18
Stability of attachment representations during adolescence: the influence of ego-identity status PETER ZIMMERMANN AND FABIENNE BECKER-STOLL The purpose of this study was to examine two core assumptions of attachment theory. First, internal working models of attachment should increase in stability during the course of development. Second, attachment is related to the adaptive solution of stage-salient issues, in adolescence, specifically to identity formation. Adolescence as a developmental period of change and reorganization might lead to changes in attachment representations by re-evaluating one’s life experiences in the course of general identity development. In a longitudinal study, the stability of attachment representations (assessed by the Adult Attachment Interview) between ages 16 and 18 was examined. In addition, the concurrent and longitudinal associations of identity status at age 16 with attachment representations at ages 16 and 18 were tested. Results showed high stability of attachment representation during that age period. Secure attachment representation was positively associated with the identity status achievement, whereas a dismissing status was positively associated with identity diffusion. Longitudinally, identity diffusion at age 16 predicted attachment representation at age 18 even after controlling for earlier attachment representation. r 2002 The Association for Professionals in Services for Adolescents. Published by Elsevier Science Ltd. All rights reserved Introduction Attachment theory is a life-span theory, which proposes that caregiving experiences from infancy on have an important influence on the development of a resilient personality (Bowlby, 1973). It describes conditions that promote or disturb the development of close relationships, and emphasizes the consequences of attachment experiences on the development of adaptive capacities or disturbances in emotional and personality development. The attachment behavior system is a biologically based security-regulation system (Bowlby, 1980) leading an individual to seek the proximity of confidants in order to reassure him or herself and to seek security and comfort in times of distress (Ainsworth, 1990). Thus, attachment behavior is activated when a person feels, e.g. frightened, sad, or ill and aims at seeking or maintaining proximity of a familiar person (Bowlby, 1969) who is perceived as helpful to cope with the situation (Ainsworth, 1990). The attachment figures ideally assist the child during emotional distress as external organizers (Grossmann and Grossmann, 1991; Cassidy, 1994) and influence the development of their children’s emotion regulation strategies. Experiences of the attachment figures’ emotional support and availability influence children’s development of an adaptive emotion regulation pattern (Bowlby, 1973; Cassidy, 1994), which is a major characteristic of a resilient personality (Eisenberg, Fabes, Guthrie, and Reiser, 2000), and which contributes to overcome adversities or risk factors also later in life (Sroufe, Carlson, Levy and Egeland, 1999). Reprint requests and correspondence to Peter Zimmermann, Institute of Psychology, University of Erlangen, 91054 Erlangen, Germany. (E-mail: [email protected]) 0140-1971/02/$35Á00 # 2002 The Association for Professionals in Services for Adolescents. Published by Elsevier Science Ltd. All rights reserved Journal of Adolescence 2002, 25, 107–124 doi:10.1006/jado.2001.0452, available online at http://www.idealibrary.com on

Stability of attachment representations during adolescence: the influence of ego-identity status

Embed Size (px)

Citation preview

Page 1: Stability of attachment representations during adolescence: the influence of ego-identity status

Journal of Adolescence 2002, 25, 107–124doi:10.1006/jado.2001.0452, available online at http://www.idealibrary.com on

Stability of attachment representations duringadolescence: the influence of ego-identity status

PETER ZIMMERMANN AND FABIENNE BECKER-STOLL

The purpose of this study was to examine two core assumptions of attachment theory.First, internal working models of attachment should increase in stability during thecourse of development. Second, attachment is related to the adaptive solution ofstage-salient issues, in adolescence, specifically to identity formation. Adolescence as adevelopmental period of change and reorganization might lead to changes inattachment representations by re-evaluating one’s life experiences in the course ofgeneral identity development. In a longitudinal study, the stability of attachmentrepresentations (assessed by the Adult Attachment Interview) between ages 16 and 18was examined. In addition, the concurrent and longitudinal associations of identitystatus at age 16 with attachment representations at ages 16 and 18 were tested. Resultsshowed high stability of attachment representation during that age period. Secureattachment representation was positively associated with the identity statusachievement, whereas a dismissing status was positively associated with identitydiffusion. Longitudinally, identity diffusion at age 16 predicted attachmentrepresentation at age 18 even after controlling for earlier attachment representation.

r 2002 The Association for Professionals in Services for Adolescents.

Published by Elsevier Science Ltd. All rights reserved

Introduction

Attachment theory is a life-span theory, which proposes that caregiving experiences frominfancy on have an important influence on the development of a resilient personality(Bowlby, 1973). It describes conditions that promote or disturb the development of closerelationships, and emphasizes the consequences of attachment experiences on thedevelopment of adaptive capacities or disturbances in emotional and personalitydevelopment.

The attachment behavior system is a biologically based security-regulation system (Bowlby,1980) leading an individual to seek the proximity of confidants in order to reassure him orherself and to seek security and comfort in times of distress (Ainsworth, 1990). Thus,attachment behavior is activated when a person feels, e.g. frightened, sad, or ill and aims atseeking or maintaining proximity of a familiar person (Bowlby, 1969) who is perceived ashelpful to cope with the situation (Ainsworth, 1990). The attachment figures ideally assistthe child during emotional distress as external organizers (Grossmann and Grossmann, 1991;Cassidy, 1994) and influence the development of their children’s emotion regulationstrategies. Experiences of the attachment figures’ emotional support and availabilityinfluence children’s development of an adaptive emotion regulation pattern (Bowlby, 1973;Cassidy, 1994), which is a major characteristic of a resilient personality (Eisenberg, Fabes,Guthrie, and Reiser, 2000), and which contributes to overcome adversities or risk factors alsolater in life (Sroufe, Carlson, Levy and Egeland, 1999).

Reprint requests and correspondence to Peter Zimmermann, Institute of Psychology, University of Erlangen,91054 Erlangen, Germany. (E-mail: [email protected])

0140-1971/02/$35�00 # 2002 The Association for Professionals in Services for Adolescents.Published by Elsevier Science Ltd. All rights reserved

Page 2: Stability of attachment representations during adolescence: the influence of ego-identity status

108 P. Zimmermann and F. Becker-Stoll

The assumption of the importance of attachment organization on later adaptation is basedon two major ideas. First, within an organizational developmental approach earlycompetencies, e.g. secure attachment organization, have been shown to increase theprobability of successfully developing the competencies of later stage-salient issues likeautonomous problem-solving or social competence (Sroufe, 1989; Cummings and Waters,2000). Secondly, based on emotional experiences with the caregivers (e.g. comfort, support,cooperation) children are supposed to develop internal working models of themselves and theattachment figures that control attachment behavior and adaptation in situations that arouse(mainly negative) emotions (Bowlby, 1980; Bretherton and Munholland, 1999; Zimmermann,1999). Given the stability of attachment experiences, Bowlby (1973) expected thedevelopment of increasingly stable patterns of cognitive, emotional and behavioral reactions.

Attachment development from infancy to adolescenceAttachment theory assumes that experiences with primary caregivers are represented ininternal working models, which consolidate during development from infancy to adolescence(Bowlby, 1973). Specifically, adolescence is supposed to be a period where internal workingmodels may finally stabilize, and as a consequence, would be rather resistant to change. Ascompetencies grow during the teenage years, attachment behavior is elicited less oftencompared to infancy. In addition, it is less likely to be expressed by seeking direct physicalproximity but rather by communicating one’s feelings and concerns to the caregivers whenneeded (Bowlby, 1980; Bretherton, 1990). Thus, internal working models as expectationsregarding the attachment figures may become less tested regarding their correspondence toactual experiences of support or rejection. This will increase the stability of already existinginternal working models of attachment.

