Early occuring maternal deppression and maternal negativity in predicting young children emotion regulation and socioemotional
J Abnorm Child Psychol (2007) 35:685–703DOI 10.1007/s10802-007-9129-0Early-occurring Maternal Depression and MaternalNegativity in Predicting Young Children’s EmotionRegulation and Socioemotional DifficultiesAngeline Maughan & Dante Cicchetti & Sheree L. Toth &Fred A. RogoschReceived: 27 April 2006 / Accepted: 13 March 2007 / Published online: 15 May 2007# Springer Science + Business Media, LLC 2007Abstract This longitudinal investigation examined the effects Keywords Early-occurring maternal depression .of maternal depression and concomitant negative parenting Maternal negativity . Emotion regulation .behaviors on children’s emotion regulation patterns and Socioemotional functioning .socioemotional functioning. One hundred fifty-one mothers Mediating and moderating processesand their children were assessed when children were approx-imately 1 1/2-, 3-, 4-, and 5-years of age. Ninety-three of the Over the past several decades, investigative efforts aimed atchildren had mothers with a history of Major Depressive elucidating the effects of maternal depression on early childDisorder (MDD) that had occurred within the first 21 months development have burgeoned (Beardslee et al. 1983; Campbellof the child’s birth, and 58 of the children had mothers without et al. 2004; Cicchetti and Toth 1995; Wickramaratne andany history of MDD. Early-occurring Initial maternal depres- Weissman 1998). Findings from these investigations revealsion predicted children’s dysregulated emotion patterns at age that children who are reared by a depressed caregiver are at4 and decreased perceived competence ratings at age 5. Initial increased risk for a range of maladaptive developmentalmaternal depression also indirectly predicted decreased child outcomes, including socioemotional (Campbell et al. 2004),social acceptance ratings at age 5 through its association with cognitive (Hay et al. 2001), and neurobiological (Ashman et al.dysregulated emotion patterns. Furthermore, the relation 2002) deficits.between maternal depression and children’s decreased social Of particular concern is the overabundance of socio-acceptance was more pronounced in those offspring with a emotional difficulties evidenced by the children of de-history of high versus low maternal negativity exposure. pressed parents. For example, children with a clinicallyFindings increase understanding of the processes by which depressed parent are 13 times more likely than children inmaternal depression confers risk on children’s socioemotional the general population to receive a childhood diagnosis ofadjustment. major depressive disorder if the onset of the parent’s depression occurred before the age of 30 (Wickramaratne and Weissman 1998) and are reported to have elevated rates ofA. Maughan (*) suicidal ideation (Garber et al. 1998) and disruptive andDepartment of Psychiatry, Children’s Hospital Boston, anxiety disorders (Radke-Yarrow et al. 1998). Moreover, theHarvard Medical School,300 Longwood Avenue, offspring of depressed caregivers have been found to be moreBoston, MA 02115, USA antisocial (Wright et al. 2000), aggressive (Zahn-Waxler et al.e-mail: email@example.com 1992), and socially withdrawn (Rubin et al. 1991) during peerD. Cicchetti interactions than the children of nondepressed caregivers.Institute of Child Development, University of Minnesota, Although our knowledge of the socioemotional sequelaeMinneapolis, MN, USA of being reared by a depressed caregiver has expanded greatly, less is known about how and why maternalS. L. Toth : F. A. Rogosch depression poses a significant risk for children’s socioemo-Department of Clinical and Social Sciences,Mt. Hope Family Center, University of Rochester, tional adjustment. For some time, theorists have assertedRochester, NY, USA that emotional problems in childhood reflect difficulties in
686 J Abnorm Child Psychol (2007) 35:685–703the regulation of emotion (Cole and Zahn-Waxler 1992). and ability to process and manage emotions effectively,Therefore, understanding how maternal depression and heightening their risk for future socioemotional difficulties.specific correlates of maternal depression (i.e., maternalnegativity) affect children’s ability to regulate emotioneffectively and the role emotion dysregulation plays in the Maternal Depression (Timing and Chronicity),emergence of socioemotional difficulties in the offspring of Maternal Negativity, and Children’s Emotiondepressed caregivers are much needed areas of research Regulation Functioning(Silk et al. 2006) and serve as the primary investigativegoals of the present prospective longitudinal investigation. Numerous findings in the literature provide evidence to support the deleterious effects of maternal depression and concomitant negative parenting behaviors on children’sGuiding Theoretical Perspective emotion regulation functioning. For example, Field et al. (1985) found that depressed caregivers exhibited lessEmotion regulation is a developmentally acquired process contingent responsivity, provided less stimulation, andthat emerges as a byproduct of both intrinsic features (i.e., showed more flat affect during interactions with theirgenetic heritability, homeostatic mechanisms, and temper- 3-month-old infants than did nondepressed mothers. Inament, Kagan 1994; Rothbart et al. 1995) and extrinsic response, the infants of depressed caregivers weresocioemotional experiences, primarily within the context of fussier, less attentive, less positive, and less active thanearly parent–child interactions (Sroufe 1996). The theoret- the offspring of nondepressed mothers. Similarly, Cohnical assumptions of the emotional security hypothesis have et al. (1990) reported increased rates of negative maternalsignificantly advanced our understanding of the extrinsic factors, including irritability, sadness, disinterest and intrusiveinfluences on emotion regulation development and how the handling, in middle-income nonworking depressed mothersquality of the interparental and parent–child relationships during interactions with their 2-month-old infants as com-can enhance or undermine children’s emotional functioning pared to nondepressed controls. The infants of these depressedand future adjustment (Cummings and Davies 1995; Davies nonworking mothers, in turn, expressed reduced rates ofand Cummings 1994). As posited by Davies and Cummings positive affect as compared to the infants of nondepressed and(1994), feelings of emotional insecurity provide an explan- depressed working caregivers. In an older sample of depressedatory construct for how compromised caregiving environ- offspring, Zahn-Waxler et al. (1990a, b) reported increasedments (e.g., those characterized by elevated levels of marital patterns of overarousal and overinvolvement and feelings ofdiscord and parenting difficulties, as is common in house- guilt in the youngsters of depressed mothers as compared toholds with a depressed caregiver) are associated with the children of healthy caregivers. In fact, longitudinalchildren’s risk for socioemotional problems. Specifically, findings have revealed early and persistent emotionalthe emotional security hypothesis proposes that children’s problems (e.g., depression, anxiety, and externalizing symp-behavioral and emotional responses to stressful events toms) from childhood to late adolescences in children ofrepresent goal-directed strategies aimed at maintaining and depressed mothers (Radke-Yarrow et al. 1998).enhancing feelings of emotional security. Although the Additional evidence indicates that the timing of maternalregulatory processes associated with emotional security depressive episodes over the course of a child’s developmenttypically serve adaptive functions, prolonged efforts to and its impact on child functioning matters (Cogill et al.preserve emotional security in conflictual and/or insensitive 1986). Specifically, depressive episodes in caregivers thatfamily environments can result in maladaptive emotion occur during early stages of development have been foundregulation processes. to have a negative effect on future child adjustment, For many children reared by a depressed caregiver, direct regardless of whether the mother’s depression subsequentlyas well as indirect interactions with that caregiver are remits or not. Murray et al. (1999), for example, reportedoften characterized by less positive, insensitive, and some- that maternal depression that occurred 2-months postpar-times unpredictable interpersonal exchanges, resulting in tum, but not current maternal depressive symptoms,children’s decreased sense of emotional security. When predicted child behavior during both caregiver and peerfeelings of emotional insecurity predominate, environmental interactions as well as maternal reports of child behaviorstressors can easily overwhelm a child’s self-regulatory problems at age 5. Specifically, Murray and colleaguesabilities. Under these conditions, emotion dysregulation— found that the children of postnatally depressed motherstypically of two forms; over- and underregulation—often evidenced higher rates of neuroticism and antisocialresults (Cole et al. 1994). As such, relational and affective behaviors and were more likely to respond negatively todisturbances in families with a depressed caregiver pose a friendly peer initiatives at age 5 than the youngsters ofsignificant threat to children’s feelings of emotional security healthy controls.
