maternal attachemnt, maternal sezitivity and infant mother intractions

Upload: elena-madalina

Post on 14-Apr-2018

222 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    1/9

    Developmental Psychology1998, Vol. 34, No" 5, 925-933 Copyright 1998 by the American Psychological Association, Inc.0012-1649/98/S3.00

    Maternal Attachment Representations, Maternal Sensitivity, and theInfant-Mother Attachment RelationshipDavid R. Pederson, Karin E. Gleason, Greg Moran, and Sandi BentoUnive r s i ty o f Wes te rn On ta r io

    The role of maternal sensitivity as a mediator accounting for the robust association between maternalattachment representations and the quality of the infant-mother attachment relationship was exam-ined. Sixty mother-infant dyads were observed at home and in the Strange Situation at 13 months,and mothers participated in the Adult Attachment Interview (AAI) within the next 6 months. Astrong association was found between AAI and Strange Situation classifications, and autonomousmothers were more sensitive at home than were nonautonomous mothers. Mothers in secure relation-ships were more sensitive at home than mothers in nonsecure relationships. Likewise, infants insecure relationships were more secure as assessed by the Waters' Attachment Q sort than infants innonsecure relationships. A test of the mediational model revealed that maternal sensitivity accountedfor 17% of the relation between AAI and Strange Situation classifications.

    A central element of current attachment theory (Cassidy,1994; Main, Kaplan, & Cassidy, 1985) is that a parent's cogni-tive representations of relationships is an important determinantof the patterning and quality of interactions with his or her infantand thus of their developing relationship. A number of studieshave found evidence for a strong association between a parent'sattachment repre sentations as assessed throug h the Adult Attach-ment Interview (George, Ka plan, & Main, 1985) and the infan t-parent relationship as reflected in the Ainswo rth, Blehar, Waters,and Wall (19 78 ) Strange Situation (e.g. , Ainsworth & Eichberg,1991; Benoit & Parker, 1994; Fonagy, Steele, & Steele, 1991;Grossmann, Fremmer-Bombik, Rudolph, & Grossmann, 1988;Main et a]., 1985; Ward & Carlson, 1995). This conclusionhas been strengthened by a meta-analytic review in which vanIJzendoorn (1995) found evidence for a robust association be-tween Adult Attachment Interview classifications and the par-ent-infant attachment relationship. In this review, involving 661dyads, 82% of parents classified as autonomous were in securerelationships with their infants, 65% of dismissing parents werein avoidant relationships, and 35% of preoccupied parents werein ambivalent relationships, thus providing support for the spe-cific predicted associations especially for the association be-tween autonomy and security. Interviews conducted prenatallyeffectively p redict relationships at 1 year of age (Ben oit &

    This research was supported by grants from the Ontario Mental HealthFoundation and from the Social Sciences and Humanities ResearchCouncil of Canada.We are grateful to the mothers and infants who participated in thisresearch; to Patricia Delmore-Ko, Kirstie Fisher, and Andrea Noonanfor assisting with home observations; and to Anne Krupka and GeorgeTarabulsy for Strange Situation coding.Correspondence concerning this article should be addressed to David R.Pederson, Karin E. Gleason, Greg Moran, or Sandi Bento, Department ofPsychology, University of Western Ontario, London, Ontario, CanadaN6A 5C2. Electronic mail m ay be sent to pederson julian.uwo.ca,[email protected], [email protected], [email protected].

    Parker, 1994; Fonagy, Steele, & Ste ele, 1 991 ; Ward & C arlson,1995), supporting the argument that the parent's attachment-related cognitions are developmental determinants of the attach-ment relationship.Because an infant has no direct access to his or her parent'scognitions, attachment theory postulates that the link betweena parent 's at tachment representations and the paren t-infant rela-tionship must be mediated through their interactions. In a secondmeta-analysis, van IJzendoorn (1995) examined the existingliterature relating Adult Attachment Interview classifications toa variety of measures of parental responsiveness. He concludedthat this association was reliable but somewhat weaker than the

    correspondence between Adult Attachment Interview classifica-tions and relationships measured in the Strange Situation.In the same report, van IJzendoorn (1995) evaluated the hy-pothesis that parental sensitivity mediates the relation betweena parent's internal working model of attachment and the attach-ment relationship. The results of his two me ta-analyses prov idedestimates of the statistical association between autonomy in theAdult Attachme nt Interview and security in the Strange Situationand between autonomy and parental sensitivity. Goldsmith andAlansky's (1987) meta-analysis of the relation between parentalsensitivity and attachment security provided an estimate of thefinal link in the model, van IJzendoorn found that 23% of thedirect association between parental attachment representations

    and the attachment relationship was mediated by parental sensi-tivity. He concluded that the larger part of the association be-tween parents' representations of attachment and the quality ofthe attachment relationship m ust be attributed to som ething otherthan sensitivity as this variable is currently conceptualized andassessed.The results of van IJzendoorn's (1995) analyses are particu-larly credible because they stem from the results of a largenumber and variety of studies. However, only three of the studiesincluded in his analyses involved assessments of all three vari-ables in his model, and in each of these the results included acomplication or anomaly. In the first study, involving a sample

    92 5

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    2/9

    926 PEDERSON, GLEASON, MORAN, AND BENTOof only 20 dyads and an early version of the Adult AttachmentInterview coding system, Grossmann et al. (1988) found anassociation betwe en maternal sensitivity du ring the first year andthe mother's attachment representations 5 years later. Maternalsensitivity at 6 months was associated with Strange Situationclassifications, but no such association was present at the 10-month assessment of maternal sensitivity, and at 2 months thedifferences were significant only between secure and am bivalentdyads (Grossmann, Grossmann, Spangler, Suess, & LJnzner,1985). In the second of these studies, van IJzendoorn, Kranen-burg, Zwart-Woudstra, van Busschbach, and Lambermon(1991) found an association between a mother's attachmentrepresentations and her sensitivity in interaction with her infantduring free pfay but found no such association for fathers. Fur-thermore, although the expected relation between sensitivitymeasures and Strange Situation classifications was found forfathers and mothers with their daughters, unexpectedly andanomalously, mothers of secure sons w ere found to be less sensi-tive than mothers of insecure sons. Finally, Ward and Carlson(1995) demonstrated that mothers classified as autonomous ina prenatal interview were m ore sensitive in an unstructured labo -ratory play session al 3 and 9 months of age than nonautono-mous m others; however, these measures of sensitivity were unre-lated to Strange Situation attachment classifications at 12months. The anomalies in the results of these three studies areparticularly troublesome because, of all the studies reviewed,only these incorporated all three of the variables comprising themediational model under evaluation. Therefore, van IJzen-doorn's (1995) provocative conclusion that maternal sensitivityplays only a modest role in mediating between maternal cogni-tion and the attachment relationship should be complementedby an analysis of observations of a single study in which a ssess-ments of maternal cognitions, moth er-infa nt interaction, and theattachment relationship arise from the same set of participants.

