identifying the psychological factors that mediate the association between parenting practices and...

37
\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 1 28-AUG-07 13:43 IDENTIFYING THE PSYCHOLOGICAL FACTORS THAT MEDIATE THE ASSOCIATION BETWEEN PARENTING PRACTICES AND DELINQUENCY* RONALD L. SIMONS Department of Sociology University of Georgia LESLIE GORDON SIMONS Department of Child and Family Development University of Georgia YI-FU CHEN Center for Family Research University of Georgia GENE H. BRODY Department of Child and Family Development University of Georgia KUEI-HSIU LIN Department of Sociology National Taipei University KEYWORDS: parenting, delinquency, theory Past research has largely ignored the developmental changes within the child that account for the association between parenting and risk for delinquency. We used structural equation modeling and data from a longitudinal study of several hundred African-American families to test the contentions of various theories regarding the sociocognitive and * This research was supported by the National Institute of Mental Health (MH48165, MH62669) and the Center for Disease Control (029136-02). Additional funding for this project was provided by the National Institute on Drug Abuse and the National Institute on Alcohol Abuse and Alcoholism. Direct correspondence to Dr. Ronald L. Simons, Department of Sociology, University of Georgia, Athens, GA 30602 (e-mail: [email protected]). CRIMINOLOGY VOLUME 45 NUMBER 3 2007 481

Upload: ronald-l-simons

Post on 21-Jul-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 1 28-AUG-07 13:43

IDENTIFYING THE PSYCHOLOGICALFACTORS THAT MEDIATE THEASSOCIATION BETWEEN PARENTINGPRACTICES AND DELINQUENCY*

RONALD L. SIMONSDepartment of SociologyUniversity of Georgia

LESLIE GORDON SIMONSDepartment of Child and Family DevelopmentUniversity of Georgia

YI-FU CHENCenter for Family ResearchUniversity of Georgia

GENE H. BRODYDepartment of Child and Family DevelopmentUniversity of Georgia

KUEI-HSIU LINDepartment of SociologyNational Taipei University

KEYWORDS: parenting, delinquency, theory

Past research has largely ignored the developmental changes within thechild that account for the association between parenting and risk fordelinquency. We used structural equation modeling and data from alongitudinal study of several hundred African-American families to testthe contentions of various theories regarding the sociocognitive and

* This research was supported by the National Institute of Mental Health(MH48165, MH62669) and the Center for Disease Control (029136-02).Additional funding for this project was provided by the National Institute onDrug Abuse and the National Institute on Alcohol Abuse and Alcoholism.Direct correspondence to Dr. Ronald L. Simons, Department of Sociology,University of Georgia, Athens, GA 30602 (e-mail: [email protected]).

CRIMINOLOGY VOLUME 45 NUMBER 3 2007 481

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 2 28-AUG-07 13:43

482 SIMONS, SIMONS, CHEN, BRODY & LIN

emotional factors that mediate the impact of parental behavior on ayouth’s risk for delinquency. Our findings largely supported the theo-ries. The impact of monitoring/discipline was indirect through low self-control and acceptance of deviant norms, whereas the effect of hostility/rejection was indirect through low self-control, hostile view of relation-ships, and acceptance of deviant norms. These two dimensions ofparenting were no longer related either to affiliation with deviant peersor to conduct problems once the effects of these psychological charac-teristics were taken into account; the impact of these parenting practiceswas completely mediated by these four cognitive/affective variables.Contrary to expectation, however, these psychological factors did notmediate any of the relationship between caretaker involvement in anti-social behavior and child conduct problems.

In the past three decades, many studies have reported that childrenexposed to inept parenting practices are at risk for conduct problems anddelinquency. Findings indicate that parental hostility and rejection, inef-fective monitoring and discipline, and low parent–child involvement arethe most powerful predictors of child conduct problems and delinquencywith parental antisocial behavior exerting a more modest effect (Loeberand Stouthamer-Loeber, 1986; Reid, Patterson, and Snyder, 2002; Simons,Simons, and Wallace, 2004). Although a multitude of studies support theseconclusions, it should be noted that most research only investigates one ortwo aspects of parental behavior. Studies rarely simultaneously investigatethe effects of several dimensions of parenting. Hence, we have limitedinformation regarding the relative importance of various dimensions ofparenting in the etiology of delinquency (Loeber and Stouthamer-Loeber,1986; Sampson and Laub, 1993).

More importantly, however, past research has largely ignored thepsychosocial mechanisms whereby parental behavior influences a child’srisk of delinquency. As children grow older, they dramatically decreasethe amount of time that they spend with parents while increasing the timespent away from home with peers. By early adolescence, parents have lim-ited opportunity to directly supervise and control the behavior of theirchild. Therefore, as children grow older, parental influence changes fromthe consequences of direct control to the indirect effects of parental social-ization. Whether deliberately or inadvertently, parents foster attitudes,beliefs, and emotions that influence the actions of their offspring whileaway from home. These cognitions and feelings mediate the effect ofparental behavior on a child’s risk of involvement in antisocial behavior;they are the psychosocial mechanisms whereby parents influence the

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 3 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 483

probability of their child becoming delinquent. Given this fact, our under-standing of the processes that link parenting and delinquency requires thatwe discover the psychosocial mechanisms that mediate this relationship.

A few criminological theories have attempted to identify the factors thataccount for the association between quality of parenting and delinquency.For example, the General Theory of Crime (Gottfredson and Hirschi,1990) emphasizes low self-control, General Strain Theory (Agnew, 1985,2001) points to feelings of anger, and Age-Graded Informal Social ControlTheory (Sampson and Laub, 1993) stresses the importance of attachmentto parents and commitment to conventional activities. Unfortunately, onlya handful of studies have investigated the mediating effects of these vari-ables. Also, no study has examined the influence of any one of thesemechanisms while controlling for the effects of the others. Thus we cur-rently have no information regarding the relative importance of these vari-ous cognitive and emotional factors as mediators of the impact of parentalbehavior.

The current study is an attempt to address these limitations of pastresearch. Using data from a longitudinal study of several hundred African-American families, we examine the extent to which various cognitive andemotional factors mediate the effect of parental behavior on the involve-ment of the child in delinquency. The target children in our investigationwere 10–12 years of age at wave 1 and 12–14 years of age at wave 2. Thus,our focus is on early involvement in delinquency. Past studies indicate thatit is these early starters who are at risk for serious adolescent delinquencyand adult crime (Moffitt, 1997; Patterson and Yoerger, 2002). Further-more, recent theorizing suggests that the causes of early starter antisocialbehavior may differ from those of late starter antisocial behavior (Chunget al., 2002; Moffitt and Caspi, 2001, 2002; Wiesner and Capaldi, 2003).The persistent deviance of early starters seems to be rooted, at least inpart, in relatively stable personal characteristics that differentiate themfrom conventional youth, whereas the experimental, short-lived deviantbehavior of late starters tends to be largely a function of peer influences. Ifthis is true, it is imperative that early and late starters not be confoundedin analyses (Moffitt, 1997). By focusing on conduct problems that occurbefore age 14, we are by definition limiting our focus to early starters (Pat-terson and Yoerger, 2002). We examine the extent to which variousparenting practices give rise to personal characteristics that account for thedelinquency of these youth.

Our measures of parenting include parental involvement, monitoringand consistent discipline, hostility and rejection, and parental antisocialbehavior. We test the extent to which psychosocial factors such as self-control, anger, a hostile view of relationships, unconventional norms, andconventional goals and values serve to mediate the effect of these various

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 4 28-AUG-07 13:43

484 SIMONS, SIMONS, CHEN, BRODY & LIN

parenting variables. Our selection of parenting behaviors and potentialmediating mechanisms was guided by past theory and research on familyprocesses and delinquency. This literature is discussed in the next section.

MEDIATORS OF PARENTAL BEHAVIOR

Antisocial acts, like conforming actions, are imbued with meaning(Giordano, Cernkovich, and Rudolph, 2002). Given their definition of thesituation, delinquent youth view their deviant behavior as appropriate andlegitimate, if not required or necessary (Anderson, 1999; Cairns, 1979;Katz, 1988). This claim suggests that important differences must existbetween conforming and delinquent youth with regard to those factorsthat shape their definitions of situations. Many social psychological studiesby sociologists (Cook, Fine, and House, 1995) and psychologists (Mischel,2004; Mischel and Shoda, 1995) have investigated the underlying psycho-logical processes that determine how situations are interpreted. In largemeasure, researchers from both disciplines agree that five general factorsare particularly important. The eminent psychologist Walter Mischel(2004) refers to these five factors as the cognitive–affective processingsystem.

The first factor includes the personal traits of individuals or their charac-teristic style of engaging the environment, such as compulsivity, extraver-sion, and self-control (Mischel, 2004). Second, people tend to possessschemata (Howard, 1995; Mischel, 2004) regarding the nature of self (e.g.,level of self-worth or self-efficacy) and others (e.g., the extent to whichpeople are trustworthy, selfish, or supportive). Developmental psycholo-gists often refer to these schemata as “working models of relationships”(Bowlby, 1969; Lyons-Ruth, 1996). Third, the perception of an event orcircumstance is influenced by the emotional state of the individual (How-ard, 1995; Isen, 1987). Depression, for example, causes individuals to cueto the negative and pessimistic features of their environment (Abramsonet al., 2002; Beck and Weishaar, 2000), whereas anger leads to the percep-tion of others as frustrating and as having bad intentions (Berkowitz,1990). Fourth, the personal goals and values of individuals color how theyview a situation and the lines of action that they find compelling (Mischel,2004; Schuman, 1995). Goals and values guide and structure the long-termprojects that individuals pursue and, therefore, the motivations that theybring into a situation. Finally, the view of a situation and response to asituation by people is circumscribed by their beliefs regarding the socialnorms that are applicable and the likely consequences of various lines ofaction (Ajzen, 1991; Schuman, 1995). These five factors are thought to

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 5 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 485

influence the stimuli or cues that are selected for attention, the signifi-cance ascribed to these stimuli, and the path of action chosen from theperceived alternatives.

In contrast to conforming youth, delinquents tend to define situations ina manner that justifies aggressive and antisocial behavior. The social psy-chological literature just reviewed suggests that these differences in per-ception are most likely rooted in discrepancies between the two groupsregarding the elements we have identified, based on the work of Mischel(2004), as the cognitive–affective processing system. In contrast to theirconventional peers, delinquents must possess personal traits, goals, norma-tive commitments, emotional states, and a view of self and others that pro-mote situational meanings conducive to deviant behavior.

