Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=rahd20

Attachment & Human Development

ISSN: 1461-6734 (Print) 1469-2988 (Online) Journal homepage: http://www.tandfonline.com/loi/rahd20

Middle Childhood Attachment Strategies:validation of an observational measure

Laura E. Brumariu, Kathryn R. Giuseppone, Kathryn A. Kerns, Magali Van deWalle, Jean-François Bureau, Guy Bosmans & Karlen Lyons-Ruth

To cite this article: Laura E. Brumariu, Kathryn R. Giuseppone, Kathryn A. Kerns, Magali Vande Walle, Jean-François Bureau, Guy Bosmans & Karlen Lyons-Ruth (2018): Middle ChildhoodAttachment Strategies: validation of an observational measure, Attachment & Human Development,DOI: 10.1080/14616734.2018.1433696

To link to this article: https://doi.org/10.1080/14616734.2018.1433696

Published online: 05 Feb 2018.

Submit your article to this journal

View related articles

View Crossmark data

Middle Childhood Attachment Strategies: validation of anobservational measureLaura E. Brumariua, Kathryn R. Giusepponea, Kathryn A. Kernsb, Magali Van de Wallec,Jean-François Bureaud, Guy Bosmanse and Karlen Lyons-Ruthf

aGordon F. Derner School of Psychology, Adelphi University, Garden City, NY, USA; bDepartment ofPsychological Sciences, Kent State University, Kent, OH, USA; cSchool Psychology and Development inContext Research Group, KU Leuven, Leuven, Belgium; dSchool of Psychology, University of Ottawa,Ottawa, ON, Canada; eParenting and Special Education Research Group, KU Leuven, Leuven, Belgium;fDepartment of Psychiatry, Harvard Medical School, Cambridge Hospital, Boston, MA, USA

ABSTRACTThe purpose of this study was to assess behavioral manifestationsof attachment in middle childhood, and to evaluate their relationswith key theoretical correlates. The sample consisted of 87 chil-dren (aged 10–12 years) and their mothers. Dyads participated inan 8-min videotaped discussion of a conflict in their relationships,later scored with the Middle Childhood Attachment StrategiesCoding System (MCAS) for key features of all child attachmentpatterns described in previous literature (secure, ambivalent, avoi-dant, disorganized-disoriented, caregiving/role-confused, hostile/punitive). To assess validity, relations among MCAS dimensionsand other measures of attachment, parenting, and psychologicaladjustment were evaluated. Results provide preliminary evidencefor the psychometric properties of the MCAS in that its behavio-rally assessed patterns were associated with theoretically relevantconstructs, including maternal warmth/acceptance and psycholo-gical control, and children’s social competence, depression, andbehavioral problems. The MCAS opens new grounds for expand-ing our understanding of attachment and its outcomes in middlechildhood.

ARTICLE HISTORYReceived 13 July 2017Accepted 24 January 2018

KEYWORDSAttachment; middlechildhood; controllingbehavior; validity; MCAS

Since the inception of parent–child attachment theory over 50 years ago, substantial progresshas been made in defining and assessing attachment patterns in children and adolescents.The Strange Situation Procedure (SSP; Ainsworth, Blehar, Waters, & Wall, 1978) has been thegold standard for measuring attachment in infancy. Adaptations of the SSP (Cassidy &Marvin,1992), as well as representational measures (e.g. story stem procedure; Bretherton, Ridgeway,& Cassidy, 1990) have been employed to assess attachment in the preschool years. To assessattachment in adolescence, representational measures (e.g. autobiographical interviews;Adult/Adolescent Attachment Interview, Main, Kaplan, & Cassidy, 1985) and questionnaires(e.g. the Experiences of Close Relationships questionnaire; Brenning, Soenens, Braet, &Bosmans, 2012), as well as behavioral measures (GPACS; Obsuth, Hennighausen, Brumariu,& Lyons-Ruth, 2014) have all been developed.

CONTACT Laura E. Brumariu lbrumariu@adelphi.edu

ATTACHMENT & HUMAN DEVELOPMENT, 2018https://doi.org/10.1080/14616734.2018.1433696

© 2018 Informa UK Limited, trading as Taylor & Francis Group

In middle childhood, attachment assessments have continued to rely mainly onrepresentational measures of attachment, including story-stem procedures (e.g. Granot& Mayseless, 2001; Kerns, Brumariu, & Seibert, 2011), autobiographical interviews (e.g.the Friends and Family Interview; Kriss, Steele, & Steele, 2012), script story assessmentsof secure base knowledge (e.g. Psouni & Apetroaia, 2014; Waters, Bosmans, Vandevivere,Dujardin, & Waters, 2015), and questionnaires (e.g. the security scale; Kerns, Aspelmeier,Gentzler, & Grabill, 2001). Nevertheless, Main et al. (1985) advanced the idea thatevidence for attachment processes should be detectable on the levels of observableattachment behavior as well as in attachment representations. However, to date, thereare few behavioral attachment assessments in middle childhood, and none that assessall forms of attachment behavior that are relevant to this period (see Bureau,Easterbrooks, & Lyons-Ruth, 2009; for scales for disorganized and controlling forms ofattachment; see Boldt, Kochanska, Grekin, & Brock., 2016; for assessment of organizedand disorganized, but not controlling, forms of attachment; see Bosmans, Braet, Heylen,& De Raedt, 2015; for assessment of security). As such, although observation of attach-ment behaviors has been critical for the advancement of our understanding of attach-ment at earlier ages, which behaviors are typical for middle childhood remains largely anempirical question.

Thus, the goals of the current study were to assess behavioral manifestations of allsix forms of post-infancy attachment patterns (secure, ambivalent, avoidant, disorga-nized-disoriented, caregiving/role-confused, hostile/punitive) in middle childhood,and to evaluate their relations with key theoretical constructs (quality of parentingand child adjustment). We developed a coding system, organized around the securebase construct, assessing different aspects of disorganized attachment as well as theorganized attachment patterns in children 9–12 years of age with a parent, as parentsare typically children’s primary attachment figures at this age (Kerns & Brumariu,2016).

Attachment behavior in middle childhood

The primary focus in attachment research has been delineating individual differences inattachment patterns for purposes of understanding the developmental significance ofattachment. Securely attached children are able to use the caregiver as a secure basefrom which to explore and as a haven of safety during times of distress (Ainsworth, 1989;Bowlby, 1969/1982). They also develop a cognitive model of the caregiver as responsive,sensitive, and available (Bowlby, 1982). Avoidantly attached children use a self-reliancestrategy when distressed in an effort to maintain contact with a caregiver who with-draws from the interaction and rejects the child when the child displays negative affect(Cassidy, 1994). Children with ambivalent attachments exhibit a heightening strategycharacterized by fearful and excessive attachment behaviors to trigger the attention of acaregiver whom they view as inconsistently responsive (Cassidy, 1994). Children withdisorganized attachments lack a consistent strategy of organizing attachment responsesunder stress (Main & Solomon, 1990), show apprehension, disorientation, or conflictbehaviors, in infancy, in the presence of the caregiver under stress, and have typicallyexperienced a caregiver who has shown frightened, frightening, or other atypicalbehavior (Lyons-Ruth & Jacobvitz, 2016).

2 L. E. BRUMARIU ET AL.

After age 3, controlling patterns of interaction emerge for most disorganized children,characterized either by organizing and supporting the parent (controlling-caregiving) orby punitive behaviors to the parent (controlling-punitive, see Lyons-Ruth & Jacobvitz,2016). These forms of attachment disorganization (disoriented, controlling-caregivingand controlling punitive) have been consistently documented in preschool (Moss,Bureau, St. Laurent, & Tarabulsky, 2011), early school age (age 8; Bureau et al., 2009),and late adolescence (age 19; Obsuth et al., 2014).

Behavioral measures of attachment that rely on separations from caregivers arewidely used with children up to age 8 (e.g. Bureau et al., 2009). However, such separa-tions do not necessarily provoke an urgent need for parental reassurance in later middlechildhood, so that finding assessment formats in this period become more challenging.To date, one other behavioral measure assessing organized and disorganized patterns ofattachment has been proposed for attachment in later middle childhood. Boldt et al.(2016) developed a coding system to assess security, avoidance, ambivalence, anddisorganization based on 10-year-olds’ naturalistic interactions with a parent. Theyfound that attachment patterns were related to a history of responsive care fromparents, as well as to attachment assessed with other methods. Findings were moreconsistent for indicators of security, avoidance, and disorganization than for ambivalentattachment. Security and disorganization were most consistently linked to measures ofchild adaptation, in the expected direction. Their approach provides one way to assessattachment, although there are some limitations to this method, in that the assessmentprocedure is long (over an hour), and the coding system does not distinguish amongforms of disorganization. Thus, there is still the strong need for a well-groundedbehavioral assessment of all attachment patterns (secure, ambivalent, avoidant, disorga-nized-disoriented, caregiving/role-confused, hostile/punitive) in later middle childhood.