J Abnorm Child Psychol (2007) 35:685–703 687 Researchers have also examined how the course or challenges, children who have developed the capacity tochronicity of maternal depressive symptoms impact mater- regulate their emotions effectively are more likely tonal behavior and child functioning (Alpern and Lyons-Ruth engage in social exchanges with peers that are positive1993). When maternal depressive symptoms are chronic, and personally rewarding. Healthy peer interactions then, inparenting behaviors are more likely to be impaired than turn, promote the development of social competence andwhen symptoms are transient or intermittent in nature. adaptive peer relationships. Alternatively, when youngstersTherefore, chronic depression in caregivers may pose more with poorly developed self-regulatory skills are confrontedof a risk to children due to their extended and unremitted with social experiences that tax their capacity to maintainexposure to negative maternal affective states and insensi- emotional control, failures of emotion regulation (e.g.,tive parenting behaviors. In support of the assertion, temper tantrums and aggressive outbursts) are likely toCampbell et al. (1995), for instance, examined the effects result. Emotion regulation difficulties in the peer arenaof depression chronicity on early mother–infant interactions increase children’s vulnerability to peer rejection andand found that chronically depressed mothers and their victimization and social isolation, as well as accompanyinginfants were less positive during mother–infant face-to-face negative social self-perceptions.interactions as compared to dyads with mothers whose Empirical findings consistently support the claim thatdepression had remitted. Additionally, the National Institute deviant emotion regulation capacities in children (i.e.,of Child Health and Human Development (NICHD) Early emotional undercontol [e.g., aggression] and/or overcontrolChild Care Research Network (1999) reported that mothers [e.g., withdrawal]) pose a significant risk for the emergence ofwho reported chronic depressive symptoms were the least dysfunctional social relationships and social self-perceptionssensitive toward their infants and rated their 3-year-old by precluding children from important socialization experi-children as less cooperative and more problematic as ences that aid in the development of socially competentcompared to mothers who endorsed some or no depression. behavior. For example, aggression in childhood has beenIn a more recent NICHD report, maternal sensitivity was linked to peer rejection (Little and Garber 1995) andfound to interact with maternal depressive symptoms to adolescent delinquent activity (Rubin et al. 1995) andpredict child attachment security, such that the children of childhood social withdrawal has been shown to predict feltdepressed mothers low in sensitivity were more likely to be insecurity, loneliness, and negative self-regard in adolescenceinsecurely attached than the youngsters of depressed care- (Rubin et al. 1995).givers rated high in sensitivity (Campbell et al. 2004). In sum, maternal depression and concomitant negativeparenting behaviors have been associated with poor socio- Emotion Regulation as a Mediatoremotional outcomes in children, and negative parenting indepressed caregivers has been found to potentiate (i.e., Given that maternal depression has been found to predictmoderate) the negative effects of maternal depression on both children’s deficits in emotion regulation abilities andchild adjustment. Additionally, depression in caregivers that their socioemotional functioning and that emotion dysregu-occurs early on during the course of a child’s development lation has been linked to problematic behavior in children,and chronic maternal depressive symptoms have both been it is reasonable to hypothesize that the relation betweenfound to have adverse effects on children’s socioemotional maternal depression and maladaptive socioemotional childwell being. outcomes may be mediated by deficits in children’s emotion regulation capacities. To our knowledge, this mediational process has not been explicitly tested longitu-Emotion Regulation and Future Socioemotional dinally in offspring of depressed mothers.Competence In contrast, support for the above mediational hypothesis has been found in children with a history of maltreatment.The ability to regulate emotion effectively plays an integral Maughan and Cicchetti (2002) reported that child maltreat-role in the acquisition of future childhood competencies, ment predicted higher rates of dysregulated emotionparticularly the establishment of peer relationships and patterns, specifically undercontrolled emotion strategies,development of social competence (Sroufe 1996). Inter- which in turn, were related to elevated levels of children’sactions with peers pose a significant challenge to pre- anxious/depressed symptoms. Additional support for theschoolers’ emotion regulation capacities because children mediating role of emotional processes in the associationmust be able to manage the normally-occurring social and between optimal and nonoptimal caregiving environmentsaffective demands of peer exchanges (e.g., remaining and parental behaviors and children’s socioemotionalorganized in response to anger and frustration and adjustment has been found in the marital conflict literaturecooperating with others) (Sroufe 1996). Given these (Davies and Cummings 1998) as well as in non-risk
688 J Abnorm Child Psychol (2007) 35:685–703samples (Eisenberg et al. 2001). For example, Davies and Davies and Forman (2002), and Maughan and CicchettiCummings found evidence for the mediating role of (2002) all of whom utilized person-centered methodologiesemotional security in the marital conflict—child adjustment to identify individual patterns of emotion regulation inrelationship and Eisenberg and colleagues reported that children. For example, in a sample of low-income mal-children’s regulation mediated the relation between both treated children, Maughan and Cicchetti (2002) identifiednegative and positive expressivity in mothers and children’s three person-centered emotion regulation patterns (EMRPs)externalizing behavior problems and social adjustment. based on the integration of children’s discrete emotional behavioral reactions and subjective reports in response to witnessed interadult anger. Children classified with adap-Assessment of Emotion Regulation: Person- Versus tive EMRPs displayed moderate levels of negative affectVariable-centered Approaches that were well modulated during and after anger exposure. In contrast, children identified with undercontrolled EMRPsAn organizational perspective on development has contributed exhibited elevated and prolonged rates of emotionality thatto important advancements in how children’s emotion regula- was often indecisive, disorganized, and not goal oriented.tion abilities are assessed (Cicchetti and Schneider-Rosen Finally, children classified with overcontrolled EMRPs1984; Sroufe 1996). Similar to attachment security (Sroufe were characterized by low levels or the absence of overtand Waters 1977), theorists view emotion regulation as an emotional behavioral reactivity and inhibition of theorganizational construct, the expression of which is manifested expression of visible signs of distress or discomfort inthrough a series of biological, emotional, and behavioral response to the witnessed exchange.reorganizations in response to stressful events. As such, the In sum, person-centered methodologies have effectivelyadaptability of an individual’s emotion regulation abilities is been utilized to examine the emotion functioning of childrennot determined by the calculation of an overall mean score on a from a variety of caregiving environments and provide a moresingle emotion variable. Rather, it is inferred from the useful means of assessing children’s increasingly complex andintegration of numerous emotion indicators and whether the dynamic emotion regulation abilities than commonly appliedemotion regulation pattern that emerges promotes or precludes variable-centered approaches.the individual’s general competence in his/her interactions withthe environment and successful attainment of subsequentdevelopmental tasks (Cicchetti and Schneider-Rosen 1986;Sroufe 1996). Design of the Present Investigation and Hypotheses Relatedly, Bergman and Mangnusson (1997) claim thatthe traditional