Gerard, Buehler - Multiple Risk Factors in The Family Environment and Youth Problem Behaviors

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Multiple Risk Factors in the Family Environment and Youth Problem Behaviors

Author(s): Jean M. Gerard and Cheryl Buehler


Source: Journal of Marriage and Family, Vol. 61, No. 2 (May, 1999), pp. 343-361
Published by: National Council on Family Relations
Stable URL: http://www.jstor.org/stable/353753 .
Accessed: 15/01/2015 22:00

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

National Council on Family Relations is collaborating with JSTOR to digitize, preserve and extend access to
Journal of Marriage and Family.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
JEAN M. GERARD AND CHERYL BUEHLER
University of Tennessee

Multiple Risk Factors in the Family Environmentand


Youth Problem Behaviors

We test three risk models (independent-additive, underscore the importance of particularlife events
interactive, and exponential) to examine how mul- and circumstances that predispose youth to adjust-
tiple riskfactors in the family environment-overt ment problems. For the most part, this research
interparental conflict, poor parenting, and eco- focuses on the potential effects of a single life
nomic hardship-operate conjointly to predict stressor (e.g., parental divorce). However, a num-
youth problem behaviors. The sample includes ber of researchershave broadened their outlook by
335 preadolescent and early adolescent youth. exploring how multiple risk factors affect youth
Findings from this study support the pattern of in- (e.g., Farrington,1991; Masten et al., 1988; Rutter,
dependent, additive effects of individual family 1979; Shaw & Emery, 1988; Werner & Smith,
stressors. Wefound no support for the idea that 1992; Williams, Anderson, McGee, & Silva,
the effects of poor parenting, overt interparental 1990). From this research, it is evident that multi-
conflict, and family economic hardship exacer- ple, stressful life events pose a greater threat to
bate one another, nor did the converse serve as children's long-term psychological well-being than
buffers. The independent, additive model explains does a single life stressor.
more variance in externalizing problem behavior Although the study of multiple risk factors en-
for youth in nondivorced, two-parent households. hances our understandingof the etiological roots of
Poor parenting is the strongest risk factor. Eco- youth maladjustment, this line of research suffers
nomic hardship is the only significant risk factor limitations with respect to issues of conceptualiza-
for youth internalizing problem behavior tion, measurement, and modeling. Assuming that
many recent stressful experiences generally are un-
favorable for a child's psychological well-being,
The identification of specific factors that place
risk often is conceptualized in terms of the amount
youth at risk for behavioral and emotional prob- or number of stressful events recently encountered
lems has concerned social scientists for years. Re-
search efforts have culminated in findings that by youth (Johnson, 1982). Risk frequently is as-
sessed using the life-events method, an approach
that involves summing the number of self-reported
stressful life experiences recently encountered to
obtain an overall index of cumulative life stress.
Departmentof Child and Family Studies,Universityof Ten- As useful as this approach is in determining the
nessee, 1215WestCumberlandAvenue,Room 115, Knoxville,
TN 37996 (jmgerard@utk.edu). amount of stress children experience at a given
time, it has several shortcomings. First, inventories
Key Words: economic hardship, interparental conflict, parent- of life events yield little information about the rel-
ing, risk, youth problem behavior. ative importance of discrete events. Second, when

Journal of Marriage and the Family 61 (May 1999): 343-361 343

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
344 Journal of Marriage and the Family

stressors are summed, interactions among specific relationships(Farington, 1991; Kovacs et al., 1984;
stressor events cannot be examined. Third, as a Maughan & Rutter, 1998; Pine, Cohen, Gurley,
function of the way they are constructed, life-event Brook, & Ma, 1998; Robins & McEvoy; 1990).
inventories assume that negative life stressors The focus of this investigation is ongoing risk
operate in a linear, additive fashion to increase to youth that occurs in the familial environment.
youths' risk of maladjustment. Although this as- Scholars generally agree that much of children's
sumption may not be correct, few researchershave socialization takes place in the context of the fam-
tested for nonlinear effects. (See Masten et al., ily, for it is through the family that the developing
1988, for an exception.) child is afforded the opportunities and experiences
An alternativeapproachto the study of multiple necessary to acquire the fundamental skills, behav-
risk factors is to examine the separate but cumula- iors, values, and knowledge that enable him or her
tive effects of individual life stressors (Luthar, to engage successfully in social relationships
1991). This method, unlike the life-events approach, (Maccoby, 1992). As such, risk researchers often
offers the advantage of exploring more complex look at the family for potential sources of stress to
patternsof risk, such as interactive and exponential youths' development.
models. Its value also becomes evident when more In conceptualizing the family environment as
than one index of youth adjustment is assessed. an arena for the transmission of risk, we draw
Research has shown that different statisticalmodels from two theoretical strands of literature.The first
may be necessary to describe the association be- is stress-coping research, which provides a frame-
tween a particularset of risk factors and different work for understanding how youth cope when
aspects of youth adjustment (Hodges, Tierey, & their immediate environment places high demands
Buchsbaum, 1984; Masten et al., 1988). on them. According to existing formulations of the
This study explores the relationship between stress process, ongoing strains have the potential
multiple risk exposure and youth maladjustment. to generate new forms of strains or exacerbate ex-
Its central goal is to determine how particularrisk isting ones. These strains may lead to the depletion
factors operate in conjunction with one another. of positive psychological resources or, alternately,
Youth maladjustmentis defined as the relative in- an overreliance on previously established negative
ability of youth to engage successfully and appro- coping strategies (Boss, 1988; Pearlin, Lieberman,
priately in interpersonal relationships over time Menaghan, & Mullan, 1981). Taking a develop-
with relative freedom from noxious social behav- mental perspective to coping with stress, Aneshen-
iors and burdensome emotions (Trotter, 1989). We sel and Gore (1991) contend that high levels of
examine youth externalizing and internalizing stress may pose a unique challenge to youth be-
problem behavior in this study because they are cause they are less skilled than adults in handling
two central indicators of maladjustment. The ex- the day-to-day struggles of life. Given a limited
ternalizing aspect of maladjustment is conceived repertoireof available coping mechanisms and im-
as outer-directed behavior (e.g., aggression, lying, mature cognitive abilities to manage the meaning
stealing) that functions maladaptively in society of stress, youth may not fare well in the face of
by producing distress in others, whereas the inter- persistent adversity. Consequently, children may
nalizing aspect is viewed as inner-directed behav- cope by aggressing against others or by internaliz-
ior (e.g., excessive fear, anxiety, depression) that ing their distress (Honig, 1986).
functions maladaptively by causing emotional dis- The second body of literature we draw from is
tress to self (Reynolds, 1992). As developmental children's risk and resiliency research, which has
outcomes, externalizing and internalizing problem demonstrated individual variation in response to
behaviors warrant empirical consideration. These high levels of stress (Garmezy, 1981; Rutter, 1979).
forms of maladjustment are frequently reasons for Researchers in this field are primarily concerned
referring youth to mental health facilities (Bor- with identification of factors that buffer youth
duin, Henggeler, & Manley, 1995; Kazdin, 1995; from high levels of environmental harm and fac-
Reynolds, 1992). Moreover, evidence of problem tors that magnify this risk. Thus, emphasis is placed
behaviors during early adolescence may have im- on the statistical interaction between individual
plications for adjustment in adulthood. Longitudi- characteristics and environmental factors, rather
nal investigationshave demonstrateda link between than their independent main effects. Risk and re-
early problem behaviors and impaired adult func- siliency variables that are frequently examined in-
tioning, including poor mental health outcomes, clude children's coping ability (e.g., Wills, Vaccaro,
substance abuse, and difficulty managing social & Benson, 1995), social support (e.g., Cohen &