To address this gap, we developed the Middle Childhood Attachment StrategiesCoding System (MCAS, Brumariu, Kerns, Bureau, & Lyons-Ruth, 2014). The main purposeof the current study was to evaluate the psychometric properties of the MCAS. TheMCAS assesses all attachment patterns documented in the literature, and it is based oncoding the child’s behavior during a parent–child discussion of a conflict in theirrelationship. We coded a conflict task as we expected that it would elicit discussion ofan emotionally salient event that would trigger attachment behaviors and also be acontext where one can observe how older children express their views, while navigatingthe potentially divergent views of the parent (Allen, Hauser, O’Connor, & Bell, 2002;Obsuth et al., 2014). Comfort expressing emotions, regardless of their affective valence,and using the parent as a secure base in potentially challenging situations, such asdiscussions of conflicts, are important features of a goal-corrected partnership (Allenet al., 2002; Bowlby, 1969/1982).

In developing the current coding system, we drew on previous literature describingbehavioral manifestations of attachment patterns in emotionally charged interactions atyounger and older ages (Bureau et al., 2009; Obsuth et al., 2014). We also incorporatedthe legacy of previous literature and its insights into child attachment strategies frominfancy to adolescence, with a particular focus on theoretical and empirical findingsregarding attachment in middle childhood (Kerns, 2008; Kerns & Brumariu, 2016).Specifically, during this age period, there is a shift toward greater co-regulation ofsecure base contact between the child and parent figure, such that attachment entails

ATTACHMENT & HUMAN DEVELOPMENT 3

a collaborative alliance whereby the child starts to use the parent as a resource ratherthan a powerful figure who solves the child’s problems (Kerns & Brumariu, 2016).Further, because previous work has shown that a dimensional approach may fit thedata better than a taxonic model (Roisman, Fraley, & Belsky, 2007), and allow for greatersensitivity in capturing subtle behaviors, we developed continuous scales rather thanclassifications.

Validating criteria: associations with other attachment measures, parenting, socialcompetence, and indexes of psychopathology

To evaluate MCAS’s validity, we relied on several criteria. A new measure of attachmentshould be associated with other measures of attachment administered close in time(Kerns et al., 2011). Our first objective was to assess the construct validity of the MCASwith other middle childhood measures of attachment. Given the secure base construct iscentral to the MCAS, we expected that MCAS security would be related to securityassessed with the security scale (Kerns et al., 2001). Further, we expected that eachMCAS attachment patterns would be associated with its counterpart assessed with astory stem task.

One key tenet of attachment theory is that parenting strategies influence the devel-opment of a secure attachment (Ainsworth, 1989; Bowlby, 1969/1982), especially car-egiving responsiveness (Solomon & George, 2008), and studies with both young childrenand children in middle childhood reveal that attachment security is linked with higherquality parenting (i.e. sensitivity, DeWolff & Van IJzendoorn, 1997; Kerns & Brumariu,2016). In addition, secure attachment follows parenting that encourages individuation, asense of personal autonomy, and open expression and acceptance of the child (Cassidy,1994; Kerns et al., 2011). Thus, we expected that more securely attached children on theMCAS would experience higher levels of warmth/engagement (i.e. acceptance) andlower levels of psychological control.

Although the distinctive parenting correlates for the insecure attachment patterns inmiddle childhood have been less extensively investigated (see Kerns & Brumariu, 2016),ambivalence is thought to arise when children experience parental inconsistency, whichinterferes with their autonomy and trust in parental availability, whereas avoidantattachment is thought to arise due to feeling parental rejection (Cassidy, 1994;Yunger, Corby, & Perry, 2005., 2005). Thus, we hypothesized that more ambivalentlyattached children would experience higher levels of maternal psychological control, andthat more avoidantly attached children would experience a combination of both lessmaternal warmth/engagement and psychological control. Studies at both younger andolder ages showed that disorganized children experience the most challenging parent-ing such as disengagement and withdrawal (Lyons-Ruth et al., 2013; Lyons-Ruth &Jacobvitz, 2016). Therefore, we speculated that disorganized-disoriented childrenwould also experience less warmth/engagement and psychological control. To ourknowledge, no study has assessed specifically relations of maternal psychological con-trol and warmth/engagement or acceptance with caregiving/role-confused and hostile/punitive patterns of attachment. However, given that children showing caregiving/roleconfusion pattern guide the parent–child interaction and their mother show greaterwithdrawal and lack of involvement (Bureau et al., 2009; Easterbrooks, Bureau, & Lyons-

4 L. E. BRUMARIU ET AL.

Ruth, 2012; Lyons-Ruth et al., 2013), we speculated that they would perceive theirmothers as showing less psychological control. Because children with higher hostile/punitive patterns exert control over the relationship via a demeaning and aggressivebehavior, and their mothers show increased hostility (Easterbrooks et al., 2012; Lyons-Ruth et al., 2013), they may perceive their mothers as showing higher levels of psycho-logical control (that has the potential to reinforce the children’s behaviors).

Another important claim of attachment theory is that the development of a secureattachment sets the stage for healthy development rather than maladjustment (Bowlby,1969/1982, 1973), and a measure of secure attachment should predict positive achieve-ment on key developmental tasks and positive psychological adjustment (Solomon &George, 2008). Thus, additional validity criteria were indexes of peer social competenceand symptoms of psychopathology. Attachment security may foster greater exploration,engagement with social partners, and competent social skills learned in the caregiver–child interaction (Kerns, Klepac, & Cole, 1996; Sroufe, Egeland, & Carlson, 1999). There issubstantial meta-analytic evidence that more securely attached children have betterpeer relationships (e.g. Groh, Fearon, Bakermans-Kranenburg, van IJzendoorn, Steele,Roisman, 2014; Pallini, Baiocco, Schneider, Madigan, & Atkinson, 2014). A small amountof literature suggests that, in middle childhood, both avoidant and disorganized childrenshow the most peer-related problems in middle childhood, and that ambivalent childrenhave fewer peer difficulties (Granot & Mayseless, 2001; Jacobvitz & Hazen, 1999; Siebert& Kerns, 2015). We, therefore, expected that security would be associated with greaterpeer social competence, whereas avoidance and disorganization-disorientation wouldbe associated with lower competence. Controlling-punitive children showed less inter-active competence with their mothers (Teti, 1999), and given that punitive behaviorsmay also be shown outside attachment relationship, we expected that children withmore hostile/punitive attachments would show lower social competence.

Finally, meta-analyses have found that secure attachment is negatively related tolower levels of internalizing problems (including depression) and externalizing problems(Fearon, Bakermans-Kranenburg, van IJzendoorn, Lapsley, & Roisman, 2010; Groh,Roisman, van IJzendoorn, Bakermans-Kranenburg, & Fearon, 2012; Madigan, Atkinson,Laurin, & Benoit, 2013; Madigan, Brumariu, Villani, Atkinson, & Lyons-Ruth, 2016). Studiesconducted specifically on middle childhood samples parallel these meta-analytic results(Kerns & Brumariu, 2016). There is some variability across individual studies, and somefindings based on middle childhood samples suggest that ambivalent children ratherthan avoidant children are more prone to maladjustment (e.g. Dubois-Comtois, Moss,Cyr, & Pascuzzo, 2013; Kochanska & Kim, 2013; Low & Webster, 2016). Meta-analyticfindings, however, show that representational measures of ambivalent attachment (butnot behavioral measures) are related to internalizing symptoms, particularly depressivesymptoms (Groh et al., 2012; Madigan et al., 2013, 2016), and that ambivalent attach-ment does not significantly relate to externalizing problems (Fearon et al., 2010; see alsoMadigan et al., 2016). Avoidant attachment is more strongly related to internalizing/depressive symptoms (Groh et al., 2012; Madigan et al., 2013, 2016) than to externalizingbehavior (Fearon et al., 2010; see also Madigan et al., 2016). Disorganized attachment isassociated with externalizing problems (Fearon et al., 2010; Madigan et al., 2016) andinternalizing problems, when is it assessed with representational measures, with stron-ger effect sizes for depression (Fearon et al., 2010; Groh et al., 2012; Madigan et al., 2013,

ATTACHMENT & HUMAN DEVELOPMENT 5

2016). Studies with younger children have shown that controlling-punitive behavior islinked with both internalizing and externalizing behavioral problems, while providingsome support for the idea that controlling-caregiving behavior may be a risk factorprimarily for internalizing problems (Dubois-Comtois et al., 2013; Lecompte & Moss,2014; Moss, Cyr, & Dubois-Comtois, 2004; O’Connor, Bureau, McCartney, & Lyons-Ruth.,2011; Teti, 1999). However, findings are mixed in adolescence (see Macfie, Brumariu, &Lyons-Ruth, 2015 for a review).

Thus, we expected that children higher on MCAS security would show lower levelsof depressive symptoms and behavioral problems. We also hypothesized that ambiva-lence would be related primarily with depressive symptoms. Although we speculatedthat avoidance might be related to both depressive symptoms and externalizingbehavioral problems, we expected these associations to be stronger for childrenwho showed more disorientation. We further tested the hypothesis that caregiving/role-confusion might be specifically linked with depressive symptoms, and that hos-tile/punitive attachment would be linked with both depressive symptoms and exter-nalizing problems.