Interestingly enough, each major theory of antisocial behavior empha-sizes a particular element of the cognitive–affective processing system asthe psychological mechanism that fosters delinquency, and in mostinstances, a specific parenting practice is identified as a cause of the devel-opment of this factor. Thus, the various theories agree that parenting con-tributes to the risk of antisocial behavior by youth through its impact onelements of the cognitive–affective processing system, but they disagreeregarding the cognitive or affective characteristics and parenting practicesthat are most crucial. Table 1 shows how this is the case. Column 1presents the five general factors that psychosocial research suggests arecritical in the interpretation of situations. Column 2 identifies particularexpressions of each of the five factors that have been posited as leading todelinquent behavior. Column 3 identifies the parenting practice that havebeen posited as giving rise to the characteristics identified in column 2, andcolumn 4 lists the various theories that make these conceptual arguments.

Beginning with the first row, table 1 shows that low self-control is apersonal trait that has been linked to delinquency, and that this argumentis posited in the Gottfredson and Hirschi (1990) General Theory of Crime(GTC). GTC argues that it is persons low in self-control who are attractedto delinquent and criminal behavior and asserts that children suffer fromlow self-control when they are raised by caretakers who fail to engage inconsistent monitoring and discipline. Thus the theory views lack of self-control as the personal characteristic that explains involvement in delin-quency and crime, and low self-control is considered to be a consequenceof insufficient parental control.

Consistent with the contentions of GTC, several studies have reportedthat low self-control predicts crime and delinquency. Recently, Pratt andCullen (2000) performed a meta-analysis of this research. Across 21 stud-ies, they found an effect size of .27, a coefficient that is far from unity.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 6 28-AUG-07 13:43

486 SIMONS, SIMONS, CHEN, BRODY & LIN

Table 1. A Social Psychological Perspective LinkingParenting Practices to Risk for Delinquency

Factors that Characteristics Associated Relevant Theories ofShape the that Promote Parental Antisocial Behavior

Meaning of Meanings BehaviorSituations Conducive to

Delinquency

Personal Low self- Monitoring/ General Theorytraits control discipline of Crime

Emotional Anger/ Hostility/ General Strain Theorystate frustration rejection Coercion Theory

Frustration/AggressionTheory

Schema of Hostile view of Hostility/ Biased Attributionself and relationships rejection Perspectiveothers Supportive Attachment Theory

involvement

Behavioral Deviance seen Monitoring/ Social Learningnorms as acceptable discipline Theory

Antisocial Social Control Theorycaregiver

Goals and Commitment to Supportive Social Learningvalues conventional involvement Theory

goals Social Control Theory

Thus, although self-control may be an important predictor of deviance,other personal characteristics are clearly important as well.

A handful of studies have investigated the extent to which the relation-ship between exposure to inept parenting during childhood and later anti-social behavior is explained by low self-control (Burt, Simons, and Simons,2006; Cochran et al., 1998; Gibbs, Giever, and Martin, 1998; Hay, 2001;Polakowski, 1994). Overall, the evidence from these studies suggests thatself-control explains at most one quarter of the relationship betweenparental control and antisocial behavior. Such findings indicate that paren-tal behavior influences the risk for deviance by youth in more ways thansimply through its impact on self-control.

The second mediating mechanism listed in table 1 is the emotional stateof anger and frustration. This factor is emphasized by Berkowitz’s (1990)revision of the frustration aggression hypothesis, Agnew’s (1992, 2001)General Strain Theory, and Colvin’s (2000) Coercion Theory. All of these

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 7 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 487

theorists argue that aversive social relations foster anger and irritability,and that these feelings increase the risk of delinquency because they fosterbelligerence and explosiveness, lower inhibition and concern with negativeconsequences, and create a desire for retaliation and revenge.

Berkowitz does not discuss the locus of aversive events, but his theorywould suggest that events such as parental hostility are likely to cause chil-dren to become angry and aggressive. Agnew (1985, 1992) and Colvin(2000) are more explicit about the role of parental behavior as a source offrustration for children. Both of these theorists contend that harsh andrejecting disciplinary practices are strains that might be expected toincrease a child’s risk for anger and delinquency.

Several studies have demonstrated that feelings of anger increase theprobability that an individual will engage in delinquent behavior (Agnew,1985; Berkowitz, 1990; Mazerolle and Piquero, 1997, 1998; Simons,Simons, and Wallace, 2004). Unfortunately, there has been limited investi-gation of the extent to which anger mediates the relation between harsh,rejecting parenting practices and delinquency. Although Broidy (2001)reported that anger mediates most of the association between these twovariables, other evidence suggests that it mediates only a small portion ofthis relationship (Brezina, 1998; Piquero and Sealock, 2000).

A distrusting, hostile view of relationships is the third potential mediat-ing mechanism listed in table 1. Two theoretical perspectives from devel-opmental psychology—the Biased Attribution Model from Ken Dodge(1980, 1986) and Attachment Theory (Ainsworth et al., 1978; Bowlby,1969; Lyons-Ruth, 1996)—posit that hostile, rejecting parenting causeschildren to develop a biased view of people and relationships thatincreases the chances that they will display conduct problems. Dodge con-tends that antisocial youth tend to believe that people are untrustworthyand exploitive, and that aggression is therefore necessary and legitimate inorder to defend oneself. This perspective leads them to attribute malevo-lent motives to others and to assume that an intimidating, confrontationalstyle of interaction is necessary to avoid exploitation.

Consistent with this hypothesis, past research has shown that this viewof relationships is strongly held by both aggressive children and institu-tionalized delinquents (Dodge, Bates, and Pettit, 1990; Dodge and New-man, 1981; Slaby and Guerra, 1988). Furthermore, Dodge and hiscolleagues (Dodge, 1991; Dodge, Bates, and Pettit, 1990) have shown thatexposure to harsh, rejecting parenting is a major cause of children devel-oping this hostile view of relationships. However, their research suggeststhat this biased attributional style only mediates a portion of the relation-ship between abusive parenting and child antisocial behavior.

Attachment theory (Ainsworth et al., 1978; Bowlby, 1969; Lyons-Ruth,1996) argues that the parent–child relationship provides the child with an

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 8 28-AUG-07 13:43

488 SIMONS, SIMONS, CHEN, BRODY & LIN

internal working model of relationships that is used in interaction withothers. Children who grow up with a caring, nurturing caregiver are said tohave a secure attachment style. They are expected to possess an optimistic,trusting model of relationships and to engage in warm, cooperative inter-actions with other people (Bretherton, 1985). Those persons exposed toharsh, rejecting parenting, on the other hand, are considered to have aninsecure attachment style. They are expected to possess a hostile, dis-trusting model of relationships that causes them to approach others withsuspicion and aggression.

Consistent with the theory, past research has reported that parental nur-turance gives rise to a secure style of interaction, whereas harsh, rejectingparenting tends to foster an insecurely attached style of interaction (Ains-worth et al., 1978; van IJzendoorn, 1995), and this association holdsregardless of race/ethnicity or social class (Bakermans-Kranenburg et al.,2004). Furthermore, studies have linked an insecure style of interaction toan increased probability of conduct problems and delinquency (Greenberget al., 1991; Renken et al., 1989; van IJzendoorn, 1997). Unfortunately,past research has rarely investigated the extent to which insecure attach-ment (i.e., a hostile, distrusting model of relationships) mediates the effectof parental nurturance and harsh discipline on a child’s risk for delin-quency. The few studies that have examined this issue find that attachmentstyle mediates only a modest portion of this relationship (Lyons-Ruth etal., 1990; Lyons-Ruth et al., 1991).

The fourth factor that influences the responses of individuals to a situa-tion is their beliefs regarding the types of role-related actions that areexpected, necessary, or justified given the circumstances. Although mostyouth possess a fairly conventional view of the norms applicable to them-selves and others, delinquents often endorse a deviant perspective thatlegitimates the use of aggression, stealing, lying, and other deviant actions.Hirschi (1969), in his formulation of social control theory, contended thatcommitment to conventional morality is a strong deterrent to delinquency.Unfortunately, this theoretical statement was not very explicit about theparental behaviors that foster these beliefs. Akers (1985, 1998), in hisapplication of social learning principles to the study of deviance and crime,has been specific about the parenting practices that influence the norma-tive framework of a child. First, he contends that children acquire conven-tional moral beliefs to the extent that parents monitor their child’s conductand reinforce conventional behavior while punishing deviant actions. Sec-ond, he emphasizes the consequences of parental modeling. Children areseen as acquiring attitudes and beliefs supporting conventional activities tothe extent that such behavior is modeled by their parents, whereas theylearn to embrace deviant actions when their parents model antisocial con-duct. We are not aware of any studies that have examined the extent to

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 9 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 489

which a deviant normative framework mediates the effect of parentalmonitoring/discipline and parental antisocial behavior on delinquency.

Commitment to conventional goals is the fifth potential mediatingmechanism listed in table 1. Hirschi (1969), in his presentation of socialcontrol theory, identified conventional goal commitments (e.g., earninggood grades or going to college) as an important bond to society thatdeters involvement in delinquent behavior. Such goal investments, heargued, provide a “stake in conformity” as conduct problems and delin-quent behavior threaten the chances of achieving these outcomes. Overthe years, several studies have reported that commitment to conventionalgoals and activities is negatively related to delinquency (Akers and Sellers,2004; Simons, Simons, and Wallace, 2004). Although Hirschi was not veryexplicit about the process whereby youth acquire conventional commit-ments, his theory is usually interpreted as suggesting that supportiveparental involvement causes children to develop an attachment to parentsthat leads them to identify with the goals and aspirations that their parentshold for them. We are not aware of any studies that have directly testedthe idea that commitment to conventional values mediates the impact ofsupportive parenting on delinquency.

Summarizing, we have identified the dimensions of parental behaviorthat various theories consider to be determinants of conduct problems anddelinquency. We have also noted the psychosocial mechanisms that thesetheories identify as the linkages between these parenting practices andchild behavior problems. These theories are usually viewed as competingexplanations regarding the manner in which parental behavior increases achild’s risk for deviant behavior. As we have noted, however, each of thevarious theoretical explanations emphasizes a mediating cognitive oraffective factor that social psychological research has shown to influencethe way individuals interpret and respond to situations. This suggests thatthe various theories represent complementary, rather than competitive,explanations for conduct problems as each identifies an important elementin the mixture of factors that leads some youth to often perceive thatdelinquent behavior is an appropriate line of action. Thus we expect tofind support for all theoretical arguments.

