attachment security and parenting quality predict children's problem-solving, attributions, and...

This article was downloaded by: [Cornell University Library]On: 13 November 2014, At: 16:43Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Attachment & Human DevelopmentPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rahd20

Attachment security and parentingquality predict children's problem-solving, attributions, and lonelinesswith peersH. Abigail Raikes a & Ross A. Thompson ba Bill & Melinda Gates Foundation , Seattle, WA, USAb University of California , Davis, USAPublished online: 26 Sep 2008.

To cite this article: H. Abigail Raikes & Ross A. Thompson (2008) Attachment security andparenting quality predict children's problem-solving, attributions, and loneliness with peers,Attachment & Human Development, 10:3, 319-344, DOI: 10.1080/14616730802113620

To link to this article: http://dx.doi.org/10.1080/14616730802113620

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/rahd20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/14616730802113620

http://dx.doi.org/10.1080/14616730802113620

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Attachment security and parenting quality predict children’s problem-solving,

attributions, and loneliness with peers

H. Abigail Raikesa* and Ross A. Thompsonb

aBill & Melinda Gates Foundation, Seattle, WA, USA; bUniversity of California, Davis, USA

The influence of early relational experience on later social understanding has evokedrich theoretical discussion but relatively little empirical inquiry. Enlisting data from theNICHD Study of Early Child Care and Youth Development, measures of the securityof attachment in infancy, toddlerhood, and early childhood, together with measures ofparenting quality (maternal sensitivity and depressive symptoms) gathered long-itudinally throughout infancy and early childhood, were used to predict differences inchildren’s thoughts and feelings about peers (i.e., social problem solving, negativeattributional biases, aggressive solutions to ambiguous social situations, and self-reported loneliness) when children were 54 months and in first grade. Relationalexperiences, especially before 36 months, were significantly predictive of later peer-related representations. Attachment security at 24 and 36 months was associated withenhanced social problem-solving skills and less loneliness, but security of attachment at15 months was nonpredictive. Early maternal sensitivity was positively associated withlater social problem-solving and negatively with aggressive responses, and earlymaternal depressive symptoms were positively associated with children’s negativeattributions. Concurrent parenting quality was also associated with children’s thoughtsand feelings about peers, but less consistently. These findings shed new light on howearly relational experiences may contribute to social information processing with peersat the end of the preschool years, and that the timing of relational influences may becrucial.

Keywords: attachment; social problem-solving; hostile attributions; loneliness

Introduction

How important are early relationships for children’s later thinking about the social world?This longstanding question of developmental science has, for the past 30 years, beenstudied most intensively within the context of attachment theory. According to this view,infants create mental representations of people and relationships based on the earlysecurity of parent–child relationships which, in turn, influence later functioning (Bowlby,1973). These mental representations, or ‘‘internal working models,’’ are believed toconstitute interpretive filters on social perception, expectations, and memory that causeyoung children to approach new social partners in a biased manner based on the securityof attachment (Bretherton & Munholland, 1999). This theoretical view has receivedgeneral support from a large literature documenting the association between earlier

*Corresponding author. Email: [email protected]

Attachment & Human Development

Vol. 10, No. 3, September 2008, 319–344

ISSN 1461-6734 print/ISSN 1469-2988 online

� 2008 Taylor & Francis

DOI: 10.1080/14616730802113620

http://www.informaworld.com

Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

measures of attachment security and the quality of later relationships, personality, andbehavior regulation. More recently, however, attachment researchers have also confirmedhypothesized associations between early security and later measures of self-concept,emotional understanding, conscience, and other variables that enlist more directlychildren’s working models of self and relationships (see reviews by Thompson, 2006, inpress).

Despite these accomplishments, the interpretation of these findings has beendebated in at least two ways. First, are these behavioral and representational outcomesattributable to the security of early relationships, or rather to continuity in the parentalpractices that initially led to a secure or insecure attachment? This question wasoriginally posed by Lamb, Thompson, Gardner, and Charnov (1985) and has remainedan important consideration, as associations between early attachment security and lateroutcomes have often been reported without intervening assessments of parentalsensitivity (e.g., Lewis, 1997). Second, are relational influences most important early inlife (when personality is initially organizing) or later, when more enduring dispositions,beliefs, and attributions may be emerging? Attachment researchers, like many otherdevelopmentalists, initially focused on parental influences in infancy when personalityprocesses (and internal working models) are most malleable, but some have arguedthat the impact of security may be most influential at the time that importantbehavioral and social-cognitive outcomes are emerging, not before (e.g., Thompson,2000). The latter view is consistent with the expectation (confirmed by the empiricalliterature) that attachment has its strongest associations with social-developmentaloutcomes when each are assessed contemporaneously or close in time (Thompson).Importantly, each question may be answered differently depending on whetherchildren’s social behaviors or children’s representations of social interactions are theoutcome of interest.

This study was designed to provide further insight into these issues concerningrelational influences on social development. Using data from the NICHD Study of EarlyChild Care and Youth Development, we sought to predict children’s peer-related thoughtsand feelings at the end of the preschool years using multiple measures of early attachmentand measures of parenting behavior obtained longitudinally throughout infancy and earlychildhood. The measures selected for this study are central to the hypothesizedconsequences of secure or insecure working models derived from attachment relationships,and include children’s negative attributions when interpreting social behavior, socialproblem-solving skills, and self-reported feelings of loneliness. These measures areimportant not only as reflections of internal working models, but also because theypredict children’s positive and negative social behavior (e.g., Orobio de Castro, Veerman,Koops, Bosch, & Monshouwer, 2002; Runions & Keating, 2007). Social problem-solvingskills are important to capable social interaction while negative social attributionsundermine it (Lemerise & Arsenio, 2000). Loneliness is based on perceptions of socialinclusion that are influenced by children’s self-regard as well as the social context(Parkhurst & Hopmeyer, 1999). As such, it is likely to be influenced by the security ofattachment and its influence on self-concept (Berlin, Cassidy, & Belsky, 1995). We used thebest-validated assessments of attachment security at three ages (infancy, toddlerhood, andearly childhood) to compare their strength of prediction to these hypothesized outcomes.We were also interested in comparing the influence of early attachment history and earlyparenting behaviors with later measures of parenting that were concurrent with theoutcome measures to determine the extent to which early or later relational influences aremost significant.

320 H.A. Raikes and R.A. Thompson

Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Early and later parenting, attachment, and children’s representations of the social world

Although early theoretical debates tended to dichotomize the effects of early attachmentsecurity and later parenting quality on developmental outcomes, virtually allattachment researchers today agree that outcomes are a function of both developmentalhistory and current experience (e.g., Sroufe, Egeland, Carlson, & Collins, 2005;Thompson, 2006). The empirical research is consistent with this view. Severallongitudinal studies have reported that preschool social competence, behavior problems,and other outcomes were predicted both by infant attachment security and bysubsequent measures of parent–child relationship quality (Easterbrooks & Goldberg,1990; Egeland, Kalkoske, Gottesman, & Erickson, 1990; Erickson, Sroufe, & Egeland,1985; Sroufe, Egeland, & Kreutzer, 1990). In an earlier report from the NICHD Studyof Early Child Care and Youth Development, the association between attachmentsecurity in infancy and preschool social competence and behavior problems wasmediated by the quality of parenting subsequent to the attachment assessment (NICHDEarly Child Care Research Network, 2006). Belsky and Fearon (2002) further reportedfrom this study that 3-year-olds’ social competence, problem behavior, language, andschool readiness were predicted by an interaction of attachment and subsequentmaternal sensitivity.

Although these studies confirm the importance of both early attachment and laterparenting, they leave unanswered the question of whether later parenting is uniquelyimportant or whether it reflects the continuing influence of earlier parenting practices(which may have initially contributed to attachment security). When researchers comparethe developmental outcomes of children who were securely attached but later experiencedinsensitive parenting with those who were secure and had subsequent sensitive care, forexample, we are still not able to fully untangle the relative contributions of parenting andattachment, because this approach uses secure attachment as a proxy for early parentalsensitivity despite the modest association between them (de Wolff & van IJzendoorn,1997). A better alternative is to determine whether measures of later parenting remainsignificantly predictive of children’s developmental outcomes after including thecontributions of comparable measures of parenting practices at earlier ages, as well asattachment security. This is the approach of this investigation. If later parenting qualitydoes not predict developmental outcomes beyond the effects of comparable measures ofearly parenting, it suggests that the important differences in parenting characteristics arethose that arise early in a child’s life. If later parenting measures remain significantlypredictive, it may be due to their unique contributions to these developmental outcomes,as well as their temporal proximity to them.

Parenting practices in early childhood might be especially influential for thedevelopment of children’s representations of the social world. Many studies documentthe importance of parenting practices to children’s social understanding. McDowell,Parke, and Spitzer (2002) found, for example, that parents’ responses to hypotheticalsocial problem solving situations were predictive of their 5-year-old children’s responses tosimilar situations, while children of mothers who produced deviant responses to socialproblems were less able to generate competent responses themselves (Pettit, Dodge, &Brown, 1988). It seems likely that these influences would begin to emerge during earlychildhood, when young children are developing the conceptual sophistication to envisionalternative approaches to social dilemmas based on their greater understanding of people’sintentions and motives (Heyman & Gelman, 1999, 2000). This is one way that parentinginfluences after infancy and toddlerhood are likely to make unique contributions to some

Attachment & Human Development 321

Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

of the cognitively-based correlates of the internal working models associated with a secureor insecure attachment.