Method

Participants

The sample included 87 mothers and their children (39 boys and 48 girls), recruited fromlocal schools (n = 82) and mental health clinics (n = 4), with mothers reporting that 12children currently or in the past had been diagnosed with a mental health disorder.Schools and clinics distributed invitation letters describing the study, and familiesvolunteered for the study. The recruitment place for 1 participant could not be identi-fied. The mean age was 11.32 (SD = .68, range = 10.17–12.83 years), with 57% in 5thgrade and 43% in 6th grade. Approximately 67% of the children were Caucasian, 17%were African-American, 6% were American-Indian, 1% were Hispanic, and 9% werebiracial. Regarding family status, 60% of mothers reported intact-family status, with33% indicating single-parent status, and 7% reporting step-family status, roughly mirror-ing current rates of two-parent (69%) and one-parent (23%) families in the U.S. (UnitedStates Census Bureau Reports, 2016). The Hollingshead’s Four Factor Index of SocialStatus was used to determine the families’ SES (Gottfried, 1985). The potential range ofcomputed scores is 8–66. In the current sample, the mean SES level was 41.17 (SD = 12.40),with scores ranging from 12 to 66, suggesting that the sample is slightly socioeconomicallyadvantaged. Families’ SES fell in the following categories: 16% major business and profes-sionals, 37% medium business and professionals, 27% skilled workers, 13% semiskilledworkers, and 3% unskilled laborers. SES data were missing for 4% of the families.

Procedure

Children and their mothers attended a 2-h laboratory session. Mothers signed consentforms and children signed written assent forms at the beginning of the session. Thenchildren and mothers completed questionnaires separately, children completed anattachment interview, and the mother–child interaction tasks were videotaped. All the

6 L. E. BRUMARIU ET AL.

observation measures were coded with different raters. Children were compensated $20and mothers $10 for their participation.

Measures

Attachment measuresThe Middle Childhood Attachment Strategies Coding System (MCAS). Mother–childdyads were videotaped while participating in three tasks. In the warm-up segment,the pairs played a game for 3 min. Then, mother–child dyads were asked to identifythe most important problem in their relationship, then to discuss it in an 8-minconversation, including why it was a problem, feelings about the problem, andpossible solutions. A list of topics of frequent parent–child conflicts (e.g. chores,allowances, sibling issues) was provided, but participants were free to choose theirown topic. In the third segment, mothers and children had to plan, in 8 min, a 1-week-long vacation for which they had unlimited funds. This task was intended tomitigate any negative mood created during the conflict task. Only the conflict taskwas coded with the MCAS.

The present coding system follows the earlier coding systems for attachment ininfancy, preschool, and the early school years in that it codes the child’s behavior.However, similar to coding systems for earlier ages [e.g. Cassidy & Marvin, withMacArthur Attachment Working Group (1992) preschool system], it considers parentalbehavior as the context within which child behavior is examined. In assigning a score oneach scale, two criteria were taken in account: (a) the behavioral manifestation, and (b)the child’s affect in the context of the interaction. The coding system was developed byobserving 20 randomly selected parent–child interactions to assess if markers of parti-cular attachment patterns are salient at this age. The research team (composed ofexperts in attachment in middle childhood and Ph.D. students) was blind to otherinformation about the families, including knowledge of responses to other measures.The following description includes examples of aspects coded.

In a secure relationship, the child is able to use the parent as a secure base and as ahaven of safety. Thus, the child talks about the issue openly (i.e. the problem is explored,processed, and addressed collaboratively) and negotiates differences. The child is able toturn to the parent during moments of vulnerability and receive a reasonably satisfyingresponse. In addition, the child may appropriately orient his body to the parent (beha-vior). In terms of affect, the child expresses feelings, both positive and negative, clearlyand directly, in accordance with the severity of the problem, and the child’s affect isgenuine and consistent with the topic of the discussion.

An ambivalently attached child displays exaggerated attachment behaviors. Thus, thechild may attempt to express his/her viewpoints, but the problem is not addressedappropriately due to a lack of collaborative problem-solving. Instead, the child entersinto a circular, fruitless conversation with the parent and the discussion does notadvance; contradiction in logic may be present. Further, the child expresses intenseemotions and uses a heightening strategy/distressed affect, with little ability to mod-ulate affect even when an appropriate response from parent is given. The child mayhave little patience or interrupts, bickers, may blame the mother or other persons, andmay show dramatic affect.

ATTACHMENT & HUMAN DEVELOPMENT 7

An avoidantly attached child uses a minimization strategy. Thus, the child avoids thetopic of disagreement and avoids related disagreements, conflicts, or emotionallycharged topics. The child may shut down/dismiss the discussion, may discuss general-ities and impersonal topics, or may look engaged with the parent, but his/her discussionseems superficial, lacks depth, is characterized by a sense of emptiness, and may have apositive spin just to terminate the discussion. The child may appear bored or disen-gaged, may present with a limited repertoire of emotions, or may exhibit emotionaldistance, passivity, or avoidance instead of warmth and mutuality.

A child with disorganized-disoriented attachment appears to lack a consistent strategyto deal with the mother. The child may present with an odd or bizarre quality ofbehavior, may show behavioral regression, may appear confused, or may show unex-pected sequences of behavior. The child may also make unusual or inappropriateremarks or sudden, out-of-context shifts away from topic. Further, the child may beunable to articulate his/her point of view, and his/her discussion of conflict may bedysregulated and accompanied by odd or anxious motor gestures (e.g. playing with andtwisting her fingers in bizarre ways). In addition, the child may become emotionallydysregulated (e.g. child laughs or cries intensely, without a reason, out-of-context orlaughs apprehensively), or show unpredictable affect and may unexpectedly “zone out,”stilling or freezing or looking disoriented and inwardly absorbed.

A child showing caregiving/role-confusion takes on functions that are more appro-priately performed by the parent or are beyond her developmental competencies. Thechild’s caregiving behavior takes different forms, depending on the context (i.e. theparental inadequacy that the child has to manage). The child may engage in entertain-ing or overly supportive behavior (e.g. making reassuring comments), may show anagenda of organizing and managing the parent, or may be in charge of engaging theparent (e.g. may show active interest in the contributions of a parent who is inappro-priately focusing on her own needs or viewpoint). Behaviorally, the child may organizethe parent by giving directions to the parent, offering suggestions for how the parentshould behave, or acting as the parent’s confidante, and may even surrender his/herview to advance the opinion of or conciliate the parent while still engaged in discussion.Some children may engage in seductive behavior, such as exhibiting flirtatious behavioror excessive inappropriate touching that pleases the parent. In terms of affect, the childmay show silly affect that amuses the parent, overbright affect or be vigilant and highlyattuned to the state of the parent (e.g. affect congruent with mother’s opinion andemotions, even if inappropriate).

A child showing hostile/punitive behavior behaves as though he or she is in charge ofthe relationship. In the service of directing the interaction, the child displays hostile,punitive, critical, hurtful or demeaning and denigrating behaviors towards the parent.Emotions expressed by the child toward the mother, directly or indirectly, are aggres-sion, frustration, and irritability. In addition, the child seems comfortable challenging theparent inappropriately.

The MCAS coding system includes rating of each interaction on six, 9-point scales,with 1 = none or slight/isolated minor evidence of a specific pattern that does notcharacterize child’s behavior and interaction overall; 3 = slight/isolated evidence of aspecific pattern that can be clearly observed/coded, but does not primarily characterizechild’s behavior and interaction overall; 5 = clear evidence of a specific pattern that is

8 L. E. BRUMARIU ET AL.

brief, and could or could not primarily characterize child’s behavior and interactionoverall; 7 = moderate evidence of a specific pattern, which characterizes most of child’sbehavior and interaction; and 9 = marked and persistent evidence of a specific patternwhich predominantly characterizes child’s behavior and interaction

The six scales are: security, sample range 1–9, M(SD) = 3.37 (2.20) ambivalence,sample range 1–7, M(SD) = 1.85 (1.61); avoidance, sample range 1–9, M(SD) = 2.15(2.30); disorientation, sample range 1–9, M(SD) = 2.48 (2.13); caregiving/role-confusion,sample range 1–9, M(SD) = 2.64 (2.37); and hostile/punitive behavior, sample range 1–9,M(SD) = 2.06 (1.91). Two coders, blind to other information about the family, rated 67interactions (ICCs ranged between .68 and .90). The interactions were discussed and anagreement was reached (used in analyses). Rater’s 1 scores were used for the additional20 interactions used in developing the coding system.

Perceptions of mother–child attachment. Children completed the (Kerns et al., 2001),a 15-item measure designed to assess children’s perception of security in parent–childrelationships. The items were administered using Harter’s (1982) “Some kids. . .otherkids. . .” format. Children read statements about two types of kids such as “some kidsgo to their mom when they are upset, but other kids do not go to their mom when theyare upset,” decided which one is most like them, and then indicated whether thestatement was really true or sort of true for them. Each item was scored on 4-pointscale, with a higher score indicating greater security. Scores were averaged across theitems (M = 3.34, SD = .45; α = .80). Previous research has shown that the security scale issignificantly related to narrative measures of attachment, parenting behaviors, andmeasures of school adjustment and peer relationships (Kerns, Schlegelmilch, Morgan,& Abraham, 2005).