application of a variable-centered approach The present investigation was designed to examine thein research on developmental psychology and psychopa- prospective longitudinal effects of maternal depression andthology has important limitations in understanding more concomitant negative parenting behaviors on children’scomplex, dynamic processes (such as children’s capacity to person-centered patterns of emotion regulation and socio-regulate emotion) due to it’s use of variables as the main emotional functioning. In addition, the mediating andunits of statistical analyses and the study of linear relation- moderating roles of emotion regulation patterns andships across individuals. Specifically, these authors argue maternal negativity in the association between maternalthat “the modeling/description of variables over individuals depression and children’s socioemotional adjustment werecan be very difficult to translate into properties character- evaluated. With the exception of maternal depression, studyizing single individuals because the information provided variables were assessed at stage-salient periods only ratherby the statistical method is variable oriented, not individual than repeatedly across assessments. For example, emotionoriented” (p.292). As such, Bergman and Mangnusson regulation patterns were evaluated when children werestress the salience of incorporating person-centered meth- approximately 4 years of age because the preschool yearsodologies in research designs, whereby overall profiles or are marked by the emergence of self-regulatory capacitiespatterns of individual functioning, including their interac- and an increase and diversification of emotion regulationtion structures, are of primary analytical interest. abilities (Cicchetti and Schneider-Rosen 1986). In addition, To date, past efforts to study the emotion regulation child social self-perceptions were assessed at age 5 becauseabilities of children from non-optimal caregiving environ- children are increasingly more autonomous and engaged inments have focused largely on variable-centered aspects of their social environments during this age period than inemotion functioning (i.e., levels of aggression, distress, and previous years. Data for the present investigation werewithdrawal), rather than on organized patterns of emotion drawn from a larger ongoing prospective longitudinalregulation. Exceptions to this include work by Cummings investigation of the effects of maternal depression onand colleagues (Cummings 1987; El-Sheikh et al. 1989), parent–child relationships and child development (Cicchetti
J Abnorm Child Psychol (2007) 35:685–703 689et al. 1998). The following hypotheses were advanced on papers, newsletters, medical offices, and community bulle-the basis of the literature: tin boards. A broad community-based sampling strategy was employed, as opposed to recruiting exclusively from(1) Children of mothers with a history of early-occurring treatment facilities, to increase generalizability of findings Major Depressive Disorder (MDD), depression that to depressed mothers with and without a treatment history. occurred between birth and the Initial evaluation, will In addition to having a child of approximately 21-months of be more likely to exhibit dysregulated emotion age, diagnostic inclusion criteria for mothers in the patterns at age 4 and have higher child self-report depressed groups required mothers to meet Diagnostic and ratings of socioemotional difficulties at age 5 than the Statistical Manual for Mental Disorders (DSM-III-R; children of mothers without a history of MDD. American Psychiatric Association 1987) criteria for MDD(2) Recurrent MDD, depression occurring across multiple at some point since the birth of their child. In order to assessment periods, will be related to greater difficulty minimize co-occurring risk factors that may accompany in emotion regulation and socioemotional functioning depression (Downey and Coyne 1990), mothers were in children. required to have at least a high school education and(3) Mothers with a history of depression will be more families could not be reliant on public assistance. negative toward their children, and maternal negativity Comparison participants were recruited by contacting will moderate the association between maternal depres- parents identified through county birth records who had similar sion and children’s emotion regulation and socioemo- socioeconomic backgrounds and who lived in the same tional outcomes. Specifically, heightened levels of geographic locations as did the depressed mothers. The control maternal negativity will potentiate the deleterious effects group mothers were screened for the presence or history of of maternal depression on children’s emotion regulation major psychiatric disorder using the Diagnostic Interview abilities and socioemotional functioning. Maternal neg- Schedule III-R (DIS-III-R; Robins et al. 1985), and only ativity will be related to greater emotion dysregulation mothers without a history of any Axis I mental disorder were and socioemotional difficulties in children. retained. Mothers in the control group who developed a(4) Dysregulated emotion patterns will be associated with psychiatric disorder over the course of the investigation were increased child self-report ratings of socioemotional excluded from the investigation (n=4). Informed consent was problems and will mediate the relation between obtained from all study participants. More depressed than maternal depression and children’s socioemotional control mothers were recruited in order to examine individual outcomes. variability in the at risk group in greater detail. Demographics Children and their mothers in the depressedMaterials and Methods and nondepressed comparison groups were comparable on a range of demographic variables at the Initial assessmentParticipants period. As shown in Table 1, offspring of depressed and nondepressed mothers did not differ in gender or age acrossOne hundred fifty-one mothers and their children from the four assessment periods. The two groups were compa-predominantly middle to upper income households were rable on maternal and paternal education, family socioeco-assessed on four separate occasions. Children were approx- nomic status (Hollingshead 1975), number of adults andimately 21 months of age (M=20.68, SD=2.40) at the Initial children in the home, and maternal race. Although theassessment period, 39 months of age (M=38.63, SD=2.73) majority of mothers in the depressed (82.8%) and nonde-at the age 3 evaluation, 51 months of age (M=51.20, SD= pressed comparison (98.3%) groups were married, a higher4.28) at the age 4 assessment, and 62 months of age (M= percentage of depressed caregivers were either separated,62.03, SD=2.72) at the age 5 evaluation period. At the divorced, or never married. Marital status was not signif-Initial assessment, 93 of the children had mothers with a icantly correlated with key study outcome variables and, ashistory of Major Depressive Disorder (MDD) that had such, was not considered further.occurred during the first 21 months following the child’sbirth, and 58 of the children had mothers without a past or Sample Retention Of the original 151 mother–child pairscurrent history of MDD or any other Axis I mental disorder. who completed the Initial assessment, 150 participated in the age 3 evaluation, 151 participated in the age 4Participant Recruitment and Study Inclusion Criteria A assessment, and 137 participated in the age 5 evaluation.community sample of mothers with a history of MDD was This represents an attrition rate of only 9% (n=14) fromrecruited through referrals from local mental health the Initial to age 5 assessment periods. Of the 14 familiesprofessionals and through advertisements in local news- that did not complete the age 5 assessment, seven were