On the other hand, parenting influences early in life may also have a significantinfluence on the development of later social cognition, especially as they influence infants’affective experiences of themselves and others that subsequently color young children’sdeveloping social attributions and self-concept. This is one way of understanding how, forexample, preverbal behavior in the still-face procedure is associated with externalizing andinternalizing behavior problems years later (see review by Adamson & Frick, 2003) or howmaternal mind-mindedness in infancy predicts later theory of mind (Meins et al., 2003). Ineach case, preverbal experience provides a foundation for later, more sophisticated, formsof social understanding.

In this study, therefore, we examined the extent to which later parenting measuressignificantly predicted individual differences in children’s thoughts and feelings about thesocial world with comparable measures of earlier parenting controlled. We examined theinfluence of two central features of parenting: maternal sensitivity and depressivesymptoms. Each is likely to be influential throughout infancy and early childhood, eventhough each may change significantly during this period. Maternal sensitivity is a coreelement of responsive parenting that contributes not only to attachment security but alsoto greater social and cognitive competence in children (e.g., Bornstein & Tamis- LeMonda,1997; Rogoff, 1990; see Laible & Thompson, 2007, for a review). Maternal depressivesymptoms are a risk factor for social skills deficits in infants and young children andmaladaptive forms of cognition about the self and social interactions in older children(e.g., Garber & Martin, 2002; Murray & Cooper, 1997). One reason for this may be theself-denigrating statements, hopelessness, and negative social attributions that arecharacteristic of depressive symptomatology. Both maternal sensitivity and maternaldepression are likely to be manifested in a family affective climate and verbal statements tochildren that influence the growth of cognition and feelings about the self and the socialworld. We expected, therefore, that differences in maternal sensitivity and depressivesymptoms would significantly predict individual differences in children’s thoughts andfeelings about peer interactions at the end of the preschool years. We were especiallyinterested in whether differences in early measures of sensitivity and depressive symptoms(in infancy and toddlerhood) or later differences in early childhood (more contempora-neous with the measures indexing peer-related thoughts and feelings) would be morestrongly predictive.

Attachment security in infancy and early childhood

Attachment theorists agree that early attachment and later parenting together predict laterbehavioral outcomes such as social competence and behavior problems. With respect tochildren’s representations of the social world, however, the influence of attachment ininfancy is much less clear. One reason is that most studies associating attachment securitywith children’s social or self representations rely on contemporaneous assessments of eachin preschoolers (Thompson, in press). From these, it is difficult to know whether early orcontemporaneous attachment is most important. Although a few studies have found anassociation between infant attachment and later social and emotional understanding (e.g.,Steele, Steele, Croft, & Fonagy, 1999, with respect to emotional understanding; Suess,Grossman, & Sroufe 1992, with respect to children’s hostile attribution bias; Cassidy,Kirsh, Scolton, & Parke, 1996, with respect to social problem-solving), little of the presentwork in this area has compared the effects of early and later measures of attachment


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

security. When infant attachment and contemporaneous attachment measures were bothincluded, Clark and Symons (2000) reported that only the latter predicted 5-year-olds’ self-concept. It is not difficult to understand why. Whereas the behavioral problems morecommonly studied in relation to early attachment may be significantly shaped byexperiences of insensitive or unresponsive parental care in infancy, children’s self-awareness and social representations rely on conceptual skills that begin to emerge afterinfancy (see McCartney, Owen, Booth, Clarke-Stewart, & Vandell, 2004, and NICHDEarly Child Care Research Network, 2004, for consistent findings from the NICHDdataset). As earlier noted, social problem-solving abilities and social attributions are basedon developing conceptions of others’ internal motives and predictions of the outcomes ofalternative social strategies that are beyond the conceptual skills of infants and toddlers.Thus although preschoolers’ social problem-solving skills and social attributions are likelyto be influenced both by the security of attachment and parenting behaviors, the formativeinfluence of a secure attachment may await early childhood when these conceptual skillsare developing most rapidly.

Such a conclusion is consistent with the view that a secure attachment is especiallyinfluential when developmental outcomes are emerging (Thompson, 2000). From thisperspective, attachment security is developmentally most important at the time that thebehavioral and representational processes influenced by internal working models aredeveloping most rapidly. If a coherent sense of self begins to take shape after the age of 3years, for example, then a secure attachment may be most strongly associated withdeveloping self-concept after this age rather than before (see, for example, Cassidy, 1988;Verschueren, Marcoen, & Shoefs, 1996). Likewise, the secure or insecure internal workingmodels associated with attachment security are most likely to influence the development ofattributional style when young children are beginning to be capable of conceptualizing theinternal motivators of peers’ social behaviors (Heyman & Gelman, 1999, 2000). These areeach cognitive capacities beyond the abilities of infants, suggesting that Strange Situationclassifications in infancy may not be as predictive of later representations of socialinteractions as of the behavioral outcomes more commonly studied by attachmentresearchers. In this study, therefore, we sought to predict individual differences inpreschoolers’ peer-related representations based on attachment security assessed in infancy(using the Strange Situation), toddlerhood (using the Attachment Q-Sort), and earlychildhood (using the modified Strange Situation developed by Cassidy, Marvin, & theMacArthur Foundation Network on the Transition From Infancy to Early Childhood,1992). Although these behavioral assessments of attachment differ in design and approach,they are the best-validated assessments in the literature and share sufficient commonvariance to each denote the ‘‘security of attachment’’ in an age-appropriate fashion(Thompson, 2006).

The present study

In sum, this study was designed to explore relational influences on children’s early peer-related representations and feelings in a developmental framework. Drawing on theconclusions of attachment research that early attachment and later parenting are eachimportant to the behavioral outcomes associated with attachment security, we sought toextend these formulations to the development of outcomes indexing social representationsand feelings at the end of the preschool years. To assess whether later parenting practicesare uniquely influential on these outcomes, we examined the predictive significance of latermeasures of maternal sensitivity and depressive symptoms with earlier measures of


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

sensitivity and depressive symptoms (and attachment security) controlled. To assesswhether attachment security is most influential early in life or later in childhood (when thefoundations of children’s conceptual models of peer interactions are being established), wecompared the predictive significance of measures of attachment in infancy, toddlerhood,and early childhood.

We hypothesized that children who were securely attached would exhibit morecompetent and fewer aggressive solutions to social problems, less loneliness, and fewernegative attributions for social behavior, consistent with prior attachment research. Weanticipated that these associations would be stronger when attachment assessments wereconducted at older ages because of the influence of security or insecurity on emergentrepresentational skills. We also hypothesized that parenting practices, specificallydifferences in maternal sensitivity and depressive symptoms, would be associated withchildren’s peer-related thoughts and feelings. In particular, we expected that children withmore sensitive mothers would exhibit more constructive social problem-solving skills andless reliance on aggressive solutions, and that maternal depressive symptoms would beassociated with children’s negative attributions for social behavior and greater loneliness.Although we anticipated that later parenting practices would yield stronger associationswith children’s peer-related thoughts and feelings, the extant literature does not provide abasis for strong hypotheses on this question. Finally, consistent with the influence of bothattachment history and current experience, we anticipated finding interactions betweenattachment security and parenting practices in the prediction of children’s peer-relatedrepresentations and feelings. Specifically, consistent with previous reports from theNICHD Study of Early Child Care and Youth Development, we anticipated thatparenting would have stronger effects on the outcomes of interest for children who wereinsecurely attached as infants (Campbell, Spieker, Burchinal, & Poe, 2006). Therefore, wetested all interactions between parenting and attachment, as there is little precedent foraddressing this question differentially for the outcome measures within existing theoreticaland empirical work.

Method

Participants

The NICHD Study of Early Child Care and Youth Development collected multiplemeasures of social and cognitive development on seven occasions over the first 6 years of1364 children’s lives. Participants were recruited in the following manner. During selected24-hour time sampling blocks in 10 sites, 8986 mothers were approached in the hospitalfollowing the birth of a child. The sites included Little Rock, AR; Irvine, CA; Lawrence,KS; Boston, MA; Philadelphia, PA; Pittsburgh, PA; Charlottesville, VA; Morganton, NC;Seattle, WA; and Madison, WO. Of the 8986 mothers who were approached, 5265 meteligibility requirements and agreed to be contacted after leaving the hospital (i.e., motherhealthy, over 18 years of age, and conversant in English; infant a singleton and healthy;family not planning to move within the next year and living in a neighborhood consideredsafe for visits). Additional details on the selection and recruiting procedures andcharacteristics of the resulting sample are described in several publications (NICHD EarlyChild Care Research Network, 2002a, b, 2003). When children were a month old, familieswere visited in their homes, and 1364 children from 10 sites across the country wereenrolled in the study. The resulting sample included 24% minority children; mothers hadon average 14.4 years of education and 14% were single (NICHD Early Child CareResearch Network, 2003). Children have been followed through adolescence, although the


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

data reported here only include assessments through the end of the first grade. At thistime, 1016 of the original 1364 children were still participating in the study. Families wholeft the study by the time children were in first grade were different from those who stayedin terms of their income, mothers’ education level, and racial background (NICHD EarlyChild Care Research Network, 2003); families leaving the study were more likely to havelow incomes, to be single-parent families, and to be ethnic minorities. In addition, not allassessments were completed for all children in the sample at each time point. Becausechildren with complete data differed from those with incomplete data according to anumber of study variables, to capture the variance in parenting and attachmenthypothesized to be important for children’s peer-related thoughts and feelings, regressionanalyses reported in the present study were not limited to cases in which there werecomplete data at all time points and, instead, regression samples were based on allavailable data. As a result, sample sizes vary across regression analyses.

