Supporting Families in a High-Risk Setting: Proximal Effects of the SAFEChildren Preventive Intervention

Abstract

Four hundred twenty-four families who resided in inner-city neighborhoods and had a child entering 1st grade were randomly assigned to a control condition or to a family-focused preventive intervention combined with academic tutoring. SAFEChildren, which was developed from a developmental-ecological perspective, emphasizes developmental tasks and community factors in understanding risk and prevention. Tracking of linear-growth trends through 6 months after intervention indicated an overall effect of increased academic performance and better parental involvement in school. High-risk families had additional benefits for parental monitoring, child-problem behaviors, and children's social competence. High-risk youth showed improvement in problem behaviors and social competence. Results support a family-focused intervention that addresses risk in low-income communities as managing abnormal challenges.

Full text

Supporting Families in a High-Risk Setting: Proximal Effects of the
SAFEChildren Preventive Intervention
Patrick Tolan, Deborah Gorman-Smith, and David Henry
University of Illinois at Chicago
Four hundred twenty-four families who resided in inner-city neighborhoods and had a child entering 1st
grade were randomly assigned to a control condition or to a family-focused preventive intervention
combined with academic tutoring. SAFEChildren, which was developed from a developmental–
ecological perspective, emphasizes developmental tasks and community factors in understanding risk and
prevention. Tracking of linear-growth trends through 6 months after intervention indicated an overall
effect of increased academic performance and better parental involvement in school. High-risk families
had additional benefits for parental monitoring, child-problem behaviors, and children’s social compe-
tence. High-risk youth showed improvement in problem behaviors and social competence. Results
support a family-focused intervention that addresses risk in low-income communities as managing
abnormal challenges.
Extensive study of prevention of antisocial behavior and pro-
motion of prosocial development has occurred in the past decade.
The accumulated evidence about the efficacy of several interven-
tions has permitted researchers to identify several characteristics
common to many of these interventions (U.S. Department of
Health and Human Services, 2001). First, preventive benefit seems
to increase when an intervention targets multiple aspects of devel-
opment and multiple areas of social functioning (Durlak & Wells,
1997; Lipsey & Wilson, 1998; Tolan & Guerra, 1994). Second, a
focus on parenting and family-relationship characteristics has
proved beneficial (Elliott & Tolan, 1998). Third, effective preven-
tive efforts for antisocial behavior link development-influencing
settings and systems, with the connection of school and family as
the most prominent example (Conduct Problems Prevention Re-
search Group, 1999; Kellam & Rebok, 1992; Spoth, Lopez Reyes,
Redmond, & Shin, 1999). For example, one prevention program
for first- through sixth-grade children, called Linking Interests of
Families and Teachers, simultaneously emphasizes improving par-
enting skills through parenting groups and teachers’ in-classroom
instructional methods to reduce aggression and improve school
functioning. The intervention also focuses on improving commu-
nication and coordination between teacher and parents. Those
randomly assigned to the intervention showed lower subsequent
aggression, with the impact greatest among the higher risk segment
of that sample (Stoolmiller, Eddy, & Reid, 2000). This example
illustrates the value of focusing on family and school risk factors
simultaneously and the importance of linking the intervention
across systems that shape child development.
Linking Interests of Families and Teachers is also similar to
several other recent efficacious preventive interventions in its
focus not only on parenting skills but also on family-relationship
qualities and the family’s management of and interaction with
other developmental systems (Henggeler, Melton, & Smith, 1992;
Redmond, Spoth, Shin, & Lepper, 1999). It may be critical to teach
more than parenting skills. In addition, it may be important to
facilitate family communication and cohesion, increase opportu-
nities for the families’ social support from others, and improve
management of challenges that the family faces to promote social
competence and reduce antisocial behavior risk (Farrington &
Welsh, 1999; Tolan, Hanish, McKay, & Dickey, 2002).
In the present study, we focus on the efficacy of a preventive
effort (SAFEChildren) based on these emerging implications from
prior prevention efforts. However, this intervention differs from
most other efforts to prevent antisocial behavior by defining in-
clusion in the study on the basis of where families live rather than
individual risk factors. It was designed to aid families in inner-
city
1
neighborhoods during the important developmental transition
of their children entering first grade. The intervention intends to
help families gain or increase parenting and family-management
skills that facilitate successful child academic and social adjust-
ment and, therefore, promote social and academic competence and
lower related risk for later antisocial behavior. In addition, the
intervention focuses on promoting important initial academic
success.
The focus and emphasis of the intervention were based on a
developmental–ecological perspective on risk and prevention ef-
forts (Tolan, Guerra, & Kendall, 1995) and viewing normal de-
velopmental transitions as a time when intervention may be par-
ticularly useful (Kellam & Rebok, 1992). In addition, it was
informed by the apparent role of the inner city in heightening risk
for antisocial behavior (Sampson, Raudenbush, & Earls, 1997) and
1
Inner city is used to refer to a set of urban poor communities that are
distinguished from other urban and other poor communities by an extreme
concentration of poverty, high rates of crime, high rates of multiple social
problems such as early pregnancy and school dropout, and low rates of
home ownership or business investment. Wilson (1987) noted this distinc-
tion and related it to a social ecology of heightened risk for children and
families.
Patrick Tolan, Deborah Gorman-Smith, and David Henry, Institute for
Juvenile Research, Department of Psychiatry, University of Illinois at
Chicago.
Correspondence concerning this article should be addressed to Patrick
Tolan, Institute for Juvenile Research, University of Illinois at Chicago,
1747 West Roosevelt Road, Chicago, IL 60608. E-mail: tolan@uic.edu
Journal of Consulting and Clinical Psychology Copyright 2004 by the American Psychological Association
2004, Vol. 72, No. 5, 855–869 0022-006X/04/$12.00 DOI: 10.1037/0022-006X.72.5.855
855
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

implications from previous prevention trials and longitudinal risk
studies of youth in poor urban communities (e.g., Metropolitan
Area Child Study [MACS] Research Group, 2002a; Tolan,
Gorman-Smith, & Henry, 2003).
Applying a Developmental–Ecological Approach to
Preventive Intervention Design
These emerging characteristics of effective prevention programs
have been accompanied by continued progress in establishing a
multivariate, developmentally conceptualized approach to risk
(Coie et al., 1993; Reid, Patterson, Dishion, & Snyder, 2002). This
approach emphasizes the interdependence among multiple risk
factors in explaining the development of antisocial behavior. There
is also recognition that the necessity and salience of a given risk
factor can vary as a function of the social ecology in which the
child develops (Kellam & Rebok, 1992; Tolan et al., 1995).
Another feature of this approach is the assumption that risk factors
may vary in importance depending on age (Loeber & Farrington,
1998). For example, peer relations emerge as an important risk
factor as a child proceeds through elementary school years and
ascend to primacy during adolescence (Coie & Dodge, 1998;
Moffitt, 1993). These prominent features of a developmental–
ecological perspective are all considered important to adequately
explain why prevalence levels of such behavior can vary by
communities and how risk contributors might vary in salience
across communities. Similarly, they are also all important in ex-
plaining variation in risk for antisocial behavior within a given
community or segment of the population (Elliott & Tolan, 1998).
This conceptualization implies that developmental and ecological
features should inform prevention efforts. Specifically, targeting
youth and families living in high-risk settings, such as inner-city
communities, may be useful. An apt target of such an effort would
be aiding the management of general developmental challenges.
Normal Developmental Transitions as Prevention
Opportunities
The developmental–ecological perspective also emphasizes that
susceptibility to and expression of risk factors are greater during
times of transition (Cicchetti & Toth, 1992; Kellam & Rebok,
1992). For example, it has been shown that there is a general drop
in academic performance during the transition from elementary to
middle school (Roeser & Eccles, 1999). This developmental mis-
match is thought to strain the coping skills of most children and
also to disproportionately affect those at higher risk prior to the
transition (Cicchetti & Toth, 1992). However, there may also be a
corresponding increased opportunity for affecting risk at these
times of transition. If the transition can be managed better then
general risk can be lessened, then it seems plausible that it may be
particularly beneficial to those entering the transition time with
greater risk.
The developmental transition into elementary school is typically
a time of significant challenges. The family must engage with a
large and demanding social institution, children have less time
under parental supervision, and influences other than the family
figure more prominently (Entwisle, Alexander, Pallas, & Cadigan,
1988). From this time on, these other systems have direct and
increasingly salient influence on child development. This transi-
tion into and adjustment during first grade also serves as the
template or framework for experience in academic pursuit and
associated social relationships (Entwisle & Alexander, 1998). De-
pending on how this transition goes, families and their children
may come to view school as a place of opportunity, social con-
nection, and competence building or as a place of frustration,
painful inequities, and failure (Mayer, 1995; Van Acker, Grant, &
Henry, 1996). The resources of the school system may affect this
transition. Children entering schools that are more strained in
regard to resources and that face more challenges among their
student population may be more at risk for a problematic transition
(Entwisle & Alexander, 1998).
As inner-city schools are often more resource strained than other
schools, children and families in these communities may face a
more perilous transition than occurs elsewhere. This transition may
be also more difficult for inner-city parents given that their chil-
dren are more likely to attend schools with lower learning expec-
tations (Entwisle et al., 1988). Also, these parents are themselves
less likely to have been successful in school or to have experienced
schooling as a positive experience (Tolan, Sherrod, Gorman-
Smith, & Henry, 2004).
School entry may also be a time of acute impact of ecological
conditions on developmental risk for inner-city children. For ex-
ample, prior research has suggested that in a general sample of
inner-city children, rates of psychopathology did not exceed na-
tional norms at the beginning of first grade. However, by second
grade the rate had increased substantially and was significantly
higher than the national prevalence level (Tolan & Henry, 1996).
Thus, we viewed assistance in adequate management of this tran-
sition as an opportunity to promote healthy development and
reduce the risk of antisocial behavior in children.
The Role of the Inner City in Risk for Antisocial
Behavior
Research has shown that neighborhoods can be a crucial influ-
ence on risk and development (Brooks-Gunn, Duncan, Klebenov,
& Sealand, 1993; Sampson et al., 1997). Particularly in the inner
city, community characteristics seem to affect child development
beyond what can be accounted for by individual and family factors
(Gorman-Smith, Tolan, & Henry, 2000; Sampson et al., 1997).
Furthermore, some recent studies have suggested that characteris-
tics of the inner city may moderate (decrease) influence of parent-
ing characteristics in regard to child risk for antisocial behavior
(Tolan & Gorman-Smith, 1997), the impact of stress on children
(Gorman-Smith, Tolan, & Henry, 1998), and children’s social
competence (Furstenberg & Hughes, 1995). Thus, raising children
in the inner city may present a greater level of challenges com-
monly faced by parents but may also present challenges to families
that are not faced elsewhere in our society (Hawkins, Catalano, &
Miller, 1992; Mason, Cauce, Gonzales, & Hiraga, 1996).
Although to our knowledge there has been very little research to
date on this issue, it appears that preventive effects can vary by
community characteristics. For example, the MACS Research
Group (2002a) reported that a multicomponent, family-focused
intervention for elementary school children lowered aggression
levels on average when provided within poor urban communities
(those with more economic and social resources than inner-city
communities). However, when provided within the inner city, the
positive effects were limited to those most at risk: those scoring
within the clinical level on the measure of aggression at pretest.
856 TOLAN, GORMAN-SMITH, AND HENRY
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