Longitudinal studies from infancy to childhood showed a high stability of infantattachment organization. Discontinuity of attachment was explained by changes in thecaregiving system. Thus, the stability of the caregiving conditions is one influential factor forthe stability of attachment organization in childhood (George and Solomon, 1999;Grossmann, Grossmann and Zimmermann, 1999). As attachment organization in childhoodalready is very stable, a high stability of attachment patterns during adolescence can beexpected. However the assessment at both ages is different. Attachment patterns in infancyand childhood mainly are assessed by observing attachment behavior patterns towards aspecific caregiver in mildly distressing situations. In adolescence and adulthood however,attachment organization often is assessed by means of the Adult Attachment Interview(AAI). The AAI is a semi-structured interview assessing the state of mind regarding one’sattachment history, i.e., the coherency of the discourse about attachment experiences, andthe emotional integration of these experiences (Main, 1991). In contrast to the methodsused in infancy, the AAI does not assess the current attachment quality to the attachmentfigures, and the classification is not relationship specific.

The resulting attachment representation or state of mind regarding attachment (Main,1991) is classified into one of four patterns that parallel the patterns of attachment quality ininfancy. Individuals classified as secure or free to evaluate coherently report about their pastattachment experiences, which can be supportive but not necessarily have to be so. Theyshow signs of integration of these experiences in their representation of their self inrelationships, and value attachment relationships. Individuals with a dismissing attachmentrepresentation show an incoherent discourse by either idealizing their caregivers, insisting onbeing unable to recall attachment related episodes, or dismiss the effect of non-supportive

Page 3: Stability of attachment representations during adolescence: the influence of ego-identity status

Stability of attachment representation 109

experiences. Moreover, some tend to derogate the need for attachment or attachmentfigures. Individuals classified as enmeshed or preoccupied with attachment show anger whentalking about their attachment relationships. They have difficulties in separating past andpresent relationships to parents, often oscillate between positive or negative appraisals ofattachment experiences, or show a passive discourse style. In addition to these three majorpatterns, subjects can receive a classification as unresolved/disorganized based on lapses inmonitoring of reasoning or discourse when reporting about loss or abuse (Hesse, 1999).

Stability of attachment representationThe AAI shows excellent test–retest reliability over 3 months from 78 percent to 90 percent(Bakermans-Kranenburg and van IJzendoorn, 1993; Sagi, van IJzendoorn, Scharf, Koren-Karie, Joels and Mayseless, 1994). Studies on short-term stability revealed a concordance ofclassification of about 86 percent to 90 percent for a time span of eleven to 18 months(Benoit and Parker, 1994; Crowell, Fraley and Shaver, 1999). Long-term stability ofattachment representation was examined in two studies. Treboux and Crowell (2001) founda stability of 85 percent of the AAI classification over the course of 18 months during thelife-transition to marriage, and a significant concordance of 82 percent with the AAIclassification 5 years later. The authors explained a change in classification by the influenceof marriage to a partner with a secure attachment representation. In a previous study(Becker-Stoll, Zimmermann and Fremmer-Bombik, 1998), we found a stability of the AAIclassifications over the course of 12 years of 72 percent. In that study, the reportedattachment experiences remained rather stable. A change in AAI classification was based ona change in the appraisal of attachment experiences. Comparable to these results inadulthood, change of the AAI classification during adulthood might be based on newsupportive experiences within close relationships, or on a reevaluation of past attachmentexperiences. Empirical evidence focusing on that specific age period is missing.

The long-term prediction of attachment in adolescence from infancy and childhoodattachment pattern revealed controversial results. Longitudinal studies examining theconcordance between infant–mother attachment patterns and attachment representationsassessed by means of the AAI did not necessarily find significant results (Zimmermann,Fremmer-Bombik, Spangler and Grossmann, 1997; Becker-Stoll and Fremmer-Bombik, 1998;Lewis, Feiring and Rosenthal, 2000; Weinfield, Sroufe, and Egeland, 2000). Similarly, thestudies did not show significant concordance between infant–father attachment and the AAIclassification in adolescence (Zimmermann et al., 1997; Becker-Stoll and Fremmer-Bombik,1998). However, there are longitudinal studies that found significant concordance betweeninfant–mother attachment and the AAI in early adulthood (Hamilton, 2000; Waters,Merrick, Treboux and Albersheim, 2000). Moreover, there is empirical evidence that ten-year-olds’ representations of parents as emotionally available and parent’s support whenchildren were ten years old predict the AAI classification at age 16 (Zimmermann andScheuerer-Englisch, 1997).

Disconcordance of attachment classifications in infancy and adolescence may be due torisk factors affecting the emotional availability of the caregivers which affects the attachmentorganization (Zimmermann et al., 1997; Water et al., 2000). However, the disconcordancemay also be explained methodologically. The comparison of the classifications based on thestrange situation and the AAI implies a development model of heterotypic continuity (Caspi,1998). The AAI assesses the current evaluation of past experiences up to late childhoodduring an interview (evaluative-declarative level), whereas the strange situation assesses

Page 4: Stability of attachment representations during adolescence: the influence of ego-identity status

110 P. Zimmermann and F. Becker-Stoll

interactive, emotion regulation strategies that can be observed as behaviour patterns(procedural level) at the end of the first year of life (Zimmermann et al., 1997). In addition, asecure pattern in the AAI can be classified when negative attachment experiences arepresented in a coherent and emotionally integrated manner during the interview resulting invaluing of attachment, showing that these experiences have been processed and re-evaluated(Main and Goldwyn, in press). Thus, disconcordance in quality of attachment organizationmight also be explained by a reappraisal of early experiences.

Reorganization during adolescenceAdolescence is a time period with changes in biological, cognitive and social areas (Lernerand Galambos, 1998) urging a re-organization process in the appraisal of self and others. Theprocess of identity formation might be interpreted as a parallel psychological process of re-evaluating one’s life-goals and commitments to specific values. A major area of change is thefamily. Adolescents and their parents spend less time together (Larson, Richards, Moneta,Holmbeck and Duckett, 1996) and mutuality in parent–teen relationships increases (Younissand Smollar, 1985). Within the family a new calibration of the reciprocal expectations ofjuveniles and their parents is necessary for the successful transition through thisdevelopmental phase (Holmbeck, 1996). Although the rate of quarrels between parentsand their adolescent children does not exceed the rate between parents themselves(Montemayor, 1988) the communication patterns are characterized by the adolescent’sstriving for autonomy. Successful communication pattern within a family include a balancebetween autonomy and relatedness (Allen and Hauser, 1996) and depend on the correctperception of the other’s viewpoint (Pratt and Arnold, 1996). Longitudinally, such acommunication pattern predicts later attachment representation (Allen and Hauser, 1996),ego-development (Allen and Land, 1999), and ego-identity status (Grotevant and Cooper,1986). Thus, open communication of own viewpoints is influential for attachmentdevelopment as well as for identity development and may offer possibilities to reevaluateformer experiences or co-construct new coherent representations.

New communication patterns and growing cognitive abilities to integrate formerly diverseconcepts (Fischer and Lamborn, 1989) facilitate the meta-monitoring of internal models ofself and parents, and reevaluations of the one’s attachment experiences become a newpossibility for change in attachment representations (Kobak and Cole, 1995). Main andGoldwyn (in press) refer to this cognitive and emotional freedom as ‘‘epistemic space’’, andsuggest that it allows individuals to evaluate their parents as attachment figures moreobjectively. As independence increases, a reevaluation of the quality of own attachmentrelationships with parents will be promoted (Allan and Land, 1999). This reevaluation ofmodels of self and others may be part of the more general process of identity development.