Measures

Attachment security

Children’s attachment security was assessed on three occasions, at 15, 24, and 36 months,using three different measures.

Strange Situation at 15 months

The first assessment, using the Strange Situation, took place in laboratories when childrenwere 15 months old. Using the standard procedure for conducting and coding the StrangeSituation (see Ainsworth, Blehar, Waters, & Wall, 1978; coding involves evaluatingchildren’s behaviors during a series of maternal departures and reunions with the child),children’s behaviors were videotaped and all tapes were sent to one location for coding.Before conferencing and reaching consensus on differences, inter-rater agreement on the 4-category system was 82% (k ¼ 0.70). Consistent with distributions reported in otherstudies, 59.6% of children, or 712, were classified as secure; 13.6%, or 162, were classifiedas avoidant; 8.5%, or 102, were classified as resistant; and 14.8%, or 177, were classified asdisorganized. The remainder of the children, 3.5% or 42, were deemed unclassifiable. Foruse in regression models, dummy codes were created to compare the mean for securechildren on outcome variables with the subgroup means for resistant, avoidant, anddisorganized children.

Attachment Q-Set

At 24 months, the Attachment Q-Set was used to measure children’s attachment security.Observers spent 2 hours with children and mothers in their homes. Immediately after thevisit, observers sorted the 90 cards of the Attachment Q-Set into nine piles ranging from‘‘very much like the child’’ to ‘‘not at all like the child.’’ Children’s scores were thencorrelated with the card sort that characterizes the behavior of an ideally ‘‘secure child’’(Waters & Deane, 1985). Results from this assessment yield a continuous score rangingfrom 71.0 to 1.0, with higher scores denoting greater security. In the NICHD sample,the mean security score was 0.29 (SD ¼ 0.21; scores ranged from 70.46 to 0.75). Beforeconducting the home visits, observers at each site received 3-days training, includingthree practice visits; observers were also tested using five test tapes. Scores on these tapes


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

were highly correlated (r ¼ 0.77). During the data collection period, reliability wasassessed within each site; across all research assistants, observers’ scores on the Q-Set werecorrelated at 0.73 (McCartney et al., 2004).

Modified Strange Situation at 36 months

At 36 months, a modified Strange Situation procedure (relying upon the MacArthurCoding system), also based on children’s behaviors during a series of maternal departuresand reunions, was used in the laboratory, following procedures outlined by Cassidy et al.(1992). This coding system yielded the same attachment categories as the Strange Situationat 15 months, but the codes used for children’s behaviors were adjusted to be age-appropriate. Three coders who obtained 75% reliability with an expert coder ratedvideotapes of the children’s attachment behavior, and the intercoder agreement on thefour-category coding system was 76% (k ¼ 0.58) before reaching consensus on differencesthrough conferencing. The majority of children were classified as secure (n ¼ 701;61.49%), followed by resistant (n ¼ 197; 17.28%), disorganized (n ¼ 187; 16.4%), andavoidant (n ¼ 55; 4.82%). For use in regression models, dummy codes were created tocompare the mean for secure children on outcome variables with the subgroup means forresistant, avoidant, and disorganized children.

Maternal depressive symptoms

Maternal depressive symptoms were measured when children were 15, 24, 36, and 54months and in first grade using a modified version of the Centers for EpidemiologicalStudies Depression Scale. This 20-item self-report scale measures symptoms of mothers’depressive symptoms over the last 3 weeks. Scores can range from 0 to 60; scores of 16 orhigher are assumed to have clinical significance. At 15 months, 15.4% of mothers hadscores on the CES-D that indicated symptoms of clinical depression; at 24 months and 36months, 16.7%; at 54 months, 19.4%; and in first grade, 15.4% had scores in the clinicalrange. Sample depressive symptoms means were similar at each age of assessment (15months M ¼ 9.05, SD ¼ 8.18; 24 months M ¼ 9.4, SD ¼ 8.63; 36 months M ¼ 9.21,SD ¼ 8.31; 54 months M ¼ 9.83, SD ¼ 8.69; and first grade M ¼ 8.39, SD ¼ 8.47).Cronbach’s alpha coefficients ranged from 0.88 to 0.91 across measurements of depressivesymptoms. Depressive symptoms scores at 54 months and in first grade were usedindividually to assess the concurrent impact of depressive symptoms on children’s negativeattributions, social problem-solving, and feelings of loneliness. To index depressivesymptoms during the first 3 years of life (referred to as ‘‘early depressive symptoms’’),scores at 15, 24, and 36 months were averaged.

Maternal sensitivity

Mothers and children were videotaped together in semi-structured 15-minute play periodsin the home at 15 and 24 months, in the laboratory at 36 and 54 months, and in first grade.Mothers were asked to interact with their children with three sets of age-appropriate toysand a 7-point scale was used to index maternal behavior. At 15 and 24 months, thesensitivity was formed from the sum of three 4-point ratings, sensitivity to nondistress,positive regard, and intrusiveness (reversed). At 36 and 54 months, three 7-point ratingswere combined to represent maternal sensitivity: supportive presence (analogous tosensitivity at earlier ages), respect for autonomy (analogous to intrusiveness, reflected),


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

and hostility (reverse scored). Two coders coded approximately 20% of the tapes andreliability was then calculated by intra-class correlations; correlations across ratersexceeded 0.80 at each age. Cronbach’s alphas for maternal sensitivity exceeded 0.70 atevery age (NICHD Early Child Care Research Network, 2003). Sample means at each ageof assessment were as follows: 15 months (M ¼ 9.39, SD ¼ 1.64); 24 months (M ¼ 9.4,SD ¼ 8.63); 36 months (M ¼ 9.21, SD ¼ 8.31); 54 months (M ¼ 9.83, SD ¼ 8.69); andfirst grade (M ¼ 8.39, SD ¼ 8.47). To measure maternal sensitivity during the first 3 yearsof life (referred to as ‘‘early sensitivity’’), scores on maternal sensitivity at 15, 24, and 36months were converted to z-scores (due to reliance on a 12-point scale at 15 and 24 monthsand a 21-point scale at 36 and 54 months and in first grade) and averaged.

Children’s negative attribution bias at 54 months

At 54 months, children were presented with cartoon drawings of ambiguous socialsituations in which the peer’s intent was ambiguous (e.g., a peer knocks over a tower ofblocks that the child had constructed) and asked to say whether the peer’s intent waspositive or negative. Four stories were presented, followed by questions by theexperimenter about what happened in the situation (‘‘Was it an accident, or was it onpurpose?’’) and how the child would respond. A point was given for each time the childinterpreted the peer’s intent as negative, with a total score of 4 possible. Cronbach’s alphafor this scale was 0.64. The mean for the sample was 1.72 (SD ¼ 1.33).

Children’s aggressive solutions and negative attributions in first grade

At the end of first grade, children were again presented with cartoon drawings of socialsituations in which the intent of the peer was ambiguous, but they were asked to explainthe peer’s intent and behavior more completely in stories with potential themes ofaggression and rejection, and were given eight stories instead of four. After each story,children were asked why they thought the peer behaved the way they did, and what theywould do about it. Children’s attributions were coded as either negative or non-negative,and their hypothesized actions were coded as either aggressive or non-aggressive. Twovariables were computed from children’s responses and used in the present study: a meanscore on the number of negative attributions (M ¼ 4.69, SD ¼ 2.04); and a mean score onthe number of aggressive solutions proposed by the child (M ¼ 1.26, SD ¼ 1.81). Possiblescores and actual scores on negative attributions and aggressive solutions ranged from0 to 8.

Children’s social problem solving skills

At 54 months, children’s social problem solving skills were assessed by presenting themwith five stories depicting social problems like wanting to make friends with another childor gaining possession of a desired toy. The child was asked what he could say or do in eachsituation to accomplish the desired goal, and was prompted for at least two solutions.Children’s scores on socially competent solutions was a composite variable, created bystandardizing and combining three variables: the number of prosocial responses generatedby the child, the variety of responses generated for solving the social problems presented,and the total number of solutions the child was able to produce for each story. Scoresranged from 76.68 to 5.54, with higher scores indicating more socially competentresponses. Cronbach’s alpha for this variable was 0.83. All coding was conducted in a


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

central location, and raters achieved adequate reliability across all variables (kappasranged from 0.83 to 0.91).

Children’s loneliness

Children’s reports of loneliness and social isolation were assessed in first grade using themeasure developed by Asher, Hymel, and Renshaw (1984). This scale was originallydeveloped for children in third through sixth grades, and was modified for youngerchildren by changing statements to questions and by changing the response scale from fivechoices to three (the possible responses to each question were yes, no, or sometimes). Thisscale had 16 items addressing feelings of loneliness (‘‘Is it easy for you to make newfriends?’’) and four filler items with questions about children’s activities. Cronbach’s alphafor the 16 items addressing loneliness was 0.77. Possible scores ranged from 16 to 48, withhigher scores indicating more loneliness, and actual scores ranged from 16 to 46. The meanfor the sample was 23.28, indicating a low degree of loneliness.

