Pathways between profiles of family functioning, child ...€¦ · problems through multiple...

Pathways between profiles of familyfunctioning, child security in theinterparental subsystem, and childpsychological problems

PATRICK T. DAVIES,a E. MARK CUMMINGS,b and MARCIA A. WINTER a

aUniversity of Rochester; andbUniversity of Notre Dame

AbstractThis study was designed to delineate pathways between systems profiles of family functioning, children’s emotionalinsecurity in the interparental relationship, and their psychological adjustment in a sample of 221 children and theirparents. Consistent with family systems theory, cluster analyses conducted with assessments of marital, coparental,and parent–child functioning indicated that families fit into one of four profiles:~a! cohesive families, characterizedby warmth, affection, and flexible well-defined boundaries in family relationships;~b! disengaged families,reflected in high levels of adversity and low levels of support across family subsystems;~c! enmeshed families,evidenced by high levels of discord and weak maintenance of relationship boundaries in the family unit; and~d!adequate families, defined by elevated parental psychological control within a larger family context of low discordand high warmth. In comparison to children in cohesive families, children in enmeshed and disengaged familiesexhibited greater signs of insecurity in the interparental relationship concurrently and internalizing and externalizingsymptoms both concurrently and 1 year later. Structural equation models revealed that a latent, multimethodmeasure of insecurity in the interparental relationship partially mediated associations between family enmeshmentand disengagement and children’s psychological symptoms 1 year later. Results are discussed in relation to howthey inform and refine a family-wide model of the emotional security hypothesis.

Family systems theory has challenged scien-tists to expand beyond understanding childdevelopment and psychopathology solely in

the context of parent–child relationship dynam-ics ~Minuchin, 1985!. Guided by the principleof holism, family systems theory asserts thatpathways linking parenting dynamics withchild functioning are fully understood in thecontext of the collective experiences in thefamily as a whole~Hayden, Schiller, Dick-stein, Seifer, Sameroff, Miller, Keitner, &Rasmussen, 1998; Mikulincer, Florian, Cowan,& Cowan, 2002!. Supporting this principle,research has repeatedly demonstrated that mar-ital functioning plays a key role in the social-ization of children in families~Cowan &Cowan, 2002; Cummings & Davies, 2002;Emery, 1982; Grych & Fincham, 1990!. Chil-dren exposed to marital discord are vulnerableto exhibiting a wide array of emotional, be-havioral, and coping difficulties~Cummings,1994!. Furthermore, marital discord is posited

This research was supported by the National Institute ofMental Health~Project R01 MH57318! awarded to PatrickT. Davies and E. Mark Cummings. Marcia A. Winter wassupported by a predoctoral National Research ServiceAward from the National Institute of Mental Health~F31MH068058!. The authors are grateful to the children, par-ents, teachers, and school administrators who partici-pated in this project and to the staff who assisted on variousstages of the project, including Courtney Forbes, MarcieGoeke–Morey, Amy Keller, Michelle Sutton, and the grad-uate and undergraduate students at the University of Roch-ester and University of Notre Dame. We also thank HarryReis for his valuable statistical advice.

Address correspondence and reprint requests to: PatrickDavies, Department of Clinical and Social Sciences inPsychology, University of Rochester, Rochester, NY14627; E-mail: [email protected].

Development and Psychopathology16 ~2004!, 525–550Copyright © 2004 Cambridge University PressPrinted in the United States of AmericaDOI: 10.10170S0954579404004651

525

to increase children’s risk for psychologicalproblems through multiple mechanisms reflect-ing transactions between the child and thelarger family system~see Margolin, Oliver, &Medina, 2001!.

In expanding on the associations betweenmarital and child functioning, the emotionalsecurity hypothesis is a process model that isdesigned to partially explain how and whyinterparental conflict is associated with childvulnerability to psychological problems. In-spired by principles of family systems theory,the emotional security hypothesis comple-ments earlier theories of parent–child relation-ships by drawing attention to the developmentaland relational significance of children’sevolving adaptation in the context of the inter-parental subsystem. Drawing from attachmenttheory, the emotional security hypothesis placesthe child’s goal of preserving a sense of secu-rity at the heart of understanding transactionsbetween the child and family. Thus, the emo-tional security hypothesis accepts the mainpresupposition of attachment theory: childemotional security in parent–child relation-ships is a key mechanism mediating the linkbetween parental socialization strategies andchild developmental outcomes. However, inassimilating principles of family systemstheory, the emotional security hypothesis dif-fers from conventional versions of attachmenttheory by more broadly conceptualizing chil-dren’s adaptation as an evolving cause andproduct of functioning across multiple familyrelationships. By extension, a main assump-tion is that other aspects of family function-ing, including the marital relationship, mayserve as sources of support or threat to chil-dren’s well-being~Cummings & Davies, 1996;Cummings, Goeke–Morey, & Papp, 2003;Davies & Cummings, 1994!.

The emotional security hypothesis specifi-cally postulates that interparental discordincreases child vulnerability to psychologicalproblems by undermining children’s sense ofsecurity or safety in the family. Children frommaritally discordant homes have sound basesfor being concerned about their sense of secu-rity. Spousal conflicts in these homes tend toescalate into dysregulated bouts of hostility,proliferate to disrupt parent–child relations,

and signify potential family instability. Thus,in the first part of this mediational chain, wit-nessing interparental discord is postulated toincrease children’s concerns about their safety,security, and well-being in subsequent con-texts of interparental problems. In the secondpart of the mediational chain, concerns aboutsecurity are proposed to sap children of biop-sychological resources and, in the process,increase their vulnerability to psychologicalproblems.

Preserving emotional security is conceptu-alized as a latent goal that can be inferred fromthree observable classes of responding:~a!emotional reactivity, characterized by intense,prolonged, dysregulated distress reactions tomarital conflict;~b! regulation of exposure toparent affect, evidenced by prolonged rigidattempts to become involved in or avoid theparental conflict; and~c! internal represen-tations of marital relations, as indexed bychildren’s evaluations of the potential conse-quences of parental difficulties for their ownwell-being. Thus, children repeatedly exposedto high levels of marital hostility are positedto experience greater difficulties preservingsecurity, as manifested in concrete signs ofdysregulated distress, avoidance of or involve-ment in parental disputes, and negative repre-sentations of the meaning marital conflict hasfor their well-being. Consistent with a pri-mary tenet of family systems theory~Byng–Hall & Stevenson–Hinde, 1991; Hinde, 1988!,these response patterns are thought to cohereover time and increase children’s probabilitiesof developing along psychopathological tra-jectories. Supporting these hypotheses, stud-ies have shown that the proposed signs ofemotional insecurity in the context of interpa-rental conflict mediate the links between his-tories of exposure to marital discord and theirpsychological problems~Davies & Cummings,1998; Davies, Harold, Goeke–Morey, & Cum-mings, 2002!.

Expanding the study of child adaptation tothe marital subsystem is a necessary, but notsufficient, condition for testing the predic-tions of family systems theories. For example,studying pathways between marital and childfunctioning in isolation from the larger familysystem is inconsistent with the principle of

526 P. T. Davies, E. M. Cummings, and M. A. Winter

holism ~Cox & Paley, 1997; Minuchin, 1974,1985; Nichols, 1999!. Thus, the emotionalsecurity hypothesis further postulates that fam-ily processes are critical to understanding thenature of associations between marital andchild functioning ~Cummmings et al., 2003;Davies et al., 2002!. For example, family pro-cesses such as coparenting and parenting prac-tices are hypothesized to mediate associationsbetween marital discord and child adjustment.Supporting this hypothesis, research has spe-cifically shown that parenting dimensions suchas parental hostility, rejection, and lax moni-toring account for part of the associationbetween marital discord and child adjustment~Gonzalez, Pitts, Hill, & Roosa, 2000; Miller,Cowan, Cowan, Hetherington, & Clingem-peel, 1993; Stocker & Youngblade, 1999!.Likewise, the implications of marital discordfor children’s security and adjustment areassumed to depend, in part, on the quality ofthe larger family system. Consistent with mod-els of developmental psychopathology, familycharacteristics are specifically proposed topotentiate or protect children from experienc-ing insecurity in the face of marital discord.For example, marital discord may assume amore benign meaning for children in the con-text of cohesive, autonomy-supportive rela-tionships in the family. By the same token, thedeleterious implications of marital discord forchildren may be amplified by family pro-cesses that reflect high levels of discord, chaos,or disengagement. Supporting this hypoth-esis, Davies et al.~2002! reported that familyinstability potentiated the link between mari-tal conflict and children’s emotional insecu-rity in the marital relationship, while signs ofharmony in the family system buffered chil-dren from experiencing insecurity in high con-flict homes.

Despite exploratory support for the valueof integrating the emotional security hypoth-esis with family systems theory, a systemsunderstanding of children’s coping and adjust-ment in the face of marital conflict will requireprogressively richer, holistic accounts of thefamily system. In accordance with family sys-tems theory~Byng–Hall & Stevenson–Hinde,1991; Mikulincer et al., 2002; Wagner & Reiss,1995!, the aim of this study was to assess

children’s security in the marital subsystemand their adjustment within a family-widemodel that captures the organization, pattern-ing, and regularity of processes in the familysystem that emerge from interrelationshipsbetween marital, coparenting, and parent–child dynamics. Pattern-based approaches,which permit classification of families intogroups on the basis of their configurations offunctioning within and across relationships,provide particularly useful tools for capturingthe higher order organization of family func-tioning across multiple family subsystems.Given the previous success of applying clus-ter analytic methods to identify patterns of fam-ily functioning ~O’Connor, Hetherington, &Clingempeel, 1998; O’Connor, Hetherington,& Reiss, 1997!, we used cluster analysis toidentify different profiles of family function-ing. Guided by developmental models of fam-ily systems~e.g., Combrinck–Graham, 1985,1990; Kretchmar & Jacobvitz, 2002; Wynne,1988!, we specifically assessed dimensions offamily functioning that reflect the quality ofemotional relationships within and acrossfamily subsystems. According to these mod-els, communication patterns of families withschool-aged children are salient developmen-tal features of relationship quality in the fam-ily system.

Although theory and research on family sys-tems have identified a number of differentconfigurations of emotional and relational func-tioning in the family ~e.g., O’Connor et al.,1997, 1998!, three profiles of family function-ing have been most consistently delineated inclinical and empirical work~Kerig, 1995;Kretchmar & Jacobvitz, 2002; Liddle &Schwartz, 2002; Minuchin, 1974; Stevenson–Hinde, 1990!. The first profile consists of cohe-sive families who exhibit relatively warm,close, and harmonious relationships acrossmultiple family subsystems. Boundaries thatmetaphorically separate relationships and fam-ily members are well-defined but flexible incohesive families, permitting family membersaccess to relationship resources and supportwhile also respecting their autonomy~Cox &Paley, 1997!. Children in cohesive familieshave been postulated to exhibit high levels ofparent–child attachment security, construc-

Family configurations and child vulnerability 527

tive coping strategies, and psychological ad-justment~Garbarino, Sebes, & Schellenbach,1984; Kerig, 1995; Stevenson–Hinde, 1990!.When interpreted within the emotional secu-rity hypothesis, witnessing well-managed mar-ital disputes within a larger context of cohesive,warm family relationships is thought to in-crease children’s confidence in their parentsto resolve any disagreements or problems inthe marriage in a way that maintains or im-proves family harmony~Davies et al., 2002!.Thus, children from cohesive families are pos-tulated to experience relatively few difficul-ties preserving their sense of security in theinterparental relationship. Success in preserv-ing emotional security, in turn, is thought tolower children’s risk for psychological adjust-ment problems.