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
Family Risk Factors 345

Wills, 1985), economic status (Duncan, Brooks- Independent-AdditiveModel


Gunn, & Klebanov, 1994), academic ability (Mas-
ten et al., 1988), and quality of family relationships An independent-additive model of risk tests
whether individual risk factors have an indepen-
(Rutter, 1978). Like Rutter (1978), we are con-
cerned with whether particular family attributes dent or main effect on the dependent variable. Cor-
operate conditionally at varying levels of risk to responding to the idea of pileup (McCubbin & Pat-
predict youth maladjustment. terson, 1983), this model assumes that individual
Three important familial correlates of youth risk factors operate in a cumulative, linear pattern
maladjustment are repeated exposure to hostile, to place youth at risk for maladjustment.Thus, the
overt interparental conflict (Buehler et al., 1998; total effect of the risk factors is the summation of
Burman, John, & Margolin, 1987; Jouriles, Mur- their individual effects (Jenkins & Smith, 1990).
For instance, Shaw and Emery (1988) investigated
phy, & O'Leary, 1989); a poor parenting environ-
ment (Conger et al., 1992, 1993; Fauber,Forehand, the relationship between four chronic family stres-
Thomas, & Wierson, 1990; Patterson, 1982); and sors (parental conflict, maternal depression, over-
economic hardship (Bolger, Patterson, Thompson, crowding in the home, and family income) and
& Kupersmidt,1995; Dodge, Pettit, & Bates, 1994; school-aged children's psychological adjustment.
Takeuchi, Williams, & Adair, 1991). In this study, These researchersfound that 5- to 12-year-old chil-
we conceptualize these three factors as potential dren displayed increasingly higher scores on mea-
stressorsfor youth. sures of internalizing and externalizing problems
Parents' use of a hostile, overt style to address as the number of risk factors increased. Their re-
their disagreements and poor parenting constitute sults supported an additive effects model of risk
risk factors to youth by reflecting negative interac- exposure. In line with these findings, our first re-
tion patterns in the family system, particularly in search question is whether overt interparentalcon-
the marital and parent-child subsystems. The na- flict, poor parenting, and economic hardship each
ture and quality of interpersonal functioning in independently influence youth problem behaviors
these dyads influences the patterns that the child in a negative fashion.
learns for dealing with other important relation- One of the limitations of the literatureis that re-
ships in his or her life (Patterson, Debaryshe, & searchers seldom have included in their analysis
Ramsey, 1989). Economically disadvantagedyouth tests for other forms of risk patterns. This has im-
often reside in environments marked by high rates portant implications because additive findings do
of violence, poor housing and schooling condi- not preclude the possibility of finding other risk
tions, and a high proportion of births to teenage patterns, such as interactive effects (Hodges et al.,
mothers (McLoyd, 1990). Being rearedunder these 1984; Masten et al., 1988). Few researchersventure
circumstances may make it difficult for youth to beyond the linear, independent-additivemodel, and
master importantdevelopmental tasks. it is difficult to tell whether multiple risk factors op-
Past research has established links among overt erate in a more complex patternto put youth at risk
interparental conflict, poor parenting, and eco- for maladjustment.One of the contributionsof this
nomic hardship and suggests that these risk factors study is that we consider more complex patternsof
often occur concomitantly (Conger et al., 1992, interactionsamong risk factors.
1993; Elder, Conger, Foster, & Ardelt, 1992).
However, few researchers have explored all three Interactive Model
of these factors in the context of a model of multi-
ple risks. In an interaction model, there is at least one condi-
tional relationship between two of the risk factors.
Conditional means that the association between
PATTERNS
OFRISK
one risk factor and a measure of child adjustment
We examine three risk models postulated in chil- depends on the level of a second factor. One type
dren's risk research: (a) the independent-additive, of interaction is the protective-versus-vulnerability
(b) the interactive, and (c) the exponential. (See model (Masten et al., 1988). This model essentially
Jenkins & Smith, 1990; Masten et al., 1988.) Un- is a statistical test of the stress-bufferingand stress-
fortunately, the distinction between these three exacerbating notion posited by risk and resiliency
models is not always clear, resulting often in the researchers. Protective factors (e.g., high socioeco-
misinterpretationof research findings. nomic status, good parenting) buffer youth from
the harmful effects of other stressors, whereas vul-

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
346 Journal of Marriage and the Family

nerability factors (low socioeconomic status, poor nential model suggests that the effects of individ-
parenting) exacerbate the harmful effects. ual risk factors multiply or potentiate each other.
Our second research question is whether good Several large studies provide support for the
parentingbuffers youth from the harmful effects of exponential risk model (Rutter, 1971; Siefer, Sam-
overt interparentalconflict and economic hardship. eroff, Baldwin, & Baldwin, 1992; Williams et al.,
We found only two studies that included statistical 1990). Rutter found quite dramaticpotentiating ef-
tests for the protective function of a good interper- fects. In his investigation, two familial risk factors
sonal relationship between parent and child. One provided a fourfold increase in the likelihood of
found support, and the other did not. Brody and child psychiatric disorder; four factors increased
Forehand (1990) discovered that the association this risk tenfold. However, with the exception of
between interparental conflict and teacher-rated Shaw and Emery (1988), few investigators have
youth internalizing problems was worse when the tried to replicate his findings. We can test this
father-childrelationship was poor than when it was model by creating a continuous risk exposure vari-
good. In contrast, Jenkins and Smith (1990) found able that represents the number of risk factors that
that good parent-child relationships were associ- youth experience and then mathematically squar-
ated with children's adjustmentin both harmonious ing this term. Thus, the third question addressed in
and disharmonious homes, suggesting that parent- this study is whether youth exposed to several
ing quality operates independently of or additively family risk factors evidence problem behavior that
with marital harmony in predicting children's exponentially increases with added risk.
adjustment. Using these three risk models, this study exam-
Rutter (1979), Hess and Camara (1979), and ines the relationships among an overt interparental
Wallerstein and Kelly (1980) found that the stress conflict style, poor parenting, economic hardship,
associated with marital conflict (and divorce) was and youth problem behaviors. This investigation
mitigated to some extent by a good relationship focuses on preadolescent and early adolescent
with at least one parent. Although widely cited for youth, a time in a child's life that is particularly
demonstrating support of the stress-buffering ef- important when encountering family risk factors.
fects of good parenting, none of these researchers Stressful environments appear to influence indi-
actually tested the interaction between marital viduals more strongly during periods of rapid de-
conflict and quality of parenting. In the absence of velopment (Anthony, 1987). Given the cognitive,
such statistical tests, the existence of buffering emotional, physical, and social changes accompa-
effects of good parenting can only be inferred. nying this period (Feldman & Elliott, 1990), chil-
The lack of formal statistical tests also plagues dren at this stage of the life cycle may be more
studies that examine the interactiverole of economic vulnerable to stressful life circumstances than
well-being with other family risk factors. Jouriles, their younger and older peers.
Bourg, and Farris (1991) found a strongerrelation-
ship between marital adjustment and problematic METHOD
child conduct in families with low socioeconomic
status than in families with high socioeconomic
status across two waves of data. However, their Sampling
findings only suggest support for the interactive Criteria and procedures. The sample and data used
effect of socioeconomic status on risk exposure, for this study are part of a larger project developed
given that their statistical analysis did not include to assess family life, with a focus on interparental
an interactionterm between marital adjustmentand conflict (Buehler et al., 1998). The families were
socioeconomic status. part of a community sample recruited from se-
lected middle schools in Knox County, Tennessee.
Schools were chosen to ensure variation in socio-
Exponential Model
economic background. Eligible families had a tar-
First introduced by Rutter (1979), this model sug- get youth in sixth, seventh, or eighth grade who
gests a cumulative effect of risk. Some degree of resided with either married parents or their single,
risk exposure can be endured, but the accumula- divorced mothers. The target youth also were a
tion of risk factors results in an accelerating or ex- product of the currentor most recent marriage.
ponential increase in youth maladjustment. The data are cross-sectional and include infor-
Whereas the additive model reflects the linear sum mation collected from youth and their teachers on
of the effects of individual risk factors, the expo- measures of youth problem behavior. Because our

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
Family Risk Factors 347

goal is to assess reliably youth problem behavior, their own time. They were paid $3.00 for each
we used multiple report data to assess sources of completed, returned questionnaire. There are
reporter bias that might result when using only teacher data for 329 youth (98%).
youth reports.
Teachers were informed that their involvement Measurement
was voluntary. Slightly over half of the homeroom
teachers participated. Teachers had a target youth Frequency of use of a hostile overt conflict style.
in a 1-hour, daily class. Students are assigned ran- Youth responded to six items assessing the fre-
domly to homerooms, thereby reducing potential quency with which their parents engaged in
sources of teacher selection bias. Letters request- overtly conflictual behaviors during disagree-
ing parental consent were sent home with youth. ments. Sample items include: "call each other
Children were given a small token of appreciation names," "threateneach other,"and "tell each other
when the letters were broughtback to class. Mailed to shut up." The response format ranges from 1
follow-up questionnaires were sent to parents who (never) to 4 (very often). Cronbach's alpha for this
had not returned the consent forms. Parental con- scale is .87.
sent was granted for 75% of the invited youth. Evidence of validity for this measure of overt
conflict is provided by previous research, which in-
Sample characteristics. Data were collected from dicated that these six items factored separately
189 girls and 146 boys, aged 10-15 years (M = from 12 items developed by Ahrons (1983) to as-
12.4; SD = .99), yielding a sample size of 335 sess the frequency of interparental disagreement.
youth. The majority were Caucasian (86%) and This distinct factor structurecharacterizedtwo dif-
lived with married parents (87%). Based on data ferent samples of youth. Additionally, in both sam-
from youth reports, the average level of formal ed- ples of youth, overt conflict style was more strongly
ucation attained by mothers and fathers was be- associated with youth problem behaviors than was
tween high school and some college. Eighty-seven the frequency of disagreement, providing evidence
percent of youth paid full price for their school of predictive validity. Additional evidence of con-
lunch, 3% received a partial subsidy, and 10% re- struct validity was amassed by correlatingthe overt
ceived free lunches. conflict measure with other measures of inter-
The representativeness of this sample was ex- parentalconflict. Overt conflict correlated .52 (p <
amined by comparing these average characteristics .001) with parent's use of a conflict style that tri-
with data provided from the 1989 and 1990 census angled the child (Buehler et al., 1998), .67 (p < .001)
on families and individuals living in Knox County with youths' perceptions of the intensity of marital
(Slater & Hall, 1996; U.S. Bureau of the Census, conflict (Buehler & Barber, 1998), and .51 (p <
1996). In 1990, 75% of individuals aged 25 years .001) with youth report on seven items from the
and older had graduated from high school. About Children's Perception Questionnaire that measure
24% had a bachelor's degree. In 1990, 13% of the hostile interparental conflict (Emery & O'Leary,
families were headed by a female. The county 1984). These correlations reflect 26%-45% shared
records did not break this down by marital status, variance, providing support of convergent validity
but, of course, the percentage of divorced or sepa- for the measure of overt conflict used in this study.
rated female heads would be smaller than 13%.
Accordingly, the sample for this part of the study Parentingquality.To assess the overall qualityof the
slightly overrepresents divorced, mother-custody parenting environment, we constructed a second-
families. In 1989, 10% of all families had incomes order composite scale aggregating youth responses
below the poverty line. This percentage is close to to items measuringthe following variables:parental
the percentage of families in this sample whose acceptance, monitoring, overly strict discipline,
children received free school lunches. and intraparentalinconsistent discipline. The deci-
sion to take this global approach to the assessment
Data Collection Procedures of parenting behaviors rests on the theoretical idea
that deficits in several key areas of parenting con-
Youth completed questionnaires in the school stitute a greater risk to youth than a deficit in one
cafeteria on three consecutive days. Trained assis- particulararea of parenting.
tants were on site to answer questions. Youth were Parental acceptance-the degree to which par-
treated to a pizza party after completing the ques- ents use affirming messages that communicate to
tionnaire. Teachers completed questionnaires on the child that he or she is valued and important-