One reason for this variation in effect may be due to the different
salience of some risk factors by community type. For example,
parenting practices may mediate the impact of social stress in some
communities but not in others (Gorman-Smith et al., 1998). The
MACS Research Group program’s aid with parenting practices
may not have resulted in general decreases in aggression within the
inner city because these parents may have needed more help with
managing the challenges of rearing children in a high-risk envi-
ronment, such as maintaining safety and maximizing benefits from
overtaxed schools.
These findings suggest that a preventive intervention that is
meant to affect risk through family influences in the inner city can
be enhanced by a focus on parenting and family functioning within
that ecological context. These include managing safety threats,
strained or underdeveloped school resources, and economic stress,
in addition to the more commonly addressed issues of parenting
practices, family organization, and communication. One avenue
for addressing these issues may be to improve social support
among families in the intervention. Such a focus expands from the
more common within-family issues to setting-related issues and to
family social relationships and beyond the family to important
institutions, such as schools. Another potentially valuable focus is
to provide information, skill training, and enhanced opportunity
for inner-city parents to be connected to and involved with con-
ventional systems, such as their children’s schools.
The SAFEChildren Preventive Intervention was designed to
incorporate these advances in understanding the risk factors of
inner-city children and to address the issues that challenge families
and schools in poor urban areas as children enter school. In prior
research, these risk factors have been linked to risk for aggression,
low academic achievement, and low social competence early in the
elementary years, which in turn are strong predictors of later
antisocial behavior (Lipsey & Wilson, 1998). On the basis of these
observations, we reasoned that there might be protective gain from
working to engage parents in their children’s academic adjustment
and achievement during first grade to bolster opportunity for the
children to achieve adequately with the basic academic skill of
reading, to support strong family relationships and parenting, and
to enable and support parents connecting with other parents and
the schools in addressing developmental and social-setting chal-
lenges. By supporting families, we hoped to promote healthy
development and reduce risk.
The SAFEChildren Preventive Intervention
In designing SAFEChildren, we combined two components.
The first component was a multiple-family group approach that
focused on (a) parenting skills, (b) family relationships, (c) under-
standing and managing expectable developmental and situational
challenges to the families, (d) increasing support among parents in
the group, (e) skills and issues in engaging as a parent with the
school, and (f) managing issues such as neighborhood problems
(e.g., violence). This component was combined with a phonic-
based reading tutoring program developed as part of the Fast Track
intervention and relates to academic improvement among high-risk
children during first grade (Conduct Problems Prevention Re-
search Group, 1999).
The intervention was begun at the outset of first grade and
involved 22 weeks of intervention. The multiple-family group
approach is seen as not only providing efficient access to more
families but also as promoting group processes that build support-
ive relationships and lessen isolation of families. Multiple-family
groups met weekly, and the children were engaged in twice weekly
30-min reading tutoring sessions. Multiple-family groups com-
bined information provision, skill practice, group problem solving,
and at-home exercises to apply session material. The tutoring
emphasized phonetics and step-by-step advancement in skills. The
multiple-family group was meant to affect the family skills and
processes listed above. Therefore, in combination with the aca-
demic support, the multiple-family group was meant to lead to
more positive attitudes about achievement and school involvement
for parents and children and lead to greater self-control, lower
aggression, and higher social competence for the children. The
intervention was intended to benefit generally as well as protect
those among this population who started school with risk-related
characteristics (e.g., poorer parenting skills and higher child
aggression).
Method
Research Participants
Because our interest was in risk inherent to the setting, we solicited
participants not because of individual or family characteristics but because
they attended schools that served inner-city neighborhoods. Five inner-city
schools that had previously collaborated in a passive longitudinal study, the
Chicago Youth Development Study, were invited to participate (Tolan et
al., 2004). That study focused on the development of risk for antisocial
behavior among youth growing up in the inner city. As part of the
agreement with the five participating schools at the outset of that longitu-
dinal study, we agreed to include them in any resulting prevention trial
effort. Therefore, all five schools were invited to and agreed to participate
in this study. To obtain the sample size needed for this study, we added two
additional neighborhood elementary schools that served similar
communities.
Most Chicago public school system elementary schools are organized as
neighborhood schools, which facilitated our interest in targeting inner-city
communities. The included schools serve communities that meet Wilson’s
(1987) definition of inner city: 40% or more below poverty level, little or
no income levels above the poverty level, crime rate higher than average
for Chicago, and a concentration of social problems (Tolan et al., 2003).
Table 1 provides a comparison of the communities in which the partici-
pants in SAFEChildren resided with statistics on selected demographic
variables and indicators of community distress for the city of Chicago and
the United States. As is evident from Table 1, these communities are more
impoverished, crime burdened, and have fewer resources than most com-
munities in Chicago or elsewhere in the country. These families reside in
communities that can be characterized as high risk.
In the spring of 1997, parents of all kindergarten children in each of the
seven schools were contacted and asked to participate in the study. A total
of 507 families were eligible to participate on the basis of their residence
within the neighborhood boundaries of the school that their children would
attend in the fall of the children’s first-grade year (although recruited in
spring of kindergarten year). Of these, 424 families (84%) consented to
participate and completed at least one baseline assessment.
Of the 424 families participating in the study, 42.5% reported their
ethnicity as African American and 57.5% as Latino. Of the target children,
51% were male, 40% lived in single-parent households, and 44% of the
primary caregivers (usually the mother) did not graduate from high school.
In addition, 59% had a family income below $20,000 per year, and 85%
had a family income below $30,000 per year in 1996. A total of 62% had
five or more people living in the household, and 57% of the families had
moved at least once in the year prior to the study. However, the majority
of families that changed addresses remained within the same
neighborhood.
857
SUPPORTING FAMILIES
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