Representations of mother–child attachment. Children completed a story stem taskin which an experimenter begins a story with an attachment theme (a “problem” thatelicits attachment behavior), and the child continues the story (see Kerns et al., 2011).Using dolls and props, the child shows and tells what s/he thinks happens in two stories.The target child and his/her mother are represented in the stories. The first story has thetheme of the child working on a homework assignment that is due tomorrow, and s/heis unsure about finishing it before bed time. The second story presents the child with thesituation of playing with a friend at the friend’s house. The child and friend get into a bigfight and the friend asks the child to leave. The child gets home and slams the door. Themother, who does not see who enters the house, calls the child’s name and asks, “Is thatyou, (child’s name)?” The experimenter then prompted the child to tell and show whathappens next.

Children were rated as secure, ambivalent, avoidant, or disorganized based on fourcriteria, adapted from Granot and Mayseless (2001): the coordination of actions, expres-sion and regulation of emotions within the dyad, the coherence of the narrative, andconstructive resolutions to the problems. (Detailed information about coding is availablefrom the authors; see also Kerns et al., 2011). For example, in a more secure relationshipthere is synchronization between parent and child in solving problems, the child is ableto express openly both positive and negative emotions, the narrative is logical andbelievable, and problems that arise are successfully addressed.

ATTACHMENT & HUMAN DEVELOPMENT 9

Two trained raters, blind to other information about the mother–child dyads, inde-pendently scored the videotaped interviews, taking into consideration the child’sresponses to both stories. A child was assigned a 5-point rating for four attachmentpatterns (security, M(SD) = 2.74 (1.27), avoidance, M(SD) = 1.91 (1.12), ambivalence, M(SD) = 1.49 (.97) and disorganization, M(SD) = 1.40 (.81). Reliabilities, computed on 50participants, were ICC = .77 for security, .65 for avoidance, .92 for ambivalence and .70for disorganization. Security ratings were negatively correlated with ambivalence ratings,r = -.37, avoidance ratings, r = -.57, and disorganization ratings, r = -.52, all ps < .001.Further, ambivalence ratings were negatively correlated with avoidance ratings, r = -.35,p < .01. This attachment representation measure has good psychometric properties (seeKerns et al., 2011).

Maternal parenting variablesObserved parenting. The conflict task was also scored for two parenting strategies:warmth/engagement and psychological control. It is important to note that theMCAS coding captures the child behavior in the context of the parent–child interac-tion, whereas the observer ratings of parenting focus on maternal behavior. Fiveraters, blind to other information about mother–child dyads, conducted the coding.Raters 1 and 2 rated warmth/engagement, and Raters 3, 4 and 5 coded psychologicalcontrol.Warmth/positive affect and interest in the child. Maternal warmth/positive affect andinterest in the child was coded using Ginni, Oppenheim, and Haimovich (2002) codingsystem. The warmth scale reflects the amount and intensity of maternal verbal and non-verbal authentic positive affect expressed towards the child and ranges from 1 = low to5 = high. The interest in child scale reflects maternal attention, interest, and focus on thechild, and ranges from 1 = low interest to 5 = high interest. ICCs (n = 12) were .76 for thewarmth and positive affect scale and .54 for the interest in child scale. The warmth andinterest scales were highly correlated (r = .82, p < .001), and therefore they wereaveraged to obtain a final scale of warmth/engagement, M(SD) = 3.40(.85), α = .86).Psychological control. The conflict task was coded for psychological control usingBarber’s (1996) system. This coding system captures 6 theoretically relevant character-istics of psychological control: constraining verbal expressions, invalidating feelings,personal attack, guilt induction, love withdrawal, and erratic emotional behavior. Thelove withdrawal characteristic was not included in the current study as the codingrevealed that our conflict task did not elicit this response. Each of the 5 characteristicsof psychological control were rated using a 4-point scale, with 0 meaning not true, and 3meaning very true that the mother displayed a specific characteristic. (ICCs) among the 3raters ranged between .74 and 1.00 on each of the 5 characteristics of psychologicalcontrol. We then averaged the raters’ ratings for the 5 characteristics to create a singleindex of psychological control (M (SD) = .39(.45), α = .71).

Children’s perceptions of parental childrearing practices. The 10-item acceptancesubscale from the Revised Children’s Report of Parental Behavior Inventory (Schaefer,1965) was used to assess child perceptions of mothers’ warmth/engagement. An exam-ple item is “my mother smiles at me very often.” Children rated their mothers’ behavioron a 3-point scale, and items were averaged (α = .83; M(SD) = 2.76 (.34).

10 L. E. BRUMARIU ET AL.

Psychological control was measured with a 17-item scale consisting of 8 items fromthe psychological control subscale from the Revised Children’s Report of ParentalBehavior Inventory (Schaefer, 1965), and 9 items proposed by Barber (1996). A sampleitem is “my mother is a person who changes the subject, whenever I have something tosay.” Children rated their mothers’ behavior on a 3-point scale, and means across itemswere computed, α = .78, M(SD) = 1.43(.31). There were some associations among theparenting variables, with the highest associations between observed psychologicalcontrol and acceptance (r = –.31, p < .001), and the lowest associations between childreported and observed maternal acceptance (r = .16, p = ns).

Child adjustmentPeer social competence. Mothers completed the Peer problems and the ProsocialBehavior scales from the Strengths and Difficulties Questionnaire (SDQ, Goodman,1997), each consisting of 5 items. Examples of items are: My child is “Rather solitary,tends to play alone,” and “Considerate of other people’s feelings,” respectively. Becausethe two scales were significantly correlated and both capture social functioning, wecreated a single variable labeled peer social competence (by reversing the items on thePeer problems scale first, and then averaging the items from both scales, α = .68, M(SD) = 1.75 (.22). The SDQ distinguishes between psychiatric and non-psychiatric sam-ples, and it is highly correlated to other established measures of child adjustment(Goodman, 1997, 2001).

Child depressive symptoms. Children completed the 10-item version of the Children’sDepression Inventory (CDI, Kovacs, 1992), assessing children’s levels of depression. Thescale is designed so that children are presented with 10 sets of three responses. Thethree response options are score 0, 1, or 2. The total score was computed by averagingthe items scores, M(SD) = .1 (.19), α = .60 in this study. The CDI has demonstratedconsistent correlations with other depression scales (e.g. Doerfler, Felner, Rowlison,Raley, & Evans, 1988).

Child behavioral problems. Mother completed the 5-item Conduct Problem scale fromthe SDQ (Goodman, 1997). Each item was scored on a 3-point scale, with a score of 0indicating that the behavior depicted by the item is not true, and a score of 2 indicatingthat the behavior is certainly true for the study child, in the last 6 months. An example ofitem is: “My child often fights with other children or bullies them.” The items wereaveraged (α = .69, M(SD) = .29(.34).

Results

Preliminary analyses

The main variables were, for most part, normally distributed and mean transformationsof two variables (MCAS hostile/punitive behavior and disorganization based on theattachment interview, that had slightly higher kurtosis scores, 4.15 and 4.62), did notchange results. Therefore, data presented here are based on the original scores. Weconducted preliminary analyses to evaluate whether demographic variables (child’s age,

ATTACHMENT & HUMAN DEVELOPMENT 11

gender; family status, SES) were related to the MCAS attachment patterns. Child’s agewas not significantly related to the MCAS variables. Boys were rated more avoidant thangirls, F(1, 85) = 4.48, p < .05, Ms(SDs) = 2.72(2.80) and 1.69(1.71). Family SES wassignificantly related to MCAS security only, r = .28, p < .05. In addition, children fromstepfamilies and single mother families (n = 35) compared to children from intactfamilies (n = 52) were rated as more disorganized-disoriented, F(1, 85) = 3.85, p = .05,Ms(SDs) = 3.01(2.36) and 2.12(1.90), more caregiving/role-confused, F(1, 85) = 4.49,p < .05, Ms(SDs) = 3.29(2.61) and 2.21(2.09), and more hostile/punitive, respectively, F(1, 85) = 5.61, p < .05, Ms(SDs) = 2.63(2.27) and 1.66(1.54). The main analyses presentedbelow were repeated, controlling for any significant demographic variables relevant tothat analysis (Tables 1–3). Changes in the results were minimal (i.e. from significant tomarginally significant or vice versa, effect sizes similar in magnitude) so for ease ofinterpretation, we focus on the bivariate correlations.

Associations among the MCAS variables

As seen in Table 1, MCAS security was significantly and negatively related to all MCASinsecure attachment patterns, although most strongly with with the disorganized pat-terns. MCAS ambivalence was significantly and negatively related to MCAS avoidanceand positively related to the MCAS hostile/punitive pattern. MCAS avoidance wasnegatively associated with both MCAS caregiving/role-confusion and hostile/punitivepatterns. Finally, MCAS disorientation was positively associated with MCAS hostility/punitive behavior.

Associations of the MCAS attachment patterns with other attachment measures

Children showing more secure behavior on the MCAS reported higher security on thesecurity scale (Table 1), and there was a trend (p < .10) for them to be rated as moresecure on the story stem interview. They were also rated less disorganized on the storystem interview. MCAS ambivalence, avoidance, and disorganization-disorientation wereeach significantly related to their counterparts from the story-stem task, and not to otherpatterns. In addition, children rated as more disorganized-disoriented or more hostile/punitive with the MCAS reported less attachment security on the security Scale. MCAScaregiving/role-confusion and hostile/punitive patterns were not significantly related tochildren’s attachment representations. No measures address caregiving/role-confusionand hostile/punitive patterns explicitly, thus, these findings may not be surprising.

