Child: care, health and development Original Article

doi:10.1111/j.1365-2214.2010.01135.x

Adoption and attachment theory the attachment models of adoptive mothers and the revision of attachment patterns of their late-adopted children
C. S. Pace and G. C. Zavattini
Sapienza, University of Rome, Rome, Italy Accepted for publication 23 May 2010

cch_1135

82..88

Abstract
Objective This study examined the attachment patterns of late-adopted children (aged 47) and their adoptive mothers during the rst 7- to 8-month period after adoption and aimed to evaluate the effect of adoptive mothers attachment security on the revision of the attachment patterns of their late-adopted children. Design We assessed attachment patterns in 20 adoptive dyads and 12 genetically related dyads at two different times: T1 (time 1) within 2 months of adoption and T2 (time 2) 6 months after T1. Methods The childrens behavioural attachment patterns were assessed using the Separation-Reunion Procedure and the childrens representational (verbal) attachment patterns using the Manchester Child Attachment Story Task. The attachment models of the adoptive mothers were classied using the Adult Attachment Interview. Results We found that there was a signicant enhancement of the late-adopted childrens attachment security across the time period considered (P = 0.008). Moreover, all the late-adopted children who showed a change from insecurity to security had adoptive mothers with secure attachment models (P = 0.044). However, the matching between maternal attachment models and late-adopted childrens attachment patterns (behaviours and representations) was not signicant. Conclusions Our data suggest that revision of the attachment patterns in the late-adopted children is possible but gradual, and that the adoptive mothers attachment security makes it more likely to occur.

Keywords adoption, adoptive mothers, attachment patterns, Internal Working Model, late-adopted children Correspondence: Cecilia S. Pace, Department of Dynamic and Clinical Psychology, University of Rome, via degli Apuli 1, Rome, 00185 Lazio, Italy E-mail: ceciliapace@fastwebnet.it

Introduction
Day-to-day interactive experiences between the child and caregiver are incorporated in Internal Working Models (IWMs) of attachment, which are representations of self, other and the relationship between self and other (Bowlby 1969, 1973, 1980; Bretherton & Munholland 2008). The IWMs are initially relationship specic but then they generalize and are the mechanisms through which early experience inuences later development, regulating affective, social and cognitive functions in adults (Cassidy & Shaver 2008; Hamilton 2000; Waters et al. 2000). IWMs serve as lters for understanding close rela-

tionships and guide attention, perception, memory and language of attachment-related experiences (Crowell et al. 2002; Mikulincer & Shaver 2008). IWMs work in an automatic way and are highly stable unless there are such signicant changes in the caregiving environment to bring about a revision of IWMs (Steele et al. 2008). Lateplaced children are those who have been adopted after 12 months of age and who have suffered at least one relationship rupture of an attachment gure (Howe 1997). They usually have experienced a wide range of difculties with their primary caregivers, ranging from neglect, rejection, role reversal, to physical, psychological and sexual abuse. Because of early

82

2010 Blackwell Publishing Ltd

Attachment in late-adopted children

83

adverse experiences late-adopted children show insecure or disorganized attachment IWMs which lead them to exhibit contradictory and deant behaviours towards their adoptive parents and severely challenge the chance for creating affectional bonds with their new caregivers (Vorria et al. 2003). Adoption of late-placed children has often been described as a natural experiment (Rutter et al. 2004; Van IJzendoorn & Juffer 2005; Van IJzendoorn et al. 2005) able to alter the course of the lives of traumatized children. In fact, being adopted, could offer to abandoned children with a troubled past the possibility to build relationships with new caregivers. These new attachments could help children revise their early insecure and/or disorganized IWMs, changing them into secure IWMs, thus breaking the intergenerational transmission of the cycle of abuse and re-orienting the development pathways of late-placed children (Schoeld & Beek 2005). Moreover, some studies with lateplaced children who were located in adoptive or foster families showed that children adopted by mothers with secure IWMs displayed more secure attachment than children who were placed with women characterized by insecure IWMs (Dozier et al. 2001; Verissimo & Salvaterra 2006; Steele et al. 2008). In the last 20 years in Italy the age of children being adopted has increased, mostly ranging from 3 to 8 years. While previous adoption studies performed in Italy assessed late-adopted childrens wellbeing a long time after adoption, the present study selected a sample of new adoptive families at the beginning of the childrens placement, when the inuence of negative pre-adoption experiences should still be traceable. This study is a brief-longitudinal research consisting of two measurements: the rst measurement (T1) within a maximum of 2 months after adoption (average 40 days after placement); the second measurement (T2) 6 months later. Our rst hypothesis is that a signicant revision intended as a transformation from insecurity towards security of the attachment patterns of late-placed children could be highlighted. The second hypothesis is that such a change may be more frequent for those children placed with adoptive mothers with secure IWMs. Third, we hypothesized links between representations of attachment among adoptive mothers secure vs. insecure and the attachment patterns of late-adopted children both at the attachment behavioural and representational levels.