690 J Abnorm Child Psychol (2007) 35:685–703Table 1 Comparison of initial depressed and nondepressed groups on general demographic variablesVariables Initial depressed Nondepressed comparisons M SD M SD TChild age (months): Time 1 20.46 (2.24) 21.05 (2.62) 1.49 Time 2 38.72 (3.26) 38.48 (1.60) <1 Time 3 51.26 (4.02) 51.11 (4.70) <1 Time 4 61.85 (3.23) 62.32 (1.57) <1Maternal education (years) 15.00 (1.66) 14.93 (1.71) <1Paternal education (years) 14.39 (2.33) 5.03 (1.94) 1.75Family Hollingshead (SES) 46.39 (11.92) 49.63 (10.73) 1.69Number adults in home 2.18 (0.83) 2.07 (0.32) <1Number children in home 1.86 (0.97) 2.10 (0.91) 1.54 % (n) % (n) χ2Child gender:Male 48.4 (45) 53.4 (31) <1Female 51.6 (48) 46.6 (27)Marital status:Married 82.8 (77) 98.3 (57) 8.85*Separated 6.5 (6) 0 (0.0)Divorced 3.2 (3) 0 (0.0)Never married 7.5 (7) 1.7 (1)Maternal race:Minority 6.5 (6) 5.2 (3) <1Non-minority 93.5 (87) 94.8 (55)SES socioeconomic status, *p<0.05.from the depressed group and seven were from the experienced a major depressive episode between the ages 4nondepressed comparison group. Level of attrition did and 5 evaluations).not differ significantly by group membership, χ2(1)=<1, For approximately 23% of the depressed mothers, the majorn.s. In addition, no significant group differences were found depressive episode that occurred during the child’s firstbetween families that did and did not complete the age 5 21 months of life was an initial onset episode and for theevaluation across important demographic characteristics, remaining depressed caregivers (77%) it was a recurrentincluding marital status [χ2(1)=1.96, n.s.], maternal race episode. The recurrence of maternal depression from the Initial[χ2(1)=1.91, n.s.], child age [t (149)=1.16, n.s.], maternal assessment to the age 5 evaluation also was examined. In theeducation [t (149)<1, n.s.], paternal education [t (149)<1, depressed group, 21.5% of the mothers were depressed at twon.s.], and family socioeconomic status [t (149)<1, n.s.]. time points, 11.8% were diagnosed with depression at three time points, and 7.5% were depressed at all four time points.Maternal Depression Characteristics As mentioned previ- Caregivers meeting diagnostic criteria for MDD at multipleously, all of the mothers in the depressed group at the Initial assessment periods were considered “recurrent” and mothersassessment period (n=93) met diagnostic criteria for major meeting diagnostic criteria for MDD at only the Initialdepressive disorder (MDD) at some point during their evaluation were considered “non-recurrent.” In the depressedchild’s first 21 months of life. At the age 3 assessment group, 59.1% (n=55) of the mothers had a non-recurrentperiod, 26.9% (n=25) of the mothers in the depressed history of depression and 40.9% (n=38) had a history ofgroup continued to meet diagnostic criteria for MDD (i.e., recurrent depression.had experienced a major depressive episode between theInitial and age 3 evaluations). At the age 4 evaluation, 18%(n=16) of the caregivers who were depressed at the Initial Procedureassessment period continued to meet criteria for MDD (i.e.,had experienced a major depressive episode between the General Overview In the larger ongoing longitudinal studyages 3 and 4 assessments). Finally, 26.7% (n=23) of the (Cicchetti et al. 1998), additional procedures and parent andmothers in the Initial depression group meet diagnostic self-report measures were obtained at each assessment.criteria for MDD at the age 5 assessment period (i.e., had However, only those relevant to current study hypotheses
J Abnorm Child Psychol (2007) 35:685–703 691are presented herein. At the Initial evaluation, mothers were consistent order across participants, and dyads were givenadministered the DIS-III-R and a demographic interview in 5-min to attempt to solve each task. The tasks in order oftheir homes by trained research assistants who were their presentation were a shape sorter game, a spoolunaware of study hypotheses and recruitment status. stringing task, a beanbag clown toss, a ring toss, an Etch-Families meeting research criteria were retained in the A-Sketch maze task, and a circus puzzle. Mothers weresample. The DIS-III-R and a follow-up demographic instructed to allow their children to solve the tasks on theirinterview were administered again to mother participants own and to give their children whatever help they saw fit.at the ages 3, 4, and 5 assessment periods. In addition, at The tasks are designed to tax the child’s problem-solvingage 3, mother–child dyads participated in six problem- skills and elicit a response from caregivers. The proceduresolving tasks from which maternal insensitivity ratings was conducted in a laboratory setting, and mother–childwere obtained. At the age 4 assessment period, mother– dyads were videotaped through a one-way mirror forchild pairs participated in an experimental paradigm subsequent coding.involving simulated interadult anger from which children’semotion regulation patterns (EMRPs) were assessed. Anger Simulation This paradigm was adapted from Cum-Finally, at age 5, a self-report measure of children’s mings and colleagues’ normative developmental researchsocioemotional functioning was obtained. on marital conflict (Cummings 1987; Cummings et al. 1989). The procedure is designed to observe children’s emotional behavioral responses and regulation strategies inMeasures response to a stressful event (i.e., interadult anger). The observational setting used was a playroom with age-Diagnostic Interview Schedule III-R (DIS-III-R) The DIS- appropriate toys and a chair for the mother. Before theIII-R (Robins et al. 1985) was developed to evaluate the procedure began, mothers were briefed about its details anddiagnostic criteria specified in the Diagnostic and Statistical their role in the simulation. They were told to fill outManual of Mental Disorders, III-R version (DSM-III-R; “pretend” paperwork during the simulation and not toAmerican Psychiatric Association 1987). The DIS-III-R initiate play or talk with their child.was used in the present study due to the longitudinal nature During the 15-min procedure, children were exposed toof the investigation. Despite the availability of more three scripted simulations of 1–1 1/2 min affectiverecently revised versions of the DIS, these were not in interactions between their mothers and a research assistant.press when the study began. No differences in diagnostic These interactions in order of their presentation werecriteria for MDD exist among the DIS-III-R and more friendly, angry, and reconciliatory, separated by 3-mincurrent versions of the diagnostic interview. This interview neutral periods, with the research assistant absent from theconsists of approximately 260 items designed to be observation room. The 3-min interspersed neutral periodsadministered in a standard sequence. Questions are an- were included in the simulation to provide an opportunityswered on a yes/no basis. Because of the structured format to observe latent or inhibited emotional behavioralof the interview, sensitive clinical judgments are not responses not otherwise exhibited in the presence of therequired and, therefore, the interview can be administered research assistant. All research assistants were females whoby lay interviewers. The DIS assesses the presence of Axis were unaware of depression status and study hypotheses.I adult psychiatric disorders and allows for the assignment Participants were exposed to similar levels of emotionalof 49 DSM-III-R diagnoses. All interviewers were trained intensity and all verbal exchanges were directed toward theto criterion reliability in the administration of the DIS, and mother and not the child andcomputer generated diagnoses were utilized. The DIS has Mothers were told to modulate their facial affect to fitundergone development and reliability and validity studies each simulation segment and not to argue back or to laugh,for use in psychiatric epidemiological field studies (Robins and only to provide responses outlined within the script.et al. 1981; Robins et al. 1982). The monitoring of maternal affect and behavior was conducted during the procedure in order to enhance theDyadic Problem-solving Tasks At age 3, mothers and their natural quality of the interadult exchange, as well as to38-month-old children completed a series of six problem- ensure consistency in response patterns across mothersolving tasks. Maternal negativity was assessed while their participants. All mother participants complied with simula-child attempted to solve the task. Tasks and maternal tion instructions. At the completion of each anger simula-coding procedures were adapted from Lewis and col- tion, the child was debriefed regarding the pretend nature ofleagues’ secondary emotion research (Alessandri and Lewis the exchange. The entire anger simulation sequence was1996; Lewis et al. 1992). The problem-solving tasks were videotaped through a one-way mirror for subsequent codingindividually presented to each mother–child dyad in a and analysis.