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
348 Journal of Marriage and the Family

is assessed using four items from the Children's parenting constructs that they represent constitute
Report of Parental Behavior Inventory (CRPBI, what scholars believe are the most critical elements
Schulderman & Schulderman, 1970). Sample of parenting (Conger et al., 1992; Maccoby &
items include: "My mother [father] is a person Martin, 1983; Patterson et al., 1989; Rollins &
who enjoys doing things with me," and "my Thomas, 1979).
mother [father] is a person who gives me a lot of
attention."Monitoring is measured with four items Economic hardship. We measured economic hard-
that ask how much parents know about their chil- ship using youths' school lunch status. Children
dren's activities (Brown, Mounts, Lamborn, & from low-income families are eligible for federally
Steinberg, 1993). Sample items include: "How mandated free lunches or lunches at a reduced
much do your parents really know about who your price if their family income does not exceed 130%
friends are?" and "how much do your parents re- and 185%, respectively, of the official poverty
ally know about what you do with your free time?" level. At the time of data collection, the official
Parental strict discipline is measured using two poverty level for a household of four ranged from
items from the CRPBI and two items from the $14,800 to $15,150 (U.S. Department of Health
Weinberger Parenting Inventory (WPI; Wein- and Human Services, 1994, 1995). For a household
berger, Feldman, & Ford, 1989). These items as- of this size, children qualified for free lunches if
sess the degree to which parents are harsh in their the family income ranged from $19,240 to
application of discipline. Sample items include: $19,695. Children qualified for reduced-price
"My mother [father] is a person who is very strict lunches if the family income ranged from $27,380
with me" (CRPBI), and "my mother [father] is a to $28,028. Information about lunch status was
person who spanks me so I will earn respect for provided by the participating schools. Economic
my elders" (WPI). Finally, inconsistent discipline hardship is coded 0 if a youth paid full price for
is measured using six items from the CRPBI. lunch, 1 if a youth paid a reduced price, and 2 if a
These items are intended to measure the degree to youth received free lunch. Few children paid re-
which parents erratically enforce rules and arbi- duced fees (approximately 4% of the sample), so
trarily use rewards and punishment. Sample items the latter two categories were collapsed into one
are: "My mother [father] is a person who only category to represent lower family economic well-
keeps rules when it suits her," and "my mother being.
[father] is a person who lets me do something one
day, and the next day I get into trouble for doing Externalizing and internalizingproblem behaviors.
the same thing." Youth problem behaviors are assessed using youth
The global parenting-quality scale was formed reports on the Youth Self-Report (YSR; Achen-
in three steps. First, youth responses to the parental bach, 1991) and teacher reports on the Teacher
acceptance and monitoring items were reversed to Report Form (TRF; Achenbach & Edelbrock,
reflect low acceptance and monitoring. Second, we 1986). The externalizing subscale consists of 30
created scales for each item in the parenting con- items. Sample items from the youth reportinclude:
structs by computing a mean score for youth "I steal at home," and "I am mean to others."The
reportsof mothers' and fathers' parentingbehavior. internalizing subscale consists of 31 items. Sample
After youth reports of mothers' and fathers' scores items from the youth reportof this scale include: "I
on each item were averaged, the aggregated scores am unhappy, sad, or depressed,"and "I feel worth-
were combined to form the global parenting scale. less or inferior."The 3-point responses range from
The rationale for averaging youth reports of moth- 0 (not true) to 2 (very true or often very true).
ers' and fathers' scores across individual items is to Cronbach's alpha for youth and teacher reports on
deal with the issue of conditionally missing data the externalizing subscale are .90 and .95, respec-
that resulted from youth in mother-headedfamilies tively. Cronbach's alpha for youth and teacher re-
who rarely saw their fathers. In cases in which a ports on the internalizing subscale are .92 and .87,
youth did not reporton his or her father'sparenting respectively. Widely used as assessment devices of
behavior, the mean score of his or her mother's be- children's behavior problems, the YSR and TRF
havior is used. Thus, conceptually this procedure have extensive evidence of internalconsistency and
accurately represents the total parenting environ- predictive validity (Achenbach, 1991; Achenbach
ment of varying family structures in this study. & Edelbrock, 1986). However, the correlation be-
Cronbach's alpha for this scale is .83. In terms of tween youth report of externalizing problem be-
content validity, the items on this scale and the four haviors and youth report of internalizing behaviors

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
Family Risk Factors 349

is .70 (49% shared variance), indicating a moder- coefficients indicate that youth reports of overt
ately high degree of comorbidity between the two conflict style and parenting quality are associated
outcome measures. Accordingly, our analysis of with youth and teacher reports of total problem
each dependent variable includes a statistical con- behavior and externalizing problem behavior, as
trol for the alternativeproblem behavior.To the ex- well as youth reports of internalizing problem be-
tent that the family risk factors in this study predict havior (but not teacher reports of internalizing).
broad behavior, youths' total problem behavior Economic hardship is associated with each mea-
also is used in the analyses. We formed this mea- sure of problem behavior.
sure by summing the externalizing and internaliz-
ing behavior scales. Cronbach'salpha for youth and The Independent-AdditiveModel
teacher reportsare .92 and .96, respectively.
A series of hierarchical multiple regression equa-
Control variables. Anticipating the likelihood that tions were computed to test for the independent-
the outcome measures may be influenced by demo- additive effects of the three family risk factors
graphic characteristics, we used dummy codes to when predicting youth problem behaviors. Entry
statistically control the variance attributable to of the variables proceeded in the following order:
youth gender, grade level, and parents' marital sta- Block 1-the alternative form of youth problem
tus. For youth gender, female youth were coded 0, behavior (this block was not included when the
and male youth were coded 1. For marital status, total score for problem behavior is used as the de-
mother-headedhouseholds were coded 0, and two- pendent variable); Block 2-youth gender, grade
parent families were coded 1. The three grade lev- level, and parents' marital status; Block 3-overt
els required two dummy variables. Using seventh conflict style, parenting quality, and economic
grade as the reference group, we coded sixth grade hardship.
1; all others were coded 0 for one variable. For the To test whether this model varies by youths'
other dummy variable, we coded eighth grade 1, gender and grade and parents' marital status, in-
and all others were coded 0. teraction terms between the individual risk factors
and each control variable were entered in a final
Data Analysis block (Block 4). The interaction terms were cre-
ated by multiplying the centered value of individ-
Data were analyzed using hierarchical multiple ual predictor variables by the centered value of
regression. This analytic procedure is well suited each dummy-coded control variable. This proce-
to research of this nature because it allows one to dure, which is intended to deal with the issue of
enter the interaction and squared terms necessary multicollinearity between variables, entailed sub-
for testing interactive and exponential patterns of tracting the mean of a variable from the individual
risk, respectively (Aiken & West, 1991). In addi- scores on that variable (Aiken & West, 1991). To
tion, this procedure helps identify variance con- maximize statistical power, a separate regression
tributed by individual variables or groups of vari- model was computed for each family risk factor
ables entered in different blocks. Final block entry and its interaction with youth gender, grade level,
differed as a function of the risk model tested, and parents' marital status.
whereas the use of control variables remained con-
sistent across analyses. For each of the three risk Total problem behaviors. Marital status was the
patterns, individual regression models were com- only significant control variable that predicted
puted for the three dependent variables (i.e., total youth reports of total problem behavior. (See col-
problem behavior, externalizing problem behav- umn 2 of Table 1.) Youth in mother-headedhouse-
ior, and internalizingproblem behavior) and across holds were at a higher risk of maladjustmentthan
informants of youth problem behaviors (i.e., youth youth residing with both parents. Overt conflict
and teachers). Reported findings are those that are style and parenting quality were significant risk
significant at the .05 level. predictors, supporting their independent, additive
influence on broad-based youth adjustment. None
RESULTS
of the changes in F associated with the interactions
between the risk factors and the control variables
The Appendix contains zero-order correlations, was significant. (See columns 4-6 of Table 1.)
means, standard deviations, and reliability coeffi- Thus, the cumulative effect of overt conflict and
cients for all variables in this study. The bivariate poor parenting influenced youth regardless of their

