25
Behavior problems in late childhood: the roles of early maternal attachment and teacher–child relationship trajectories Erin E. O’Connor a *, Brian A. Collins b and Lauren Supplee c a Department of Teaching and Learning, New York University; b Department of Teaching and Curriculum, Hunter College; c Administration for Children, Youth and Families (Received 11 February 2010; final version received 14 December 2011) The purposes of the current study were: (1) to examine the roles of early maternal attachment relationships and teacher–child relationships during childhood for externalizing and internalizing behaviors in late childhood, and (2) to investigate teacher–child relationships, as well as externalizing and internalizing behaviors in early childhood as possible mechanisms linking early maternal attachment relationships to behavior problems in late childhood. Longitudinal data from the National Institute of Child Health and Human Development Early Child Care Research Network Study of Early Child Care and Youth Development (N ¼ 1140 mothers and children) were used in this investigation. There were three main findings. First, insecure/other maternal attachment relationships in early childhood (i.e., 36 months) were associated with externalizing and internalizing behaviors in late childhood (Grade 5). Second, elevated levels of teacher–child conflict during childhood were associated with externalizing behaviors in late childhood whereas low levels of teacher–child closeness were associated with internalizing behaviors. Third, the effects of insecure/other attachment on externalizing and internalizing behaviors in late childhood were mediated through teacher–child relationships during childhood and early externalizing and internalizing behaviors. Implications for attachment theory are discussed. Keywords: maternal attachment; teacher–child relationships; behavior problems; insecure/other attachment; longitudinal Introduction Associations between maternal attachment relationships and children’s externalizing behaviors, including over activity, impulsivity, or aggression, and internalizing behaviors, including depression and social withdrawal, in early childhood are well established (for reviews, see Thompson, 2008; Weinfield, Sroufe, Egeland, & Carlson, 2008). More specifically, children with avoidant and ambivalent attach- ments, compared to those with secure attachments, show higher levels of externaliz- ing and/or internalizing behaviors whereas those with disorganized attachments appear to be at especial risk for elevated levels of externalizing and internalizing behaviors (e.g., Cassidy & Berlin, 1994; NICHD Early Child Care Research Network, 2001; O’Connor, Bureau, McCartney, & Lyons-Ruth, 2011; Teti, 1999; van IJzendoorn, Schuengel, & Bakermans-Kranenburg, 1999; Weinfield et al., 2008). *Corresponding author. Email: [email protected] Attachment & Human Development Vol. 14, No. 3, May 2012, 265–288 ISSN 1461-6734 print/ISSN 1469-2988 online Ó 2012 Taylor & Francis http://dx.doi.org/10.1080/14616734.2012.672280 http://www.tandfonline.com

Behavior problems in late childhood: the roles of early maternal attachment and teacher–child relationship trajectories

Embed Size (px)

Citation preview

Behavior problems in late childhood: the roles of early maternal

attachment and teacher–child relationship trajectories

Erin E. O’Connora*, Brian A. Collinsb and Lauren Suppleec

aDepartment of Teaching and Learning, New York University; bDepartment of Teaching andCurriculum, Hunter College; cAdministration for Children, Youth and Families

(Received 11 February 2010; final version received 14 December 2011)

The purposes of the current study were: (1) to examine the roles of early maternalattachment relationships and teacher–child relationships during childhood forexternalizing and internalizing behaviors in late childhood, and (2) to investigateteacher–child relationships, as well as externalizing and internalizing behaviors inearly childhood as possible mechanisms linking early maternal attachmentrelationships to behavior problems in late childhood. Longitudinal data from theNational Institute of Child Health and Human Development Early Child CareResearch Network Study of Early Child Care and Youth Development (N ¼ 1140mothers and children) were used in this investigation. There were three mainfindings. First, insecure/other maternal attachment relationships in early childhood(i.e., 36 months) were associated with externalizing and internalizing behaviors inlate childhood (Grade 5). Second, elevated levels of teacher–child conflict duringchildhood were associated with externalizing behaviors in late childhood whereaslow levels of teacher–child closeness were associated with internalizing behaviors.Third, the effects of insecure/other attachment on externalizing and internalizingbehaviors in late childhood were mediated through teacher–child relationshipsduring childhood and early externalizing and internalizing behaviors. Implicationsfor attachment theory are discussed.

Keywords: maternal attachment; teacher–child relationships; behavior problems;insecure/other attachment; longitudinal

Introduction

Associations between maternal attachment relationships and children’s externalizingbehaviors, including over activity, impulsivity, or aggression, and internalizingbehaviors, including depression and social withdrawal, in early childhood are wellestablished (for reviews, see Thompson, 2008; Weinfield, Sroufe, Egeland, &Carlson, 2008). More specifically, children with avoidant and ambivalent attach-ments, compared to those with secure attachments, show higher levels of externaliz-ing and/or internalizing behaviors whereas those with disorganized attachmentsappear to be at especial risk for elevated levels of externalizing and internalizingbehaviors (e.g., Cassidy & Berlin, 1994; NICHD Early Child Care ResearchNetwork, 2001; O’Connor, Bureau, McCartney, & Lyons-Ruth, 2011; Teti, 1999;van IJzendoorn, Schuengel, & Bakermans-Kranenburg, 1999; Weinfield et al., 2008).

*Corresponding author. Email: [email protected]

Attachment & Human Development

Vol. 14, No. 3, May 2012, 265–288

ISSN 1461-6734 print/ISSN 1469-2988 online

� 2012 Taylor & Francis

http://dx.doi.org/10.1080/14616734.2012.672280

http://www.tandfonline.com

At preschool and early school age, different subtypes of disorganized attachmenthave been identified, including controlling and disorganized strategies (i.e., insecure/other category in the Cassidy and Marvin system for preschoolers), which vary in thelevel of disorganization. Controlling children ‘‘actively attempt to control or directthe parent’s attention and behavior, and assume a role which is usually consideredmore appropriate for a parent with reference to a child’’ (Main & Cassidy, 1988,p. 418). Insecure/other children continue to exhibit anomalous, out-of-contextbehaviors in relation to the parent, similar to the disorganized behaviors of infancy(Main & Cassidy, 1988; Moss, Cyr, & Dubois-Comtois, 2004; Wartner, Grossman,Fremmer-Bombok, & Suess, 1994). Little research has examined variation inexternalizing and internalizing behaviors among children evidencing the controllingand insecure/other subtypes of disorganized attachment; however, a small body ofresearch shows insecure/other children at greatest risk for elevated levels ofexternalizing and internalizing behaviors (e.g., O’Connor, Bureau, et al., 2011; Teti,1999).

Our understanding of early maternal attachment relationships and their roles inchildren’s psychological development is limited, however, as relatively few empiricalstudies exist on early maternal attachment relationships and children’s externalizingand internalizing behaviors in late childhood. In particular, little research exists onthe mechanisms through which early maternal attachment relationships influencechildren’s later externalizing and internalizing behaviors. In his review of attachmentresearch, Thompson (2008) notes that there will be important further advances inour understanding of the associations between early maternal attachment relation-ships and later psychological development when research is conducted to investigatethe intervening processes that connect them. This paucity of research on mediatingmechanisms is surprising as Bowlby (1969/1982) framed attachment theory as apathway model (Weinfield et al., 2008). According to Bowlby, early insecurity is arisk factor for externalizing and internalizing behaviors in later childhood due to theinitiation in early childhood of pathways which, only when followed, increase thelikelihood of later pathological conditions (Bowlby, 1973; Weinfield et al., 2008).

Theoretical and empirical work suggests that teacher–child relationships mayprovide an important pathway connecting early maternal attachment relationshipsto children’s externalizing and internalizing behaviors in late childhood (e.g.,Goossens & van IJzendoorn, 1990; Howes & Hamilton, 1992; Ladd & Burgess, 1999;Ladd, Birch, & Buhs, 1999; O’Connor & McCartney, 2006; Sroufe, Fox, & Pancake,1983; Thijs & Koomen, 2009). No empirical research, however, has been conductedto examine children’s cumulative experiences in their relationships with teachers as apotential intervening mechanism. Thus, the purpose of the current study was toinvestigate teacher–child relationship trajectories as a mediator in the associationbetween early maternal attachment relationships and children’s externalizing andinternalizing behaviors in late childhood.

Internal working models

Most theorists hypothesize that Internal Working Models (IWMs) of the self andrelationships developed through experiences in the early maternal attachmentrelationship are responsible for associations between maternal attachment relation-ships and children’s psychological development (Bowlby, 1969/1982; Bretherton &Munholland, 2008; Moss, Bureau, Beliveau, Zdebik, & Lepine, 2009). These models

266 E.E. O’Connor et al.

act as filters through which incoming information (e.g., behavior of others) isprocessed (e.g., Thompson, 2008). Due to sensitive and responsive maternal care,secure children tend to view themselves as worthy of care and others as trustworthy,whereas, due to insensitive and unresponsive care, insecure children are more at riskfor seeing themselves as unworthy and others as untrustworthy (see Verschueren,Doumen, & Buyse, 2012). Children use these models to organize behavioral stra-tegies to regulate their emotions, and ensure that some of their attachment needs aremet even in the face of inconsistent and/or unresponsive care, as in the case ofchildren with avoidant, ambivalent, and controlling attachments. However, childrenwith insecure/other attachments appear to lack a representational strategy to gainfeelings of safety and security (Teti, 1999).