Random assignment to intervention. After completing a baseline as-
sessment during the spring of their children’s kindergarten year, families
were randomly assigned to the intervention group or to a nonintervention
control group. Random assignment was done within classrooms to control
for school effects. Because research has found lower retention rates for
intervention compared with control participants, 55% of the children with
consent in each classroom were randomly assigned to the intervention
condition, and 45% were assigned to the control condition. Once random-
ized assignment was complete, the children remained in that condition for
evaluation purposes until the completion of the study.
2
Retention. Of the participating families initially agreeing to participate
in the study, 401 (95%) completed all five waves of the assessment in the
first phase of the study. There was no significant difference in attrition
based on ethnicity, gender, or other demographic variables that might be
related to risk: parental educational attainment,
␹
2
(4, N⫽424) ⫽2.37, ns;
parental occupation,
␹
2
(6, N⫽424) ⫽4.98, ns; receipt of public aid,
␹
2
(1,
N⫽424) ⬍1.00, ns; child–school bonding, F(6, 210) ⫽1.95, ns;
children’s social competence, F(15, 672) ⫽1.25, ns; family-relationship
characteristics, F(6, 227) ⬍1.00, ns; parenting practices, F(5, 227) ⬍1.00,
ns; self-control, F(15, 672) ⫽1.05, ns. Of those families randomly as-
signed to the intervention condition, 178 (82%) completed the intervention
(defined as participating in at least half of the family sessions combined
with the child’s participation in the tutoring intervention). However, the
majority of those participating completed the full intervention (78%). No
significant differences between completers and noncompleters of the in-
tervention were found.
Procedures
Five waves of data were collected for all participants to measure inter-
vention effects. In Wave 3 (midintervention), only teacher data were
collected owing to financial constraints. Data were collected from parents
and children at the other four points: Pretest 1 (end of kindergarten), Pretest
2 (beginning of first grade), posttest assessment (end of first grade), and a
6-month follow-up. Individual interviews were conducted with the primary
and secondary caregiver(s) (when the latter was available) and with the
target child. African American and Latino interviewers recruited from
communities of the study conducted interviews as two-person teams. After
obtaining informed consent and explaining the procedures to the family,
the two-person teams interviewed the parent(s) individually with laptop
computers. The child interview was completed with paper and pencil
methods and in private, although the caregivers were given the opportunity
to examine the questionnaires their children would be completing before
the interview began.
Prior to the study outset, interviews were piloted with families and
children in English and Spanish to ensure understanding so that the
families could respond fully. All measures were translated for Spanish
language interviews and then back to English prior to testing and use.
Every interview completed was checked by the project coordinator and
data-management staff for proper consent, missing data, accuracy, and
out-of-range values. A random sample of 20% of the cases received a
follow-up phone call from the project director to assess whether interview
procedures were followed. Teachers provided data regarding each child at
each of the data collections plus at midintervention.
Measures
We assessed four major targets of proximal effects for the intervention:
(a) child’s school functioning, (b) child’s behavior, (c) child’s social
competence, and (d) parenting and family relationships. Within each, we
often measured multiple variables usually from multiple sources. We also
measured important covariates and differentiated high-risk families and
children. Prior to evaluating effects, we conducted factor analyses and
other appropriate psychometric analyses to ensure that scales were reliable
and valid for this sample. In addition, where constructs were expected to be
measured from multiple sources, we tested measurement models, modify-
ing scaling when fit was not as expected. In some cases, cross-source
relations were not supported by the confirmatory factor analysis. In those
cases, we retained separate scales by source.
Child’s school functioning. Two constructs in this area were assessed.
Reading achievement was measured by administering the child the Wood-
cock Diagnostic Reading Battery (Woodcock, 1997). This is a comprehen-
sive set of individually administered tests that measures important dimen-
sions of reading achievement and closely related abilities. Four sections
were administered: (a) letter-word identification, (b) word attack, (c)
passage comprehension, and (d) incomplete words. These subscales from
these sections were combined for a total reading score in a manner
consistent with the published instructions and supported by our confirma-
tory factor analysis. In addition, confirmatory factor analysis affirmed no
substantial correlation between the measures of school bonding and read-
ing achievement. School bonding was measured with two subscales in the
child’s self-report version of the Behavioral Assessment System for Chil-
dren (BASC): the Attitude Toward Teacher and Attitude Toward School
subscales (Reynolds & Kamphaus, 1998). The BASC is a widely used,
published measure, with substantial evidence for reliability and validity
with school-age children as young as 8 years old. For the present sample,
the subscales had internal consistency reliabilities of .62 and .57, respec-
tively. Higher scores on these subscales indicate greater levels of school
bonding. Although both subscales measure aspects of school bonding, our
confirmatory factor analysis indicated that they should be regarded as
distinct constructs (correlations between the scales were modest, from .26
to .37). Therefore, we did not combine these subscales into a single
composite scale for analysis. The Attitude Toward Teacher and Attitude
2
As in any random assignment to intervention within a setting, there is
potential for contamination. However, this is most likely to work against
the finding of significant differences attributable to the intervention. In this
case, tutoring was individually offered only to those assigned to interven-
tion. Family groups were conducted after school or at other locations.
These design features would seem to help lessen the potential for control
group members to obtain benefits from intervention.
Table 1
Census and Crime Characteristics of SAFEChildren Schools and
Neighborhoods
Variable SAFEChildren
Schools Chicago United
States
Race and ethnicity (%)
a,b
Caucasian 25.13 45.48 80.35
African American 52.69 39.03 12.03
Other 22.17 15.49 7.62
Hispanic 32.03 19.23 8.81
Language spoken at home (%)
a
English 64.13 70.88 86.18
Spanish 29.76 17.39 7.53
Community poverty and crime (%)
1-year mobility rate
a
19.80 21.37 20.89
Poverty rate
a,b
28.24 21.62 13.12
Unemployment rate
a
16.53 11.33 6.31
Owner-occupied housing
a
37.69 41.50 64.20
Medical facilities per 100,000
c
47.45 111.86
Violent crimes per 100,000
d
6,944.15 824.84 922.70
a
Source: 1990 U.S. Census (U.S. Census Bureau, 1992).
b
Percentages
add to more than 100% because other and Hispanic ethnicity categories are
independent.
c
Statistics derived from national yellow pages and 1990
U.S. Census (U.S. Census Bureau, 1992).
d
Source: Federal Bureau of
Investigation Uniform Crime Reports, 1997 (Federal Bureau of Investiga-
tion, 1992).
858 TOLAN, GORMAN-SMITH, AND HENRY
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

Toward School subscales were the only self-report scales used from the
BASC.
Child’s behavior and social competence. Three aspects of a child’s
behavior were assessed with the Teacher Observations of Classroom Ad-
aptation—Revised (TOCA–R) and Parent Observations of Classroom Ad-
aptation—Revised (POCA–R; Kellam, Brown, Rubin, & Ensminger,
1983): aggression, hyperactivity, and concentration. Both versions are
frequently used and validated measures of a child’s behavior as observed
by the teacher and parent. Parents and teachers are given a structured
interview to elicit their experiences with the child and their observations of
different behaviors and characteristics. The interview does not specify
location of the observations of the behaviors, but teachers and parents
report their observations of the child’s behaviors that may affect adaptation
to school. Confirmatory factor analyses with our sample supported the
original scale content and the choice to combine each of the POCA–R and
TOCA–R subscales. Higher scores refer to higher aggression, hyperactiv-
ity, and concentration. Child’s social competence was measured with
parent and teacher versions of the BASC. The BASC contains three
subscales meant to assess child social competence. Confirmatory factor
analyses supported combining parent and teacher reports on each of the
three subscales: Leadership (
␣
⫽.70–.78), Adaptability (
␣
⫽.71–.81), and
Social Skills (
␣
⫽.80–.86). However, confirmatory factor analyses did not
support combining all into a single indicator. Higher scores on the cross-
source composites indicate higher levels of the respective indicator of
social competence.
Parenting and family relationships. Parenting characteristics were as-
sessed with the Parenting Practices Questionnaire (Gorman-Smith, Tolan,
Zelli, & Huesmann (1996). This 46-item scale measures two primary
constructs, discipline practices and monitoring, which are higher order
factors derived from five scales: (a) Discipline Effectiveness (
␣
⫽.61), (b)
Discipline Avoidance (
␣
⫽.66), (c) Positive Parenting (
␣
⫽.81), (d)
Involvement (
␣
⫽.73), and (e) Supervision (
␣
⫽.59). Each scale is scored
such that higher scores represent more effective or better parenting prac-
tices. The scales and factor structure have been validated with several
samples, including initial development with a sample from the same
neighborhoods (Gorman-Smith et al., 1996). The published scoring that
produces these higher order factor scores (discipline and monitoring) was
used in this analysis.
To assess family relationship characteristics, we used the Family Rela-
tionships Scale (Tolan, Gorman-Smith, Huesmann, & Zelli, 1997), which
yields scores for cohesion, beliefs about family, and structure (organiza-
tion). These are derived as higher order factors from a 35-item scale
measuring six aspects of family-relationship characteristics: (a) cohesion
(
␣
⫽.80), (b) beliefs about family (
␣
⫽.84), (c) deviant beliefs (
␣
⫽.71),
(d) support (
␣
⫽.63), (e) organization (
␣
⫽.53), and (f) communication
(
␣
⫽.57). Scales are scored such that higher scores mean stronger and
more functional family-relationship characteristics. The scale has been
validated with several samples, many of which are demographically equiv-
alent to this sample (see Tolan et al., 1997, for details on development and
initial psychometric validation). Reports of multiple informants may be
combined to produce cross-source scores when available. The same factor
structure has been confirmed for single sources (August, Realmuto, Hekt-
ner, & Bloomquist, 2001). Single-informant scaling of the higher order
constructs (cohesion, beliefs, and structure) was used in these analyses.
Parental involvement in child’s education was measured with the parent
and teacher forms of the Fast Track Parent Involvement Scales (Conduct
Problems Prevention Research Group, 1999). Although the item content of
each version differs somewhat, each contains the same three subscales. For
our sample there was high internal consistency for each scale: parent
version of Parent Endorsement of School (
␣
⫽.90), Parent Involvement
(
␣
⫽.68), and Quality of Relationship With the Teacher (
␣
⫽.91); teacher
version of Parent Endorsement of School (
␣
⫽.93), Parent Involvement
(
␣
⫽.76), and Quality of Relationship With the Teacher (
␣
⫽.74).
Confirmatory factor analyses showed that the three subscales composed
one higher order factor within each source but did not support cross-source
composites (whether as three cross-source scales or two higher order,
cross-source scores). Therefore, we used the within-source composites of
the three scales to form two assessed constructs: Teacher-Rated Parent
Involvement in Child’s Education and Parent Self-Rated Involvement in
Child’s Education. Higher scores on these composites indicate greater
parental involvement in school.
Covariates and moderators. We assessed and included in the analyses
the following variables that might moderate the effects of the intervention:
parents’ marital status; family’s income; and child’s gender, ethnicity, and
school at the time of random assignment. Demographic variables were
measured with a background information questionnaire administered as
part of the parent interview.
Differentiating High-Risk Families and Youth With High-
Risk Behavior
Because all families and children were included by virtue of where they
lived, this intervention is universal in one sense. However, we know from
other prevention trials that within a general population, intervention effects
may be limited to or be greater for high-risk groups (e.g., Hawkins,
Catalano, Kosterman, Abbott, & Hill, 1999; Olds et al., 1998; Reid et al.,
2002). In the present case, prior analysis suggests risk-related heterogene-
ity in parenting and family functioning and in child-externalizing behaviors
(Gorman-Smith, Tolan, Henry, Quintana, & Lutovsky, in press). We
wanted to incorporate these two approaches to risk differentiation into the
evaluation of the intervention. We applied a strategy that others have used
to focus on the impact of high-risk groups (e.g., August et al., 2001;
Stoolmiller et al., 2000). We created terms that categorize families as high
risk owing to poorer family functioning and children as high risk owing to
child-behavior problems. This approach permits analyses that compare
high risk with high risk (whether high-risk families or children) across
conditions to ascertain the impact of the intervention for those at high risk
(on the basis of family and then on behavioral differentiation), in addition
to the overall comparisons. This approach has an additional advantage of
permitting interpretation of subgroup variations in effects without having
to attempt dissembling the meaning of a statistical relation between an
outcome and multidimensional, higher order interaction terms, which in
this case are composed of continuous and categorical predictors (e.g.,
Risk ⫻Intervention Condition ⫻Time).
High-risk family designation for these analyses was coded if their pretest
scores fell at or below one standard deviation below the mean score for the
average of Waves 1 and 2 on composite scores of the scales from either the
Family Relationship Scale or the Parenting Practices Questionnaire. This
procedure resulted in 100 (23.5%) families being classified as at high
pretest risk. High risk due to child behavior designation was coded based
on a composite of aggression, hyperactivity, concentration scale scores
across Waves 1 and 2 using the TOCA–R and POCA–R as sources. If a
child was more than one standard deviation above the mean on this
composite score for the whole sample, he or she was designated as
high-risk due to behavior. This procedure resulted in 86 (20.0%) children
being classified as high risk at pretest. Overlap between the two high-risk
groups was moderate. (n⫽46 for both high-risk classifications). Approx-
imately one third of the children that were high risk due to behavior were
from families classified as high risk.
Results
Analysis Plan
Our primary interest was in testing the effects of the intervention
on growth trajectories of the intervention targets that are thought to
affect or mark risk. Accordingly, our analysis plan centered on
comparing the growth trajectories of participants assigned to in-
tervention with those assigned to the control group. We included
all participants who were randomly assigned to conditions in these
859
SUPPORTING FAMILIES
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