    van IJzendoorn's (1995) conclusion is qualified further by aconsideration of the difficulty of establishing a robust em piricallink between sensitivity and a ttachment security (see d e Wolff &van IJzendoorn, 1997; Goldsmith & Alansky, 1987, for re-view s). Even those studies that have reported a significant asso-ciation between maternal sensitivity and infant attachm ent secu-rity (e.g., Egeland & Farber, 1984; Grossmann, et al., 1985)have failed to replicate the magnitude of the effects reported byAinsworth, Bell, and Stayton (1971). Other research has failedto find the expected association (e.g ., Mangelsdorf, Gunnar,Kestenbaum, Lang, & Andreas, 1990; Rosen & Rothbaum, 1993;Seifer, Schiller, Sameroff, Resnick, & Riordan, 1996; Ward &Carlson, 1995). It has been suggested (e.g., Ainsworth & Mar-vin, 1995; Pederson & Moran, 1995a; Pederson, et al., 1990)that much of this difficulty in providing empirical confirmationof the central role of maternal sensitivity may be related to thesubstantial challenge involved in the valid description of infa nt-mother interactions, rather than indicating that maternal sensitiv-ity is less important to the development of attachment relation-ships than postulated by attachment theory. Brief observationsin the relatively stress-free environment of the home or in labora-tory play sessions may not provide observers with an adequatebasis for distinguishing the subtle variations in maternal behav-ior that are associated with variations in maternal sensitivity.

    Pederson and Moran (1995a) described three aspects of their

    procedures designed to increase the power of home observationsto distinguish crucial nuances of maternal and infant behavior.First, the observations are designed to simulate a busy householdin which the mother's attention is divided between her infant'ssignals and other tasks. Second, maternal sensitivity is assessedusing the Maternal Behavior Q set (Pederson & Moran, 1995b;Pederson et al., 1990), a set of 90 items descriptive of maternalinteractive behaviors that are ranked to reflect the observed inter-actions of a particular dyad. Among other advantages, the Q-setprocedures orient the observers to attachment-relevant maternalbehaviors. Third, the observers take extensive notes that areused as the basis for a debriefing interview following the visit.The debriefing interview is designed to highlight the infant'ssignals for proximity, affective co mm unication, and com fort andthe mother's responses to these signals. Using these procedures,Pederson and Moran (1996) provided empirical support forattachment theory's central prediction of a robust associationbetween a mother's sensitivity at home and her attachment rela-tionship with her infant. In this study, the correlation betweenmaternal sensitivity Q-sort scores derived from home observa-tions at 12 months and attachment security assessed in theStrange Situation at 18 months w as .60, which is a substantiallystronger relation than the average correlation of .24 reported inde Wolff and van IJzendoorn's (1997) review.

    A psychometrically sound assessment of parental sensitivityis a prerequisite to the evaluation of a model in which sensitivitymediates the association between adult representations of attach-ment and the parent-infant relationship. Any error in the mea-surement of sensitivity will attenuate the statistical strength ofthe mediated path (Baron & Kenny, 1986). Considerable effortshave been made to promote the accuracy and standardization ofassessing both attachment cognitions with the Adult AttachmentInterview and the attachment relationship w ith the Strange Situa-tion. In both cases, coders are expected to receive training andcertification to assure uniform coding standards, and codingis based on permanent records (i.e., transcripts for the AdultAttachment Interview and videotapes for the Strange Situation)that allow the repeated review of puzzling or ambiguous aspects.It is reasonable to assume that these efforts to assure standard-ized coding have enhanced the psychometric characteristics ofthese measures and contributed to the robustness of the associa-tion between Adult Attachment Interview autonomy and StrangeSituation security. In sharp contrast, m easures of parental sensi-tivity, including those used in van IJzendoorn's (1995) analyses,are diverse and have not been subject to a similar rigorousprocess of standardization. The mediating path is a product ofthe correlations between autonomy and sensitivity and betweensensitivity and security (Pedhazur, 1982). Thus, correlationsgreater than .7 for each pair would be required for this m ediatingroute to account for a share of the variance comparable withthat attributed to the direct path between autonomy and security(estimated by van IJzendoorn, 1995, to be .47).

    These various considerations suggest that the results of vanIJzendoorn's (1995) meta-analysis should not be taken as con-clusive evidence of the inadequacy of a developmental modelthat proposes that maternal sensitivity is the principal routelinking a parent's attachment-related cognitions and the qualityof the relationships with their infant. The primary goal of thecurrent study is to examine this issue directly using a research

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    3/9

    MATERNAL ATTACHMENT REPRESENTATIONS 927design that involves the assessment of all three factors with thesame participants and using observational measures that havebeen demonstrated in previous research to provide a valid andreliable assessment of meaningful variation in maternalsensitivity.

    In addition to this primary objective, we also designed thepresent study to provide more information about the relationbetween attachment behavior in the home and the attachmentrelationship as assessed in the Strange Situation. Home-basedattachment behavior was assessed by the Waters (1995) Attach-ment Q set and the Pederson and Moran (1995a) system forclassifying attachment relationships on the basis of home obser-vations. The Attachment Q set consists of 90 short descriptionsof infant behavior that are sorted to characterize the observedbehavior. Seifer et al. (1996), Vaughn and Waters (1990), andPederson and Moran (1996) reported that infants in secure rela-tionships in the Strange Situation had higher Attachment Q-sortsecurity scores compared with infants in nonsecure relationships(see van IJzendoorn, Vereijken, & Riksen-Walraven, in press,for a review). Using attachment theory and research, Pedersonand Moran (1995a) developed descriptive criteria for classifyingattachment relationships from home observations. Pedersonand Moran (1996) reported an overall congruence of 71 %between these classifications and Strange Situation attachmentclassifications.

    MethodParticipants

    Mothers of 60 physically healthy 12-month-old infants were recruitedfrom a registry of mothers who during their postpartum hospital stayhad expressed interest in participating in child development studies. Allinfants (33 female and 27 male) were observed at home with theirmothers between 12 and 16 months of age (M = 13 months ) . All ofthe mothers were Caucasian; their ages ranged from 21 to 44 years (M= 31.7years). Mothers' and fathers' highest level of education rangedfrom 9 to 27 years in school, averaging 15.2 years for fathers and 15.0years for mothers. Family income was approximately $45,000 Canadiandollars per year, which is slightly higher than the average family incomein London, Ontario.

