., Watt, N. R, West, S. G., Hawkins, J. D., Asamow, J. R., Markman, H.
J ., Ramey, S. L., Shure, M. B., & Long, B. (1993). The science of prevention: A
conceptual framework and some directions for a national research program. American
Psychologist, 48, 1013 - 1022.

Coll, C.G., Crnic, K., Lamberty, G., Wasik, B.H., Jenkins, R, Garcia, H.V., &
McAdoo, HP. (1996). An integrative model for the study of developmental
competencies in minority children. Child Development, 67, 1891 - 1914.

Conger, R., Elder, G., Jr., Lorenz, F., Simons, R., & Whitbeck, L. (1992). A
family process model of economic hardships and influences on adjustment of early
adolescent boys. Child Development, 63, 526 - 541.

Conger, R. D., McCarty, J. A., Yang, R K., Lahey, B. B., & Kropp, J. P. (1984).
Perception of child, child-rearing values and emotional distress as mediating links
between environmental stressors and observed maternal behavior. Child Development,
55, 2234 - 2247.

Connelly, C. D., & Straus, M. A. (1992). Mother’s age and risk for physical

abuse. Child Abuse & Neglect, 16, 709 - 718.

80

Conrad, 8., Gross, D., Fogg, L., & Ruchala, P. (1992). Maternal confidence,
knowledge, and quality of mother-toddler interactions: A preliminary study. Infant
Mental Health Journal, 13(4), 353 - 362.

Crnic, K., & Acevedo, M. (1995). Everyday stresses and parenting. In M. H.
Bomstein (Ed), Handbook of parenting (V 01. Vol.4: Applied and practical parenting, pp.
277 - 297). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.

Crnic, K. A., & Greenberg, M. T. (1990). Minor parenting stresses with young
children. Child Development, 61, 1628 - 1637.

Crnic, K. A., Greenberg, M. T., Ragozin, A. 8., Robinson, N. M., & Basham, R.
B. (1983). Effects of stress and social support on mothers and premature and full-term
infants. Child Development, 54, 209 - 217.

Cummings, E. M., & Davies, P. T. (1994). Maternal depression and child
development. Journal of Child Psychology and Psychiatry, 35(1), 73 - 112.

Deater-Deckard, K. & Scarr, S. (1996). Parenting stress among dual-earner
mothers and fathers: Are there gender differences? Journal of Family Psychology, 10
(1), 45 — 59.

DeLongis, 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 Psycholog, 54(3), 486-495.

Dickstein, S., Seifer, R., Hayden, L. C., Schiller, M., Sameroff, A. J., Keitner, G.,
Miller, I., Rasmussen, S., Matzko, M., & Magee, K. D. (1998). Levels of family
assessment: II. Impact of maternal psychopathology on family functioning. Journal of

Family Psychology, 12(1), 23 - 40.

81

Dore, M. M., Doris, J. M., & Wright, P. (1995). Identifying substance abuse in
maltreating families: A child welfare challenge. Child Abuse & Neglect, 19(5), 531 - 543.
Dukewich, T. L., Borkowski, J. G., & Whitman, T. L. (1996). Adolescent
mothers and child abuse potential: An evaluation of risk factors. Child Abuse & Neglect,

20(11), 1031 - 1047.

Egeland, B., Breitenbucher, M., & Rosenberg, D. (1980). Prospective study of
the significance of life stress in the etiology of child abuse. Journal of Consulting and
Clinical Psychology, 48(2), 195 - 205.

Egeland, B., & Brunnquell, D. (1979). An at-risk approach to the study of child
abuse. Journal of the American Academy of Child Psychiatry, 18, 219 - 235.

Egeland, B., Deinard, A., Brunnquell, D., Phipps-Yonas, S., & Chricton, L.
(1979). A prospective study of the antecedents of child abuse: University of Minnesota.

Esdaile, S. A., & Greenwood, K. M. (1995). Issues of parenting stress: A study
involving mothers of toddlers. Journal of Family Studies, 1(2), 153 - 165.

Evans, A. L. (1980). Personality characteristics and disciplinary attitudes of child
abusive mothers. Child Abuse & Neglect, 4, 179 - 187.

Fennell, D. C., & Fishel, A. H. (1998). Parent education: An evaluation of STEP
on abusive parents’ perceptions and abuse potential. Journal of Child and Adolescent
Psychiatric Nursing, 11(3), 107 - 120.

Fitzgerald, H.E., Strommen, E.A. & McKinney, J .P. (1982). Developmental

psychology: The infant and young child. Homewood, IL: Dorsey Press.

82

Frankel, K. A., & Harmon, R. J. (1996). Depressed mothers: They don’t always
look as bad as they feel. Journal of the American Academy of Child and Adolescent
Psychiatry, 35, 289 - 298.

Fraser, M. W. (1997). The ecology of childhood: A multisystems perspective. In
M. W. Fraser (Ed.), Risk and resilience in childhood: An ecological perspective (pp.
320). Washington, DC: National Association of Social Workers Press.

Fuller, J. (1989). Maternal self-confidence: A reflection of expectations. Paper
presented at the American Nurses’ Association Council of Nurse Researchers
International Conference, Washington, DC.

Fulton, A. M., Murphy, K. R., & Anderson, S. L. (1991). Increasing adolescent
mothers’ knowledge of child development: An intervention program. Adolescence,
26(101), 73 - 81.

Garbarino, J. (1976). A preliminary study of some ecological correlates of child
abuse: The impact of socioeconomic stress on mothers. Child Development, 47, 178 -
185.

Garbarino, J. (1977). The human ecology of child maltreatment: A conceptual
model for research. Journal of Marriage and the Family, 721 - 735.

Garbarino, J ., & Sherman, D. (1980). High-risk neighborhoods and high-risk
families: The human ecology of child maltreatment. Child Development, 51, 188 - 198.

Gelardo, M. S., & Sanford, E. E. (1987). Child abuse and neglect: A review of

the literature. School Psychology Review, 16(2), 137 - 155.

83

 

Gelfand, D. M., Teti, D. M., & Fox, C. E. R. (1992). Sources of parenting stress
for depressed and nondepressed mothers of infants. Journal of Clinical Child Psychology,
21(3), 262 - 272.

Gelles, R. J. (1973). Child abuse as psychopathology: A sociological critique and
reformulation. American Journal of Orthopsychiatry, 43, 611 - 621.

Gil, D. (1970). Violence against children: Physical child abuse in the United
States. Cambridge, MA: Harvard University Press.

Gil, D. (1975). Unraveling child abuse. American Journal of Orthopsychiatry,
45, 346 - 348.

Glachan, M. (1991). Child abuse: A social and cultural phenomenon. Early
Childhood Development and Care, 74, 95 - 102.

Goerge, R.M. & Lee, B.J. (1997). Abuse and neglect of children. In R.A.
Maynard (Ed), Kids having kids: Economic costs and social consequences of teen
pregnancy, (pp. 205 — 230). Washington, DC: The Urban Institute Press.

Goodnow, J. J. (1988). Parents’ ideas, actions and feelings: Models and methods
from developmental and social psychology. Child Development, 59, 286 - 320.

Harman, K. & Luster, T. (1991). Influence of parent, child and contextual factors
on the quality of the home environment. Infant Mental Health Journal, 12 (1), l7 — 30.