Due to IWMs, children are assumed to apply early attachment-related expecta-tions to future interactions and behave so as to elicit from others responses in linewith these expectations (e.g., Bowlby, 1969/1982). Secure models result in positiveattribution biases that lead children to interpret others’ behaviors positively and torespond appropriately. On the other hand, insecure models tend to result in negativeattribution biases that cause children to act aggressively and/or to withdraw, as wellas to negatively interpret even benign behaviors (e.g., Dodge & Newman, 1981). Theunorganized and dysregulated models of disorganized children appear to place thesechildren at especial risk for aggressive and potentially withdrawing behaviors (e.g.,Bureau & Moss, 2010; Granot & Mayseless, 2001; Verschueren & Marcoen, 1999).With developmental gains in cognitive, linguistic, and behavioral skills, IWMsbecome more solidified and elaborate, and may exert increasing influence onindividual’s behavioral biases and interpretive tendencies (Sroufe, Carlson, Levy, &Egeland, 1999).

The potential pathway of teacher–child relationships

Based on their IWMs, children may attempt to engender interactions from subse-quent caregivers, notably teachers, similar to those of their mothers. Consequently,children tend to develop relationships with teachers consistent with their earlymaternal attachment relationships (e.g., Sroufe & Fleeson, 1986). These relationshipsmay, in turn, reinforce earlier behavioral tendencies developed within the maternalattachment relationship.

An attachment perspective characterizes the affective quality of teacher–childrelationships most often by two relatively independent constructs: closeness, theamount of warmth and positive affect in the relationship; and conflict, the amount ofdiscordance and anger in the relationship (e.g., Baker, 2006; Pianta & Nimetz, 1991;Spilt & Koomen, 2009). A high quality relationship is marked by high levels ofcloseness and low levels of conflict. In general, secure children evidence higherquality relationships with teachers than insecure children (e.g., Howes & Matheson,1992; Pianta, Nimetz, & Bennett, 1997; Sroufe et al., 1983). Several studies suggestthat avoidant and insecure/other children are at especial risk for low-qualityrelationships (O’Connor & McCartney, 2006; O’Connor, Bureau, et al., 2011;O’Connor, Dearing, & Collins, 2011; Teti, 1999). Teacher–child relationships are, inturn, relatively robust predictors of externalizing and internalizing behaviors withespecially strong associations noted between conflict and externalizing behaviors(e.g., Birch & Ladd, 1998; Henricsson & Rydell, 2004; Howes, 2000; Ladd &Burgess, 1999; Pianta & Nimetz, 1991). Consequently, the teacher–child relationship

Attachment & Human Development 267

is expected to be a pathway linking early maternal attachment relationships withexternalizing and internalizing behaviors in later childhood.

An important consideration in research examining the teacher–child relationshipduring childhood as a mechanism linking early maternal attachment relationshipsand later externalizing and internalizing behaviors is the dynamic nature of therelationship. The teacher–child relationship is considered a dynamic system whosequality changes with alterations of the environment in which the relationship existsand the individuals within the relationship (e.g., Kontos, 1992; Mantzicopoulos,2005; Pianta & Walsh, 1996). For example, in their study of teacher–child relation-ships from kindergarten through sixth grade, Jerome, Hamre, and Pianta (2009)found significant, non-linear decreases in closeness and increases in mean levelsof conflict over time. Changes in closeness and conflict during childhood haveimportant consequences for children’s externalizing and internalizing behaviors aschildren’s cumulative experiences in the relationship are associated with psycholo-gical development.

Recently, researchers have used person-centered analyses, which identify sub-groups of individuals with varying developmental trajectories, to examine associa-tions between developmental trajectories for teacher–child relationships duringchildhood and children’s externalizing and internalizing behaviors. More specifically,in their examination of overall relationship quality, as rated by combined closenessand conflict, O’Connor, Dearing, and colleagues (2011) identified four groups ofchildren with varying developmental trajectories for overall relationship quality, asrated by combined closeness and conflict, from first through fifth grade: Poorworsening (4%), Poor improving (7%), Strong worsening (16%), and Strong (73%).Members of the Strong group were less likely to evidence externalizing behaviorsthan their peers in the other groups. In addition, variation was evident inexternalizing behaviors among children in each of the groups. Children in thePoor worsening and Strong worsening groups were more likely than their peers inthe other groups to evidence levels of externalizing behaviors of clinical significance.A decrease in overall relationship quality was the most detrimental in regards tochildren’s externalizing behaviors. In relation to internalizing behaviors, childrenwith elevated levels of internalizing behaviors in early childhood, who were in theStrong group, were less likely to evidence elevated levels of these same behaviors inmiddle childhood than their peers in the other groups.

Individual variation in children’s relational trajectories appears to have impor-tant implications for children’s externalizing and internalizing behaviors. Interest-ingly, in another study (O’Connor & McCartney, 2007) using person-centeredanalyses to examine overall relationship quality during the first few years ofschooling, children with early insecure/other attachments were more likely than theirpeers to evidence a developmental trajectory marked by a significant decrease inrelationship quality between pre-kindergarten and third grade. Variation inchildren’s cumulative experience in the teacher–child relationship may thus be animportant mechanism linking early maternal attachment relationships to laterexternalizing and internalizing behaviors.

There are no published studies that have used person-centered analyses toexamine conflict and closeness in the teacher–child relationship across childhood.The few studies that have used person-centered analyses to investigate relationshipquality over the course of childhood have examined overall relationship quality.Previous research, however, indicates that a person-centered approach to studying

268 E.E. O’Connor et al.

closeness and conflict would enhance our understanding of these aspects of therelationship and their influence on children’s development. For example, Jerome andcolleagues (2009) found that significant individual variation remained in closenessand conflict in teacher–child relationships from kindergarten to sixth grade afteraccounting for average changes in these aspects of the relationship over time,suggesting individual growth trajectories for closeness and conflict vary substantiallyaround group means. Furthermore, research indicates separate, independent effectsof closeness and conflict in the relationship on externalizing and internalizingbehaviors (Birch & Ladd, 1998; Henricsson & Rydell, 2004; Howes, 2000; Ladd &Burgess, 1999; Pianta & Nimetz, 1991). Consequently, it is important to identifydevelopmental trajectories for these separate aspects of the relationship. Accord-ingly, the first goal of this study was to identify developmental trajectories of conflictand closeness between teachers and their students from pre-kindergarten, when themajority of children in the current study began full-time school, through fifth grade.

The potential pathway of early behavior problems

Teacher–child relationships may be only one of multiple pathways through which earlymaternal attachment relationships influence externalizing and internalizing behaviorsin late childhood. More specifically, previous theoretical and empirical researchsuggests that externalizing and internalizing behaviors in early childhood also may beimportant links connecting early maternal attachment relationships and externalizingand internalizing behaviors in late childhood. According to an early starter pathwaytheory, externalizing and internalizing behaviors in early childhood set children ontrajectories for later elevated levels of these same behaviors through continuity ofbehavior, as well as through recursive cycles that reinforce and escalate early behavioraltendencies (e.g., Dodge, Lansford, Burks, Bates, Pettit, Fontaine, et al., 2003; Patterson& Bank, 1989). Empirical research supports these theories, and links externalizing andinternalizing behaviors in early and late childhood (e.g., Patterson, Capaldi, & Bank,1991; Slemming, Sørensen, Thomsen, Obel, & Linnet, 2010).

Early externalizing and internalizing behaviors may be additional, or possiblyalternative, paths linking early maternal attachment relationships to externalizingand internalizing behaviors in late childhood. Furthermore, externalizing andinternalizing behaviors in early childhood may be related to teacher–child relationalcloseness and conflict, as well as behavior problems in late childhood, resulting inendogeneity bias. More specifically, the effects of relational closeness and conflict onexternalizing and internalizing behaviors in late childhood may partly reflect theinfluences of early externalizing and internalizing behaviors as research suggests thatearly elevated levels of externalizing and internalizing behaviors are related torelationships with teachers marked by lower levels of closeness and higher levels ofconflict (e.g., Henricsson & Rydell, 2004). Thus, in order to accurately assess the roleof teacher–child relationships as mechanisms linking early maternal attachmentrelationships to later externalizing and internalizing behaviors, it is necessary toconsider an early starter pathway as well.

The present study

In the present study, we investigated closeness and conflict in teacher–childrelationships and externalizing and internalizing behaviors in early childhood as

Attachment & Human Development 269

potential mechanisms through which early maternal attachment relationships relateto externalizing and internalizing behaviors in late childhood. This study extends theliterature in three significant ways. First, we empirically tested mediators theorized toexplain associations between early maternal attachment relationships and externaliz-ing and internalizing behaviors during late childhood. Second, we consideredteacher–child relationships as potential mechanisms linking early maternal attach-ment relationships to externalizing and internalizing behaviors in late childhood.Third, we identified developmental trajectories for closeness and conflict in theteacher–child relationship over the course of childhood, and examined thesetrajectories as pathways through which early maternal attachment relationshipsinfluence externalizing and internalizing behaviors in late childhood.

In our analyses we used data from the National Institute of Child Health andHuman Development Early Child Care Research Network Study of Early ChildCare and Youth Development (NICHD SECCYD). This data set enables exami-nation of the implications of early maternal attachment relationships for laterexternalizing and internalizing behaviors in a large, national sample of childrenstudied longitudinally. The data set is also large enough to provide power to examinepotential mediating mechanisms between early maternal attachment relationshipsand later development (Thompson, 2008). Previous studies using this sample haveyielded valuable information about maternal attachment and externalizing andinternalizing behaviors (see Thompson, 2008, for a review), and about teacher–childrelationships and externalizing and internalizing behaviors (e.g., Jerome et al., 2009;NICHD Early Child Care Research Network, 2004; O’Connor, 2010; O’Connor &McCartney, 2007; O’Connor, Dearing, et al., 2011). None, however, examined thepathways relating early maternal attachment relationships and children’s externaliz-ing and internalizing behaviors in late childhood.