Family income

Information on family income was collected at each assessment period from 6 months tothe end of first grade. Mothers’ and fathers’ income from all sources was included in thecalculation of family income, as was receipt of income from public programs such as foodstamps and welfare payments from Temporary Assistance to Needy Families. The totalfamily income was adjusted for the number of people in the household to create an‘‘income-to-needs’’ ratio. The income-to-needs ratio is considered a more accuratedepiction of the resources available to the family than total family income, and was used asa control variable to account the family’s income level at the time of the outcomeassessment.

Results

Bivariate analyses

Before beginning multivariate analyses, bivariate relations between predictor variables(parenting, attachment security, and control variables) and children’s peer-relatedthoughts and feelings at 54 months and in first grade were assessed. Consistent withhypotheses, results indicated that both early attachment security and parenting wereassociated with children’s peer-related representations at 54 months and in first grade.These associations were generally modest although significant, however, and relationalpredictors varied in the strength of their association with different social cognitiveoutcomes. In a similar manner, attachment security assessed at 15 months (in the StrangeSituation), 24 months (Q-Set security score), and 36 months (Modified Strange Situation)varied in their associations with the social cognitive outcomes at each age. Correlationsbetween continuous variables appear in Table 1, and ANOVA results reporting meandifferences between attachment classifications at 15 and 36 months appear in Table 2.Family income and child race were significantly associated with both predictor andoutcome measures, and child gender was associated with the outcome measures.Consequently, family income, race, and child gender were also included in each regressionmodel as control variables. In addition, significant differences were noted between researchsites (this has been typical in analyses of these data), so site was also included in allregression models as a control variable.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Table

1.

Correlationsbetweenpredictorandoutcomevariables.

1.

Income

toneeds

54months

2.

Income

toneeds

springof

1st

grade

3.

Gender

(1¼

male)

4.

Race

(1¼

white,

0¼

other)

5.

Early

maternal

sensitivity

6.

Maternal

sensitivity,

54months

7.

Maternal

sensitivity,

springof

1st

grade

8.

Early

maternal

depressive

symptoms

9.

Maternal

depressive

symptoms,

54months

10.

Maternal

depressive

symptoms,

springof

1st

grade

11.

Q-Set

security

score

12.

Socially

competent

solution

13.

Negative

attributions,

54months

14.

Negative

attributions,

springof

1st

grade

15.

Aggressive

responses

16.

Loneliness

2.74***

37.04

7.00

4.21**

.24**

.02

5.34**

.38**

7.11**

.41**

6.24**

.26**

7.00

.29**

.52***

7.27**

.28**

.05

.38**

.55***

.49***

87.24**

7.27**

7.02

7.16**

7.34***

7.24**

7.21**

97.22**

7.26**

.04

7.17**

7.31***

7.18**

7.24**

.56***

10

7.17**

7.21**

7.01

7.13**

7.23***

7.20**

7.13**

.56***

.51***

11

.11**

.14**

7.13**

.15**

.30**

.22**

.17**

7.17**

7.11**

7.09**

12

.08*

.09**

7.14**

.09**

.24**

.14**

.14**

7.11*

7.12**

7.07*

.15**

13

7.08*

7.05

.07*

7.06

7.11**

7.10**

7.10**

.07**

.06

.09*

.7.06

.02

14

7.11*

7.08*

.06

7.07*

7.16**

7.12**

7.12**

.14**

.10**

.02

7.09**

7.12**

.12**

15

7.09**

7.10*

.07*

7.12**

7.23**

7.19**

7.22**

.11**

.11**

.10**

7.08**

7.11**

.13**

.36***

16

7.13**

7.12**

.02

7.14**

7.15**

7.21**

7.15**

.10**

.07*

.05

7.14**

7.09**

.08**

.05

.12**

Note:*p5

.05;**p5

.01;***p5

.001.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Table

2.

Meandifferencesin

children’sproblem

solving,attributions,andlonelinessacross

attachmentclassificationsat15and36months.

15-m

onth

StrangeSituation

36-m

onth

modified

StrangeSituation

AB

CD

AB

CD

Sociallycompetentsolutions

(F(3,976)¼

.86,NS)at

70.16

(n¼

136)

0.13

(n¼

608)

0.15

(n¼

81)

70.10

(n¼

152)

71.04*

(n¼

48)

0.23

(n¼

593)

70.16

(n¼

153)

70.14

(n¼

148)

15months;(F(3,941)

¼5.21;p5

.01)

at36months

A5

B

Negativeattributions,

54months

(F(3,944)¼

1.43,NS)

at15months;

(F(3,910)¼

2.54;NS)

at36months

1.82

(n¼

130)

1.74

(n¼

592)

1.44

(n¼

77)

1.71

(n¼

146)

2.18

( n¼

46)

1.65

(n¼

576)

1.76

(n¼

149)

1.82

(n¼

142)

Loneliness

(F(3,931)¼

2.56,NS)at

23.98

(n¼

134)

23.09

(n¼

571)

22.72

(n¼

81)

23.75

(n¼

146)

23.46

(n¼

43)

22.87

(n¼

567)

24.18*

(n¼

145)

24.03

(n¼

141)

15months;(F(3,895)

¼3.60;p5

.05)

at36months

C4

B

Aggressiveresponses

(F(3,951)¼

2.41,

p5

.10)A

4B

1.62þ

(n¼

134)

1.19

(n¼

590)

1.06

(n¼

80)

1.22

(n¼

148)

1.32

(n¼

47)

1.19

(n¼

575)

1.22þ

(n¼

148)

C4

B

1.54

(n¼

145)

at15months;

(F(3,914)¼

1.51;NS)

at36months

Negativeattributes,

firstgrade

4.63

(n¼

134)

4.67

(n¼

590)

4.76

(n¼

80)

4.65

(n¼

148)

4.81þ

(n¼

47)

4.57

(n¼

575)

5.03

(n¼

148)

4.61

(n¼

145)

(F(3,951)¼

.07,NS)

at15months;

(F(3,914)¼

2.18;

p5

.10)at36months

A4

B

Note:þp5

.10;*p5

.05.In

instanceswheresignificantormarginallysignificantgroupdifferencesbetweensecure

andinsecure

childrenwerediscovered,thedirectionof

effectindicatingwhichsubgroupdiffered

from

secure

childrenandthesignificance

level

isnoted.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Regression models

Hierarchical regression models were used to assess the contribution of parenting andattachment security after accounting for control variables (family income, child gender,child race, and site, the latter included through nine dummy variables). Becauseconcurrent parenting was measured closer in time to children’s peer-related thoughtsand feelings, the measures of concurrent maternal sensitivity and maternal depressivesymptoms were entered on the second step of the model. Measures of early parenting werethen included in the third step of the regression, and the fourth step tested attachmentrelationship effects. Any relevant interactions between parenting and attachment measureswere included in the fifth step of the model.

To test for statistical interactions between parenting and attachment security,interaction terms were created by centering continuous variables (maternal sensitivity,maternal depressive symptoms, and the Attachment Q-Set security scores) andmultiplying them by one another. For interaction terms involving the Strange Situationat 36 months, centered parenting variables were multiplied by the dummy codes createdfrom attachment classifications. Because we were interested in learning whether theeffects of attachment were dependent on later parenting quality, interaction terms werecalculated when main effects of attachment security were apparent. Beta weights fornon-significant interactions are not reported in the tables. In addition, post-hocsignificance testing and line plotting was conducted upon finding a statisticallysignificant interaction, to assess the statistical significance and the nature of the groupdifferences detected by the interaction.

Previous reports from the NICHD Early Child Care Research Network havedocumented that differences often emerge between children’s attachment classificationsat 15 months and at 36 months (NICHD Early Child Care Research Network, 2001) and,in addition, each measure of attachment security appears to explain a unique portion ofvariance in children’s outcomes (e.g., McCartney et al., 2004). For this reason and also totest study hypotheses regarding the association of attachment security at differentdevelopmental stages with children’s peer-related thoughts and feelings, regression modelswere created with all three attachment measures. Categorical attachment measures werecoded using secure children as the reference group, so that the means for each subgroup ofinsecure children were compared with means for secure children in each regression model.The results for the models for children’s problem-solving, attributions, and loneliness at 54months are reported first, followed by results for children’s problem-solving, attributions,and loneliness in first grade.