    Home Visit ProcedureTwo female observers conducted semistructured home visits lastingapproximately 2 hr. Cognitive andmotor development of the infant wasassessed by one observer using the Bayley Scales of Infant Development(Bayley, 1969), while the mother completed the Attachment Q set (Wa-

    ters, 1995) under the supervision of the second observer. Following theseprocedures, the mother was interviewed to obtain information about thefamily's demographic background and the infant's health and develop-mental history. During the interview, the infant was free to play, allowingobservation of how both the mother and infant functioned while themother's attention wasdivided between the interviewer's questions andattending to her infant.Both observers kept running notes during the visit, describing infantand maternal behavior and interactions, devoting particular attention tothe infant's secure base behavior, affective sharing, fussiness, and resis-tance to interaction with the mother and to the mother's availability, herresponses to the infant's signals, and her monitoring the infant whileoccupied with the Q set and interview. Upon completion of the homevisit, the observers were interviewed by another researcher who had no

    knowledge of the dyad. In this debriefing interview, the observers wereasked to describe the visit using both their notes and recollections toclarify and consolidate the experiences of the visit. During the interview,the observers were careful not to reveal any identifying information suchas names or physical characteristics of the participants. The role of thenaive interviewer was to reflect his or her understanding gained fromthe observers' accounts of the infant, the mother, and the relationshipuntil the observers and interviewer were satisfied with the accuracy ofthe portrayal of the home visit. Finally, infant-mother dyads were classi-fied by consensus as secure, avoidant, or ambivalent according to thedescriptions of attachment relationships in the home developed by Peder-son and Moran (1995a) .

    In the Pederson andMoran (1995a) system, mothers in relationshipsclassified as secure monitor their infants and respond promptly andeffectively to their infants' signals for contact and reassurance. Infantsin secure relationships clearly show differential attachment behavior totheir mothers in comparison to the visitors, are comforted by their moth-ers ' responses, and engage in secure base behaviors. There are threesubclassifications of secure (B) relationships that parallel the Bl, B3,and B4 Strange Situation classifications. In Bl relationships, the infantsare sociable with both mother and visitors; however, they are moreresponsive to mother. Both mother and infant engage in distal affectivesharing. Secure base behavior is conspicuous for B3 infants; mothersare responsive anddelight in their infants. B4 infants and their mothersenjoy physical contact and are more attentive to each other than to thevisitors-Mothers in avoidant relationships are unresponsive to signals for com-fort: however, they may be intrusive in assisting on cognitive tasks.Infants in avoidant relationships may be more sociable with visitorsthan with their mothers, seldom seek contact with and attempt to beemotionally independent of their mothers, especially when distressed. InA1-ignoring relationships, both mothers and infants are independent ofeach other. The mothers' attention is focused on tasks or on interactionswith the visitors. Likewise, infants are engaged with the visitors or inexploration and play away from their mothers. In Al-teaching relation-ships, the interactions are focused on cognitive tasks, but there is surpris-ingly little affective sharing around achievem ents on these tasks. M othersoften respond to negative affect by distraction with food or toys. Mothersand infants in A2 relationships show mixed strategies during the homeobservations. Although the mothers often ignore their infants, especiallywhen involved with tasks, they are available to their infants at othertimes in the visit. Infants may approach their mothers more frequentlythan infants in other forms of avoidant relationships; however, they, likeother avoidant infants, attempt to soothe themselves when upset.

    Negative affect dominates the interactions for ambivalent (C) rela-tionships. Mothers in ambivalent relationships are unpredictable; some-times they may be responsive and at other times seemingly obliviousto their infants' signals for comfort. They may restrict their infants'exploration andappear to be overprotective. Infants in ambivalent rela-tionships are irritable and seek extensive contact with their mothers, butthe contact does not appear to be satisfying. They engage in limitedexploration away from their mothers. Like the Strange Situation classifi-cations, Cl infants appear to be angry and C2 infants appear helpless.(See the Appendix to Pederson and Moran, 1995a, for more details aboutthe classification procedures.) In the present study, each observer wasdebriefed by a different interviewer on 21 of the 60 home v isits; interraterclassification agreement for these dyads was 86 % (K = .74, z = 4.63, p< .001). Following the debriefing meeting, each observer independentlycompleted the Attachment Q sort, Maternal Behavior Q sort, and theAinsworth et al. (1971) ratings of maternal acceptance, accessibility,cooperation, and sensitivity.

    The Attachment Q set, Version 3.0 (Waters, 1995) w as used to describeinfant attachment behavior. An infant's security score was the correlationbetween the sort summarizing his or her behavior and the criterion sort

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    4/9

    928 PEDERSON, GLEASON, MORAN, AND BENTOfor a prototypically secure infant. Items indicative of security describedthe infant's secure base behavior; enjoyment of physical contact, andaffective sharing and the absence of fussy o r angry behaviors. The Mater-nal Behavior Q set, Version 2.0 (Pederson & Moran, 1995b) was usedto describe mothers' sensitivity during the home visits. The MaternalBehavior Q set consists of 90 items that are sorted in a manner analogousto the Attachment Q set. A mother's sensitivity score was the correlationbetween this sort and a sort describing a prototypically sensitive mother.Items indicative of sensitivity refer to the mother monitoring the infant'sbehavior and state, consistency and timing of her responses, andappro-priateness of her interventions.Strange Situation Procedure

    The infant-mother dyads were observed in the Ainsworth StrangeSituation within 2 weeks of the home visit. By using criteria describedin Ainsworth et al ., (1978 ) , the infant-mother attachment relationshipswere classified as secure, avoidant, or ambivalent by coders who hadattained reliability with the Carlson and Sroufe (1993) Strange Situationreliability videotapes. In brief, secure infants greeted or approachedtheir mothers in the reunion episodes and returned to play. Avoidantinfants were not overtly distressed by separations and ignored motherin the beginning of the reunion episodes. Ambivalent infants did notengage in exploratory play, were distressed during separation, and wereangry and petulant in the reunion episodes. Seventeen Strange Situationswere coded by a second coder with a 94% agreement with the first coder(K = .91 , z = 5.19, p < .001) . None of the coders participated inthe home observations for the dyads they coded. The disorganized-disoriented classification was not used because insensitivity is theoreti-cally not predictive of this form of nonsecure relationship (Main &Hesse, 1990).Adult Attachment Interview P rocedure

    Mothers were interviewed using the Adult Attachment Interview(George et al., 1985) in a third session conducted eidier at the universityor in the mother's home within 6 months of the Strange Situation. Thetranscripts were rated on the 18 scales described by Main and Goldwyn(1994) . These ratings together with Main and Goldwyn's descriptionsof autonomous, dismissing, or preoccupied attachment representationswere used to classify each transcript. Transcripts classified as autono-mous were rated as moderately or highly coherent in that general descrip-tions of childhood relationships with parents were convincingly sup-ported by episodic memories and the narrative was internally consistent.Dismissing transcripts were characterized by a positive depiction ofchildhood parental relationships that were either not supported or contra-dicted by episodic memories. Transcripts classified as preoccupied in-cluded passages indicating current anger when describing parental be-havior or vague and unfocused discussion.