We test this idea using two waves of data from a sample of several hun-dred African-American caretakers and their children. We assume that it ispersistent exposure to ineffective parenting that causes the relativelyenduring psychosocial characteristics identified in table 1 to develop overtime. Thus, in our analyses, we examine the extent to which parentingpractices summed across the two waves of data account for increases in thepsychosocial risk factors between waves. Then, we investigate whether

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 10 28-AUG-07 13:43

490 SIMONS, SIMONS, CHEN, BRODY & LIN

these psychosocial developments mediate the association between persis-tent exposure to ineffective parenting practices and increases in delin-quency between waves 1 and 2.

In performing our analyses, we control for family income and commu-nity disadvantage. Past research has shown that these two factors increasea child’s risk for delinquency and have a disruptive effect on parentalbehavior (Burton and Jarrett, 2000; Simons et al., 2002; White and Rogers,2000). Furthermore, it may well be that that they are related to somepsychosocial mechanisms included in our study. Therefore, we controlledfor these two factors in order to eliminate any confounding influence thatthey might be exerting on our findings.

One of the strongest and most consistent findings in criminologicalresearch is the association between delinquency and affiliation with devi-ant peers. Longitudinal studies suggest that, although there is a reciprocalrelationship between the two variables, the predominant causal flow isfrom affiliation with deviant peers to involvement in delinquent behavior(Elliott and Menard, 1996; Warr, 2002). Some have argued that part of theeffect of parental behavior on risk for delinquency is mediated by theimpact of parents on their child’s peer group affiliation (Conger andSimons, 1997; Simons et al., 2001; Warr, 2005). It is usually assumed thatparents influence peer affiliations by encouraging their child to join onepeer group over another (Brown et al., 1993) or by promoting participa-tion in conventional activities such as organized sports and other extracur-ricular activities (Ladd, Profilet, and Hart, 1992). In addition to theseavenues, however, it may be that they affect their child’s friendshipchoices indirectly through one or more of the psychosocial mechanismsidentified in table 1. It may be, for example, that individuals with a hostileview of relationships or low self-control tend to select delinquent friends.Thus, we include a measure of affiliation with deviant peers in our analy-ses in an effort to determine the extent to which the effect of the variousparenting behaviors and psychosocial factors upon risk for delinquency ismediated by involvement in a deviant peer group.

METHODS

SAMPLE

We tested the relationships posited by the various theories of devianceusing two waves of data from the Family and Community Health Study(FACHS), a multisite investigation of neighborhood and family effects onhealth and development (Conger et al., 2002; Simons et al., 2002). TheFACHS sample consists of several hundred African-American families liv-ing in Georgia and Iowa. Each family included a child who was in fifth

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 11 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 491

grade at the time of recruitment. Interviews were conducted with the tar-get child and his/her primary caregiver.

Families were recruited from neighborhoods that varied on demo-graphic characteristics, specifically racial composition (percent AfricanAmerican) and economic level (percent of families with children livingbelow the poverty line). Block groups (BGs) were used to identify neigh-borhoods. Using 1990 census data, BGs were identified in both Iowa andGeorgia in which the percent of African-American families was highenough to make recruitment economically practical (10 percent or higher),and in which the percent of families with children living below the povertyline ranged from 10 percent to 100 percent. Using these criteria, 259 wereidentified (115 in Georgia and 144 in Iowa). The study families were ran-domly selected and recruited from rosters of all African-American fami-lies in these BGs that had a fifth grader.

Most study families were recruited by telephone. However, afterrepeated unsuccessful attempts to make telephone contact, or if a poten-tial participant did not have a telephone, a staff member attempted tomake face-to-face contact. If the potential participant was no longer at theaddress, we asked neighbors for information regarding his or her newaddress.

Two waves of data were collected from the Georgia and Iowa familiesusing identical research procedures. The first wave was collected in 1998and the second in 2000. At wave 1, the participants were 867 African-American children (400 boys and 467 girls; 462 in Iowa and 405 in Geor-gia) and their primary caregivers. The children were 10 to 12 years old(mean of 10.5 years) at wave 1 of data collection. Seven hundred andthirty-eight of the children (361 boys and 418 girls) and their caregiverswere interviewed again at wave 2. This rate of response was 86 percent.Analyses indicated that the families who did not participate at wave 2 didnot differ significantly from those who did with regard to caregiver incomeand education or child’s age, gender, school performance, or delinquency.

A primary caregiver was defined as a person living in the same house-hold as the target child and who was responsible for most of the child’scare. Most (84 percent) of the primary caregivers were the target child’sbiological mother (6 percent were the child’s father, 6 percent were thechild’s grandmother). Their mean age was 37.1 years and ranged from 23to 80 years. They reported an average of 4.5 children living in their homes.Education ranged from less than high school (19 percent) to advancedgraduate degrees (3 percent). The mode was a high-school degree (41 per-cent). Median income for the study families was $29,756. No significantdifference was found in income or education of the primary caregiverbetween the Iowa and Georgia subsamples.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 12 28-AUG-07 13:43

492 SIMONS, SIMONS, CHEN, BRODY & LIN

PROCEDURES

Before data collection began, four focus groups in Georgia and four inIowa examined and critiqued the self-report instruments. Each group wascomposed of 10 African-American women who lived in neighborhoodssimilar to those from which the study participants were recruited. Thefocus groups and pilot tests did not indicate a need for changes in anyinstruments used in the current article.

To enhance rapport and cultural understanding, African-American uni-versity students and community members served as field researchers tocollect data from the families in their homes. Before data collection, theresearchers received 1 month of training in the administration of the self-report instruments. Two home visits, each of which lasted 2 hours, weremade to each family. The second visit occurred within 7 days of the firstvisit. During the first visit, informed consent was obtained; primarycaregivers consented to their own and their children’s participation, andthe children agreed to participate.

At each home visit, self-report questionnaires were administered to thecaregiver and the child in an interview format. Each interview was con-ducted privately between one participant and one researcher, with noother family members present. The instruments were presented on laptopcomputers. Questions appeared in sequence on the screen, which both theresearcher and the participant could see. The researcher read each ques-tion aloud, and the participant entered a response using the computerkeypad.

MEASURES

Our analyses used measures of parenting practices, psychosocial mecha-nisms, deviant peers, and conduct problems collected at waves 1 and 2.Children were considered to be the most valid sources of information con-cerning their own cognitive processes, feelings, and antisocial behavior.Parents often possess limited information concerning their child’s beliefs,emotions, and delinquency. Therefore, child self-reports were used to formour measures of the psychosocial mechanisms and delinquency. Parentingpractices, on the other hand, represent a very different type of phenome-non. They involve actions that are so habitual and routine that both par-ents and children often have a difficult time providing accuratedescriptions (Furman et al., 1989; Simons, Johnson, and Conger, 1994).Furthermore, reports from family members about family processes are aptto be biased by the emotional state, attributional style, or personality ofthe respondent (Lorenz et al., 1991). Given these problems, we combinedchild and parent reports to construct our parenting measures. This multi-method approach was assumed to provide a more comprehensive and

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 13 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 493

valid depiction of parental behavior than measures based only on a singlereporter. The specific measures used were as follows.

DEPENDENT AND INDEPENDENT VARIABLES

CONDUCT PROBLEMS

This construct was measured using child self-reports on the conduct dis-order section of the Diagnostic Interview Schedule for Children, Version 4(DISC-IV). The DISC-IV covers Diagnostic Statistical Manual-IV (Amer-ican Psychiatric Association, 1994) criteria for diagnoses. The DISC wasdeveloped over a 15-year period of research on thousands of children andparents, and it has demonstrated reliability and validity (Shaffer et al.,1993). Version IV became available in 1995 and represents a modest revi-sion of the DISC-III based on findings from the Methods for the Epidemi-ology of Child and Adolescent Mental Disorders (MECA) project(Shaffer et al., 1993). The conduct disorder section contains a series ofquestions regarding how often during the preceding year the respondentengaged in 26 deviant acts such as shoplifting, physical assault, lying, set-ting fires, cruelty to animals, vandalism, burglary, and robbery. The scalecan be used to construct symptom counts or diagnoses. Symptom countswere used in this study. The coefficient alpha was above .90 at both waves1 and 2.

PARENTING PRACTICES

The items for the scales were adapted from instruments developed forthe Iowa Youth and Families Project (IYFP) (Conger and Elder, 1994;Conger et al., 1992). These measures have been shown to have high valid-ity and reliability. For example, analyses from IYFP have shown that par-ent reports on these instruments correlate with child reports and withobserver ratings (Conger et al., 1992; Simons, 1996), and they predict vari-ous dimensions of child behavior across a several year period (Simons etal., 1998; Simons et al., 2001). Focus group feedback before data collectionindicated that these items are meaningful to African-American parentsand capture what they consider to be the important dimensions of effec-tive parenting.

To assess monitoring/discipline, the primary caregivers completed a 12-item scale consisting of five questions concerned with monitoring (e.g.,How often do you know who your child is with when he/she is away fromhome?) and seven with consistency of discipline (e.g., How often do youpunish your child for something at one time and then at other times notpunish him/her for the same thing?). The response format for the itemsranged from 1 (never) to 5 (always). These same items were reworded sothat the target child could use this scale to rate the parenting behavior of

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 14 28-AUG-07 13:43

494 SIMONS, SIMONS, CHEN, BRODY & LIN

the primary caregiver. The caregiver and child-report items were summedto form a composite 24-item measure of parental control. The coefficientalpha for the instrument was approximately .85 at both waves 1 and 2.Scale scores for the two waves were summed to form a measure of persis-tent monitoring and discipline. The coefficient alpha for this combinedmeasure was .75.

A composite measure of hostility/rejection was formed from a 3-itemscale completed by the primary caregivers (e.g., How often do you hit yourchild with a belt or paddle? How often do you lose your temper and yell atyour child?) and a 14-item scale completed by the target child (How oftendoes your parent push, grab, hit, or shove you? How often does your par-ent shout or yell at you?). To provide equal weighting of caregiver andchild responses, scores on the scales were standardized before summingthem. Reliability for the composite scale was approximately .70 at eachwave. Scale scores were summed across waves to obtain a measure of per-sistent hostility/rejection. The coefficient alpha was .79 for this combinedmeasure.

A measure of supportive involvement was formed by summing caregiverresponses to a seven-item scale that focused on the extent to which theytalk with their children about their activities and concerns and try to helpthem solve everyday problems (e.g., How often do you talk to your childabout the things that bother him or her? How often do you ask your childwhat he or she thinks before deciding on family matters that involve himor her?). The target child responded to these same seven items, and a com-posite measure was constructed by summing the caregiver and childreports. The coefficient alpha for the composite scale was approximately.85 at both waves. Scale scores were summed across waves to establish ameasure of persistent supportive involvement. The coefficient alpha forthis combined measure was .83.