Methods

Participants

We used data from the first three phases of the NICHD SECCYD (see NICHDEarly Child Care Research Network, 2001, for extensive recruitment and datacollection information). A total of 1364 mothers and children were enrolled in thestudy, with sample demographics of 24% ethnic minority, 11% mothers without ahigh school education, and 14% single mothers at the child’s birth (NICHD EarlyChild Care Research Network, 1997). Our analyses relied on attachment-relateddata collected in the lab at 36 months, and included the 1140 children and familieswith data for this measure.

Measures

Behavior problems

Externalizing and internalizing behaviors were assessed at 36 months and fifth grade(96 months) using the Child Behavior Checklist (CBCL; Achenbach, 1991, 1992;Achenbach, Edelbrock, & Howell, 1987), which contains items describing a broadrange of child behaviors. At 36 months the 99-item Child Behavior Checklist-2/3(CBCL; Achenbach et al., 1987; Achenbach, 1992) was used, and at fifth grade the118-item Child Behavior Checklist was used (Achenbach, 1991). The respondent was

270 E.E. O’Connor et al.

asked how well each item describes the child currently or within the last six months.Higher scores indicate more problems. Raw scores were converted into standard tscores, based on normative data for children of the same age. Scores above 50demonstrate that the raw score was above the mean in the norming population. TheCBCL is one of the most widely usedmeasures of psychological well-being for childrenand has well established psychometric qualifications, good test-retest reliability, andconcurrent and predictive validity (Achenbach, 1991). Because teachers rated one keypredictor variable (teacher–child relationship quality), parental ratings of behaviorproblems were used to avoid problems with shared-rater variance.

Maternal attachment

A modified Strange Situation procedure was used to assess attachment patterns at 36months (see Cassidy, Marvin, & the MacArthur Working Group on Attachment,1992, for a complete description of the measure). The mother and child were invitedto make themselves comfortable in a room and, after three minutes, the mother left.The first separation lasted three minutes, unless the child was distressed. After athree-minute reunion, the mother left again for five minutes, as more time away fromthe mother may be needed to induce anxiety in older children. Children’s behaviorswere classified according to a system developed by the MacArthur Working Groupon Attachment (see Cassidy, Marvin, & the MacArthur Working Group onAttachment, 1992), as secure, ambivalent, avoidant, controlling, or insecure-other.Secure children resume calm, comfortable interactions with their mothers uponreunion. Avoidant children maintain polite neutrality toward their mothers and limitemotional expressions towards them, even after reunion. Ambivalent children arefussy, helpless, whiny, and/or resistant toward their mothers. Controlling childrentake charge of the reunion. Insecure/other children do not demonstrate a coherentattachment strategy during reunion and/or display disorganization according toinfancy criteria. Three coders passed the minimum 75% agreement classificationswith Jude Cassidy on a set of 21 test tapes.

Teacher–child conflict and closeness

A shortened version of the Student–Teacher Relationship Scale (STRS; Pianta,1992) was used to assess teacher perceptions of the quality of the teacher–childrelationship at pre-kindergarten, kindergarten, first, second, third, fourth, and fifthgrades. Using a five-point Likert scale, teachers rate how applicable statements are totheir current relationship with a particular child. The closeness subscale includeseight items, and is an index of the amount of warmth and open communication inthe relationship (e.g., ‘‘I share an affectionate, warm relationship with this child’’),with possible scores from 8 to 40. The conflict subscale includes seven items, andmeasures how much the relationship is marked by antagonistic and disharmoniousinteractions (e.g., ‘‘This child and I always seem to be struggling with each other’’),with possible scores from 7 to 35. Teachers completed the STRS each Spring. STRSscores evidence convergent and predictive validity (e.g., Birch & Ladd, 1997;Doumen, Verschueren, Buyse, De Munter, Max, & Moens, 2009; Howes &Hamilton, 1992; Howes & Ritchie, 1999; Pianta & Nimetz, 1991). Importantly,research indicates that the conflict subscale taps a different underlying construct ofchild behavior than assessments of externalizing behaviors (Doumen, Verschueren,

Attachment & Human Development 271

Buyse, Germeijs, Luyckx, & Soenens, 2008). Cronbach’s a s in the NICHDSECCYD for the STRS ranged from .85 to .94.

Demographic controls

The ratio of family income-to-needs was based on the poverty threshold for theappropriate family size (US Census, 1999). Family income information was collectedevery six months during the course of the study. An average income-to-needs ratiowas computed from 36 months through fifth grade.

Missing data

A total of 1140 children completed the Strange Situation at 36 months. Nagincluster analysis, the person-centered method of analysis we used to modeldevelopmental trajectories for teacher–child relational closeness and conflict, isrobust to missing data and can accurately estimate individuals’ trajectories withone or more data points (Nagin, 1999, 2005). Therefore, we were able to include allchildren who had at least one assessment of closeness and conflict. However,missing values for continuous measures of income-to-needs, externalizing orinternalizing behaviors eliminated 378 children from the dataset. To recover theselost subjects and values, we used multiple imputations (Rubin, 1987; Schafer, 1997)applying a Markov chain Monte Carlo method (Schafer, 1997), which usessimulation from a Bayesian prediction distribution. This was appropriate asthe data were normally distributed and appeared missing at random (Kellum,Rebok, Ialongo, & Mayer, 1994). Ten imputations were performed with a burn-inperiod of 500 to prevent starting values from effecting final parameter estimates.We used SAS PROC MIANALYZE, which aggregates analyses results, tocalculate final parameter estimates (Schafer, 1997). Descriptive statistics andregression models estimated using only original values were similar to those usingimputed values.

Analysis plan

First, we conducted person-centered Nagin cluster analyses to identify prototypicgrowth curves for closeness and conflict in the teacher–child relationship from pre-kindergarten through fifth grade to account for individuals’ cumulative experiencesin these relationships from pre-kindergarten, when most children entered school fulltime, through fifth grade, when behavior problems were assessed. Nagin clusteranalysis uses a multinomial modeling strategy to identify groups of individuals basedon developmental trajectories (see Nagin, 1999, for a complete description), underthe assumption that the population consists of separate groups of individuals withdistinct developmental trajectories and that within-group differences are minimaland non-informative in comparison to between-group differences (Nagin, 1999).Group curves are chosen to represent developmental trajectories that best describethe data (Nagin, 1999). Parameters can vary across groups, allowing for differentiallyshaped trajectories.

Next, we built a hierarchical multiple regression model to examine associationsbetween the maternal attachment relationship and externalizing and internalizingbehaviors at fifth grade. We specified a hierarchical model such that externalizing

272 E.E. O’Connor et al.

and internalizing behaviors were regressed on three blocks of predictor variables.The three blocks of variables were: (1) demographics and maternal attachment, (2)early externalizing and internalizing behaviors, and (3) teacher–child relationships.The demographics and maternal attachment block included demographic controlvariables of child gender and family income-to-needs and variables for attachmentrelationship patterns at 36 months. More specifically, in this block, we included fourdummy variables to represent avoidant, ambivalent, insecure/other, or controllingattachment at 36 months. Secure attachment served as the reference group. Weincluded the control variables of child gender and family income-to-needs so as toexamine the effects of early maternal attachment relationships on children’s laterexternalizing and internalizing behaviors over and above the effects of thesedemographic characteristics, which in previous research were related to externalizingand internalizing behaviors (e.g., Shaw, 2005).

The early externalizing and internalizing behaviors block included externalizingand internalizing behaviors at 36 months—measured concurrent to maternalattachment. The teacher–child relationships block included dummy variables forconflict and closeness group trajectory membership obtained through the Naginanalyses. Multicollinearity was not a problem in any model (tolerance statisticsranged from .80 to .90).

We examined the predictors hierarchically to assess the separate contributionsof each of the two blocks of potential mediators. More specifically, we omitted eachof the two blocks of variables for teacher–child relationships and earlyexternalizing and internalizing behaviors in turn from the full model so that wecould assess the unique contribution of each set. Removing each of the blocksindividually allowed us to account for any potential endogeneity bias resultingfrom associations between early externalizing and internalizing behaviors andteacher–child relational closeness and conflict. Using this modeling strategy, wewere able to examine the stability of each of the attachment relationship variablesacross model specifications and to investigate indications of mediation through aparticular mechanism. An increase in the b value for a variable with the removal ofa block of variables indicates that the effect of the initial variable on the outcome ismediated through one or more of the removed variables (Baron & Kenney, 1986).Complete mediation is suggested when the coefficient associated with the predictorchanges from zero to significance when the mediator is removed from the model.Partial mediation is indicated when the association between the predictor andoutcome is significant when the mediator is in the model but increases in magnitudewhen the mediator is removed from the model.

When mediation was suggested because of an increase in one or more ofthe b coefficients for the attachment variables, we conducted formal tests ofmediation to elucidate relations among the variables. We performed Sobel’s z testswith individual variables. The Sobel’s z test indicates whether the difference betweenthe coefficient associated with the predictor variable of interest excluding andincluding the mediator is greater than zero. If so, mediation is indicated in that theeffect of the predictor on the outcome occurs through the relationship between thepredictor and the mediator. We used the Sobel’s z test to test for mediated effects, asthis test is less vulnerable to Type I errors than other tests of mediation. Specifically,the error rates for Sobel’s z tests are less than .05. Additionally, Sobel’s z tests haveadequate power to identify small effects for sample sizes of 1000 or greater(MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002).