Haskett, M. E., Johnson, C. A., & Miller, J. W. (1994). Individual differences in
risk of child abuse by adolescent mothers: Assessment in the perinatal period. Journal of

Child Psychology and Psychiatry, 35(3), 461 - 476.

84

Haskett, M.E., Smith Scott, S., & Fann, KB. (1995). Child Abuse Potential
Inventory and parenting behavior: Relationships with high-risk correlates. Child Abuse
& Neglect, 19(12), 1483 — 1495.

Haynes, C. F., Cutler, C., Gray, J., & Kempe, R. S. (1984). Hospitalized cases of
nonorganic failure to thrive: The scope of the problem and short-term lay health visitor
intervention. Child Abuse & Neglect, 8, 229 - 242.

Herrenkohl, E. C., & Herrenkohl, R. C. (1979). A comparison of abused children
and their nonabused siblings. Journal of the American Academy of Child Psychiatry, 18,
260 - 269.

Herrenkohl, E. C., Herrenkohl, R. C., Rupert, L. J ., Egolf, B. P., & Lutz, J. G.
(1995). Risk factors for behavioral dysfunction: The relative impact of maltreatment,
SES, physical health problems, cognitive ability, and quality of parent-child interactions.
Child Abuse & Neglect, 19(2), 191 - 203.

Hetherington, E. M., Cox, M., & Cox, R (1982). Effects of divorce on parents
and children. In M. E. Lamb (Ed.), Nontraditional families: Parenting and child
development (pp. 233 - 288). I-Iillsdale, NJ: Lawrence Erlbaum Associates, Inc.

I-Iillson, J. M. C ., & Kuiper, N. A. (1994). A stress and coping model of child
maltreatment. Clinical Psychology Review, 14(4), 261 - 285.

Holden, E. W., & Banez, G. A. (1996). Child abuse potential and parenting stress
within maltreating families. Journal of Family Violence, 11(1), 1 - 12.

Holden, E. W., Willis, D. J ., & Foltz, L. (1989). Child abuse potential and
parenting stress: Relationships in maltreating parents. Psychological Assessment: A

Journal of Consulting and Clinical Psychology, 1(1), 64 - 67.

85

Holden, G. W., & Ritchie, K. A. (1989). Linking extreme marital discord, child
rearing and child behavior problems: Evidence from battered women. Austin: University
of Texas, Dept. of Psychology.

Houck, G. M., & King, M. C. (1989). Child maltreatment: Family characteristics
and developmental consequences. Issues in Mental Health Nursing, 10, 193 - 208.

Howes, P.W., Cicchetti, D., Toth, S.L., & Rogosch, RA. (2000). Affective,
organizational, and relational characteristics of maltreating families: A systems
perspective. Journal of Family Psychology, 14 (1), 95 — 110.

Howze, D. C., & Kotch, J. B. (1984). Disentangling life event, stress and social
support: Implications for the primary prevention of child abuse and neglect. Child Abuse
& Neglect, 8, 401 - 409.

Hoyle, RH. & Panter, AT (1995). Writing about structural equation models. In
Structural Equation Modeling, (Ed.) RH. Hoyle. Thousand Oaks, CA: Sage.

Jackson, S., Thompson, RA., Christiansen, B.H., Colman, RA. et al. (1999).
Predicting abuse-prone parental attitudes and discipline practices in a nationally
representative sample. Child Abuse and Neglect, 23 (1), 15 — 29.

Jones, E. D., & McCurdy, K. (1992). The links between types of maltreatment
and demographic characteristics of children. Child Abuse & Neglect, 16, 201 - 215.

Justice, B., & Duncan, D. F. (1976). Life crisis as a precursor to child abuse.
Public Health Reports, 91(2), 110 - 115.

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 (2), 169 - 192.

86

Kotch, J. B., Browne, D. C., Ringwalt, C. L., Stewart, P. W., Ruina, E., Holt, K.,
Lowman, B., & Jung, J .-W. (1995). Risk of child abuse or neglect in a cohort of low-
income children. Child Abuse & Neglect, 19(9), 1115 - 1130.

Kowal, L. W., Kottmeier, C. P., Ayoub, C. C., Komives, J. A., Robinson, D. S., &
Allen, J. P. (1989). Characteristics of families at risk of problems in parenting: Findings
from a home-based secondary prevention program. Child Welfare, 68(5), 529 - 53 8.

Lahey, B. B., Conger, R. D., Atkeson, B. M., & Treiber, F. A. (1984). Parenting
behavior and emotional status of physically abusive mothers. Journal of Consulting and
Clinical Psychology, 52(6), 1062 - 1071.

Larsen, J. J ., & McCreary Juhasz, A. (1985). The effects of knowledge of child
development and social-emotional maturity on adolescent attitudes toward parenting.
Adolescence, 20(80), 823 - 839.

Liaw, F .-R, & Brooks-Gum, J. (1994). Cumulative familial risks and low
birthweight children’s cognitive and behavioral development. Journal of Clinical Child
Psychology, 23(4), 360 - 372.

Linsky, A. S., Straus, M. A., & Colby, J. P. (1985). Stressful events, stressful
conditions and alcohol problems in the United States: A partial test of Bales’s theory.
Journal of Studies on Alcohol, 46(1), 72-80.

Luster, T. & Brophy-Herb, H. (2000). Adolescent mothers and their children.
The WAIMH Handbook of Infant Mental Health (V 01., IV): Infant Mental Health in

Groups at High Risk.

87

Luster, T. & Dubow, E. (1990). Predictors of the quality of the home
environment that adolescent mothers provide for their school-age children. Journal of
Youth & Adolescence, l9 (5), 475 — 494.

Luster, T. & Rhoades, K. (1989). The relation between child-rearing beliefs and
the home environment in a sample of adolescent mothers. Family Relations, 38, 317 —
322.

MacCallum, R. C., & Widaman, K. F. (1999). Sample size in factor analysis.
Psychological Methods, 4(3), 84—99.

Macphee, E. (1981). Manual: Knowledge of Infant Development. Unpublished
manuscript.

Mash, E. J ., & Johnston, C. (1990). Determinants of parenting stress: Illustrations
from families of hyperactive children and families of physically abused children. Journal
of Clinical Child Psychology, 19(4), 313 - 328.

Mash, E. J ., Johnston, C., & Kovitz, K. (1983). A comparison of the mother-child
interactions of physically abused and non-abused children during play and task situations.
Journal of Clinical Child Psychology, 12(3), 337 - 346.

McClennan, S., & Harkless, G. (1998). Self-efficacy, stress and parental
adaptation: Applications to the care of childbearing families. Journal of Family Nursing,
4(2), 198 - 215.

McCubbin, M. A., & McCubbin, H. I. (1993). Family coping with health crises:
The resiliency model of family stress, adjustment and adaptation. In C. Danielson, B.
Hamel-Bissell, & P. Winstead-Fry (Eds), Families, health and illness (pp. 21 - 64). New

York: Mosby.

88

McCurdy, K. (1995). Risk assessment in child abuse prevention programs. Social
Work Research, 19(2), 77 - 87.

McGrath, M. M., & Meyer, E. C. (1992). Maternal self-esteem: From theory to
clinical practice in a special care nursery. Children’s Health Care, 21(4), 199 - 205.