CAREGIVER ANTISOCIAL BEHAVIOR

This construct was measured using 15 questions from the CompositeInternational Diagnostic Instrument (CIDI) developed at the Universityof Michigan for the NIMH National Comorbidity Study (Kessler, 1991). Avariety of studies have demonstrated the validity of the diagnostic classifi-cations rendered by the CIDI (Wittchen, 1994). The antisocial personalitysection of the instrument contains a series of questions regarding howoften during the preceding year the respondent had engaged in acts suchas lying, stealing, skipping work, physical assault, threatening someonewith a weapon, and hitting their partner. The coefficient alpha for this 29-item symptom scale was approximately .80 at each wave. A measure ofpersistent involvement in antisocial behavior was obtained by summing

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 15 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 495

the scores for waves 1 and 2. The coefficient alpha for this combined mea-sure was .75.

MEDIATING VARIABLES

LOW SELF-CONTROL

A measure of low self-control was formed by combining the 17 itemsfrom the Kendall and Williams (1982) inventory of self-constraint (e.g., Ifyou find that something is really difficult, you get frustrated and quit; Youwould rather have a small gift today than a large gift tomorrow) with theEysenck and Eysenck (1977) six-item scale of risk-taking tendency (e.g.,You enjoy taking risks; You would prefer doing something dangerousrather than sitting quietly). The resulting 23-item measure captured thevarious elements of low self-control described by Gottfredson and Hirschi(1990) in their General Theory of Crime (e.g., impulsivity, insensitivity,physicality, risk-taking, short-sightedness). The response format for theitems ranged from 1 (not at all true) to 3 (very true), and the coefficientalpha for the instrument was .77 at wave 1 and .81 at wave 2.

HOSTILE VIEW OF RELATIONSHIPS

A nine-item scale developed for the current project was used to assessthis construct (Simons et al., 2006). The items focus on the extent to whichthe respondent takes a cynical view of people’s motives (e.g., When peopleare friendly, they usually want something from you; You have often beenlied to) and believes that violence is often necessary to achieve respect andto obtain fair treatment (e.g., People will take advantage of you if youdon’t let them know how tough you are; If you don’t let people know youwill defend yourself, they will think you are weak and take advantage ofyou). The response format ranged from 1 (strongly disagree) to 4 (stronglyagree). The coefficient alpha for the scale was .66 at wave 1 and .72 atwave 2.

ANGER/FRUSTRATION

This construct was assessed using four items from the DISC-IV thatassess relatively enduring feelings of anger and frustration. The itemsfocused on how often the respondent feels grouchy, annoyed, mad, orunfairly treated. The response format for the items ranged from 1 (lessthan once a week) to 4 (nearly every day). The coefficient alpha for thismeasure was .64 at wave 1 and .75 at wave 2.

CONVENTIONAL GOALS

The target children responded to 11 items regarding their dedication toschool (e.g., You try very hard in school), educational plans (e.g., How far

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 16 28-AUG-07 13:43

496 SIMONS, SIMONS, CHEN, BRODY & LIN

would you like to go in school?), and concern with long-term goals (e.g., Ioften think about the goals I have for the future). The response format forthese items varied from three to five alternatives. Therefore the itemswere standardized before summing them to form a measure of commit-ment to conventional activities. The coefficient alpha for the scale was .71at wave 1 and .74 at wave 2.

ACCEPTABILITY OF DEVIANCE

The target children completed a four-item scale that asked how they feelabout kids their age having sex, smoking, drinking, or using drugs. Theresponse format for each item was as follows: 1) You think it is bad, 2) youthink it is neither bad nor good, 3) you think it is good, and 4) you think itis very good. The coefficient alpha for the scale was .68 at wave 1 and .77at wave 2.

AFFILIATION WITH DEVIANT PEERS

The target children self-reported their affiliation with deviant peers usingan instrument adapted from the National Youth Survey (Elliott, Huizinga,and Menard, 1989). They were asked how many of their close friends (1 =none, 3 = half, 5 = all) had engaged in each of 19 delinquent acts. The actsvaried from relatively minor offenses such as skipping school to more seri-ous violations such as stealing something worth more than $25. Theresponses to the items were summed to obtain a total score concerning theextent to which the respondents’ friends engage in deviant behavior. Thecoefficient alpha for the scale was .88 at both waves.

CONTROL VARIABLES

HOUSEHOLD INCOME

Primary caregivers and secondary caregivers (where appropriate)reported the amount of money that they had earned during the previousyear from employment, child support, government payments, and so forth.We summed these measures to generate household income.

COMMUNITY DISADVANTAGE

The 673 respondents used in the current analyses resided in 259 blockgroups. A community disadvantage measure was constructed by summingfive U.S. Census items for the block group in which each study familyresided: per capita income, proportion of adults receiving public assis-tance, proportion of households below the poverty level, proportion ofadults unemployed, and proportion of single mother households. Eachrespondent was assigned a community disadvantage score based on thesum of these items for his or her block group. Previous research has used

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 17 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 497

such variables to index community socioeconomic status (Sampson,Raudenbush, and Earls, 1997; Sucoff and Upchurch, 1998), and in a studyof African-American adults, a similar operationalization was found to be asensitive index of neighborhood disadvantage (Cutrona et al., 2000). Thecoefficient alpha for the measure was .84.

RESULTS

Complete data for the measures used in this article were available for673 families. At wave 2, most of the youth (76 percent of boys and 64percent of girls) reported that during the preceding year they had engagedin at least one behavior included on the DISC measure of conductproblems. The number of symptoms reported ranged from 0 to 26. Over-all, 23 percent of boys and 13 percent of girls scored above six. The mostfrequently reported behaviors involved theft, fighting, and difficulties atschool.

Table 2 presents the correlation matrix for the study variables. The tableshows that, with one exception, there are significant relationships betweenthe parenting and the psychosocial variables. The exception involves Anti-social Caregiver, which does not show a significant association with anypsychosocial variables. Furthermore, all psychosocial and parenting vari-ables are significantly related to the wave 2 assessments of both Affiliationwith Deviant Peers and Conduct Problems. The only exception is Antiso-cial Caregiver, which is related to Conduct Problems but not to Affiliationwith Deviant Peers. Finally, a significant correlation exists between Affili-ation with Deviant Peers and Conduct Problems.

Statistically, a variable may be said to function as a mediator to theextent that it accounts for the association between a predictor and someoutcome. As noted by Baron and Kenny (1986) in their classic article, thefollowing conditions must be met to establish a mediation model. First,there must be a significant association between the predictor and the out-come variable. Second, the mediator must be significantly related to boththe predictor and the outcome variable. And, finally, controlling theeffects of the mediator must significantly reduce the association betweenthe predictor and the outcome variable. We used structural equation mod-eling (SEM) (MPLUS 3.0, Muthen and Muthen, 2004) to establish theextent to which these conditions were met for the hypothesized mediatorspresented in table 1.

The use of SEM afforded several advantages. Chief among them is thatit provides tests of significance for indirect or mediational effects. Further-more, it allows one to correct for attenuation in the estimates of associa-tions between study variables caused by measurement error. This issue is

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 18 28-AUG-07 13:43

498 SIMONS, SIMONS, CHEN, BRODY & LIN

Tab

le 2

.C

orre

lati

on M

atri

x fo

r th

e St

udy

Var

iabl

es (

N =

673

)

12

34

56

78

910

1112

1314

1516

1. S

ex--

2. M

onito

ring/

disc

iplin

e-.

04--

3. S

uppo

rtiv

e in

volv

emen

t-.

05.5

0*

--

4. H

ostil

ity/r

ejec

tion

.04

-.22

*-.

23*

--

5. A

ntis

ocia

l car

egiv

er-.

04-.

12*

-.01

.10

*--

6. F

amily

inco

me

.02

-.12

*.0

4-.

07-.

05--

7. C

omm

unity

dis

adva

ntag

e.0

3-.

05.0

2.0

7.0

0-.

28*

--

8. L

ow s

elf-

cont

rol

.00

-.20

*-.

21*

.16

*.0

2.0

5.0

3--

9. H

ostil

e vi

ew o

f rel

atio

nshi

ps.0

7-.

11*

-.08

*.1

5*

.01

-.17

*.1

1*

.22

*--

10. A

nger

/frus

trat

ion

-.03

-.10

*-.

12*

.20

*-.

04-.

04.1

0*

.20

*.2

1*

--

11. C

onve

ntio

nal g

oals

-.06

.22

*.1

8*

-.08

*-.

02-.

02.0

1-.

26*

-.01

-.12

*--

12. D

evia

nce

acce

ptab

le-.

09*

-.13

*-.

09*

.08

*.0

4-.

02.0

8.1

3*

.11

*.0

8-.

14*

--

13. D

evia

nt p

eer

grou

pT

1.1

2*

-.19

*-.

07.1

9*

.00

-.13

*.1

3*

-.01

.10

*.0

7-.

03-.

01--

14. D

evia

nt p

eer

grou

p T

2.0

0-.

18*

-.11

*.2

2*

.05

-.07

.13

*.2

9*

.30

*.2

1*

-.13

*.2

5*

.36

*--

15. C

ondu

ct p

robl

ems

T1

.15

*-.

20*

-.19

*.2

7*

.01

-.09

*.1

0*

.14

*.1

2*

.11

*-.

14*

.14

*.3

3*

.36

*--

16. C

ondu

ct p

robl

ems

T2

.11

*-.

27*

-.21

*.2

7*

.15

*-.

07.0

7.3

2*

.23

*.3

5*

-.22

*.2

1*

.21

*.4

3*

.45

*--

Mea

n.4

614

377

51.4

1.0

00 †

0 †0 †

0 †0 †

22.6

223

.21.

712.

85S

tan

dar

d d

evia

tio

n.5

028

1713

.88

3.9

6.12

3.34

2.76

1.87

1.60

4.66

4.72

2.63

3.13

25,9

1222

,484

NO

TE

S: V

aria

ble

cons

ists

of

resi

dual

s ob

tain

ed b

y re

gres

sing

wav

e 2

scor

es o

n w

ave

1 sc

ores

.*

p <

.05

; †

The

mea

n is

0 a

s th

e va

riab

le c

onsi

sts

of t

he r

esid

uals

fro

m r

egre

ssin

g w

ave

2 sc

ores

on

wav

e 1

scor

es.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 19 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 499

important when testing for mediation, as a measurement error in a mediat-ing variable prevents one from totally controlling the effects of the media-tor on the dependent variable when estimating the effects of theindependent variable on the dependent variable (Baron and Kenny, 1986).Since all constructs in our model were measured by a single indicator,1 wefollowed the suggestion of Bollen (1989) and used the reliability of eachindicator to correct for attenuation caused by measurement error.