692 J Abnorm Child Psychol (2007) 35:685–703Perceived Competence and Social Acceptance for Young from videotaped recordings using Cummings’ (1987)Children This self-report measure (Harter and Robin 1984) person-centered classification system. Consistent with pastis a downward extension of the Perceived Competence research (Cummings 1987; El-Sheikh et al. 1989; MaughanScale for Children (Harter 1982) and is designed to assess and Cicchetti 2002), coding of children’s EMRPs followedperceived competence and social acceptance in preschool- a multi-step procedure. First, using a coding schemeaged children and older using a pictorial format. The adapted from Cummings and colleagues’ prior workmeasure assesses children’s perceptions of their competence (Davies 1998), 16 discrete emotional behavioral responsesin four domains, including cognitive competence (e.g., were coded as either present or absent during 30-s intervalsgood at puzzles, knows alphabet), physical competence (e.g., across the simulation’s six event sequences. Discretecan tie shoes, good at hopping), peer acceptance (e.g., has behaviors included sadness/crying, whining, anxiety/freez-lots of friends, gets asked to play with others), and maternal ing, anger, physical aggression, verbal aggression, object-acceptance (e.g., Mom reads to you, Mom cooks your related aggression, dysregulated aggression/lose of control,favorite foods). The cognitive competence and physical smiling/laughing, preoccupation, verbal concern, requestscompetence domains make up the perceived competence to leave, shutting out, getting comfort from mother,scale and the peer acceptance and maternal acceptance helping/instructing mother, and comforting/protectingdomains make up the perceived social acceptance scale. mother.Child responses were scored on a 4-point scale, 1 indicating Using children’s coded discrete emotion responses, threelow self-perceptions of competence/acceptance and 4 indi- different aspects of emotion modulation were then codedcating high self-perceptions of competence/acceptance. during the anger and reconciliation periods of the simula-Previous research has demonstrated that this measure is tion. These included latency to onset of emotional reactiv-reliable and has good discriminate, predictive, and conver- ity, rise time to peak emotional arousal, and verbalgent validity (Brody et al. 2004; Harter and Robin 1984). In regulation of emotion (Frodi and Thompson 1985). Latencythe present study, alpha coefficients for the perceived to onset of emotional reactivity was operationally definedcompetence and social acceptance scales were 0.68 and as the number of 30-s intervals between anger exposure and0.73, respectively. the onset of expressed emotion. Rise time to peak arousal was operationally defined as the number of 30-s intervals between the child’s initial exposure to interadult anger andCoding his/her highest frequency or peak level of discrete emo- tional reactivity. Verbal regulation of emotion was opera-Maternal Negativity Maternal negativity ratings were based tionally defined as the frequency with which the childon the frequency of negative maternal verbalizations and talked about the witnessed angry exchange with his/heraffective and behavioral expressions toward the child caregiver. Children’s verbal concern discrete code duringduring the six problem-solving tasks (Alessandri and Lewis the anger and reconciliation periods was used as an1996). Maternal negativity ratings included any facial indicator of the child’s verbal processing of the affectiveexpression (e.g., a frown and/or angry affect), vocalization experience.(e.g., “No!”; “Can’t you do anything right!”), and/or bodily Finally, tapes were viewed a second time to capture aexpression (e.g., shaking head and looking away) of sense of children’s regulatory profiles by assessing chil-disappointment, rejection, and/or hostility directed toward dren’s global emotional reactivity and self-regulatorythe child and/or his/her performance. Coders viewed abilities before, during, and after anger exposure. Withvideotapes and recorded the frequency of negative maternal children’s coded discrete emotional behavioral responseverbalizations and affective and behavioral expressions profiles across the six event sequences (friendly exchangeduring each 5-min problem-solving task procedure. Mater- through post-reconciliation neutral period) and emotionnal negativity frequency scores were summed across the six modulation scores in hand, coders classified children asproblem-solving tasks. This summed score was used in exhibiting either an adaptively regulated, undercontrolled,subsequent analyses. or overcontrolled EMRP based on answers to the following Coders were two trained bachelor level research assistants overall profile assessments: (1) Does the child exhibitwho were unaware of maternal depression status and study generally low, moderate, or high levels of emotionalhypotheses. A reliability analysis was conducted on 20% of behavioral reactivity in response to the witnessed ex-the sample and adequate reliability was attained. The intra- change?; (2) Does the child exhibit any overt displays ofclass correlation for maternal negativity total scores was 0.87. dysregulated emotion (e.g., presence of dysregulated aggression/loss of control, responses are aimless and/orEmotion Regulation Patterns (EMRPs) Children’s EMRPs disorganized)?; (3) What is the duration of the child’sin response to the simulated anger procedure were coded emotional responses across the simulation procedure (e.g.,
J Abnorm Child Psychol (2007) 35:685–703 693do responses subside or continue after witnessed reconcil- and child socioemotional adjustment at age 5. Next, theiation)?; (4) Are emotional behavioral responses congruent relations among maternal negativity, child EMRPs, andwith procedural demands (e.g., elevated reactivity during child socioemotional outcomes were evaluated. Finally,anger exposure and amelioration of responses during paths from early-occurring Initial maternal depression (i.e.,conflict resolution)?; and (5) Is the child able to effectively depression that has occurred at some time during the firstmodulate the intensity and duration of his/her emotional 21 months of the child’s life) to children’s dysregulatedbehavioral responses (e.g., latency period between anger EMRPs at age 4 and socioemotional problems at age 5 wereexposure and emotion expression, rise time to peak arousal, evaluated. Gender of child was not included as a predictorand ability to verbally process witnessed event with mother)?. because it was correlated with only one of five key outcomeTable 2 displays example responses to the above overall variables (i.e., age 5 perceived competence ratings, r=0.18,profile assessments for the three EMRP classifications. p<0.05). The effect of child gender on perceived compe- Overall, 58 children were classified with adaptively tence scores was statistically controlled by generating anregulated EMRPs, 83 were designated with undercontrolled unstandardized residualized variable utilizing linear regres-EMRPs, and ten were classified with overcontrolled sion techniques (Cohen et al. 2002), whereby perceivedEMRPs. Detailed profile descriptions of the three EMRPs competence scores were regressed on child gender. Thisare provided in the results section. A reliability analysis of unstandardized residualized perceived competence variablediscrete emotional reactivity and emotion modulation codes was used in all subsequent analyses.and overall EMRP ratings were conducted on 25% of thesample (i.e., 38 tapes). Coders were trained graduate level Emotional Response Profiles of Children’s EMRPsresearch assistants who were unaware of maternal depres-sion status and study hypotheses. All interrater disagree- Discrete emotional reactivity and emotion modulationments were resolved by discussion. Intraclass correlation codes from the simulated anger procedure were examinedcoefficients for the 16 discrete emotion codes ranged from to assess the emotional response profiles of children’s0.75 to 1.00 and coefficients for each of the three emotion person-centered EMRPs. First, the 16 discrete emotionalmodulation codes were 1.00. Kappas for the three EMRPs reactivity codes were combined using factor analyticwere as follows: adaptively regulated, κ=0.78; under- procedures to create composite indexes of children’scontrolled, κ=0.88; and overcontrolled, κ=0.74. emotional responses due to low frequency rates of individual discrete codes. Total frequency scores were calculated for each discrete emotion code by summing frequency counts across two of the anger simulationResults segments (angry and reconciliation periods). Principal components analysis extraction method was used and aOverview of Data Analytic Strategy varimax rotation was employed. Five factors with eigen- values greater than 1 emerged and a loading cutoff of 0.3 orEmotional response profiles of children’s person-centered greater was used to determine factor structure (PedhazurEMRPs are first presented. Then, the proposed hypotheses and Schmelkin 1991). After rotation, the percentage of totalwere evaluated in several stages. First, correlations among variance accounted for by the five factors was 66.4%. Thekey study variables are presented, followed by a more discrete emotion codes that adequately loaded on the anger/detailed examination of the effects of maternal depression hostility factor included anger, physical aggression, verbalon maternal negativity at age 3, children’s EMRPs at age 4, aggression, object-related aggression, and dysregulatedTable 2 Example responses to overall profile assessments for the three EMRP classifications Adaptively Regulated EMRP Undercontrolled EMRP Overcontrolled EMRPOverall Profile Assessments1. Level of emotional behavioral reactivity Moderate High Low2. Overt displays of dysregulated emotion No Yes No3. Duration of emotional behavioral responses Moderate Extended Brief4. Responses congruent with procedural demands Yes No No5. Effectively modulate emotion responses Yes No NoEMRPs emotion regulation patterns.