late-placed children aged 4 to 7, without a diagnosis of psychiatric disorders, mental retardation or physical disabilities and their 15 adoptive mothers (n = 35). The mean age of the children (9 male and 11 female) was 71.65 months (SD = 12.4), almost 6 years old. Some 75% (n = 15) of the children come from a foreign country (international adoption), while 25% (n = 5) of them were adopted within their own country (national adoption). Within the sample, 85% of the children (n = 17) had been institutionalized for periods from 7 to 67 months (mean = 27.88 months, SD = 20.39) before adoption. The adoptive mothers mean age was 44.13 (SD = 4.05) years, they had on average 15.2 (SD = 3.12) years of education (53.3% of mothers had university level education). We used a control sample which consisted of 23 persons, 12 children and their 11 biological mothers with the same characteristics of the adopted sample (age, gender, etc.) in order to verify the effects of practice of the Separation-Reunion Procedure (SRP) (Main & Cassidy 1988, see below).

Measures
Child attachment behaviour
The SRP (Cassidy 1988; Main & Cassidy 1988) is a laboratory observation procedure aimed at identifying and classifying differing attachment behavioural patterns of the child to his or her parent at pre-school and school-age. The child experiences two episodes of separation from and reunion with his or her caregiver: the rst lasts 1015 min and the second lasts 4560 min. The system is based on a close analysis of the childs response to reunion with the parent during the two reunions (3 to 5 min) following the two separations.

Child attachment representation

The Manchester Child Attachment Story Task (MCAST, Goldwyn et al. 2000; Green et al. 2000) is a doll-play vignette completion method to enable identication and detailed classication of internal representations of attachment relationships in young pre-school and school-age children. The interviewer shows the child a dolls house and a set of dolls, the child has to choose the doll which represents himself (or herself) and the doll which represents his or her mother. The four core vignettes are intended to activate the attachment system and show the following attachment-related themes: nightmare, hurt knee, tummy-ache, losing caregiver in a shopping centre. As the MCAST requires that the children understand and speak Italian well, it was administered only during the second

Methods
Sample
The total number of participants in this pilot-study was 58 individuals. The adopted sample consisted of 20 recently

2010 Blackwell Publishing Ltd, Child: care, health and development, 37, 1, 8288

84

C.S. Pace and G.C. Zavattini

Table 1. Attachment classications in children and adults Children: the SRP and MCAST classications Secure (B) Insecure avoidant (A) Insecure ambivalent (C) Disorganized disorientated (D) Mothers: the AAI classications Secure free/autonomous (F/A) Insecure dismissing (Ds) Insecure preoccupied (P) Unresolved with respect to loss and abuse (U)

AAI, Adult Attachment Interview; MCAST, Manchester Child Attachment Story Task; SRP, Separation-Reunion Procedure.

The inter-coder agreement on the four classications ranged from 88.5% (kappa = 0.77) for the AAI, to 93.4 (kappa = 0.90) for the MCAST and to 93.8% (kappa = 0.91) for SRP. The childrens SRP and MCAST classications had no association to the following variables: child age, child gender, type of adoption (national/international) and length of institutionalization. The mothers AAI classications had no association to mother age, profession, years of education and years of marriage.

Procedure
observation, when the late-adopted children specically those who came from foreign countries had acquired a better competence in Italian.1 Adoptive families were recruited from National Health Services designed to select adoptive parents and evaluate familys adjustment post-adoption, and Agencies for International Adoption. Two mother-child observation sessions took place in the video-recording laboratory at Sapienza, University of Rome. Both sessions lasted roughly 1 h and a half and were videorecorded. During both sessions, the child-mother dyads were observed within the SRP. During T1 each mother was administered the AAI, which was tape-recorded and transcribed verbatim. During T2, each child had to complete the MCAST. At the beginning of the rst session, written informed consent was obtained from the mothers after the procedures regarding their own and their childrens participation in the study had been fully explained to them (Law 675/96 on the Protection of the individual and other aspects related to the management of personal data).