Table 3 presents the results obtained when our two outcomes, affiliationwith deviant peers and conduct problems, were regressed on the parentingconstructs and the control variables. The analyses control for wave 1assessments of each outcome so that the analyses examine the extent towhich the parenting constructs explain increases in affiliation with deviantpeers and conduct problems over time. The table reports the results forthe total sample, but it should be noted that we used the model stakingprocedure (Bollen, 1989) to examine the extent to which any of the coeffi-cients differed by sex of respondent. This process entailed testing for dif-ferences in chi-square between models that constrained the relationshipsfor the boys’ and girls’ models to be identical versus models that allowedspecific relationships to vary across groups. The only significant differencewas for the stability of conduct problems where the coefficient was .54 forgirls and .38 for boys.

The results in table 3 show that monitoring/discipline fosters a decreaseand hostility/rejection an increase in the two outcomes. An antisocialcaregiver is associated with an increase in conduct problems but is notassociated with affiliation with deviant peers. Finally, contrary to thebivariate findings reported in table 2, supportive involvement is notrelated to either outcome once the effects of the other parenting variablesare taken into account. Given these findings, the next step was to investi-gate the extent to which the effects of monitoring/discipline, hostility/rejection, and antisocial caregiver are mediated by the various cognitiveand affective variables presented in table 1.

To examine this issue, we expanded our model to include the cognitive

1. Potentially, our four parenting constructs could be treated as having two indica-tors by either considering child and parent reports as two separate indicators orconsidering the wave 1 and 2 measures as separate indicators. However, both ofthese approaches produce complex models that tax available degrees of freedom,especially when the analyses are separated by sex of child. Furthermore, twoindicator models are notoriously unstable. Finally, even if we could treat theparenting constructs as having two indicators, we still had only single indictorsfor the remaining constructs. Given these concerns, we decided that the bestapproach was to consider all constructs as having a single indicator and to correctfor attenuation by adjusting for the internal consistency of each measure.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 20 28-AUG-07 13:43

500 SIMONS, SIMONS, CHEN, BRODY & LIN

Table 3. Standardized Regression Coefficients RelatingAffiliation with Deviant Peers and ConductProblems to the Exogenous andEndogenous Variables (N = 673)

Affiliation with ConductDeviant Peers T2 Problems T2

Affiliation with deviant peers T1 .341* —Conduct problems T1 — .408*Monitoring/discipline –.124* –.175*Supportive involvement .007 .003Hostility/rejection .158* .148*Antisocial caregiver .046 .134*Sex –.040 .030Family income .032 .019Community disadvantage .102* .015Model fit indices Chi-square = 50.645, d.f. = 18, p = .000;

CFI = .959; RMSEA = .054 (.037,.071); SRMR = .037

*p < .05.

and affective factors as endogenous variables that might mediate the asso-ciation between parental behavior and the two outcomes. We estimatedthis model as fully recursive. The results are presented in table 4. The firstfour columns report the extent to which parental behavior predictsincreases in the cognitive and affective variables. These analyses used theresiduals obtained when each cognitive and affective variable assessed attime 2 was regressed upon its respective time 1 scores. We used thisapproach, rather than simply controlling for the time 1 scores for each ofthese variables, in order to save degrees of freedom. Again, we tested forsex differences using the model staking procedure, and as before, the onlysignificant difference was in the association between the wave 1 and wave2 assessments of conduct problems.

Table 4 shows that monitoring/discipline is associated with increasedself-control, whereas hostility/rejection is associated with decreased self-control. In addition, hostility/rejection is related to increases in a hostileview of relationships and anger/frustration, whereas monitoring/disciplineis associated with an increased commitment to conventional goals and adecreased acceptance of deviance. These results are roughly consistentwith the predictions of table 1. The glaring exception is the finding thatsupportive/involvement is not significantly related to changes in any cogni-tive or affective variables.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 21 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 501

Tab

le 4

.St

anda

rdiz

ed R

egre

ssio

n C

oeff

icie

nts

Rel

atin

g A

ffili

atio

n w

ith

Dev

iant

Pee

rs a

ndC

ondu

ct P

robl

ems

to t

he E

xoge

nous

and

End

ogen

ous

Var

iabl

es (

N =

673

)

Exo

geno

us V

aria

bles

D L

owD

Hos

tile

D A

nger

/D

Com

mit

D D

evia

nce

Aff

iliat

ion

Con

duct

Self

-V

iew

of

Frus

trat

ion

Con

vent

iona

lA

ccep

tabl

eD

evia

ntP

robl

ems

Con

trol

Rel

atio

nshi

psG

oals

Pee

rs T

2T

2

Con

duct

pro

blem

s T

1.3

72*

Mon

itor

ing/

disc

iplin

e–.

161*

–.06

3–.

043

.243

*–.

156*

–.01

5–.

084

Supp

orti

ve i

nvol

vem

ent

–.10

8–.

004

–.05

0.0

45.0

02.0

28.0

10H

osti

lity/

reje

ctio

n.1

07*

.138

*.2

08*

–.01

5.0

52.0

78.0

23A

ntis

ocia

l ca

regi

ver

–.01

4–.

019

–.07

9.0

16.0

07.0

39.1

66*

Sex

–.01

8.0

63–.

042

–.04

4–.

101*

–.04

6.0

64**

Fam

ily i

ncom

e.0

98*

–.14

3*.0

04–.

052

.027

.022

–.00

5C

omm

unit

y di

sadv

anta

ge.0

44.0

52.0

84*

.002

.074

.041

–.03

2L

ow s

elf-

cont

rol

.199

*.1

46*

Hos

tile

vie

w o

f re

lati

onsh

ips

.188

*.0

42A

nger

/fru

stra

tion

.086

*.2

55*

Con

vent

iona

l go

als

–.02

7–.

056

Dev

ianc

e ac

cept

able

.197

*.0

76*

Aff

iliat

ion

wit

h de

vian

t pe

ers

.375

*.2

00*

Mod

el f

it i

ndic

esC

hi-s

quar

e =

95.

474,

d.f.

= 3

0, p

= .

000;

CF

I =

.94

5; R

MSE

A =

.06

0 (.

047,

.07

4);

SRM

R =

.03

7

*p <

.05

; **

p <

.10

.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 22 28-AUG-07 13:43

502 SIMONS, SIMONS, CHEN, BRODY & LIN

Column 6 in table 4 reports the extent to which all exogenous andendogenous variables in the model are related to affiliation with deviantpeers at time 2, controlling for time 1 scores on this outcome. Thus, again,our focus is on explaining change over time. The table shows that all cog-nitive and affective variables are related to changes in affiliation with devi-ant peers except commitment to conventional goals. Furthermore,although both monitoring/discipline and hostility/rejection were related tochange in affiliation with deviant peers in table 3, neither of these parent-ing variables has an effect once the cognitive/affective variables are in themodel.

Column 7 reports the extent to which the various exogenous and endog-enous variables, including affiliation with deviant peers, are associatedwith change in conduct problems. Four endogenous variables—low self-control, anger/frustration, deviance acceptable, and affiliation with deviantpeers—show significant associations. Turning to the parenting variables,the effects of monitoring/discipline and hostility/rejection on change inconduct problems (see table 3) are no longer significant after the cognitiveand affective variables are included. This is not the case, however, for anti-social caretaker, which remains significant. The coefficient for this variableis actually slightly larger than was the case in table 3.

To obtain a more parsimonious model, we performed the analyses againbut only included paths with a t > 1.5. The chi-square difference betweenthe fully recursive and reduced models was not significant (Dc2

(38) = 35.146,p = 1.00). The reduced model is presented in figure 1. The figure providesa visual depiction of the pattern of findings reported in table 4, whichmakes it easier to discern the avenues whereby the parenting variablesproduce their effects. The figure indicates that monitoring/discipline hasan indirect effect on both affiliation with deviant peers and conductproblems through its association with low self-control and devianceacceptable. Harsh/rejecting parenting shows an indirect effect on affilia-tion with both deviant peers and conduct problems through its impact onlow self-control and anger/frustration. In addition, it influences affiliationwith deviant peers through its association with a hostile view ofrelationships.

Table 5 reports that all of these indirect effects are significant.2 Further-more, the figure shows that contrary to the findings for the other parentingvariables, none of the effects of antisocial caregiver are indirect through

2. MPLUS provides two kinds of tests for the significance of indirect effects: thedelta and bootstrapping methods. We performed both tests and found the resultswere consistent across methods. The t-values derived from the delta method arereported in table 5. Note that the indirect effect of monitoring/discipline throughdeviance acceptable is only significant using a one-tailed test. However, such atest is justified, as the direction of this effect was predicted to be negative.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 23 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 503

Fig

ure

1.R

educ

ed S

truc

tura

l E

quat

ion

Mod

el (

N =

673

)

.09*

.1

6*

-.16

*

.23* .2

8* .11*

-.17

*

Sex

Mon

itori

ng/

Dis

cipl

ine

Sup

port

ive

Invo

lvem

ent

Hos

tility

/ R

ejec

tion

Ant

isoc

ial

Car

egiv

er

Com

mun

ity

Dis

adva

ntag

e

Hou

seho

ldIn

com

e

Low

Sel

f-C

ontr

ol

Hos

tile

Vie

wof

Rel

atio

nshi

ps

Ang

er/

Frus

trat

ion

Con

vent

iona

lG

oals

Dev

iant

Peer

s T

2(.

36)

Con

duct

Prob

lem

s T

2(.

53)

.19*

Chi

-squ

are

= 1

30.6

2, d

.f. =

68

CFI

= .9

5 R

MSE

A =

.04

(.03

, .05

) R

2 in p

aren

thes

is

+p

< .1

0, *

p <

.05

∆χ

2 (38)

=35

.146

(p

= 1

.00)

Dev

ianc

eA

ccep

tabl

e

-.22

*

-.12

*

.07+

-.08

+

.20*

.18*

.09*

.21* .0

7+.1

8*

.27*

.08*

.22*

.07+

.37*

Con

duct

Prob

lem

s T

1

.40*

.08+

Dev

iant

Peer

sT

1

NO

TE

: B

old

lines

ind

icat

e pa

ths

sign

ific

ant

at p

< .