    Each Adult Attachment Interview transcript was independently codedby at least two raters (for all transcripts, at least one of these initialcoders had achieved reliability with Mary Main and Erik Hesse). In thecase of interrater disagreements, transcripts were coded by a third rater,and a final classification was derived by consensus. The initial twocoders agreed on 80% of the three-category classifications (K = .66, z= 7.00, p < .001) .Given the multiple measures used in this study, care was exercised toensure the independence of each assessment. All names, occupations,and any other potentially identifying information were removed fromAdult Attachment Interview transcripts. Transcripts and Strange Situa-tion tapes of the same dyads were assigned different identification codesand were coded in different orders months apart.Results

    The demographic background variables of paternal education,maternal age and education, and family income were not sig-

    nificantly associated with the Adult Attachment Interview classi-fications, Strange Situation classifications, or the Q-sort sensitiv-ity scores. Thus, there was no statistical reason to use thesevariables as covariates in the subsequent analyses.

    Adult Attachment Interview and Strange SituationCo rrespondenceThe distribution of mothers classified as dismissing, autono-

    mous, and preoccupied in the Adult Attachment Interview andthe corresponding Strange Situation relationship classificationsof avoidant, secure, and ambivalent is presented in Table 1.The predicted association between mothers' Adult AttachmentInterview classifications and the dyads' Strange Situation classi-fications was tested using Cohen's kappa (Cohen, 1960) becausethis statistic controls for chance agreement based on the distribu-tion of participants across categories. For 44 of the 60 infant-mother dyads, Adult Attachment Interview and Strange Situationclassifications corresponded as predicted (73%; K = .56, z -5.89, p < .001). By using the autonomous-nonautonomousversus secure-nonsecure groupings, 48 of 60 infant-motherdyads were placed in concordant attachment categories (80%;K = .60, z = 4.63, p < .001).

    Adult Attachment Interview and Home ObservationsA r test was used to test the hypothesis that autonomous

    mothers were more sensitive than nonautonomous mothers. Con-sistent with this hypothesis, Q-sort sensitivity scores for autono-mous mothers (M = 0.57, SD - 0.39) were significantly higherthan for nonautonomous mothers (M = 0.27, SD = 0.63), r(44)= 2.24, p < .05, using an unequal variance test. For descriptivepurposes, sensitivity scores were also analyzed as a function ofthe three Adult Attachment Interview classifications and weresignificantly related to these classifications, F(2, 57) = 5.08, p< .01 . Duncan's multiple range post hoc tests (p < .05) revealedthat dismissing mothers (M = 0.11, SD - 0.67) were signifi-cantly less sensitive than autonomous (M = 0.57, SD = 0.39)and preoccupied mothers (M 0.56, SD = 0.49). The sensitiv-ity of autonomous and preoccupied mothers did not differ sig-nificantly. Similar results were found in parallel analyses usingthe Ainsworth ratings of sensitivity. The security scores derivedfrom the Attachment Q sort were not significantly related to theAdult Attachment Interview classifications, r (58) = 1.72, p >.05.

    Table 1Adult Attachment Interview and Strange SituationClassification Correspondence

    AdultAttachmentInterviewDismissingAutonomousPreoccupiedTotal

    Avoidant114116

    Strange Situation classificationSecure

    426232

    Ambivalent32712

    Total18321060

    Note, K = .56 (z = 5.89, p < .001).

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    5/9

    MATERNAL ATTACHMENT REPRESENTATIONS 929The frequency distribution of mothers' Adult Attachment In-terview classifications and the Pederson and Moran (1995a)home attachment classifications are shown in Table 2. Thirty-six of 60 infant-mother pairs were placed in matching homeattachment classifications (60%; K = .30, z = 3.23, p < . 01) .

    Thus, mothers' adult attachment classifications are related toinfant-mother attachment relationships assessed in both theStrange Situation and at home.Home Observations and Strange SituationClassifications

    Q-sort sensitivity scores of mothers in secure relationships(M = 0.68, SD = 0.27) were significantly higher than sensitivityscores of mothers in nonsecure relationships (M = 0.14, SD =0.27), ?(36) = 4.27, p < .001, using an unequal variance test,thus providing support for the basic hypothesis that sensitivityis related to attachment sec urity as assessed in the Strange Situa-tion. As with the Adult Attachment Interview classifications,sensitivity scores were also analyzed as a function of the threeattachment groups for descriptive purposes. These scores weresignificantly related to classifications in the Strange Situation,F ( 2 , 57) = 20.35, p < .00 1. Follow-up tests determined thatmothers classified as being in avoidant relationships in theStrange Situation were significantly less sensitive (M = - 0 . 1 3 ,SD = 0.65) than mothers in secure (M = 0.67, SD = 0.27) orambivalent (M = 0.50, SD = 0.37) relationships; the ambivalentand secure groups were not significantly different. Additionalanalyses using the Ainsworth ratings yielded comparable results.

    Security scores from the Attachment Q sorts (Waters, 1995)were significantly higher for infants classified as secure (M =0.47, SD = 0.30) than as nonsecure (M = 0.05, SD = 0.34)in the Strange Situation, /(58) = 5.12,/? < .001 . For descriptivepurposes, the security scores were analyzed as a function of thethree Strange Situation classifications, F(2, 57) = 13.29, p

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    6/9

    930 PEDERSON, GLEASON, MORAN, AND BENTO

    MaternalSensitivityFigure I. Path diagram for the relationship between Adult AttachmentInterview and Strange Situation classifications mediated by sensitivity.

    1996), firmly establishes the validity of our home observationprocedures and the associated measures of maternal sensitivity,and suggests that they form a credible basis for the evaluationof the role of maternal sensitivity as a mediator between adultattachment cognitions and the attachment relationship.