analyses, regardless of attrition, dosage, or missing data. This
intent-to-treat approach to analysis ensures that the randomness of
assignment to conditions is preserved (Shadish, Hu, Glaser, Kow-
nacki, & Wong, 1998) and that the comparisons reported reflect
differences in growth trajectories associated with intervention.
We used a random regression approach to growth curve analysis
to test hypotheses related to differential effects of the intervention
on growth trends in the predictor variables. Growth curve models
of multiple data points over time provide more reliable estimates
of differential change because of intervention effects than do
pre–post means comparisons (Muthe´n & Curran, 1997). In part
this is because growth curve models allow for individual variation
in measurement points within waves, such as those found in
studies such as the present study, and can model random variation
in effects (Bock, 1989; Gibbons et al., 1993). Random regression
models are a class of mixed-effects linear models well suited to
growth curve analysis of multiple repeated-measures data. Such
models have been developed by Bock (1983, 1989), Gibbons et al.
(1993), and Bryk and Raudenbush (1992). Random effects regres-
sion models assume that the available data at any given point of
measurement estimate the group growth trend and each individu-
al’s deviation from the group trend at that point of measurement.
This assumption permits valid estimates of slopes and intercepts
with cases that have missing waves of data, regardless of whether
the data are missing at random (for reasons unrelated to the
variables under study). They were developed, in part, because of
the unreliability and limited precision of focusing only on pre–post
intercept differences in evaluating intervention effects (Gibbons et
al., 1993). They have been lauded as suited to prevention because
they are sensitive to effects that emerge over time and consistent
with a developmental approach to intervention effects (Muthe´n&
Curran, 1997). The focus of the analysis and interpretation of
effects is on the difference in growth attributable to intervention,
rather than on the relative level between groups at any given point
in time.
We used 2 two-level models in these outcome analyses. The
Level-1 models predicted an outcome variable from an individual
intercept and linear and quadratic slopes terms for wave of mea-
surement (expressed in years), which was centered at the date of
the posttest assessment. The Level-1 equations also included terms
for family income and parental marital status at each wave of
measurement. Thus, parental marital status and income were
treated as time-varying covariates in these analyses. Quadratic
terms for slope were included to accurately model the shape of the
growth curves. However, we limited interpretation of intervention
effects to linear growth differences because our interest was in
group differences in the linear rate of change associated with the
intervention (see Barnes, Reifman, Farrell, & Dintcheff, 2000;
Bryk & Raudenbush, 1992). The Level-2 equations predicted the
Level-1 intercepts and slope terms by intervention condition, gen-
der, ethnicity, and the child’s school at the time of random assign-
ment to conditions.
We first fit random regression models to assess effects of the
intervention on all participants (overall effects model). We then
refit the models with terms for both risk group designations,
focusing on the interaction between risk group and assigned inter-
vention condition (high-risk family and high-risk children models).
A term representing the interaction between risk groups was in-
cluded in each of the high-risk models to account for the associ-
ation between risk groups. However, we did not enter a term for
the three-way interaction between both risk groups and interven-
tion conditions. These models allowed us to assess the effects of
the intervention on high-risk families (as defined by their family
relationships and parenting practices scores) and high-risk children
(as defined by their externalizing scores) over and above any
overall or general Level l effects. Thus, we report three sets of
results: overall analyses, high-risk families, and then high-risk
children.
The high-risk models included Level-2 terms for each risk group
designation and the interaction between that risk group and inter-
vention condition. Because a significant Intervention ⫻Risk
Group interaction might not necessarily indicate a slope difference
between intervention and control high-risk groups, we constructed
planned comparisons evaluating the differences in linear slopes by
intervention condition within the high-risk group. These compar-
isons were linear contrasts with one degree of freedom.
All of these models were intent-to-treat analyses. They included
data from all participants randomly assigned to condition, regard-
less of their level of participation in the intervention. Random error
terms were entered for individual intercepts and for linear and
quadratic slopes. Random error terms for the quadratic slopes were
deleted if initial analyses revealed no individual-level variation in
quadratic slope terms.
Overall Effects
Table 2 reports means and standard deviations by condition and
wave for all participants. Table 3 summarizes the initial compar-
isons that focused on overall effects of the intervention (the entire
intent-to-treat sample without consideration of differential impact
by initial risk status). These models were fit without terms for
high-risk children or high-risk families and without interactions
between risk groups and intervention condition. The slope coeffi-
cients and standard errors reported in the control rows represent
the linear growth trend for control participants, and thus the
expected developmental trend, in the absence of intervention of
each variable. The slope estimates and standard errors in the
intervention rows report the linear growth trends for participants
randomly assigned to the intervention. The degrees of freedom,
significance tests, and effect sizes relate to differences in linear
growth from the control condition.
3
Child’s school functioning. There was a significant slope dif-
ference for groups in academic achievement. Although the rate of
growth in reading ability was naturally steep, intervention partic-
ipants grew in overall reading ability at a more rapid rate for the
reading composite (slope effect size: d⫽0.17). Converted to
grade-equivalent scores, the intervention participants were 0.44
grade-equivalent years ahead of the control participants by the
middle of second grade (at 6-month follow-up). The intervention
3
Slope difference effect sizes are not directly equatable to the more
commonly reported intercept difference effect sizes. The former represent
relative difference in growth per unit time, whereas the latter represent
differences at a given point in time, albeit controlling for prior differences
in level. Because slope differences are measures of expectable increases in
differences over time, the same magnitude effect size does not equate to the
same impact. Typically, slope effect sizes are smaller than what might be
found in intercept effect size estimates. A modest or small slope difference
can translate to a large impact over time. We report linear estimates here,
controlling for any nonlinear effects.
860 TOLAN, GORMAN-SMITH, AND HENRY
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

Table 2
Means, Standard Deviations, and Numbers Assessed by Wave of Measurement and Condition
Outcome condition
Baseline Pretest Posttest Follow-up
M SD n M SD n M SD n M SD n
Child’s school functioning
WDRB total reading composite
Control 0.95 0.54 199 1.54 1.03 167 2.36 1.62 163 3.12 1.90 172
Treatment 1.00 0.85 220 1.20 0.85 202 2.66 1.80 200 3.51 2.49 197
Attitude toward school
Control ⫺0.36 1.12 199 0.07 1.06 169 ⫺0.10 1.02 178 0.35 1.06 185
Treatment ⫺0.37 1.02 221 ⫺0.05 0.96 212 ⫺0.14 1.04 208 0.21 1.04 213
Attitude toward teacher
Control ⫺0.14 0.92 199 0.03 1.01 169 0.01 1.03 178 0.13 1.00 185
Treatment ⫺0.12 1.02 220 ⫺0.03 0.99 212 ⫺0.07 1.04 208 0.04 1.04 213
Child behavior and social competence
Aggression
Control 0.03 1.00 196 0.01 0.96 170 0.26 1.03 165 0.04 1.20 183
Treatment 0.01 1.02 217 0.01 1.03 210 0.41 1.16 205 0.25 1.12 211
Hyperactivity
Control 0.09 0.95 197 0.04 0.94 170 0.12 0.92 165 ⫺0.05 1.03 183
Treatment 0.00 0.95 217 ⫺0.04 1.05 210 0.13 0.99 205 0.02 1.07 211
Concentration
Control 0.23 0.90 196 0.02 1.02 170 0.21 1.30 165 0.21 1.21 183
Treatment 0.31 0.88 217 ⫺0.01 0.99 210 0.07 1.31 205 0.16 1.24 211
Social skills
Control 0.33 1.06 197 ⫺0.01 0.95 167 0.18 1.03 165 0.26 1.05 183
Treatment 0.23 1.15 218 ⫺0.02 1.04 206 0.08 1.21 205 0.10 1.19 211
Leadership
Control 0.27 0.98 196 0.01 1.00 170 0.24 1.06 165 0.14 1.03 183
Treatment 0.35 0.96 218 0.00 1.00 210 0.09 1.17 205 0.09 1.13 211
Adaptability
Control 0.47 0.96 196 0.01 0.98 170 0.15 0.99 165 0.25 1.06 183
Treatment 0.37 1.03 218 ⫺0.01 1.02 210 ⫺0.03 1.09 205 0.10 1.16 211
Parenting and family relationships
Parental monitoring
Control 0.08 1.03 185 0.22 0.88 173 0.17 0.87 178 0.22 0.81 181
Treatment ⫺0.03 1.15 212 ⫺0.18 1.06 212 0.20 0.95 199 0.05 0.96 206
Parental discipline
Control 0.07 0.99 185 0.03 1.00 173 ⫺0.17 1.04 178 0.08 1.03 181
Treatment 0.22 0.97 212 ⫺0.03 1.00 212 ⫺0.27 1.05 199 ⫺0.02 1.06 206
Family cohesion
Control ⫺0.07 0.99 198 ⫺0.04 1.03 173 ⫺0.27 0.99 181 ⫺0.23 0.92 183
Treatment 0.01 0.98 220 0.04 0.98 213 ⫺0.26 0.98 206 ⫺0.25 0.93 211
Family beliefs
Control 0.00 1.28 197 0.01 1.41 173 ⫺0.53 1.37 181 ⫺0.25 1.39 183
Treatment 0.01 1.30 220 ⫺0.01 1.27 213 ⫺0.47 1.40 206 ⫺0.06 1.38 211
Family structure
Control 0.15 1.37 198 0.02 0.94 173 0.04 1.15 181 ⫺0.23 1.10 183
Treatment 0.22 1.44 220 ⫺0.02 1.05 213 ⫺0.11 1.03 206 ⫺0.17 1.07 211
Teacher-rated parent involvement in
child’s education
Control 0.61 1.06 195 0.00 1.00 195 0.53 1.12 177 0.21 1.07 190
Treatment 0.62 1.04 222 0.00 1.00 221 0.64 1.03 222 0.34 1.15 217
Parent self-rated involvement in
child’s education
Control 0.23 0.85 198 0.16 1.00 173 ⫺0.11 1.18 181 ⫺0.09 1.07 183
Treatment 0.05 1.05 220 ⫺0.13 0.98 213 ⫺0.25 1.19 206 ⫺0.13 1.10 211
Note. WDRB ⫽Woodcock Diagnostic Reading Battery.
861
SUPPORTING FAMILIES
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