05.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 24 28-AUG-07 13:43

504 SIMONS, SIMONS, CHEN, BRODY & LIN

Table 5. Significance of the Indirect Effects Relating theParenting Variables to Affiliation with DeviantPeers and Conduct Problems (N = 673)

Predictors Mediators Affiliation with ConductDeviant Peers Problems

Monitoring/ D Low self-control –.045 (t = –3.125) –.039 (t = –3.316)discipline D Hostile view of relationships — —

D Anger/frustration —D Conventional goals — —D Deviance acceptable –.035 (t = –3.125) –.013 (t = –1.786)

Supportive D Low self-control — —involvement D Hostile view of relationships — —

D Anger/frustration — —D Conventional goals — —D Deviance acceptable — —

Hostility/rejection D Low self-control .024 (t = 2.323) .021 (t = 2.237)D Hostile view of relationships .030 (t = 2.953) —D Anger/frustration .021 (t = 2.121) .060 (t = 4.175)D Conventional goals — —D Deviance acceptable — —

Antisocial D Low self-control — —caregiver D Hostile view of relationships — —

D Anger — —D Conventional goals — —D Acceptance of convention — —

D Low self-control Affiliation with deviant peers — .044 (t = 3.643)D Hostile view ofrelationships Affiliation with deviant peers — .040 (t = 3.482)D Anger/frustration Affiliation with deviant peers — .019 (t = 2.091)D Conventional goals Affiliation with deviant peers — —D Devianceacceptable Affiliation with deviant peers — –.045 (t = –2.034)

the cognitive/affective variables. Rather, there is a direct path from antiso-cial caregiver to conduct problems.

Finally, three cognitive/affective variables are related to conductproblems: low self-control, anger/frustration, and deviance acceptable.Affiliation with deviant peers is also related to conduct problems, and thefigure shows that these three cognitive/affective variables, as well as hos-tile view of relationships, influence conduct problems through their impacton affiliation with deviant peers. Table 5 reports that all four of these indi-rect effects are significant.

DISCUSSION

Although there is widespread acceptance of a link between quality ofparenting and delinquency, two important issues remain to be resolved.First, we have only limited information regarding those aspects of parentalbehavior that are most consequential for the development of child and

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 25 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 505

adolescent behavior problems. Studies rarely investigate simultaneouslythe effects of several parenting practices, and hence, we have little infor-mation regarding the relative importance of various dimensions of parent-ing in the etiology of delinquency. Our analyses indicated that persistentexposure to three caregiver patterns of behavior predicted increases inchild conduct problems: expressions of hostility and rejection, the absenceof monitoring and discipline, and involvement in antisocial activity. Theeffect size was roughly comparable for these three parenting behaviors.Two of these constructs—high parental hostility and rejection and lowparental monitoring and discipline—also predicted increased affiliationwith deviant peers. Surprisingly, parental support and involvement wasnot related to change in either affiliation with deviant peers or conductproblems once the effects of the other parenting variables were taken intoaccount. Although supportive involvement may be an important determi-nant of other child outcomes (e.g., self-esteem or school achievement), ourresults suggest that it has little impact on risk for delinquency.

Having established that these dimensions of caretaker behavior arerelated to a youth’s affiliation with deviant peers and delinquency, wewere concerned with identifying the psychological characteristics thatmediate these associations. As children grow older, there is a dramaticincrease in the amount of time that they spend away from home withpeers, thereby limiting the extent to which parents can directly superviseand control their behavior. During this time, parental influence becomesincreasingly indirect through the attitudes, beliefs, and emotions that theyhave intentionally or unintentionally fostered in their children. Some chil-dren acquire cognitive and affective characteristics that make antisocialbehavior an attractive, appropriate, or compelling response to many situa-tions. Gottfredson and Hirschi (1990) label these characteristics that pre-dispose a youth to crime and analogous activities as “criminality.”

As noted, Gottfredson and Hirschi assert that criminality is largely amatter of low self-control. Youth are attracted to criminal and deviantbehavior when their caretakers fail to provide the monitoring and supervi-sion necessary to teach the child self-control. We also noted, however, thatthe various theories of antisocial behavior differ regarding the psychologi-cal factor that they consider to be the mediator of the impact of parentingon risk for delinquency. Although Gottfredson and Hirschi (1990) stressthe importance of self-control, other theorists emphasize the importanceof anger (Agnew, 1992; Berkowitz, 1990), a hostile view of relationships(Dodge, 1980, 1986), acceptance of deviant norms (Akers, 1998), or com-mitment to conventional goals (Hirschi, 1969).

Although these various theories are usually seen as competing theoreti-cal perspectives, there is reason to believe that they might actuallyrepresent complementary perspectives. Findings from social psychology

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 26 28-AUG-07 13:43

506 SIMONS, SIMONS, CHEN, BRODY & LIN

indicate that a wide variety of psychological factors such as personal traits,emotions, view of relationships, and norms and values all combine to influ-ence the meaning that a person attributes to a situation. This result sug-gests that each of the theories of antisocial behavior may contain anelement of truth and that together they present a rather comprehensivepicture of the avenues whereby parental behavior influences a child’s riskfor delinquency.

Consistent with this idea, we found that four psychological characteris-tics—low self-control, a hostile view of relationships, anger/frustration,and acceptance of deviant norms—made a unique contribution in predict-ing conduct problems. Importantly, each of these variables mediated someportion of the effect of parenting practices on affiliation with deviant peersand conduct problems. Parental monitoring and discipline reducedinvolvement with deviant peers and delinquency by fostering self-controlas suggested by the General Theory of Crime (Gottfredson and Hirschi,1990), and by teaching norms that discourage deviant behavior as positedby the social learning and social control theories (Akers, 1998). However,although the General Theory of Crime emphasizes the importance ofparental monitoring and discipline to the exclusion of other parenting vari-ables, we found that parental hostility and rejection also influences risk ofaffiliation with deviant peers and delinquency through self-control. Hay(2001) recently reported a similar finding. In addition to its indirect effectthrough self-control, we found that parental hostility and rejection alsoincreased the chances of affiliation with deviant peers and delinquency byfostering anger and frustration as suggested by General Strain Theory(Agnew, 1992), Coercion Theory (Colvin, 2000), and Frustration Aggres-sion Theory (Berkowitz, 1990). Finally, consonant with both AttachmentTheory (Bowlby, 1969; Lyons-Ruth, 1996) and Biased Attribution Per-spective Theory (Dodge, 1980, 1986), parental hostility and rejectionincreased affiliation with a deviant peer group by fostering a hostile viewof relationships. Contrary to these two theories, however, a hostile view ofrelationships did not have a direct effect on delinquency. Rather, a hostileview of relationships only increased risk of delinquency indirectly byincreasing involvement with deviant peers.

The results did contain two surprises. First, although monitoring and dis-cipline predicted child commitment to conventional goals, commitment toconventional goals, contrary to both social learning (Akers, 1998) andsocial control theories (Hirschi, 1969), was not associated with either affili-ation with deviant peers or conduct problems. Past research has not evalu-ated the consequences of conventional goals while controlling for theother psychological variables in our model. Thus, it may be that conven-tional goals are simply not an important predictor of involvement withdeviant peers or delinquency once the effect of the other psychological

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 27 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 507

variables is taken into account. However, our measure of conventionalgoals was largely limited to educational commitments and aspirations. Per-haps a broader measure that included occupational, financial, religious,and family aspirations would have detected an effect.

Second, the psychological factors did not mediate any of the relation-ship between caretaker involvement in antisocial behavior and child con-duct problems. Based on Social Learning Theory (Akers, 1998), weexpected that a child’s beliefs regarding the acceptability of deviant behav-ior would mediate the association between caretaker antisocial behaviorand delinquency. Perhaps this association indicates that deviant behaviormodeled by caretakers conveys scripts for antisocial behavior that areenacted by youth with little conscious thought. What is more likely thecase, however, is that our measure of deviant norms, which focused on theacceptability of sex, smoking, drinking, and illicit drug use, was too narrowto detect the mediation effect. Caretaker antisocial behavior includedbehaviors such as fighting, lying, and stealing. We suspect that a measureof children’s beliefs about various deviant acts that included these behav-iors is likely to demonstrate the expected mediation effect.

In general, our findings indicate that each of the theories included in ourstudy identifies an avenue whereby parental behavior influences risk fordelinquency. These theories are often treated as competing perspectives asif only one of them can be correct. However, given the complexity ofhuman behavior, it would be truly amazing if a single psychological mech-anism, such as anger or low self-control, completely accounted for the linkbetween parenting and delinquency. And such a finding would be contraryto the wealth of psychosocial research showing that a wide variety of cog-nitive and emotional factors influence people’s definitions of, andresponses to, situations. Hence, we believe the various theories of deviantbehavior included in our analysis should be viewed as complimentaryframeworks to be combined to form a more comprehensive theory ofdelinquent behavior.

It should be noted, however, that our model only explained about 25percent of the variance in conduct problems and affiliation with deviantpeers. Presumably factors relating to neighborhood context account formuch of the remaining unexplained variance. Elsewhere we have arguedthat a comprehensive theory of antisocial behavior would combine bothfamily and community influences and would specify the psychologicalmechanisms whereby each exerts its effect (Simons et al., 2005). Thus, justas the current study investigated the psychological processes that mediateparental behavior, so there is a need for studies that examine the psycho-logical mechanisms whereby community factors exert their influence. Cur-rently, we have virtually no information regarding the cognitive andemotional processes that mediate the impact of community disadvantage,

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 28 28-AUG-07 13:43

508 SIMONS, SIMONS, CHEN, BRODY & LIN

collective efficacy, or community crime on risk for delinquency. Thisresearch might begin by investigating the possibility that the psychologicalmediators of community characteristics are the same as those that mediatethe impact of parental behavior (e.g., level of self-control, view of relation-ships, or deviant norms).

The data used in the current study afforded several advantages. Forexample, it was longitudinal and contained a variety of family and psycho-logical variables. It is also the case, however, that our analyses were ham-pered by certain limitations in the data. Perhaps the most glaring is thehomogeneity of the sample. All study families were African American. Inpart, this focus is warranted given the high incidence of delinquency andcrime in the African-Americans community (Gibbs, 1998). On the otherhand, this lack of diversity in our sample raises the question of whetherour findings generalize to other racial/ethnic groups. We cannot think ofany reason why the psychological mechanisms whereby parenting effectsdelinquency would vary by cultural group, but this is really an empiricalquestion. Studies with diverse samples are needed to establish the general-izability of our results.