/ Mercer, F. T., & F erketich, S. L. (1994). Predictors of maternal role competence
by risk status. Nursing Research, 43(1), 38 - 43.

Merenda, P., F. (1997). A guide to the proper use of factor analysis in the conduct
and reporting of research: Pitfalls to avoid. Measurement and Evaluation in Counseling
and Development, 30, 156-164.

Miller, C.L., Miceli, P.J., Whitman, T.L., & Borkowski, J .G. (1996). Cognitive
readiness to parent and intellectual-emotional development in children of adolescent
mothers. Developmental Psychology, 32 (3), 533 — 541.

Milner, J. S. (1989a). Additional cross-validation of the Child Abuse Potential
Inventory. Psychological Assessment: A Journal of Consulting and Clinical Psychology,
11(3), 219 - 223.

Milner, J. S. (1989b). Applications and limitations of the Child Abuse Potential
Inventory. Early Childhood Development and Care, 42(3 ), 85 - 97.

Milner, J. S. (1989c). Applications of the Child Abuse Potential Inventory.
Journal of Clinical Psychology, 45(3), 450 - 454.

Milner, J. S., & Ayoub, C. (1980). Evaluation of “at risk” parents using the Child

Abuse Potential Inventory. Journal of Clinical Psychology, 36(4), 945 - 948.

89

Milner, J. 8., Gold, R G., Ayoub, C., & Jacewitz, M. M. (1984). Predictive
validity of the Child Abuse Potential Inventory. Journal of Consulting and Clinical
Psychology, 52(5), 879 - 884.

Milner, J. S., Gold, R G., & Wimberley, R C. (1986). Prediction and
explanation of child abuse: Cross-validation of the Child Abuse Potential Inventory.
Journal of Consulting and Clinical Psychology, 54(6), 865 - 866.

Milner, J. S., & Robertson, K. R. (1985). Development of a random response
scale for the Child Abuse Potential Inventory. Journal of Clinical Psychology, 41(5), 639
- 643.

Milner, J. S., & Robertson, K. R. (1989). Inconsistent response patterns and the
prediction of child maltreatment. Child Abuse & Neglect, 13, 59 - 64.

Milner, J. S., & Wimberley, R. C. (1979). An inventory for the identification of
child abusers. Journal of Clinical Psychology, 35(1), 95 - 101.

Milner, J. S., & Wimberley, R. C. (1980). Prediction and explanation of child
abuse. Journal of Clinical Psychology, 3 6(4), 875 - 884.

Monroe, L. D., & Schellenbach, C. J. (1989). Relationship of Child Abuse
Potential Inventory scores to parental responses: A construct validity study. Child and
Family Behavior Therapy, 11(3/4), 39 - 58.

Olds, D., Kitzman, H., Cole, R, & Robinson, J. (1997). Theoretical foundations
of a program of home visitation for pregnant women and parents of young children.
Journal of Community Psychology, 25 (1), 9 — 25.

Palmer, D. (1983). An attributional perspective on labeling. Exceptional

Children, 43, 423 — 429.

90

Passino, A.W., Whitman, T.L., Borkowski, J .G., Schellenbach, C.J., Maxwell,
S.E., Keogh, D., & Rellinger, E. (1993). Personal adjustment during pregnancy and
adolescent parenting. Adolescence, 28, 97 — 121.

Pearlin, L. I., Menaghan, E. G., Lieberman, M. A., & Mullan, J. T. (1981). The
stress process. Journal of Health and Social Behavior, 22, 337 - 356.

Perry, M. A., Wells, E. A., & Doran, L. D. (1983). Parent characteristics in
abusing and nonabusing families. Journal of Clinical Child Psychology, 12(3), 329 - 336.

Peterson, J ., & Hawley, D. (1998). Effects of stressors on parenting attitudes and
family functioning in a primary prevention program. Family Relations, 47(3), 221 - 227.

Pianta, R. C., & Egeland, B. (1990). Life stress and parenting outcomes in a
disadvantaged sample: Results of the Mother-Child Interaction Project. Journal of
Clinical Child Psychology, 19(4), 329 - 336.

Pruitt, D. L., & Erickson, M. T. (1985). The Child Abuse Potential Inventory: A
study of concurrent validity. Journal of Clinical Psychology, 40(1), 1001.

Quittner, A. L., Glueckauf, R. L., & Jackson, D. N. (1990). Chronic parenting
stress: Moderating versus mediating effects of social support. Journal of Personality and
Social Psychology, 59(6), 1266 - 1278.

Ragozin, A.S., Basham, RB., Crnic, K.A., Greenberg, M.T., & Robinson, NM.
(1982). Effects of maternal age on parenting role. Developmental Psychology, 18, 627 —
634.

Raykov, T., Tomer, A., & Nesselroade, J .R. (1991). Reporting structural
equation modeling results in Psychology and Aging: Some proposed guidelines.

Psychology and Aging, 6 (4), 499 -- 503.

91

Reis, J ., Barbera-Stein, L., & Bennett, S. (1986). Ecological determinants of
parenting. Family Relations, 35, 547 - 554.

Roberts, RE., Lewinsohn, P.M., & Seeley, J .R (1991). Screening for adolescent
depression: A comparison of depression scales. Journal of the American Academy of
Child and Adolescent Psychiatry, 30 (l), 58 — 66.

Robertson, K. R, & Milner, J. S. (1985). Convergent and discriminant validity of
the Child Abuse Potential Inventory. Journal of Personality Assessment, 49(1), 86 - 88.

Rodgers, A. Y. (1998). Multiple sources of stress and parenting behavior.
Children and Youth Services Review, 20(6), 525 - 546.

Rodriguez, C. M., & Green, A. J. (1997). Parenting stress and anger expression
as predictors of child abuse potential. Child Abuse & Neglect, 21 (4), 367 - 377.

Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American
Journal of Orthopsychiatry, 5 7, 316 - 331.

Sameroff, A. (1975). Transactional models in early social relations. Human
Development, 18, 65 - 79.

Sameroff, A. J., Bartko, W. T., Baldwin, A., Baldwin, C., & Seifer, R (1998).
Family and social influences on the development of child competence. In M. Lewis, C.

F eiring et al. (Eds), Families, risk and competence (pp. 161 - 185). Mahwah, NJ:
Erlbaum.

Sameroff, A. J ., & Chandler, M. J. (1975). Reproductive risk and the continuum
of caretaking casualty. In F. D. Horowitz, M. Hetherington, S. Scarr-Salapatek, & G.
Siegal (Eds), Review of child development research (Vol. 4). Chicago: University of

Chicago Press.

92

Sameroff, A. J ., & F eil, L. A. (1985). Parental concepts of development. In I. E.
Sigel (Ed.), Parental belief systems (pp. 83 - 105). Hillsdale, NJ: Erlbaum.

Sameroff, A. J ., & Seifer, R. (1983). Familial risk and child competence. Child
Development, 54, 1254 - 1268.

Sameroff, A. J ., & Seifer, R. (1995). Accumulation of environmental risk and
child mental health. In H. E. Fitzgerald, B. M. Lester, & B. Zuckerman (Eds), Children
of poverty (pp. 233 - 254). New York: Garland Publishing.

Sameroff, A. J ., Seifer, R, Zax, M., & Barocas, R. (1987). Early indicators of
developmental risk: The Rochester longitudinal study. Schizophrenia Bulletin, 13, 383 -
393.