Socially competent solutions

With respect to parenting influences, concurrent maternal sensitivity was not a significantpredictor, but concurrent maternal depressive symptoms were negatively associated withsocially competent solutions. Adding concurrent parenting variables increased theamount of variance accounted for by 1% over that accounted for by control variables(F(14, 902) ¼ 3.45, p 5 .001; R2 ¼ .05, R2 change; p 5 .001). When including earlyparenting variables in the next step of the model, early maternal sensitivity was stronglyand positively associated with children’s socially competent responses. Early maternaldepressive symptoms were not related to children’s competent responses and concurrentmaternal depressive symptoms were no longer significantly related when earlier parentingwas introduced into the model. In this third step of the model, adding early parenting


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

influences accounted for an additional 3% of the variance in children’s sociallycompetent solutions (F(16, 902) ¼ 5.03, p 5 .001; R2 ¼ .08), and produced a significantR2 change (p 5 .001). On the fourth step of the regression, Attachment Q-Set securityscores and Strange Situation classifications at 36 months accounted for a significantportion of variance in socially competent solutions; avoidant children at 36 monthsproduced fewer socially competent solutions than secure children, and children withhigher scores on the Attachment Q-Set produced more socially competent solutions. Themodel accounted for 10% of the variance in children’s socially competent solutions(F(23, 879) ¼ 4.15, p 5 001), and adding the measures of attachment produced asignificant R2 change (R2 change ¼ .02; p 5 .01). Finally, in the fifth step, interactionsbetween early maternal depressive symptoms and children’s attachment security at 24and 36 months were included in the regression models because attachment had yieldedsignificant main effects in the preceding analyses. Results indicated that the impact ofmaternal depressive symptoms on children’s socially competent solutions variedaccording to children’s attachment security at 24 months: maternal depressive symptomswere less negatively associated with socially competent solutions for children who weremore secure at 24 months. A plot of this interaction appears in Figure 1, showing thatmaternal depressive symptoms were significantly and negatively associated withchildren’s socially competent solutions for insecure children, but not for children insecure relationships. The final model including the significant interaction betweenattachment security at 24 months and maternal depressive symptoms accounted for 11%of the variance in children’s socially competent solutions (F(24, 902) ¼ 4.35, p 5 .001),and adding the interaction significantly increased the amount of variance accounted forby the model (R2 change ¼ .01; p 5 .001). There was not a significant interactionbetween maternal depressive symptoms and children’s attachment security at 36 months,and so results are not reported. In the final step of the model, boys produced fewersocially competent solutions than girls, but none of the other control variables wassignificant. Results appear in Table 3.

Figure 1. Interaction between depressive symptoms at 54 months and attachment at 24 months, inrelation to socially competent solutions to social problems.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Table 3. Summary of regression results for children’s problem solving, attributions, and lonelinessat 54 months.

Socially competentsolutions

(F(24, 878 ¼ 4.35;R2 ¼ 0.10)

Negative attributions(F(23, 871 ¼ 1.5;

R2 ¼ 0.06)

B SE B B SE B

Step 1SiteIncome to needs ratio 0.03 0.03 0.04 70.04 0.02 70.09*Gender 70.70 0.16 70.15*** 0.15 0.09 0.06Race 0.56 0.22 0.09* 70.23 0.13 70.07

Step 2SiteIncome to needs ratio 0.00 0.03 0.01 70.03 0.02 70.07Gender 70.68 0.16 70.14*** 0.15 0.09 0.06Race 0.38 0.23 0.06 70.16 0.13 70.04Maternal sensitivity, 54 months 0.05 0.03 0.06 70.04 0.02 70.08*Mat. depressive symptoms, 54 months 70.02 0.01 70.08* 0.00 0.00 0.02

Step 3SiteIncome to needs ratio 70.00 0.03 70.01 70.03 0.02 70.07Gender 70.61 0.16 70.13*** 0.15 0.09 0.06Race 0.23 0.23 0.04 70.13 0.14 70.04Early maternal sensitivity 0.90 0.17 0.23*** 70.04 0.10 70.02Early mat. depressive symptoms 70.00 0.01 70.03 0.01 0.01 0.04Maternal sensitivity, 54 months 70.03 0.03 70.03 70.03 0.02 70.07Mat. depressive symptoms, 54 months 70.01 0.01 70.08 70.00 0.00 70.00

Step 4SiteIncome to needs ratio 70.02 0.03 70.03 70.03 0.02 70.06Gender 70.56 0.12 70.13*** 0.15 0.09 0.06Race 70.04 0.24 70.01 70.13 0.14 70.04Early maternal sensitivity 0.82 0.17 0.21*** 70.03 0.10 70.01Early mat. depressive symptoms 70.00 0.01 70.02 0.00 0.01 0.04Maternal sensitivity, 54 months 70.04 0.03 70.04 70.03 0.02 70.06Mat. depressive symptoms, 54 months 70.01 0.01 70.04 70.00 0.00 70.01Avoidant, 15 months 0.05 0.24 0.01 70.03 0.14 70.00Resistant, 15 months 0.10 0.28 0.01 70.28 0.17 70.06Disorganized, 15 months 70.12 0.22 70.02 70.11 0.13 70.03Q-Set security score, 24 months 0.86 0.41 0.07* 70.08 0.24 70.01Avoidant, 36 months 71.03 0.36 70.09** 0.44 0.22 0.07*Resistant, 36 months 70.27 0.22 70.04 0.06 0.13 0.02Disorganized, 36 months 70.22 0.23 70.03 0.04 0.13 0.01

Step 4 (reported only for significant statistical interactions)SiteIncome to needs ratio 70.02 0.03 70.03Gender 70.54 0.16 70.11***Race 70.05 0.24 70.01Early maternal sensitivity 0.80 0.17 0.20***Early mat. depressive symptoms 70.04 0.01 70.01Maternal sensitivity, 54 months 70.04 0.03 70.04Mat. depressive symptoms, 54 months 70.01 0.01 70.05Avoidant, 15 months 0.05 0.24 0.01

(continued)


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Negative attributions at 54 months

None of the control variables significantly predicted children’s negative attributions at 54months, nor did early or concurrent maternal sensitivity or maternal depressive symptoms;no significant R2 change resulted when including concurrent parenting variables in thesecond step of the model, or when including early parenting variables in the third step ofthe model. Avoidant children produced more negative attributions than secure children,but adding attachment variables did not significantly increase the amount of varianceaccounted for. The final model explained 4% of the variance in children’s negativeattributions (F(22, 871) ¼ 1.53; p 5 .10). Results appear in Table 3.

Loneliness

Adding concurrent parenting influences accounted for an additional 1% of the variancebeyond that accounted for by control variables in children’s feelings of loneliness (F(14,813) ¼ 4.05, p 5 .000; R2 ¼ .07), and produced a significant R2 change (p 5 .05).Maternal sensitivity when children were in first grade was negatively associated withchildren’s reports of loneliness, but there was no effect of maternal depressive symptoms onchildren’s feelings of loneliness. Likewise, in the next step of the model, adding earlyparenting influences accounted for an additional 1% of the variance in children’s feelings ofloneliness (F(16, 813) ¼ 4.00, p 5 .001; R2 ¼ .08), and produced a significant R2 change(p 5 .05). Concurrent maternal sensitivity was no longer significant, and early maternalsensitivity emerged as a significant and negative predictor of children’s loneliness in firstgrade. In the fourth step of the model, attachment security at 24 months was a significantnegative predictor of children’s reports of loneliness, with insecure children reporting moreloneliness, and children who were resistant at 36 months reported more loneliness thanchildren who were secure at 36 months. There were no effects of either concurrent or earlyparenting on children’s feelings of loneliness. The final model accounted for 10% of thevariance in children’s loneliness (F(23, 813) ¼ 3.78, p 5 .001), and adding the threeattachment measures significantly increased the amount of variance accounted for (R2

change ¼ .02; p 5 .001). There were no significant interaction effects between maternalsensitivity and children’s attachment security at either 24 or 36 months. In the final step ofthe model, children’s race was a significant predictor of self-reported loneliness, withnonwhite children reporting more loneliness than white children. Results appear in Table 4.

Table 3. (Continued).

Socially competentsolutions

(F(24, 878 ¼ 4.35;R2 ¼ 0.10)

Negative attributions(F(23, 871 ¼ 1.5;

R2 ¼ 0.06)

B SE B B SE B

Resistant, 15 months 0.05 0.28 0.01Disorganized, 15 months 70.09 0.22 70.01Q-Set security score, 24 months 0.87 0.41 0.07*Avoidant, 36 months 71.05 0.36 70.10**Resistant, 36 months 70.28 0.22 70.06Disorganized, 36 months 70.19 0.23 0.03Dep. sympt.6Attachment, 24 months 0.13 0.05 0.09**

Note: þp 5 .10; *p 5 .05; **p 5 .01; ***p 5 .001.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Table

4.

Summary

ofregressionresultsforchildren’sproblem-solving,attributions,andlonelinessin

firstgrade.

Lonelinessattributions

Aggressivesolutions

Negative

(F(23,813¼

3.78;R2¼

0.10)

(F(23,842¼

4.02;R2¼

0.10)

F(23,842¼

2.64;R2¼

0.07)

BSE

BB

SE

BB

SE

B

Step1

Site

**

Gender

70.03

0.36

70.00

0.19

0.12

0.05

0.22

0.14

0.05

Race

72.09

0.52

70.15***

70.57

0.18

70.12**

70.14

0.20

70.03

Incometo

needsratio

70.15

0.07

70.09*

70.04

0.02

70.06

7.04

0.03

70.05

Step2

Site

**

Gender

0.03

0.36

0.01

0.23

0.12

0.06

0.23

0.14

0.06

Race

71.52

0.54

70.11**

70.28

0.18

70.06

70.03

0.21

70.01

Incometo

needsratio

70.13

0.07

70.07

70.00

0.02

70.01

70.03

0.03

70.04

Maternalsensitivity,firstGrade

70.17

0.07

70.10*

70.10

0.02

70.17***

70.04

0.03

70.06

Mat.depressivesymp.,firstgrade

70.00

0.02

70.01

0.02

0.01

0.10*

0.00

0.01

0.00

Step3

Site

**

Gender

70.07

0.36

70.01

0.18

0.12

0.05

0.21

0.14

0.05

Race

71.21

0.56

70.09**

70.14

0.19

70.03

0.07

0.22

0.01

Incometo

needsratio

70.08

0.07

70.05

0.01

0.02

0.02

70.01

0.03

70.01

Earlymaternalsensitivity

70.84

0.41

70.10*

70.39

0.14

70.13**

70.25

0.16

70.07

Earlymat.dep.symptoms

0.04

0.03

0.05

0.02

0.01

0.06

0.04

0.01

0.14**


70.10

0.07

70.06

70.07

0.02

0.11**

70.01

0.03

70.02

Maternaldep.symp.,firstGrade

70.03

0.03

70.05

0.01

0.01

0.05

70.02

0.01

70.07

Step4

Site

**

Gender

70.14

0.36

70.01

0.17

0.12

0.05

0.23

0.14

0.06

Race

71.21

0.58

70.09*

70.14

0.19

70.03

0.04

0.22

0.01

Incometo

needsratio

70.07

0.07

70.04

0.02

0.02

0.03

70.01

0.03

70.01

Earlymaternalsensitivity

70.49

0.41

70.06

70.37

0.14

70.12**

70.24

0.16

70.07

Earlymat.dep.symptoms

0.04

0.03

0.05

0.02

0.01

0.06

0.04

0.01

0.15**


70.09

0.07

70.05

70.06

0.03

70.10*

70.01

0.03

70.02

Mat.dep.symptoms,firstGrade

70.03

0.03

70.05

0.01

0.01

0.05

70.02

0.01

70.07

(continued)


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Table

4.