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
TABLE 1. HIERARCHICALREGRESSION ANALYSES OF OVERT CONFLICT, PARENTING QUALITY, AND
ECONOMIC HARDSHIP PREDICTING YOUTH AND TEACHER REPORTS OF TOTAL PROBLEM BEHAVIOR

YouthReportTotalProblemBehavior Teache
Family Overt Parenting Economic Fam
Risk Conflict Quality Hardship Ris
Control Factors Interactions Interactions Interactions Control Fact
Variable Block 1 Block 2 Block 3a Block 3b Block 3c Block 1 Bloc
Summaryof block entry
TotalR2(in %) 4.5** 28.6*** 29.6 30.3 28.9 5.8*** 10.6
R2change (in %) 4.5** 24.1*** 1.0 1.6 .3 5.8*** 4.9*
Betas
Youthgender .08 .02 .03 .02 .02 .13* .1
Youthgrade 1 (sixth) .02 -.02 -.00 -.03 -.02 .14* .0
Youthgrade2 (eighth) .06 -.05 -.03 -.06 -.04 -.05 -.08
Parent'smaritalstatus -.20** -.11* -.12* -.08 -.10* -.09 -.04
Overtconflict .12* .10* .10* .11* .0
Parentingquality .44*** .43*** .45*** .44*** .15
Economichardship .02 .03 .03 .04 .16
Overtconflict x youth gender .07
Overtconflict x youth grade 1 .03
Overtconflict x youth grade 2 .09
Overtconflict x parent'smaritalstatus .03
Parentingqualityx youth gender -.02
Parentingqualityx youth grade 1 -.05
Parentingqualityx youth grade2 .04
Parentingqualityx parent'smaritalstatus -.11*
Economichardshipx youth gender .04
Economichardshipx youth grade 1 -.01
Economic hardshipx youth grade 2 .03
Economichardshipx parent'smaritalstatus -.05
Note: n = 335.
*p < .05. **p < .01. ***p < .001.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
TABLE 2. HIERARCHICALREGRESSION ANALYSES OF OVERT CONFLICT, PARENTING QUALITY, AND
ECONOMIC HARDSHIP PREDICTING YOUTH AND TEACHER REPORTS OF EXTERNALIZINGPROBLEM BEHAV

YouthReportExternalizingProblemBehavior TeacherReportE
Alternate Family Overt Parenting Economic Alternate Fam
Problem Risk Conflict Quality Hardship Problem Ri
Behavior Control Factors Interactions Interactions Interactions Behavior Control Fac
Variable Block 1 Block 2 Block 3 Block 4a Block 4b Block 4c Block 1 Block 2 Blo
1lt ,Fh +1 - .rt
Nummary ot oIUOKentry
TotalR2(in %) 49.1*** 53.2*** 58.0*** 58.5 59.4* 58.9 15.7*** 18.4* 20.7
R2change (in %) 49.1*** 4.1*** 4.8*** .5 1.3* .9 15.7*** 2.7* 2.3
Betas
Internalizingproblem
behavior .70*** .69*** .59*** .59*** .58*** .60*** .40*** .37*** .3
Youthgender .19*** .15**1* . 15**1 .15*** .13** .1
Youthgrade 1 (sixth) .00 -.02 -.02 -.03 .01 .07 .0
Youthgrade2 (eighth) .05 -.00 -.00 -.01 .04 -.03 -.0
Parent'smaritalstatus -.08* -.05 -.06 -.04 -.06 -.04 -.0
Overtconflict .02 .03 .00 .02 .0
Parentingquality .24*** .23*** .24*** .23*** .1
Economic hardship .01 .02 .03 -.01 -.0
Overtconflict x
youth gender .03
Overtconflict x
youth grade 1 -.07
Overtconflict x
youth grade2 -.03
Overtconflict x
parent'smaritalstatus .02
Parentingqualityx
youth gender .07*
Parentingqualityx
youth grade 1 -.07
Parentingqualityx
youth grade2 -.02
Parentingqualityx
parent'smaritalstatus -.07*
Economic hardshipx
youth gender .02
Economic hardshipx
youth grade 1 .16*
Economic hardshipx
youth grade 2 .13
Economic hardshipx
parent'smaritalstatus .02
Note: n = 335.
*p<.05. **p<.01. ***p<.001.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
352 Journal of Marriage and the Family

gender, grade level, and their parents' marital sta- (See column 3 of Table 3.) The negative beta co-
tus. (The significant beta coefficient for the interac- efficient for this variable may reflect the suscepti-
tion between parenting quality and parent's marital bility of female youth to overcontrolled behavior.
status was not interpreted because the relevant Youth reports reveal no significant associations
change in F was not significant.) between familial risk factors and internalizing
Gender was the only significant control vari- problem behavior. Thus the additive effects model
able that predicted teacher reports of total prob- was not supported. Significant interactions
lem behavior, reflecting higher scores on this emerged between parent quality and youth gender
index for male youth. (See row 5 and column 7 of and between economic hardship and youth grade.
Table 1.) Like youth reports of total problem be- However, these were not interpreted because the
havior, teacher reports supported the independent- overall change in F was not significant. Teacher
additive model of risk. Poor parenting quality and reports reveal a significant association between
economic hardship both emerged as significant economic hardship and internalizing problem be-
risk factors. However, the association between havior. (See row 12 of Table 3.) This association
economic hardship and total problem behavior is similar, regardless of youth gender, youth
depended on parents' marital status. Youth in eco- grade, and parent's marital status.
nomically disadvantaged two-parent households
scored higher on total problem behavior than The Interactive Model
youth in low-income, mother-headed households.
Thus, the independent-additive model is sup- To test the possibility that risk factors interact with
ported for youth from two-parent families, but not one another in a conditional fashion, we created
for youth from single-parent families when we three interactionterms: overt conflict style x parent-
focus on teacher reports of total problem behav- ing quality, overt conflict style x economic hard-
ior. Poor parenting influences youth regardless of ship, and parenting quality x economic hardship.
their gender, grade level, or their parent's marital The interactionterms were entered in Block 4 after
status. direct relationships between the dependent vari-
able and the alternativeproblem behavior, the con-
Externalizing problem behaviors. Gender was the trol variables, and the familial risk factors were
only significant control variable that predicted taken into account.
youth reports of externalizing problem behavior, The findings do not support the interactive
indicating greater risk of undercontrolledbehavior model. None of the 18 interaction terms tested was
for male youth. (See column 3 of Table 2.) Only significant, ruling out the buffering or exacerbat-
parenting quality was a significant family risk fac- ing effects of particular familial variables for this
tor. The parenting quality x gender interaction sample of youth. Although parenting quality x
terms and the parenting quality x marital status in- economic hardship reached a probability level
teraction terms also were significant. The former between .05 and .10 for teacher reports of total
finding indicates that the relationshipbetween poor problem behavior and externalizing problem be-
parenting quality and externalizing problem behav- havior, its inclusion in these regression models did
ior is stronger for males than females, but signifi- not result in a significant F change.
cant for both. The latter finding indicates that the
relationship between poor parenting quality and The Exponential Model
externalizing problem behavior is significant for
both youth in mother-headedhouseholds and youth We conducted a final set of analyses to examine
in two-parent families. However, this relationship whether incremental increases exposure to risk
is more detrimentalto youth in single-parent fami- result in exponential increases in levels of youth
lies. Consistent with their reports of total problem maladjustment. To conduct these analyses, a new
behavior, teacher reports of externalizing problem variable, named "risk," was formed to represent
behavior indicate that gender, parenting quality, varying levels of risk exposure. Based on youth re-
and economic hardshipx parent's marital status are ports, youth were classified as having zero, one,
significant predictorvariables. two, or three risk factors. For risk factors measured
on a continuous scale (overt conflict and parenting
Internalizing problem behaviors. Gender was the quality), the cutoff point used to classify at-risk
only significant control variable that predicted youth was a score 1 standard deviation above the
youth reports of internalizing problem behavior. mean. For economic hardship, youth receiving