Attachment & Human Development 273

Table 1. Descriptive statistics for study variables.

M (SD) %

Externalizing behaviors (36 m) 50.27 (7.68)Externalizing behaviors (5th grade) 51.19 (7.93)Internalizing behaviors (36 m) 50.91 (7.53)Internalizing behaviors (5th grade) 50.51 (7.95)Average income-to-needs (36 m–5th grade) 3.87 (2.81)Female 51Secure attachment 61Avoidant attachment 5Ambivalent attachment 17Controlling attachment 10Insecure/other attachment 7Teacher–child closeness (54 m) 33.25 (4.60)Teacher–child closeness (Kindergarten) 34.29 (5.11)Teacher–child closeness (1st grade) 33.99 (4.88)Teacher–child closeness (2nd grade) 33.71 (4.79)Teacher–child closeness (3rd grade) 33.14 (4.84)Teacher–child closeness (4th grade) 32.55 (4.86)Teacher–child closeness (5th grade) 31.93 (5.10)Teacher–child conflict (54 m) 11.09 (3.86)Teacher–child conflict (Kindergarten) 10.60 (5.19)Teacher–child conflict (1st grade) 11.03 (5.11)Teacher–child conflict (2nd grade) 10.94 (5.18)Teacher–child conflict (3rd grade) 11.56 (5.77)Teacher–child conflict (4th grade) 11.20 (5.39)Teacher–child conflict (5th grade) 11.42 (5.35)

Results

Descriptive statistics

Table 1 contains descriptive statistics. Means and standard deviations for continuousvariables and percentages for dichotomous variables are presented. The distributionof attachment classifications was similar to those in a meta-analysis of attachmentstudies (van IJzendoorn, Goldberg, Kroonenberg, & Frenkel, 1992). Average scoresfor closeness and conflict showed very little change over time. At each age, mostteachers reported relatively high levels of closeness and low levels of conflict in theirrelationships with the study children. However, at each age, the standard deviationsfor the scores were relatively large indicating a fair amount of individual variation incloseness and conflict at each time point.

Nagin cluster analyses

Results for the Nagin cluster analysis for teacher–child closeness are presented inFigure 1. Fit statistics (available for all models upon request) indicated that a four-group model (BIC: 78724) in which one group was defined by a constant trajectory,one group by a linear trajectory, and two groups by cubic trajectories, was the bestfit, and that model fit was good to excellent. The closeness groups were characterizedas: Stable low, Moderate incliners, High decliners, and High. Children in the Stablelow group demonstrated consistently low levels of closeness in their relationships, asindexed by scores in the lowest quartile for the sample, across time. Children in the

274 E.E. O’Connor et al.

Moderate incliners group evidenced moderately low levels of closeness in theirrelationships at pre-kindergarten, as indexed by scores in the second quartile, and alinear increase in closeness in their relationships from pre-kindergarten through fifthgrade. Children in the High decliners group demonstrated high levels of closeness inpre-kindergarten, as indexed by values in the highest quartile, and a non-lineardecrease in amounts of closeness from pre-kindergarten through fifth grade. Lastly,children in the High group showed high levels of closeness at pre-kindergarten, asindexed by values in the highest quartile, and a slight non-linear oscillation incloseness from pre-kindergarten through fifth grade.

Results for teacher–child conflict are presented in Figure 2. Fit statistics indicatedthat a six-group model (BIC: 78647) in which all groups were defined by a cubictrajectory was the best fit, with the fit being very good to excellent. The groups werecharacterized as Low peakers, Low, Low incliners, High incliners, Moderatedecliners, and Moderate peakers. Children in the Low peakers group demonstratedrelatively low levels of conflict at pre-kindergarten, as evidenced by scores in thelowest quartile, with a relatively sharp increase in conflict at third grade followed bydecreases at fourth and fifth grade. Children in the Low group evidenced low levelsof conflict, as indexed by scores in the lowest quartile, over time with some slightoscillation in levels across time. Children in the Low incliners group showed lowlevels of conflict in pre-kindergarten, with scores in the lowest quartile, and non-linear increases in conflict over-time with the largest increases at third and fifthgrade. Children in the High incliners group demonstrated high levels of conflict, withscores in the highest quartile, across the years. Children in the Moderate declinersgroup evidenced moderate levels of conflict at pre-kindergarten, with scores in thesecond quartile, and a slight increase at kindergarten followed by a decrease overtime. Lastly, children in the Moderate peakers group demonstrated moderate levels

Figure 1. Closeness trajectories for teacher–child relationships from pre-kindergartenthrough fifth grade.

Attachment & Human Development 275

of conflict, as indexed by scores in the second quartile, in pre-kindergarten followedby increases up until third grade and then decreases in amount of conflict throughfifth grade. Group analyses for closeness and conflict were also performed usinghierarchical cluster analysis with similar results being obtained. Associations amongthe closeness and conflict group trajectories were low to moderately low with thestrongest correlation between the High closeness and Low conflict groups (r ¼ .22,p 5 .001) (statistics available upon request), which is consistent with previousresearch (e.g., Pianta, Steinberg, & Rollins, 1995), indicating that conflict andcloseness represent relatively independent dimensions of relationship quality (e.g.,Spilt & Koomen, 2009).

Regression models

Results for externalizing and internalizing behaviors at fifth grade are presented inTables 2 and 3. In the teacher–child relationships block, we included three dummyvariables to represent membership in the Stable low, Moderate incliners, and Highdecliners closeness groups. High closeness was the reference group. In regards toconflict, we included five dummy variables to represent membership in the Lowpeakers, Low incliners, High incliners, Moderate decliners, and Moderate peakersgroups. Low served as the reference group.

Table 2 shows the results for externalizing behaviors. In Model 1, the full model,no significant associations were evident between the attachment relationshipvariables and externalizing behaviors. Of note, in Models 2 and 3, respectively,which included blocks for either early externalizing and internalizing behaviors orteacher–child relationships, insecure/other attachment was a significant predictorof externalizing behaviors. Teacher–child relationships were significant predictors of

Figure 2. Conflict trajectories for teacher–child relationships from pre-kindergarten throughfifth grade.

276 E.E. O’Connor et al.

externalizing behaviors. Children in the Low peakers, Low incliners, High incliners,Moderate decliners, and Moderate peakers groups evidenced significantly higherlevels of externalizing behaviors than their peers in the Low group. No significantdifferences, however, in externalizing behaviors were evident between children ineach of the closeness groups. Externalizing behaviors at 36 months were alsopositively related to externalizing behaviors at fifth grade.

Table 3 displays the results for internalizing behaviors. In Model 1, we note thatsimilar to externalizing behaviors, insecure/other was the only attachment style thatwas significantly associated with internalizing behaviors. Interestingly, as seen inModels 2 and 3, the effect of insecure/other attachment for internalizing behaviorsremained significant after including variables for either early externalizing andinternalizing behaviors or teacher–child relationships. A slightly different patternwas evident for internalizing than externalizing behavior in regards to teacher–childrelationships. More specifically, children in the Stable low group for closenessevidenced significantly higher levels of internalizing behaviors at fifth grade thantheir peers in the High group. No differences, however, were found in internalizingbehaviors among the conflict groups. Early internalizing and externalizing behaviorswere significantly associated with internalizing behaviors at fifth grade.

Mediation analyses

Bivariate correlations and variation in the stability of the insecure/other attachmentvariable across model specifications suggested that teacher–child relationships and

Table 2. Multiple regression models predicting externalizing behavior at fifth grade.

Model 1 Model 2 Model 3

Block 1: Demographics & maternal attachmentChild gender .01 7.03 7.01Family income 7.05 7.09** 7.13**Ambivalent (36 m) 7.04 7.02 7.01Avoidant (36 m) 7.03 7.04 7.03Insecure/other (36 m) .04 .06* .07*Controlling (36 m) 7.01 .03 .01

Block 2: Early externalizing and internalizing behaviorsExternalizing behavior (36 m) .39*** .42***Internalizing behavior (36 m) .05 .04

Block 3: Teacher–child relationshipsStable low closeness1 7.02 7.02Moderate incliners closeness 7.04 7.03High decliners closeness .02 7.02Low peakers conflict2 .08** .11**Low incliners conflict .17*** .21***High incliners conflict .19*** .22***Moderate decliners conflict .10*** .15***Moderate peakers conflict .14*** .20***

R2 .32*** .22*** .16***DR2a .10*** .16***

***p 5 .001; **p 5 .01; *p 5 .05.1High closeness was used as a reference category for the three closeness predictor variables.2Low conflict was used as a reference category for the five conflict predictor variables.aIn comparison to Model 1.