Associations of the MCAS attachment patterns with parenting

As shown in Table 2, MCAS security was consistently related to parenting. Children withmore secure behavior based on the MCAS ratings had mothers who were rated lower inpsychological control and were higher on warmth/engagement (i.e. acceptance), regard-less of the measurement approach to parenting (ratings of interaction from observers orchildren’s reports). To capture how much variance in MCAS security is explained by theparenting variables, we conducted a regression analysis. The results showed that, aftercontrolling for family SES, the parenting variables explained an additional 23% of

12 L. E. BRUMARIU ET AL.

Table1.

Associations

amon

gtheMCA

Sattachmentpatterns

andof

theMCA

Sattachmentpatterns

with

otherattachmentmeasures.

Variables

MCA

Ssecurity

MCA

Sam

bivalence

MCA

Savoidance

MCA

Sdisorientation

MCA

Scaregiving

/role-con

fusion

MCA

Sho

stile/

punitive

1.MCA

S(observerratin

g)MCA

Ssecurity

−.22*

−.26*

−.56***

−.30**

−.42***

MCA

Sam

bivalence

−.22*

.06

−.14

.26*

MCA

Savoidance

.04

−.28**

−.24*

MCA

Sdisorganization-disorientation

.03

.32*

MCA

Scaregiving

/role-confusion

−.11

2.Securityscale(child

repo

rt)

.22*

(.21+)

−.02

−.14

−.44***

(−.43***)

.17(.19+)

−.24*

(−.23*)

3.Attachmentrepresentatio

ns(story

stem

s)Security

.19+

(18+)

−.15

−.11(−.18+)

−.15(-.11)

.07(.11)

−.08(-.04)

Ambivalence

−.01(−.01)

.22*

−.17(−.09)

.19+

(.14)

.05(.01)

.10(.04)

Avoidance

−.15(−.17)

−.01

.23*

(.19+)

−.07(−.06)

−.18+

(−.18

+)

.05(.06)

Disorganizatio

n−.24*

(−.20+)

.03

.11(.13)

.26*

(.21*)

.03(−.04)

.06(−.02)

+p<.10;*p

<.05;**p<.01;***p

<.001.Inparenthesis:correlations

ofMCA

Scontrolling

forSES(security),sex(avoidance),andfamily

status

(intactvs.non

-intact;alldisorganized

patterns).

ATTACHMENT & HUMAN DEVELOPMENT 13

variance in MCAS security, F change (4, 82) = 4.78, p < .001, with maternal warmth/engagement and acceptance as marginally significant predictors, B (SE) = .53(.28),p = .06, and B (SE) = 1.17(.67), p = .08).

MCAS ambivalent attachment was associated with psychological control reported bythe child, but was not related to observer ratings of parenting (regression analyses withall parenting variables as predictors, R2 = .02, p = ns, B (SE) perceived psychologicalcontrol = 1.03 (.60), p = .09). The mothers of children who showed more avoidance onthe MCAS were rated as less warmly engaged and more psychologically controlling byobservers, and their children also perceived them as more psychologically controlling.Collectively, parenting explained an additional 16% of variance in MCAS avoidance, aftercontrolling for child sex, F change (4, 81) = 4.02, p < .01, and observed maternal warmth/engagement remained a unique predictor, B (SE) = –.75 (.30), p < . 05. MCAS disorga-nization was also associated with parenting. Specifically, greater disorganization-disor-ientation was related to more psychological control as reported by the child or rated byobservers, and with lower perceived parental acceptance. After controlling for familystatus, parenting explained an additional 15% of variance in MCAS disorganization-disorientation, F change (4, 81) = 3.93, p < .01, and children reported maternal controland acceptance emerged as unique predictors (B (SE) = 1.52 (.74), p < . 05 and B(SE) = −1.2 (.68), p = .08. Caregiving/role-confusion and Hostile/punitive behaviorsrated with the MCAS were not related to parenting dimensions (although there was amarginally significant association between MCAS hostile/punitive behavior and observerratings of psychological control) and respective regression analyses with parenting

Table 2. Associations of the MCAS attachment patterns with parenting.

Variables

Parentalacceptance(child report)

Warmth/engagement(observed)

Psychologicalcontrol

(child report)

Psychologicalcontrol

(observed)

MCAS security .27* (.31**) .39*** (.35***) −.20+ (-.25*) −.39*** (−.34***)MCAS ambivalence −.02 −.16 .21* −.01MCAS avoidance −.13 (−.15) −.37** (−.37**) .25* (.26*) .23* (.24*)MCAS disorganization-disorientation −.30** (−.29**) −.19+ (−.17) .34** (.32**) .26* (.24*)MCAS caregiving/role-confusion −.04 (−.02) .06 (.10) −.14 (-.19+) .03 (−.01)MCAS hostile/punitive −.12 (−.10) −.16 (−.13) .06 (.02) .20+ (.16)

+p < .10; *p < .05; **p < .01; ***p < .001. In parenthesis: correlations of MCAS controlling for SES (security), sex(avoidance), and family status (intact vs. non-intact; all disorganized patterns).

Table 3. Associations of the MCAS attachment patterns with social competence and psychologicaladjustment.

VariablesSocial competence(mother report)

Depressive symptoms(child report)

Behavioral problems(mother report)

MCASSecurity .30** (.25*) −.08 (−.04) −.40*** (−.38***)Ambivalence −.11 .30** .16Avoidance .18+ (.18) −.20+ (−.13) −.06 (−.04)Disorganization-disorientation −.15 (−.08) .26* (.22*) .34 ** (.29**)Caregiving/role-confusion −.17 (−.10) −.11 (−.19+) .18+ (.11)Hostile/punitive −.44***(−.39***) .24* (.19+) .31** (.23*)

+p < .10; *p < .05; **p < .01; ***p < .001. In parenthesis: correlations of MCAS controlling for SES (security), sex(avoidance), and family status (intact vs. non-intact; all disorganized patterns).

14 L. E. BRUMARIU ET AL.

variables as predictors, controlling for family status, were not significant, adjustedR2s = .04 (no significant predictors).

Associations of the MCAS patterns with peer social competence and childadjustment

As hypothesized (see Table 3), more securely attached children were rated by theirmothers as showing better peer social competence and less behavioral difficulties. Moreambivalently attached children reported higher levels of depressive symptoms, whereasavoidance was not related to child adjustment. Children rated higher on disorganization-disorientation also reported more depression, and were seen by their mother as show-ing more behavioral problems. There were no significant associations between MCAScaregiving/role-confusion and adjustment, although children rated as more hostile/punitive were less socially competent and displaying more behavioral problems asreported by their mothers, and they self-reported more depressive symptoms.

Discussion

The first important finding of this study was that all organizations of attachmentbehavior described at younger and older ages can be reliably rated in later middlechildhood based on an 8-min interaction with a caregiver that includes a discussion of atopic of conflict. While questionnaires and interviews have provided valuable insightsinto attachment in middle childhood, identifying key behavioral features of attachmentincreases our understanding of attachment at this age (Bosmans & Kerns, 2015).

Importantly, the MCAS scores were associated with each other in expected ways, withthe most consistent associations emerging for MCAS security. Specifically, as expectedbased on theory, MCAS security was negatively and significantly related to all otherinsecure MCAS patterns. In line with other work (e.g. Kerns et al., 2011), MCAS ambiva-lence and avoidance were negatively related. Although not a focus of the current study,it should be also noted that boys were rated more avoidant than girls, providing somesupport for Del Giudice’s (2009) hypothesis that sex differences in attachment arise inmiddle childhood. Given that evidence for sex differences is mixed (Bakersman-Kranenburg & van IJzendoorn, 2009; Del Giudice, 2009), the question of potential sexdifferences in MCAS should be further evaluated. Among the distinct forms of disorga-nized behavior, consistent with findings at younger ages (Bureau et al., 2009), MCASdisorganized-disoriented behavior and hostile/punitive behavior were moderately andpositively associated, suggesting the importance of capturing these behaviors dimen-sionally rather than relying on mutually exclusive categories. All three aspects ofdisorganized behavior showed some distinct associations with all other organizedMCAS dimensions. For example, of the three markers of disorganization, all werenegatively related to security, hostile-punitive and role-confused behaviors were relatednegatively to avoidance, and only hostile/punitive behaviors was related to ambivalence.