    A primary purpose of the present study was to examine thehypothesis that maternal sensitivity is a mediator that wouldstatistically explain the association betwe en m aternal attachmen trepresentations and the attachment relationship. We found asubstantially higher relationship between maternal sensitivityand attachment security in the Strange Situation than van Llzen-doorn (1995) used in his assessment of the model. Still, in bothvan IJzendoorn's analysis and in the present study, the magni-tude of the mediational path was virtually identical and ac-counted for less than 25 % of the association between representa-tional autonomy and attachment security.Although the results of this study failed to provide strongsupport for the mediational role of maternal sensitivity, a generaldevelopmental model that hypothesizes that the infant's interac-tional experience with the parent mediates the association be-tween the parent's attachment representations and the attach-ment relationship remains compelling. It is difficult to imaginedevelopmental processes in which the association between pa-rental represen tations and the manifest relationship be tween par-ent and infant is not mediated by some aspect of their interac-tions. An explanation of the weakness of the statistical supportfor the mediational model should first consider the adequacy ofthe assessment of maternal sensitivity. In the present study, the

    magnitude of the mediating path was no greater than that esti-mated by van IJzendoorn (1995), even though the relation be-tween maternal sensitivity and attachment security was substan-tially stronger than the estimate used by van IJzendoom, Thus,a more robust measure of maternal sensitivity failed to increasethe statistical strength of the mediated path relative to the directpath. It is still possible that maternal sensitivity does mediate theinfluence of maternal attachment representations on attachmentsecurity, but empirical substantiation would require home visitsrepeated over the first year (Ainswo rth et al., 1971; Ainsworth &Marvin, 1995). Isabella (1993) found that assessments of mater-nal behavior early in the first year were more clearly related toStrange Situation classifications than later assessments. He ar-gued that measures of sensitivity taken before the infant has

    had opportunity to accommodate to the mother provide a moreaccurate reflection of the mother's initial caregiving strategies.A plausible extension of this analysis is that these early assess-ments may also be more strongly related to maternal attachmentrepresentations.Our failure to fully support a mediational role for maternalsensitivity might be accounted for by considering that maternalsensitivity, as traditionally conceptualized in attachment theory,is but one of many aspects of the infant-mother interactionpotentially influenced by the mother's representations of attach-ment. It may be more fruitful to broaden our conceptualizationof the mediator variables in van IJzendoorn's (1995) modelrather than pursuing more refined measu res of maternal sensitiv-ity or rejecting the m odel itself. This approach c alls for recastingthe model to include multiple aspects of infant-mother interac-tion while retaining the model's essential structure. Maternalsensitivity, conceptualized as the mother's ability to recognizeand respond effectively to the child's needs and communicative

    signals, is retained as an important aspect of these interactions;however, other domains of mother-infant interactive behaviorcould be added to the mediational m odel. To function as media-tors, these variables must be related to the mother's attachmentrepresentations and to attachment security and be conceptuallydistinct from measures of sensitivity.A useful criterion for selecting likely variables for such explo-ration might be their relation to the conceptual link betweenmaternal cognitions and the attachment relationship. Cassidy(1994) considered emotion regulation to be one such link. Writ-ers from a variety of theoretical perspectives agree that theinfant-mother relationship provides a context for the socializa-tion of emotions (e.g., Calkins, 1994; Cassidy, 1994; Cole, Mi-chel, & Teti, 1994; Cummings & Davies, 1996; Gergely & Wat-son, 1996; Goldberg, MacKay-Soroka, & Rochester, 1994; Ma-latesta & Haviland, 1985; Stern, 1985), although there is lackof consensus on the particular socialization process. For exam-ple, Stern (1985) underscored the role of interactional syn-chrony, Malatesta and Haviland (1985) argued for the role ofthe mother's mirroring of the infant's affective states, and Ger-gely and Watson (1996) merged these two ideas and proposeda biofeedback model in which infants learn to identify theiraffective states because they detect the contingency betweentheir state and the mother's e xaggerated facial e xpression of thatstate. The mother's role is presumably influenced by the accu-racy of her perceptions (e.g., Emde, Osofsky, & Butterfield,1993), her socialization goals (e.g., Gottman, Katz, & Hooven,

    1996; Malatesta & Haviland, 1985), and her own chronic emo-tional state (Field, 1992; Zahn-Waxler & Wagner, 1993). It isreason able to expect that at least some of these maternal factorsmay also be related to her attachment representations (Cowan,1996).Cassidy (1994) explicitly described the relation betweenAdult Attachment Interview classifications and the socializationof emotions. She suggested that autonomous parents attend toand accept their infants' emotional expressions, thus facilitatingthe communicative aspects of affect. Bell and Ainsworth (1972)observed that infants whose mothers were responsive to theircrying in the first 6 mon ths had higher ratings of com municativecompetence by 12 months. In our home observations at 12months, crying is relatively rare. Several items in the Maternal

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    7/9

    MATERNAL ATTACHMENT REPRESENTATIONS 931Behavior Q set refer to maternal responses to crying, but be-cause of the relatively low frequency of crying in this study,these items make only a minor contribution to the summaryscore. Other research contexts and assessment procedures areneeded to assess the mother's perception of and responses toher infant's crying and other affective states.Other aspects of maternal interactions that are likely to medi-ate the impact of a parent's attachment representations on theattachment relationship may be those related to the role playedby these representations in structuring these interactions. Kaye's(1982) apprenticeship model of infant socialization is readilyadapted to a consideration of such interactions and the associ-ated developmental processes. In Kaye's model, the parent con-structs an interactional scaffold for the infant's social behavior.Although it was developed some time before the study of attach-ment cognitions was fully developed, it seems reasonable toextend the model to this area (Moran, Pederson, & Tarabulsy,199 6). As w ith emotional aspects of interaction, it is possible toconstruct differential predictions regarding the role of cognitivestructuring for parents in different attachment relationships withtheir infants. Autonomous parents would view their role as sup-porting their infants' establishment of individuality; early inter-actions would involve parents' increasing attributions of inde-pendent intentions, desires, and beliefs to the infant. As Fonagy(1994; Fonagy, Steele, Steele, Moran, & Higgitt, 1991) andMain (1991) have proposed, parental representations of attach-ment are likely to generate different attributions regarding theirinfants' representational processes and consequently influencetheir early interactions. According to this line of reasoning, ourobservations and research might profit by a closer examinationof such representational aspects of interaction.

    Although the results of the present study do not entirely sup-port the model originally proposed by van Uzendoorn (1995),they can be taken to suggest that further attempts to bridge thetransmission gap should focus on the development of descrip-tions that are capable of capturing aspects of the infant-motherinteraction beyond those encompassed by the traditional notionof sensitivity. Emotion regulation and cognitive scaffolding havebeen discussed here but are not intended to exhaust all possiblemediators.

    Additional domains of early mother-infant interaction mayenhance the statistical adequacy of a mediational model but willnot eliminate the difficulties assoc iated w ith their m easurem ent.Observations of interac tion, of necessity, are less structured thaneither the Adult Attachment Interview or the Strange Situationand thus are inevitably associated with higher levels of measure-ment error. Moreover, the establishment of an em pirical assoc ia-tion between attachment cognitions and relationships is facili-tated by the fact that both measures are categorical and havethe same structure, that is, two parallel categories designed toreflect the same underlying social developmental factors.