Seifer, R, Sameroff, A. J ., Baldwin, C. P., & Baldwin, A. (1992). Child and
family factors that ameliorate risk between 4 and 13 years of age. Journal of the
American Academy of Child and Adolescent Psychiatry, 31(5), 893 - 903.

Shaner, J. M., Peterson, K. L., & Roscoe, B. (1985). Older adolescent females’
knowledge of child development norms Adolescence, 20(77), 53 - 59.

Sharma, 8., Singh, M. B., & Sighu, K. (1985). Factors affecting child
deve10pment of rural and urban expectant mothers. Indian Psychological Review, 28(3),
23 - 30.

Simons, R, Beaman, J ., Conger, R, & Chao, W. (1993). Childhood experience,
conceptions of parenting, and attitudes of spouse as determinants of parental behavior.

Journal of Marriage and the Family, 55, 385 - 396.

93

Simons, R, & Johnson, C. (1996). The impact of marital and social network
support on quality of parenting. In G. Pierce, R. Sarason, & I. Sarason (Eds), Handbook
of social support and the family (pp. 269 - 287). New York: Plenum.

Simons, R. L., Whitbeck, L. B., Melby, J. N., & Wu, C. (1994). Economic
pressure and harsh parenting. In R. D. Conger & G. H. Elder (Eds), Families in troubled
times: Adapting to change in rural America (pp. 207 - 222). New York: Adeline.

Spicer, S. J ., & Franklin, C. (1994). Exploratory effects of social support, stress,
and locus of control on the conflict tactics of parents at risk for child maltreatment.
Journal of Social Service. Research, 19(3-4), 1 - 22.

Spinetta, J ., & Rigler, D. (1972). The child-abusing parent: A psychological
review. Psychological Bulletin, 7, 296 - 304.

Steele, B. F ., & Pollack, C. B. (1974). A psychiatric study of parents who abuse '
infants and smallichildren. In R. E. Helfer & C. H. Kempe (Eds), The battered child.
Chicago, IL: University of Chicago Press.

Stevens, J. H. (1984). Child deve10pment knowledge and parenting skills. Family
Relations, 33, 237 - 244.

Straus, M. A. (1983). Ordinary violence, child abuse and wife beating: What do
they have in common? In D. Finkelhor (Ed.), The dark side of families (pp. 213 - 234).
Beverly Hills, CA: Sage.

Straus, M. A., Gelles, R. J ., & Steinmetz, S. K. (1980). Behind closed doors:

Violence in the American family. Garden City, NJ: Anchor Books.

94

Teti, D. M., & Gelfand, D. M. (1991). Behavioral competence among mothers of
infants in the first year: The mediational role of self-efficacy. Child Development, 65(5),
918 - 929.

Trickett, P. K., & Kuczinski, L. (1986). Children’s misbehavior and parental
discipline strategies in abusive and nonabusive families. Developmental Psychology, 22,
115 - 123.

Twentyman, T. C., & Plotkin, R. C. (1982). Unrealistic expectations of parents
who maltreat their children: An educational deficit that pertains to child development.
Journal of Clinical Psychology, 38, 497 - 503.

US. Department of Health and Human Services. Child Maltreatment 1998:
Reports from the States to the National Child Abuse and Neglect Data System.
Washington, DC: US. Government Printing Office, 2000.

Walker, L. O., Crain, H., & Thompson, E. (1986). Mothering behavior and
maternal role attainment during the postpartum period. Nursing Research, 35(6), 352 -
355.

Webster-Stratton, C. (1990). Stress: A potential disrupter of parent perceptions
and family interactions. Clinical Child Psychology, 19, 302 - 313.

Wekerle, C. & Wolfe, DA. (1993). Prevention of child physical abuse and
neglect: Promising new directions. Clinical Psychology Review, 13 (6), 501 — 540.

Whipple, E. E., & Webster-Stratton, C. (1991). The role of parental stress in

physically abusive families. Child Abuse & Neglect, 15, 279 - 291.

95

Whissell, C., Lewko, J ., Carriere, R, & Radford, J. (1990). Test scores and
sociodemographic information as predictors of child abuse potential scores in young
female adults. Journal of Social Behavior and Personality, 5(3), 199 - 208.

Wilcox, H., Field, T., Prodromidis, M., & Scafidi, F. (1998). Correlations
between the BDI and CES-D in a sample of adolescent mothers. Adolescence, 33 (131),
565 — 574.

Williams, S., Anderson, J ., McGee, R, & Silva, P. A. (1990). Risk factors for
behavioral and emotional disorder in preadolescent children. Journal of the American
Academy of Child and Adolescent Psychiatry, 29, 413 - 419.

Wolfe, D. A. (1985). Child-abusive parents: An empirical review and analysis.
Psychological Bulletin, 97(3), 462 - 482.

Wolfe, D. A. (1987). Child abuse: Implications for child development and
psychopathology. Newbury Park, California: Sage.

Zich, J .M., Attkisson, C.C., & Greenfield, TR (1990). Screening for depression
in primary care clinics: The CES-D and the BDI. International Journal of Psychiatry in
Medicine, 20 (3), 259 — 277.

Zuravin, S. J. (1989). The ecology of child abuse and neglect: Review of the
literature and presentation of data. Violence and Victims, 4(2), 101 - 120.

Zuravin, S. J. & DiBlasio, F .A. (1996). The correlates of child physical abuse

and neglect by adolescent mothers. Journal of Family Violence, 11, 149 - 166.

96

APPENDIX

97

1 . on
2. Par tress

4.

 

98

 

Table 3: Missing Data/Valid Cases (N=95)

 

 

 

 

 

 

 

 

Variable Total # % of % with Complete
data points for scale missing missing scale cases/0
(items x subjects) individual scores missing items

items

Risk load 25 x 95 = 2375 0% 1.05% 98.95%

Depression:

BDI 21 x 95 = 1995 .01% 11.6% 93.68%

CES-D 20 x 95 = 1900 0% 88.4% 11.6%

Parenting Stress: 95 79%

PSI 23 x 95 = 2185 .73% 0

Knowledge of

Infant

Development: 14 x 95 = 1330 .03% 0 95.79%

KIDI

Parenting skill 1 x 95 = 95 5.26% n/a 94.73%

Abuse Potential:

CAPI 77 x 95 = 7315 .72% 0 77.89%

 

 

 

 

 

99

 