The second important set of findings from the study provided support for thevalidity of the newly developed scales. The MCAS dimensions converged with twomeasures of attachment in middle childhood: the security Scale and the story steminterview. MCAS security significantly but modestly overlapped with the two others

ATTACHMENT & HUMAN DEVELOPMENT 15

assessments of security. MCAS security was also associated with lower levels ofdisorganization based on the story stem interview. Further, MCAS avoidance, ambiva-lence, and disorganization-disorientation were significantly and modestly related totheir corresponding attachment dimensions on the story stem interview. Overall,findings might suggest that the more extreme patterns are easier to capture with anobservational coding measure. These findings are consistent with other studies wheremultiple measurement approaches of attachment in middle childhood were includedand compared (e.g. Di Folco, Messina, Zavattini, & Psouni, 2017; Kerns, Mathews,Koehn, Williams, & Siener-Ciesla, 2015; Psouni & Apetroaia, 2014). The current patternof findings is also in line with the moderate effects sizes found in the attachmentobservation study by Boldt et al. (2016), even though they used a different observa-tional strategy and coding system. Similarly, although not discussing specificallyobservation based measures, Psouni and Apetroaia (2014) argued that differenttypes of attachment measures are only modestly associated because they each capturespecific aspects of the attachment representations (e.g. questionnaires vs. interviews;specific and explicit vs. global and implicit representations of relationships). Bosmansand Kerns (2015) proposed that these different aspects reflect variation in the extent towhich attachment measurement approaches capture strategic versus automatic pro-cesses, and research confirms that measures that differ with regard to the level ofautomaticity correlate less (Moors, 2016). Consequently, because also in the currentstudy the assessment methods measure different levels of automaticity, these modestassociations are in line with theoretical expectations (see Bosmans & Kerns, 2015). AsBosmans and Kerns (2015) pointed out, each measure that incorporates the securebase construct may capture unique variance and each might provide unique informa-tion regarding mechanisms of the attachment system.

To further evaluate the construct validity of the MCAS, we investigated its relationswith parenting. MCAS security scores were related to higher maternal warmth/engage-ment (i.e. acceptance) and lower levels of psychological control, assessed with childquestionnaires or observer coding. A similar set of correlates have been reported inprevious work using questionnaires or interviews of attachment (Kerns & Brumariu,2016). Interestingly, when controlling for SES and the other parenting variables, bothself-reported and observed maternal acceptance remained relevant predictors, furtherconfirming the key role of supportive parenting for attachment security (e.g. Kerns et al.,2011). As expected, more ambivalently attached children perceived their mothers ashigher on psychological control, and this association was weaker when controlling forthe other parenting variables. More avoidantly attached children on the MCAS experi-enced less psychological control, regardless of the measurement approach, and lessobserved maternal warmth/engagement. Notably, observed maternal warmth/engage-ment remained a robust predictor of avoidance when the effects of all parentingvariables where controlled for. Our findings are in line with the available yet limiteddata on insecure-organized attachment and parenting at this age (e.g. Boldt et al., 2016;Yunger et al., 2005; see Kerns & Brumariu, 2016), and suggest that there could be somedistinctive associations for avoidance and ambivalence, with avoidance being uniquelylinked to low parental warmth.

Among the three forms of disorganization, only MCAS disorganization-disorientationshowed a pattern of associations with the parenting measures. Children rated higher on

16 L. E. BRUMARIU ET AL.

disorganization-disorientation reported lower maternal acceptance and higher psycho-logical control, and their mothers were rated by observers as higher in psychologicalcontrol. The findings regarding children reported maternal control and acceptanceremained relatively robust in regression analyses when the effect of all parentingvariables was assessed and controlling for family status type. Thus, children higher ondisorientation perceived their mothers negatively and experienced more maternalunpredictable emotional behavior, personal attacks or guilt induction. Our findings areconsistent with meta-analytic findings showing that infant D attachment is linked withmaternal disrupted communication, marked by negative-intrusive behaviors, contradic-tory communication, frightened/disoriented behavior, and withdrawal (Madigan et al.,2006). Interestingly, whereas disorganized-disoriented behavior and the organizedsecure and insecure patterns were strikingly associated with parenting, no associationsemerged for MCAS hostile/punitive and caregiving/role-confused dimensions, suggest-ing that middle childhood parenting strategies of parents of these children may not becaptured in traditional parenting constructs such as acceptance and control. Whereasstudies of attachment in late middle childhood have not assessed these attachmentdimensions due to lack of measures, so there is little data for comparison, it is possiblethat aspects of parenting not evaluated in this study, such as parental helplessness,passive/withdrawn behavior, punitive behavior or lack of competence and reliance onthe child for instrumental and emotional support may be key parental behaviors signal-ing the child’s role-confusion (Bureau et al., 2009; Easterbrooks et al., 2012; Obsuth et al.,2014). The parenting rearing practices nonetheless did explain a significant variance inthe organized and disorganized-disoriented attachment behaviors, and this study couldbe extended by examining associations of the MCAS with other parenting practices inmiddle childhood (e.g. encouraging independent decision-making, mastery, safebehavior).

Finally, we evaluated the relations of the MCAS with measures of child adjustment.Consistent with previous studies (Pallini et al., 2014; Madigan et al., 2016), childrenrated more secure showed greater social competence and fewer externalizing beha-vioral problems, although contrary to our expectation, security was not associatedwith depressive symptoms. In line with representational studies of attachment(Brumariu & Kerns, 2010; Madigan et al., 2016), MCAS ambivalence was related tochildren’s depression. However, children rated as more avoidant did not reportgreater depression and their mothers did not perceive them as less socially compe-tent or behaviorally challenging, in spite of speculations that their self-reliance andlack of a responsive attachment relationship may predispose them to maladjustment(Carlson & Sroufe, 1995). The latter result also parallels meta-analytic findings basedon representational measures of attachment (Madigan et al., 2016). Meta-analyticfindings generally indicate that the organized forms of insecurity (ambivalence,avoidance) are not as maladaptive as disorganized patterns (Fearon et al., 2010;Madigan et al., 2016; Van IJzendoorn, Schuengel, & Bakermans-Kranenburg, 1999).Thus, they may entail risk for maladaptation only in conjunction with other riskfactors. For example, given that some studies found a link between ambivalenceand externalizing problems (Dubois-Comtois et al., 2013; Kochanska & Kim, 2013; Low& Webster, 2016), it would be important to explore under which conditions thisassociation is found. The findings that MCAS disorientation showed significant

ATTACHMENT & HUMAN DEVELOPMENT 17

relations to both depression and behavioral problems are in line with this interpreta-tion and with a plethora of data indicating that disorganized attachment is a moreproblematic form of attachment insecurity (e.g. Brumariu & Kerns, 2010; Fearon et al.,2010; Madigan et al., 2016; Van IJzendoorn et al., 1999).

Notably, MCAS hostile/punitive behavior was associated with poorer social compe-tence, greater depression, and higher levels of externalizing behavioral problems. Thesefindings parallel those based on younger and older ages described earlier (Lecompte &Moss, 2014; Moss et al., 2004; O’Connor et al., 2011; Obsuth et al., 2014; Teti, 1999) andextend them into middle childhood. Finally, MCAS caregiving/role-confused behaviorwas not significantly related to children’s depression (or other adjustment indexes,including those reported by mothers). Whether this represents idealizing tendenciesfor the child is unknown due to the lack of data. It is also possible that the effects of role-confusion become stronger in high risk environments. Since Solomon and George (2008)reported that caregiving children may be idealized by their mothers and seen as “largerthan life,” it is possible that mothers in these dyads are gratified by the child’s adapta-tion and do not see it as problematic.

Several limitations of this study should be taken into account. Directions of causalitycannot be inferred as our study is cross-sectional. Longitudinal studies and interventiondesigns are necessary to sort out potential bi-directional effects of attachment withparenting and child adjustment. While our measurement approach of including mater-nal, child, and observer measures limited the effects of shared-method variance, includ-ing diagnostic interviews of psychopathology would extend this work. We only assessedchildren’s attachment behaviors with mothers. As fathers play a significant role insupporting children’s navigation of the social world (Grossmann, Grossmann, Kindler,& Zimmerman, 2008), assessing behavioral manifestations of attachment with fatherswould add to our understanding of attachment processes in middle childhood. It isimportant to consider whether findings of this study (e.g. that boys were more avoidantthan girls) might be a result of focusing on mother–child relationships. There is clearly aneed for more studies on father–child attachment to understand whether father–childand mother–child attachment are similar in their influence in middle childhood (e.g. DiFolco et al., 2017).

The MCAS showed promise as a behavioral measure of attachment that can beused with children 9–12 years of age. Nevertheless, additional work is needed tomore fully validate the measure, including examining how it is related to autobio-graphical interview measures of attachment such as the FFI (Kriss et al., 2012) orChild Attachment Interview (Schmueli-Goetz, Target, Fonagy, & Datta, 2008) or toaspects of child functioning (e.g. emotion regulation, Kerns & Brumariu, 2016; self-image, Psouni, Di Folco, & Zavattini, 2015), not included in this study. Although wedid use the full range for six scales, the mean for the security rating was lower thanexpected. This might be because the scoring criteria were conservative, or it mightbe because of the nature of the sample (55% of children scored high on a ques-tionnaire measure of anxiety, Brumariu & Kerns, 2015). Further, this is the first studythat attempted to capture caregiving/role-confused and hostile/punitive aspects ofdisorganized behavior in older middle childhood, and the coding system may requireadjustments. Nonetheless, the hostile-punitive profile was associated with the high-est level of child problems in the study, as reported by child and mother. This is in

18 L. E. BRUMARIU ET AL.

line with the expected maladaptive nature of this behavioral pattern. Also, childrenshowing caregiving/role-confused profiles were rated low on security and also lowon avoidance, which is expected, given the degree of involvement of the child inmanaging the interaction with the parent. Both sets of findings suggest that thecurrent observation and coding system did reveal theoretically meaningful patternsof middle childhood disorganized attachment behavior. However, the results dopoint to the need for further work on the correlates and establishing meaningfulindicators of middle childhood caregiving/role-confusion and to some extent, of thehostile/punitive pattern, as there were fewer findings for these MCAS dimensions,particularly in relation to other attachment measurement approaches and parenting.For example, it is important to include teacher and peer ratings in the study of thesepatterns as well as to assess abdication of parental role (e.g. ratings for parentalwithdrawal and failure to play an appropriate parental structuring role in the inter-action, Obsuth et al., 2014).