Attachment and identity statusIdentity formation is a major developmental task during adolescence (Kroger, 2000) and isrelated to adjustment (Waterman, 1992). According to Marcia (1980) it can be classifiedinto four distinct identity statuses. The identity status achievement represents the successfulsolution of that task, with coherent commitment to specific values based on exploration andpersonal experiences (Berzonsky and Adams, 1999). Identity diffusion is characterized by alack of personal commitments and low exploration regarding new experiences. Foreclosure ischaracterized by commitments without own exploration, and moratorium is a status wherethe adolescent is still exploring with no final decision regarding own values or goals. Within a

Page 5: Stability of attachment representations during adolescence: the influence of ego-identity status

Stability of attachment representation 111

developmental perspective there is empirical evidence for a transition from diffusion toachievement, specifically for relational identity (Meeus, 1996; Meeus, Iedema, Helsen andVollebergh, 1999) but elements of regression into other statuses during specific phases ofadolescence are possible (Waterman, 1985).

From an attachment perspective it is theoretically expected and has been empiricallyshown that adaptation in adolescence is promoted by a secure attachment organization ininfancy as well as by a concurrent secure attachment organization (Kobak and Sceery, 1988;Zimmmermann and Grossmann, 1997; Allan, Moore, Kuperminc and Bell, 1998; Allan andLand, 1999). Based on the organizational-developmental approach, one classic stage-salientissue for adolescence should be related to attachment security. although attachment patternsin adolescence, the development of the ego-identity-status achievement as a measure ofadjustment in adolescence, as assessed by means of the AAI, may not be the direct result ofattachment patterns in infancy it has been shown empirically that there is continuity at theprocedural level of emotion regulation strategies (i.e. attachment behavior as emotionrelated behavior) from early attachment patterns to later emotion regulation patterns oradaptation (Sroufe, 1989) independent of the AAI (Zimmermann, Maier, Winter andGrossmann, 2001).

Marcia (1989) proposed that a secure attachment would promote the development of theego-identity status achievement because an adolescent will feel free to explore his or herenvironment and has a secure base to discuss own experiences and attitudes within thefamily. In contrast, he expected adolescents with insecure attachment organizations toexplore less and either to develop into the ego-identity status diffusion (in case of anavoidant/dismissing attachment) or into foreclosure (in case of an insecurely-ambivalentattachment pattern). Empirical studies have shown that measures of attachment (SeparationAnxiety Test or self-report questionnaires) are related to concurrent identity status (Krogerand Haslett, 1988; Lapsley, Rice and Fitzgerald, 1990), but did not find a longitudinalprediction (Kroger and Haslett, 1988) from attachment (assessed with the SAT) to lateridentity. Studies using the AAI as an attachment measure did find concurrent associationsbetween a secure AAI classification and the achievement identity status and dismissing AAIclassification and identity diffusion (Zimmermann, Gliwitzky and Becker, 1992; Zimmer-mann, 2000).

Theoretically expected is an influence of attachment organization on identitydevelopment. This cannot be inferred from concurrent assessments of attachment andidentity. Kroger and Haslett (1988) found stability of identity and instability in attachmentstyle assessed by means of the Separation Anxiety Test. Moreover, they found that identitypredicted attachment style and not vice versa. From a developmental perspective, thesefindings were unexpected because identity changes during adolescence, as it is a majordevelopmental task whereas attachment organization should consolidate. The effect ofachievement on attachment representation might result from the cognitive flexibility, aninformation processing style that is related to the identity status achievement (Berzonsky andNeimeyer, 1994). Achievers show a high ego-development, a highly differentiated andintegrated self-construct system, and a more coherent narrative about their past experiencesand their current (and future) life (Berzonsky and Adams, 1999). In contrast, individuals inthe identity diffusion status are characterized by a more avoidant, incoherent, and brittlecognitive style.

The present study examines the stability of attachment representation during lateadolescence and the concurrent and longitudinal associations with identity status at age 16.

Page 6: Stability of attachment representations during adolescence: the influence of ego-identity status

112 P. Zimmermann and F. Becker-Stoll

Recent studies on stability of attachment representations focus on adults not on adolescents,and mainly cover a shorter period of time. We expected that attachment representation, asan assessment of internal working models would remain rather stable even during an ageperiod of changes and re-organization of social and individual norms and expectations. Morespecifically, we expected a high stability of the AAI during adolescence at the classificationlevel and the normative level of continuity (Caspi, 1998).

Regarding the relation between attachment and identity we expect a replication of earlierfindings (Zimmermann, 2000), i.e. significant positive associations between attachmentsecurity and the identity status achievement and between dismissing attachmentrepresentation and the identity status diffusion. Although such associations might reflectthe fact that attachment promotes identity development this correlation design might alsoreflect other causal effects. As the AAI does not assess the current attachment relationshipor current attachment behavior but the evaluation of earlier experiences, we expect that theidentity status achievement might promote the ability to re-appraise these experiences.Longitudinally, the identity status achievement therefore might promote security ofattachment representation. Both identity status as well as attachment representation (notbehavior) is based on the ability to present a coherent sense of self within the society (in caseof identity) or in relationships (in case of attachment). Thus, specifically we expected effectson two specific mega-items, (i.e. sub-scales) of the AAI, coherency and integration ofattachment experiences.

Method

Sample and proceduresThis study is a follow-up of the Regensburg longitudinal study, an ongoing study onattachment and socio-emotional development of children with no discernible risk at time ofrecruitment, which started in 1980 (Wartner, Grossmann, Fremmer-Bombik and Suess,1994). Subjects were seen at the University laboratory twice. The first data collection tookplace when subjects were age 16 together with their mothers, where their attachmentrepresentation was assessed for the first time by means of the Adult Attachment Interview(George et al., 1985). The adolescents completed the California Adult Q-Sort (CAQ). Inaddition, their mothers, and six to nine months later their best same-sex friend rated themwith the Q-Sort. Subjects were seen again at age 18 for the second data collection.

The sample for the first data assessment consists of 43 subjects (79?60% of the originalsample, 63% female) at age 16. During the second data collection at age 18 a total of 41adolescents participated. Two adolescents, who have been interviewed at age 18, did notparticipate at age 16.

MeasuresAdult Attachment Interview (George et al., 1985). The Adult AttachmentInterview (AAI) is a semi-structured interview that focuses on attachment-relevantexperiences in childhood, such as being upset, hurt, separated, threatened, rejected, orhaving experienced loss. Other questions aim at understanding the individual’s integration ofexperiences with caregivers into their view of their self and the relationship to one’scaregivers. The AAI was designed to assess an individual’s state of mind with regard toattachment (Main, 1991).

Page 7: Stability of attachment representations during adolescence: the influence of ego-identity status

Stability of attachment representation 113

The interviews were transcribed and rated, using a German version of the AdultAttachment Interview Q-Sort (Kobak, 1993; Zimmermann, 1994), which consists of 100items based on Main and Goldwyn’s (in press) scoring system. The items assess coherency ofthe transcript, representations of the relationships to the attachment figures, integration ofexperiences, valuing of attachment, and other aspects relevant to describe attachmentrepresentation patterns. The AAI-Q-sort method shows significant concordance withindependent classifications following the Main and Goldwyn system (in press) between 96percent (Kobak et al., 1993), 84 percent (Allen et al., 1998), and 80 percent (Zimmermann,Fremmer-Bombik and Becker-Stoll, 1997).