. «A.,—logo 3n Gena—moo: Zufix om mow—om moo—.3 Ea 00333338

\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.: <3. o _ N u a m o q a 3 3 Z 3
o Z”: 098 .\
_ Em. ”mo 3min...
m 053 wmo ..3 -.:
w GEE max :3 -._o :2...
A 93:05. .3 MA... -. 3 .8
m _ < Non: ..NA .3... :8 now .Nw...
o 452m. )3 now .: -.o_ .3 loo
3 32:85 .3 low -.~o on Joe :3 ..3...
3 38:5 .3 .m _ ........ -. 3 ANN... .: .31.... AN? .3
3 t 5303 .2 .31.... o; Juo... sow .51... low -.o_ .21...
_o n 5998 .3 .31... now AB... .3 .31... .8 -.oo .33.... .31....
2 :5” wow .3 .33... J3 :3 .3 .31.... Joe .ou .oo..;.... .31.... 33......
3 E<o 5 Lo :3 .: ..oo -._o .3 nos :3 so.“ .3 nos :3
mb. .........
3 W03 .3 .31.... )2 AMA... o; .33.... Jon no. .31... .31... .31... .33... -.oo
3 0055. now -.3 now .om not. .8 -.3 L _ .3 Jon -.oo loo .3
3 .560 two 33...... :8... .3 .3 -.3 :3 .8... -.o3 38 .. -._o :3 AZ -.o3
3 .360 .oo :8... -.oo .No .oo :3 ._o .3 :3 13*... 23... 13...... .3
”Hgm *** 43.2..
3 amigo -.~o :3 .34 :2 .oo -.3 .33.. -.~o umo Awo... )3 ..3...... Lo
3 Doc Lo -._ 3 AMA... now Job ..3 .3 .3 JG )3 now ..No .3
3 3833mm -.oo 33 Joe -.~o :3 :3 no. .3 now now -.o3 JG .oS
no 55.5% .3... .3 no; .3 -.o_ ..3 low .3 .3 .3 ._o .3 a:
2 m5: .3 .3 .3. -. E .3 .3 ..3 ..oo .3 .3 on .3 -.:
MN >30on :3 23*... wow moo mom :3 ._o -.o_ a: :3 :3 -.3 now
No be $3: -.: Lo .8 .3 -.oo -.o_ :3 -.: Joe ..2 -._o 23 now
MA 09:33 .om .oo -.oA :3 -.3 .3 -.: .om .34 .3 now .33.. -.3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3 >332“: J3 Joo J3 J3 .3 J3 .oo 32 .3 J3 J8 J8 J3
3 2.5. now 33 .22 .oo J3 .32 .oo .3 J3 .oo .8 .8 .8 Joo
3 025:0 =2 .3 .3 .oo J3 .3 .3 32 Joe .3 .33 J3 .o3 Joo
3 2.2933029 .3 .o3 .3 Joe Joe .oo J3 J3 32 .33 Joe 32 32
3 2332va J8 .32 .3... .2 2 J8 J3 .3... Joe .3 J3 .33 J 3 33
3 0.3853 .3 .33 .31. J3 J3 J3... .3 J3 J3... J3 J3 J3... .3
32 9.395 .33 J3 .3 Jou .3 J3 .3 .2 2 J3 J3... J3 J3... J 2 o
3 Sam mom .3... .3 .31... J22 .3 .oo .3. J3 .3 .3 32 .oo J3
3 2.3 m=m2 .3 .3 33 J3 33...... .22 J3 .33 32...... .3 .3 .3 J3
3 ho was J32 Joe .8 .3 .8 .3 .32.... J3 J3 J3 .32 .32 .32
”Ow.— ***
3 U055. 33 Jon .3 .3 .3 J3 .32 J3 J2 .. J3 J8 J31. J3
3 2352.82. .oo .3 .3 Joo .8 J3 J2 J 3 32 J3 J8 J3 J8
3 2222838. J3 Jom J3 .8 .3 .3 .3... J 3 J2 J3 .3 .33 .32
3.3..
3 [$1320 .32 J3 .3... .3 J2 J3 .3... J3 J3 J33 J2 J8 .3
mvm—GH: +3.... ..22.. *** .37... *** ..2..*
3 9532.2 J3 .8 J3 J8 J2 .3 .33.. J 3 Joa .3 .33., .3 .8
3o r02 :8 J3 32 .3... J3 J3 J3 .3 J3... J3 .3 .8 J3 .8
32 2.33.203 J 3 .3 J8 .3 J3 .8 Jom .8 . 2 3 .3 2 .. .32 ... .3 J 3
3 Damn? .oo J3 .3 .3 J8 J3 .8 .8 J3 J3...... J3 J3 J3
352.8 I... .........
3 30: J3... J3 Joo J8 J3 J8 .oo .3 J3... J3 J3 J3 .3
333 1....
3 3535. J3. .3 .33 J32 .. J8 J8 .8 J3 J3. J3 J8 .3 Joe
:2
3 2:88 J3 J3 .3 .8 .22 J8 .3 J3 J3 J3 J 3 J32 .. .32
3 WE. 255.. .3 .33 32 J3 J8 J3 .8 .2o .8 J3 32 .8 J2 2
0:. 5.13? 3. 3 3 3 3 3 3 No 32 3 3 3
3 09:5. .3

 

 

101

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3 H3305 J3 .3
3 .33 J3 J3 .3
:85 .1...
3 Eur—own J 3 J2...... J3 .3
3 Gov J3 J3 J3 .3... .3...
3 335mm .3 J3 .2 .2 .3... .31...
3 302—mm .3 .2 J3 .3 J3 J3 J3
2 m5: .3 .3 .3 J2 J3 J3 J3 .3
52.2.. ***
3 >953» J3 J3 .2 .3 .31.... .33.... .33.... J3 J3...
3 he 32.: .3 .3 .2 .3 .3 J3 J3 .3 .2 J3
3 60235 .31. J3 J3 J2 .3... J3 .3... .3 J2 .3 .3...
3 >c\=om_ J3 J3 J3 .3 .31.... .2... .3 J3 J3 .3 J3 .3
3 2:32am .3 J3 .3 J3 .31.... .3 .3 3 .3 .2 J3 .31...
3 0:33 3 .3 J3... J3 J3 .31... .3 .3 J3 .3 .3 J3 .3...
mm 2838 .3 J3 .3 J3 .3 J3 .3 J3 .3... .3 J 3 .3
3 632:8 J3 J2 J3 .3 .33.... .3 .3 J3 J3 .3... J3 .2
3 05. J2 J3 J3 J2 .33.... .3 .3 J3 J2 .3 J3 J3
8:5 1.... I.
2 mccmvcm J3 J3... JNN... J3 .31.... .3 .3 .2 J2 .3 .3 .3
3 Fwnmmom .3 J3 J3 J3 .33.... J3 .3 .3 .3 .3 .2 .31....
3 FE m=m_ .3 .3 J2 J3 J3 .3 .3 J3 .3 .3 J3 J3
3 wow so: J3 J3 J2 J3 33...... .3. J3 J2 .. .3 .3 .3 J3
3 015% J3... J3 J2 .3 .31... .2 J3 J3 .3 .3 .2 J3
3 3:68”. .3 J3 J3 J3 .21.... .2 .3 .3 .3 .3 .3 .3
3 3:038. J3 J2 .2 J3 .31... .2 J3 J31. .2 .3 .3 J3
3 mmsmfi J3 J3 .2 N .31. .33... .3... .3 .3 J2 .3... .3 J3
332 .........
3 Cumin—v. .3 J3 .3 .3 .3 3 .3 J2 .3 .3 .2 J3
3 be So J3 J 3 .3... J3 3 J3 J2... .3 .3 J3 .3 .J2

 

 

 

 

102

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

S 3:53... .3 -.2 :3 -. 2 w .3 -.S .3 .3 -.3 .3 .3 .3

3 6238 -.3 -.3 .3 .3... .31.. .3 .3 -.3 -.3 b3 .3 .3

3... 3963: -.No... -.3 .3 .No .31... .3... .3... -.3 -. 2 m .31.... .oo .2

3 3358 .3 -.3 -.ow ..o... .31.... .3 .3 -.3 .3 .3 -.3 .3
2 anmm

3 2:88 -.3 -.3 .03 .3 .31.... .3 .3 .2 -.3 .oo .8... -.3

3 WE.»— -.oo -. 2 m -.3 .3 .No .3 .2 -.o3 -.3 .3 -. 2 o .3
rocmmsm ......

9: «Elna—... 3 Na 3 3 3 we 3 3 mm 3 3 3

no 2:: ma. .3

3 03.30 E .3 .33.,

3 magma“. .8 .3 3...