Equally important would be assessing how the MCAS converges with and comple-ments the Iowa Attachment Behavior (IAB) measure (Boldt et al., 2016), another recentlydeveloped behavioral measure of attachment patterns. Both the MCAS and the IAB haveshown associations with other attachment measures, parenting, and child adjustment.The MCAS may have greater utility as part of a larger study given the shorter timedemand, and it also allows for distinguishing among three forms of disorganizedattachment. An empirical study comparing the two systems, both their convergenceas well as the predictive validity of each, would be important for more decisivelyevaluating the relative merits of each.

In summary, this study extended the developmental literature by providing initialvalidation for an observational measure of attachment in middle childhood designedto assess all attachment patterns described in the literature. Thus, the MCAS opensnew grounds for assessing continuity and change of attachment manifestationsacross childhood and for expanding our understanding of attachment and its out-comes in middle childhood.

Acknowledgment

This research was supported by a grant from the Fund for Psychoanalytic Research, AmericanPsychoanalytic Association, awarded to Laura E. Brumariu and Karlen Lyons-Ruth. We are particu-larly indebted to the families who have given their time to participate in the study. For traininginformation regarding the use and scoring of the Middle Childhood Attachment Strategies CodingSystem, please contact Laura E. Brumariu at lbrumariu@adelphi.edu

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This research was supported by a grant from the Fund for Psychoanalytic Research, AmericanPsychoanalytic Association, awarded to Laura E. Brumariu and Karlen Lyons-Ruth.

ATTACHMENT & HUMAN DEVELOPMENT 19

References

Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: Apsychologicalstudy of the strange situation. Hillsdale, NJ: Lawrence Erlbaum.

Ainsworth, M. S. (1989). Attachments beyond infancy. American Psychologist, 44(4), 709.Allen, J. P., Hauser, S. T., O’Connor, T. G., & Bell, K. L. (2002). Prediction of peer-rated adult hostility

from autonomy struggles in adolescent–Family interactions. Development and Psychopathology,14, 123–137.

Bakermans-Kranenburg, M. J. (2009). No reliable gender differences in attachment across thelifespan. Behavioral and Brain Sciences, 32(1), 22–23.

Barber, B. K. (1996). Parental psychological control: Revisiting a neglected construct. ChildDevelopment, 67, 3296–3319.

Boldt, L. J., Kochanska, G., Grekin, R., & Brock, R. L. (2016). Attachment in middle childhood:Predictors, correlates, and implications for adaptation. Attachment & Human Development, 18(2), 115–140.

Bosmans, G., Braet, C., Heylen, J., & DeRaedt, R. (2015). Children’s attentional processing of motherand proximity seeking. Plos One, 10(4), e0124038.

Bosmans, G., & Kerns, K. A. (Eds.). (2015). Attachment in middle childhood: Progress and prospects.New Directions for Child and Adolescent Development, 148, 1–14.

Bowlby, J. (1969/1982). Attachment and loss: Attachment (Vol. 1). New York, NY: Basic Books.Bowlby, J. (1973). Attachment and loss: Separation (Vol. 2). New York, NY: Basic Books.Brenning, K. M., Soenens, B., Braet, C., & Bosmans, G. (2012). Attachment and depressive symptoms

in middle childhood and early adolescence: Testing the validity of the emotion regulationmodel of attachment. Personal Relationships, 19(3), 445–464.

Bretherton, I., Ridgeway, D., & Cassidy, J. (1990). Assessing internal working models of the attach-ment relationship: An attachment story completion task for 3-year-olds. In M. T. Greenberg, D.Cicchetti, & E. M. Cummings (Eds.), Attachment in the preschool years: Theory, research, andintervention (pp. 273–308). Chicago, IL: University of Chicago Press.

Brumariu, L. E., & Kerns, K. A. (2010). Parent–Child attachment and internalizing symptoms inchildhood and adolescence: A review of empirical findings and future directions. Developmentand Psychopathology, 22, 177–203.

Brumariu, L. E., & Kerns, K. A. (2015). Mother-child emotion communication and childhood anxietysymptoms. Cognition and Emotions, 29, 416–431.

Brumariu, L. E., Kerns, K. A., Bureau., J.-F., & Lyons-Ruth, K. (2014). Middle Childhood AttachmentStrategies Coding System (Version 2). Unpublished Manual. Adelphi University/Harvard MedicalSchool.

Bureau, J. F., Easterbrooks, M. A., & Lyons-Ruth, K. (2009). Attachment disorganization and con-trolling behavior in middle childhood: Maternal and child precursors and correlates. Attachment& Human Development, 11, 265–284.

Carlson, E. A., & Sroufe, L. A. (1995). Developmental psychopathology: Vol. 1. Contribution ofattachment theory to developmental psychopathology. Wiley: Oxford, England.

Cassidy, J. (1994). Emotion regulation: Influences of attachment relationships. Monographs of theSociety for Research in Child Development, 59(2/3), 228–249.

Cassidy, J., & Marvin, R. S., with McArthur Attachment Working Group. (1992). Attachment organi-zation in 2½ to 4½ year olds: Coding manual (4th ed.). Charlottesville: Unpublished codingmanual, University of Virginia.

Del Giudice, M. (2009). Sex, attachment, and the development of reproductive strategies.Behavioral and Brain Sciences, 32(1), 1–21.

DeWolff, M. S., & Van IJzendoorn, M. H. (1997). Sensitivity and attachment: A meta-analysis onparental antecedents of infant attachment. Child Development, 68(4), 571–591.

Di Folco, S., Messina, S., Zavattini, G. C., & Psouni, E. (2017). Attachment to mother and father attransition to middle childhood. Journal of Child and Family Studies, 26(3), 721–733.

20 L. E. BRUMARIU ET AL.

Doerfler, L. A., Felner, R. D., Rowlison, R. T., Raley, P. A., & Evans, E. (1988). Depression in childrenand adolescents: A comparative analysis of the utility and construct validity of two assessmentmeasures. Journal of Consulting and Clinical Psychology, 56(5), 769.

Dubois-Comtois, K., Moss, E., Cyr, C., & Pascuzzo, K. (2013). Behavior problems in middle childhood:The predictive of maternal distress, child attachment, and mother-child interactions. Journal ofAbnormal Child Psychology, 41, 1311–1324.

Easterbrooks, M. A., Bureau, J. F., & Lyons-Ruth, K. (2012). Developmental correlates and predictorsof emotional availability in mother–child interaction: A longitudinal study from infancy tomiddle childhood. Development and Psychopathology, 24(1), 65–78.

Fearon, R. P., Bakermans-Kranenburg, M. J., Van IJzendoorn, M. H., Lapsley, A. M., & Roisman, G. I.(2010). The significance of insecure attachment and disorganization in the development ofchildren’s externalizing behavior: A meta-analytic study. Child Development, 81(2), 435–456.

Ginni, M., Oppenheim, D., & Haimovich, T. (2002). The Joint Story Telling (JST) task scales(Unpublished manual). Department of Psychology, University of Haifa, Israel.

Goodman, R. (1997). The strengths and difficulties Questionnaire: A research note. Journal of ChildPsychology and Psychiatry, 38(5), 581–586.

Goodman, R. (2001). Psychometric properties of the strengths and difficulties questionnaire.Journal of the American Academy of Child & Adolescent Psychiatry, 40(11), 1337–1345.

Gottfried, A. W. (1985). Measures of socioeconomic status in child development research: Data andrecommendations. Merrill-Palmer Quarterly, 31, 85–92.

Granot, D., & Mayseless, O. (2001). Attachment security and adjustment to school in middlechildhood. International Journal of Behavioral Development, 25(6), 530–541.

Groh, A. M., Fearon, R., Bakermans-Kranenburg, M. J., Van IJzendoorn, M. H., Steele, R. D., &Roisman, G. I. (2014). The significance of attachment security for children’s social competencewith peers: A meta-analytic study. Attachment and Human Development, 16, 103–136.

Groh, A. M., Roisman, G. I., Van IJzendoorn, M. H., Bakermans-Kranenburg, M. J., & Fearon, R. (2012).The significance of insecure and disorganized attachment for children’s internalizing symptoms:A meta-analytic study. Child Development, 83(2), 591–610.

Grossmann, K., Grossmann, K. E., Kindler, H., & Zimmermann, P. (2008). A wider view of attachment andexploration: The influence of mothers and fathers on the development of psychological securityfrom infancy to young adulthood. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory,research, and clinical applications (2nd ed., pp. 857–879). New York: Guilford Press.

Harter, S. (1982). The perceived competence scale for children. Child Development, 53, 87–97.Jacobvitz, D., & Hazen, N. (1999). Developmental pathways from infant disorganization to child-

hood peer relationships. In J. Solomon & C. George (Eds.), Attachment disorganization (pp. 127–159). New York, NY: Guilford Press.