Table 3
Estimates, Significance Tests, and Effect Sizes by Treatment Condition for Overall Growth
Models
Risk marker–outcome variable and
assigned condition Slope estimate SE df t Effect size
(Cohen’s d)
Child’s school functioning
WDRB total reading composite
Control 1.27 0.07
Treatment 1.53 0.07 1067 2.76** 0.17
Attitude toward school
Control 0.41 0.09
Treatment 0.35 0.08 1132 ⫺0.52 0.03
Attitude toward teacher
Control 0.13 0.09
Treatment 0.09 0.09 1131 ⫺0.25 0.01
Child behavior and social competence
Aggression
Control ⫺0.06 0.09
Treatment 0.06 0.08 1116 1.03 0.06
Hyperactivity
Control ⫺0.13 0.08
Treatment ⫺0.03 0.07 1117 0.90 0.05
Concentration
Control 0.12 0.08
Treatment 0.20 0.08 1116 0.68 0.04
Social skills
Control 0.24 0.09
Treatment 0.12 0.08 1111 ⫺0.97 0.06
Leadership
Control 0.06 0.08
Treatment 0.09 0.07 1117 0.28 0.02
Adaptability
Control 0.22 0.09
Treatment 0.15 0.08 1117 ⫺0.65 0.04
Parenting and family relationships
Parental monitoring
Control 0.03 0.08
Treatment 0.10 0.08 1107 0.67 0.04
Parental discipline
Control 0.11 0.09
Treatment 0.13 0.08 1107 0.18 0.01
Family cohesion
Control ⫺0.13 0.08
Treatment ⫺0.18 0.07 1144 ⫺0.48 0.03
Family beliefs
Control ⫺0.10 0.13
Treatment 0.11 0.12 1143 1.19 0.07
Family structure
Control 0.27 0.11
Treatment 0.17 0.10 1144 ⫺0.69 0.04
Teacher-rated parent involvement in
child’s education
Control 0.01 0.10
Treatment 0.14 0.09 1115 1.03 0.06
Parent self-rated involvement in
child’s education
Control ⫺0.17 0.09
Treatment 0.04 0.08 1144 1.77† 0.10
Note. The slope estimates for the tvalues represent the coefficient for linear change over time (in years) from
the growth model. Asterisks beside the tvalues for treatment indicate the level of significance for H
0
:
Slope
Treatment
Slope
Control
⫽0. WDRB ⫽Woodcock Diagnostic Reading Battery.
†p⬍.10. ** p⬍.01.
862 TOLAN, GORMAN-SMITH, AND HENRY
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

participants were somewhat above the national average for mid-
second-grade reading skills (grade equivalent ⫽2.9 level achieve-
ment, school month ⫽2.6), and the control participants were
slightly below the national average (grade equivalent ⫽2.4, school
month ⫽2.6).
There was no significant difference by intervention group in
these comparisons on school bonding. We find it notable that all
children reported stable, highly positive attitudes toward their
teachers and school, rendering any differentiation owing to inter-
vention very unlikely.
Child’s behavior and social competence. No significant dif-
ferences were found in the overall model for measures of a child’s
behavior or social competence.
Parenting and family relationship characteristics. Control
participants showed declining scores over time on the Parent
Self-Rated Involvement in Child’s Education, whereas interven-
tion parents maintained a stable score level over time, yielding a
significant difference (slope effect size: d⫽0.10). Overall com-
parisons did not show any significant effects for any of the other
measures of parenting and family-relationship characteristics.
Effects for High-Risk Families
Table 4 reports the linear slope estimates and standard errors for
participants assigned to control and intervention conditions and
reports significance tests and effect sizes within the high-risk
families of differences in linear growth between those assigned to
control and intervention conditions. These families (N⫽100; 45
control and 55 intervention) were defined by their scores on family
relationships and parenting practices measures. Effects found here
represent those specific to high-risk families if no significant
difference was found for the overall comparison or represent
differentiated (heightened) effects of high-risk families when a
significant difference was found in the overall comparison.
Child’s school functioning. Within the high-risk families,
there were no significant differences between treatment and con-
trol for academic achievement or school bonding. This suggests
that the effects in the overall model were not different for high-risk
families.
Child’s behavior and social competence. There was a decrease
over time in aggression among children in high-risk families in the
intervention, whereas the slope for controls was essentially flat,
yielding a significant difference (slope effect size: d⫽0.12). Also,
there was a significantly steeper, more positive slope in concen-
tration for intervention than for control participants (slope effect
size: d⫽0.13). Children’s social competence had a nonsignificant
effect, with children from high-risk families in the intervention
showing a positive linear slope on adaptability, whereas controls
remained essentially unchanged (slope effect size: d⫽0.10).
Parenting and family relationship characteristics. The inter-
vention resulted in significantly greater improvement in parental
monitoring among intervention than among control high-risk fam-
ilies (slope effect size: d⫽0.14). The effects were not different
from the overall effects for any other parenting and family-
relationship characteristic scales.
Effects for High-Risk Children
Table 5 summarizes the results related to the subgroup of
children defined as high risk because of high levels of externaliz-
ing behavior. As was the case in Table 4, Table 5 presents linear
slope estimates and standard errors for high-risk children assigned
to control and intervention conditions and presents significance
tests and effect sizes related to the differences in linear growth
between those assigned to control and intervention conditions. As
with the high-risk family comparisons, significant differences be-
tween conditions represent effects limited to high-risk children if
not found in the overall comparison or represent a heightened
effect if a significant difference was found for that variable in the
overall comparison.
Child’s school functioning. No additional significant differ-
ences were noted for academic achievement and school bonding
for high-risk children, suggesting that there is no difference in
effects from those noted in the overall model.
Child’s behavior and social competence. Of the children des-
ignated high-risk by behavior, those assigned to the intervention
had a negative slope on aggression, whereas those assigned to the
control condition had a slightly positive slope. The difference in
slopes was statistically significant (slope effect size: d⫽0.16).
Also, children in the intervention condition had a decreasing slope
for hyperactivity, whereas those in the control condition had an
increasing slope (slope effect size: d⫽0.10). Finally, high-risk
children in the treatment condition showed improved leadership
ratings, whereas controls remained essentially unchanged, yielding
a significant difference (slope effect size: d⫽0.10).
Parenting and family relationship characteristics. Among
high-risk children, families in the intervention condition showed a
slightly increasing level of Parent Self-Rated Involvement in
Child’s Education,whereas controls showed a substantial de-
crease, yielding a significant difference (slope effect size: d⫽
0.14).
Discussion
In this study, we focused on the proximal effects of a preventive
intervention that included families that lived in a high-risk set-
ting—an inner-city community. Other than residence location,
inclusion was not based on individual or family characteristics
thought to relate to risk for antisocial behavior. This is a different
approach to inclusion than that used in most prevention trials.
However, prevalence of antisocial behaviors such as violence and
delinquency is related to residence in an inner-city community.
Therefore, such a focus seems valuable if the goal is to reduce
overall prevalence of such behavior (Sampson, 1997).
This study is also different from many other prevention trials by
its focus on facilitating success during a common but challenging
developmental transition, occurring under uncommonly challeng-
ing conditions (Cicchetti & Toth, 1992). More commonly, preven-
tion efforts are meant to build or remediate risk characteristics that
differentiate a given individual or family within a population (Coie
et al., 1993; Conduct Problems Prevention Research Group, 1999)
or reduce risk-related behaviors within a given population (Ialongo
et al., 1999; Stoolmiller et al., 2000). Here, the intent was to
determine whether a family-focused intervention with academic
support for the child could lead to normative academic achieve-
ment and adequate child-behavioral functioning while aiding par-
enting skills and family characteristics to better manage the tasks
that accompany the transition into elementary school.
The results suggest that the intervention has some benefits for
the overall sample. The general effects seem limited to supporting
863
SUPPORTING FAMILIES
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