    As an alternative to interactive processes, it could be arguedthat the concordance of parent attachment representations andparent-infant attachment relationships is a reflection of sharedgenes. Such a hypothesis is not untenable, given the evidence ofgenetic correlates of parenting behavior (Braungart, Plomin, &Fulker, 1992; Plomin, Reiss, Hetherington, & Howe, 1994) andof personality, including measures of infant temperament andsociability (Robinson, Kagan, Reznick, & Corley, 1992; Rowe,

    1993). There undoubtedly may be temperamental influences oninfant-mother interactions, especially with regard to resistanceand distress (Calkins & Fox, 1992; Fox, 1995; Goldsmith &Alansky, 1987; Kagan, 1989; Kagan, Arcus, & Snidman, 19 93;Mangelsdorf et al., 1990; Moran & Pederson, in press; Vaughnet al., 1992). However, empirical evidence has been unable toconsistently relate infant temperament measures to attachmentsecurity in the Strange Situation (Mangelsdorf et al., 1990;Seifer et al., 1996; Sroufe, 1 985 ). Likewise, parental personalitymeasures are seldom directly related to Strange Situation mea-sures of attachment security (Mangelsdorf et al., 1990).

    Perhaps more important, even a model that emphasized therole of genetic factors would most likely involve mediationthrough some aspects of infant-mother interaction. That is, itis difficult to imagine a genetic link so direct that it was notmanifest in systematic variation in patterns of interaction aswell as in the attachment relationship itself. Thus, it seemsunlikely that genetically shared temperamental or personalityfactors would acc ount for a substantial portion of the associationbetween mother and infant working models of attachment.Conceptually, the most challenging finding in the presentstudy is that both autonomy and sensitivity we re strongly relatedto security but only modestly to each other. Thus, rather thanautonomy affecting attachment security because autonomousmothers are more sensitive, an appreciable portion of the influ-ence of autonomy on security appears to be independent of theinfluence of sensitivity. Conversely, there is a substantial portionof the influence of sensitivity on security that is surprisinglyindependent of the influence of autonomy on security. This pat-tern can also be seen in other measures derived from homeobservations in that both the home-based attachment relation-ship classifications and Q-sort security scores were strongly

    related to Strange Situation security classifications but weakly,if at all, to autonomy in the Adult Attachment Interview. Thesefindings are difficult if not impossible to reconcile with thetraditional explanation that maternal interactive behavior medi-ates the link between maternal representations of attachmentand the attachment relationship; they invite a substantial recon-ceptualization of the model. We are unable at this point to offera comprehensive alternative but suggest that any new modelmust acknowledge that the interrelations of mental representa-tions, the attachment relationship, and mother-infant interactionare fundamentally bidirectional and constitute a true controlsystem. The unidirectional causal model of representations in-fluencing interactions that affect the relationship may simply beinadequate to capture this system.In this study, we also confirmed previous rep orts (e.g ., Peder-son & M oran, 1996; Seifer et al., 1996; Vaughn & Waters, 1990)of the association between attachment security measured in thehome with the Attachment Q set and in the Strange Situation.Unlike the findings reported by Das Eiden, Teti, and Corns(1995) and by Posada, Waters, Crowell , and Lay (1995) with3- to 5-year old children, the Q-sort measure of attachmentsecurity was not significantly related to maternal attachmentrepresentations in the present infant sample, suggesting thatthere may be d evelopmental shifts in the role of maternal attach-ment representations in the expression of security at home orin other nonstressful interactions. The significant associationbetween classifications of attachment relationships expressed at

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    8/9

    932 PEDERSON, GLEASON, MORAN, AND BENTOhome using the Pederson and Moran (1995a) descriptions andin the Strange Situation confirmed the validity of our homeclassification procedures. These home-based relationship classi-fications were also modestly related to the mother's attachmentrepresentations derived from the Adult Attachment Interview. Itis hoped the validity of both the interactional and relationshipassessments of attachment in more naturalistic settings will stim-ulate research in these contexts.

    ReferencesAinsworth, M. D. S. (1 98 2). Attachment; Retrospect and prospect. InC. M. Parkes & J. Stevenson-Hinde (Eds.), The place of attachment

    in human behavior (pp. 3-30). New York: Basic Books.Ainsworth, M . D. S., Bell, S. M. V., & Stayton, D. J. (1 97 1). Individualdifferences in Strange Situation b ehavior of one-year-olds. In H. R.Schaffer (Ed.), The origins of human social relations (pp . 17 -57 ) .New %rk: Academic Press.Ainsw orth, M . D. S., Blehar, M. C , Waters, EM & Wall, S. (1978).

    Patterns of attachmen t: A psycholog ical study of the Strange Situa-tion. Hillsdale, NJ: Erlbaum.Ainsworth. M. D. S., & Eichberg, C. G. (1 99 1) . Effects on infan t-mother attachment of mother's unresolved loss of an attachment figureor other traumatic experience. In P. Marris, J. Stevenson-Hinde, & C.Parkes (Eds.) Attachment across the life cycle (pp . 160 -183) . New\brk: Routledge,Ainsworth, M. D . S., & Marvin, R. S. (199 5). O n the shaping of attach-ment theory and research: An interview with Mary D. S. Ainsworth(Fall, 1994). Monographs of the Society for Research in Child Devel-

    opment, 60(2-3, Seria l No. 244) , 3-21.Baron, R. M., & Kenny, D. A. (198 6). The moderator-m ediator variabledistinction in social psychological research: Conceptual, strategic, andstatistical considerations. Journal of Personality and Social Psychol-ogy, 51, 1173-1182 .