3 :3” 8m .31.... .3 .3 -.ox

mo 939.6 .31.... .No .31. .3 .31...

8 mag—US .3... ...... .3 .33. .8 .3... ...... .31....

3 55m mom .3 38...... .8... .3 .8 ... 33...... .3

3 Ca mam. .3 .3 .2 -.om -.3 -.2 -.3 .3

3 we: no: .31 -.om .8 -.3 .8 ... .3 .38. -. _ m .3

3 U055: .31... .3 .31. .2 .39.... .31... 8 ...... 2 -.3 .39.

3 3:682. .33... .3 .3 -.S .31. .31... 38...... .3 -.2 .3 .31...

3.3 5:088. .No .2 .3 -.3 .3 .3 3 -.2 .2 .31.... No .3

3 933:: .oo ..3 ..oo -.om .3 .8... .31. ..8 -.3 .3 .31.. .No

3 6:23.; .2 b3 -. 3 .3 -.3 -.S -.3 ..3 -.3 .3 ..3 -. _ 3

3o ho in -.om .om -. 3 :3... .2 .3 .03 ..8 13...... .3 .3 .3

3. 3:503 .3 -. 3 .3 -.3 .3 .8 .S -.o3 -.ow .oo -.o3 .3

.3 U338 .3 .3 .3 -.3 .8... -.oo .3 .3 -.S -.3 .S .No

3m 32. .3 .3 .8 -.oo .3 3 .3 -.3 -.3 .2 -.3 ..3
“.39:

33 Edi .33.. .3 .3... .3 .3 .31... 3 .3 -.oo .3 .3 83......
59: E

 

 

 

 

103

 

104

Table 12: Oneway AN OVAs comparing Clusters formed by Child Abuse Potential

Scores

Clusters were compared on the following variables: child age and sex, maternal age and
education, marital status, 1- vs. 2-parent household, receiving TANF and Medicaid,

number of people in household and number of children in household, yearly rent/heating
bills, income, risk load, knowledge of infant development, self-rated parenting skill,
depression, parenting stress, and all individual risk factors.

Only significant differences are displayed in Table 12.

 

 

 

 

 

 

 

 

 

Descriptives
N _M_ (s_d) Std. 95% Confi'gnce Min Max
error Interval for M
Risk load 1 6 8.8 (2.9) 1.2 5.8 — 11.9 6 l
2 28 6.0 (3.2) .6 4.8 — 7.2 1 16
3 21 7.4 (3.6) .8 5.8 - 9.0 3 14
4 40 5.0 (2.6) .4 4.2 — 5.9 2 14
Total 95 6.1 (3 .2) .3 5.4 — 6.7 1 16
Depression 1 6 25.0 (6.4) 2.6 18.3 — 31.7 20 35
2 28 11.9 (5.9) 1.1 9.6 — 14.1 2 23
3 21 19.5 (7.0) 1.5 16.3 - 22.7 5 35
4 40 6.1 (4.3) .7 4.7 - 7.4 0 17
Total 95 11.9 (8.3) .9 10.3 — 13.6 0 35
Parenting 1 6 65.2 (14.9) 6.1 49.5 — 80.8 43 7:37—
Stress 2 28 45.2 (10.6) 2.0 41.1 —49.3 26 70
3 21 58.1 (14.8) 3.2 51.4—64.9 35 86
4 40 36.1 (11.4) 1.8 32.5 — 39.8 24 79
Total 95 45.5 (15.6) 1.6 42.3 - 48.7 24 87
Nutritional 1 6 .67 (.52) .21 .12 — 1.21 0 1
deficit 2 28 .11 (.31) e e - .23 0 1
3 20 .10 (.31) e e- .24 0 1
4 40 .18 (.38) e e - .30 0 1
Total 94 .17 (.38) e e - .25 O 1
Chronic 1 6 .137 (.41) .17 .26 - .60 0 1
illness 2 28 .21 (.42) e e - .38 0 1
3 20 e e e - .15 0 1
4 40 e e e - e 0 1
Total 94 e e e - .16 0 1
Unstable 1 6 .17L(.41) .17 -.26— .60 o 1
residence 2 28 e (.19) e e - .11 0 1
3 20 .25 (.44) e e - .46 0 1
4 40 e (.22) e e - .12 0 1
Total 94 e (.30) e e - .16 O 1
Single 1 6 .67 (.52) .21 .12 — 1.21 O 1
parent 2 28 .39 (.50) e .20 - .59 0 l

 

 

 

 

 

 

 

 

 

105

 

 

 

 

 

 

 

 

 

 

3 20 .85 (.37) e .68 - 1.02 O 1

4 40 .45 (.50) e .29 - .61 0 1

Total 94 .53 (.50) e .43 - .63 0 1

Teen parent 1 6 .83 (.41 ) .17‘ .40 - 1.26 0 1
2 28 .39 (.50) e .20 - .59 0 1

3 20 .60 (.50) .11 .36 - .84 0 1

4 40 .25 (.44) e .11 - .39 0 1

Total 94 .40 (.49) e .30 - .51 0 1

Distress 1 6 240.83 (11.1) 4.53 229.18 - 252.49 219 249
(subscale of 2 28 99.89 (26.7) 5.05 89.54 - 110.25 53 158
CAPI) 3 21 187.43 (32.3) 7.04 172.74 - 202.12 123 261

4 40 19.85 (16.9) 2.67 14.45 - 25.25 0 62

Total 95 94.44 (78.9) 8.10 78.37 - 110.52 0 261

Rigidity 1 6 26.00 (23.02) 9.40 1.84 - 50.16 5 56
(subscale of 2 28 9.93 (8.68) 1.64 6.56 - 13.30 0 37
CAPI) 3 21 18.67 (15.41) 3.36 11.65 - 25.68 3 59

4 40 7.60 (8.57) 1.36 4.86 - 10.34 0 28

Total 95 11.89 (12.76) 1.31 9.30 - 14.49 0 59

Unhappiness 1 6 39.83 (10.26) 4.19 29.06 - 50.61 27 53
(subscale of 2 28 15.82 (10.25) 1.94 11.85 - 19.80 0 36
CAPI) 3 21 32.24 (13.95) 3.04 25.89 - 38.59 8 55

4 40 7.68 (6.44) 1.02 5.61 - 9.74 0 27

Total 95 17.54 (14.72) 1.51 14.54 - 20.54 0 55

Problems 1 6 16.00 (10.6) 4.33 4.87 - 27.13 0 29
with Child 2 28 4.82 (6.54) 1.24 2.28 - 7.36 0 27.
(subscale of 3 21 6.48 (5.93) 1.29 3.78 - 9.18 0 21
CAPI) 4 40 2.85 (5.23) .83 1.18 - 4.52 O 19
Total 95 5.06 (6.88) .71 3.66 - 6.46 O 29