694 J Abnorm Child Psychol (2007) 35:685–703aggression/lose of control. The active concern factor tivity composites and three emotion modulation codes (i.e.,included verbal concern, helping/instructing mother, and latency to onset, rise time to peak arousal, and verbalcomforting/protecting mother discrete codes. Preoccupa- regulation of emotion) (See Table 3). As shown in Table 3,tion, anxiety/freezing, and get comfort from mother codes children classified with adaptively regulated EMRPs dis-all adequately loaded on the fear/support seeking factor. played moderate frequencies of emotional reactivity thatThe discrete emotion codes that loaded on the high distress emerged and peaked in a timely manner and whosefactor included whining and sadness/crying. Finally, the processing of the emotional event with caregivers appearedavoidance factor included shutting out and requests to leave to be effective in helping the child to modulate his/herresponses. The discrete emotion codes that comprised each arousal and resume baseline arousal levels. In contrast,of the five factors were then summed to create five youngsters classified with undercontrolled EMRPscomposite indices of children’s emotional reactivity in exhibited elevated frequencies of emotional reactivity inresponse to the friendly, angry, and reconciliation periods response to witnessed anger. These children’s heightenedof the simulation. Comparison of mean frequencies of child reactivity often extended into the reconciliation period,emotional reactivity composites during the three simulation even after witnessing the conflict resolution between theperiods revealed that the manipulation was successful in mother and the confederate. Although undercontrolledeliciting emotional behavioral reactions in response to children attempted to verbally process the emotional eventinteradult anger exposure (e.g., higher emotional reactivity with caregivers, this strategy did not appear to be effectivescores during the angry than the friendly simulation period). in modulating their affective arousal, as evidenced by Then, general linear model (GLM) multivariate proce- elevated emotional reactivity and verbal regulation ofdures followed by analysis of variance (ANOVAs) and a emotion scores in both the angry and reconciliation periods.priori contrasts were performed to examine overall and Finally, children classified with overcontrolled EMRPsindividual EMRP differences on the five emotional reac- displayed low rates of emotional reactivity and appearedTable 3 Mean frequencies of emotional response profiles of children’s person-centered emotion regulation patterns (EMRPs) across the angerand reconciliation simulation periods EMRPsEmotional Adaptively regulated Under-controlled Over-controlledResponse parameters (n=58) (n=83) (n=10) F(2,148) Planned contrastsEmotional reactivity compositesa Anger/hostility Angry periodc 0.19 (0.52) 0.83 (1.01) 0.10 (0.32) 11.88*** U>O, U>AR Reconciliation periodc 0.18 (0.56) 0.67 (1.08) 0.00 (0.00) 6.71** U>O, U>AR Fear/support seeking Angry period 5.93 (3.70) 8.59 (4.87) 2.70 (2.00) 12.22*** U>O, U>AR, AR>O Reconciliation period 2.55 (1.93) 6.62 (4.98) 1.60 (1.90) 21.69*** U>O, U>AR High distress Angry periodc 0.24 (0.52) 0.68 (0.94) 0.00 (0.00) 7.44** U>O, U>AR Reconciliation period 0.17 (0.83) 0.79 (1.32) 0.20 (0.63) 5.73** U>AR Active concern Angry period 2.12 (2.16) 3.16 (2.77) 0.00 (0.00) 8.81*** U>O, U>AR, AR>O Reconciliation period 0.49 (1.32) 0.92 (1.48) 0.00 (0.00) 3.00+ U>O, U>AR Avoidance Angry periodc 0.27 (0.50) 0.33 (0.57) 0.25 (0.73) .22 Reconciliation periodc 0.14 (0.47) 0.33 (0.61) 0.39 (0.63) 2.36+ U>ARLatency to onsetb c 0.13 (0.37) 0.14 (0.44) 1.21 (1.58) 17.37*** U<O, AR<ORise time to peak arousalb 0.70 (1.16) 1.95 (3.01) 4.20 (5.41) 8.67*** U<O, AR<U, AR<OVerbal regulation of emotiona Angry period 1.67 (1.37) 2.35 (1.90) 0.00 (0.00) 10.26*** U>O, U>AR, AR>O Reconciliation period 0.19 (0.48) 0.69 (0.97) 0.00 (0.00) 8.65*** U>O, U>ARAR adaptively regulated EMRPs, U undercontrolled EMRPs, O overcontrolled EMRPs.a Values represent mean frequencies (SDs).b Values represent mean number of 30-second intervals (SDs).c Square root transformed variables due to non-normal univariate distributions.+ p<0.10, *p<0.05, **p<0.01, ***p<0.001.
J Abnorm Child Psychol (2007) 35:685–703 695to restrict emotion expression in response to the angry distribution of children of mothers with a history ofexchange. These youngsters also minimized the importance depression at the Initial assessment period and children ofof the mother in helping to modulate emotional arousal, as nondepressed mothers across the two EMRP groupsindicated by their failure to verbally process the stressful (adaptively regulated and dysregulated) was examined. Aevent with caregivers. Unlike their adaptively regulated and significantly higher percentage of dysregulated EMRPs wasundercontrolled peers, overcontrolled children did not found in children of mothers with early-occurring Initialexhibit other-directed regulatory strategies and tended to depression (73.1%) as compared to children of nondepressedturn inward in response to the stressful event. comparisons (43.1%), χ2(1)=13.58, p<0.001, rΦ =0.30. The emotional response profiles of each of the three Next, the effect of recurrent maternal depression (i.e.,EMRPs were consistent with those reported in previous depression that occurred across multiple assessment peri-investigations (Cummings 1987; El-Sheikh et al. 1989; ods) on children’s EMRPs was then examined. Children inMaughan and Cicchetti 2002). Due to the low frequency of the recurrent versus non-recurrent maternal depressionchildren displaying overcontrolled EMRPs (n=10), under- groups and youngsters in the nondepressed comparisoncontrolled and overcontrolled youngsters were combined group were compared. 74.2% of the recurrent group andinto a dysregulated EMRP group (similar results were 72.6% of the non-recurrent group had dysregulatedfound with or without combining the two groups). Based EMRPs, contrasting with 43.1% of the nondepressed group.on this grouping, 93 (61.6%) of the children in the total Thus no differential effect related to recurrent depressionsample displayed dysregulated EMRPs and 58 (38.4 %) was observed [recurrent versus non-recurrent maternalexhibited adaptively regulated patterns. This adaptively depression, χ2(1)<1, p=0.87, rΦ =0.01].regulated/dysregulated EMRP designation was used in allsubsequent analyses. Maternal Depression and Children’s Later Social Self-PerceptionsCorrelations among Study Variables The effect of maternal depression at the Initial assessmentCorrelations among key study variables at the Initial, ages period on children’s social self-perceptions at age 5 was3, 4, and 5 assessment periods are summarized in Table 4. examined. t tests were conducted and, as hypothesized, children in the Initial maternal depression group had lowerMaternal Depression and Children’s EMRPs self-report ratings of perceived competence at age 5 (M= −0.06, SD=0.43) than did children in the nondepressedAs shown in Table 4, Initial maternal depression was comparison group (M=0.11, SD=0.32), t (135)=2.47, p<significantly correlated with children’s age 4 dysregulated 0.05, d=0.44. Initial maternal depression, however, wasEMRPs. Maternal depression at ages 3 and 4, however, was unrelated to child social acceptance ratings, t (135)<1, n.s.,unrelated to children’s dysregulated emotion patterns. To d=0.01. Also, ages 3, 4, and 5 maternal depression wereassess the relation between early-occurring maternal de- not significantly correlated with either of the age 5 childpression and dysregulated EMRPs in more detail, the social self-perception ratings.Table 4 Correlation matrix of study variablesVariables 1 2 3 4 5 6 7 8Initial 1. Maternal Depression –Age 3 2. Maternal Depression 0.35*** – 3. Maternal Negativity −0.08 −0.01 –Age 4 4. Maternal Depression 0.18* 0.36*** −0.10 5. Dysregulated EMRPs 0.30*** 0.13 0.06 −0.03 –Age 5 6. Maternal Depression 0.29** 0.40*** −0.07 0.39*** 0.01 – 7. Perceived Competencea −0.21* −0.08 0.11 0.00 −0.21* −0.03 – 8. Social Acceptance 0.00 −0.03 0.16+ 0.14 −0.18* −0.02 0.55*** –EMRPs emotion regulation patterns.a Unstandardized residualized variable that controls for gender.+ p<0.10, *p<0.05, **p<0.01,***p<0.001.