At ages 16 and 18, two independent coders blind to other results of the study, rated eachinterview. One rater at each assessment was reliably trained by Main and Hesse, thesereliable coders trained the other raters. The combined ratings were correlated with theprovided expertly rated prototypes of the three main patterns of attachment representation.Composite reliability of the prototypic Q-sorts range from r(100)=0?92 to r(100)=0?96(Kobak, 1993). Correlations with these prototypes represent subject’s similarity to thespecific attachment representation prototypes, and lead to continuous scores for secure,dismissing, and preoccupied attachment representations. In addition, the method offers ascore for deactivation of attachment relevant thoughts and feeling during the AAI. Eachinterview is checked for reliability regarding the concordance of the distribution of all 100items (Kobal et al., 1993), which is a very conservative test compared to the classical rateragreement of the attachment classification of all interviews. Both ratings are combined, inorder to increase reliability, which is standard for Q-Sort assessments (Block, 1978; Ozer,1993). At age 16, the averaged combined reliability was 0?81 (Spearman-Brown) with arange from 0?63 to 0?93. This results in a reliability of both raters at the dimensional scorelevel of r=0?93 for the dimension secure, r=0?94 for the dimenstion dismissing, r=0?91 forthe dimension preoccupied, and r=0?92 for deactivation. At age 18, the average combinedreliability was 0?80 (Spearman-Brown) with a range from 0?61 to 0?92. At age 18, this refersto a correlation at the dimensional score level of r=0?91 for the dimension secure, r=0?90for dismissing, r=0?87 for preoccupied, and r=0?89 for deactivation, showing at highreliability.

Subjects were also classified into the categorial variable secure vs. insecure or secure,dismissing, and emeshed based on the dimensional scores. A score above zero on the securityscale led to a classification of the attachment representation as secure, below zero to aclassification as insecure. The scores for dismissing and preoccupied were standardized (dueto different variances). The higher standardized score led either to a classification asdismissing or enmeshed. The unresolved/disorganized classification is derived from single Q-Sort-items assessing specific lapses in monitoring or reasoning during discourse of loss orabuse. The concordance at the categorical level at age 18 was 95 percent (kappa 0?90) for asecure vs. insecure comparison 93 percent (kappa 0?86) for a three-way comparison ofsecure, dismissing, and enmeshed attachment representation, and 93 percent (kappa 0?87)for a four-way comparison including the unresolved status.

The single AAI-Q-Sort-Items can be aggregated to mega-items that assess specific aspectsof the AAI (Zimmermann, 1999). This offers the opportunity for particular analysis of thecontent, the formal quality, and the psychological processing apparent in the AAI. For thisanalysis, items are aggregated to mega-items regarding three major aspects of attachmentrepresentation. The content level is assessed by the validly reported attachment experiencesof support by mother (at age 16: a=0?88; at age 18: a=0?80), and support by father (at age 16:

Page 8: Stability of attachment representations during adolescence: the influence of ego-identity status

114 P. Zimmermann and F. Becker-Stoll

a=0?85; a=0?80). The reported experiences were only rated as support if the generaldescriptions were confirmed by reported episodes. The second aspect is the formal discoursequality operationalized as mega-items for coherency (at age 16: a=0?96; at age 18: a=0?94)and attachment relevent childhood memories (at age 16: a=0?95 at age 18: a=0?93). Thethird aspect is the psychological processing of attachment experiences assessed by the mega-item integration of attachment experiences (at age 16: a=0?88; at age 18: a=0?92), (i.e. theacceptance of possible negative experiences and a perspective why parents behaved the waythey did); and an overall attitude regarding valuing of attachment (at age 16: a=0?92;a=0?93). Each mega-item has a possible range on a nine-point-scale.

California Adult Q-Sort (Block, 1978). The California Adult Q-Sort (CAQ; Block,1978) consists of 100 items, which have to be sorted in a person-centered approach tocharacterize an individual’s ipsative personality structure (Ozer, 1993). The items capture awide range of personality features and are sorted in a forced, near-normal distribution on ninecategories. The subjects rated themselves by means of the CAQ, and were rated by theirmothers, and their best friend at the first two assessments at age 16 and age 16?6. In order toincrease reliability the mean of these three Q-sort-ratings was calculated as age 16assessment. Each subject’s combined Q-Sort was correlated with prototypic Q-sorts (Mallory,1984) describing the identity-statuses achievement and diffusion, as they have beenconceptualized by Marcia (1980). The correlation coefficients were used as raw-scores re-presenting each subject’s similarity with the prototype. The prototypes have been developedby experts in identity research, showed good reliability and validity (Mallory, 1984), and havebeen used in studies on personality development (Helson, 1993; Cramer, 1995).

Results

The primary question of interest was the stability of attachment representations. Thus thefirst step of the analysis was to examine the stability of attachment representations at thelevel of mean scores, categorical classification, the level of normative stability, i.e. the relativeposition within the sample, and at the level of mega-items. Next, we considered theconcurrent and longitudinal associations between identity status and attachmentrepresentation. Finally, we examined the prediction of attachment representation at age18 by identity status at age 16 over and above the stability of attachment representation.

In order to examine a general trend in attachment development, we first tested thestability of means of the AAI dimensions. Paired t tests for each of the four AAI dimensionsdid not reveal significant changes in mean scores. The means (with standard deviations inparenthesis) for the AAI dimensions secure, dismissing, preoccupied, and deactivation were0?03 (0?62), 0?02 (0?56), �0?05 (0?30), and 0?05 (0?33) at age 16 and 0?19 (0?59), �0?13(0?55), �0?10 (0?29) and �0?02 (0?33) at age 18, respectively. Thus, there is stability ofattachment representation at the level of means and no significant trend to more security orinsecurity of attachment representation from age 16 to age 18.

As a next step, the stability of attachment representation classifications in a 2� 2 design(secure vs. insecure) and a 3� 3 design (secure vs. dismissing vs. enmeshed) were tested. Asshown in Table 1, the 2� 2 analysis revealed a significant concordance of the secure vs.insecure classification of attachment representations at ages 16 and 18 of 77 per cent(w2=12?4, po0?001, kappa=0?55, po0?01). The 3� 3 analysis revealed a significant

Page 9: Stability of attachment representations during adolescence: the influence of ego-identity status

Table 1 Classifications of attachment representations at ages 16 and 18 (2� 2 analysis and3� 3 analysis)

AAI at age 18 Total AAI at age 18 Total

AAI at age 16 Secure Insecure Secure Dismissing Enmeshed

Secure 16 2 18 Secure 16 2 18Insecure 7 14 21 Dismissing 4 6 1 11

Enmeshed 3 4 3 10

Total 23 16 39 23 12 4 39

Stability of attachment representation 115

concordance of 64 percent (w2=15?3, po0?01, kappa=0?42, po0?01). Thus, at thecategorical level there is significant stability of the AAI between ages 16 and 18. Only oneadolescent, classified as dismissing at both ages, was classified in addition as unresolved withregard to loss (U) at age 16, and again at age 18. The 4� 4 analysis revealed a significantconcordance of 67 percent (Kappa=0?47, po0?01).

Next, we examined normative stability within the sample. As shown in Table 2, therespective AAI-dimensions at age 16, secure, dismissing, preoccupied, and deactivation didcorrelate significantly positive with each corresponding dimension at age 18. The dimensionpreoccupied showed the highest stability within the sample. The overall correlation patternrevealed the expected significant negative associations between secure attachmentrepresentations at age 16, and the dimensions dismissing, preoccupied, and deactivation atage 18. This provides evidence for discriminate validity of the AAI dimensions.

Next, we analysed stability at the level of mega-items of attachment representation.Pearson correlations between the mega-items of the AAI-Q-Sort at ages 16 and 18 revealedthat all mega-items at age 16 were significantly positively associated with their respectivecounterparts at age 18. At the level of reported attachment experiences, the longitudinalcorrelations were r=0?59 (po0?001) for support by father, r=0?50 (po0?001) for support bymother (r=0?50, po0?001). Correlations of mega-items assessing the formal quality ofdiscourse were r=0?46 (po0?01) for coherency, and r=0?49 (po0?001) for child hood memory.The longitudinal correlations for mega-items assessing the processing of attachment experienceswere 0?51 (po0?001) for integration of attachment experiences, and r=0?49 (po0?001) forvaluing attachment.