Problems 1 6 35.0 (3.29) 1.34 31.55 - 38.45 32 38
with Family 2 28 13.93 (12.77) 2.41 8.98 - 18.88 0 38
(subscale of 3 21 25.81 (12.2) 2.66 20.26 - 31.36 1 38
CAPI) 4 40 4.13 (7.18) 1.14 1.83 - 6.42 0 32
Total 95 13.76 (14.18) 1.46 10.87 - 16.65 0 38

Problems 1 6 22.83 (.75) .31 22.04 - 23.62 22 24
with Others 2 28 11.50 (6.87) 1.30 8.84 - 14.16 0 24
(subscale of 3 21 16.52 (6.55) 1.43 13.54 - 19.51 1 24
CAPI) 4 40 4.05 (5.09) .81 2.42 - 5.68 0 17
Total 95 10.19 (8.32) .85 8.49 - 11.88 0 24

 

 

 

 

 

 

 

 

106

 

Post Hoc Tests

 

 

 

Risk load

Student-Newman-Keuls a'b
QCL_2 Cluster Subset for alpha = .05
Number of Case N 1 2

40 6.03
2 28 6.00
3 21 7.38 7.38
1 6 8.83
Sig. .095 .198

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean
of the group sizes is used. Type I error levels are

 

 

 

 

not guaranteed.
Depression

Student-NewmanoKeuls 8'”
QCL_2 Cluster Subset for alpha = .05
Number of Case N 1 2 3 4
4 40 6.05
2 28 11.86
3 21 19.52
1 6 25.00
Sig. 1.000 1.000 1.000 1.000

 

 

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.545.

b. The group sizes are unequal. The harmonic mean of the group sizes is
used. Type I error levels are not guaranteed.

 

 

 

 

Parenting Stress
Student-Newman—Keuls “b
QCL_2 Cluster Subset for alpha = .05
Number of Case N 1 2 3
4 40 36.13
2 28 , 45.18
3 21 58.14
1 6 65.17
Sig. 1.000 1.000 .125

 

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean of the
group sizes is used. Type I error levels are not guaranteed.

107

 

A_RISK4 Nutritional deficit

Student-Newman-Keuls a'b

 

 

 

 

QCL_2 Cluster Subset for alpha = .05
Number of Case N 1 2

3 .flf

2 28 .11

4 40 .18

1 6 .67
Sig. . .841 1.000

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
3. Uses Harmonic Mean Sample Size = 14.421.

b- The group sizes are unequal. The harmonic mean
of the group sizes is used. Type I error levels are
not guaranteed.

A_RISK5 Chronic illness (trend)

Student-Newman-Keuls 3",

 

 

 

 

 

 

 

 

Subset

for alpha
QCL_2 Cluster = ~05
Number of Case N 1

40 2.56502"
3 20 5.00E-02
1 6 .17
2 28 .21
Sig. .297
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.421. _

b- The group sizes are unequal. The harmonic mean
of the group sizes is used. Type I error levels are
not guaranteed.

108

A_RISK7 Unstable residence

Student-Newman-Keuls a'b

 

 

 

 

Subset

for alpha
QCL_2 Cluster = -05
Number of Case N 1
2 err-3757m-
4 40 5.00E-02
1 6 .17
3 20 .25
Sig. .1 96

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.421.

b- The group sizes are unequal. The harmonic mean
of the group sizes is used. Type I error levels are
not guaranteed.

A_RISK16 Single parent

Student-Newman-Keuls 8'”

 

 

 

 

Subset

for alpha
QCL_2 Cluster = -05
Number of Case N 1
I T—T
4 40 .45
1 6 .67
3 20 .85
Sig. .056

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Hannonic Mean Sample Size = 14.421.

b- The group sizes are unequal. The harmonic mean
of the group sizes is used. Type I error levels are
not guaranteed.

109

A_RlSK21 Teenage parent

Student-Newman-Keuls 8'”

 

 

 

QCL_2 Cluster ' Subset for alpha = .05
Number of Case N 1 2

40 .25
2 28 .39
3 20 .60 .60
1 6 .83
Sig. .1 17 .185

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.421.

b- The group sizes are unequal. The harmonic mean
of the group sizes is used. Type I error levels are

 

 

 

 

not guaranteed.

Distress (subscale of CAPI)
Student-Newman-Keuls a'b
QCL_2 Cluster Subset for alpha = .05
Number of Case N 1 2 3 4
4 40 1 9.
2 28 99.89
3 21 187.43
1 6 240.83
Sig. 1 .000 1.000 1 .000 1 .000

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean of the group sizes is
used. Type I error levels are not guaranteed.

 

 

 

Rigidity (subscale of CAPI)
Student-NewmanoKeuls a'b
00L 2 Cluster Subset for algha = .05
Number of Case N 1 2
4 40 7.60
2 28 9.93
3 21 18.67
1 6 26.00
Sig. .590 .092

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean
of the group sizes is used. Type I error levels are
not guaranteed.

110

 

Student-Newman-Keuls a'b

Unhappiness (subscale of CAPI)

 

QCL_2 Cluster

Subset for alpha = .05

 

 

 

 

 

 

 

 

Number of Case N 1 2 3 4
40 7.58
2 28 1 5.82
3 21 32.24
1 6 39.83
Sig. 1 .000 1 .000 1.000 1 .000

 

 

Means for groups in homogeneous subsets are displayed.

a. Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean of the group sizes is

used. Type I error levels are not guaranteed.

Student-Newman-Keuls a‘b

Problems with Child (subscale of CAPI)

 

 

 

 

 

 

QCL_2 Cluster SUbSEt for alpha = .05
Number of Case N 1 2

4 40 235

2 28 4.82

3 21 6.48

1 6 16.00
Sig. .260 1.000

 

 

 

Means for groups in homogeneous subsets are displayed.
3° Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean

Student-Newman-Keuls a'b

of the group sizes is used. Type I error levels are
not guaranteed.

Problems with Family (subscale of CAPI)

 

 

 

QCL_2 Cluster Subset for alpha = .05

Number of Case N 1 2 3 4

f 40 4.13

2 28 13.93

3 21 25.81

1 6 35.00
Sig. 1 .000 1 .000 1 .000 1 .000

 

 

 

 

 

 

Means forgroups in homogeneous subsets are displayed.

a. Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean of the group sizes is

used. Type I error levels are not guaranteed.

111

 

 

Student-Newman-Keuls a'b

Problems with Oflrers (subscale of CAPI)

 

 

 

 

 

QCL_2 Cluster Subset for alpha = .05

Number of Case N 1 2 3 4

4 4O '10?

2 28 11.50

3 21 16.52

1 6 22.83
Sig. 1 .000 1 .000 1 .000 1.000

 

 

 

 

 

 

 

Means for groups in homogeneous subsets are displayed.

a. Uses Harmonic Mean Sample Size = 14.545.

b- The group sizes are unequal. The harmonic mean of the group sizes is

used. Type I error levels are not guaranteed.

112

 

'1‘.