Enmeshed or chaotic families constitute thesecond profile of family functioning. Unlikecohesive families, these families exhibit highlevels of conflict, hostility, and coercive con-trol across multiple family subsystems. Dif-fuse boundaries exist between subsystems~e.g.,marital and coparenting and parent–child sub-systems! and individuals, as stress and dys-function in one individual or relationship is arelatively seamless product and cause of dys-function in the larger family system~Kretch-mar & Jacobvitz, 2002; Marvin & Stewart,1990; Stevenson–Hinde, 1990!. In the bidirec-tional conceptualization of family systemstheory, the psychological difficulties of eachindividual are conceptualized as both causesand sequelae of discordant, weak boundarieswithin and across family relationships~e.g.,Wood, 1985!. Accordingly, children fromenmeshed families are thought to be at risk fordeveloping a wide array of psychological~e.g.,internalizing, externalizing! problems~Kerr,2003; Tolan & Mitchell, 1989!. According tothe emotional security hypothesis, child con-cern about security in the face of marital dif-ficulties is one mechanism that may accountfor links between enmeshed family processesand child psychological symptoms. Enmeshedfamily processes that emotionally pull or coaxchildren into family problems may amplify theimpact marital conflict has on children’s senseof security. In turn, prolonged difficulties inpreserving security in the marital subsystem

are thought to increase children’s risk for mal-adjustment~Coyne, Downey, & Boergers,1992; Davies & Forman, 2002; Davies et al.,2002!.

Disengaged or “separate” families, whichcomprise a third profile of family function-ing, may resemble enmeshed families in exhib-iting high levels of hostility and discord inrelationships. However, unlike enmeshedfamilies, the boundaries between and withinmarital, coparenting, and parent–child rela-tionships are characterized as overly rigid,inflexible, distant, and emotionally cold~Bel-sky, Rovine, & Fish, 1989; Kretchmar &Jacobvitz, 2002; Minuchin, 1974; Stevenson–Hinde, 1990!. According to family systemstheory, disengaged family systems are thoughtto increase child psychological problems~Kerig, 1995!. Although the emotional secu-rity hypothesis shares the prediction thatchildren from disengaged families are atheightened risk for experiencing externaliz-ing symptoms, it also hypothesizes that inter-parental hostility expressed in the context ofa disengaged, cold family system amplifieschildren’s concerns about the consequencesinterparental discord have for their own wel-fare and the welfare of their families. Theresulting insecurity is further posited toincrease psychological difficulties. A deriva-tive hypothesis is that children’s security inthe interparental conflict is hypothesized topartially account for the vulnerability of chil-dren from disengaged families~Davies & For-man, 2002!.

Additional family profiles have also beenidentified, but with less consistency. On theone hand, some empirical and clinical worksupports the existence of a family profile char-acterized by a relatively cohesive marriage butsome form of dysfunction in the parent–childsubsystem. For example, research using com-munity samples has identified profiles in whichdysfunction in the family is isolated to theparent–child subsystem~O’Connor et al., 1997,1998!. Likewise, systems theorists have spe-cifically described a “detouring” family con-figuration in which parents are proposed tobecome preoccupied with the difficulties facedby their children in an attempt to deflect atten-tion away from more serious marital strife


~Buchanan & Waizenhofer, 2001; Cox & Paley,1997; Emery, 1989; Minuchin, 1974!. On theother hand, other research supports the exis-tence of family profiles characterized by sig-nificant difficulties in the spousal relationshipmay be embedded in the context of close-ness and cohesion in other family subsystems~O’Connor et al., 1998!. Inconsistencies inidentifying other family configurations hinderthe development of predictions about theexistence and meaning of alternative familyprofiles.

The overarching goal of this study was toexamine children’s emotional security andadjustment in the context of the broader pat-tern of family relationships. Thus, our first spe-cific aim was to identify the distinct forms offamily configurations within our communitysample of kindergarten children and their par-ents. Consistent with family systems theory, itwas hypothesized that at least three forms offamily functioning would emerge: cohesive,enmeshed, and disengaged families. Althoughthe complexity of the family system also pointsto the possibility of identifying other familyconfigurations, no additional hypotheses wereoffered in light of the inconsistencies in theliterature. Our second aim was to examinewhether these different family configurationsare differentially linked with children’s secu-rity in the marital relationship and their psy-chological symptoms. Consistent with ourintegration of the emotional security hypoth-esis and family systems theory, we specifi-cally hypothesized that family enmeshment anddisengagement would increase children’s emo-tional insecurity in the marital subsystem. Inturn, emotional insecurity is hypothesized tobe associated with child psychological symp-toms. To rigorously delineate associationsbetween profiles of family functioning, childemotional security in the marital subsystem,and child psychological symptoms, we uti-lized ~a! multiple informant ~i.e., mother,father! reports of functioning in multiplefamily subsystems,~b! multiple methods andtasks for assessing child security in the mari-tal subsystem, and~c! multiple informant~i.e., parents, teacher! reports of child psycho-logical symptoms across two measurementoccasions.

Method

Participants

The data for this study were drawn from alarger investigation focusing on linkages be-tween family processes and child coping andpsychological adjustment. The original sam-ple of 236 kindergarten children and their fam-ilies in the first wave of the study were recruitedthrough local school districts and communitycenters in a moderate-sized metropolitan areain the Northeast and a small city in the Mid-west. Families who expressed an interest inparticipating by returning a post card or call-ing our research office were included in thelarger project if they met the following crite-ria: ~a! the family had a child in kindergarten,~b! the kindergarten child lived with two pri-mary caregivers for at least 3 years,~c! theprimary caregivers and children were fluent inEnglish, and~d! the primary caregivers andchildren were willing to participate. The reten-tion rate over the 1-year period between thefirst and second waves was 98%~n 5 231!.Families who participated in both measure-ment waves were included in the present studyif mothers, fathers, and children completed allthe measures of family and child functioningused in this article. The resulting sample con-sisted of 221 mothers, fathers, and children.The average age of the children at Wave 1 was6.0 years~SD5 0.48!, with 56% of the sam-ple consisting of girls~n5124! and 44% con-sisting of boys~n 5 97!.

Sociodemographic data reflected that theparticipating families were relatively represen-tative of the households in the counties fromwhich our sample was drawn~i.e., St. JosephCounty, IN; Monroe County, NY!. Medianannual family income of the families wasbetween $40,000 and $54,999, with 9.2% ofthe families reporting income below $17,000per year. The average number of years of edu-cation completed by mothers and fathers were14.5~SD52.32! and 14.6~SD52.71!, respec-tively. The majority of the sample identifiedthemselves as White~77.1%!, followed bysmaller, but relatively representative~for thecounties!, proportions of Black~15.9%!, His-panic~4.0%!, Asian~1.1%!, and Other0Mixed


~1.8%! family members. A large percentage ofmothers and fathers reported being the biolog-ical parents of their children~95.5 and 87.8%,respectively!, followed by relatively smallerpercentages of adoptive parents~2.3 and 3.2%,respectively!, stepparents~0.5 and 5.9%, re-spectively!, and other types of guardianship~1.7 and 3.1%, respectively!. At the first wave,mothers, fathers, and children lived in the samehousehold for an average of 5.4 years~SD51.00!.

Procedure

Data for this study were collected at two mea-surement occasions spaced 1 year apart. Ateach wave, families visited the laboratories atone of the two research sites. The laboratorieswere designed to be comparable in size andquality and included:~a! an observation roomthat was designed to resemble a family room~e.g., couch, pictures, lamps, end tables! andequipped with audiovisual equipment to cap-ture family interactions, and~b! interviewrooms for completing confidential surveymeasures.

Simulated Phone Argument Task (SPAT).Dur-ing the first measurement occasion, childrenand their mothers participated in the SPATto assess child reactivity to interparentalconflict. During this procedure, children wit-nessed live simulations of their parents engag-ing in a conflict and a subsequent resolutionover the telephone. Each exchange lastedapproximately 1 min, and was followed by a3-min free period to assess children’s imme-diate and delayed reactivity. The conflict scriptrevolved around a relatively trivial disagree-ment regarding whether the father had com-pleted a task requested by the mother~i.e.,stopped at the store, made a phone call, madean appointment!. The mothers were instructedto convey mild irritation, frustration, and angertoward their partner as they normally wouldat home. For the simulated resolution, themother was asked to communicate a moderatelevel of understanding, caring, and warmth inher tone of voice. Although the simulationsindicated to the child that the father was onthe other end of the phone, an experimenter

was actually on the phone feeding the motherthe lines from the script. Videotaped recordsof the task were later coded for child overtreactivity to conflict.

Several procedures were instituted duringa presimulation briefing and training sessionto ensure that mothers accurately followed thescript and expressed the desired level and typeof affect for each emotional exchange. First,mothers listened to a standard, audiotaped sam-ple of the content and affective tone of theconflict and resolution. Second, mothers prac-ticed the script with the experimenter until theywere able to accurately and comfortably con-vey the content and affective tone of theexchanges. Third, mothers were encouragedto convey the same emotional level and qual-ity of their successful practice run in the actualtask with their children. Fourth, in feeding themother the lines during the procedure, theexperimenter simulated the affective tone andlevel for the mother to emulate.

To maximize the ecological validity of theexchanges, parents were asked during the firstlaboratory visit to schedule the second visitinvolving only the mother and child during atime in which the father was accessible byphone. In addition, mothers were asked tochoose one of three script options that mostrealistically reflected the nature of interparen-tal activities. Finally, mothers were given theopportunity to alter some of the wording ofthe scripts to more accurately reflect the con-tent and nature of actual interparental con-flicts and resolutions that occur in the home.However, to ensure that internal validity wasnot compromised, changes to the script wereonly implemented if they did not significantlyalter the intensity, meaning, and affective toneof the disagreement or resolution.

SPAT interview.Immediately following eachof the two free periods of the SPAT, the exper-imenter conducted a structured interview withthe children to assess their self-reported emo-tionality and behavioral impulses during thesimulated interparental conflict and resolution.

Structured interviews and questionnaires.AtWave 1, children completed the Children’s Per-ception of Interparental Conflict Scale for


Young Children~CPIC-Y; Grych, 2000!, amodified, interview version of the CPIC scalesdesigned to assess children’s internal represen-tations of interparental relations~Grych, Seid,& Fincham, 1992!. Mothers and fathers alsocompleted survey assessments of interparen-tal functioning and parenting at the first mea-surement occasion, while mothers, fathers, andteachers completed assessments of children’spsychological symptoms at both measure-ment occasions.

Measures

Child emotional insecurity.Different sets ofjudges rated each of the key constructs~e.g.,behavioral distress, behavioral regulation ofconflict exposure, involvement impulses! de-rived from behavioral and open-ended narra-tive responses to the SPAT.