Attachment & Human Development 277

early externalizing and internalizing behaviors were mediators in associationsbetween early maternal attachment relationships and later externalizing andinternalizing behaviors. A necessary requirement for tests of mediation is that asignificant association exists between the main predictor, mediator variables, and theoutcome. Insecure/other attachment was related to externalizing (r ¼ .12, p 5 .01)and internalizing (r ¼ .14, p 5 .001) behaviors at fifth grade. Insecure/otherattachment was also associated with membership in the Stable low closeness teacher–child relationship group (r ¼ .18, p 5 .001), as well as membership in the Low conflictgroup (r ¼ 7 .18, p 5 .001), the Low peakers conflict group (r ¼ .14, p 5 .001), andthe Low incliners conflict group (r ¼ .20, p 5 .001). In addition, insecure/otherattachment was related to early externalizing (r ¼ .10, p 5 .01) and internalizing(r ¼ .12, p 5 .01) behaviors. Membership in the Stable low closeness group wasassociated with internalizing behaviors at fifth grade (r ¼ .11, p 5 .01), whereasmembership in the Low peakers (r ¼ .09, p 5 .05) and Low incliners (r ¼ .17,p 5 .001) groups were associated with externalizing behaviors at fifth grade. Inaddition, early externalizing behaviors (r ¼ .48, p 5 .001) were related to externalizingbehaviors at fifth grade and early internalizing behaviors were related to internalizingbehaviors at fifth grade (r ¼ .38, p 5 .001). Furthermore, in Models 2 and 3,for externalizing and internalizing behaviors the standardized beta-coefficientassociated with insecure/other attachment increased with the removal of the blocksfor teacher–child relationships and early behavior problems, respectively, and the effectof insecure/other attachment either became significant in the case of externalizingbehaviors or increased in significance in the case of internalizing behaviors.

Table 3. Multiple regression models predicting internalizing behavior at fifth grade.

Model 1 Model 2 Model 3

Block 1: Demographics & maternal attachmentChild gender .06* .05 .05Family income .02 7.03 7.04Ambivalent (36 m) 7.03 7.02 .01Avoidant (36 m) .03 .03 .03Insecure/other (36 m) .07* .10** .11**Controlling (36 m) .03 .03 .04

Block 2: Early externalizing and internalizing behaviorsExternalizing behavior .20*** .21***Internalizing behavior .21*** .21***

Block 3: Teacher–child relationshipsStable low closeness1 .08** .09**Moderate incliners closeness 7.04 7.03High decliners closeness .01 7.01Low peakers conflict2 .02 .03Low incliners conflict .05 .02High incliners conflict .03 .04Moderate decliners conflict .01 .03Moderate peakers conflict .05 .04

R2 .19*** .16*** .06***DR2a .03* .13***

***p 5 .001; **p 5 .01; *p 5 .05.1High closeness was used as a reference category for the three closeness predictor variables.2Low conflict was used as a reference category for the five conflict predictor variables.aIn comparison to Model 1.

278 E.E. O’Connor et al.

Sobel’s z-tests confirmed that teacher–child conflict was a mediating mechanismin models predicting fifth grade externalizing behaviors. The effect of insecure/otherattachment was mediated by several of the variables for conflict group membership:Low peakers (z ¼ 1.89, p ¼ .05) and Low incliners (z ¼ 2.21, p 5 .05). In addition,Sobel’s z indicated that the effect of an insecure/other attachment on internalizingbehaviors was partially mediated through membership in the Stable low closenessgroup (z ¼ 7 2.05, p 5 .05). In regards to early behavior problems, Sobel’s zdemonstrated that the effects of insecure/other attachment on externalizingbehaviors in fifth grade were partially mediated through early externalizing beha-viors (z ¼ 2.99, p 5 .05), whereas the effects of insecure/other attachment oninternalizing behaviors at fifth grade were partially mediated through early inter-nalizing behaviors (z ¼ 2.01, p 5 .05).

Discussion

Based on attachment theory, as outlined by Bowlby (1969/1982), early maternalattachment relationships tend to shape children’s later psychological development.Few empirical studies of potential mechanisms explaining this effect, however,have been conducted. The NICHD SECCYD provided a much-needed data setfor a large-scale investigation of early maternal attachment relationships andexternalizing and internalizing behaviors in later childhood that enabled the use ofmediation statistics. Findings indicate that teacher–child relationships, as well asearly externalizing and internalizing behaviors, are important pathways throughwhich early maternal attachment relationships influence children’s later psycholo-gical development.

Attachment in early childhood and externalizing and internalizing behavior in latechildhood

Our study results indicate associations between early maternal attachmentrelationships and externalizing and internalizing behaviors during late childhood.More specifically, insecure/other children demonstrated elevated levels of bothexternalizing and internalizing behaviors. These results are generally in accord withpast work that has found higher rates of externalizing and internalizing behaviorsamong children with disorganized attachment patterns (as defined by either acontrolling or insecure/other attachment pattern) (e.g., Carlson, 1998; Lyons-Ruth,Easterbrooks, & Cibelli, 1997; van IJzendoorn et al., 1999). Our study extendsprevious research by examining variation in the psychological development ofchildren evidencing different subtypes of disorganized attachment, and indicatesthat insecure/other children are most at-risk for multifaceted maladjustment in latechildhood. Results are consistent with previous studies indicating a generalmaladaptive behavioral profile for these children (Cicchetti & Barnett, 1991; Mosset al., 2004; O’Connor, Dearing, et al., 2011; Teti, 1999). The current findings likelyreflect insecure/other children’s disjointed IWMs, and subsequent collapse ofattentional and behavioral strategies and difficulties self-regulating (e.g., Moss,Rousseau, Parent, St-Laurent, & Saintonge, 1998; Teti, 1999). Furthermore, thesefindings are in line with Bowlby’s work tracing childhood depression to experiencesof separation from or loss of maternal attachment figures, especially if thoseexperiences induced hopelessness and helplessness as is often experienced by

Attachment & Human Development 279

children with insecure/other maternal attachment relationships (DeKlyen &Greenberg, 2008; Lyons-Ruth et al., 1997).

Interestingly, associations existed between early insecure/other maternal attach-ment relationships and later externalizing and internalizing behaviors in statisticalmodels that controlled for these same behaviors assessed concurrent to maternalattachment in early childhood. These results indicate that insecure/other attachmentrelationships are also related to changes in externalizing and internalizing behaviorsbetween early and late childhood. These findings are somewhat consistent withrecent research, which found a minimal effect of early maternal attachment dis-organization on externalizing behaviors at grade 1 but a steeper rate of growth inexternalizing behaviors among disorganized boys than their peers from grades 1 to 6with relatively large differences in externalizing behaviors evident between dis-organized boys and their peers at sixth grade (Fearon & Belsky, 2011).

It is possible that insecure/other children evidence increases in externalizingand internalizing behaviors in late childhood as during this period IWMs serve moreof a self-regulation function than in early childhood (Zimmermann, 1999). By latechildhood, when children’s physical proximity to parents in times of stress isdiminished, children’s abilities to represent parental availability, via their IWMs, instressful contexts appear to play increasingly important roles in children’s adaptivefunctioning (Kerns, Klepac, & Cole, 1996; Kerns, Tomich, Aspelmeier, & Contreras,2000; Moss et al., 2009). In fact, some scholars theorize that it is during laterchildhood that IWMs become more of a structure for self-regulation within theperson at which point the behavior-regulating function of the attachment systemdevelops into one of evaluation and prediction (Moss et al., 2009; Zimmermann,1999). Consequently, the behavioral manifestations of these models may increasewith age resulting in a continued influence of early maternal attachment relation-ships on children’s psychological development throughout childhood. Insecure/otherchildren’s lack of a coherent model may thus have an increasingly negative effect ontheir development. Additional research is needed to explore the role of IWMs inassociations between early insecure/other attachment relationships and externalizingand internalizing behaviors in later childhood.

It is of note that, in line with other research among relatively low-risk samples,we did not find associations between the other, organized insecure attachmentpatterns (i.e., avoidant, ambivalent, and controlling) and externalizing and inter-nalizing behaviors in late childhood (e.g., Moss et al., 2009). It is possible thatchildren with other types of insecure attachment relationships are able to use theirIWMs to regulate stress well enough to not evidence elevated levels of externalizingand internalizing behaviors, especially in relatively low-risk contexts (Moss et al.,2009).

Teacher–child relationships and behavior problems

We conducted person-centered longitudinal analyses of teacher–child relationalcloseness and conflict from pre-kindergarten through fifth grade to understandchildren’s cumulative experiences in the relationship from early through late child-hood. Results indicated four groups for closeness and six groups for conflict.Trajectories for conflict and closeness were important predictors of children’s exter-nalizing and internalizing behaviors in fifth grade. Children in each of the groupsevidencing elevated levels of conflict demonstrated higher levels of externalizing

280 E.E. O’Connor et al.

behaviors than their peers in the Low group. Additionally, children in the Stable lowgroup for closeness had higher levels of internalizing behaviors than their peers inthe High group. Conflict and closeness trajectory group memberships remainedsignificant predictors of externalizing and internalizing behaviors at fifth grade,respectively, after accounting for the effects of externalizing and internalizingbehaviors in early childhood. These results indicate that children’s experiences inrelationships with teachers may change externalizing and internalizing behaviorsdeveloped in early childhood. Children appear capable of learning new behavioralstrategies through relationships with teachers (e.g., Baker, Grant, & Morlock, 2008).

Our results support previous findings of a more consistent association betweenconflict and externalizing behaviors and between closeness and internalizing beha-viors (e.g., Buyse, Verschueren, Doumen, Van Damme, & Maes, 2008; Hamre &Pianta, 2001; Henricsson & Rydell, 2004; Ladd & Burgess, 2001). Conflict in therelationship may lead children to interpret others’ behaviors negatively and torespond with aggressive behavior (e.g., Dodge et al., 2003). Interestingly, thestandardized effects of conflict group membership on externalizing behaviors weregreatest for groups in which children evidenced an increase in conflict over-time. Thelater elementary school years mark important periods in development during whichchildren attempt to renegotiate relationships with adults to gain more independencein the relationship while also relying on the caregiver to provide support andregulation (Blos, 1967). Increasing levels of conflict may be especially detrimentalduring these time periods as children may act out if they are unable to gain neededsupport from the teacher–child relationship as they become more independent.