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
TABLE 3. HIERARCHICALREGRESSION ANALYSES OF OVERT CONFLICT, PARENTING QUALITY, AND
ECONOMIC HARDSHIP PREDICTING YOUTH AND TEACHER REPORTS OF INTERNALIZINGPROBLEM BEHAV

YouthReportInternalizingProblemBehavior TeacherReportI
Alternate Family Overt Parenting Economic Alternate Fam
Problem Risk Conflict Quality Hardship Problem R
Behavior Control Factors Interactions Interactions Interactions Behavior Control Fac
Variable Block 1 Block 2 Bloc]k 3 Block 4a Block 4b Block 4c Block 1 Block 2 Blo

Summaryof block entry


TotalR2(in %) 49.1*** 51.4*** 52.2 52.6 52.9 53.2 15.7*** 17.2 23.8
R2change (in %) 49.1*** 2.3*** .7 .5 .8 1.0 15.7*** 1.5 6.6
Betas
Externalzing problem
behavior .70*** .72*** .6;7*** .67*** .68*** .68*** .39*** .38*** .3
Youthgender -.14*** -.1j *** -.14*** -.14*** -.15*** -.02 -.0
Youthgrade 1 (sixth) -.01 .0 ) .02 .01 -.02 .09 .0
Youthgrade2 (eighth) -.02 -.0r3 -.02 -.03 -.06 -.02 -.0
Parent'smaritalstatus -.03 -.022 -.02 -.02 -.01 -.06 -.0
Overtconflict .0(6 .04 .06 .06 -.0
Parentingquality .0(6 .06 .07 .06 .0
Economic hardship .011 .00 -.00 .03 .2
Overtconflict x
youth gender .02
Overtconflict x
youth grade 1 .09
Overtconflict x
youth grade2 .10
Overtconflict x
parent'smaritalstatus -.01
Parentingqualityx
youth gender .09*
Parentingqualityx
youth grade I .04
Parentingqualityx
youth grade2 .04
Parentingqualityx
parent'smaritalstatus -.00
Economic hardshipx
youth gender .01
Economic hardshipx
youth grade 1
Economic hardshipx
youth grade2 -.12
Economichardshipx
parent'smaritalstatus -.05
Note: n = 335.
*p<.05. **p<.01. ***p<.001.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
TABLE 4. HIERARCHICALREGRESSION ANALYSES OF RISK AND RISK SQUARED PREDICTING YOUTH AND TEACHER REPORTS OF

YouthReportTotalProblemBehavior Tea
Control Risk Risk2 Control
Variable Block 1 Block 2 Block 3 Block 1
Summaryof block entry
TotalR2(in %) 4.5** 18.8*** 20.8** 5.8***
R2change (in %) 4.5** 14.3*** 2.0** 5.8***
Betas
Youthgender .08 .05 .06 .13*
Youthgrade 1 (sixth) .02 -.07 -.08 .14*
Youthgrade2 (eighth) .06 .01 .01 -.05
Parent'smaritalstatus -.20*** -.14** -.13** -.09
Risk -.39*** .40***
Risk2 -.26**
Note: n = 335.
*p <.05. **p <.01. ***p < .001.

TABLE 5. HIERARCHICALREGRESSION ANALYSES OF RISK AND RISK SQUARED PREDICTING YOUTH AND TEACHER REPORTS OF EXTE

YouthReportExternalizingProblemBehavior TeacherRepo
Alternate Alternate
Problem Problem
Behavior Control Risk Risk2 Behavior Co
Variable Block 1 Block 2 Block 3 Block 4 Block 1 Bl
Summaryof block entry
TotalR2(in %) 49.1*** 53.2*** 54.8*** 54.9 15.7*** 18.
R2change (in %) 49.1*** 4.1*** 1.5*** .1 15.7*** 2.
Betas
Internalizingproblembehavior .70*** .69*** .65*** .65*** .40*** .
Youthgender .19*** .18*** .18*** .
Youthgrade 1 (sixth) .00 -.03 -.03 .
Youthgrade2 (eighth) .05 .03 .03 -.0
Parent'smaritalstatus -.08* -.07 -.07 -.0
Risk .14*** .11***
Risk2 .03
Note: n = 335.
*p < .05. **p <.01. ***p <.001.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
Family Risk Factors 355

either a partial or a full lunch subsidy were consid-


* *
It. *
0 Cl \0 t
* ered at risk. The newly formed variable was entered
*q 10 Cl) 0000
in Block 3 after entry of the alternative behavior
0 problem in Block 1 and the control variables in
0
CO .0.0-
CA 0) Block 2. In Block 4, the squared risk variable was
0 .u
entered to test for exponential effects of multiple
0 *
0
* *
0
*
*
*
risk exposure (Aiken & West, 1991).
2c *
&-
* cCC'0I-t.l-
p00000- The risk term was significant across informants
0
0 00- I II
CO
,.
0h
and dependent variables, suggesting a linear pattern
0
0 of adjustment to risk exposure. (See Tables 4-6.)
0
z ct
CD
Additionally, the curvilinear risk term reached sta-
tistical significance in three of the six regression
a) *
z0 00 0\ - models. These were for youth and teacherreportsof
0 ClC( 00000
0
H I II total problem behavior as well as youth reports of
z 0.
U
c. internalizing problem behavior. Contrary to the
0
steep increases in problem behavior expected for
H m
0
0 potentiated risk, the findings are more consistent
0
0 2CO s -
with a threshold effect of multiple risk exposure.
0 - $ .0
-
*
*
0
0
rCO0 t- r- The data indicated that the most dramatic increase
0
0 in these measures of problem behavior is between
0
H zero and one risk factor, with a slight leveling off at
0
z .0 n I greater levels of risk exposure. Closer examination
0
H
of these findings reveal that 47% of youth classified
0
0 as having a risk factor of one (where problem be-
* *
0
.0.0
*0 havior scores were at their highest) experience poor
z ,- ,O

0
cuq parentingquality,thus indicating the saliency of this
0
0
risk factor in predicting youth problem behavior.
0
ci
a)
*- **
ICCl Cl Cl ~ Although the curvilinear term accounts for only a
0
0
0
small portion of the variancein these outcome mea-
CO
0 .0.0 sures beyond that of the linear risk term (approxi-
& * 0 * *
000--N c
00
0
"O
0
t cr
mately 3%), the consistency of this finding across
0
informantsoffers reliable evidence of how high risk
0
z a. exposure manifests itself in this sample of youth.
0 ,.
- Cl
0
C .0 * * DISCUSSION
* *
0
0 (-y This study tests various risk models that may ex-
plain how multiple risk factors predict preadoles-
z cent and early adolescent problem behaviors. The
>CO
0
three risk models are based on different assump-
2
CO
s -

0 0 $.0 tions about the relationship between exposure to


0 .0 CO 0
0
0
0.00
<o multiple risks and youth problem behaviors. These
at- OIt
models have different implications for youth con-
0 o fronted with multiple adversities. However, the
0
0
v research lacks sufficient tests of these models, and
0 our understanding of what exposure to multiple
0 risks means is limited. Our findings extend the
0
o literature on youth adjustment to family stress by
.4
0
v addressing whether complex patterns of cumula-
H * tive risk provide better explanations of youth prob-
" lem behaviors than do simpler additive models.
ajv ^ We first tested whether overt interparentalcon-
i .& flict, poor parenting quality, and family economic
hardship jointly predict youth problem behaviors

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
356 Journal of Marriage and the Family