Child emotional reactivity.Behavioral and sub-jective expressions of distress in response tothe simulated interparental exchanges wereobtained as indicators of emotional reactivityto interparental conflict. To assess children’sbehavioral distress to interparental discord,judges coded the presence of specific distressbehaviors for eight 30-s intervals during the4-min videotaped segment of the conflict andpostconflict periods. Two judges indepen-dently coded 45~21%! randomly selectedchildren’s responses to the SPAT to determineinterrater reliability. The specific behavioralcodes and their corresponding interraterreliability coefficients were as follows:~a!freezing—tense, motionless, or “fixed in place”for $5 s ~k 5 .73!; ~b! anxiety— expressionsof anxiety through facial expressions and0orphysical gestures~k 5 .70!; ~c! whining—whimpering, fretting, complaining, or whin-ing ~k 5 .54!; ~d! anxious smiling—smilingor laughing to self or parent that is clearly ananxious response to the simulated conflict~k5.85!; ~e! sadness—facial expressions, verbalstatements and tone, and0or posture reflect sad-ness or depressed affect~k 5 .59!; ~f ! loss ofcontrol—aimless, disorganized, uncontrolledquality to child activity~k 5 .50!; ~g! anger—facial expression and0or posture reflects anger~k 5 .66!; and ~h! interpersonal hostility—

verbal or physical hostility directed toward theself or others~k 5 .93!. Frequency counts ofthe eight distress behaviors over the 4-minobservational period were summed to form anoverall measure of child behavioral distress.

Ameasureofsubjectivedistresswasobtainedfromfour interviewquestions in thepairofSPATinterviews. In the interview following the sim-ulated conflict, children specifically respondedwith yes or no answers to two questions: “Didyou feel scared@sad# when your parents weretalking on the phone?” To obtain assessmentsof prolonged subjective distress, children alsoresponded with “yes” or “no” answers to twoquestions after the simulated phone resolution,“Did you feel scared@sad# when your parentswere talking the second time?” Immediatelyafter affirmative responses to these questions,children answered the question, “How much didyou feel that way?” using response alternativesranging from 15 very little to 55 a whole lot.Consistent with earlier research~Davies & For-man, 2002!, fearful and sad responses weresummed to form a single measure of children’ssubjective distress.

Child regulation of conflict exposure.Judgesrated the presence of specific avoidance andinvolvement behavior codes in the eight 30-sintervals of the interparental conflict andpostconflict free period to assess overt formsof child regulation of conflict exposure. Dueto the very low frequency of specific in-volvement behaviors recorded during and fol-lowing the phone dispute, the involvementbehaviors were dropped from the analyses. Todetermine interobserver reliability, judges inde-pendently coded the behaviors of 47~21%! ofthe children in the sample. The specific behav-ioral avoidance codes and their correspondinginterobserver reliability coefficients were asfollows: ~a! shutting out, which is any behav-ioral attempt to avoid or escape the interper-sonal exchanges or the parents~k 51.00!; and~b! social withdrawal, which is unresponsive-ness, indifference, and disinterest in responseto bids, requests, or commands made by themother ~k 5 1.00!. Frequency counts of thetwo types of avoidance over the 4-min obser-vational period were summed to form an over-all measure of child behavioral avoidance.


To assess impulses to become involved inthe interparental exchanges, children firstresponded to the question, “What did you wantto do when your parents were talking on thephone?” in the interview following the simu-lated conflict. In the interview after the simu-lated resolution, children answered a similarquestion, “What did you want to do when yourparents were talking the second time?”, toobtain an assessment of prolonged impulsesto intervene in interparental relations. For eachquestion, children were first asked to selectone of eight response cards containing car-toon illustrations of the following child actions:playing, walking away from the parents, hud-dling, watching the parents, helping with tasks,getting in between the parents, comforting oneof the parents, and yelling at the parents. Afterselecting a card, children were asked to de-scribe what was happening in the picture. Toencourage children to rely on their ownthoughts, feelings, and impulses in their nar-ratives, the cartoon illustrations were specifi-cally designed to be “projective” in nature,providing little or no information regardingthe emotional states or the sexes of the char-acters. Involvement impulses were defined asdispositions to become emotionally or behav-iorally involved in the interparental inter-actions. Ratings of involvement impulsesconsisted of: 1, none; 2, mild—a single, weakor ambiguous sign of involvement; 3, mod-erate—a clear, but still relatively mild, dispo-sition to become involved; 4, intense —considerable disposition to become involvedthat requires moderate investment, risk, or fore-thought; and 5, very intense— considerabledisposition to become involved that requires ahigh degree of investment, risk, or forethought.Intraclass correlation coefficients, which indexinterrater reliability with 100% of the sample,were .97 and .92 for ratings of involvementimpulses in the conflict and the resolution,respectively.

Internal representations of interparentalrelations.To assess children’s negative repre-sentations of interparental relations, we uti-lized the CPIC-Y~Grych, 2000!. The CPIC-Y,which consists of 34 items, is a modified inter-view version of the CPIC Scale for use with

young children~Grych et al., 1992!. Childrenverbally respond to each statement by answer-ing yes if the statement describes them or theirfamily ~1! or no if the statement does notdescribe them or their family~0!. Given ourtheoretical emphasis on assessing children’sperceptions of the implications interparentalconflict has for their own welfare, only itemsfrom the Perceived Threat Scale were includedin the present study. The six items comprisingthe Perceived Threat Scale assess children’sappraisals of the degree to which interparentalconflict is threatening to their well-being~e.g.,“When my parents argue, I’m afraid some-thing bad will happen”;a5 .84 in this sample!.

Family processes

Mothers and fathers independently completeda battery of questionnaires designed to assessfunctioning in the spousal, coparental, andparent–child subsystems.

Interparental conflict.Parents completed theO’Leary–Porter Scale~OPS; Porter & O’Leary,1980! to assess the frequency of child expo-sure to interparental hostility. Each item~e.g.,“How often do you and0or your partner dis-play verbal [email protected]., raised voices, criti-cism, etc.# in front of your child?”! is rated ona scale of 0~never! to 4 ~very often!. Test–retest reliability, internal consistency, and theconcurrent validity of the OPS are well docu-mented~e.g., Grych et al., 1992; Porter &O’Leary, 1980!. Alpha coefficients for themother and father reports on the OPS in thissample were .81 and .79, respectively. Motherand father reports of interparental hostility wereaveraged together in light of their substantialintercorrelation,r ~221! 5 .61,p , .001.

Interparental affection.To obtain an assess-ment of interparental affection, parents com-pleted the Love and Affection subscale of theManaging Affect and Differences Scale~Arel-lano & Markman, 1995!. Participants rated howmuch they agreed or disagreed with 10 state-ments~e.g., “I am affectionate toward my part-ner.”! by selecting a response from a scaleranging from 1 ~strongly disagree! to 5~strongly agree!. The Love and Affection sub-


scale has adequate internal consistency~as5.86 and .85 in this sample for mothers andfathers, respectively! and the validity of themeasure is supported by its links with higherlevels of relationship satisfaction and per-ceived relational efficacy and lower levels ofrelationship problems. Mother and fatherreports were averaged together to form a sin-gle measure of interparental affection in lightof their moderate correlation,r ~221! 5 .55,p , .001.

Coparental disagreements.Parents completedeight items from the original 21-item Child-Rearing Disagreements~CRD! Scale to assessthe frequency of parental disagreements overchild-rearing issues over the past year~Jouriles,Murphy, Farris, Smith, Richters, & Waters,1991!. Parents responded to each item on a6-point continuum, ranging from 1~never! to6 ~daily; e.g., irritating disagreements about“being too hard on our child”!. Support forthe reliability and validity of the CRD is wellestablished. Alpha coefficients in the presentsample were .79 and .74 for mothers andfathers, respectively. Because mother and fatherreports were moderately correlated,r ~221! 5.38, they were averaged together to form asingle measure.

Parental acceptance.Parental reports on theParental Acceptance subscale of the ParentalAcceptance and Rejection Questionnaire~Rohner, 1990! were used to assess parentacceptance and warmth. Mothers and fatherscompleted the instrument twice to assess theirown and their partner’s acceptance using a5-point scale ranging from 1~never! to 5~always; e.g., “You talk to your child in a warm,affectionate way.”!. Internal consistencies forthe four measures ranged from .89 to .93. Giventhe moderate correlations among the four mea-sures~mean r 5 .38!, the Acceptance sub-scales were averaged together to form a singleassessment of parental acceptance.

Parental psychological control.Mothers andfathers reported on their own and their partner’suse of psychologically controlling parentingbehaviors using the Control Through Guilt~e.g., “You let your child know you feel hurt

by the things he0she does”!, Instilling Per-sistent Anxiety ~e.g., “You think that anymisbehavior is serious and will have future con-sequences”!, and Intrusive~e.g., “You keep acareful check on your child to make sure s0hehas the right kind of friends.”! subscales fromthe Child Report of Parenting Behaviors Inven-tory ~Margolies & Weintraub, 1977; Schluder-mann & Schludermann, 1970!. Items on eachof the five-item subscales were rated on 5-pointLikert scales ranging from 1~never! to 5~always!. The subscales have satisfactory inter-nal consistency and previous research has sup-ported the convergent and discriminant validityof the measures~Fauber, Forehand, Thomas, &Wierson, 1990; Schwarz, Barton–Henry, &Pruzinsky, 1985!. Because of the moderate tostrong correlations among the three subscales~rs5 .31–.67!, the subscales were summed toformmother reportsofherownandherpartner’spsychological control~as5 .85 and .83, respec-tively! and father reports of his own and hispartner’s psychological control~as 5 .83 forboth scales!. The four measures of parental psy-chological control, in turn, were strongly cor-related, meanr ~221! 5 .50. Thus, they wereaveraged together to form a single assessmentof parental psychological control.

Poor parental discipline.Parental reports onthe Inconsistent Discipline subscale of the Ala-bama Parenting Questionnaire~APQ! wereused to assess parental behavioral control~Shelton, Frick, & Wootton, 1996!. Mothersand fathers completed the six-item subscaletwice to assess their own and their partner’sinconsistent discipline using a 5-point contin-uum ranging from 1~never! to 5 ~always; e.g.,“You threaten to punish your child and thendo not actually punish him0her”!. The valid-ity of the APQ scales is supported by its sig-nificant associations with child psychiatricsymptoms and diagnoses~Frick, Christian, &Wootton, 1999; Wootton, Frick, Shelton, & Sil-verthorn, 1997!. Internal consistencies for thefour measures ranged from .66 to .73 in thepresent sample. The four measures of motherand father reports of their own and theirpartner’s inconsistent discipline practices wereaveraged together into a single composite inlight of their intercorrelations~meanr 5 .41!.


Child maladjustment

Mothers and fathers completed the symptomssubscales from the Child Behavior Checklist~CBCL; Achenbach, 1991! at Waves 1 and 2.Following recommendations by Achenbach~1991!, the Anxious0Depressed, Withdrawn,and Somatic Symptoms subscales weresummed to obtain separate mother and fatherreports of internalizing symptoms, while theDelinquent Behavior and Aggressive Behav-ior subscales were summed to obtain motherand father reports of externalizing symptoms.The CBCL has demonstrated reliability~as..84 for all scales in the present sample!, test–retest reliability, and validity~Achenbach,1991!. Mother and father reports were aver-aged together to form parsimonious parentreports of internalizing and externalizing symp-toms at each wave, in light of the substantialempirical overlap across the informants~i.e.,meanr 5 .51!.