It is also of note that only children who evidenced relationships marked byconsistently low levels of closeness were at-risk for higher levels of internalizingbehaviors. These results suggest that it may be through repeated distant interactionswith teachers that children develop expectations for relationships characterized byrejection, which lead to withdrawal. On the other hand, having close relationshipswith teachers likely provides children with emotional support and security thatencourage positive behaviors and exclude withdrawing behaviors in social contexts,such as at school (Birch & Ladd, 1998; Buyse, Verschueren, Verachtert, & VanDamme, 2009; Ladd et al., 1999). Interestingly, closeness in the therapeutic relation-ship has been found to reduce internalizing behavior problems (Mufson & Dorta,2003). A close teacher–child relationship may provide similar support for children inthe classroom (Baker et al., 2008).

Mediating mechanisms

According to Bowlby’s (1969/1982) formulation of attachment theory, earlymaternal attachment relationships have special significance for development asthese relationships frame the child’s subsequent transactions with the environment,and initiate pathways related to later development (Sroufe et al., 1999; Weinfieldet al., 2008). Our findings are in-line with this theoretical framework. More speci-fically, this study suggests that early maternal attachment relationships are related tolater externalizing and internalizing behaviors in part through their effects oncloseness and conflict in subsequent teacher–child relationships, as well as theirinitiation of externalizing and internalizing behaviors in early childhood.

Mediational analyses indicated that conflict in the teacher–child relationship wasa pathway through which an early insecure/other maternal attachment relationship

Attachment & Human Development 281

influenced externalizing behaviors in late childhood. Children with insecure/otherattachments were more likely than their secure peers to be in a group that experi-enced increased conflict from pre-kindergarten through fifth grade, and consequentlyto evidence elevated levels of externalizing behaviors at fifth grade. On the otherhand, closeness in the teacher–child relationship served as a pathway for an earlyinsecure/other maternal attachment relationship to affect later internalizing beha-viors. Insecure/other children were more likely to experience low levels of relationalcloseness with teachers during elementary school and in turn to demonstrate higherlevels of internalizing behaviors at fifth grade.

Insecure/other attachment likely influences closeness and conflict in the relation-ship through children’s working models and associated behaviors. Disjointed andinconsistent behaviors characterize the insecure/other attachment pattern, making itdifficult for teachers to assess these children’s needs, and to provide appropriateresponses thereby limiting the amount of closeness in the relationship. Associationsbetween insecure/other attachment and conflictual teacher–child relationships inlater childhood are somewhat in line with recent research that found insecure/otherchildren reported higher levels of conflict than their peers in their representations ofmaternal relationships in later childhood (see Moss et al., 2009). Children maytransfer these models of conflictual caregiving relationships with their mothers totheir relationships with teachers. It is of note that insecure/other children were morelikely that their secure peers to be in a group evidencing increases in conflict in theirrelationships with teachers over the course of childhood; however, they were nomore likely than their secure peers to be in a group that demonstrated elevated levelsof conflict with teachers in their pre-kindergarten relationships. These findings may,in part, be explained by researchers’ hypotheses that IWMs play a larger role inchildren’s psychological development in later rather than early childhood (Mosset al., 2009).

The effects of insecure/other attachment relationships on externalizing andinternalizing behaviors in later childhood also partially occurred through elevatedlevels of the same behaviors in early childhood. Insecure/other children evidencedelevated levels of externalizing and internalizing behaviors in early childhood, whichwere in turn related to externalizing and internalizing behaviors in fifth grade.Externalizing and internalizing behaviors in early childhood likely place children ontrajectories for later elevated levels of the same behaviors through continuity ofbehavior (e.g., Birch & Ladd, 1998; Patterson et al., 1991).

Interestingly, our results suggested that the effects of insecure/other attachment onexternalizing behaviors in late childhood are fully mediated through the combinedeffects of teacher–child relationships and early externalizing behaviors. On the otherhand, the effects of insecure/other attachment on internalizing behaviors at fifth gradewere only partially mediated through the combined effects of the teacher–childrelationship and early internalizing behaviors. Insecure/other attachment remained asignificant predictor of internalizing behaviors even when accounting for the effects ofearly internalizing behaviors, as well as closeness and conflict in the teacher–childrelationship. This later finding is in accord with previous research that foundassociations between early disorganized attachment and behavior problems, especiallyinternalizing, in later childhood even after controlling for the effects of early behaviorproblems and experiences in caregiving relationships (e.g., Carlson, 1998;Moss, Smolla,Cyr, Dubois-Comtois, Mazzarello, & Berthiaume, 2006; Shaw & Vondra, 1995;Vondra, Shaw, Swearingen, Cohen, & Owens, 2001).

282 E.E. O’Connor et al.

In regards to internalizing behaviors in later childhood, early internalizingbehaviors and relationships with teachers low in closeness appear to be only two ofseveral pathways through which early insecure/other attachment influences laterinternalizing behaviors. Attachment disorganization, especially insecure/otherattachment, involves a collapse of attentional and behavioral strategies that leadthese children to view themselves as helpless and defenseless and makes themespecially vulnerable to internalizing behaviors (e.g., Moss et al., 1998). Thesechildren may evidence increases in internalizing behaviors as they encounter personaldisappointment, achievement failure, and loss-related stressors, which are moreprevalent in later developmental periods (Brumariu & Kerns, 2010; Goodyear, 2001;Lakdawalla, Hankin, & Mermelstein, 2007).

Limitations and future directions

The current study has multiple limitations and more work is needed to betterunderstand the complex relations between early maternal attachment relationships,early behavior problems, teacher–child relationships, and children’s psychologicaldevelopment. First, the current study relied on teacher reports of relationship qualityand maternal reports of behavior problems. It is essential to use observationalmeasures of teacher–child relationships and behavior problems in the home andschool. Second, we did not consider mother–child relationships in later childhood,as well as father–child relationships, as potential pathways. It is important toinvestigate further other potential mechanisms through which early maternalattachment relationships relate to later behavior problems. Third, the NICHDSECCYD sample is relatively low-risk. It is essential to examine mechanisms linkingearly maternal attachment to later development among higher-risk samples, such asthose with children of adolescent parents or living in high-crime areas (e.g.,Thompson, 2008).

Conclusion

In sum, results indicate both a teacher–child relationship and early starter pathwaylinking early maternal attachment relationships to later externalizing andinternalizing behaviors, and support the pathway model regarding early maternalattachment and later psychopathology originally outlined by Bowlby (1969/1982,1973). Findings are also in-line with Bowlby’s (1969/1982) formulation thatattachment-related disruptions in early childhood, as is the case of children withinsecure/other attachments, negatively impact children’s later psychosocial function-ing through children’s development along deviant pathways (Madigan, Moran,Schuengel, Pederson, & Otten, 2007). The current study expanded on previousresearch by demonstrating that the genesis of behavior problems, related to attach-ment disorganization, lies not only in deregulated mother–child dyadic relationships(see Madigan et al., 2007) but also in unfavorable teacher–child relationships.

References

Achenbach, T.M. (1991).Manual for the Child Behavior Checklist and 1991 profile. Burlington,VT: Department of Psychiatry, University of Vermont.

Achenbach, T.M. (1992). Manual for the Child Behavior Checklist 2–3 and 1992 profile.Burlington, VT: Department of Psychiatry, University of Vermont.

Attachment & Human Development 283

Achenbach, T.M., Edelbrock, C., & Howell, C.T. (1987). Empirically based assessment of thebehavioral/emotional problems of 2- and 3-year-old children. Journal of Abnormal ChildPsychology, 15, 629–650. doi: 10.1007/BF00917246

Baker, J.A. (2006). Contributions of teacher–child relationships to positive school adjustmentduring elementary school. Journal of School Psychology, 44, 211–229. doi: 10.1016/j.jsp.2006.02.002

Baker, J.A., Grant, S., & Morlock, L. (2008). The teacher–student relationship as adevelopmental context for children with internalizing or externalizing behavior problems.School Psychology Quarterly, 23, 3–15. doi: 10.1037/1045-3830.23.1.3

Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in socialpsychological research: Conceptual, strategic, and statistical considerations. Journal ofPersonality and Social Psychology, 51, 1173–1182. doi: 10.1037/0022-3514.51.6.1173

Birch, S.H., & Ladd, G.W. (1997). The teacher–child relationship and children’s earlyschool adjustment. Journal of School Psychology, 35, 61–79. doi: 10.1016/S0022-4405(96)00029-5.

Birch, S.H., & Ladd, G.W. (1998). Children’s interpersonal behaviors and the teacher–childrelationship. Developmental Psychology, 34, 934–946. doi: 10.1037/0012-1649.34.5.934

Blos, P. (1967). The second individuation process of adolescence. Psychoanalytic Study of theChild, 22, 162–186.

Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation. New York, NY: Basic Books.Bowlby, J. (1982). Attachment and loss, Vol. 1: Attachment. New York, NY: Basic Books.

(Original work published 1969)Bretherton, I., & Munholland, K.A. (2008). Internal working models in attachment

relationships: Elaborating a central construct in attachment theory. In J. Cassidy &P.R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications(2nd ed., pp. 102–127). New York, NY: Guilford Press.