Table 4
Estimates, Significance Tests, and Effect Sizes by Treatment Condition for High-Risk Families
Risk marker–outcome variable and
assigned condition Slope estimate SE df t Effect size
(Cohen’s d)
Child’s school functioning
WDRB total reading composite
Control 1.16 0.16
Treatment 1.48 0.15 1057 1.47 0.09
Attitude toward school
Control 0.29 0.21
Treatment 0.49 0.21 1122 0.74 0.04
Attitude toward teacher
Control 0.13 0.22
Treatment 0.13 0.21 1121 0.00 0.00
Child behavior and social competence
Aggression
Control ⫺0.03 0.19
Treatment ⫺0.55 0.19 1106 ⫺2.02* 0.12
Hyperactivity
Control 0.13 0.19
Treatment ⫺0.16 0.18 1107 ⫺1.19 0.07
Concentration
Control 0.16 0.19
Treatment 0.68 0.18 1106 2.08* 0.13
Social skills
Control 0.17 0.21
Treatment 0.41 0.20 1101 0.88 0.05
Leadership
Control 0.11 0.18
Treatment 0.41 0.18 1107 1.24 0.07
Adaptability
Control 0.05 0.20
Treatment 0.49 0.19 1107 1.70† 0.10
Parenting and family relationships
Parental monitoring
Control 0.14 0.19
Treatment 0.70 0.18 1097 2.26* 0.14
Parental discipline
Control 0.49 0.21
Treatment 0.28 0.20 1097 ⫺0.77 0.05
Family cohesion
Control 0.08 0.18
Treatment ⫺0.08 0.18 1134 ⫺0.67 0.04
Family beliefs
Control ⫺0.07 0.29
Treatment ⫺0.09 0.28 1133 ⫺0.05 0.00
Family structure
Control 0.54 0.23
Treatment 0.98 0.23 1134 1.41 0.08
Teacher-rated parent involvement in
child’s education
Control 0.05 0.23
Treatment 0.43 0.22 1105 1.27 0.08
Parent self-rated involvement in
child’s education
Control ⫺0.27 0.20
Treatment 0.14 0.20 1134 1.58 0.19
Note. The slope estimates for the tvalues represent the coefficient for linear change over time (in years) from
the growth model. Asterisks beside the tvalues for treatment indicate the level of significance for H
0
:
Slope
Treatment
Slope
Control
⫽0. WDRB ⫽Woodcock Diagnostic Reading Battery.
†p⬍.10. * p⬍.05.
864 TOLAN, GORMAN-SMITH, AND HENRY
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

Table 5
Estimates, Significance Tests, and Effect Sizes by Treatment Condition for High-Risk Children
(Because of Behavior)
Risk marker–outcome variable and
assigned condition Slope estimate SE df t Effect size
(Cohen’s d)
Child’s school functioning
WDRB total reading composite
Control 1.19 0.17
Treatment 1.28 0.17 1057 0.38 0.02
Attitude toward school
Control 0.43 0.22
Treatment 0.35 0.23 1122 ⫺0.25 0.01
Attitude toward teacher
Control 0.01 0.23
Treatment 0.22 0.24 1121 0.64 0.04
Child behavior and social competence
Aggression
Control 0.09 0.20
Treatment ⫺0.66 0.21 1106 ⫺2.64* 0.16
Hyperactivity
Control 0.18 0.19
Treatment ⫺0.29 0.20 1107 ⫺1.72† 0.10
Concentration
Control 0.23 0.20
Treatment 0.65 0.21 1106 1.54 0.09
Social skills
Control 0.31 0.22
Treatment 0.32 0.23 1101 0.04 0.00
Leadership
Control ⫺0.03 0.19
Treatment 0.42 0.20 1107 1.67† 0.10
Adaptability
Control 0.06 0.21
Treatment 0.36 0.21 1107 1.05 0.06
Parenting and family relationships
Parental monitoring
Control 0.04 0.20
Treatment 0.30 0.20 1097 0.95 0.06
Parental discipline
Control 0.55 0.22
Treatment 0.27 0.22 1097 ⫺0.92 0.06
Family cohesion
Control 0.05 0.19
Treatment ⫺0.13 0.20 1134 ⫺0.69 0.04
Family beliefs
Control ⫺0.41 0.31
Treatment ⫺0.04 0.32 1133 0.86 0.05
Family structure
Control 0.35 0.25
Treatment 0.71 0.26 1134 1.02 0.06
Teacher-rated parent involvement in
child’s education
Control ⫺0.06 0.24
Treatment 0.31 0.25 1105 1.09 0.07
Parent self-rated involvement in
child’s education
Control ⫺0.58 0.21
Treatment 0.14 0.22 1134 2.43* 0.14
Note. The slope estimates for the tvalues represent the coefficient for linear change over time (in years) from
the growth model. Asterisks beside the tvalues for treatment indicate the level of significance for H
0
:
Slope
Treatment
Slope
Control
⫽0. WDRB ⫽Woodcock Diagnostic Reading Battery.
†p⬍.10. * p⬍.05.
865
SUPPORTING FAMILIES
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

adequate gains in academic skill and maintaining initial levels of
parental involvement with school. Intervention families remained
engaged in school, whereas the norm was to disengage, and their
children progressed academically at a rate comparable with the
national rate, whereas the norm was to drop farther behind national
standards. Thus, the general effect seems to be of promoting a
positive and effective start to school in these high-risk communi-
ties. As such a strong start has often been cited as one of the most
important protective factors against antisocial behaviors and
school failure, this overall effect seems valuable (Moffitt, 1993).
In addition to these overall effects, there are additional preven-
tive benefits for high-risk families (those with less adequate par-
enting skills and family-relationship quality at pretest) and for
high-risk children (with high levels of problem behaviors at pre-
test). High-risk families in the intervention evidenced improve-
ment in the critical skill of monitoring, and their children showed
improvements in behavior and social competence. For children
considered high-risk because of their behavior, the intervention
seems to help with several aspects of behavior and to aid social
competence, in addition to the overall effects found. These in-
cluded lessening aggression and hyperactivity, whereas the control
participants’ level increased. Thus, in addition to aiding children
by promoting a good start in school, the intervention seems to also
specifically benefit parents in need of skills and to benefit children
with risky behavior who would otherwise be on a trajectory of
increasing aggression and hyperactivity. In addition to reducing
risk behaviors among these children, the intervention also seems to
improve characteristics related to social competence.
Given the moderate overlap of membership in the high-risk
groups, the analyses suggest this general approach may beneficial
for those at high risk, even if there is a different basis for such risk.
In both cases, the effects are more extensive for developmental
influences related to the reason for risk designation but are not
limited only to those characteristics. It may be that the intervention
has benefits that extend to child behavior for families with parent-
ing risk through improving parenting, or it could be that improving
child behavior facilitates use and consolidation of intervention
resources and skills for parenting. Similarly, for children at risk
because of behavior, improving parenting may be a benefit of child
improvement or may facilitate such change. It may also be that the
effects on parenting and behavior of both groups are due to some
overlap in membership. Unfortunately, the sample size with such
overlap and the complexity of interpreting further levels of inter-
action preclude more specific analyses at this point. However, it
does seem that the intervention affects specific risk characteristics
even when offered as nonspecific.
The behavior effects may be important because aggression,
concentration problems, and low social competence are among the
strongest predictors of later risk (Moffitt, 1993; Tolan & Guerra,
1994). Social competence gains have been shown to reduce ag-
gression in two similar studies; the patterns uncovered here sug-
gest that promoting social competence may be valuable in coping
with risk-related stress in the inner city (MACS Research Group,
2002b; Tolan et al., 2002). Similarly, improving parenting and
family-relationship skills has well-documented benefits for reduc-
ing later risk (Hawkins et al., 1992; Tolan & Guerra, 1994). Aiding
high-risk youth at an early stage while improving parenting skills
of parents with less capability may promote resilience in the face
of environmental threats (Luthar, Cicchetti, & Becker, 2000).
Thus, although the impact of the program has complexity, it
might be summarized overall as aiding normal development and
functioning in a setting in which development is often compro-
mised with additional effects specific to high-risk children and
families among the inner-city population (Lochman, Wells, &
Murray, in press). This pattern of results suggests that there is
merit in a universally applied, family-focused intervention in high-
risk communities. The results are consistent with studies of other
efficacious prevention efforts that promoted sound family func-
tioning and appropriate and normative developmental accomplish-
ments, albeit not in high-risk communities per se (Hawkins et al.,
1999; Redmond et al., 1999). However, the current set of analyses
cannot indicate whether the emphasized processes are the avenues
through which the benefits were attained. In addition, significant
effects were not found for all comparisons, and for the overall
comparison there were limited effects.
Although the findings point to the potential value of interven-
tions that support normative functioning during important devel-
opmental transitions in high-risk settings, the findings are less
clear on the value of selective versus universal inclusion (Aos,
Phipps, Barnoski, & Lieb, 2001). It could be argued that because
effects were found across more areas for those with high-risk
characteristics, concentrating on this portion of the population is
warranted. However, this would seem to be dismissing the overall
benefits that promote academic progression and parental involve-
ment in school when the pattern otherwise is just the opposite.
These effects extended to these high-risk families and youth as
well. Also, this design leaves unclear how excluding the lower-risk
families might affect the processes and effects of the family
groups. For example, it may be that the high-risk families were
able to compare their expectations with those of better functioning
families and, therefore, were supported in their own skill devel-
opment. Such groups are also thought to facilitate connection
between neighbors, which should improve overall monitoring of
children and reduce risk. With selective inclusion, many of the
most motivated and capable parents may be excluded in such
network development.
Although many effects were concentrated in the high-risk sam-
ple, the general and specific effects were not negligible. Effect
sizes calculated were for linear slope differences controlling for
any quadratic or cubic effects, whereas more typically reported
intervention effect sizes are based on intercept difference. In other
words, the effects were tested as the relative difference in linear
gain per unit of time after any nonlinear tendencies were con-
trolled, whereas more typically the effect sizes represent a set level
of difference attributable to the intervention at the point of last
measurement. Slope effect sizes are typically smaller in absolute
value than intercept differences (Barnes et al., 2000). Slope effect
sizes of the magnitude found here suggest increasing differences
over time that may lead to substantial differences, even if the
actual growth patterns reach an asymptote at some point. Such an
expectation would be consistent with the findings of a first-grade
preventive intervention trial conducted by Ialongo and colleagues
(Ialongo et al., 1999; Ialongo, Poduska, Werthamer, & Kellam,
2001). Analyzing proximal effects of a classroom-enhancement
intervention, Ialongo et al. (1999) found an effect size equivalent
to a slope of 0.14 standard deviations per year, an effect similar in
magnitude to those obtained in this study. This slope difference led
to greater differences between groups as time passed, such that at
5 years postintervention there was a quite substantially lower
866 TOLAN, GORMAN-SMITH, AND HENRY
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