Table 13: Output for Model 1

Amos Version 4 by James L. Arbuckle

Your model contains the following variables

RISKLOAD observed endogenous
STRESS observed endogenous
DEP observed endogenous
AB_POT observed endogenous
SKILL observed endogenous
e1 unobserved exogenous
e3 unobserved exogenous
e2 unobserved exogenous
e4 unobserved exogenous
65 unobserved exogenous

Number of variables in your model: 10

Number of observed variables: 5

Number of unobserved variables: 5
Number of exogenous variables: 5
Number of endogenous variables: 5

Summary of Parameters

Weights Covariances Variances Means Intercepts Total
Fixed: 5 0 0 0 0 5
Labeled: 0 0 0 0 0 0
Unlabeled: 8 1 5 O O 14

 

 

 

 

Total: 13 1 5 0 0 19

NOTE: The model is recursive.
Sample size: 95

Model: Default model

113

Computation of degrees of fi'eedom

Number of distinct sample moments: 15
Number of distinct parameters to be estimated: 14

Degrees of freedom: 1

0e 2 0.0e+000 -1.3461e-001 1.00e+004
1e 0 8.1e+000 0.0000e+-000 8.79e-001
2e 0 l.1e+001 0.0000e+000 l.92e-001
3e 01.6e+001 0.00006+000 1.94e-001
4e 01.9e+001 0.0000e+000 4.94e-002
5e 01.9e+001 0.0000e+000 3.69e-003
6e 01.9e+001 0.0000e+000 2.04e-005

Minimum was achieved

Chi-square = 0.759
Degrees of freedom = 1
Probability level = 0.384

 

1.45985277450e-I-002 0 1.00e+004
1.552356480076001 18 8836-001
4.36901631557e+000 3 0.00e+000
8.69541143622e-001 1 1.01e+000
7.59495792027e-001 1 l.04e+000
7.5905 7894867e-001 1 1.00e+000
7.5905 78825 1 8e-001 l 1 .00e+000

 

 

Maximum Likelihood Estimates

Regression Weights: Estimate S.E. C.R. Label
DEP <------- RISKLOAD 0.587 0.261 2.246
STRESS < ------ RISKLOAD 1.078 0.491 2.194
SKILL <---------- DEP -0.023 0.014 -1.670
SKILL < ------- STRESS -0.016 0.007 -2.106
AB_POT <----- SKILL 7.397 7.317 1.011
AB_POT < ------ DEP 7.224 1.014 7.128

AB_POT < ------ STRESS
AB_POT < ...... RISKLOAD

2.566 0.543 4.722

5.744 2.199 2.613

114

Standardized Regression Weights: Estimate

 

DEP <.--...... RISKLOAD 0.226
STRESS <-.-.- RISKLOAD 0.221
SKILL < ...... .- DEP -0.193
SKILL <------.- STRESS .0243
AB_POT < ..... .- SKILL 0.065
AB_POT <-------- DEP 0.528
AB_POT <-..----- STRESS 0.352

AB_POT <----- RISKLOAD 0.161

 

 

 

 

Covariances: Estimate S.E. C.R. Label
e3 < > e2 ‘ 65.868 14.260 4.619
Correlations: Estimate
e3 < > e2 0.542
Variances: - Estimate S.E. C.R. Label

 

e1 10.068 1.469 6.856
e3 228.603 33.345 6.856
e2 64.632 9.428 6.856
e5 0.851 0.124 6.856

64 4283.051 624.747 6.856

Squared Multiple Correlations: Estimate

 

RISKLOAD 0.000

DEP 0.051
STRESS 0.049
SKILL 0.149

AB_POT 0.665

115

Total Efi‘ects

RISKLOAD DEP STRESS SKILL

 

DEP 0.587 0.000 0.000 0.000

STRESS 1.078 0.000 0.000 0.000
SKILL -0.031 -0.023 -0.016 0.000
AB_POT 12.525 7.052 2.450 7.397

Standardized Total Effects

RISKLOAD DEP STRESS SKILL

 

DEP 0.226 0.000 0.000 0.000

STRESS 0.221 0.000 0.000 0.000
SKILL -0.097 -0.193 -0.243 0.000
AB_POT 0.352 0.515 0.336 0.065

Direct Effects

RISKLOAD DEP STRESS SKILL

 

DEP 0.587 0.000 0.000 0.000

STRESS 1.078 0.000 0.000 0.000
SKILL 0.000 -0.023 -0.016 0.000
AB_POT 5.744 7.224 2.566 7.397

Standardized Direct Effects

RISKLOAD DEP STRESS SKILL

 

DEP 0.226 0.000 0.000 0.000
STRESS 0.221 0.000 0.000 0.000
SKILL 0.000 -0.193 -O.243 0.000
AB_POT 0.161 0.528 0.352 0.065

116

Indirect Effects

RISKLOAD DEP STRESS SKILL

 

DEP 0.000 0.000 0.000 0.000
STRESS 0.000 0.000 0.000 0.000
SKILL -0.031 0.000 0.000 0.000
AB_POT 6.782 -0.173 -0.1 16 0.000

Standardized Indirect Effects

RISKLOAD DEP STRESS SKILL

 

DEP 0.0000 0.0000 0.0000 0.0000

STRESS , 0.0000 0.0000 0.0000 0.0000
SKILL -0.0971 0.0000 0.0000 0.0000
AB_POT 0.1904 -0.0126 -0.0159 0.0000

 

 

 

 

Modification Indices

Covariances: M.I. Par Change
Variances: M.I. Par Change
Regression Weights: M.I. Par Change
Summary of models

 

Model NPAR CMIN DF P CMIN/DP

 

Default model 14 0.759 1 0.384 0.759
Saturated model 15 0.000 0
Independence model 5 161.057 10 0.000 16.106

 

Model RMR GFI AGFI PGFI

 

Default model 5.607 0.997 0.952 0.066
Saturated model 0.000 1.000
Independence model 350.925 0.579 0.368 0.386

117

DELTAl RHOl DELTA2 RH02
Model NFI RFI IFI TLI CFI

 

Default model 0.995 0.953 1.002 1.016 1.000
Saturated model 1.000 1.000 1.000
Independence model 0.000 0.000 0.000 0.000 0.000

Model PRATIO PNFI PCFI

 

Default model 0.100 0.100 0.100
Saturated model 0.000 0.000 0.000
Independence model 1.000 0.000 0.000

Model NCP L0 90 HI 90

 

Default model 0.000 0.000 6.3 13
Saturated model 0.000 0.000 0.000
Independence model 151.057 1 13.616 195.939

Model FMIN F0 L0 90 HI 90

 

Default model 0.008 0.000 0.000 0.067
Saturated model 0.000 0.000 0.000 0.000
Independence model 1.713 1.607 1.209 2.084

Model RMSEA L0 90 HI 90 PCLOSE

 

Default model 0.000 0.000 0.259 0.437
Independence model 0.401 0.348 0.457 0.000

Model AIC BCC BIC CAIC

 

Default model 28.759 30.668 87.045 78.513
Saturated model 30.000 32.045 92.450 83.308
Independence model 171.057 171.739 191.874 188.826

Model ECVI L0 90 HI 90 MECVI

 

Default model 0.306 0.309 0.376 0.326
Saturated model 0.319 0.319 0.319 0.341

118

Independence model 1.820 1.421 2.297 1.827

HOELTER HOELTER
Model .05 .01

 

Default model 476 822
Independence model 11 14

Execution time summary:

Minimization: 0.060
Miscellaneous: 0.650
Bootstrap: 0.000
Total: 0.710

119

  

     

      
   

EUI 5

1110111111111

 

lllll '

6