in an additive fashion. Few researchers have ex- remained the same when correlated error was con-
amined all three of these factors conjointly in the trolled). Third, we conducted follow-up regression
context of a multiple risk model. The results of analyses to determine whether the association be-
this study highlight the importance of this type of tween youth reports of poor parenting and their
exploration and support the idea of cumulative ef- own reports of problem behaviors could be ex-
fects of multiple family risk factors on youths' plained by a hostile attributionbias held by youth.
psychological adjustment. Youth who have negative beliefs about the world,
Youth perceptions of hostile, overt interparental in general, and about social relationships, in par-
conflict and poor parenting explained variance in ticular, may describe their parents in a more nega-
youth reports of total problem behaviors. Some tive fashion. We used a six-item measure of youth
scholars have argued that marital conflict influ- hostile attributionbias (e.g., "I think most people
ences youth outcomes by contaminating parental can't be trusted, there are some people I really
behavior and the general quality of parent-childre- hate") and found that the addition of this measure
lations (Erel & Burman, 1995; Fauber et al., 1990). as the final block in the regression analysis ex-
For example, Harold and Conger (1997) found in a plained additional variance in youths' total prob-
sample of 451 families in ruralIowa that one of the lem behaviors, but that the association between
major pathways from marital conflict to adolescent poor parenting and problem behaviors barely
problem behaviors was through observed and changed (before Beta = .24, after Beta = .21). Thus,
youth-perceived parental hostility toward the it does not seem that the association between poor
youth. Because their design was longitudinal, they parenting and youth problem behaviors can be
provided evidence that hostility in the marital rela- accounted for by youths' hostile outlook on social
tion may actually spill over to the parent-childrela- and family life.
tionship. However, we found that the main effects Teachers' reports of total problem behaviors
of overt conflict remained, even when accounting were explained by youth reports of poor parenting
for poor parenting. This may occur by shaping and family economic hardship. The robustness of
children's cognitions and perceptions (Davies & these findings need to be highlighted because each
Cummings, 1998; Grych & Fincham, 1990), their variable was assessed by a different informant.
ability to regulate their emotions (Gottman & Katz, Teachers reported on problem behaviors, youth re-
1989), and through their particular coping re- ported on parenting, and schools provided data to
sponses (Jenkins, Smith, & Graham, 1989; create the measure of family economic hardship.
O'Brien, Margolin, & John, 1995). Tittsworth (1996) found that parents' involvement
One of the potential limitations of these find- with children's school life (e.g., helping with
ings is that youth reported on conflict and parent- homework and attending school activities) was
ing, as well as their own problem behaviors. Thus, closely associated with parental monitoring. Thus,
these findings may reflect rater bias, more than the association between poor parenting and youth
they reflect valid, substantive findings about fam- problem behaviors may reflect a lack of involve-
ily risk and youth problem behaviors. Although ment by parents in their child's academic life. This
rater bias may account for a small portion of the lack of parental involvement and structure may
association between youth reports of family risk fuel misbehavior in the classroom and may indi-
and youth problem behaviors, it does not account cate an absence of a connection between parents
for it to any great extent, and the major conclusions and teachers that might help mitigate such mis-
remain intact. First, at the zero-order level, youths' behavior. The association between poverty and
perceptions of overt interparentalconflict and poor youth problem behaviors may be explained by
parenting are associated with total problem behav- possible links among the educational quality of
iors reported by both youth and teachers. Thus, at home environments, academic difficulties, and in-
best, only a part of the correlationcan be accounted creased problem behaviors. Poor families have
for by single informant measurement. Second, fewer resources to commit to educationally related
using latent constructanalysis, Buehler et al. (1998) tools such as computers, books, and calculators.
demonstrated that youth reports of overt conflict This type of impoverishment is associated with
and youth problem behaviors contain little corre- academic failure (Downey, 1994), which, in turn,
lated error in this sample of youth, and the corre- may prompt or sustain misbehavior or emotional
lated error that is present does not alter the associ- distress in school.
ation between interparental conflict and youth Only poor parenting was associated with youth
problem behavior (i.e., the structural paths reports of externalizing problem behavior (con-

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
Family Risk Factors 357

trolling for internalizing problem behavior). economic status mitigate the harmful effects asso-
These results replicate previous findings and re- ciated with the risk level of these variables. This
affirm the role that parenting quality plays in chil- finding has important implications because it ap-
dren's psychosocial adjustment. Males and youths pears that poor parenting quality, the most consis-
in mother-headed households were particularly tent predictor of youth problem behaviors in this
susceptible to the risk associated with ineffective study, is detrimentalto youth well-being regardless
parenting. This is consistent with results of previ- of other positive familial processes and resources
ous researchers who found that divorced mothers considered importantto youth adjustment (i.e., co-
make fewer developmentally appropriatedemands, operativestyles of addressingconflict or high socio-
are less affectionate and less consistent in their economic status).However, more researchis needed
discipline, and generally manage children, particu- to explore the interactive relationships among
larly boys, less effectively, compared with mothers these factors before reaching definitive conclusions
in two-parent families (Hetherington, Cox, & about the stress-buffering or stress-exacerbating
Cox, 1987; Simons, 1996). effects of these particularfamily attributes.
Independent, additive effects were detected in Although the findings supportthe idea of cumu-
teachers' reports of externalizing problem behav- lative risk, our analysis also reveals a threshold ef-
ior. Both parenting quality and economic hardship fect of exposure to multiple risks on youth problem
were significant predictors.Additionally, economic behavior.We attributethis finding to the overriding
hardship interacted with the marital status of influence of parenting quality in predicting youth
youths' parents.Contraryto what one might expect, problem behavior, which may, as a result, have at-
youths in economically disadvantaged two-parent tenuated the relationships among overt conflict,
households display higher levels of externalizing economic hardship, and youth problem behavior.
problem behavior than youths in economically Poor parenting is a strong risk factor and saturates
disadvantaged mother-headed households. This the model. If these findings are replicated, one of
may indicate the influential nature of fathers' par- the positive implications is that individual risk fac-
enting quality in times of economic crisis. Liker tors in the family do not seem to potentiate the
and Elder (1983) found that the economic hardship harmful effects of one another. Although youth
stemming from the Great Depression was particu- who experience two of the risk factors obviously
larly distressing to fathers and that they manifested have more stress to cope with than those who expe-
this distress through the use of unsupportive, rience only one (as shown in the results of the
harsh, and rejecting behavior toward their children. additive models), these risk factors don't seem to
Youths in two-parent families may be more likely compound one another to any great extent. This
to be the recipients of this type of behavior, given relative independence has implications for research
the limited amount of contact many children of di- on children's coping responses to family stress. Re-
vorce have with their fathers (Furstenberg, Nord, searchers may want to begin to explore whether
Peterson, & Zill, 1983). Thus, aversive familial general or more specific coping behaviors help
processes particular to two-parent families may mitigate the effects of family stressors on youths'
be the operative mechanism that accounts for the difficulties with socialization.
finding that cumulative risk is more detrimental in Less variance was accounted for when predict-
this family type. ing internalizing problem behaviors than when
We did not find additive effects for internalizing predicting pure externalizing problem behaviors.
problem behavior. However, this analysis reveals To the extent that individualrisk factors are special-
that economic status is an important predictor of ized predictors of youth externalizing and internal-
internalizing problem behavior. The transition to izing problem behaviors, our analysis may not have
adolescence is a time of increasing psychological accounted for the variables most strongly related
vulnerability for youth (Harter, 1990). Economic to internalizing problem behavior. A different pat-
hardshipmay increase this vulnerabilityby leading tern may have emerged for internalizing problem
youth to draw negative social comparisons as they behavior if our analysis included other predictors
move away from the family and begin to expand of this outcome measure (e.g., maternaldepression,
their social network. peer rejection,youths' negative social comparisons).
Contrary to the few other studies reporting The findings must be weighed carefully against
buffering effects of positive family attributes, this the limitations of this study. First, this study does
study did not provide evidence that low levels of not adequately deal with the chronicity of family
overt conflict, positive parenting quality, and high risk. Chronic risk may result in severer maladjust-

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
358 Journal of Marriage and the Family