Finally, it should also be noted that our findings are specific to earlyinvolvement in delinquency. The processes whereby parenting influenceslate onset delinquency may be quite different (Moffitt, 1997; Patterson andYoerger, 2002). Furthermore, it is undoubtedly the case that a different setof psychosocial factors accounts for the effect of parenting upon otherdevelopmental outcomes such as internalizing problems or academic suc-cess. Parental behavior may affect adolescent depression, for example,through its influence upon personal traits such as optimism or self-efficacy.Our analyses only speak to the psychological processes that link parentingto early onset antisocial behavior.

REFERENCES

Abramson, Lyn Y., Lauren B. Alloy, Benjamin I. Hankin, Gerald J. Haef-fel, Donald G. MacCoon, and Brandon E. Gibb. 2002. Cognitive vul-nerability-stress models of depression in a self-regulatory andpsychobiological context. In Handbook of Depression, eds. Ian H. Got-lib and Constance L. Hammen. New York: Guilford.

Agnew, Robert. 1985. A revised strain theory of delinquency. SocialForces 64:151–67.

Agnew, Robert. 1992. Foundation for a general strain theory of crime anddelinquency. Criminology 30:47–87.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 29 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 509

Agnew, Robert. 2001. Building on the foundation of general strain theory:Specifying the types of strain most likely to lead to crime and delin-quency. Journal of Research in Crime and Delinquency 38:319–61.

Ainsworth, Mary D. Salter, Mary C. Blehar, Everett Waters, and SallyWall. 1978. Patterns of Attachment: A Psychological Study of theStrange Situation. Hillsdale, NJ: Erlbaum.

Ajzen, Icek. 1991. The theory of planned behavior. Organizational Behav-ior and Human Decisions Processes 50:179–211.

Akers, Ronald L. 1985. Deviant Behavior: A Social Learning Approach.Belmont, CA: Wadsworth.

Akers, Ronald L. 1998. Social Learning and Social Structure: A GeneralTheory of Crime and Deviance. Boston, MA: Northeastern UniversityPress.

Akers, Ronald L., and Christine S. Sellers. 2004. Criminological Theories:Introduction, Evaluation, and Application, 4th ed. Los Angeles, CA:Roxbury.

American Psychiatric Association. 1994. Diagnostic and Statistical Manualof Mental Disorders, 4th ed., DSM-IV. Washington, DC: AmericanPsychiatric Association.

Anderson, Elijah. 1999. Code of the Streets: Decency, Violence, and theMoral Life of the Inner City. New York: W. W. Norton & Company.

Bakermans-Kranenburg, Marian J., Marinus H. van IJzendoorn, and Pie-ter M. Kroonenberg. 2004. Differences in attachment security betweenAfrican-American and white children: Ethnicity or socio-economic sta-tus? Infant Behavior and Development 27:417–33.

Baron, Reuben M., and David A. Kenny. 1986. The moderator-mediatorvariable distinction in social psychological research: Conceptual, strate-gic, and statistical considerations. Journal of Personality and SocialPsychology 51:1173–82.

Beck, Aaron T., and Marjorie Weishaar. 2000. Cognitive therapy. In Cur-rent Psychotherapies, 6th ed., eds. Raymond J. Corsini and DannyWedding. Itasca, IL: Peacock.

Berkowitz, Leonard. 1990. On the formation and regulation of anger andaggression: A cognitive-associationistic analysis. American Psycholo-gist 45:494–503.

Bollen, Kenneth A. 1989. Structural Equations with Latent Variables. NewYork: Wiley.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 30 28-AUG-07 13:43

510 SIMONS, SIMONS, CHEN, BRODY & LIN

Bowlby, John. 1969. Attachment and Loss, vol. I, Attachment. New York:Basic Books.

Bretherton, Inge. 1985. Attachment theory: Retrospect and prospect.Monographs of the Society for Research in Child Development 50:3–35.

Brezina, Timothy. 1998. Adolescent maltreatment and delinquency: Thequestion of intervening processes. Journal of Research in Crime andDelinquency 35:71–99.

Broidy, Lisa. 2001. A test of general strain theory. Criminology 39:9–33.

Brown, B. Bradford, Nina Mounts, Susie D. Lamborn, and LaurenceSteinberg. 1993. Parenting practices and group affiliation in adoles-cence. Child Development 64:467–82.

Burt, Callie B., Ronald L. Simons, and Leslie G. Simons. 2006. A longitu-dinal test of the effects of parenting and the stability of self-control:Negative evidence for the General Theory of Crime. Criminology44:353–96.

Burton, Linda M., and Robin L. Jarrett. 2000. In the mix, yet on the mar-gins: The place of families in urban neighborhoods and child develop-ment research. Journal of Marriage and the Family 63:1114–35.

Cairns, Robert B. 1979. Social Development: The Origins and Plasticity ofInterchanges. San Francisco, CA: W.H. Freeman.

Chung, Ick-Joong, Karl G. Hill, J. David Hawkins, Lewayne D. Gilchrist,and Daniel S. Nagin. 2002. Childhood predictors of offense trajecto-ries. Journal of Research in Crime and Delinquency 39:60–90.

Cochran, John K., Peter B. Wood, Christine S. Sellers, Wendy Wilkerson,and Mitchell B. Chamlin. 1998. Academic dishonesty and low self-con-trol: An empirical test of a general theory of crime. Deviant Behavior19:227–55.

Colvin, Mark. 2000. Crime and Coercion: An Integrated Theory of ChronicCriminality. New York: St. Martins Press.

Conger, Rand D., and Glen H. Elder. 1994. Families in Troubled Times:Adaptation to Change in Rural America. New York: Aldine.

Conger, Rand D., and Ronald L. Simons. 1997. Life course contingenciesin the development of adolescent antisocial behavior: A matching lawapproach. In Developmental Theories of Crime and Delinquency, ed.Terence Thornberry. New Brunswick, NJ: Transaction.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 31 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 511

Conger, Rand D., Lora Ebert-Wallace, Y. Sun, Ronald L. Simons, VonnieMcLoyd, and Gene H. Brody. 2002. Economic pressure in AfricanAmerican families: A replication and extension of the Family StressModel. Developmental Psychology 38:179–93.

Conger, Rand D., Kathy J. Conger, Glen H. Elder, Frederick O. Lorenz,Ronald L. Simons, and Les B. Whitbeck. 1992. A family process modelof economic hardship and influences on adjustment of early adolescentboys. Child Development 63:526–41.

Cook, Karen S., Gary A. Fine, and James. S. House, eds. 1995. SociologicalPerspectives on Social Psychology. Boston, MA: Allyn and Bacon.

Cutrona, Carolyn E., Daniel W. Russell, Robert M. Hessling, and PhyllisA. Brown. 2000. Direct and moderating effects of community contexton the psychological well being of African American women. Journalof Personality and Social Psychology 79:1088–101.

Dodge, Kenneth A. 1980. Social cognition and children’s aggressivebehavior. Child Development 51:162–70.

Dodge, Kenneth A. 1986. A social information processing model of socialcompetence in children. In Minnesota Symposium in Child Psychology,ed. Michael P. Perlmuter. Hillsdale, NJ: Lawrence Erlbaum Associates.

Dodge, Kenneth A. 1991. The structure and function of reactive andproactive aggression. In The Development and Treatment of ChildhoodAggression, eds. Debra J. Pepler and Kenneth H. Rubin. Hillsdale, NJ:Lawrence Erlbaum Associates.

Dodge, Kenneth A., and Joseph P. Newman. 1981. Biased decision-mak-ing processes in aggressive boys. Journal of Abnormal Psychology90:375–97.

Dodge, Kenneth A., John E. Bates, and Gregory S. Pettit. 1990. Mecha-nisms in the cycle of violence. Science 250:1678–83.

Elliott, Delbert S., and Scott Menard. 1996. Delinquent friends and delin-quent behavior: Temporal and developmental patterns. In Delinquencyand Crime: Current Theories, ed. J. David Hawkins. New York: Cam-bridge University Press.

Elliott, Delbert S., David Huizinga, and Scott Menard. 1989. MultipleProblem Youth: Delinquency, Substance Use, and Mental HealthProblems. New York: Springer-Verlag.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 32 28-AUG-07 13:43

512 SIMONS, SIMONS, CHEN, BRODY & LIN

Eysenck, Sybil Bianca Giuletta, and Hans J. Eysenck. 1977. The place ofimpulsiveness and venturesomeness in a dimensional system of person-ality description. British Journal of Social and Clinical Psychology16:57–68.

Furman, Wyndol, Laura Jones, Duane Buhrmester, and Terry Adler. 1989.Children’s, parents’, and observers’ perspectives on sibling relation-ships. In Sibling Interaction Across Cultures: Theoretical and Method-ological Issues, ed. Patricia G. Zukow. New York: Springer-Verlag.

Gibbs, John J., Dennis Giever, and Jamie S. Martin. 1998. Parental man-agement and self-control: An empirical test of Gottfredson and Hir-schi’s general theory. Journal of Research in Crime and Delinquency35:40–70.

Giordano, Peggy C., Stephen A. Cernkovich, and Jennifer I. Rudolph.2002. Gender, crime, and desistance: Toward a theory of cognitivetransformation. American Journal of Sociology 107:990–1064.

Gottfredson, Michael R., and Travis Hirschi. 1990. A General Theory ofCrime. Stanford, CA: Stanford University Press.

Greenberg, Mark T., Matthew L. Speltz, Michelle DeKlyen, and Marya C.Endriga. 1991. Attachment security in preschoolers with and withoutexternalizing problems: A replication. Development and Psychopathol-ogy 3:413–30.

Hay, Carter. 2001. Parenting, self-control, and delinquency: A test of self-control theory. Criminology 39:707–36.

Hirschi, Travis. 1969. Causes of Delinquency. Berkeley, CA: University ofCalifornia Press.

Howard, Judith A. 1995. Social cognition. In Sociological Perspectives onSocial Psychology, eds. Karen S. Cook, Gary A. Fine, and James S.House. Boston, MA: Allyn and Bacon.

Isen, Alice M. 1987. Positive affect, cognitive processes, and social behav-ior. In Advances in Experimental Social Psychology, vol. XX. ed. L.Berkowitz. New York: Academic.

Katz, Jack. 1988. Seductions of Crime: Moral and Sensual Attractions inDoing Evil. New York: Basic Books.