Brumariu, L.E., & Kerns, K.A. (2010). Parent–child attachment and internalizing symptomsin childhood and adolescence: A review of empirical findings and future directions.Development and Psychopathology, 22, 177–203. doi: 10.1017/S0954579409990344

Bureau, J.-F., & Moss, E. (2010). Behavioural precursors of attachment representations inmiddle childhood and links with child social adaptation. British Journal of DevelopmentalPsychology, 28, 657–677. doi: 10.1348/026151009X468062

Buyse, E., Verschueren, K., Doumen, S., Van Damme, J., & Maes, F. (2008). Classroomproblem behavior and teacher–child relationships in kindergarten: The moderating role ofclassroom climate. Journal of School Psychology, 46, 367–391. doi: 10.1016/j.jsp.2007.06.009

Buyse, E., Verschueren, K., Verachtert, P., & Van Damme, J. (2009). Predicting schooladjustment in early elementary school: Impact of teacher–child relationship quality andrelational classroom climate. The Elementary School Journal, 110, 119–141. doi: 10.1086/605768

Carlson, E.A. (1998). A prospective longitudinal study of attachment disorganization/disorientation. Child Development, 69, 1107–1128. doi: 10.2307/1132365

Cassidy, J., & Berlin, L.J. (1994). The insecure/ambivalent pattern of attachment: Theory andresearch. Child Development, 65, 971–991. doi: 10.2307/1131298.

Cassidy, J., Marvin, R.S., & the MacArthur Working Group on Attachment. (1992).Attachment organization in preschool children: Coding guidelines. Unpublished manuscript,MacArthur Working Group on Attachment, Seattle, WA.

Cicchetti, D., & Barnett, D. (1991). Attachment organization in maltreated preschoolers.Development and Psychopathology, 3, 397–411. doi: 10.1017/S0954579400005009

DeKlyen, M., & Greenberg, M.T. (2008). Attachment and psychopathology in childhood. InJ. Cassidy & P.R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinicalapplications (2nd ed., pp. 637–665). New York, NY: Guilford Press.

Dodge, K.A., & Newman, J.P. (1981). Biased decision-making processes in aggressive boys.Journal of Abnormal Psychology, 90, 375–379. doi: 10.1037/0021-843X.90.4.375

Dodge, K.A., Lansford, J.E., Burks, V.S., Bates, J.E., Pettit, G.S., Fontaine, R., et al. (2003).Peer rejection and social information processing factors in the development of aggressivebehavior problems in children. Child Development, 74, 374–393. doi: 10.1111/1467-8624.7402004

284 E.E. O’Connor et al.

Doumen, S., Verschueren, K., Buyse, E., De Munter, S., Max, K., & Moens, L. (2009).Further examination of the convergent and discriminant validity of the Student–TeacherRelationship Scale. Infant and Child Development, 18, 502–520.

Doumen, S., Verschueren, K., Buyse, E., Germeijs, V., Luyckx, K., & Soenens, B. (2008).Reciprocal relations between teacher–child conflict and aggressive behavior in kindergar-ten: A three-wave longitudinal study. Journal of Clinical Child and Adolescent Psychology,37, 588–599. doi: 10.1080/15374410802148079

Fearon, P.R.M., & Belsky, J. (2011). Infant–mother attachment and the growth ofexternalizing problems across the primary-school years. Journal of Child Psychology andPsychiatry, 52, 782–791.

Goodyear, I.M. (2001). Life events: Their nature and effects. In I.M.Goodyear (Ed.), Thedepressed child and adolescent (2nd ed., pp. 204–232). New York, NY: CambridgeUniversity Press.

Goossens, F.A., & van IJzendoorn, M.H. (1990). Quality of infants’ attachments toprofessional caregivers: Relation to infant–parent attachment and day-care characteristics.Child Development, 61, 832–837. doi: 10.2307/1130967

Granot, D., & Mayseless, O. (2001). Attachment security and adjustment to school in middlechildhood. International Journal of Behavioural Development, 25, 530–541. doi: 10.1080/01650250042000366

Hamre, B.K., & Pianta, R.C. (2001). Early teacher–child relationships and the trajectory ofchildren’s school outcomes through eighth grade. Child Development 72, 625–638. doi:10.1111/1467-8624.00301

Henricsson, L., & Rydell, A.M. (2004). Elementary school children with behaviorproblems: Teacher–child relations and self-perception. A prospective study. MerrillPalmer Quarterly Journal of Developmental Psychology, 50, 111–138. doi: 10.1353/mpq.2004.0012

Howes, C. (2000). Social-emotional classroom climate in child care, child–teacher relation-ships and children’s second grade peer relations. Social Development, 9, 191–204. doi:10.1111/1467-9507.00119.

Howes, C., & Hamilton, C.E. (1992). Children’s relationships with child care teachers:Stability and concordance with parental. Child Development, 63, 867–878. doi: 10.2307/1131239

Howes, C., & Matheson, C.C. (1992). Contextual constraints on the concordance of mother–child and teacher–child. In R.C. Pianta (Ed.), Beyond the parent: The role of other adults inchildren’s lives (pp. 25–40). San Francisco, CA: Jossey-Bass.

Howes, C., & Ritchie, S. (1999). Attachment organizations in children with difficult lifecircumstances. Development and Psychopathology, 11, 251–268. doi: 10.1017/S0954579499002047

Jerome, E.M., Hamre, B.K., & Pianta, R.C. (2009). Teacher–child relationships fromkindergarten to sixth grade: Early childhood predictors of teacher-perceived conflict andcloseness. Social Development, 18, 915–945. doi: 10.1111/sode.2009.18.issue–410.1111/j.1467-9507.2008.00508.x

Kellum, S., Rebok, G., Ialongo, N., & Mayer, L. (1994). The course and malleability ofaggressive behavior from early first grade into middle school: Results of a developmentalepidemiologically-based preventive trial. Journal of Child Psychology and Psychiatry, 35,259–281. doi: 10.1111/j.1469-7610.1994.tb01161.x

Kerns, K.A., Klepac, L., & Cole, A.K. (1996). Peer relationships and pre-adolescents’perceptions of security in the child–mother relationship. Developmental Psychology, 32,457–466.

Kerns, K.A., Tomich, P.I., Aspelmeier, J.E., & Contreras, J.M. (2000). Attachment-basedassessments of parent–child relationships in middle childhood. Developmental Psychology,36, 614–626.

Kontos, S. (1992). The role of continuity and context in children’s relationships withnonparental adults. In R.C. Pianta (Ed.), Beyond the parent: The role of other adults inchildren’s lives (pp. 109–119). New York, NY: Jossey-Bass.

Ladd, G.W., & Burgess, K.B. (1999). Charting the relationship trajectories of aggressive,withdrawn, and aggressive/withdrawn children during early grade school. ChildDevelopment, 70, 910–929. doi: 10.1111/1467-8624.00066

Attachment & Human Development 285

Ladd, G.W., & Burgess, K.B. (2001). Do relational risks and protective factors moderate thelinkages between childhood aggression and early psychological and school adjustment?Child Development, 72, 1579–1601. doi: 10.1111/1467-8624.00366

Ladd, G.W., Birch, S.H., & Buhs, E.S. (1999). Children’s social and scholastic lives inkindergarten: Related spheres of influence? Child Development, 70, 1373–1400. doi:10.1111/1467-8624.00101

Lakdawalla, Z., Hankin, B.L., & Mermelstein, R. (2007). Cognitive theories of depression inchildren and adolescents: A conceptual and quantitative review. Clinical Child and FamilyPsychology Review, 10, 1–24. doi: 10.1007/s10567-006-0013-1

Lyons-Ruth, K., Easterbrooks, M.A., & Cibelli, C.D. (1997). Infant attachment strategies,infant mental lag, and maternal depressive symptoms: Predictors of internalizing andexternalizing problems at age 7. Developmental Psychology, 33, 681–692. doi: 10.1037/0012-1649.33.4.681

MacKinnon, D.P., Lockwood, C.M., Hoffman, J.M., West, S.G., & Sheets, V. (2002). Acomparison of methods to test mediation and other intervening variable effects.Psychological Methods, 7, 83–104. doi: 10.1037/1082-989X.7.1.83

Madigan, S., Moran, G., Schuengel, C., Pederson, D.R., & Otten, R. (2007). Unresolvedmaternal attachment representations, disrupted maternal behavior and disorganizedattachment in infancy: Links to toddler behavior problems. Journal of Child Psychologyand Psychiatry, 48, 1042–1050.

Main, M., & Cassidy, J. (1988). Categories of response to reunion with the parent at age six:Predictable from infant attachment classifications and stable over a 1-month period.Developmental Psychology, 24, 415–526. doi: 10.1037/0012-1649.24.3.415

Mantzicopoulos, P. (2005). Conflictual relationships between kindergarten children and theirteachers: Associations with child and classroom context variables. Journal of SchoolPsychology, 43, 425–442. doi: 10.1016/j.jsp.2005.09.004

Moss, E., Bureau, J.F., Beliveau, M.J., Zdebik, M., & Lepine, S. (2009). Links betweenchildren’s attachment behavior at early school-age, their attachment-related representa-tions, and behavior problems in middle childhood. International Journal of BehavioralDevelopment, 33, 155–166. doi: 10.1177/0165025408099012

Moss, E., Cyr, C., & Dubois-Comtois, K. (2004). Attachment at early school age anddevelopmental risk: Examining family contexts and behavior problems of controlling-caregiving, controlling-punitive, and behaviorally disorganized children. DevelopmentalPsychology, 40, 519–531. doi: 10.1037/0012-1649.40.4.519

Moss, E., Rousseau, D., Parent, S., St-Laurent, D., & Saintonge, J. (1998). Correlates ofattachment at school age: Maternal reported stress, mother–child interaction, andbehavior problems. Child Development, 69, 1390–1405. doi: 10.2307/1132273

Moss, E., Smolla, N., Cyr, C., Dubois-Comtois, K., Mazzarello, T., & Berthiaume, C. (2006).Attachment and behavior problems in middle childhood as reported by adult andchild informants. Development and Psychopathology, 18, 425–444. doi: 10.1017/S0954579406060238

Mufson, L., & Dorta, K.P. (2003). Interpersonal psychotherapy for depressed adolescents. InA.E. Kazdin & J.R. Weisz (Eds.), Evidence-based psychotherapies for children andadolescents (pp. 148–164). New York, NY: Guilford Press.