696 J Abnorm Child Psychol (2007) 35:685–703 Next, one-way ANOVAs with three groups (recurrent First, hierarchical logistic and linear regressions weredepression, nonrecurrent depression, and comparisons) performed to examine the moderating role of age 3 maternalwere performed to determine the extent to which recurrent negativity in the association between early-occurringmaternal depression was related to children’s socioemo- Initial maternal depression and children’s dysregulatedtional outcomes at age 5. Findings revealed that the EMRPs at age 4 and socioemotional adjustment at age 5.recurrent nature of the mother’s depression over time was Given that the main effects of the predictors on EMRPsnot associated with age 5 child self-report ratings of and socioemotional outcomes have already been dis-perceived competence or social acceptance (F (2, 134)=3.04, cussed, only the results of the interaction terms aren.s.; F (2, 134)<1, n.s., respectively). reported. In each analysis, the outcome variable was Findings suggest that the effects of early-occurring Time regressed onto the predictor and moderator variables at1 maternal depression on children’s later perceived compe- step 1 and then onto the interaction term at step 2.tence do not cumulate over time, such that depression Findings revealed that the interaction of early-occurringbeyond 21 months does not provide a significant increment Initial maternal depression and age 3 maternal negativityin the prediction of 5-year-old perceived competence scores. in predicting children’s dysregulated emotion patterns was marginally significant [Â=0.08, Wald statistic F (1,Maternal Negativity and its Relation to Maternal 148)=3.44, p<0.07, Cohen’s d=0.15], contributing onlyDepression and Children’s EMRPs and Socioemotional 2% of the variance to R2 (ΔR2 =0.02).Outcomes With respect to children’s socioemotional outcomes, the interaction effect of early-occurring Initial maternal depres-Findings indicated that Initial and age 3 maternal depres- sion and age 3 maternal negativity on child self-reportsion were not associated with the negativity caregivers ratings of social acceptance at age 5 was significant [β=provided their children at age 3 (see Table 4). Additionally, −0.47, p<0.05, Cohen’s d=0.38] and uniquely accountedage 3 maternal negativity was unrelated to children’s age 4 for an additional 3% of the variance, ΔR2 =0.03. To clarifydysregulated emotion patterns (r=0.06, n.s.), perceived this significant interaction effect, plots of the regressioncompetence scores at age 5 (r=0.11, n.s.) and only slopes at high (+1 SD) and low (−1 SD) levels of maternalmarginally associated with age 5 child social acceptance negativity for children in the Initial depressed and nonde-ratings (r=0.16, p<0.10). pressed groups were created (see Fig. 1). The significant interaction effect indicated that the degree of the relation-EMRPs and Children’s Later Socioemotional Functioning ship between maternal negativity and children’s social acceptance ratings is dependent upon whether the childThe degree to which children’s EMRPs at age 4 were had a mother with or without early-occurring depression.related to their self-report ratings of socioemotional As illustrated in Fig. 1, the regression lines determinedadjustment 1 year later was then assessed. T tests revealed separately for high and low levels of maternal negativitythat children with adaptively regulated EMRPs at age 4 have significantly different slopes, revealing that the effectsreported higher ratings of both perceived competence (M= of maternal negativity at age 3 on children’s self-report0.11, SD=0.36) and social acceptance (M=3.09, SD=0.52) ratings of social acceptance 2 years later are greater forat age 5 than did youngsters classified with dysregulated C hild S ocial-A cceptance R atingsemotion patterns (M=−0.06, SD=0.40; M=2.89, SD=0.53, 3.5respectively), [t (135)=2.48, p<0.05, d=0.44; t (135)=2.06, 3p<.05, d=0.37, respectively]. 2.5Developmental Pathways to Children’s Dysregulated 2EMRPs and Socioemotional Difficulties 1.5Next, mediating and moderating processes of the relation 1between maternal depression and children’s emotion regu- 0.5lation and socioemotional difficulties were investigated. 0Given that maternal depression that occurred after the Depressed NondepressedInitial assessment period and recurrent episodes of depres-sion over time did not predict children’s EMRPs at age 4 or Low Maternal Negativity (-1SD) High Maternal Negativity (+SD)their socioemotional adjustment at age 5, maternal depres- Fig. 1 Significant moderating effect of age 3 maternal negativity onsion at the Initial evaluation (i.e., early-occurring maternal the relation between initial maternal depression and child self-reportdepression) served as the depression predictor variable. ratings of social acceptance at age 5
J Abnorm Child Psychol (2007) 35:685–703 697early-occurring Initial depressed offspring than youngsters with outcome variables. To test the third and final conditionof nondepressed comparisons. Maternal negativity, however, of mediation (i.e., when predictor and mediator variablesfailed to moderate the association between Initial maternal are evaluated together, the unique effect of the predictordepression and children’s perceived competence scores at variable on the outcome variable is reduced or eliminated,age 5 [β=0.07, p=0.74, Cohen’s d=0.06]. while the significant effect if the mediator remains), SEM Next, the mediational role of dysregulated EMRPs in the techniques using Amos (Analysis of Moment Structures)association between early-occurring Initial maternal depres- 4.0 software were utilized (Arbuckle and Wothke 1999).sion and age 5 socioemotional outcomes was examined. The maximum likelihood (ML) method was used toGuided by the procedures outlined by Baron and Kenny estimate model parameters.(1986), study findings thus far have met two out of the To determine mediation, model building techniques (Klinethree necessary conditions of mediation: (1) predictor 1998) were used to compare two hierarchically nested pathvariable (i.e., Initial maternal depression) must account for models, the direct effects model and the mediation model.significant variance in mediator (i.e., dysregulated EMRPs) The chi-square difference statistic (χ2 difference) was usedand outcome (i.e., child socioemotional problems) variables to assess the significance of the improvement in fit with theand (2) mediator variable must be significantly associated addition of the mediation path(s). To further assess whetherFig. 2 Direct effects modelwith Initial maternal depression Initial Age 4 Age 5as a predictor of dysregulatedEMRPs at age 4 and perceived (a) Direct Effects Modelcompetence and social accep- Perceivedtance scores at age 5 (a). Medi- Competenceaational/indirect model with (R2 = .05)initial depression indirectly pre- -.16 (-.13)+dicting child social acceptance -.17 (-.13)+ratings through its association Maternal Dysregulatedwith dysregulated EMRPs (b) Depression EMRPs -.19 (-.21)* .06 (.06) Social Acceptance (R2 = .04) χ 2 (3) = 62.83*** NFI = .92 CFI = .93 RMSEA = .09 (b) Mediation/Indirect Model Perceived Competencea (R2 = .07) -.16 (-.13)+ -.16 (-.13)+ Maternal Dysregulated Depression .30 (.30)* EMRPs (R2 = .09) .06 (.06) -.19 (-.21)* Social Acceptance (R2 = .03) χ 2 (1) = 48.67*** NFI = .94 CFI = .94 RMSEA = .08 Note. EMRPs = emotion regulation patterns; Standardized coefficients are outside brackets and unstandardized coefficients are inside brackets; Significance levels were determined by the critical ratios of the unstandardized coefficients; + p < .10. * p < .05; *** p < .001; a = Unstandardized residualized variable that controls for gender
698 J Abnorm Child Psychol (2007) 35:685–703or not mediation is present, the Sobel test (Sobel 1982) was (Cummings 1987; Davies and Forman 2002; Zahn-Waxlerperformed to assess whether the indirect effect of the et al. 1990a, b). Findings suggest that infancy and earlypredictor variable on the outcome variable via the mediator toddlerhood are particularly vulnerable periods of develop-was significantly different from zero. Given that the effects ment for the long term effects of maternal depression onof the predictors in the direct effects models have been children’s capacity to regulate emotion in the preschooldiscussed in prior sections of this report, only the results of years.the mediational models and model comparisons are reported Contrary to study hypotheses, neither depressive episodesfor the sake of parsimony. beyond the 21 month period, nor the recurrence of the mother’s The mediating role of dysregulated EMRPs in the depression, contributed to predicting children’s emotionrelation between early-occurring Initial maternal depression regulation abilities at age 4 years. These findings were notand children’s age 5 self-reported ratings of perceived entirely surprising given that the literature addressing thecompetence was examined. Children’s ratings of social association between recurrent maternal depression and childacceptance also were included in the model to examine the affective responses has been somewhat inconsistent (Campbellindirect effects of maternal depression on social acceptance et al. 1995; Moore et al. 2001). Variability across inves-through its association with children’s dysregulated tigations in how recurrent depression is operationalized, theEMRPs. Dysregulated EMRPs could not serve as a timing of the mother’s depression over the course of themediator because maternal depression failed to predict child’s development, and the severity of symptoms assessedsocial acceptance in prior analyses. The direct effects model (clinical versus subclinical levels of depression) may contrib-is displayed in Fig. 2a. The