Table 2 Pearson correlations between the dimensions of the AAI at ages 16 and 18

AAI at age 18

Secure Dismissing Preoccupied Deactivation

AAI at age 16Secure 0?51*** �0?48** �0?39 �0?36*Dismissing �0?46** 0?48** 0?24 0?43**Preoccupied �0?45** 0?32* 0?64*** �0?07Deactivation �0?32* 0?41** �0?06 0?49**

***pr0?001; **pr0?01; *pr0?05 (p two-tailed).

Page 10: Stability of attachment representations during adolescence: the influence of ego-identity status

116 P. Zimmermann and F. Becker-Stoll

Attachment and identity in adolescence: concurrent and longitudinal analysis.We next examined the concurrent and longitudinal associations between the adolescents’attachment representations at ages 16 and 18 and their ego-identity status at age 16. Pearsoncorrelations between attachment representation and ego-identity status at age 16 revealedthat security of attachment representation was significantly positively associated withidentity achievement and marginally negatively associated with identity diffusion (see Table3). For a dismissing attachment representation, there was a similar but reversed pattern. Apreoccupied state of mind with regard to attachment was marginally negatively associatedwith identity diffusion, whereas deactivation of attachment relevant thoughts and feelingsdid correlate negatively with achievement, although failed to reach statistical significance.

The longitudinal associations between identity at age 16 and attachment representation atage 18 replicated the concurrent association at age 16 to a great extend. Both identityachievement and identity diffusion at age 16 were related to the AAI dimensions secure,dismissing, and deactivation. Again there was no significant correlation between ego-identitystatuses and preoccupation.

The longitudinal prediction of attachment representation by ego-identity status might beunanticipated, as theoretically expected, a secure attachment organization (which impliesexperienced parental support) should promote the formation of an achievement identitystatus. However, the AAI assesses the state of mind regarding childhood attachmentexperiences (at a declarative level) and does not assess attachment behaviour (procedurallevel) or concurrent support by parents. Thus, in a next step we analyzed whether theinfluence of identity on attachment representation is mainly based on its effect on state ofmind scales, i.e. coherency in the AAI and the processing of attachment experiences. Inorder to test the specific influence of identity, concurrent and longitudinal correlationsbetween identity status at age 16 and AAI-mega-items at ages 16 and 18 were computed. Asshown in Table 4, both concurrently and longitudinally identity achievement and identitydiffusion were significantly associated with formal discourse quality in the AAI (i.e.coherency and childhood memory), and the integration of attachment experiences. Thereported attachment experiences with parents and the mega-item valuing attachment werenot significantly related to identity status, but showed a statistical trend at somemeasurement points.

Table 3 Concurrent and longitudinal correlations between ego-identity status at age 16 andthe dimensions of the AAI at ages 16 and 18

AAI at age 16

Identity at age 16 Secure Dismissing Preoccupied Deactivation

Achievement 0?34* �0?33* �0?22 �0?28+Diffusion �0?29+ 0?27+ 0?24+ 0?20

AAI at age 18

Identity at age 16 Secure Dismissing Preoccupied Deactivation

Achievement 0?33* �0?38* �0?06 �0?42*Diffusion �0?36* �0?39* 0?16 0?38*

**pr0?01; *pr0?05; +pr0?10 (p two-tailed).

Page 11: Stability of attachment representations during adolescence: the influence of ego-identity status

Table 4 Concurrent and longitudinal correlations between identity-status at age 16 andAAI mega items at ages 16 and 18

Ego-identity status at age 16

Achievement Diffusion

AAI at AAI at

AAI-mega-items Age 16 Age 18 Age 16 Age 18

Reported attachment experiencesSupport by mother 0?07 0?13 �0?9 �0?22Support by father 0?27+ 0?26 �0?23 �0?30+

Formal discourse qualityCoherency 0?32* 0?36* �0?27+ �0?39*Childhood memory 0?46** 0?38* �0?43** �0?37*

Processing of attachment experiencesIntegration of attachment experiences 0?39* 0?32* �0?33+ �0?33*Valuing attachment 0?22 0?22 �0?19 �0?29+

**pr0?05; *pr0?05; +pr0?10 (p two-tailed).

Stability of attachment representation 117

The cross-lag correlations from identity at age 16 to attachment representation at age 18imply causality, although may be interpreted with caution (Rogosa and Willett, 1985).Longitudinal correlations can be spurious effects of the association with a third variable.Thus, we next computed hierarchical regressions with each AAI dimension at age 18 asdependent variable, the respective AAI dimension at age 16 entered as first predictor, andego-identity status entered a second predictor. As can be seen from Table 5, identity diffusionsignificantly predicted a dismissing and deactivated attachment representation over andabove the stability of the respective attachment dimensions. For the secure dimension only a

Table 5 Hierarchical regression analysis for the prediction of the AAI dimensions at age 18

Age 18 outcome variablesAge 16 predictors Beta R2 FDR2

Secure at age 18Secure at age 16 0?45** 0?26** 12?8**Diffusion at age 16 �0?26+ 0?32** 3?4+

Dismissing age 18Dismissing at age 16 0?48** 0?23** 11?4**Diffusion at age 16 0?30* 0?32** 4?5*

Preoccupied at age 18Preoccupied at age 16 0?64*** 0?41** 25?4***Preoccupied at age 16 0?02 0?41** 0?02

Deactivation at age 18Deactivation at age 16 0?49** 0?24** 11?5**Diffusion at age 16 0?31* 0?34** 5?4**

***pr0?01; **pr0?01; *pr0?05 (p two-tailed).

Page 12: Stability of attachment representations during adolescence: the influence of ego-identity status

118 P. Zimmermann and F. Becker-Stoll

statistical trend was found. Preoccupation was not predicted by identity diffusion. Similarregression analyses with identity achievement at age 16 as second predictor of attachmentrepresentation at age 18 revealed statistical trend in the prediction of dismissing at age 18(b=�0?27, po0?10), and a significant prediction of deactivation at age 18 (b=�0?33,po0?05). Identity achievement did not significantly account for additional variance in theprediction of the AAI dimensions secure and preoccupied at age 18 after controlling for therespective AAI dimensions at age 16.

Discussion

The present study examined the stability of attachment representations between ages 16 and18, in relation to ego-identity-status at age 16. Derived from attachment theory and basedon studies on the stability of the AAI in adulthood, we expected to find significant stability ofattachment representations during adolescence, despite the many changes and individualand social reorganizations typical for that age period.

The results showed stability of attachment representations at several levels of analysis. Themeans of the AAI dimensions did not change significantly. Thus, there is no generaldevelopment from insecurity of attachment representation during middle adolescence tomore security of attachment representation in late adolescence. Whereas Steinberg (1988)had found that emotional distance between adolescents and their parents increases as afunction of puberty, this does seem to affect attachment representation. However, theindividual developmental status varies tremendously within the each age group. Thus,assessing pubertal development and attachment representation may clarify existing relations.