The Teacher Report Form~TRF; Achen-bach, 1991! was also completed by thechildren’s kindergarten and first grade teach-ers at Waves 1 and 2, respectively. Consistentwith recommendations~Achenbach, 1991!, theAnxious0Depressed, Withdrawn, and SomaticSymptoms subscales were summed to form aninternalizing symptoms measure, while theDelinquent Behavior and Aggressive Behav-ior subscales were summed together to pro-vide a teacher assessment of externalizingsymptoms. Alpha coefficients for scales in thepresent sample exceeded .86 for each of thescales.

Results

For descriptive purposes, Table 1 shows themeans, standard deviations, and intercorrela-tions among the measures of family process,child insecurity, and child psychological symp-toms. Mean levels of behavior problems inour sample were comparable to or higher thanother community samples for parent reportsinternalizing and externalizing symptoms atboth time points and teacher reports of inter-nalizing symptoms at both time points~seeAchenbach, 1991; Clarke–Stewart, Allhusen,McDowell, Thelen, & Call, 2003!. However,

teacher reports of externalizing symptoms inour sample were notably lower than other com-munity and high-risk samples, with our meansranging between 4.01 and 5.95 in comparisonwith other means ranging between 8.16 and8.20 in other samples~Achenbach, 1991; Lynch& Cicchetti, 1998!. Proportions of childrenwho were above the clinical cut point~t . 63!on the child maladjustment measures rangedbetween 16 and 24% for parent reports of psy-chological symptoms and between 2 and 9%for teacher reports of psychological symptoms.

The first section of the primary analysespresents pattern-based results pertaining to theidentification of distinct profiles of family func-tioning across multiple family subsystems. Theremaining sections present analyses that weredesigned to examine the nature of interrela-tionships among identified profiles of familyfunctioning, child insecurity, and child psy-chological symptoms.

Cluster analyses

Identification of the profiles of family func-tioning occurred in two analytical steps. In thefirst step, we sought to specify the number ofclusters in the data set by submitting the stan-dardized scores for the six measures of fam-ily process to a hierarchical agglomerativecluster analysis. Hierarchical agglomerativecluster analyses are especially effective indetermining the number of the clusters in adata set. Ward’s hierarchical agglomerativecluster method with squared Euclidian dis-tance was selected for use on the basis ofgeneral recommendations by quantitative psy-chologists; the availability of stopping rulesfor determining the number of clusters~e.g.,dendogram!; and its ability to recover theunderlying structure of the clusters while tak-ing into account level, shape, and scatter inidentifying cluster profiles~Borgen & Bar-nett, 1987; Lorr, 1994; Milligan & Cooper,1985; Skinner & Blashfield, 1982!. In merg-ing cases into groups at each successive stage,the Ward’s method is specifically designed tomaximize within-group homogeneity by min-imizing within-group error variance for all pro-file variables.


Table 1. Means, standard deviations, and intercorrelations for measures of family process, child emotional security, and child psychological symptoms

Variables M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Family processes1. Interparental hostility 11.93 4.782. Interparental affection 42.75 5.542.493. Coparent conflict 14.77 3.79 .40 2.144. Parent psych. control 13.89 2.24 .132.08 .325. Parent acceptance 86.71 6.432.18 .16 .18 .226. Inconsistent discipline 11.11 2.22 .322.11 .46 .33 2.08

Child emotional security7. Subjective distress 1.34 2.65 .152.15 .16 .04 2.14 .128. Involvement impulses 3.92 1.96 .182.02 .23 .18 2.04 .11 .269. Threat Appraisals 2.22 2.14 .192.10 .24 .18 .01 .16 .27 .14

10. Behavioral distress 2.42 2.89 .03 .01 .20 .082.05 .19 .13 .10 .0711. Behavioral avoidance 0.21 0.56 .132.08 .17 .10 2.06 .18 .12 .02 .09 .37Child psychological symptoms12. T1 Externalizing~T! 4.32 6.89 .00 2.07 .16 .19 2.08 .11 .19 .10 .11 .19 .0813. T2 Externalizing~T! 5.24 7.65 .10 2.10 .11 .20 2.06 .12 .15 .10 .07 .20 .10 .5514. T1 Internalizing~T! 5.36 5.70 .02 2.01 .09 .05 2.17 .00 .06 .14 .02 .05 .04 .35 .1415. T2 Internalizing~T! 5.91 6.50 .07 2.11 .04 .09 2.10 .06 .17 .07 .17 .07 .01 .15 .27 .2416. T1 Externalizing~P! 10.13 6.68 .17 2.06 .29 .23 2.22 .38 .07 .10 .07 .21 .24 .30 .27 .032.0117. T2 Externalizing~P! 9.48 6.60 .20 2.10 .33 .22 2.18 .31 .11 .08 .08 .20 .19 .29 .37 .01 .01 .8418. T1 Internalizing~P! 7.15 5.41 .17 2.07 .22 .11 2.18 .27 .09 .08 .06 .25 .23 .06 .05 .13 .09 .61 .5219. T2 Internalizing~P! 7.07 5.25 .17 2.06 .24 .10 2.18 .25 .16 .09 .05 .22 .20 .08 .13 .08 .24 .55 .66 .77

Note:Correlation coefficients of .13 and above are significant atp # .05. ~T!, teacher report;~P!, parent report.

53

5

The results of our analyses indicated thatthe four-cluster solution outperformed alter-native cluster solutions across most domainsof evaluation. First, the dendogram and ag-glomeration schedule showed that the increasein distance coefficients was particularly pro-nounced when four clusters were successivelymerged into three clusters. Thus, the four-factor solution was supported on the basis ofstopping rules. Second, the four-cluster solu-tion was effective in identifying family pro-files that were significantly different from eachother on the family variables submitted to thecluster analysis. Third, from a theoretical per-spective, the four-cluster solution most closelyresembled family profiles identified in previ-ous family systems work~e.g., Kerig, 1995;Minuchin, 1974!.

In the second analytical step, we sought totest the stability of the four-cluster solutionand maximize the accuracy of the assignmentof families to the clusters by conducting a non-hierarchical cluster analysis with the same data.Because hierarchical agglomerative methodsof cluster analysis successively merge individ-uals into progressively smaller sets of clus-ters, participants assigned to clusters in theearly iterations of the analysis may not resem-ble the changing properties of the increas-ingly larger clusters formed in the subsequentiterations and final solution. In contrast, non-hierarchical clustering methods permit reclas-sification of cases to clusters in an effort tomaximize the between-cluster variance in rela-tion to within-cluster variance~Grych, Jouriles,Swank, McDonald, & Norwood, 2000!. There-fore, we conducted a nonhierarchical clusteranalysis~i.e., K-means! to obtain a more pre-cise classification of participants into the fourclusters. Given that any single clusteringmethod can generate idiosyncratic, artificial,or skewed solutions, replicating the clustersolution usingK-means analysis holds the addi-tional advantage of testing the stability of theclusters.

Consistent with conventional quantitativeprocedures~Fals–Steward, Schafer, & Birchler,1993!, the initial cluster centers generated fromthe hierarchical analysis were used as startingpoints for theK-means analysis. Comparisonof the four cluster solutions obtained from the

hierarchical andK-means analyses indicatedthat there was considerable correspondence inthe classification of families across the meth-ods~81% agreement;k 5 .71!. Moreover, thestrong correlations between the mean profilesof the family process variables of the clustersgenerated by the Ward’s andK-means algo-rithms indicated that the shape of the profilesfrom the two cluster methods were highly sim-ilar, with rs ranging from .92 to 1.00. Giventhe robustness of the four-cluster solutionacross different methods of analysis, we fol-lowed the practice of using the results of theK-means analysis as our final solution~seeGrych et al., 2000!.

Table 2 shows the means and standard devi-ations of the standardized measures of familyprocess for each of the four family clustersidentified in theK-means analysis~also seeFigure 1!. The first cluster, which consisted of46% of the sample~n 5 102!, displayed lowlevels of spousal hostility, coparental disagree-ments, and parental psychological control andinconsistent discipline along with moderate tohigh levels of spousal affection and parentalacceptance. On the basis of earlier research~Kerig, 1995!, these families were interpretedas exhibiting a cohesive profile of function-ing. Parents of families in the second cluster~n518; 8%! reported very high levels of spou-sal hostility, coparenting disagreements, psy-chological control, and inconsistent discipline,while reporting modest to moderate levels ofspousal affection and parental acceptance rel-ative to the other clusters. Because the highlevels of coparenting difficulties and parentalpsychological control were interpreted toreflect considerable difficulties in regulatingaffect and maintaining roles across familysubsystems, we labeled these families asenmeshed. The third cluster consisted of 29%of the sample~n 5 64!. Although these fami-lies generally reported moderately high levelsof spousal hostility, coparental disagreements,psychological control, and inconsistent dis-cipline, the distinctiveness of this cluster liesin the extremely low levels of spousal andparental affection~i.e., parental acceptance!.Accordingly, these families were labeled asdisengaged. The fourth cluster of families~n537; 17%! closely resembled cohesive families


in reporting harmonious marriages~i.e., lowspousalhostilityandcoparentingdisagreements,high spousal affection! and healthy function-ing along two of the three dimensions parent-ing ~i.e., high acceptance, modest levelsinconsistent discipline!. However, unlike cohe-sive families, they reported using relatively highlevels of psychological control tactics with theirchildren. Given the predominance of healthyfunctioning, these families were interpreted asdisplaying adequate~but not optimal! func-tioning ~e.g., Lynch & Cicchetti, 1997!.

Links between profiles of family functioningand child emotional insecurity

To examine whether the profiles of family func-tioning were associated with dimensions ofchildren’s insecurity, the five measures of childinsecurity were submitted as dependent vari-ables to a series of analyses of covariance,with the four profiles of family functioningserving as the independent variable and childgender serving as the covariate. Significantanalyses of variance were followed up withplanned comparisons~t tests! to examinewhether children in each of the three nonopti-mal profiles of family functioning ~i.e.,enmeshed, withdrawn, adequate! differed sig-nificantly from children in the cohesive fam-ily typology in their reactivity to maritaldifficulties. The results, which are presented

in Table 3, indicated that all five child responsesdiffered significantly as a function of familytypology. Results of the planned comparisonsindicated that children from enmeshed fami-lies experienced significantly higher levels ofreactivity to conflict than children from cohe-sive families across all five assessments of inse-curity, while children from disengaged familiesexhibited higher levels of the three subjectiveindices of insecurity~i.e., subjective distress,involvement impulses, threat appraisals! thanchildren from cohesive families. Conversely,despite experiencing elevated threat apprais-als, children from adequate families exhibitedsimilar levels of security to children in cohe-sive families.

Links between profiles of family functioningand child psychological symptoms

To examine associations between family pro-files and child psychological symptoms overthe two measurement occasions, a series ofrepeated measures multivariate analyses ofcovariance were conducted for each depen-dent measure~i.e., teacher and parent reportsof internalizing and externalizing symptoms!using family profile as the between-subjectsfactor and time~i.e., measurement occasion!as the within-subjects factor. Child genderserved as a covariate in each analysis. Givenour primary aim of understanding links

Table 2. Means, standard deviations, and statistical comparisons of the four family profilesof functioning on the six defining variables

Cohesive~C!

~n 5 102!

Enmeshed~E!

~n 5 18!

Disengaged~D!

~n 5 64!

Adequate~A!

~n 5 37!