mediation/indirect model was ute to the lack of uniformity in findings. Additionally, studiesthen evaluated, whereby the path from Initial maternal that examine recurrent or chronic depression, including thedepression to children’s dysregulated EMRPs was free to present investigation, often fail to discern whether thevary. Findings revealed that the mediation/indirect model mother’s depression was persistent and uninterrupted byyielded a significantly better fit to the data than did the periods of healthy adjustment. Interruptions in the course ofdirect effects model, χ2 difference = 14.16, p < 0.001. the mother’s illness may provide greater opportunity forResults supported the indirect effect of early-occurring positive parent–child interactions and thereby promoteInitial maternal depression on children’s ratings of social healthy child adjustment. Consequently, it is not yet possibleacceptance at age 5 through its association with age 4 to draw a clear conclusion regarding the effects of chronicdysregulated emotion patterns. The Sobel test was per- depression in caregivers on children’s emotion regulationformed to determine the significance of the dysregulated development.EMRP mediation effect and a marginally significant effect Also in accordance with the extant maternal depressionemerged (critical ratio=−1.62, p=0.096). literature (Campbell et al. 2004; Cicchetti and Toth 1995), child socioemotional functioning was related to maternal depression history. Specifically, the children of mothersDiscussion with early-occurring depression reported lower ratings of perceived competence at 5 years of age than did theThe findings presented in this investigation increase the youngsters of healthy caregivers. Consistent with EMRPunderstanding of the emotional sequelae of being reared by a findings, major depressive episodes that occurred duringdepressed mother, explicate the associations between specific the late toddler and early preschool periods (i.e., at the agescorrelates of maternal depression (i.e., maternal negativity) and 3, 4, and 5 assessments) and the recurrence of the mother’schild outcomes, and provide insight into the processes through depression over time did not predict child social self-which maternal depression confers risk on offspring. Each of perceptions at age 5 years.the study’s main findings are examined in turn. Findings, thus far, demonstrate the salience of early- As predicted, early-occurring maternal depression (i.e., occurring maternal depression during the infancy and earlydepression that occurred during the first 21 months of the toddler periods, as compared to the recurrence of depres-child’s life) was related to emotion regulation development sion later on in development, on children’s emotionat 4 years of age. Specifically, results indicated that regulation abilities and socioemotional functioning in theapproximately 73% of the children of mothers with early- preschool years. Prior investigations have found that earlyoccurring depression exhibited dysregulated emotion patterns episodes of maternal depression have differential effects onin response to witnessed anger compared with only 43% of the child outcomes than subsequent depressive episodes and thatyoungsters of nondepressed caregivers. The percentages of the risk to offspring of depressed caregivers persists beyonddysregulated emotion patterns identified in the depressed and the acute stages of the mother’s early depression. Dawson etnondepressed comparison groups were comparable to percen- al. (2003), for instance, found that the number of monthstages reported in other middle class, low risk samples mothers were depressed from birth to age 2 was a stronger
J Abnorm Child Psychol (2007) 35:685–703 699predictor of children’s socioemotional functioning at age and negative affective exchanges are likely to affect31/2 years than depression occurring after 2 years of age. developing neural substrates and circuitry that mediate In explaining why early exposure to maternal depression self-regulatory behaviors (Dawson et al. 1994).can affect the development of emotion regulation capacities Lastly, it is important to keep in mind that depressedand place children at heightened risk for future socioemo- caregivers may pass onto their offspring an increased genetictional difficulties, several related developmental processes risk for depression and associated regulatory deficits. Asare possible. First, infancy is marked by rapid changes in such, increased genetic loading or heritability for psychopa-cognitive, emotional, and neurobiological development and thology may account for the emotion regulation and socio-primary caregivers play a central role in helping children emotional difficulties identified in the children of mothersnavigate critical developmental issues that are associated with early-occurring depression (Weissman et al. 2005).with the emergence of self-regulatory skills (e.g., caregiver- Taken together, early exposure to maternal depression isguided regulation of emotional arousal and dyadic emotion likely to precipitate a cascade of psychosocial, neurobiolog-regulation through the attachment relationship) (Sroufe ical, and genetic processes that interfere with the acquisition1996). Evidence clearly indicates that depressive symptoms of adaptive self-regulatory capacities in youngsters andin mothers exert a negative effect on emotion regulation increase children’s risk for future emotion regulation andchallenges of infancy by interfering with the mother’s ability socioemotional difficulties. The presence of several possibleto provide supportive, responsive, and consistent care that processes through which early-occurring maternal depres-scaffolds the development of competent early regulatory sion negatively impacts children’s emotional developmentprocesses. From an emotional security hypothesis perspec- is consistent with general systems theory’s principle oftive, these early parenting difficulties (as well as possible equifinality (Cicchetti and Rogosch 1996; von Bertalanffyconcomitant discordant interparental relations) likely under- 1968), which proposes that a diversity of pathways can leadmine children’s emotional security by threatening their to the same outcome. It is likely that multiple mechanismsfeelings of protection and safety within the context of the of risk are involved in the transmission of emotion regu-parent–child relationship and overall family milieu. In- lation and socioemotional deficits from depressed motherscreased insecurity in youngsters then often results in to their children and that these mechanisms interact in di-heightened distressed states and emotion regulation difficul- verse and complex ways to influence the early emotional ad-ties. Regardless of whether the mother’s depression later justment of the children of depressed caregivers (Goodmanremits, these early experiences and their effects on the child’s and Gotlib 1999).developing emotion system are carried forward over the Study findings also supported the prediction thatcourse of development and incorporated into new patterns of dysregulated emotion patterns in children represent aadaptation. From a developmental psychopathology and developmental liability, making the adaptation to laterorganizational perspective, maladaptive functioning on stage-salient issues more challenging (e.g., emerging self-earlier issues of emotion regulation in infancy increases the concept and peer relations). Specifically, children identifiedprobability of continued incompetence in modulating with dysregulated emotion patterns at 4 years of ageheightened affective states in later years (Cicchetti 1990). reported lower ratings of perceived competence and social A second explanation for the salience of early-occurring acceptance 1 year later than did children with adaptivelymaternal depression on children’s future emotion regulation regulated patterns. The relation between emotion regulationand socioemotional functioning comes from growing evi- deficits and deleterious socioemotional outcomes has beendence in the neurobiological literature regarding the biolog- well documented in the literature (Eisenberg et al. 1997).ical processes that accompany and influence the development Contrary to study predictions, maternal depression (bothof emotion expression and regulation (Davidson et al. 2000) early and more recent episodes) did not predict maternaland the impact of early maternal depression on these pro- negativity ratings at age 3. Despite evidence in the literaturecesses (Dawson and Ashman 2000). Accelerated changes in for the association between maternal depression andbrain structure and function occur during the first 2 years of negativity toward offspring (National Institute of Childlife (Cicchetti and Curtis 2006), making this a period of Health and Human Development (NICHD) Early Childdevelopment that is particularly vulnerable to adverse or Care Research Network 1999), the lack of a link betweenatypical infant experiences. Specifically, infancy is a period these variables has been reported by others. Dawson et al.defined by increased synaptogenesis and neuronal pruning (2003), for example, failed to find differences betweenwhere neural patterns and connections are established, and depressed and nondepressed mothers on measures ofindividual variability in neural patterning is hypothesized to maternal warmth and encouragement during mother–childoccur as a function of environmental input (Edelman 1987). tasks at age 31/2 years. Inconsistent findings in the linkTherefore, early interactions between a depressed mother between maternal depression and maternal negativity mayand her infant that are defined by noncontingent, distressing, be due to differences across studies in the severity and