portion of the intervention children meeting diagnostic criteria for
conduct disorder than among controls. Of course, whether the
initial effects found here continue, grow, or diminish can only be
determined by follow-up analyses. Such data collection is under-
way and will be the focus of ensuing studies.
Although the immediate results are promising, the overall im-
pact on parenting and family-relationship characteristics was rel-
atively limited. The slope differences for all scales favored the
intervention, although not statistically significant by comparison.
It may be that the skills gained will lead to differences that are
more substantial over time as they are consolidated in patterns of
family interaction. It may also be that the intervention needs to be
bolstered in regard to important skills, such as discipline practices,
and family characteristics, such as organization (Forgatch & De-
Garmo, 1999). It may also be that the measures applied are not as
sensitive to changes in family practices as other methods, such as
observation (Stoolmiller et al., 2000).
It could be, however, that the results point to the limitation, as
a general intervention, of focusing on parenting skills and reme-
diating deficits in families in high-risk communities. It may be that
these families need more support that promotes their capabilities
rather than skill building, particularly if they are already function-
ing adequately. A study with a similar sample suggested that
family functioning levels do not differentiate across urban poor
and inner-city communities (Gorman-Smith et al., 1998). More
emphasis on aiding family functioning rather than attempting to
improve skills and practices may be useful. Although this inter-
vention included such emphasis, unfortunately one weakness of
the study is that our measurement package did not include assess-
ment of family skills, interfamily support, and use of support
resources.
This study can also contribute to the literature on risk and
prevention because the growth curve measurement and analysis
approach can shed some illumination on developmental processes
in high-risk communities. For example, the quick and substantial
drop in parental involvement in school that characterized the
control condition might not have been noted if this had not been
traced over multiple data points and school years. Similarly, the
pattern of a substantial increase in absolute reading ability, while
there is a widening disparity in relative ability compared with
national norms, that was found among nonintervention families
might not have been identified if a growth curve modeling ap-
proach had not been used.
In addition, the growth modeling approach may help in under-
standing how prevention related to development can vary in the
nature of the beneficial effect. In some instances, the preventive
effect is to enhance ongoing positive growth. In other cases, the
preventive effect is to induce growth in which there is stability
otherwise, to maintain functioning when the norm is to have
declining functioning, and to lessen the rate of decline in func-
tioning. Although these few data points over a short a period of
time cannot specify which pattern ought to be assumed, the vari-
ations identified here illustrate the importance of a growth ap-
proach and a developmental framework (Kellam & Rebok, 1992;
McArdle & Epstein, 1987).
Overall, the results suggest a promising intervention that ap-
pears to benefit families and children living in high-risk commu-
nities, with more specific and extensive benefits for high-risk
families and children within those communities. The intent of this
intervention was to support healthy development and effective
parenting during an important, and relatively early, stage of de-
velopment to buffer risk related to living in impoverished, crime-
ridden, and often isolating communities. These are proximal ef-
fects that are limited to 6 months after intervention, and the
ultimate outcomes of interest are not evident until mid-to-late
adolescence (e.g., school dropout, drug abuse, delinquency); there-
fore, the ultimate impact must await determination.
References
Aos, S., Phipps, P., Barnoski, R., & Lieb, R. (2001). The comparative costs
and benefits for programs to reduce crime. Olympia: Washington State
Institute for Public Policy.
August, G. J., Realmuto, G. R., Hektner, J. M., & Bloomquist, M. L.
(2001). An integrated components preventive intervention for aggressive
elementary school children: The Early Risers program. Journal of Con-
sulting and Clinical Psychology, 69, 614–626.
Barnes, G. M., Reifman, A. S., Farrell, M. P., & Dintcheff, B. A. (2000).
The effects of parenting on the development of adolescent alcohol
misuse: A six-wave latent growth model. Journal of Marriage & the
Family, 62, 175–186.
Bock, R. D. (1983). Within-subject experimentation in psychiatric re-
search. In R. D. Gibbons & M. W. Dysken (Eds.), Statistical and
methodological advances in psychiatric research (pp. 59–90). New
York: Spectrum Books.
Bock, R. D. (1989). Measurement of human variation: A two-stage model.
In R. D. Bock (Ed.), Multilevel analysis of educational data (pp. 319–
342). Orlando, FL: Academic Press.
Brooks-Gunn, J., Duncan, G. J., Klebenov, P. K., & Sealand, N. S. (1993).
Do neighborhoods influence child and adolescent development? Amer-
ican Journal of Sociology, 99, 353–395.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models:
Applications and data analysis methods. Newbury Park, CA: Sage.
Cicchetti, D., & Toth, S. L. (1992). The role of developmental theory in
prevention and intervention. Development and Psychopathology, 4,
489–493.
Coie, J. D., & Dodge, K. A. (1998). Aggression and antisocial behavior. In
W. Damon (Ed.), Handbook of child psychology (pp. 779–862). New
York: Wiley.
Coie, J. D., Watt, N. F., West, S. G., Hawkins, J. D., Asarnow, J. R.,
Markman, H. J., et al. (1993). The science of prevention: A conceptual
framework and some directions for a national research program. Amer-
ican Psychologist, 48, 1013–1022.
Conduct Problems Prevention Research Group. (1999). Initial impact of
the fast track prevention trial for conduct problems: I. The high-risk
sample. Journal of Consulting and Clinical Psychology, 67, 631–647.
Durlak, J. A., & Wells, A. M. (1997). Primary prevention mental health
programs for children and adolescents: A meta-analytic review. Ameri-
can Journal of Community Psychology, 25, 115–152.
Elliott, D., & Tolan, P. H. (1998). Youth violence, prevention, intervention,
and social policy: An overview. In D. Flannery & R. Hoff (Eds.), Youth
violence: A volume in the psychiatric clinics of North America (pp.
3–46). Washington, DC: American Psychiatric Association.
Entwisle, D. R., & Alexander, K. L. (1998). Facilitating the transition to
first grade: The nature of transition and research on factors affecting it.
Elementary School Journal, 98, 351–364.
Entwisle, D. R., Alexander, K. L., Pallas, A. M., & Cadigan, D. (1988). A
social psychological model of the schooling process over first grade.
Social Psychology Quarterly, 51, 173–189.
Farrington, D., & Welsh, B. (1999). Delinquency prevention using family-
based interventions. Children and Society, 13, 287–303.
Forgatch, M. S., & DeGarmo, D. S. (1999). Parenting through change: An
effective parenting training program for single mothers. Journal of
Consulting and Clinical Psychology, 67, 711–724.
Furstenberg, F. F., & Hughes, M. E. (1995). Social capital and successful
867
SUPPORTING FAMILIES
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