M SD M SD M SD M SD F~3, 217! Post Hoc

Interparental hostility 20.53 0.65 0.56 1.02 0.93 0.8820.42 0.64 56.72*** E, D. C, AInterparental affection 0.44 0.6820.27 1.15 20.83 1.01 0.36 0.68 33.73*** C, A. D, ECoparenting

disagreements 20.41 0.69 1.71 1.21 0.44 0.8920.45 0.46 49.99*** E. D . C, AParent psychological

control 20.65 0.56 1.75 0.85 0.00 0.65 0.95 0.69 108.43*** E. A . D . CParent acceptance 0.04 0.79 0.18 1.1520.72 0.88 1.04 0.55 37.24*** A. C, E . DInconsistent parent

discipline 20.39 0.85 1.70 0.83 0.22 0.8520.15 0.72 34.74*** E. C, D, A; D . C

*** p # .001.


Figure 1. The means of the four family profiles of functioning on the six defining variables.

53

8

between family systems profiles and child psy-chological functioning, only family profilemain effects and Family Profile3 Time inter-actions are described. For each set of analy-ses, family profile main effects and FamilyProfile3 Time interactions were further eval-uated at each measurement occasion to pin-point whether child symptoms accompaniedor followed the assessment of family profiles.

Table 4 provides a summary of the findingsfrom these analyses. Relative to children incohesive families, children in enmeshed fam-ilies experienced significantly higher levels ofparent-reported internalizing and externaliz-ing symptoms at Time 1 and 2 and higher lev-els of teacher-reported externalizing symptomsat Time 2. In comparison with children in cohe-sive families, children in disengaged familiesdisplayed greater externalizing symptomsacross both informants and measurement occa-sions. Children in disengaged families alsoexperienced elevated levels of parent-reportedinternalizing symptoms. Finally, children inadequate families were generally comparableto cohesive families in their psychologicalsymptoms with one exception: they exhibitedhigher levels of teacher-reported externaliz-ing symptoms than did children in cohesivefamilies at Time 2.

Links between child emotional insecurityand psychological symptoms

To examine links between children’s signs ofinsecurity in the martial subsystem and their

psychological symptoms over the two mea-surement occasions, we conducted a series ofrepeated-measures multiple regression analy-ses. Each form of child conflict reactivity wassubmitted as a predictor in separate modelspredicting each of the four measures of psy-chological symptoms. Time~i.e., first or sec-ond measurement occasion! served as awithin-subjects factor in the model, and childgender was treated as a covariate. Given ouraim of delineating concurrent and prospectivelinks between child conflict reactivity and childpsychological symptoms, only child insecu-rity main effects and Child Insecurity3 Timeinteractions are described. Significant maineffects or interactions were followed up withregression analyses designed to specify thenature and magnitude of the relationshipbetween the child insecurity variable and thepsychological symptomatology measure ateach measurement occasion.

Table 5 shows the results of the modelspredicting child externalizing symptoms.Children’s subjective and behavioral distressreactions to interparental conflict predictedboth parent and teacher reports of externaliz-ing symptoms. With the exception of the non-significant link between subjective distress andparent reports of externalizing symptoms atTime 1, follow-up regression analyses indi-cated that child subjective and behavioral signsof insecurity were significantly associated withparent and teacher reports of externalizingsymptoms across both time points. Childbehavioral avoidance was also significantly

Table 3. The relation of family profiles of functioning to indices of children’s emotionalinsecurity in the interparental relationship

Family Profile

Cohesive Enmeshed Disengaged Adequate

M SD M SD M SD M SD F~3, 216! h2

Subjective distress 0.92 1.97 2.33 4.01* 1.92 3.16** 1.03 2.30 2.67* .036Involvement impulses 3.58 1.75 5.06 2.42** 4.16 1.92* 3.88 2.14 3.70** .049Threat appraisals 1.70 1.92 3.39 2.35*** 2.52 2.06** 2.57 2.41* 4.66** .061Behavioral distress 2.25 2.51 4.56 4.00** 2.53 3.20 1.65 2.19 4.39** .057Behavioral avoidance 0.15 0.48 0.61 1.04*** 0.23 0.56 0.14 0.35 4.00*** .053

*p # .05. **p # .01. *** p # .001.


associated with parent, but not teacher, reportsof externalizing symptoms at Time 1 and 2.

The results of the models predicting inter-nalizing symptoms, which are shown inTable 6, indicate that child insecurity mea-sures were consistent predictors of internaliz-ing symptoms. Children’s subjective distressto conflict was a consistent predictor of both

parent and teacher reports of Time 2 internal-izing symptoms. However, the robustness ofthe other measures of insecurity as predictorsof internalizing symptoms varied across thetype of informant. Children’s behavioral dis-tress and avoidance reactions to conflict pre-dicted parent, but not teacher, reports ofinternalizing symptoms at both time points.

Table 4. Statistical comparisons of the family profiles for parent and teacher reports ofchild internalizing and externalizing symptoms

Family Profile

Cohesive Enmeshed Disengaged Adequate FamilyFamily3 Time

M SD M SD M SD M SD F~3, 216! F ~3, 216!

Internalizing symptomsTeacher report

T1 5.13 5.68 6.13 6.10 6.39 6.47 3.84 3.56F 5 1.93 F 5 2.10T2 4.72 4.94 6.43 5.48 6.95 7.47 6.97 7.49h2 5 .026 h2 5 .028

Parent reportT1 6.46 5.25 11.14** 7.55 7.87 5.18 5.86 3.96 F 5 5.66*** F 5 0.37T2 6.22 4.53 10.77*** 8.24 7.84* 5.13 6.30 4.69h2 5 .073 h2 5 .005

Externalizing symptomsTeacher report

T1 3.00 4.95 5.74 11.12 5.84** 6.52 4.32 6.89 F 5 4.97** F 5 0.58T2 3.35 5.76 6.88* 7.94 7.58*** 9.12 5.80* 7.24h2 5 .065 h2 5 .008

Parent reportT1 8.27 6.02 15.97*** 10.12 11.80*** 6.13 9.51 5.01 F 5 11.35*** F 5 0.97T2 7.49 5.41 14.14*** 9.58 11.46*** 6.56 9.30 6.05h2 5 .136 h2 5 .013

*p # .05. **p # .01. *** p # .001.

Table 5. Repeated measures general linear model analyses predicting child externalizingsymptoms from indicators of child emotional insecurity in the interparental relationship

Externalizing Symptoms~T! Externalizing Symptoms~P!

Time 1 Time 2 Time 1 Time 2

Child Insecurity Variables F ~1, 218! B SE B SE F~1, 218! B SE B SE

Model 1: subjective distress 9.89** .53** .18 .49** .20 3.93* .27 .17 .37* .17Subjective Distress3 Time 0.07 1.06

Model 2: involvement impulses 2.58 — — 1.40 — —Involvement Impulses3 Time 0.00 0.27

Model 3: perceived threat 2.37 — — 1.68 — —Perceived Threat3 Time 0.18 0.06

Model 4: behavioral distress 11.44*** .46** .16 .52** .17 11.22** .49** .15 .47** .15Behavioral Distress3 Time 0.15 0.06

Model 5: behavioral avoidance 2.37 — — 11.30** 2.84** .78 2.17** .78Behavioral Avoidance3 Time 0.14 2.14

Note: ~T!, teacher report;~P!, parent report.*p # .05. **p # .01. *** p # .001.


Conversely, children’s subjective impulses tointervene and appraisals of threat were asso-ciated with teacher, but not parent, reports ofinternalizing symptoms. In the model of teacherreports of internalizing symptoms, follow-uptests of the Marginal Perceived Threat3 Timeinteraction indicated that perceived threat wassignificantly linked with teacher-reported inter-nalizing symptoms at Time 2.

Mediational paths between familyfunctioning, emotional security,and child symptoms

Structural equation modeling~SEM! with max-imum likelihood estimation was used to testthe mediational role of emotional insecurityin the links between profiles of family func-tioning and child internalizing and externaliz-ing symptoms. Linkages among the profilesof family functioning, emotional insecurity, andchild psychological symptoms were consis-tently found for planned comparisons involv-ing disengaged and enmeshed families but notfor adequate families. Therefore, consistentwith recommendations by Baron and Kenny~1986!, our analyses examined whether thehigher levels of psychological symptoms ex-hibited by children in disengaged and en-meshed families relative to children in cohesivefamilies were partially mediated by childemotional insecurity in interparental relation-

ships. Given the potentially distinct nature ofmediational paths involving enmeshed anddisengaged families, separate structural equa-tion models were conducted to compare~a!enmeshed versus cohesive families and~b! dis-engaged versus cohesive families. Thus, in eachmodel, the exogenous variable represented adummy variable that contrasted one of the twomaladaptive profiles~1! with the cohesive pro-file ~0!. The five indicators of emotional inse-curity each served as the manifest indicatorsof a latent construct of emotional insecurity inthe interparental relationship. To provide amore conservative test of our mediationalhypotheses that temporally separates assess-ments of security from child symptoms,parent and teacher reports of psychologicalsymptoms at Time 2 were utilized as manifestindicators of latent variables of internalizingand externalizing symptoms.

If the SEM results met criteria for media-tion outlined by Baron and Kenny~1986!, wealso tested the significance of the indirect ormediational paths involving emotional secu-rity using the two methods developed byMacKinnon, Lockwood, Hoffman, West, andSheets~2002!. In the first method, the productof the two path coefficients in the indirect path~i.e., paths running from family type to emo-tional security and emotional security to childmaladjustment! is divided by its standard errorand compared with critical values of the sam-

Table 6. Repeated measures general linear model analyses predicting child internalizingsymptoms from indicators of child emotional insecurity in the interparental relationship

Internalizing Symptoms~T! Internalizing Symptoms~P!

Time 1 Time 2 Time 1 Time 2

Child Insecurity Variables F ~1, 218! B SE B SE F~1, 218! B SE B SE

Model 1: subjective distress 5.19* .15 .15 .41** .16 4.21* .20 .14 .33* .14Subjective Distress3 Time 1.78 1.75

Model 2: involvement impulses 3.90* .41* .20 .23 .22 1.79 — —Involvement Impulses3 Time 0.49 0.09

Model 3: perceived threat 3.35† .06 .18 .49** .20 0.74 — —Perceived Threat3 Time 3.39† 0.06

Model 4: behavioral distress 1.31 — — 14.51** .47** .12 .40** .12Behavioral Distress3 Time 0.08 0.54

Model 5: behavioral avoidance 0.20 — — 12.08** 2.26** .64 1.85** .63Behavioral Avoidance3 Time 0.09 0.88

Note: ~T!, teacher report;~P!, parent report;~C!, child report.†p , .07. *p # .05. **p # .01.


pling distribution for product terms. The result-ing z' provides a test of the significance of theproposed indirect path. In the second method,z scores are obtained for each of the indirectpaths by dividing the respective path coeffi-cients by their standard errors. TheP statistic,which is the product of the twozscores, is thencalculated and compared with the sampling dis-tribution for the product of two normal randomvariables~see Elliot & Reis, 2003, for moredetails!.