ment. This is especially important with respect to NOTE


economic hardship. For instance, children living in
This study was supported, in part, by a grant from the
persistent poverty are at greater risk for maladjust- University of Tennessee to Cheryl Buehler. We acknowl-
ment than those who experience transientfinancial edge Christine Anthony, Ambika Krishnakumar,Sharon
instability (Duncan et al., 1994). This study's mea- Pemberton, and Gaye Stone for their contributions to this
sure of economic status reflects youths' current study. We thank Jo Lynn Cunningham, Greer Litton Fox,
and the anonymous reviewers for their comments on ear-
financial situation and gives no indication of the lier drafts. We also thank Samuel Bratton of the Knox
historic economic situations of the families. Risk County School System for his supportand the many fam-
researchers interested in the influence of this vari- ilies and teachers who participatedin this project.
able would benefit from obtaining measures of
transient and chronic economic hardship (Dubow REFERENCES
& Ippolito, 1994).
Second, this study accounts for only a small Achenbach, T. M. (1991). Manual for the Youth Self-
Report and 1991 Profile. Burlington: University of
portion of variance in youth problem behaviors and. Vermont, Department of Psychiatry.
suggests the important role of other risk factors. Achenbach, T. M., & Edelbrock, C. S. (1986). Manual
Our focus was stress in the family. However, a for the Teacher's Report Form of the Child Behavior
more exhaustive examination of multiple risk fac- Profile. Burlington: University of Vermont, Depart-
ment of Psychiatry.
tors could include individual difference variables
Ahrons, C. R. (1983). Predictors of paternal involvement
such as constitutional and cognitive factors (e.g., postdivorce: Mothers' and fathers' perceptions. Jour-
temperament or personality and intelligence), as nal of Divorce, 6, 55-69.
well as environmental factors such as neighbor- Aiken, L. S., & West, S. G. (1991). Multiple regression:
hood poverty, exposure to crime, peer influence, Testing and interpreting interactions. Newbury Park,
CA: Sage.
and the availability of social support (Barber &
Aneshensel, C. S., & Gore, S. (1991). Development,
Olsen, 1997; Garmezy, 1981). stress, and role restructuring: Social transitions of
Finally, the low number of risk factors experi- adolescence. In J. Eckenrode (Ed.), The social context
enced by the majority of families in this sample of coping. New York: Plenum Press.
Anthony, E. J. (1987). Risk, vulnerability, and re-
may bear directly on our failure to detect an expo- silience: An overview. In E. J. Anthony & B. J.
nential pattern of risk. Rutter (1979) focused his Cohler (Eds.), The invulnerable child (pp. 3-48). New
investigation on a high-risk population that experi- York: Guilford.
enced a multiplicity of risk factors. Thus, there Barber, B. K., & Olsen, J. A. (1997). Socialization in
context: Connection, regulation, and autonomy in the
may be a relationship between the degree of expo-
family, school, and neighborhood, and with peers.
sure to risk and the manner in which exposure to Journal of Adolescent Research, 12, 287-315.
risk manifests itself in youth problem behaviors. Bolger, K. E., Patterson, C. J., Thompson, W. W., & Ku-
Researchers can address this question by testing persmidt, J. B. (1995). Psychosocial adjustmentamong
these various risk models across community and children experiencing persistent and intermittentfamily
economic hardship. Child Development, 66, 1107-
clinical samples.
1129.
This study makes a significant contribution to Borduin, C. M., Henggeler, S. W., & Manley, C. M.
the literatureson family stress and children's risk in (1995). Conduct and oppositional disorders. In V. B.
the socialization context by clearly delineating and VanHasselt & M. Hersen (Eds.), Handbook of adoles-
cent psychopathology: A guide to diagnosis and treat-
testing three importantrisk models. We found that ment (pp. 349-383). New York:Lexington Books.
most of the effects of risk in the family indepen-
Boss, P. (1988). Family stress management. Newbury
dently and cumulatively influence difficulties in Park, CA: Sage.
youths' socialization, represented by the presence Brody, G. H., & Forehand, R. (1990). Interparentalcon-
of problem behaviors. We found little evidence of flict, relationshipwith the noncustodial father,and ado-
the buffering or exacerbating effects of family lescent post-divorce adjustment. Journal of Applied
Developmental Psychology, 11, 139-147.
stressors, and we found no evidence of extreme Brown, B. B., Mounts, N., Lamborn, S. D., & Stein-
potentiation of the stressors. These findings con- berg, L. (1993). Parenting practices and peer group
tribute to the literature by detailing the patterns affiliation in adolescence. Child Development, 64,
467-482.
among family stressors. This detail can be used to
inform theoretical formulations and interventions Buehler, C., & Barber,B. K. (1998, July). Intensity of in-
terparental conflict and youth problem behaviors: The
related to children's coping with family stress.
moderating role of content, chronicity, and resolution.
Paper presented at the 1998 bi-annual conference of
the InternationalSociety for the Study of Behavioural
Development, Berne, Switzerland.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
Family Risk Factors 359

Buehler, C., Krishnakumar,A., Stone, G., Anthony, C., Furstenberg, F. F., Jr., Nord, C. W., Peterson, J. L., &
Pemberton, S., Gerard, J., & Barber, B. K. (1998). In- Zill, N. (1983). The life course of children of divorce:
terparental conflict styles and youth problem behav- Marital disruption and parental contact. American So-
iors: A two-sample replication study. Journal of Mar- ciological Review, 48, 656-668.
riage and the Family, 60, 119-132. Garmezy, N. (1981). Children under stress: Perspectives
Burman, B., John, R. S., & Margolin, G. (1987). Effects on antecedents and correlates of vulnerability and re-
of marital and parent-child relations on children's ad- sistance to psychopathology. In A. Rubin, J. Arnoff,
justment. Journal of Family Psychology, 1, 91-108. A. Barclay, & R. Zucker (Eds.), Further exploration in
Cohen, S., & Wills, T. A. (1985). Stress, social support, personality (pp. 196-268). New York:Wiley.
and the buffering hypothesis. Psychological Bulletin, Gottman, J. M., & Katz, L. F. (1989). Effects of marital
98, 310-357. discord on young children's peer interaction and
Conger, R. D., Conger, K. J., Elder, G. H., Lorenz, F. O., health. Developmental Psychology, 25, 373-381.
Simons, R. L., & Whitbeck, L. B. (1992). A family Grych, J. H., & Fincham, F. D. (1990). Marital conflict
and children's adjustment: A cognitive-contextual
process model of economic hardship and adjustment
of early adolescent boys. Child Development, 63, 526- framework. Psychological Bulletin, 2, 267-290.
541. Harold, G. T., & Conger, R. D. (1997). Marital conflict
and adolescent distress: The role of adolescent aware-
Conger, R. D., Conger, K. J., Elder, G. H., Lorenz, F. 0., ness. Child Development, 68, 333-350.
Simons, R. L., & Whitbeck, L. B. (1993). Family eco-
Harter,S. (1990). Self and identity development. In S. S.
nomic stress and adjustment of early adolescent girls. Feldman & G. R. Elliot (Eds.), At the threshold: The
Developmental Psychology, 29, 206-219. developing adolescent (pp. 352-387). Cambridge, MA:
Davies, P. T., & Cummings, E. M. (1998). Exploring HarvardUniversity Press.
children's emotional security as a mediator of the link Hess, R. D., & Camara,K. A. (1979). Post-divorce family
between marital relations and child adjustment. Child relationships as mediating factors in the consequences
Development, 69, 124-139. of divorce for children. Journal of Social Issues, 35,
Dodge, K. A., Pettit, G. S., & Bates, J. E. (1994). Social- 79-96.
ization mediators of the relation between socioeco- Hetherington, E. M., Cox, M., & Cox, R. (1987). Long-
nomic status and child conduct problems. Child Devel- term effects of divorce and remarriage on the adjust-
opment, 65, 649-665. ment of children. In S. Chess & T. Alexander (Eds.),
Downey, D. B. (1994). The school performance of chil- Annual progress in child psychiatry and child develop-
dren from single-mother and single-father families: ment 1986 (pp. 407-429). New York:Brunner/Mazel.
Economic or interpersonal deprivation? Journal of Hodges, W. F., Tierney, C. W., & Buchsbaum, H. K.
Family Issues, 15, 129-147. (1984). The cumulative effect of stress on preschool
Dubow, E. F, & Ippolito, M. F. (1994). Effects of poverty children of divorced and intact families. Journal of
and quality of the home environmenton changes in the Marriage and the Family, 46, 611-617.
academic and behavioral adjustment of elementary Honig, A. (1986). Stress and coping in children. Young
school-age children. Journal of Clinical Child Psychol- Children, 41, 50-63.
ogy, 23, 401-412. Jenkins, J. M., & Smith, M. A. (1990). Factors protect-
Duncan, G. J., Brooks-Gunn, J., & Klebanov, P. K. ing children living in disharmonious homes: Maternal
(1994). Economic deprivation and early childhood reports. Journal of the American Academy of Child
development. Child Development, 65, 296-318. and Adolescent Psychiatry, 29, 60-69.
Elder,G. H., Jr.,Conger,R. D., Foster,E. M., & Ardelt,M. Jenkins, J. M., Smith, M. A., & Graham, P. J. (1989)
(1992). Families under economic pressure. Journal of Coping with parental quarrels. American Academy of
Family Issues, 13, 5-37. Child and Adolescent Psychiatry, 28, 182-189.
Emery, R. E. (1982). Interparentalconflict and children of Johnson, J. H. (1982). Life events as stressors in child-
discord and divorce. Psychological Bulletin, 92, 310- hood and adolescence. In B. B. Lahey & A. E.
Kazdin (Eds.), Advances in clinical child psychology
330.
(Vol. 5, pp. 219-253). New York:Plenum Press.
Emery, R. E., & O'Leary, K. D. (1984). Marital discord
and child behavior problems in a nonclinic sample. Jouriles, E. N., Bourg, W. J., & Farris, A. M. (1991).
Marital adjustment and child conduct problems: A
Journal of Abnormal Child Psychology, 10, 411-420.
comparison of the correlation across subsamples.
Erel, O., & Burman, B. (1995). Interrelatednessof marital Journal of Consulting and Clinical Psychology, 59,
relations and parent-childrelations:A meta-analyticre- 354-357.
view. Psychological Bulletin, 118, 106-132. Jouriles, E. N., Murphy,C. M., & O'Leary, K. D. (1989).
Farrington,D. P. (1991). Childhood aggression and adult Interspousal aggression, marital discord, and child
violence: Early precursors and later-life outcomes. In problems. Journal of Consulting and Clinical Psychol-
D. J. Peeler & K. H. Rubin (Eds.), The development ogy, 57, 453-455.
and treatment of childhood aggression (pp. 5-29). Kazdin, A. E. (1995). Conduct disorders in childhood
Hillsdale, NJ: Erlbaum. and adolescence (2nd ed.). Thousand Oaks, CA: Sage.
Fauber,R., Forehand, R., Thomas, A. M., & Wierson, M. Kovacs, M., Feinberg, T. L., Crouse-Novak, M., Pau-
(1990). A mediational model of the impact of marital lauskas, S. L., Pollock, M., & Finkelstein, R. (1984).
conflict on adolescent adjustment in intact and di- Depressive disorders in childhood: II. A longitudinal
vorced families: The role of disruptedparenting. Child study of the risk for a subsequent major depression.
Development, 61, 1112-1123. Archives of General Psychiatry, 41, 643-649.
Feldman, S. S., & Elliott, G. R. (1990). At the threshold: Liker, J. K., & Elder, G. H., Jr. (1983). Economic hard-
The developing adolescent. Cambridge, MA: Harvard ship and marital relations in the 1930s. American
University Press. Sociological Review, 48, 343-359.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
360 Journal of Marriage and the Family