Nagin, D.S. (1999). Analyzing developmental trajectories: A semiparametric, group-basedapproach. Psychological Methods, 4, 139–157. doi: 10.1037/1082-989X.4.2.139

Nagin, D.S. (2005). Group-based modeling of development. Cambridge, MA: HarvardUniversity Press.

NICHD Early Child Care Research Network. (1997). The effects of infant child care oninfant–mother attachment security: Results of the NICHD study of early child care. ChildDevelopment, 68, 860–879.

NICHD Early Child Care Research Network. (2001). Nonmaternal care and familyfactors in early development: An overview of the NICHD Study of Early Child Care.Journal of Applied Developmental Psychology, 22, 559–579. doi: 10.1016/S0193-3973(01)00092-2

NICHD Early Child Care Research Network. (2004). Trajectories of physical aggression fromtoddlerhood to middle childhood. Monographs of the Society for Research in ChildDevelopment, 69, 1–143. Serial No. 278. doi: 10.1007/s12103-010-9074-2

286 E.E. O’Connor et al.

O’Connor, E. (2010). Teacher–child relationships as dynamic systems. Journal of SchoolPsychology, 48, 187–218. doi: 10.1016/j.jsp.2010.01.001

O’Connor, E., & McCartney, K. (2006). Testing associations between mother–child andteacher–child relationships. Journal of Educational Psychology, 26, 301–326. doi: 10.1037/0022-0663.98.1.87

O’Connor, E., & McCartney, K. (2007). Examining teacher–child relationships andachievement as part of an ecological model of development. American EducationalResearch Journal, 44, 340–369. doi: 10.3102/0002831207302172

O’Connor, E., Bureau, J.-F., McCartney, K., & Lyons-Ruth, K. (2011). Risks and outcomesassociated with disorganized/controlling patterns of attachment at age three years in thenational institute of child health & human development study of early child care and youthdevelopment. Infant Mental Health Journal, 32, 450–472. doi: 10.1002/imhj.20305

O’Connor, E., Dearing, E., & Collins, B.A. (2011). Teacher–child relationship trajectories:Predictors of behavior problem trajectories and mediators of child and family factors.American Educational Research Journal, 48, 120–162. doi: 10.3102/0002831210365008

Patterson, G.R., & Bank, C.L. (1989). Some amplifying mechanisms for pathologic processesin families. In M.R. Gunnar & E. Thelen (Eds.), Systems and development. The Minnesotasymposia on child psychology, Vol. 22 (pp. 167–209). Hillsdale, NJ: Lawrence ErlbaumAssociates.

Patterson, G.R., Capaldi, D., & Bank, L. (1991). An early starter model for predictingdelinquency. In D.J. Pepler & K.H. Rubin (Eds.), The development and treatment ofchildhood aggression (pp. 139–168). Hillsdale, NJ: Lawrence Erlbaum.

Pianta, R.C. (1992). The Student Teacher Relationship Scale. University of Virginia,Charlottesville.

Pianta, R.C., & Nimetz, S.L. (1991). Relationships between children and teachers:Associations with classroom and home behavior. Journal of Applied DevelopmentalPsychology, 12, 379–393. doi: 10.1016/0193-3973(91)90007-Q

Pianta, R.C., & Walsh, D.J. (1996). High-risk children in schools: Constructing sustainingrelationships. New York, NY: Routledge.

Pianta, R.C., Nimetz, S.L., & Bennett, E. (1997). Mother–child relationships, teacher–childrelationships, and school outcomes in preschool and kindergarten. Early ChildhoodResearch Quarterly, 12, 263–280. doi: 10.1016/S0885-2006(97)90003-X

Pianta, R.C., Steinberg, M.S., & Rollins, K.B. (1995). The first two years of school: Teacher–child relationships and deflections in children’s classroom adjustment. Development andPsychopathology, 7, 295–312. doi: 10.1017/S0954579400006519

Rubin, D.B. (1987). Multiple imputation for nonresponse in surveys. New York, NY:Wiley.

Schafer, J.L. (1997). Analysis of incomplete multivariate data. New York, NY: Chapman andHall.

Shaw, D.S. (2005). The development of aggression in early childhood. In H.E. Fitzgerald,R.A. Zucker, & K. Freeark (Eds.), The crisis in youth mental health: Critical issues andeffective programs (pp. 183–203). New York, NY: Greenwood Publishing Group.

Shaw, D.S., & Vondra, J.I. (1995). Infant attachment security and maternal predictors of earlybehavior problems: A longitudinal study of low-income families. Journal of AbnormalChild Psychology, 23, 335–357. doi: 10.1007/BF01447561

Slemming, K., Sørensen, M.J., Thomsen, P.H., Obel, C.B., & Linnet, K.M. (2010). Theassociation between preschool behavioural problems and internalizing difficulties at age10–12 years. European Child & Adolescent Psychiatry, 19, 787–795. doi: 10.1007/s00787-010-0128-2

Spilt, J.L., & Koomen, H.M.Y. (2009). Widening the view on teacher–child relationships:Teachers’ narratives concerning disruptive versus nondisruptive children. SchoolPsychology Review, 38, 86–101.

Sroufe, L.A., & Fleeson, J. (1986). Attachment and the construction of relationships. In W.Hartup & Z. Rubin (Eds.), Relationships and development (pp. 51–72). Hillsdale, NJ:Erlbaum.

Sroufe, L.A., Carlson, E.A., Levy, A.K., & Egeland, B. (1999). Implications of attachmenttheory for developmental psychopathology. Development and Psychopathology, 11, 1–13.doi: 10.1017/S0954579499001923

Attachment & Human Development 287

Sroufe, L.A., Fox, N.E., & Pancake, V.R. (1983). Attachment and dependency indevelopmental perspective. Child Development, 54, 1615–1627. doi: 10.2307/1129825

Teti, D.M. (1999). Conceptualizations of disorganization in the preschool years: An integration(Vol. xxiii). New York, NY: Guilford Press.

Thijs, J.T., & Koomen, H.M.Y. (2009). Toward a further understanding of teachers’ reportsof early teacher–child relationships: Examining the roles of behavior appraisals andattributions. Early Childhood Research Quarterly, 24, 186–197. doi: 10.1016/j.ecresq.2009.03.001

Thompson, R.A. (2008). Early attachment and later development: Familiar questions, newanswers. In J. Cassidy & P.R. Shaver (Eds.), Handbook of attachment: Theory, research,and clinical applications (2nd ed., pp. 348–365). New York, NY: Guilford Press.

US Census. (1999). Statistical abstract of the United States: 1999. U.S. Bureau of the Census,Washington, DC.

van IJzendoorn, M.H., Goldberg, S., Kroonenberg, P.M., & Frenkel, O.J. (1992). The relativeeffects of maternal and child problems on the quality of attachment: A meta-analysis ofattachment in clinical samples. Child Development, 63, 840–858.

van IJzendoorn, M.H., Schuengel, C., & Bakermans-Kranenburg, M.J. (1999). Disorganizedattachment in early childhood: Meta-analysis of precursors, concomitants, and sequelae.Development and Psychopathology, 11, 225–250. doi: 10.1017/S0954579499002035

Verschueren, K., & Marcoen, A. (1999). Representation of self and social emotionalcompetence in kindergartners: Differential and combined effects of attachment to motherand to father. Child Development, 70, 183–201. doi: 10.1111/1467-8624.00014

Verschueren, K., Doumen, S., & Buyse, E. (2012). Relationships with mother, teacher, andpeers: Unique and joint effects on young children’s self-concept, Attachment & HumanDevelopment, 14, 233–248.

Vondra, J.I., Shaw, D.S., Swearingen, L., Cohen, M., & Owens, E.B. (2001). Attachmentstability and emotional and behavioral regulation from infancy to preschool age.Developmental and Psychopathology, 13, 13–33. doi: 10.1017/S095457940100102X

Wartner, U.G., Grossman, K., Fremmer-Bombik, E., & Suess, G. (1994). Attachmentpatterns at age six in South Germany: Predictability from infancy and implications forpreschool behavior. Child Development, 65, 1014–1027. doi: 10.2307/1131301

Weinfield, N.S., Sroufe, L.A., Egeland, B., & Carlson, E.A. (2008). Individual differences ininfant–caregiver attachment: Conceptual and empirical aspects of security. In J. Cassidy &P.R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications(2nd ed., pp. 78–102). New York, NY: Guilford Press.

Zimmermann, P. (1999). Structure and functions of internal working models of attachmentand their role for emotion regulation. Attachment & Human Development, 1, 291–306.

288 E.E. O’Connor et al.

Copyright of Attachment & Human Development is the property of Routledge and its content may not be

copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.