Figure 2 contains the results of the familyenmeshment models in the prediction of inter-nalizing and externalizing symptoms, respec-tively. Examining the mediational model firstrequires demonstrating that the proposed pre-dictor is associated with the outcome. Thus, inthe first stage of analyses, we specified mod-els in which family enmeshment predicted childinternalizing and externalizing symptoms afterconstraining paths between emotional insecu-rity and internalizing and externalizing symp-toms to 0. Estimates of the standardized pathcoefficients from these models, which areshown in the brackets in Figure 2, indicatethat family enmeshment was a significant pre-dictor of subsequent internalizing and exter-nalizing symptoms. Accordingly, the pathsbetween emotional insecurity and child psy-chological symptoms were subsequently esti-mated in the second stage of analyses. AsFigure 2 shows, the chi squares and goodnessof fit indices ~GFI! suggest that these modelsprovide a good fit to the data for internalizingsymptoms~x2 5 28.64,df517p5 .04, GFI5.94, adjusted GFI@AGFI# 5 .88, root meansquare error analysis@RMSEA# 5 .08! andexternalizing symptoms~x2 5 24.28,df517,p 5 .11, GFI5 .95, AGFI5 .89, RMSEA5.06!. In support of the measurement portion ofthe models, indicator loadings for each of thelatent constructs in the models were statisti-cally significant and, on average, moderate inmagnitude~meanl 5 .44!. Consistent withour mediational hypotheses, family enmesh-ment was associated with greater child emo-tional insecurity, which in turn, predictedhigher levels of child internalizing and exter-nalizing symptoms at Time 2. Further support-ing the mediational role of insecurity, themagnitude of the standardized paths between

family enmeshment and child internalizing andexternalizing symptoms was substantiallyreduced when insecurity was included in themodel~74 and 45% reductions, respectively!.Follow up tests using the two methods recom-mended by MacKinnon et al.~2002! indicatedthat the indirect pathway involving emotionalsecurity was significant in the prediction ofinternalizing ~z' 5 1.79, p , .05, P 5 7.80,p , .05! and externalizing~z'51.62,p , .05,P 5 5.49,p , .05! symptoms.

Figure 3 contains the results of the modelsestimating linkages between family disengage-ment and child internalizing and externalizingsymptoms. To examine the mediational pre-condition requiring significant associationsbetween family disengagement and child symp-toms, we first specified models in which fam-ily disengagement predicted child internalizingand externalizing symptoms after constrain-ing paths between emotional insecurity andinternalizing and externalizing symptoms to 0.Estimates of the standardized path coeffi-cients from these models are shown in thebrackets in Figure 3. In both models, familydisengagement was a significant predictor ofsubsequent internalizing and externalizingsymptoms. Accordingly, to examine meda-tional hypotheses, the paths between emo-tional insecurity and child psychologicalsymptoms were subsequently estimated in themodels. As Figure 2 shows, the chi squaresand GFIs suggest that these models provide agood fit to the data for internalizing symp-toms~x2 519.03,df518,p5 .39, GFI5 .97,AGFI 5 .95, RMSEA5 .02! and externalizingsymptoms~x2 516.33,df517,p5 .50, GFI5.98, AGFI5 .95, RMSEA5 .00!. Support forthe measurement component of the two mod-els is evidenced by moderate and statisticallysignificant indicator loadings for each of thelatent constructs~meanl5 .43!. Moreover, thepattern of findings in the structural componentof the model was consistent with our media-tional hypotheses for the prediction of internal-izing symptoms. Family disengagement waslinkedwithgreater insecurity,which in turn,pre-dicted greater child internalizing symptoms 1year later. In further support of the mediationalhypotheses, the magnitude of the standardizedpath between family disengagement and child


internalizing was reduced by 37%~from .35 to.22! after emotional insecurity was included inthe model. Follow-up tests using the methodsrecommended by MacKinnon et al.~2002! indi-cated that the indirect pathway involving emo-tional security was significant in the prediction

of internalizing symptoms~z'51.50,p , .05,P 5 5.29, p , .05!. In contrast, the nonsig-nificant path between child insecurity andexternalizing symptoms did not support themediational hypotheses in the prediction ofchild externalizing symptoms.

Figure 2. Structural equation models testing the mediational role of children’s emotional insecurity inthe interparental relationship in pathways between family enmeshment and children’s internalizing andexternalizing symptoms.


Discussion

In this study, our goal was to integrate theemotional security hypothesis within a frame-work inspired by family systems theory. Inthe emotional security hypothesis and, morebroadly, the discipline of developmental psy-

chopathology, considerable attention has beendevoted to identifying the coping and adapta-tional patterns of children that develop incontexts of family adversity and ultimatelyincrease their vulnerability to psychopathol-ogy ~Cicchetti, Cummings, Marvin, & Green-berg, 1990; Cicchetti & Toth, 1995; Cummings

Figure 3. Structural equation models testing the mediational role of children’s emotional insecurity inthe interparental relationship in pathways between family disengagement and children’s internalizingand externalizing symptoms.


& Davies, 1996; Shields, Ryan, & Cicchetti,2001!. However, relatively less is known abouthow children cope, adapt, and function in rela-tion to the broader constellation of functioningin the family system. With its aim of decipher-ing the relational meaning of specific familycharacteristics in the wider organization of thefamily unit, family systems conceptualizationsprovide useful heuristics for generating hypoth-eses regarding the interplay between configu-rations of functioning in the family unit andchildren’s stress, coping, and psychopath-ology. Thus, integrating these frameworksgenerated a novel set of research questionsaddressed in this study. More specifically, candistinct configurations of relationship and inter-action properties in the family system be iden-tified through empirical classification methods?If so, are the distinct configurations of familysystems conceptualizations developmentallymeaningful? In other words, are individual dif-ferences in children’s emotional security in theinterparental relationship and their psycholog-ical adjustment associated with different rela-tional configurations in the family? Likewise,does children’s emotional insecurity in the inter-parental relationship partially mediate oraccount for why children exposed to distressedfamily configurations are at risk for develop-ing psychological problems?

Addressing whether family configurationshold developmental utility in understandingchild outcomes first required demonstratingthat specific types of family configurationsderived from family systems theory can bereliably identified. Using a variety of classi-fication schemes, earlier empirical and clini-cal work has consistently identified threeprimary profiles of family functioning charac-terized by cohesion, enmeshment, and disen-gagement. Consistent with this work, three ofthe four family profiles identified in our clus-ter analyses closely resembled these configu-rations. The first cluster consisted of cohesivefamilies who displayed low levels of hostility,conflict, and constraints on autonomy and highlevels of warmth and supportiveness acrossmarital, coparental, and parent–child subsys-tems. Thus, the proportion of families classi-fied as cohesive~i.e., roughly half! and thenature of their profiles of functioning are in

accordance with previous classifications ofcohesive families based on displays of warmth,closeness, and maintenance of boundaries thatpermit family members access to resourceswithout jeopardizing their autonomy~e.g.,Kerig, 1995; Stevenson–Hinde, 1990!.The sec-ond cluster exhibited considerable difficultiesregulating negative affect and maintaining pos-itive affect in the context of diffuse and irreg-ular boundaries between subsystems~e.g.,coparenting difficulties, inconsistent discipline,psychological control!. Given its considerableconceptual overlap with earlier systems classi-fications of enmeshed or chaotic families andits rare occurrence~i.e., less than 10% of thesample!, families in this cluster were labeled asenmeshed~Gilbert & Christensen, 1988; Kerig,1995; Kerr, 2003!.Although families in the thirdcluster resembled enmeshed families in report-ing high levels of marital conflict, features ofdistinction for the third cluster included partic-ularly low levels of positive affect across fam-ily systems and diminished engagement in thefamily. Thus, these families were character-izedas “disengaged”or “separate”~Belskyetal.,1989;Kretchmar&Jacobvitz,2002;Stevenson–Hinde, 1990!. The final cluster, which did notreadily fit in the tripartite classification scheme,consisted of “adequate” families who exhib-ited elevated levels of parental psychologicalcontrol in a broader context warmth, harmony,and support across multiple family subsystems.

From a developmental psychopathologyperspective~Cicchetti, 1993!, the value of ourclassification of family profiles hinges, in largepart, on its power to account for individualdifferences in child coping and adaptation overtime. Analyses of associations between familyprofiles and child functioning indicated thatthe children in enmeshed and disengaged fam-ilies exhibited consistently greater signs ofinsecurity than the children in the cohesivefamilies. In contrast, children in adequate andcohesive families were largely indistinguish-able along indices of emotional security in themarital relationship. In inspecting the familyprofiles, elevated discord in marital and co-parenting relationships were the commondenominators that distinguished the high-risk~i.e., enmeshed, disengaged! and low-risk fam-ily profiles ~i.e., cohesive, adequate!. Within


the emotional security hypothesis, these find-ings are consistent with the assumption thatheightened interparental discord plays a keyrole in increasing child concerns about secu-rity in the marital subsystem~Cummings et al.,2003; Davies et al., 2002!.

However, two important variations aroundthese general conclusions highlight the utilityof adopting a family systems conceptualiza-tion of emotional security. First, children inenmeshed families exhibited the most wide-spread signs of insecurity in the interparentalrelationship despite being exposed to levelsof interparental discord that were comparableto children in disengaged families. Enmesh-ment was specifically associated with ele-vated levels of both subjective~i.e., subjectivedistress, involvement impulses, threat apprais-als! and overt~i.e., behavioral distress, avoid-ance! signs of child insecurity in the face ofthe simulated interparental conflict, while chil-dren in disengaged families only evidencedhigher levels of subjective indices of insecu-rity. These findings suggest that exposure tosimilar levels of marital discord may assumedifferent meaning for children depending onthe broader organization of functioning in thefamily system. Specifically, the high levelsof coparenting disagreements, parental psy-chological control, and inconsistent disci-pline in enmeshed families may signify diffuserelationship boundaries in the family that fuelconcerns about insecurity in the interparentalsubsystem by “breaking down any affectivedistance between the children and the par-ents’ distress”~Coyne et al., 1992; p. 222!.For example, diffuse boundaries betweenparents and children and family role confu-sion emerging out of psychological controland inconsistent discipline may intensifychildren’s insecurity in the face of interparen-tal difficulties by constraining their auton-omy, compromising their perceived personalefficacy, and heightening their feelings of guiltand dependency in the family~Davies & For-man, 2002!.

Second, despite considerable similarity inlevels of security among children in adequateand cohesive families, children in adequatefamilies did appraise conflicts between par-ents as more threatening to their well-being

than children in cohesive families. Thus,parent–child difficulties were associated withincreases in children’s perceptions of threat inthe interparental relationship even in the con-text of warmth and harmony in the broaderfamily system. These findings suggest thatsources of family adversity outside the copa-rental and marital subsystems may undermine~albeit in a limited way! specific componentsof children’s security in the interparental rela-tionship above and beyond the impact of expo-sure to interparental discord. One interpretationis that parental use of intrusive and threaten-ing strategies for managing child behavior inotherwise harmonious family units may actu-ally foster child appraisals of parents as poten-tially threatening, vulnerable, or volatile figures~Rogosch, Cicchetti, &Aber, 1995!. These neg-ative parental representations, in turn, mayincrease children’s negative appraisals of theimplications interparental problems have fortheir own welfare~Davies et al., 2002!.