Luthar, S. S. (1991). Vulnerability and resilience: A Rutter, M. (1979). Protective factors in children's re-
study of high-risk adolescents. Child Development, sponses to stress and disadvantage. In M. W. Kent &
62, 600-616. J. E. Rolf (Eds.), Primary prevention of psychopathol-
Maccoby, E. E. (1992). The role of parents in the social- ogy: III. Promoting social competence and coping in
ization of children: An historical overview. Develop- children (pp. 49-74). Hanover, NH: University Press
mental Psychology, 28, 1006-1010. of New England.
Maccoby, E. E., & Martin, J. A. (1983). Socialization in Schulderman, E., & Schulderman, S. (1970). Replicabil-
the context of the family: Parent-child interaction. In ity of factors in children's report of parent behavior
P. H. Mussen (Series Ed.) & E. M. Hetherington (Vol. (CRPBI). Journal of Psychology, 96, 15-23.
Ed.), Handbook of child psychology: Vol. 4. Social- Shaw, D. S., & Emery, R. E. (1988). Chronic family ad-
ization, personality, and social development (4th ed., versity and school-age children's adjustment. Journal
pp. 1-101). New York:Wiley. of the American Academy of Child and Adolescent
Masten, A. S., Garmezy,N., Tellegen, A., Pellegrini, D. S., Psychiatry, 27, 200-206.
Larkin, K., & Larsen, A. (1988). Competence and Siefer, R., Sameroff, A. J., Baldwin, C. P., & Baldwin,
stress in school children: The moderating effects of A. L. (1992). Child and family factors that ameliorate
individual and family qualities. Journal of Child Psy- risk between 4 and 13 years of age. Journal of the
chology and Psychiatry, 29, 745-764. American Academy of Child and Adolescent Psychia-
Maughan, B., & Rutter, M. (1998). Continuities and dis- try, 31, 893-903.
continuities in antisocial behavior from childhood to Simons, R. L. (1996). Understandingdifferences between
adult life. In T. H. Ollendick & R. J. Prinz (Eds.), Ad- divorced and intact families. Thousand Oaks, CA:
vances in clinical child psychology (Vol. 20, pp. 1-47). Sage.
New York:Plenum Press. Slater, C. M., & Hall, G. E. (Eds.). (1996). County and
McCubbin, H., & Patterson,J. (1983). Family transitions: city extra annual metro city and county data book.
Adaptation to stress. In H. McCubbin & C. Figley Lanham, MD: Beman.
(Eds.), Stress and the family: Coping with normative Takeuchi, D. T., Williams, D. R., & Adair, R. K. (1991).
transitions (pp. 5-25). New York:Brunner/Mazel. Economic stress in the family and children's emo-
McLoyd, V. C. (1990). The impact of economic hard- tional and behavioral problems. Journal of Marriage
ship on Black families and children: Psychological and the Family, 53, 1031-1041.
distress, parenting, and socioemotional development. Tittsworth, S. P. (1996). Overt conflict style and youth
Child Development, 61, 311-346. academic achievement: Thefunction of importantlink-
O'Brien, M., Margolin, G., & John, R. S. (1995). Relation ing mechanisms. Unpublished master's thesis, Univer-
among marital conflict, child coping, and child adjust- sity of Tennessee, Knoxville.
ment. Journal of Clinical Child Psychology, 24, 346- Trotter, B. B. (1989). Coparental conflict, competition,
361. and cooperation and parents' perceptions of their chil-
Patterson, G. R. (1982). Coercive family process: A fam- dren's social-emotional well-being following marital
ily learning approach (Vol. 3). Eugene, OR: Castalia. separation. Unpublished doctoral dissertation, Univer-
Patterson, G. R., DeBaryshe, B. D., & Ramsey, E. sity of Tennessee, Knoxville.
(1989). A developmental perspective on antisocial be- U.S. Bureau of the Census. (1996). County and city data
havior. American Psychologist, 44, 329-335. books. Washington, DC: U.S. Government Printing
Pearlin, L. I., Lieberman, M. A., Menaghan, E. G., & Office.
Mullan, J. T. (1981). The stress process. Journal of U.S. Department of Health and Human Services. (1994,
Health and Social Behavior, 22, 337-356. February 10). Annual update of the Health and Human
Pine, D. S., Cohen, P., Gurley, D., Brook, J., & Ma, Y. Services poverty guidelines (p. 6277). Federal Register.
(1998). The risk for early-adulthood anxiety and de- U.S. Department of Health and Human Services. (1995,
pressive disorders in adolescents with anxiety and de- February9). Annual update of the Health and Human
pressive disorders.Archives of General Psychiatry, 55, Services poverty guidelines (p. 7772). Federal Register.
56-64. Wallerstein, J. S., & Kelly, J. B. (1980). Surviving the
Reynolds, W. M. (1992). Internalizing disorders in chil- breakup: How children and parents cope with divorce.
dren and adolescents. New York:Wiley. New York:Basic Books.
Reynolds, W. M. (1995). Depression. In V. B. VanHas- Weinberger,D. A., Feldman, S. S., & Ford, M. E. (1989).
selt & M. Hersen (Eds.), Handbook of adolescent Validation of the WeinbergerParenting Inventoryfor
psychopathology: A guide to diagnosis and treatment preadolescents and their parents. Unpublished manu-
(pp. 297-348). New York:Lexington Books. script.
Robins, L. N., & McEvoy, L. (1990). Conduct problems Werner, E. E., & Smith, R. S. (1992). Overcoming the
as predictors of substance abuse. In L. Robins & M. odds: High risk children from birth to adulthood.
Rutter (Eds.), Straight and devious pathways from Ithaca, NY: Cornell University Press.
childhood to adulthood (pp. 182-204). Cambridge, Williams, S., Anderson, J., McGee, R., & Silva, P. A.
England: Cambridge University Press. (1990). Risk factors for behavioral and emotional dis-
Rollins, B. C., & Thomas, D. L. (1979). Parentalsupport, order in preadolescent children. Journal of the Ameri-
power, and control techniques in the socialization of can Academy of Child and Adolescent Psychiatry, 29,
children. In W. R. Burr, R. Hill, F. I. Nye, & I. L. 413-419.
Reiss (Eds.), Contemporary theories about the family Wills, T. A., Vaccaro, D., & Benson, G. (1995). Coping
(Vol. 1, pp. 317-364). New York:Free Press. and competence in adolescent alcohol and drug use. In
Rutter,M. (1971). Parent-childseparation:Psychological J. L. Wallander& L. J. Siegel (Eds.), Adolescent health
effects on the children. Journal of Child Psychology problems: Behavioral perspectives (pp. 160-178). New
and Psychiatry, 12, 233-260. York:The Guilford Press.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions
Family Risk Factors 361

APPENDIX
DESCRIPTIVE STATISTICSAND CORRELATIONS

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
1. Overtconflict .46 -.25 .02 -.04 -.15 .34 .32 .30 .14 .14 .08
2. Parentingquality .12 .13 .14 -.13 .51 .52 .42 .17 .18 .09
3. Economic hardship .00 -.32 -.21 .13 .13 .13 .24 .13 .33
4. Youthgender .00 .11 .06 .16 -.03 .12 .14 .03
5. Youthgrade .05 .02 .04 .00 -.19 -.15 -.16
6. Parent'smaritalstatus -.18 -.18 -.17 -.09 -.07 -.09
7. Youthreporttotal problem
behavior .91 .93 .27 .24 .21
8. Youthreportexternalizing
behavior .70 .24 .26 .15
9. Youthreportinternalizing
behavior .25 .21 .22
10. Teacherreporttotal problem
behavior .91 .74
11. Teacherexternalizing
behavior .40
12. Teacherinternalizing
behavior
M 1.4 1.45 NA NA NA NA 21.51 11.15 10.40 4.71 2.68 2.03
SD .57 .29 NA NA NA NA 17.17 8.78 9.82 8.94 6.57 4.00
Cronbach'salpha .85 .83 NA NA NA NA .92 .90 .92 .96 .95 .87
Note: n = 335. Correlationsabove .10 were significantat the p < .05 level. Correlationsabove .13 were significantat the
p < .01 level.

This content downloaded from 128.235.251.160 on Thu, 15 Jan 2015 22:00:14 PM


All use subject to JSTOR Terms and Conditions

You might also like