(Continued).

Lonelinessattributions

Aggressivesolutions

Negative

(F(23,813¼

3.78;R2¼

0.10)

(F(23,842¼

4.02;R2¼

0.10)

F(23,842¼

2.64;R2¼

0.07)

BSE

BB

SE

BB

SE

B

Avoidant,15months

0.69

0.54

0.05

0.33

0.18

0.06

70.16

0.21

70.03

Resistant,15months

70.99

0.64

70.05

70.18

0.22

70.03

70.09

0.25

70.01

Disorganized,15months

0.13

0.50

0.01

0.01

0.17

0.00

70.07

0.20

70.01

Q-Set

security

score,24months

72.81

0.94

70.11*

70.02

0.32

70.02

0.00

0.37

0.01

Avoidant,36months

70.45

0.86

70.02

70.08

0.29

70.00

0.10

0.34

0.00

Resistant,36months

1.23

0.50

0.08*

70.02

0.17

70.00

0.41

0.20

0.08*

Disorganized,36months

0.67

0.51

0.05

0.12

0.17

0.03

70.08

0.20

70.01

þp5

.10;*p5

.05;**p5

.01;***p5

.001.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Aggressive solutions

None of the control variables was significantly associated with children’s aggressivesolutions. When including concurrent parenting variables, maternal sensitivity wasnegatively associated with aggressive solutions and maternal depressive symptoms werepositively associated with aggressive solutions. Adding concurrent parenting influencesaccounted for an additional 3% of the variance in children’s aggressive solutions (F(14,842) ¼ 5.30, p 5 .001; R2 ¼ .08), and produced a significant R2 change (p 5 .001). In thenext step, adding early parenting variables to the model further increased the varianceaccounted for; early maternal sensitivity and concurrent maternal sensitivity were bothsignificant and negative predictors of aggressive solutions, but early and concurrentmaternal depressive symptoms were not associated with aggressive solutions (F(16,842) ¼ 5.47, p 5 .001; R2 ¼ .10), and produced a significant R2 change (p 5 .001).Children whose mothers were less sensitive both early in life and concurrent to themeasurement of aggressive solutions were more likely to generate aggressive solutions.There were no significant effects of attachment on children’s aggressive solutions. The finalstep of the model accounted for 10% of the variance in children’s aggressive solutions(F(23, 842) ¼ 4.02, p 5 .001). Results appear in Table 4.

Negative attributions in first grade

Neither early maternal sensitivity nor maternal depressive symptoms were significantlyassociated with negative attributions; the R2 change in the second step of the model wasnot significant. In the third step of the model, adding early parenting influences accountedfor an additional 1% of the variance in children’s negative attributions (F(16, 842) ¼ 3.41,p 5 .001; R2 ¼ .06), and produced a significant R2 change (p 5 .001). Children ofmothers who reported many symptoms of depression early in their lives made morenegative attributions in first grade. In the fourth step of the model, there was a significantdifference between resistant and secure children at 36 months. Children who were resistantat 36 months made more negative attributions than children who were securely attached.No other effects of attachment were apparent. The final model accounted for 7% of thevariance in children’s negative attributions (F(23, 842) ¼ 2.62, p 5 .001), and adding allthree measures of attachment did not significantly increase the amount of varianceaccounted for beyond the variance accounted for by parenting and the control variables.There were no significant interaction effects between parenting quality and children’sattachment security at 36 months. None of the control variables was significant in the finalregression. Results appear in Table 4. A summary of all results appears in Table 5.

Discussion

This study examined how the security of attachment in early childhood, and the quality ofearly parenting (in the first 3 years of life) and later parenting (at the end of the preschoolyears) were associated with children’s peer-related thoughts and feelings at 54 months andin first grade. Three main conclusions emerge from the findings. First, the security ofattachment is associated with individual differences in how children represent peerinteractions. Children who were securely attached at 24 and 36 months exhibited moreadaptive social attitudes and expectations for peer interactions as much as 4 years later,even with the influences of parenting quality throughout this period controlled. Thissupports the view of attachment theorists that working models of close relationships


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

developed in early childhood can influence later peer-related representations. There aretwo elaborations to this conclusion. First, attachment security was more consistentlyrelated to some aspects of children’s peer-related thoughts and feelings than others.Attachment security figured more prominently in children’s reports of loneliness and theirsocial problem solving skills than in their negative peer attributions. Second, attachmentsecurity at 24 and 36 months was more predictive of children’s peer-related thoughts andfeelings than attachment security at 15 months, which had little predictive significance inthis study. Thus, the developmental stages during which children are securely attachedmay have implications for how attachment security provides a foundation for children’sconceptions of peer relationships in the preschool and early school years.

A second conclusion is that the quality of parenting is also associated with howchildren represent peer interactions, beyond the effects of a secure attachment. This wastrue even when attachment and parenting quality were assessed concurrently, and is notsurprising in light of the modest empirical association between them (see de Wolff & vanIJzendoorn, 1997) and that attachment measures children’s representations of arelationship while parenting measures assess one partner’s behavior. Not surprisingly,different attributes of parenting (sensitivity and depressive symptoms) were differentially

Table 5. Summary of findings.

Negativeattributions54 m

Sociallycompetentsolutions Loneliness

Aggressivesolutions

Negativeattributions1 s

Concurrentdepressivesymptoms

Concurrentmaternalsensitivity

Negativeassociation

Early depressivesymptoms

Positiveassociation

Early maternalsensitivity

Positiveassociation

Negativeassociation

15 monthattachmentsecurity


Positiveassociation

Negativeassociation


Avoidant mademore negativeattributionsthan secure

No othergroupdifferences

Avoidantlower thansecure


Resistanthigher thansecure


Resistant mademore negativeattributionsthan secure


Interactions Insecurechildrenmore affectedby depressivesymptoms

Note: All associations reported in this chart were significant at or lesser than the p 5 .05 level.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

associated with children’s peer-related thoughts and feelings, with sensitivity morestrongly predictive of children’s social problem-solving solutions, and depressivesymptoms associated with negative attributional style. This study is also one of the firstto document that maternal depressive symptoms have a negative association withchildren’s social cognitive processing that is independent of both maternal sensitivity andchildren’s attachment security.

Somewhat more surprising is that parenting practices in the first 3 years were stronglyassociated with children’s peer-related representations at the end of the preschool years,even with later parenting practices controlled. Both parenting quality and attachmentearly in life were predictive, suggesting that the groundwork for children’s cognitive andemotional processing of social situations is founded long before social cognitive biases arereadily apparent. This groundwork may lie in the emotional climate of the family,influences on preverbal representations of relationships or of the self, specific features ofthe quality of care (e.g., maternal mind-mindedness), or other factors that merit furtherinvestigation. This is the first large-scale study examining the roots of such processing inthis manner, and the results offer a provocative framework for reconsidering the origins ofbiases in cognitive and emotional processing about social interactions.

Third, consistent with prior research, attachment security also interacted with thequality of subsequent parenting in predicting children’s peer-related representations. Forchildren who were insecurely attached at 24 months, having a mother with moredepressive symptoms at 54 months resulted in significantly diminished social problem-solving skills compared with insecure children whose mothers did not report high levels ofdepressive symptoms. In a sense, insecurity and maternal depressive symptoms werecumulative risks to children’s social problem-solving skills at the end of the preschoolyears. In this interaction as well as in the other findings of this study, both attachmentsecurity and parenting quality were important.

Early attachment security and children’s problem-solving, attributions, and loneliness

Consistent with the views of attachment theorists that a secure attachment should beassociated with greater skill in establishing and maintaining close relationships,attachment security was associated with more competent social problem-solving skillsand diminished reports of loneliness in the first grade. These aspects of children’s peer-related thoughts and feelings may be most sensitive to children’s attachment historybecause they measure children’s beliefs about their abilities to develop positiverelationships with others and their strategies for maintaining close relationships. Thesefindings are consistent with the results of other research indicating that securely attachedchildren are also more proficient at emotional understanding, conscience development,and other aspects of social understanding (e.g., Raikes & Thompson, 2006; Steele et al.,1999; Thompson, Laible, & Ontai, 2003), and suggest that greater exploration of therepresentational foundations of secure children’s greater competence in close relationshipsis warranted.