Examining the associations between spe-cific family profiles and child psychologicalsymptoms revealed a similar pattern to themodels predicting child emotional security.Although children in adequate and cohesivefamilies were comparable to each other acrossvirtually all the measures of psychologicaladjustment, enmeshment and disengagementwere consistently associated with elevatedlevels of psychological symptoms. Familyenmeshment was a consistent predictor of bothconcurrent and subsequent internalizing andexternalizing symptoms in children, whereasfamily disengagement was more consistentlyassociated with child externalizing symptomsthan child internalizing symptoms. This pat-tern of findings is largely consistent with afamily systems perspective on emotional secu-rity ~e.g., Coyne et al., 1992; Davies & For-man, 2002!. According to such a perspective,unsupportive, disengaged family relation-ships are especially likely to promote dis-missing patterns of coping characterized byinterpersonal disregard, alienation, and con-duct problems. In contrast, diffuse boundariesbetween subsystems in enmeshed families arepostulated to emotionally draw or pull chil-dren into family difficulties and, as a resultamplify worries and distress. In addition to


serving as a foundation for the developmentof internalizing symptoms, this constellationof worries, preoccupation, and distress mayfoster externalizing symptoms as childrenassume roles as family “decoys” or “detours”in deflecting attention away from more seri-ous family issues to child behavior problems~Emery, 1989; Kerig, 1995; Stormshak, Speltz,DeKlyen, & Greenberg, 1997!.

Our final aim was to examine whetherchildren’s emotional security in the interparen-tal relationship mediated the link betweenenmeshed and disengaged family configura-tions and child psychological symptoms. Sup-porting the preconditions of mediation andearlier empirical findings~Davies & Cum-mings 1998; Davies & Forman, 2002; Davieset al., 2002!, indices of insecurity were con-sistently associated with parent and teacherreports of internalizing and externalizing symp-toms both concurrently and prospectively. Inaccordance with the emphasis on affectivemechanisms in the emotional security hypoth-esis, behavioral and subjective distress re-sponses to interparental conflict were theindices of insecurity that were most consis-tently associated with child symptoms. Theresults of the prospective, multimethod, andmultiinformant SEM analyses further indi-cated that emotional insecurity partially medi-ated links between:~a! family enmeshment andchildren’s internalizing and externalizing symp-toms 1 year later and~b! family disengage-ment and child internalizing symptoms 1 yearlater. These results provide specific supportfor our hypothesis that exposure to familyenmeshment or disengagement sensitizes childconcerns about their emotional security, asmanifested in high levels of behavioral andself-reported distress, avoidance, involve-ment, and negative representations of interpa-rental relations. Emotional insecurity, in turn,not only predicted child symptomatology, butalso accounted for part of the associationbetween enmeshed and disengaged familyfunctioning and child symptomatology. Con-sistent with the outcome specificity hypoth-esis~Davies et al., 2002!, emotional insecuritywas a particularly robust mediator of associa-tions between family enmeshment and disen-gagement and child internalizing symptoms.

According to this hypothesis, the worry, vigi-lance, and preoccupation with interparental dif-ficulties that reflect emotional insecurity maycohere and crystallize into broader maladap-tive coping patterns~e.g., learned helpless-ness, rumination, self-blame! over time and,in the process, increase children’s specific vul-nerability to internalizing symptoms.

Several limitations of this study and corre-sponding directions for future research meritdiscussion. Results from our classification pro-cedure for identifying family configurationsare open to alternative interpretations. Forexample, our analysis of the enmeshed clusterhinges on the assumption that high levels ofpsychological control, inconsistent discipline,and coparenting disagreements are reflectiveof boundary dissolution across family subsys-tems. Likewise, accepting our interpretationof the existence of a disengaged family typol-ogy presupposes that diminished positive affectin conjunction with high levels of hostility anddiscord in the family system are definingfeatures of disengagement and alienation infamily relationships. Thus, replicating thesefindings using different assessment techniquesis an important empirical direction in this earlystage of research~see Kerig, 1995; Lindahl &Malik, 2001!. Likewise, although there is con-siderable conceptual overlap between our pat-terns of functioning in our family profiles andthe profiles identified in family systems theoryand research~e.g., Kerig, 1995; Stevenson–Hinde, 1990!, explanations derived from otherconceptual frameworks may also be employedto readily account for our findings~e.g., sociallearning theory!. For example, the lack of sup-port for the mediational role of emotionalsecurity in associations between family disen-gagement and externalizing symptoms may beinterpreted to suggest that other emotional,social–cognitive, or behavioral mechanisms areoperating as mediators. Interpreted withinsocial learning theory, the direct paths betweenfamily disengagement and child externalizingsymptoms may specifically reflect children’smodeling of callous, hostile behavioral stylesin the family. Likewise, cohesive, disengaged,and enmeshed profiles of family functioningmay reflect underlying secure, avoidant, andresistant patterns of attachment, respectively.


Given the well-established links between indi-vidual differences in attachment and child psy-chological adjustment, it is plausible that partof the association between family configura-tions and child maladjustment may beaccounted for by parent–child attachmentsecurity ~Cicchetti & Howes, 1991; Marvin& Stewart, 1990, Stevenson–Hinde, 1990!.Caution also should be exercised in general-izing findings to other samples. As models ofdevelopmental psychopathology underscore~Cicchetti & Howes, 1991; Cowan & Cowan,2002!, the substance, function, correlates, andsequelae of family configurations in high riskpopulations may be different from those ofour community sample.

Finally, although our prospective design isan advance over the predominant use of cross-sectional studies in the literature, our designand analyses cannot disentangle the temporalordering of relationships between family pro-files, child security, and child adjustment. Illus-trating the limitations characteristic of short-term longitudinal designs, the high stability ofpsychological symptoms across our relativelynarrow longitudinal window precluded the useof more powerful models~e.g., autoregressivedesigns! designed to identify individual differ-ences in psychological symptom changes overtime. Issues regarding directionality raise fur-ther questions about the operation of alterna-tive pathways and processes that may help toexplain the mediational paths in this study.

Thus, although the mediational results are con-sistent with our prediction that patterns of fam-ily adversity increase child vulnerability toadjustment problems by undermining theirsense of security in the interparental relation-ship, transactional models of developmentalpsychopathology and family systems theoriesshare the assumption that associations betweenfamily and child functioning are products ofbidirectional, reciprocal pathways of influ-ences in the family~Cicchetti & Howes, 1991;Minuchin, 1985!. Thus, the mediational path-ways may reflect the impact children’s copingpatterns and adaptation have on the welfare oftheir parents’ marriage and the family~Davies& Cummings, 1994; Schermerhorn, Cum-mings, & Davies, 2004!.

In summary, the present findings supportthe utility of understanding children’s emo-tional security in the interparental relationshipin the context of broader family relational pat-terns encompassing the marital, coparenting,and parent–child subsystems. Supporting pre-dictions, specific configurations of family func-tioning characterized by enmeshment anddisengagement were especially robust pre-dictors of children’s emotional insecurity andpsychological symptomtalogy. Furthermore,children’s difficulties preserving their secu-rity in the interparental relationship partiallyexplained why family disengagement andenmeshment were associated with greater psy-chological problems in children.

References

Achenbach, T. M.~1991!. Integrative guide for the 1991CBCL04–18, YSR, and TRF profiles. Burlington, VT:University of Vermont, Department of Psychiatry.

Arellano, C. M., & Markman, H. J.~1995!. The ManagingAffect and Differences Scale~MADS!: A self-reportmeasure assessing conflict management strategies incouples.Journal of Family Psychology, 9, 319–334.

Baron, R. M., & Kenny, D. A.~1986!. The moderator-mediator variable distinction in social psychologicalresearch: Conceptual, strategic, and statistical consid-erations.Journal of Personality and Social Psychol-ogy, 51, 1173–1182.

Belsky, J., Rovine, M., & Fish, M.~1989!. The develop-ing family system. In M. R. Gunnar & E. Thelen~Eds.!,The Minnesota Symposium on Child Psychology: Vol.22. Systems and development~pp. 119–166!. Hills-dale, NJ: Erlbaum.

Borgen, F. H., & Barnett, D. C.~1987!. Applying clusteranalysis in counseling psychology research.Journalof Counseling Psychology, 34, 456–468.

Buchanan, C. M., & Waizenhofer, R.~2001!. The impactof interparental conflict on adolescent children:Considerations of family systems and family struc-ture. In A. Booth, A. C. Crouter, & M. Clements~Eds.!, Couples in conflict~pp. 149–160!. Mahwah,NJ: Erlbaum.

Byng–Hall, J., & Stevenson–Hinde, J.~1991!. Attachmentrelationships within a family system.Infant MentalHealth Journal, 120, 187–200.

Cicchetti, D. ~1993!. Developmental psychopathology:Reactions, reflections, projections.DevelopmentalReview, 13, 471–502.

Cicchetti, D., Cummings, E. M., Marvin, R. S., & Green-berg, M. T. ~1990!. An organizational perspectiveon attachment beyond infancy. In M. T. Greenberg,D. Cicchetti, & E. M. Cummings~Eds.!, Attachmentin the preschool years~pp. 51–95!. Chicago: Univer-sity of Chicago Press.

Cicchetti, D., & Howes, P. W.~1991!. Developmental psy-chopathology in the context of the family: Illustra-


tions from the study of child maltreatment.CanadianJournal of Behavioural Science, 23, 257–281.

Cicchetti, D., & Toth, S. L.~1995!. A developmental psy-chopathology perspective on child abuse and neglect.Journal of the American Academy of Child and Ado-lescent Psychiatry, 34, 541–565.

Clarke–Stewart, K. A., Allhusen, V. D., McDowell, D. J.,Thelen, L., & Call, M. D.~2003!. Identifying psycho-logical problems in young children: How do motherscompare with child psychiatrists.Journal of AppliedDevelopmental Psychology, 23, 589–624.

Combrinck–Graham, L.~1985!. A developmental modelfor family systems.Family Process, 24, 139–150.

Combrinck–Graham, L.~1990!. Developments in familysystems theory and research.Journal of the AmericanAcademy of Child and Adolescent Psychiatry, 29,501–512.

Cowan, P.A., & Cowan, C. P.~2002!. Interventions as testsof family systems theories: Marital and family relation-ships in children’s development and psychopathology.Development and Psychopathology, 14, 731–759.

Cox, M. J., & Paley, B.~1997!. Families as systems.AnnualReview of Psychology, 48, 243–267.

Coyne, J. C., Downey, G., & Boergers, J.~1992!. De-pression in families: A systems perspective. In D.Cicchetti & S. L. Toth~Eds.!, Rochester symposiumon developmental psychopathology: Vol. 4. Develop-mental perspectives on depression~pp. 211–249!.Rochester, NY: University of Rochester Press.

Cummings, E. M.~1994!. Marital conflict and children’sfunctioning.Social Development, 3, 16–36.

Cummings, E. M., & Davies, P.~1996!. Emotional secu-rity as a regulatory process in normal developmentand the development of psychopathology.Develop-ment and Psychopathology, 8, 123–139.

Cummings, E. M., & Davies, P. T.~2002!. Effects of mar-ital discord on children: Recent advances and emerg-ing themes in process-oriented research.Journal ofChild Psychology and Psychiatry, 43, 31–63.

Cummings, E. M., Goeke–Morey, M. C., & Papp, L. M.~2003!. A family-wide model for the role of emotionin family functioning. Marriage & Family Review,34, 13–34.

Davies, P. T., & Cummings, E. M.~1994!. Marital con-flict and child adjustment: An emotional securityhypothesis.Psychological Bulletin, 116, 387–411.

Davies, P. T., & Cummings, E. M.~1998!. Exploringchildren’s emotional security as a mediator of the linkbetween marital relations and child adjustment.ChildDevelopment, 69, 124–139.