The quality of attachment representation in adolescence remained relatively stable overthe course of two years. The significant stability of 77 percent is comparable to studies withadults over a shorter time span (Bakermans-Kranenburg and van IJzendoorn, 1993; Benoitand Parker, 1994). The stability is even more impressive as the subjects are younger and thetime-span between the two assessments is longer compared to studies with adults. Researchon the continuity of personality traits has shown that stability decreases as a function of timebetween measurements, and increases with subject’s age at the time of the first assessment(Caspi, 1998). At the level of continuous scores for attachment representation, we foundsignificant normative stability (Caspi, 1998) that is clearly in the expected range of stabilityof personality traits for the age group studied here (Roberts and Friend-DeVecchio, 2000). Inthis sample, especially the preoccupied state of mind remained stable. Preoccupation ischaracterized by extended and irrelevant discourse about attachment experiences withoutfinal conclusive appraisal. This might be understood as the expression of a still ongoingprocess of individuation from the parents combined with a self-centered perspective, and alack of distance from the own viewpoint. As most subjects still lived with their parents, apsychological distance in appraising one’s attachment experiences might be a difficult task.Studies comparing adolescents who already have left home might show different results.

However, the significant correlations of AAI dimensions do not imply an absolutedeterminism of attachment representation from age 16 on. Bowlby’s (1973) hypothesis thatfrom adolescence on attachment patterns and related personality characteristics wouldremain relatively stable can be confirmed partly. There is significant stability but there aresome adolescents who change in their attachment representations. Differences inclassification may be accounted for by measurement error at each assessment due to non-

Page 13: Stability of attachment representations during adolescence: the influence of ego-identity status

Stability of attachment representation 119

perfect rater agreement. This decreases the expectable continuity that can be foundempirically. However, change in attachment representation might also result from areappraisal of attachment experiences.

Based on an earlier study on stability of attachment representation over the course of12 years (Becker-Stoll et al., 1998) the hypothesis was tested that discontinuity can be causedby changes in the evaluation and emotional integration of attachment experiences, and thecoherency of the state of mind regarding attachment because of re-appraisal of childhoodexperiences. However, in contrast to the results with the mothers of the adolescents studiedhere, we found similar stability for the reported attachment history as well as for the formaldiscourse qualities (e.g. coherency), and the processing of attachment experiences (e.g.integration of experiences). Thus, during adolescence stability or change is observable at allaspects of the AAI.

The significant associations between ego-identity status and attachment representationreplicate earlier findings with 16-year-olds in different samples (Zimmermann et al., 1992;Zimmermann, 2000). Security of attachment representations is concurrently and long-itudinally related to the identity status achievement whereas a dismissing attachmentrepresentation is associated with identity diffusion. Based on attachment theory, theconcurrent associations might be interpreted as evidence for the promoting effect ofattachment on ego-identity, because attachment security is a basis for the exploration of newroles, attitudes, and relationships. Longitudinally however, ego-identity status predicted laterattachment representation over and above the autonomous stability of attachmentrepresentation.

Although the additional variance explained by each ego-identity status was small, identitydiffusion will lead to a stabilization of dismissing attachment representation and inhibit achange to more security. This influence is mainly based on the longitudinal prediction ofcoherency and the understanding and emotional integration of attachment experiences.From a methodological perspective, the classification of both the ego-identity status and theattachment representation is based on coherency of representation. Identity diffusion ischaracterized by lack of clear commitments, an avoidant information processing style, and anincoherent life perspective (Berzonsky and Neimeyer, 1992; Berzonsky and Adams, 1999).Similarly, a dismissing attachment representation is characterized by an avoidant informationprocessing style and an incoherent perspective on the attachment history (Kobak and Sceery,1988; Zimmermann, 1999). Thus, coherency of the representation of either one’s values andlife-goals (in case of identity) or one’s attachment experiences (in case of the AAI) might bea common underlying methodological factor.

From a developmental perspective however, the security an adolescent gains from thecommitment to specific, personal goals based on personal experiences may provide her orhim with the emotional freedom to appraise the parent’s caregiving more objectively. Theperception of parents as persons with supportive and unsupportive characteristics within aspecific current life situation promotes the coherent appraisal, understanding, and emotionalintegration of earlier attachment experiences, and the development an (earned) secureattachment representation (Kobak and Cole, 1995).

Ego-identity status was not related to support from parents in childhood as validly reportedduring the AAI. However, concurrent parental support and emotional availability issignificantly related to an achievement ego-identity status in adolescence (Zimmermann,2000). Thus, an effect of attachment on the successful solution of stage-salient issues (e.g.identity formation in adolescence) might be based on the experiences of current parental

Page 14: Stability of attachment representations during adolescence: the influence of ego-identity status

120 P. Zimmermann and F. Becker-Stoll

support and their emotional availability. The AAI, in contrast, assesses the evaluation of pastexperiences and the coherency of this evaluation.

Several limitations of this study have to be acknowledged. First, the sample size is small, sothat generalizations of the results of this study are restricted. Second, ego-identity status wasassessed by means of the CAQ instead of an interview approach as commonly used (Marcia,1980). More domain specific assessments of identity might have been included (Meeus et al.,1999). These limitations might be addressed in new studies or replications of longitudinalassessments of attachment and identity during adolescence.

In conclusion, we found that there is reasonable stability of attachment representationsduring adolescence. Changes in the development of attachment representation inadolescence are influenced longitudinally by identity status. This emphasizes two aspectsof attachment theory. First, attachment patterns in adolescence are stable despite this periodof rapid change. Second, attachment and measures of concurrent adaptation are significantlyrelated supporting the idea of the important influence of attachment organization on thedevelopment of a resilient personality.

In conclusion, we found high autonomous stability of attachment representations duringadolescence. The development of security in the AAI during adolescence is to some extentinfluenced by the ego-identity status. Both ego-identity status and attachment representa-tion are related to flexibility, differentiation, and accuracy in cognitive style (Berzonsky andAdams, 1999) or flexibility in appraisal, emotional, and behavioral regulation strategies(Kobak and Cole, 1995; Bretherton and Munholland, 1999; Zimmermann, 1999). Acoherent representation of both past attachment relationships (Grossmann, Grossmann,Winter and Zimmermann, in press) and of current life-goals (Waterman, 1992) mightpromote optimal adaptation. The reciprocal influences of working models of one’sattachment figures and the self in attachment relationships on the hand, and identity as aworking model of the society and the self in society on the other hand, might help to gaininsights into developmental influences on adaptation during adolescence.

Acknowledgement

The first assessment and analysis was supported by the Deutsche Forschungsgemeinschaft.The second assessment and analysis was supported by the Koehler-Stiftung, Munchen. Datawere collected during our work do at the University of Regensburg. The authors would like tothank Inge Loher for the data collection, Annie Bernier and Gottfried Spangler for theircomments on an earlier draft of this paper, two anonymous reviewers and the editor for theirvery valuable and helpful comments. Finally, we would like to express our gratitude to thefamilies for their cooperative and enduring participation.

References

Ainsworth, M. D. S. (1990). Some considerations regarding theory and assessment relevant toattachments beyond infancy. In Attachment in the Preschool Years, M. T. Greenberg, D. Cicchettiand C. M. Cummings (Eds). Chicago: University of Chicago Press, pp. 1–39.

Allen, J. P. and Hauser, S. T. (1996). Autonomy and relatedness in adolescent-family interactions aspredictors of young adults’ states of mind regarding attachment. Development and Psychopathology,8, 793–809.

Page 15: Stability of attachment representations during adolescence: the influence of ego-identity status

Stability of attachment representation 121

Allen, J. P. and Land, D. (1999). Attachment in Adolescence. In Handbook of AttachmentFTheory,Research, and Clinical Applications, J. Cassidy and P. R. Shaver (Eds). New York: The Guilford Press,pp. 319–332

Allen, J. P., Moore, C. M., Kuperminc, G. P. and Bell, K. L. (1998). Attachment and adolescentpsychosocial functioning. Child Development, 69, 1406–1419.