development among at-risk youth. Journal of Marriage & the Family,
57, 580–592.
Gibbons, R. D., Hedeker, D., Elkin, I., Waternaux, C., Kraemer, H.,
Greenhouse, J. B., et al. (1993). Some conceptual and statistical issues in
analysis of longitudinal psychiatric data. Archives of General Psychia-
try, 50, 739–750.
Gorman-Smith, D., Tolan, P. H., & Henry, D. (1998). The relation of
community and family to risk among urban-poor adolescents. In P.
Cohen, C. Slomkowski, & L. K. Robins (Eds.), Historical and geo-
graphical influences on psychopathology (pp. 349–367). Mahwah, NJ:
Erlbaum.
Gorman-Smith, D., Tolan, P. H., & Henry, D. B. (2000). A developmental–
ecological model of the relation of family functioning to patterns of
delinquency. Journal of Quantitative Criminology, 16, 169–198.
Gorman-Smith, D., Tolan, P. H., Henry, D., Quintana, E., & Lutovsky, K.
(in press). The SAFEChildren Prevention Program. In P. Tolan, J.
Szapocznik, & S. Sombrano (Eds.), Developmental approaches to pre-
vention of substance abuse and related problems. Washington, DC:
American Psychological Association.
Gorman-Smith, D., Tolan, P. H., Zelli, A., & Huesmann, L. R. (1996). The
relation of family functioning to violence among inner-city minority
youth. Journal of Family Psychology, 10, 115–129.
Hawkins, J. D., Catalano, R. F., Kosterman, R., Abbott, R., & Hill, K. G.
(1999). Preventing adolescent health-risk behaviors by strengthening
protection during childhood. Archives of Pediatrics and Adolescent
Medicine, 153, 226–324.
Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective
factors for alcohol and other drug problems in adolescence and early
adulthood: Implications for substance abuse prevention. Psychological
Bulletin, 112, 64–105.
Henggeler, S. W., Melton, G. B., & Smith, L. A. (1992). Family preser-
vation using multisystemic therapy: An effective alternative to incarcer-
ating serious juvenile offenders. Journal of Consulting and Clinical
Psychology, 60, 953–961.
Ialongo, N., Poduska, J., Werthamer, L., & Kellam, S. (2001). The distal
impact of two first-grade preventive interventions on conduct problems
and disorder in early adolescence. Journal of Emotional & Behavioral
Disorders, 9, 146–160.
Ialongo, N., Werthamer, L., Kellam, S. G., Brown, C. H., Wang, S., & Lin,
Y. (1999). Proximal impact of two first-grade preventive interventions
on the early risk behaviors for later substance abuse, depression, and
antisocial behavior. American Journal of Community Psychology, 27,
599–641.
Kellam, S. G., Brown, C. H., Rubin, B. R., & Ensminger, M. E. (1983).
Paths leading to teenage psychiatric symptoms and substance use: De-
velopmental epidemiological studies in Woodlawn. In S. B. Guze, F. J.
Earls, & J. E. Barrett (Eds.), Childhood psychopathology and develop-
ment (pp. 17–47). Chicago: University of Chicago Press.
Kellam, S. G., & Rebok, G. W. (1992). Building developmental and
etiological theory thorough epidemiologically based preventive inter-
vention trials. In J. McCord & R. Tremblay (Eds.), Preventing antisocial
behavior: Interventions from birth through adolescence (pp. 162–195).
New York: Guilford Press.
Lipsey, M. W., & Wilson, D. B. (1998). Effective intervention for serious
juvenile offenders: A synthesis of research. In R. Loeber & D. Far-
rington (Eds.), Serious and violent juvenile offenders: Risk factors and
successful interventions (pp 313–341). Thousand Oaks, CA: Sage.
Lochman, J. E., Wells, K. C., & Murray, M. (in press). The Coping Power
Program: Preventive intervention at the middle school transition. In P. H.
Tolan, J. Szapocznik, & S. Sambrano (Eds.), Developmental approaches
to preventing substance abuse, ages 3 to 14. Washington, DC: American
Psychological Association.
Loeber, R., & Farrington, D. P. (1998). Serious and violent juvenile
offenders: Risk factors and successful interventions. Thousand Oaks,
CA: Sage.
Luthar, S. S., Cicchetti, D., & Becker, B. (2000). The construct of resil-
ience: A critical evaluation and guidelines for future work. Child De-
velopment, 71, 543–562.
Mason, C. A., Cauce, A. M., Gonzales, N., & Hiraga, Y. (1996). Neither
too sweet nor too sour: Problem peers, maternal control, and problem
behavior in African American adolescents. Child Development, 67,
2115–2130.
Mayer, G. R. (1995). Preventing antisocial behavior in the schools. Journal
of Applied Behavior Analysis, 28, 467–478.
McArdle, J. J., & Epstein, D. (1987). Latent growth curves within devel-
opmental structural equation models. Child Development, 58, 110–133.
Metropolitan Area Child Study Research Group. (2002a). A cognitive–
ecological approach to preventing aggression in urban settings: Initial
outcomes for high-risk children. Journal of Consulting and Clinical
Psychology, 70, 179–194.
Metropolitan Area Child Study Research Group. (2002b). Social cognitive
effects of a prevention program for elementary school urban youth.
Unpublished manuscript, University of Illinois at Chicago.
Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent anti-
social behavior: A developmental taxonomy. Psychological Review,
100, 674–701.
Muthe´n, B. O., & Curran, P. J. (1997). General longitudinal modeling of
individual differences in experimental designs: A latent variable frame-
work for analysis and power estimation. Psychological Methods, 2,
371–402.
Olds, D., Henderson, C. R., Cole, R., Eckenrode, J., Kitzman, H., Luckey,
D., et al. (1998). Long-term effects of nurse home visitation on chil-
dren’s criminal and antisocial behavior: 15-year follow-up of a random-
ized controlled trial. Journal of the American Medical Association, 280,
1238–1244.
Redmond, C., Spoth, R., Shin, C., & Lepper, H. S. (1999). Modeling
long-term parent outcomes of two universal family-focused preventive
interactions: One-year follow-up results. Journal of Consulting and
Clinical Psychology, 67, 975–984.
Reid, J. B., Patterson, G. R., Dishion, T., & Snyder, J. (2002). Antisocial
behavior in children and adolescents: A developmental analysis and
model for intervention. Washington, DC: American Psychological
Association.
Reynolds, C. R., & Kamphaus, R. W. (1998). BASC: Behavioral Assess-
ment System for Children—Manual. Circle Pines, MN: American Guid-
ance Service.
Roeser, R. W., & Eccles, J. S. (1999). Academic functioning and mental
health in adolescence: Patterns, progressions, and routes from childhood.
Journal of Adolescent Research, 14, 135–174.
Sampson, R. J. (1997). The embeddedness of child and adolescent devel-
opment: A community-level perspective on urban violence. In J. Mc-
Cord (Ed.), Violence and childhood in the inner city (pp. 31–77).
Cambridge, MA: Cambridge University Press.
Sampson, R. J., Raudenbush, S., & Earls, F. (1997). Neighborhoods and
violent crime: A multilevel study of collective efficacy. Science, 277,
918–924.
Shadish, W. R., Hu, X., Glaser, R. R., Kownacki, R., & Wong, S. (1998).
A method for exploring the effects of attrition in randomized experi-
ments with dichotomous outcomes. Psychological Methods, 3, 3–22.
Spoth, R., Lopez Reyes, M., Redmond, C., & Shin, C. (1999). Assessing a
public health approach to delay onset and progression of adolescent
substance use: Latent transition and log-linear analyses of longitudinal
family preventive intervention outcomes. Journal of Consulting and
Clinical Psychology, 67, 619–630.
Stoolmiller, M., Eddy, J. M., & Reid, J. B. (2000). Detecting and describ-
ing preventive intervention effects in a universal school-based random-
ized trial targeting delinquent and violent behavior. Journal of Consult-
ing and Clinical Psychology, 68, 296–306.
Tolan, P. H., & Gorman-Smith, D. (1997) . Families and the development
of urban children. In H. J. Walberg, O. Reyes, & R. P. Weissberg (Eds.),
868 TOLAN, GORMAN-SMITH, AND HENRY
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.

Urban children and youth: Interdisciplinary perspectives on policies
and programs (pp. 67–91). Thousand Oaks, CA: Sage.
Tolan, P. H., Gorman-Smith, D., & Henry, D. (2003). The developmental
ecology of urban males’ youth violence. Developmental Psychology, 39,
274–291.
Tolan, P. H., Gorman-Smith, D., Huesmann, L. R., & Zelli, A. (1997).
Assessment of family relationship characteristics: A measure to explain
risk for antisocial behavior and depression in youth. Psychological
Assessment, 9, 212–223.
Tolan, P. H., & Guerra, N. G. (1994). What works in reducing adolescent
violence: An empirical review of the field. Boulder: University of Col-
orado, Center for the Study and Prevention of Youth Violence.
Tolan, P. H., Guerra, N. G., & Kendall, P. C. (1995). A developmental–
ecological perspective on antisocial behavior in children and adoles-
cents: Toward a unified risk and intervention framework. Journal of
Consulting and Clinical Psychology, 63, 579–584.
Tolan, P. H., Hanish, L., McKay, M., & Dickey, M. (2002). Evaluating
process in child and family interventions: Aggression prevention as an
example. Journal of Family Psychology, 16, 220–236.
Tolan, P. H., & Henry, D. (1996). Patterns of psychopathology among
urban poor children: Comorbidity and aggression effects. Journal of
Consulting and Clinical Psychology, 64, 1094–1099.
Tolan, P. H., Sherrod, L., Gorman-Smith, D., & Henry, D. (2004). Building
protection, support, and opportunity for inner youth and their families. In
K. Maton, B. Ledheather, & A. Solarz (Eds.), Positive youth develop-
ment: Research and policy (pp. 235–260). Washington, DC: American
Psychological Association.
U.S. Census Bureau. (1992). Census of population and housing, 1990:
Summary Tape File 3 on CD-ROM (machine-readable files). Washing-
ton, DC: Author.
U.S. Department of Health and Human Services. (2001). Youth violence: A
report of the Surgeon General. Washington, DC: Government Printing
Office.
U.S. Federal Bureau of Investigation. (2004). Uniform crime reports for
the United States, 1995 to present (GOVT. DOCS. J1.14/7). Washing-
ton, DC: Author.
Van Acker, R., Grant, S. G., & Henry, D. (1996) Teacher and student
behavior as a function of risk for aggression. Education and Treatment
of Children, 19, 316–334.
Wilson, W. J. (1987). The truly disadvantaged: The inner city, the under-
class, and public policy. Chicago: University of Chicago Press.
Woodcock, R. W. (1997). Woodcock Diagnostic Reading Battery: Exam-
iner’s manual. Itasca, IL: Riverside Publishing.
Received March 11, 2003
Revision received January 17, 2004
Accepted January 17, 2004 䡲
Instructions to Authors
For Instructions to Authors, please consult the February 2004 issue of the volume or
visit www.apa.org/journals/ccp and click on Submission Guidelines.
869
SUPPORTING FAMILIES
This
document
is
copyrighted
by
the
American
Psychological
Association
or
one
of
its
allied
publishers.
This
article
is
intended
solely
for
the
personal
use
of
the
individual
user
and
is
not
to
be
disseminated
broadly.