Davies, P. T., & Forman, E. M.~2002!. Children’s pat-terns of preserving emotional security in the interpa-rental subsystem. Child Development, 73, 1880–1903.

Davies, P. T., Harold, G. T., Goeke–Morey, M., & Cum-mings, E. M. ~2002!. Children’s emotional securityand interparental conflict.Monographs of the Societyfor Research in Child Development, 67, 1–129.

Elliot, A. J., & Reis, H. T.~2003!. Attachment and explo-ration in adulthood.Journal of Personality and SocialPsychology, 85, 317–331.

Emery, R. E.~1982!. Interparental conflict and the chil-dren of discord and divorce.Psychological Bulletin,92, 310–330.

Emery, R. E.~1989!. Family violence.American Psychol-ogist, 44, 321–328.

Fals–Stewart, W., Schafer, J., & Birchler, G. R.~1993!.An empirical typology of distressed couples that is

based on the areas of change questionnaire.Journalof Family Psychology, 7, 307–321.

Fauber, R. E., Forehand, R., Thomas, A. M., & Wierson,M. ~1990!. A mediational model of the impact of mar-ital conflict on adolescent adjustment in intact anddivorced families: The role of disrupted parenting.Child Development, 61, 1112–1123.

Frick, P. J., Christian, R. E., & Wootton, J. M.~1999!. Agetrends in association between parenting practices andconductproblems.BehaviorModification,23, 106–128.

Garbarino, J., Sebes, J., & Schellenbach, C.~1984!. Fam-ilies at risk for destructive parent–child relations inadolescence.Child Development, 55, 174–183.

Gilbert, R. K., & Christensen, A.~1988!. The assessmentof family alliances.Advances in Behavioral Assess-ment of Children and Families, 4, 219–252.

Gonzales, N. A., Pitts, S. C., Hill, N. E., & Roosa, M. W.~2000!. A mediational model of the impact of interpa-rental conflict on child adjustment in a multiethnic,low-income sample.Journal of Family Psychology,14, 365–379.

Grych, J. H.~2000!. The Children’s Perception of Inter-parental Conflict Scale for young children. Unpub-lished manuscript, Marquette University.

Grych, J. H., & Fincham, F. D.~1990!. Marital conflictand children’s adjustment: A cognitive–contextualframework.Psychological Bulletin, 108, 267–290.

Grych, J. H., Jouriles, E. N., Swank, P. R., McDonald, R.,& Norwood, W. D. ~2000!. Patterns of adjustmentamong children of battered women.Journal of Con-sulting and Clinical Psychology, 68, 84–94.

Grych, J. H., Seid, M., & Fincham, F. D.~1992!. Assess-ing marital conflict from the child’s perspective.ChildDevelopment, 63, 558–572.

Hayden, L. C., Schiller, M., Dickstein, S., Seifer, R.,Sameroff, A. J., Miller, I., Keitner, G., & Rasmussen,S.~1998!. Levels of family assessment I: Family, mar-ital, and parent–child interaction.Journal of FamilyPsychology, 12, 7–22.

Hinde, R. A. ~1988!. Introduction. In R. A. Hinde & J.Stevenson–Hinde~Eds.!, Relationships within families:Mutual influences~pp. 1–4!. Oxford: Clarendon Press.

Jouriles, E. N., Murphy, C., Farris, A. M., Smith, D. A.,Richters, J. E., & Waters, E.~1991!. Marital adjust-ment, childrearing disagreements, and child behaviorproblems: Increasing the specificity of the maritalassessment.Child Development, 62, 1424–1433.

Kerig, P. K.~1995!. Triangles in the family circle: Effectsof family structure on marriage, parenting, and childadjustment.Journal of Family Psychology, 9, 28–43.

Kerr, M. ~2003!. Multigenerational family systems theoryof Bowen and its application. In G. P. Sholevar~Ed.!,Textbook of family and couples therapy: Clinical appli-cations ~pp. 103–126!. Washington, DC: AmericanPsychiatric Publishing.

Kretchmar, M. D., & Jacobvitz, D. B.~2002!. Observingmother–child relationships across generations: Bound-ary patterns, attachment, and the transmission of care-giving. Family Process, 41, 351–374.

Liddle, H. A., & Schwartz, S. J.~2002!. Attachment andfamily therapy: Clinical utility of adolescent–familyattachment research.Family Process, 41, 455–476.

Lindahl, K. M., & Malik, N. M. ~2001!. The system forcoding interactions and family functioning. In P. K.Kerig & K. M. Lindahl ~Eds.!, Family observationalcoding systems: Resources for systemic research~pp.77–91!. Mahwah, NJ: Erlbaum.


Lorr, M. ~1994!. Cluster analysis: Aims, methods, andproblems. In S. Stack & M. Lorr~Eds.!, Differentiat-ing normal and abnormal personality~pp. 179–195!.New York: Springer.

Lynch, M., & Cicchetti, D.~1997!. Children’s relation-ships with adults and peers: An examination of ele-mentary and junior high school students.Journal ofSchool Psychology, 35, 81–99.

Lynch, M., & Cicchetti, D. ~1998!. An ecological–transactional analysis of children and contexts: Thelongitudinal interplay among child maltreatment,community violence, and children’s symptomatology.Development and Psychopathology, 10, 235–258.

MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M.,West, S. G., & Sheets, V.~2002!. A comparison ofmethods to test mediation and other intervening vari-able effects.Psychological Methods, 7, 83–104.

Margolies, P. J., & Weintraub, S.~1977!. The revised56-item CRPBI as a research instrument: Reliabilityand factor structure.Journal of Clinical Psychology,33, 472–476.

Margolin, G., Oliver, P., & Medina, A.~2001!. Concep-tual issues in understanding the relation between inter-parental conflict and child adjustment: Integratingdevelopmental psychopathology and risk0resilienceperspectives. In J. Grych & F. Fincham~Eds.!, Childdevelopment and interparental conflict~pp. 9–38!. NewYork: Cambridge University Press.

Marvin, R. S., & Stewart, R. B.~1990!. A family systemsframework for the study of attachment. In M. Green-berg, D. Cicchetti, & E. M. Cummings~Eds.!, Attach-ment in the preschool years: Theory, research, andintervention ~pp. 51–86!. Chicago: University ofChicago.

Mikulincer, M., Florian, V., Cowan, P. A., & Cowan, C. P.~2002!. Attachment security in couple relationships:A systemic model and its implications for familydynamics.Family Process, 41, 405–434.

Miller, N. B., Cowan, P. A., Cowan, C. P., Hetherington,E. M., & Clingempeel, W. G.~1993!. Externalizing inpreschoolers and early adolescents: A cross-study rep-lication of a family model.Developmental Psychol-ogy, 29, 3–18.

Milligan, G. W., & Cooper, M. C.~1985!. An examinationof procedures for determining the number of clustersin a data set.Psychometrica, 50, 159–197.

Minuchin, P. ~1985!. Families and individual develop-ment: Provocations from the field of family therapy.Child Development, 56, 289–302.

Minuchin, S.~1974!. Families and family therapy. Cam-bridge, MA: Harvard University Press.

Nichols, W. C.~1999!. Family systems therapy. In S. W.Russ & T. H. Ollendick~Eds.!, Handbook of psycho-therapies with children and families. Issues in clin-ical child psychology~pp. 137–151!. New York:Kluwer0Plenum.

O’Connor, T. G., Hetherington, E. M., & Clingempeel,W. G. ~1997!. Systems and bidirectional influences infamilies. Journal of Social and Personal Relation-ships, 14, 491–504.

O’Connor, T. G., Hetherington, E. M., & Reiss, D.~1998!.Family systems and adolescent development: Sharedand nonshared risk and protective factors in nondi-vorced and remarried families.Development and Psy-chopathology, 10, 353–375.

Porter, B., & O’Leary, K. D.~1980!. Marital discord and

childhood behavior problems.Journal of AbnormalChild Psychology, 8, 287–295.

Rogosch, F. A., Cicchetti, D., & Aber, J. L.~1995!. Therole of child maltreatment in early deviations in cog-nitive and affective processing abilities and later peerrelationship problems.Development and Psychopa-thology, 7, 591–609.

Rohner, R. P.~1990!. Handbook for the study of parentalacceptance and rejection. Storrs, CT: University ofConnecticut, Center for the Study of Parental Accep-tance and Rejection.

Schludermann, E., & Schludermann, S.~1970!. Replica-bility of factors in children’s report of parent behavior~CRPBI!. Journal of Psychology, 76, 239–249.

Schermerhorn, A. C., Cummings, E. M., & Davies, P. T.~2004!. A developmental perspective on children asagents in the family. Manuscript submitted forpublication.

Schwarz, J. C., Barton–Henry, M. L., & Pruzinsky, T.~1985!. Assessing child-rearing behaviors: A compar-ison of ratings made by mother, father, child, and sib-ling on the CRPBI.Child Development, 56, 462–479.

Shelton, K. K., Frick, P. J., & Wootton, J.~1996!. Assess-ment of parenting practices in families of elementaryschool-age children.Journal of Clinical Child Psy-chology, 25, 317–329.

Shields, A., Ryan, R. M., & Cicchetti, D.~2001!. Narra-tive representations of caregivers and emotion dysreg-ulation as predictors of maltreated children’s rejectionby peers.Developmental Psychology, 37, 321–337.

Skinner, H. A., & Blashfield, R. K.~1982!. Increasing theimpact of cluster analysis research: The case of psy-chiatric classification.Journal of Consulting and Clin-ical Psychology, 50, 727–735.

Stevenson–Hinde, J.~1990!. Attachment within family sys-tems: An overview.Infant Mental Health Journal, 11,218–227.

Stocker, C.M., & Youngblade, L.~1999!. Marital conflictand parental hostility: Links with children’s siblingand peer relationships.Journal of Family Psychology,13, 598–609.

Stormshak, E. A., Speltz, M. L., DeKlyen, M., & Green-berg, M. T.~1997!. Observing family interaction dur-ing clinical interviews: A comparison of familiescontaining preschool boys with and without disrup-tive behavior.Journal of Abnormal Child Psychol-ogy, 25, 345–357.

Tolan, P. H., & Mitchell, M. E.~1989!. Families and thetherapy of antisocial and delinquent behavior.Jour-nal of Psychotherapy and the Family, 6, 29–48.

Wagner, B., & Reiss, D.~1995!. Family systems and devel-opmental psychopathology: Courtship, marriage, ordivorce? In D. Cicchetti & D. J. Cohen~Eds.!, Devel-opmental psychopathology: Vol. 1. Theory and meth-ods~pp. 696–730!. New York: Wiley.

Wood, B. ~1985!. Proximity and hierarchy: Orthogonaldimensions of family interconnectedness.Family Pro-cess, 24, 487–507.

Wootton, J. M., Frick, P. J., Shelton, K. K., & Silverthorn,P. ~1997!. Ineffective parenting and childhood con-duct problems: The moderating role of callous-unemotional traits.Journal of Consulting and ClinicalPsychology, 65, 301–308.

Wynne, L. C.~1988!. An epigenetic model of family pro-cesses. In C. J. Falicov~Ed.!, Family transitions: Con-tinuity and change over the life cycle~pp. 81–106!.New York: Guilford Press.


Pathways between profiles of family functioning, child ...€¦ · problems through multiple...

Documents

Transcript of Pathways between profiles of family functioning, child ...€¦ · problems through multiple...