Children with insecure attachments exhibited less positive social representationalcharacteristics. Resistant children were more lonely than were securely attached children,consistent with findings reported by Berlin, Cassidy, and Belsky (1995) demonstrating thatchildren who were resistant as infants reported more loneliness in kindergarten. Inaddition, resistant children made more negative social attributions in first grade thansecurely attached children, and avoidant children made more negative attributions at 54months than secure children, which partially replicates the findings of Cassidy et al. (1996)


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

of negative peer attributions in insecurely attached 5-year-olds. However, attachmentsecurity was not related to children’s aggressive responses to social problems in first grade.

This is one of the first studies to establish that attachment security assessed at differentdevelopmental stages has unique associations with children’s later peer-related thoughtsand feelings. Attachment security at 24 and 36 months was consistently related toindividual differences in children’s representations of peer interactions, in contrast toattachment security at 15 months, which had no bearing on future processing in thissample. Attachment relationships may be especially important for children’s developingpeer-related representations as young children make the transition to representationalthought and gain greater insight into other people’s minds, both of which occur beginningat 2–3 years old. This is not to conclude that infant attachment status is irrelevant to laterachievements in social understanding in light of a number of studies linking attachment ininfancy to later emotion understanding and social processing (e.g., Cassidy et al., 1996;Steele et al., 1999; Suess et al., 1992; but see Steele, Steele, & Johansson, 2002). Takentogether, these findings suggest that greater investigation of the processes linkingattachment in infancy and early childhood to later advances in social knowledge arecertainly warranted.

Parenting and children’s problem-solving, attributions, and loneliness

The results of this study support the expectation that both early attachment security andparenting quality are important to early social cognitive development, and that each hasdistinct influences. Parenting quality predicted children’s peer-related thoughts andfeelings even after accounting for the role of early attachment. This is especially notable inthe instance of maternal sensitivity, which was significantly related to social problem-solving skills at 54 months and aggressive solutions in first grade even after accounting forattachment security. However, the effects of parenting were also variable across measuresof children’s peer-related representations and depended on the timing of the assessment.Overall, the discovery of stronger effects from early than later parenting on children’ssocial representations suggests that children’s early relationship experiences, encompassingbut not limited to the development of attachment security, form the basis for how theythink about social interactions as much as 4 years later.

Maternal sensitivity during children’s first 3 years was significantly positively related tochildren’s social problem solving competence at 54 months, and negatively to aggressiveresponses in first grade. This is consistent with the view that sensitive responding providesyoung children with a model of socially competent and constructive behavior. Attachmentsecurity at 24 months also predicted social problem solving at 54 months, suggesting thatmaternal sensitivity (a behavioral quality of the caregiver) and attachment security (arelationship variable) have unique influences on common features of later socialrepresentation. This should not be surprising (from sensitive care children can derivebehavioral models, from security implicit rules of relating to others) and although they arerelated, the modest empirical association between maternal sensitivity and attachmentsecurity also suggests that they are likely to have unique influences on common outcomes(see de Wolff & van IJzendoorn, 1997). The joint and interactive contributions of theseearly relational influences merits further study.

Likewise, maternal depressive symptoms in the first 3 years of life were positivelyassociated with children’s negative attributions in first grade, consistent with thehypothesis that maternal depressive symptoms negatively impact children’sperceptions of others (Schultz & Shaw, 2003). Later parenting was also related to


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

children’s peer-related thoughts and feelings, although these effects were less prevalent.Maternal sensitivity in first grade was negatively associated with children’s aggressivesolutions to social problems in first grade. Concurrent maternal depressive symptoms hadno association with children’s peer-related thoughts and feelings. While this studyestablishes that parenting quality contributes to children’s social representations, futureresearch should more broadly investigate how diverse aspects of parenting contribute tochildren’s cognitive and emotional processing about peer interactions.

This study also points to new avenues for understanding when and how maternaldepressive symptomatology affects young children. In light of the large body of researchdemonstrating the negative effects of maternal depression on children’s development,identifying how early exposure to depressive symptoms contributes to biases in socialattributions years later could lead to important insights regarding the intergenerationaltransmission of depression. For instance, while this study does not provide samples ofmaternal–child discourse, it is possible that the effects of depressive symptoms areapparent in how mothers talk about the social and emotional world with their children,which in turn becomes embedded in children’s processing about social interactions. It isalso important to note that the sample used in this study was largely normative, with arelatively low percentage (between 16 and 20% across the first 6 years of children’s lives) ofmothers whose scores on the depression index met clinical cut-offs. These findings indicatethat even some exposure to depressive symptoms has an impact on children’s thoughts andfeelings about peer interactions, and stronger effects on children’s peer-related thoughtsand feelings may have been discovered in a sample with higher rates of maternaldepression. Future research should attempt to replicate the findings reported here within asample with wider variability of maternal depression scores and with more attention to themechanisms that may link maternal depressive symptoms to children’s attributions andfeelings, to further elucidate the effects of depression on young children’s thoughts andfeelings about peer interactions (see Raikes & Thompson, 2006).

That early parenting is associated with children’s peer-related thoughts and feelingsin the preschool and early school years is particularly notable. Existing work has shownthat responsive parenting in the first 3 years facilitates children’s burgeoning insight intoothers’ minds (e.g., Carpendale & Lewis, 2004). How children’s early relationshipexperiences contribute to their ability to think effectively about social interactions laterin life may hinge on the cascading influence of early social understanding: children whogain more insight into others’ minds during the first 3 years of life may preserve thisadvantage because their early knowledge facilitates more effective cognitive processingabout social interactions in the preschool and early school years. Because no measures ofearly social knowledge were available in this study, it is not possible to determinewhether early parenting affects later peer-related representations by increasing children’searly social knowledge, yet findings from this study clearly suggest that more attentionshould be devoted to children’s relationship experiences, the acquisition of socialknowledge in the first 3 years of life, and associations with later processing about peerinteractions.

It is noteworthy that there was also a significant interaction between attachment andparenting quality which indicated that social problem solving skills among children whowere less securely attached at 24 months were more susceptible to the negative effects ofmaternal depressive symptoms when children were 54 months. This suggests that, in someaspects, the effects of early attachment security and parenting quality are compoundingand cumulative, rendering some young children at greater risk of later deficits in thinkingeffectively about peer interactions. However, because only one significant interaction


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

between parenting and attachment was discovered, care should be taken when interpretingthis finding and efforts should be made to replicate it in the future.

Conclusion

Overall, then, it is clear that a young child’s early relational experience (encompassingattachment security and specific parenting practices) is an important source ofunderstanding for the development of children’s peer-related thoughts and feelings, andprovide a route by which relationship quality affects social development. These findingsdemonstrate that the mental models used to process peer interactions have their roots inearly parent–child relationships; how children think about the social world is based notonly on messages they receive contemporaneously from parents, but also on the behavioralrepresentations formed during infancy and early childhood. However, the security ofattachment in infancy was not predictive of later individual differences in peer-relatedrepresentations. Rather, attachment security at 24 and/or 36 months, when children arebeginning to construct the mental representations they will enlist into social representa-tions in later childhood, was most predictive. In addition, maternal sensitivity anddepressive symptoms before children reach 3 years also seem to have an especiallysignificant role in children’s social representations. This study is a first step in identifyinghow children form mental models used to process peer interactions, in this case bydemonstrating not only the importance of attachment relationships but also the potentialsignificance of parenting during the first 3 years of life for children’s conceptions of socialrelationships. At the same time, the quality of parenting later in the preschool years exertsits own important influence on children’s conceptions of peer interactions.

These results provide new avenues for inquiry on the role of parent–child relationshipsfor later social development. As suggested by Dweck and London (2004), it may be mostappropriate to view children’s representations and the mental processing that is built uponthem as mediators between children’s experiences and their social behavior. Such a view isat the heart of the heuristic power of Bowlby’s (1973) concept of the ‘‘internal workingmodel.’’ Although this study focused on the roots of mental processing rather than linkingsuch processing to social behavior, this study offers glimpses into the complexdevelopmental roots of the representations that children use in developing peer-relatedthoughts and feelings. It also demonstrates that future research on the development ofthese representations should attempt to account for multiple relational influences, and theinteractions between them, to most accurately explain how children translate relationshipexperiences into mental models.

References

Adamson, L., & Frick, J. (2003). The still face: A history of a shared experimental paradigm.Infancy, 4, 451–473.

Ainsworth, M., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychologicalstudy of the strange situation. Hillsdale, NJ: Erlbaum.

Asher, S., Hymel, S., & Renshaw, P. (1984). Loneliness in childhood.Child Development, 55, 1456–1464.Belsky, J., & Fearon, R.M.P. (2002). Early attachment security, subsequent maternal sensitivity, and

later child development: Does continuity in development depend upon continuity of caregiving?Attachment & Human Development, 4, 361–387.

Berlin, L.J., Cassidy, J., & Belsky, J. (1995). Loneliness in young children and infant–motherattachment: A longitudinal study. Merrill-Palmer Quarterly, 41, 91–103.