    Bayley, N. (1969). The Bay ey Scales of Infant Development. New York:Psychological Corporation.Bell, S. M., & Ainsworth, M. D. S. (1972). Infant crying and maternalresponsiveness. Child Development, 43, 1171-1190 .Benoit, D., & Parker, K. C. H. ( 19 94 ). Stability and transm ission ofattachment across three generations. Child Development, 65, 1 4 4 4 -1456.Braungart, J. M., Plomin, R., & Fulker, D. W. (1992). Genetic mediationof the home environment during infancy: A sibling adoption study ofthe HOME. Developmental Psychology, 28, 1048-1055 .Calkins, S. D. (1 99 4) . Origins and outcom es of individual differencesin emotion regulation. Monograph.1; of the Society for Research in

    Child Development, 59(Seria l No. 240) , 53-72.Calkins, S. D., & Fox, N. A. ( 19 92 ). The relations among infant temper-ament, security of attachment, and behavioral inhibition at twenty-four months. Child Development, 63, 1456-1472 ,Carlson, E., & Sroufe, L. A. (1993, Spring). Reliability in attachmentclassifications. News letter of the Society for Researc h in Child Devel-

    opment, pp. 4, 12.Cassidy, J. (1994). Emotion regulation: Influences of attachment rela-tionships. Monographs of the Society for Re search in Child De velop-ment, 59(Seria l No. 240) , 228-249.Cohen, J. (I960). A coefficient of agreement for nominal scales. Educa-tional and Psychological Measurement, 20, 3 7 - 4 6 .Cole, P. M , Michel, M. K., & Teti, L. O. (199 4). T he development ofemotion regulation and dysregulation: A clinical perspective. Mono-graphs of the Society for Research in Child Develop ment, 59 (SerialN o. 240) , 73 -100 .Cow an, P. A. (1 99 6) . Meta-thoughts on the role of meta-emotion in

    children's development: Comment on Gottman et al. (1996). Journalof Family Psychology, 10, 2 7 7 - 2 8 3 .

    Cumm ings, M., & Davies, P. (19 96 ). Em otional security as a regulatoryprocess in normal development and the development of psychopathol-ogy. Development and Psychopathology, 8, 1 2 3 - 1 3 9 .

    Das Eiden, R., Teti, D. M., & Corns, K.M. (1995). Maternal workingmodels of attachment, marital adjustment, and the parent-child rela-tionship. Child Development, 66, 1504-1518 .

    de Wolff, M. S ., & van LTzendoorn, M. H. (1 99 7) . Sensitivity and attach-ment: A meta-analysis on parental antecedents of infant attachment.Child Development, 68, 5 7 1 - 5 9 1 .

    Egeland, B., & Farber, A. E. (1 98 4). Infant-mother attachment: Factorsrelated to its development and changes over time. Child Development,55 , 7 5 3 - 7 7 1 .

    Emde, R . N., Osofsky, J. D., & Butterfield, P. M. (19 93) . The IFEELpictures: A new instrument for interpreting emotions. Madison, CT:International Universities Press.

    Field, T. (1992). Infants of depressed mothers. Development and Psy-chopathology, 4, 4 9 - 6 6 .

    Fonagy, P. (1994, October). The significance of the development ofmetacog nitive contro l over mental representations in parenting andinfant development. Paper presented at the New %rk PsychoanalyticSociety, New "fork.

    Fonagy, P., Steele, H., & Steele, M. (1991). Maternal representationsof attachment during pregnancy predict the organization of infant-mother attachment at one year of age. Child Development, 62, 891 905.

    Fonagy, P.. Steele, M., Steele, H.. Moran, G. S., & Higgitt, A. C. (1991).The capacity for understanding mental states: The reflective self inparent and child and its significance for security of attachment. InfantMental Health Journal, 13, 2 0 0 - 2 1 7 .

    Fox, N. A. (1995). Of the way we were: Adult memories about attach-ment experiences and their role in determining infant-parent relation-ships: A commentary on van Uzendoorn (1995). Psychological Bulle-tin, 117, 4 0 4 - 4 1 0 .

    George, C , K aplan, N., & Main, M. (19 85 ). Adult Attachment Interview.Unpublished manuscript, Department of Psychology, University ofCalifornia, Berkeley.

    Gergely, G., & Watson, J. S. (1996). The social biofeedback theory ofparent affect mirroring: The development of emotional self-awarenessand self-control in infancy. International Journal of Psychoanalysis,77 , 1181-1212 .

    Goldberg, S., MacKay-Soroka, S., & Rochester, M. (1994). Affect,attachment, and maternal responsiveness. Infant Behavior and Devel-opment, 17, 3 3 5 - 3 3 9 .

    Goldsmith, H. H., & Alansky, J. A. (1987). Maternal and infant temper-amental predictors of attachment: A meta-analytic review. Journal ofConsulting and Clinical Psychology, 55, 8 0 5 - 8 1 6 .

    Gottman, J. M., Katz, L. F., & Hooven, C. (1996). Parental meta-emo-tion philosophy and the emotional life of families: Theoretical modelsand preliminary data. Journal of Family Psychology, 10, 2 4 3 - 2 6 8 .

    Grossmann, K., Fremmer-Bombik, E., Rudolph, J., & Grossmann, K. E.( 1 9 8 8 ) . Maternal attachment representations as related to patterns ofinfant-mother attachment and maternal care during the first year. InR. A. Hinde & J. Stevenson-Hinde (E ds.), Relationships within fami-lies (pp. 241-260). Oxford, England: Oxford University Press.Grossm ann, K., Grossm ann, K. E., Spangler, G., Suess, G., & Unzner,L. (1985). Maternal sensitivity and newborns' orientation responsesas related to quality of attachment in Northern Germany. Monographsof the Society for Research in Child Development, 50(1-2, Ser ia lNo. 209) , 233 -256 .

    Isabella, R. (1993). Origins of attachment: Maternal interactive behavioracross the first year. Child Development, 64, 605-624 .

  • 7/30/2019 Maternal Attachemnt, Maternal Sezitivity and Infant Mother Intractions

    9/9

    MATERNAL ATTACHMENT REPRESENTATIONS 933Kagan, J. (1989). Temperamental contributions to social behavior.

    American Psychologist, 44, 6 6 8 - 6 7 4 .Kagan, J., Arcus, K., & Snidman, N. (1993). The idea of temperament:Where do we go from here? In R. Plomin & G. E. McClearn (Eds.),

    Nature, nurture, and psychology (pp. 197-210) , Washington, DC:American Psychological Association.Kaye, K. (198 2) . The mental and social life of babies. Chicago: Univer-sity of Chicago Press.Main, M. (1991). Metacognitive knowledge, metacognitive monitoring,and singular (coherent) vs. multiple (incoherent) models of attach-ment: Findings and directions for future research. In P. Harris, J.Stevenson-Hinde, & C . Parkes (Eds.) Attachment across the lifecycle(pp . 129-159). New York: Routledge.Main, M., & Goldwyn, R. (1994 ) . Adult attachment scoring and classi-

    fication systems. Unpublished manuscript, Department of Psychology,University of California, Berkeley.Main, M., & Hesse, E. (1990). Parents' unresolved traumatic experiencesare related to infant disorganized attachment status : Ts frightening and/or frightened parental behavior the linking mechanism? In M.