Bakermans-Kranenburg, M. J. and IJzendoorn, M. H. (1993). A psychometric study of the adultattachment interview: reliability and discriminant validity. Developmental Psychology, 29, 870–879.

Becker-Stoll, F. and Fremmer-Bombik (1997). Adolescent–mother Interaction and Attachment: ALongitudinal Study. Poster presented at the biennial meeting of the Society for Research in ChildDevelopment. Washington, DC.

Becker-Stoll, F., Zimmermann, P. and Fremmer-Bombik, E. (1998). Continuity and Discontinuity ofAttachment Representations: A Twelve-year Longitudinal Study. Poster presented at the XVth Meetingof the International Society for the Study of Behavioral Development, Bern, Switzerland.

Benoit, D. and Parker, K. C. H. (1994). Stability and transmission of attachment across threegenerations. Child Development, 65, 1444–1456.

Berzonsky, M. D. and Adams, G. R. (1999). Reevaluating the identity status paradigm: still useful after35 years. Developmental Review, 19, 557–590.

Berzonsky, M. D. and Neimeyer, G. J. (1994). Ego identity status and identity processing orientation:the moderating role of commitment. Journal of Research in Personality, 28, 425–435.

Block, J. (1978). The Q-sort Method in Personality Assessment and Psychiatric Research. Palo Alto, CA:Consulting Phychologist Press.

Bowlby, J. (1969). Attachment and Loss, Vol. 1: Attachment. New York: Basic Books.Bowlby, J. (1973). Attachment and Loss, Vol. 2: Separation: Anxiety and Anger. New York: Basic Books.Bowlby, J. (1980). Attachment and Loss, Vol. 3: Loss: Sadness and Depression. New York: Basic Books.Bretherton, I. (1990). Open communication and internal working models: their role in the

development of attachment relationships. In Socio-emotional Development, Nebraska Symposiumon Motivation, R. A. Thompson (Ed.). Lincoln: University of Nebraska Press, pp. 57–113.

Bretherton, I. and Munholland, K. A. (1999). Internal working models in attachment relationships:a construct revisited. In Handbook of Attachment, J. Cassidy and P. R. Shaver (Eds). New York:Guilford, pp. 89–111.

Caspi, A. (1998). Personality development across the life course. In Handbook of Child Psychology.Vol. III: Social, Emotional and Personality Development, W. Damon and N. Eisenberg (Eds). NewYork: Wiley & Sons, pp. 311–388.

Cassidy, J. (1994). Emotion regulation: influences of attachment relationships. In The Development ofEmotions Regulation: Biological and Behavioral Considerations, N. A. Fox (Ed.). Monographs of theSociety of Research in Child development, 59. Chicago: University of Chicago Press, pp. 228–249.

Cramer, P. (1995). Identity, narcissism, and defense mechanisms in late adolescence. Journal of Researchin Personality, 29, 341–361.

Crowell, J. A., Fraley, R. C. and Shaver, R. P. (1999). Measurement of Individual Differences inAdolescent and Adult Attachment. In Handbook of Attachment: Theory, Research, and ClinicalApplications, J. Cassidy and P. R. Shaver (Eds). New York: Guilford Press, pp. 434–468.

Eisenberg, N., Fabes, R. A., Guthrie, I. K. and Reiser, M. (2000). Dispositional emotionality andregulation: their role predicting quality of social functioning. Journal of Personality and SocialPsychology, 78, 136–157.

Fishcher, K. W. and Lamborn, S. D. (1989). Mechanisms of variations in developmental levels:cognitive and emotional transitions during adolescence. In Transition Mechanisms in ChildDevelopment: The Longitudinal Perspective, A. de Ribaupierre (Ed.). Cambridge: CambridgeUniversity Press.

George, C., Kaplan, N. and Main, M. (1985). An Adult Attachment Interview. Unpublished manuscript,University of California at Berkeley.

George, C. and Solomon, J. (1999). Attachment and caregiving: the caregiving behavioral system. InHandbook of Attachment: Theory, Research, and Clinical Applications, J. Cassidy and P. R. Shaver(Eds). New York: Guilford Press, pp. 649–670.

Grossmann, K. E. and Grossmann, K. (1991). Attachment quality as an organizer of emotional andbehavioral responses in a longitudinal perspective. In Attachment Across the Life Cycle, C. M. Parkes,J. Stevenson-Hinde and P. Marris (Eds). London/New York: Tavistock/Routledge, pp. 93–114.

Page 16: Stability of attachment representations during adolescence: the influence of ego-identity status

122 P. Zimmermann and F. Becker-Stoll

Grossmann, K. E. and Grossmann, K. and Zimmermann, P. (1999). A wider view of attachment andexploration: stability and change during the years of immaturity. In Handbook of Attachment Theoryand Research, J. Cassidy and P. Shaver (Eds). New York: Guilford, pp. 760–786.

Grossmann, K. E. and Grossmann, K., Winter, M. and Zimmermann, P. (in press). Attachmentrelationships and appraisal of partnership: from early experience of sensitive support to laterrelationship representation. In Paths to Successful Development, L. Pulkinnen and A. Caspi (Eds).Cambridge, U.K.: Cambridge University Press.

Grotevant, H. D. and Cooper, C. R. (1985). Patterns of interaction in family relationships and thedevelopment of identity exploration in adolescence. Child Development, 56, 415–428.

Hamilton, C. (2000). Continuity and discontinuity of attachment from infancy through adolescence.Child Development, 71, 690–694.

Helson, R. (1993). Comparing longitudinal studies of adult development: toward a paradigm of tensionbetween stability and change. In Studying Lives through Time. Personality and Development, D. C.Funder, R. D. Parke, C. Tomlinson-Keasy and K. Widaman (Eds). Washington, D.C.: AmericanPsychological Association, pp. 93–119.

Hesse, E. (1999). The Adult Attachment Interview. Historical and Current Perspectives, In Handbookof AttachmentFTheory, Research, and Clinical Applications, J. Cassidy and P. R. Shaver (Eds). NewYork: the Guilford Press, pp. 395–433.

Holmbeck, G. H. (1996). A model of family relational transformations during the transition toadolescence: parent–adolescent conflict and adaptation. In Transitions through Adolescence.Interpersonal Domains and Context, J. A. Graber, J. Brooks-Gunn and A. C. Petersen (Eds).Mahwah, NJ: Lawrence Erlbaum, pp. 167–200.

Kobak, R. R. and Cole, C. (1995). Attachment and metamonitoring: Implications for adolescentautonomy and psychopathology. In Rochester Symposium on Development and Psychopathology:Vol. 5. Disorders of the Self, D. Cicchetti (Ed.). Rochester, NY: University of Rochester Press,pp. 267–297.

Kobak, R. R., Cole, H. E., Ferenz-Gillies, R., Fleming, W. S. and Gamble, W. (1993). Attachmentand emotion regulation during mother–teen problem solving: a control theory analysis. ChildDevelopment, 64, 231–245.

Kobak, R. R. and Sceery, A. (1988). Attachment in late adolescence: working models, affect regulation,and representations of self and others. Child Development, 59, 135–146.

Kobak, R. R. (1993). The Adult-Attachment-Interview-Q-Sort. Unpublished manuscript. University ofDelaware.

Kroger, J. and Haslett, S. J. (1988). Separation-individuation and ego-identity status in lateadolescence: a two-year longitudinal study. Journal of Youth and Adolescence, 17, 59–79.

Kroger, J. (2000). Ego identity status research in the new millennium. International Journal of BehavioralDevelopment, 24, 145–148.

Lapsley, K. D. and Rice, K. (1988). The ‘‘new look’’ at the imaginary audience and personal fable:Toward a general model of adolescent ego development. In Self, Ego, and Identity, IntegrativeApproaches, K. D. Lapsley and F. C. Power (Eds). New York: Springer, pp. 109–129.