Bornstein, M., & Tamis-LeMonda, C. (1997). Maternal responsiveness and infant mental abilities:Specific predictive relations. Infant Behavior and Development, 20, 283–296.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

Bowlby, J. (1973). Attachment and loss. Vol. 2: Separation: Anxiety and anger. New York: Basic.Bretherton, I., & Munholland, K. (1999). Internal working models revisited. In J. Cassidy &

P.R. Shaver (Eds.), Handbook of attachment: Theory, research and clinical applications (pp. 89–111). New York: Guilford.

Campbell, S.B., Spieker, S., Burchinal, M., & Poe, M.D. (2006). Trajectories of aggression fromtoddlerhood to age 9 predict academic and social functioning through age 12. Journal of ChildPsychology and Psychiatry, 47, 791–800.

Carpendale, J., & Lewis, C. (2004). Constructing an understanding of mind: The development ofsocial understanding within social interaction. Behavioral and Brain Sciences, 27, 79–96.

Cassidy, J. (1988). Child–mother attachment and the self in six-year-olds. Child Development, 59,121–134.

Cassidy, J., Kirsh, S., Scolton, K., & Parke, R. (1996). Attachment and representations of peerrelationships. Developmental Psychology, 32, 892–904.

Cassidy, K., Marvin, R.S., & the MacArthur Foundation Network on the Transition From Infancyto Early Childhood. (1992). Attachment organization in three- and four-year olds: Procedures andcoding manual. Unpublished manuscript, University of Virginia.

Clark, S., & Symons, D. (2000). A longitudinal study of Q-sort attachment security and self-processes at age 5. Infant and Child Development, 9, 91–104.

de Wolff, M., & van IJzendoorn, M. (1997). Sensitivity and attachment: A meta-analysis on parentalantecedents of infant attachment. Child Development, 68, 571–591.

Dweck, C., & London, B. (2004). The role of mental representation in social development. Merrill-Palmer Quarterly, 50, 428–444.

Easterbrooks, M., & Goldberg, W. (1990). Security of toddler–parent attachment: Relationto children’s sociopersonality functioning during kindergarten. In M. Greenberg, D. Cicchetti, &E. Cummings (Eds.), Attachment in the preschool years (pp. 221–244). Chicago: University ofChicago Press.

Egeland, B., Kalkoske, M., Gottesman, N., & Erickson, M. (1990). Preschool behavior problems:Stability and factors accounting for change. Journal of Child Psychology and Psychiatry, 31, 891–909.

Erickson, M., Sroufe, L., & Egeland, B. (1985). The relationship between quality of attachment andbehavior problems in preschool in a high-risk sample. In I. Bretherton & E. Waters (Eds.),Growing points in attachment theory and research. Monographs of the Society for Research inChild Development, 50(Serial No. 209), 147–166.

Garber, J., & Martin, N.C. (2002). Negative cognitions in offspring of depressed parents:Mechanisms of risk. In S.H. Goodman & I.H. Gotlib (Eds.), Children of depressed parents (pp.121–154). Washington, DC: American Psychological Association.

Heyman, G., & Gelman, S. (1999). The use of trait labels in making psychological inferences. ChildDevelopment, 70, 604–619.

Heyman, G., & Gelman, S. (2000). Preschool children’s use of trait labels to make inductiveinferences. Journal of Experimental Child Psychology, 77, 1–19.

Laible, D.J., & Thompson, R.A. (2007). Early socialization: A relational perspective. In J. Grusec &P. Hastings (Eds.), Handbook of socialization (rev. ed., pp. 181–207). New York: Guilford.

Lamb, M.E., Thompson, R.A., Gardner, W.P., & Charnov, E.L. (1985). Infant–mother attachment.Hillsdale, NJ: Erlbaum.

Lemerise, E.A., & Arsenio, W.F. (2000). An integrated model of emotion processes and cognition insocial information processing. Child Development, 71, 107–118.

Lewis, M. (1997). Altering fate: Why the past does not predict the future. New York: Guilford.McCartney, K., Owen, M.T., Booth, C.L., Clarke-Stewart, A., & Vandell, D.L. (2004). Testing a

maternal attachment model of behavior problems in early childhood. Journal of ChildPsychology and Psychiatry, 45, 765–778.

McDowell, D.J., Parke, R.D., & Spitzer, S. (2002). Parent and child cognitive representations ofsocial situation and children’s social competence. Social Development, 11, 469–486.

Meins, E., Fernyhough, C., Wainwright, R., Clark-Carter, D., Gupta, M., Fradley, E., et al. (2003).Pathways to understanding mind; Construct validity and predictive validity of maternal mind-mindedness. Child Development, 74, 1194–1211.

Murray, L. & Cooper, P. (Eds.). (1997). Postpartum depression and child development. New York:Guilford.

NICHD Early Child Care Research Network (2001). Child-care and family predictors of preschoolattachment and stability from infancy. Developmental Psychology, 37, 847–862.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

NICHD Early Child Care Research Network (2002a). Early child care and children’s developmentprior to school entry. American Educational Research Journal, 39, 133–164.

NICHD Early Child Care Research Network (2002b). The relation of global first grade classroomenvironment to structural classroom features, teacher, and student behaviors. Elementary SchoolJournal, 102, 367–387.

NICHD Early Child Care Research Network (2003). Social functioning in first grade: Associationswith earlier home and child care predictors and with current classroom experiences. ChildDevelopment, 74, 1639–1662.

NICHD Early Child Care Research Network (2004). Affect dysregulation in the mother–childrelationship in the toddler years: Antecedents and consequences. Development and Psycho-pathology, 16, 43–68.

NICHD Early Child Care Research Network (2006). Infant–mother attachment: Risk andprotection in relation to changing maternal caregiving quality over time. DevelopmentalPsychology, 42, 38–58.

Orobio de Castro, B., Veerman, J.W., Koops, W., Bosch, J., & Monshouwer, H.J. (2002). Hostileattribution of intent and aggressive behavior: A meta-analysis. Child Development, 73, 916–934.

Parkhurst, J.T., & Hopmeyer, A. (1999). Developmental change in the sources of loneliness inchildhood and adolescence: Constructing a theoretical model. In K.T. Rotenberg & S. Hymel(Eds.), Loneliness in childhood and adolescence (pp. 56–80). New York: Cambridge UniversityPress.

Pettit, G., Dodge, K., & Brown, M. (1988). Early family experience, social problem solving patterns,and children’s social competence. Child Development, 59, 107–120.

Raikes, H.A., & Thompson, R.A. (2006). Family emotional climate, attachment security, and youngchildren’s emotion understanding in a high-risk sample. British Journal of DevelopmentalPsychology, 24, 89–104.

Rogoff, B. (1990). Apprenticeship in thinking. New York: Oxford University Press.Runions, K.C., & Keating, D.P. (2007). Young children’s social information processing: Family

antecedents and behavioral correlates. Developmental Psychology, 43, 838–849.Schultz, D., & Shaw, D. (2003). Boys’ maladaptive social information processing, family emotional

climate, and pathways to early conduct problems. Social Development, 12, 440–460.Sroufe, L.A., Egeland, B., Carlson, E., & Collins, W. (2005). The development of the person: The

Minnesota study of risk and adaptation from birth to maturity. New York: Guilford.Sroufe, L.A., Egeland, B., & Kreutzer, T. (1990). The fate of early experience following

developmental change: Longitudinal approaches to individual adaptation in childhood. ChildDevelopment, 61, 1363–1373.

Steele, H., Steele, M., Croft, C., & Fonagy, P. (1999). Infant–mother attachment at one year predictschildren’s understanding of mixed emotions at six years. Social Development, 8, 161–178.

Steele, M., Steele, H., & Johansson, M. (2002). Maternal predictors of children’s social cognition: Anattachment perspective. Journal of Child Psychology and Psychiatry, 43, 1–12.

Suess, G.J., Grossman, K.E., & Sroufe, L.A. (1992). Effects of infant attachment to mother andfather on quality of adaptation in preschool: From dyadic to individual organization of self.International Journal of Behavioral Development, 15, 43–65.

Thompson, R.A. (2000). The legacy of early attachments. Child Development, 71, 145–152.Thompson, R.A. (2006). The development of the person: Social understanding, relationships, self,

conscience. In W. Damon & R.M. Lerner (Series Eds.) & N. Eisenberg (Vol. Ed.), Handbook ofchild psychology: Vol. 3. Social, emotional, and personality development (6th ed., pp. 24–98). NewYork: Wiley.

Thompson, R.A. (in press). Early attachment and later development: Familiar questions, new answers. InJ. Cassidy & P.R. Shaver (Eds.),Handbook of attachment (2nd ed.). New York: Guilford.

Thompson, R.A., Laible, D.J., & Ontai, L.L. (2003). rly understanding of emotion, morality, and theself: Developing a working model. In R.V. Kail (Ed.), Advances in child development andbehavior: Vol. 31 (pp. 137–171). San Diego: Academic.

Verschueren, K., Marcoen, A., & Schoefs, V. (1996). The internal working model of the self,attachment, and competence in five-year-olds. Child Development, 67, 2493–2511.

Waters, E., & Deane, K. (1985). Defining and assessing individual differences in attachmentrelationships: Q-methodology and the organization of behavior in infancy and early childhood.In I. Bretherton & E. Waters (Eds.) Growing points of attachment theory and research.Monographs for the Society for Research in Child Development, 50, 41–65.


Dow

nloa

ded

by [

Cor

nell

Uni

vers

ity L

ibra

ry]

at 1

6:43

13

Nov

embe

r 20

14

attachment security and parenting quality predict children's problem-solving, attributions, and...

Documents