Kendall, Philip C., and Carolyn L. Williams. 1982. Assessing the cognitiveand behavioral components of children’s self-management. In Self-Management and Behavior Change, eds. Paul Karoly and Fredrick H.Kanfer. New York: Pergamon Press.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 33 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 513

Kohlberg, Lawrence. 1984. Essays on Moral Development, vol. II, The Psy-chology of Moral Development. New York: Harper and Row.

Ladd, Gary W., Susan M. Profilet, and Craig H. Hart. 1992. Parents’ man-agement of children’s peer relations: Facilitating and supervising chil-dren’s activities in the peer culture. In Family-Peer Relationships:Modes of Linkage, eds. Ross D. Parke and Gary W. Ladd. Hillsdale,NJ: Lawrence Erlbaum Associates.

Loeber, Rolf, and Magda Stouthamer-Loeber. 1986. Family factors as cor-relates and predictors of juvenile conduct problems and delinquency.In Crime and Justice, vol. VII, eds. Michael Tonry and Norval Morris.Chicago: University of Chicago Press.

Lorenz, Frederick O., Rand Conger, Ronald L. Simons, Les B. Whitbeck,and Glen H. Elder. 1991. Economic pressure and marital quality: Anillustration of the method variance problem in the causal-modeling offamily processes. Journal of Marriage and the Family 53:375–88.

Lyons-Ruth, Karlen. 1996. Attachment relationships among children withaggressive behavior problems: The role of disorganized early attach-ment patterns. Journal of Consulting and Clinical Psychology 64:64–73.

Lyons-Ruth, Karlen, David B. Connell, Henry U. Grunebaum, and SheilaBotein. 1990. Infants at social risk: Maternal depression and familysupport services as mediators of infant development and security ofattachment. Child Development 61:85–98.

Lyons-Ruth, Karlen, Betty Repachyoli, Sara McLeod, and Eugenia Silva.1991. Disorganized attachment behavior in infancy: Short-term stabil-ity, maternal and infant correlates and risk-related sub-types. Develop-ment and Psychopathology 3:377–96.

Mazerolle, Paul, and Alex Piquero. 1997. Violent responses to strain: Anexamination of conditioning influences. Violence and Victims12:323–43.

Mazerolle, Paul, and Alex Piquero. 1998. Linking exposure to strain withanger: An investigation of deviant adaptations. Journal of CriminalJustice 26:195–211.

Mischel, Walter. 2004. Toward an integrative science of the person. AnnualReview of Psychology 55:1–22.

Mischel, Walter, and Yuichi Shoda. 1995. A cognitive-affective system the-ory of personality: Reconceptualizing situations, dispositions, dynam-ics, and invariance in personality structure. Psychological Review102:246–68.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 34 28-AUG-07 13:43

514 SIMONS, SIMONS, CHEN, BRODY & LIN

Moffitt, Terrie E. 1997. Adolescent-limited and life-course-persistentoffending: A complementary pair of developmental theories. In Devel-opmental Theories of Crime and Delinquency, ed. Terence P.Thornberry. New Brunswick, NJ: Transaction.

Moffitt, Terrie E., and Avshalom Caspi. 2001. Childhood predictors differ-entiate life-course persistent and adolescence-limited antisocial path-ways among males and females. Development and Psychopathology13:355–75.

Muthen, Linda K., and Bengt O. Muthen. 2004. Mplus 3.0 (Computer pro-gram). Los Angeles, CA: Muthen and Muthen.

Patterson, Gerald R., and Karen Yoerger. 2002. A developmental modelfor early- and late-onset delinquency. In Antisocial Behavior in Chil-dren and Adolescents, eds. John B. Reid, Gerald R. Patterson, andJames Snyder. Washington, DC: American Psychological Association.

Piquero, Nicole L., and Miriam D. Sealock. 2000. Generalizing generalstrain theory: An examination of an offending population. JusticeQuarterly 17:449–84.

Polakowski, Michael. 1994. Linking self- and social control with deviance:Illuminating the structure underlying a general theory of crime and itsrelation to deviant activity. Journal of Quantitative Criminology10:41–78.

Pratt, Travis C., and Francis T. Cullen. 2000. The empirical status of Gottf-redson and Hirschi’s General Theory of Crime: A meta-analysis. Crim-inology 38:931–64.

Reid, John B., Gerald R. Patterson, and James Snyder. 2002. AntisocialBehavior in Children and Adolescents: A Developmental Analysis andModel for Intervention. Washington, DC: American PsychiatricAssociation.

Renken, Bruce, Byron Egeland, Denice Marvinney, Sarah Mangelsdorf,and L. Alan Sroufe. 1989. Early childhood antecedents of aggressionand passive-withdrawal in early elementary school. Journal of Person-ality 57:257–81.

Sampson, Robert J., and John H. Laub. 1993. Crime in the Making: Path-ways and Turning Points Through Life. Cambridge, MA: Harvard Uni-versity Press.

Sampson, Robert J., Stephen W. Raudenbush, and Felton Earls. 1997.Neighborhoods and violent crime: A multilevel study of collective effi-cacy. Science 277:918–24.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 35 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 515

Schuman, Howard. 1995. Attitudes, beliefs, and behavior. In SociologicalPerspectives on Social Psychology, eds. Karen S. Cook, Gary A. Fine,and James S. House. Boston, MA: Allyn and Bacon.

Shaffer, David, Mary Schwab-Stone, Prudence Fisher, Patricia Cohen,John Piacentini, M. Davies, C. Keith Conners, and Darrel Regier. 1993.The Diagnostic Interview Schedule for Children-Revised Version(DISC-R). Preparation, field testing, inter-rater reliability, and accept-ability. Journal of the American Academy of Child and Adolescent Psy-chiatry 32:643–50.

Simons, Ronald L., ed. 1996. Understanding Differences Between Divorcedand Intact Families: Stress, Interaction, and Child Outcome. ThousandOaks, CA: Sage.

Simons, Ronald L., Christine Johnson, and Rand D. Conger. 1994. Harshcorporal punishment versus quality of parental involvement as anexplanation of adolescents’ maladjustment. Journal of Marriage andthe Family 56:591–607.

Simons, Ronald L., Leslie G. Simons, and Lora E. Wallace. 2004. Families,Delinquency, and Crime: Linking Society’s Most Basic Institution toAntisocial Behavior. Los Angeles, CA: Roxbury.

Simons, Ronald L., Wei Chao, Rand Conger, and Glen H. Elder. 2001.Quality of parenting as mediator of the effect of childhood defiance onadolescent friendship choices and delinquency: A growth curve analy-sis. Journal of Marriage and Family 63:63–79.

Simons, Ronald L., Christine Johnson, Rand Conger, and Glen H. Elder.1998. A test of latent trait versus life course perspectives on the stabil-ity of adolescent antisocial behavior. Criminology 36:217–44.

Simons, Ronald L., Leslie G. Simons, Callie H. Burt, Gene H. Brody, andCarolyn Cutrona. 2005. Collective efficacy, authoritative parenting,and delinquency: A longitudinal test of a model integrating commu-nity- and family-level processes. Criminology 43:989–1030.

Simons, Ronald L., Kuei-Hsu Lin, Leslie C. Gordon, Gene Brody, VelmaMurry, and Rand D. Conger. 2002. Community contextual differencesin the effect of parental behavior on child conduct problems: A mul-tilevel analysis with an African American sample. Journal of Marriageand the Family 64:331–45.

Slaby, Ronald G., and Nancy G. Guerra. 1988. Cognitive mediators ofaggression in adolescent offenders: I. Assessment. Developmental Psy-chology 24:580–8.

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 36 28-AUG-07 13:43

516 SIMONS, SIMONS, CHEN, BRODY & LIN

Sucoff, Clea A., and Dawn M. Upchurch. 1998. Neighborhood context andthe risk of childbearing among metropolitan-area black adolescents.American Sociological Review 63:571–85.

van IJzendoorn, Marinus H. 1995. Adult attachment representations,parental responsiveness, and infant attachment: A meta-analysis on thepredictive validity of the Adult Attachment Interview. PsychologicalBulletin 117:387–403.

van IJzendoorn, Marinus H. 1997. Attachment, emergent morality, andaggression: Toward a developmental socioemotional model of antiso-cial behavior. International Journal of Behavioral Development21:703–27.

Warr, Mark. 2002. Companions in Crime: The Social Aspects of CriminalConduct. Cambridge, U.K.: Cambridge University Press.

Warr, Mark. 2005. Making delinquent friends: Adult supervision and chil-dren’s affiliations. Criminology 43:77–106.

White, Lynn K., and Stacy J. Rogers. 2000. Economic circumstances andfamily outcomes: A review of the 1990s. Journal of Marriage and Fam-ily 62:1035–51.

Wiesner, Margit, and Deborah M. Capaldi. 2003. Relations of childhoodand adolescent factors to offending trajectories of young men. Journalof Research in Crime and Delinquency 40:231–62.

Wittchen, Hans U. 1994. Reliability and validity studies of the WHO-Composite International Diagnostic Interview (CIDI): A criticalreview. Journal of Psychiatric Research 28:57–84.

Ronald L. Simons is Distinguished Professor of Research in the Depart-ment of Sociology and a research fellow in the Institute for BehavioralResearch at the University of Georgia. Much of his research has focusedon the manner in which family processes, peer influences, and communityfactors combine to influence deviant behavior across the life course. Hehas also completed work on domestic violence and the effect of racial dis-crimination on child development.

Leslie Gordon Simons is an assistant professor of child and familydevelopment at the University of Georgia. Her research investigates theeffect of family structure, parenting practices, community processes, andreligion on adolescent outcomes such as conduct problems, risky sex, anddepression.

Yi-Fu Chen is a management information specialist with the Center for

\\server05\productn\C\CRY\45-3\CRY302.txt unknown Seq: 37 28-AUG-07 13:43

IDENTIFYING PSYCHOLOGICAL FACTORS 517

Family Research at the University of Georgia. His research includes cross-cultural comparisons of typologies of delinquent behavior, cultural similar-ities in the effect of parental behavior on child development, and the effi-cacy of parenting programs designed to reduce drug use.

Gene H. Brody is Regents Professor in the Department of Child andFamily Development and director of the Center for Family Research atthe University of Georgia. His research is concerned with the impact offamily and community processes on the development of externalizing andinternalizing problems in children and adolescents.

Kuei-Hsiu Lin is an associate professor of sociology at National TaipeiUniversity. Much of her work has focused on the manner in which family,peer, and community processes combine to influence child and adolescentinvolvement in deviant behavior.