Kerns, K. A., Aspelmeier, J. E., Gentzler, A. L., & Grabill, C. M. (2001). Parent–child attachment andmonitoring in middle childhood. Journal of Family Psychology, 15(1), 69.

Kerns, K. A., & Brumariu, L. E. (2016). Attachment in middle childhood. In J. Cassidy (Ed.), Handbookof attachment: Theory, research, and clinical applications (pp. 349–365). New York, NY: GuilfordPress.

Kerns, K. A., Brumariu, L. E., & Seibert, A. (2011). Multi-method assessment of mother-childattachment: Links to parenting and child depressive symptoms in middle childhood.Attachment & Human Development, 13(4), 315–333.

Kerns, K. A. (2008). Attachment in middle childhood. In J. Cassidy & P. R. Shaver (Eds.), Handbook ofattachment: Theory, research, and clinical applications (2nd ed., pp. 366–382). New York, NY:Guilford Press.

Kerns, K. A., Klepac, L., & Cole, A. (1996). Peer relationships and preadolescents’ perceptions ofsecurity in the child-mother relationship. Developmental Psychology, 32(3), 457.

Kerns, K. A., Mathews, B. L., Koehn, J. A., Williams, C. T., & Siener-Ciesla, S. (2015). Assessing bothsafe haven and secure base support in parent–child relationships. Attachment & HumanDevelopment, 17, 337–353.

ATTACHMENT & HUMAN DEVELOPMENT 21

Kerns, K. A., Schlegelmilch, A., Morgan, T. A., & Abraham, M. M. (2005). Assessing attachment inmiddle childhood. In K. A. Kerns & R. A. Richardson (Eds.), Attachment in Middle Childhood (pp.46–70). New York: Guilford Press.

Kochanska, G., & Kim, S. (2013). Early attachment organization with both parents and futurebehavior problems: From infancy to middle childhood. Child Development, 84(1), 283–296.

Kovacs, M. (1992). Children’s Depression Inventory (CDI). Toronto: Multi Health Systems.Kriss, A., Steele, H., & Steele, M. (2012). Measuring attachment and reflective functioning in early

adolescence: An introduction to the Friends and Family Interview. Research in Psychotherapy:Psychopathology, Process and Outcome, 15(2), 87–95.

Lecompte, V., & Moss, E. (2014). Disorganized and controlling patterns of attachment, role reversal,and caregiving helplessness: Links to adolescents’ externalizing problems. American Journal ofOrthopsychiatry, 84(5), 581–589.

Low, J. A., & Webster, L. (2016). Attention and executive functions as mediators of attachment andbehavior problems. Social Development, 25, 646–664.

Lyons-Ruth, K., Bureau, J. F., Easterbrooks, M. A., Obsuth, I., Hennighausen, K., & Vulliez-Coady, L.(2013). Parsing the construct of maternal insensitivity: Distinct longitudinal pathways associatedwith early maternal withdrawal. Attachment & Human Development, 15(5–6), 562–582.

Lyons-Ruth, K., & Jacobvitz, D. (2016). Attachment disorganization from infancy to adulthood:Neurological correlates, parenting contexts, and pathways to disorder. In J. Cassidy & P. Shaver(Eds), Handbook of attachment: Theory, research, and clinical applications (pp. 667–695). NewYork, NY: Guilford Press.

Macfie, J., Brumariu, L. E., & Lyons-Ruth, K. (2015). Parent–child role-confusion: A critical review ofan emerging concept. Developmental Review, 36, 34–57.

Madigan, S., Atkinson, L., Laurin, K., & Benoit, D. (2013). Attachment and internalizing behavior inearly childhood: A meta-analysis. Developmental Psychology, 49(4), 672.

Madigan, S., Bakermans-Kranenburg, M. J., Van IJzendoorn, M. H., Moran, G., Pedersen, D., & Benoit,D. (2006). Unresolved states of mind, anomalous parental behavior, and disorganized attach-ment: A review and meta-analysis of a transmission gap. Attachment & Human Development, 8,89–111.

Madigan, S., Brumariu, L. E., Villani, V., Atkinson, L., & Lyons-Ruth, K. (2016). Representational andquestionnaire measures of attachment: A meta-analysis of relations to child internalizing andexternalizing problems. Psychological Bulletin, 142(4), 367–399.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move tothe level of representation. Monographs of the Society for Research in Child Development, 50(1 &2), 66–104.

Main, M., & Solomon, J. (1990). Procedures for identifying infants as disorganized/disorientedduring the ainsworth strange situation. In M. T. Greenberg, D. Cicchetti, & E. M. Cummings(Eds), Attachment in the preschool years: Theory, research, and intervention (pp. 121–160).Chicago, IL: University of Chicago Press.

Moors, A. (2016). Automaticity: Componential, causal, and mechanistic explanations. AnnualReview of Psychology, 67, 263–287.

Moss, E., Bureau, J.-F., St. Laurent, D., & Tarabulsky, G. M. (2011). Understanding disorganizedattachment at preschool and school age: Examining divergent pathways of disorganized andcontrolling children. In J. Solomon & C. George (Eds.), Disorganized attachment and caregiving(2nd ed., pp. 52–79). New York, NY: Guilford Press.

Moss, E., Cyr, C., & Dubois-Comtois, K. (2004). Attachment at early school age and developmentalrisk: Examining family context and behavior problems of controlling- caregiving, controlling-punitive, and behaviorally disorganized children. Developmental Psychology, 40(4), 519–532.

O’Connor, E., Bureau, J. F., McCartney, K., & Lyons-Ruth, K. (2011). Risks and outcomes associatedwith disorganized/controlling patterns of attachment at age three years in the national instituteof child health & human development study of early child care and youth development. InfantMental Health Journal, 32(4), 450–472.

22 L. E. BRUMARIU ET AL.

Obsuth, I., Hennighausen, K., Brumariu, L. E., & Lyons-Ruth, K. (2014). Disorganized behavior inadolescent–parent interaction: relations to attachment state of mind, partner abuse, andpsychopathology. Child Development, 85(1), 370–387.

Pallini, S., Baiocco, R., Schneider, B. H., Madigan, S., & Atkinson, L. (2014). Early child–parentattachment and peer relations: A meta-analysis of recent research. Journal of FamilyPsychology, 28(1), 118–123.

Psouni, E., & Apetroaia, A. (2014). Measuring scripted attachment-related knowledge in middlechildhood: The Secure Base Script Test. Attachment & Human Development, 16, 22–41.

Psouni, E., Di Folco, S., & Zavattini, J.-C. (2015). Scripted secure base knowledge and its relation toperceived social acceptance and competence in early middle childhood. Scandinavian Journal ofPsychology, 56(3), 341–348.

Roisman, G. I., Fraley, R. C., & Belsky, J. (2007). A taxometric study of the adult attachmentinterview. Developmental Psychology, 43(3), 675.

Schaefer, E. S. (1965). Children’s reports of parental behavior: An inventory. Child Development, 36(2), 413–424.

Shmueli-Goetz, Y., Target, M., Fonagy, P., & Datta, A. (2008). The child attachment interview: Apsychometric study of reliability and discriminant validity. Developmental Psychology, 44(4),939–956.

Siebert, A., & Kerns, K. A. (2015). Early mother-child attachment: Longitudinal prediction to thequality of peer relationships in middle childhood. International Journal of BehavioralDevelopment, 39, 130–138.

Solomon, J., & George, C. (2008). The measurement of attachment security and related con-structs in infancy and early childhood. In J. Cassidy & P. R. Shaver (Eds.), Handbook ofattachment: Theory, research, and clinical applications (2nd ed., pp. 383–416). New York, NY:Guilford Press.

Sroufe, L. A., Egeland, B., & Carlson, E. A. (1999). One social world: The integrated development ofparent–child and peer relationships. In W. A. Collins & B. Laursen (Eds.), Relationships asdevelopmental contexts (pp. 241–261). Hillsdale: Erlbaum.

Teti, D. M. (1999). Conceptualizations of disorganization in the preschool years. In J. Solomon & C.George (Eds.), Attachment disorganization (pp. 213–242). New York: Guilford Press.

United States Census Bureau Reports (2016, November 17). The majority of children live with twoparents. Article retrieved from https://www.census.gov/newsroom/press-releases/2016/cb16-192.html

Van IJzendoorn, M. H., Schuengel, C., & Bakermans-Kranenburg, M. (1999). Disorganized attach-ment in early childhood: Meta-analysis of precursors, concomitants, and sequelae. Developmentand Psychopathology, 11, 225–250.

Waters, T. E. A., Bosmans, G., Vandevivere, E., Dujardin, A., & Waters, H. S. (2015). Secure baserepresentations in middle childhood across two western cultures: Associations with parentalattachment representations and maternal reports of behavior problems. DevelopmentalPsychology, 51, 1013–1025.

Yunger, J. L., Corby, B. C., & Perry, D. G. (2005). Dimensions of attachment in middle childhood. InK. A. Kerns & R. A. Richardson (Eds.), Attachment in middle childhood (pp. 89–114). New York:Guilford Press.

ATTACHMENT & HUMAN DEVELOPMENT 23