    Greenberg, D. Cicchetti, & M. Cummings (Eds.), Attachment in thepreschool years (pp. 121-160). Chicago: Chicago University Press.Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, child-hood, and adulthood: A move to the level of representation. Mono-

    graphs of the Society for Research in Child Development, 5 0 ( 1 - 2Seria l No. 209) , 66-104.Malatesta, C. Z., & Haviland, J. M. ( 19 85 ). Signs, symbols, and social-ization. In M. Lewis & C. Saarni (Eds.), The socialization of emotion(pp . 89-115) . New York; Plenum Press .Mangelsdorf, S., Gunnar, M., Kestenbau m, R., Lang, S., & Andreas , D.( 1 9 9 0 ) . Infant proneness-to-distress temperament, maternal personal-ity, and mother-infant attachment: Associations and goodness of fit.Child Development, 61, 8 2 0 - 8 3 1 .

    Moran, G., & Pederson, D. R. (in pre ss). Proneness to distress andambivalent mother-infant relationships observed in the home. InfantBehavior and Development.Moran, G., Pederson, D. R., & Tarabulsy, G. M. (1996). Le role de latheorie d'attachment dans l'analyse des interactions mere-enfant du-rant la petite enfance: Descriptions precises et interpretations signifi-catives [The role of attachment theory in the analysis of mother-infantinteractions: Targeted descriptions and meaningful interpretations ] . InG. M. Tarabulsy & R. Tessier (Series & Volume Eds.), D'Enfance:Vol. 1. Families et meres: le diveloppement social de Venfant encontexte. Quebec City, Quebec, Canada: Les Presses de TUniversitedu Quebec.

    Pederson, D. R., & Moran, G. (1995a). A categorical description of in-fant-m other relationships in die home and its relation to Q-sort m easuresof infant-mother interaction. Monographs of the Society for Researchin Child Development, 60(2-3, Serial No. 244), 111-145.

    Pederson, D. R., & Moran, G. (1995b). Maternal Behavior Q-set [Ap-pendix B ]. Monographs of the Society for Research in Child Develop-ment, 60(2-3, Seria l No. 244) , 247-254.

    Pederson, D. R., & M oran, G. (1 99 6). E xpressions of the attachmentrelationship outside of the Strange Situation. Child Development, 67,9 1 5 - 9 2 7 .Pederson, D. R., Moran, G., Sitko, C , Campbell, K., G hesquire, K., &Acton, H. (1990). Maternal sensitivity and the security of infant-mother attachment: A Q-sort study. Child Development, 61, 1 9 7 4 -1983.Pedhazur, E. J. (1982). Multiple regression in behavioral research: Ex-

    planation and prediction (2nd ed .). Fort Worth, TX: H olt, Rinehart &Winston.

    Plomin, R., Reiss, D., Hetherington, E. M., & Howe, G. W. (1 99 4).Nature and nurture: Genetic contributions to measures of family envi-ronment. Developmental Psychology, 30, 3 2 - 4 3 .Posada, G., Waters, E., Crowell, J. A., & Lay, K.-L. (1995). Is it easierto use a secure mother as a secure base? Attachment Q-sort correlatesof the Adult Attachment Interview. Monographs of the Society for

    Research in Child Development, 60(2-3, Seria l No. 244) , 133-145.Robinson, J. L., K agan, J., Reznick, J. S., & Corley, R. (1 99 2). Theheritability of inhibited and uninhibited behavior: A tw in study. Devel-opmental Psychology, 28, 1030-1037 .

    Rosen, K. S., & Rothbaum, F. (1993). Quality of parental caregivingand security of attachment. Developmental Psychology, 29, 3 5 8 - 3 6 7 .Rowe, D. C. (1993). Genetic perspectives on personality. In R. Plomin &G. E. McClearn (Eds.), Nature, nurture, and psychology (pp. 179195). Washington, DC: American Psychological Association.Seifer, R., Schiller, M., Sameroff, A., Resnick, S., & Riordan, K. (1996).Attachment, maternal sensitivity, and infant temperament during thefirst year of life. Developmental Psychology, 32, 1 2 - 2 5 .Sroufe, L. A. (1985). Attachment classification from the perspective of

    infant-caregiver relationships and infant temperament. Child Devel-opment, 56, 1-14 .Stern , D. N. (198 5) . The interpersonal world of the infant: A view

    from psychoanalysis and developmental psychology. New York: BasicBooks.van IJzendoorn, M. H. (19 95 ). Adult attachment representations, paren-tal responsiveness, and infant attachment: A meta-analysis on the pre-dictive validity of the Adult Attachment Interview. Psychological Bul-

    letin, 117, 1-17.van IJzendoorn, M . H., Kranenburg, M . J., Zw art-Woudstra, H. A., vanBusschbach, A. M., &Lambermon, M. W. E. (1991). Parental attach-ment and children's socio-emotional development: Some findings onthe validity of the Adult Attachment Interview in the Netherlands.

    International Journal of Behaviora l Developme nt, 14, 3 7 5 - 3 9 4 .van IJzendoorn, M. H., Vereijken, C. M. J, L., & Riksen-W alraven,

    J. M. A. (in press). Is the attachment Q-sort a valid measure of attach-ment security in young children? In B. E. Vaughn & E. Waters (Eds.),Patterns of secure base behavior: Q-sort perspectives on attachmentand caregiving. Hillsdale, NJ: Erlbaum.

    Vaughn, B. E., Stevenson-Hinde, J., Waters, E., Kotsaftis, A ., Lefever,G.B., Shouldice, A., Trudel, M., & Belsky, J. (1992). Attachmentsecurity and temperament in infancy and early childhood: Some con-ceptual clarifications. Developmental Psychology, 28, 4 6 3 - 4 7 3 .Vaughn, B. E., & Waters, E. (19 90 ). Attachment behavior at home andin the laboratory: Q-sort observations and Strange Situation classifi-cations of one-year-olds. Child Development, 61, 1 9 6 5 - 1 9 7 3 .Ward, M. J., & Carlson, E. A. (19 95) . Associations among adult attach-ment representations, maternal sensitivity, and infant-mother attach-ment in a sample of adolescent mothers. Child Development, 66, 6 9 -79 .Waters, E. (19 95 ). The Attachment Q-set, Version 3.0. [Appendix A] .Monographs of the Society for Research in Child Development, 60(2-

    3, Seria l No. 244) , 234-246.Zahn-Waxler, C , & Wagner, E. (1 99 3) . Careg ivers' interpretations ofinfant emotions: A comparison of depressed and well mothers. InR. N. Em de, J. D. Osofsky, & P. M. B utterfield ( Ed s.) , The IFEEL

    pictures: A new instrument for interpreting emotions (pp . 175 -184) .Madison, CT: International Universities Press.Received November 22, 1996Revision received January 15, 1998Accepted January 15, 1998