Lapsley, K. D. and Rice, K. and FitzGerald, D. (1990). Adolescent attachment, identity, adjustment tocollege: implications for the continuity of adaptation hypothesis. Journal of Counseling andDevelopment, 68, 561–565.

Larson, R. W., Richards, M. H., Moneta, G., Holmbeck, G. and Duckett, E. (1996). Changes inadolescents’ daily interactions with their families from ages 10 to 18: disengagement andtransformation. Developmental Psychology, 32, 744–754.

Lerner, R. M. and Galambos, N. L. (1998). Adolescent development: challenges and opportunities forresearch, programs, and policies. Annual Review of Psychology, 49, 413–446.

Lewis, M., Feiring, C. and Rosenthal, S. (2000). Attachment over time. Child Development, 71, 707–720.

Main, M. and Goldwyn, R. (in press). Adult Attachment Scoring and Classification Systems. Unpublishedmanuscript, University of California, Berkeley.

Main, M. (1991). Metacognitive knowledge, metacognitive monitoring and singular (coherent) versusmultiple (incoherent) model of attachment: findings and directions for future research. InAttachment Across the Life Cycle, C. M. Parkes, J. Stevenson-Hinde and P. Marris (Eds). London/New York: Tavistock/Routledge, pp. 127–159.

Page 17: Stability of attachment representations during adolescence: the influence of ego-identity status

Stability of attachment representation 123

Main, M., Kaplan, N. and Cassidy, J. (1985). Security in infancy, childhood, and adulthood: a move tothe level of representation. In Growing Points in Attachment Theory and Research. Monographs of theSociety for Research in Child Development, I. Bretherton and E. Waters (Eds). 50, 66–106.

Mallory, M. E. (1989). Q-sort definition of ego identity status. Journal of Youth and Adolescence, 18,399–412.

Marcia, J. E. (1988). Common proceses underlying ego-identity, cognitive/moral development andindividuation. In Self, Ego, Identity, D. K. Lapsley and F. C. Power (Eds). New York: Springer,pp. 211–225.

Marcia, J. E. (1980). Identity in adolescence. In Handbook of Adolescent Psychology, J. Adelson (Ed.).New York: Wiley, pp. 159–187.

Meeus, W. (1996). Studies on identity development in adolescence: an overview of research and somenew data. Journal of Youth and Adolescence, 25, 569–599.

Meeus, W., Iedema, J., Helsen, M. and Vollebergh, W. (1999). Patterns of adolescent identitydevelopment: review of literature and longitudinal analysis. Developmental Review, 19, 419–461.

Ozer, D. J. (1993). The Q-Sort-method and the study of personality development. In Studying Livesthrough Time: Approaches to Personality and Development, D. C. Funder, R. D. Parke, C. Tomlinson-Keasey and Widaman (Eds). Washington, DC: American Psychological Association, pp. 147–168.

Roberts, B. W. and Friend-DelVecchio, W. (2000). Consistency of personality traits from childhood toold age: a quantitative review of longitudinal studies. Psychological Bulletin, 126, 3–25.

Rogasa, D. and Willett, J. B. (1985). Understanding correlates of change by modeling individualdifferences in growth. Psychometrika, 50, 203–228.

Sagi, A., Van Ijzendoorn, M. H., Scharf, M., Koren-Karie, N., Joels, T. and Mayseless, O. (1994).Stability and discriminant validity of the Adult Attachment Interview: a psychometric study inyoung Israeli adults. Developmental Psychology, 30, 988–1000.

Sroufe, L. A. (1989). Relationships, self and individual adaptation. In Relationship Disturbances in EarlyChildhood. A Developmental Approach, A. J. Sameroff and R. N. Emde (Eds). New York: Basicbooks, pp. 70–94.

Sroufe, L. A., Carlson, E. A., Levy, A. K. and Egeland, B. (1999). Implications of attachment theory fordevelopmental psychopathology. Development and Psychopathology, 11, 1–13.

Steinberg, L. (1988). Reciprocal relation between parent-adolescent distance and pubertal maturation.Developmental Psychology, 24, 122–129.

Treboux, D. and Crowell, J. (2001). Are Attachment Representations Stable Across Phases of Development?Transitions to Marriage and to Parenting. Poster presented at the Biennial meeting of the Society forResearch on Child Development, Minneapolis.

Wartner, U., Grossmann, K., Fremmer-Bombik, E. and Suess, G. (1994). Attachment patterns at age sixin South Germany: predictability from infancy and implications for preschool behavior. ChildDevelopment, 65, 1014–1027.

Waterman, A. S. (1985). New Directions for Child Development. Identity in Adolescence: Processes andContent. San Francisco: Jossey-Bass.

Waterman, A. S. (1992). Identity as an aspect of optimal psychological functioning. In AdolescentIdentity Formation, G. R. Adams, T. P. Gullotta and R. Montmayor (Eds). Newbury Park, CA: Sage,pp. 50–72

Waters, E. and Cummings, M. (2000). A secure base form which to explore close relationships. ChildDevelopment, 71, 164–172.

Waters, E., Merrick, S. Treboux, D., Crowell, J. and Albersheim, L. (2000). Attachment security ininfancy and early adulthood: a twenty-year longitudinal study. Child Development, 71, 684–690.

Weinfield, N. S. Sroufe, L. A. and Egeland, B. (2000). Attachment from infancy to early adulthood in ahigh-risk sample: continuity, discontinuity, and their correlates. Child Development, 71, 695–702.

Zimmermann, P. (1994). A German version of the Adult Attachment Interview Q-sort. Unpublishedmanuscript, University of Regensburg.

Zimmermann, P. (1999). Structure and functioning of internal working models of attachment and theirrole for emotion regulation. Attachment and Human Development, 1, 55–71.

Zimmermann, P. (2000). L’attachmement a l’adolescence: measure, developpement et adaptation(Attachment in adolescence: Development, assessment, and adaptation). In Attachment etDeveloppement, S. Larose, G. M. Tarabulsy, D. R. Pederson and G. Moran (Eds). Quebec UniversityPress, pp. 181–204.

Page 18: Stability of attachment representations during adolescence: the influence of ego-identity status

124 P. Zimmermann and F. Becker-Stoll

Zimmermann, P. and Grossmann, K. E. (1997). Attachment and adaptation in adolescence. InDevelopment of Interaction and Attachment: Traditional and Non-traditional Approaches, W. Koops,J. B. Hoeksma and D. C. van den Boom (Eds). Amsterdam: North-Holland, pp. 271–280.

Zimmermann, P. and Scheuerer-Englisch, H. (1997). Attachment at Age Ten and Age Sixteen. Posterpresented at the Biennial Meeting of the Society for Research on Child Development,Washington.

Zimmermann, P., Becker-Stoll, F. and Fremmer-Bombik, K. (1997). Die Erfassung der Bindungsrepra-sentation mit dem Adult Attachment Interview: Ein Methodenvergleich. (The assessment ofattachment representation with the Adult Attachment A comparison of methods). Kindheit undEntwicklung, 6, 173–182.

Zimmermann, P., Fremmer-Bombik, E., Spangler, G. and Grossmann, K. E. (1997). Attachment inadolescence: a longitudinal perspective. In Development of Interaction and Attachment: Traditionaland Non-traditional Approaches, W. Koops, J. B. Hoeksma and D. C. van den Boom (Eds).Amsterdam: North Holland, pp. 281–292

Zimmermann, P., Maier, M., Winter, M. and Grossmann, K. E. (2001). Attachment and emotionregulation of adolescents during joint problem-solving with a friend. International Journal ofBehavioral Development, 25, 331–342.