Mothers attachment models

The Adult Attachment Interview (AAI, George et al. 1984, 1985, 1996; Main et al. 2002; Bakermans-Kranenburg & Van IJzendoorn 2009) is a well-known semi-structured interview consisting of 20 questions and it lasts for approximately 1 h. The AAI is able to assess the attachment representation in adults regarding their early attachment experiences. During the AAI the mothers are asked to talk about their relationships with their attachment gures during childhood, and to support these descriptions with specic memories; to refer to attachmentrelated topics, such as crisis, illness, separation, loss and abuse; to reect on how attachment experiences had inuenced their adult personality.

Statistical methods
Statistical analyses were run using Statistical Package for the Social Science (SPSS, Version 12.0). The data analysis was carried out by categorizing children and mothers into secure (B and F) and insecure (A, C, D and Ds, P, U) groups. Our data were analysed using non parametric methods, adequate for nominal and categorical data: Fishers exact test, McNemar test and rphi (Siegel & Castellan 1988).

Measure coding
To guarantee the accuracy of the measurement, each of the attachment classication instruments SRP, AAI, MCAST were codied by two independent, double-blinded and wellqualied raters, each of whom was unaware of the coding assigned by the other to the same instrument (in this sense the rater was blind). Then we applied an index of raters agreement, the Cohen K, which calculated the inter-raters reliability. The cases of disagreement were codied by a third rater and they were discussed together by the three raters. All the coding systems required both 1 to 9 scores on ordinal scales and the attribution of three main classications, plus one secondary category (disorganized and unresolved), as reported in Table 1.
We administered the Peabody Picture Vocabulary Test-Revised (P.P.V.T.-R., Dunn & Dunn 2000) to check the childrens comprehension of the Italian language.
1

Results
The revision of attachment behavioural patterns in late-placed children T1
At T1 the distribution of the attachment behavioural categories in the SRP of the 20 late-placed children was 15% secure, 50% avoidant, 35% ambivalent. Biological children, instead, showed 67% secure patterns 25% avoidant patterns and 8% ambivalent

2010 Blackwell Publishing Ltd, Child: care, health and development, 37, 1, 8288

Attachment in late-adopted children

85

patterns. Overall in the adopted sample 85% of patterns were insecure compared with the 33% in the control sample. At the beginning of the adoption, the attachment behavioural organizations of the late-adopted children, recently living with their new caregivers, are signicantly more insecure than those of their peers who grew up in their own families of origin (Fishers exact test, P = 0.005).

The inuence of the adoptive mothers attachment models

The AAI classications of the adoptive mothers were distributed as follows: 67% secure, 20% dismissing, 13% preoccupied (including one mother who was assigned to the unresolved category). The results showed that adoptive mothers with a secure AAI classication are more likely to have children who get better- in terms of being able to transform their attachment behavioural strategies from insecure to secure than mothers assessed as insecure by the AAI, whose children remain stable, maintaining in T2 their previous insecure patterns of attachment (Fishers exact test, P = 0.044). The second hypothesis therefore seemed to be supported as well (see Table 2).

T2
After 6 months, the frequencies of child attachment categories in the SRP were 50% secure, 30% avoidant, 20% ambivalent in the late-placed sample and 67% secure and 33% avoidant in the control sample. The Fishers Exact Test no longer indicated a signicant difference (P = 0.30) between adopted/non adopted groups on the secure/insecure classications.

Association between maternal attachment and children attachment Mothers attachment models and childrens attachment behaviours
We pointed out that the matching between the AAI categories of the adoptive mothers and the SRP classications of the lateplaced children increased from 25% (rphi = -0.34, P = 0.13, n.s.) in T1 to 60% (rphi = 0.22, P = 0.33, n.s.) in T2, considering the two-way coding system. This result highlighted a signicant growing increase of overlap (35%, McNemar test, P = 0.008) between the attachment patterns in the adoptive dyads. See Fig. 1.

The change between T1 and T2

Seven out of the 17 late-placed children who were assessed as insecure in SRP in T1 were classied as secure in T2. Hence the SRP insecure categories were reshaped in 41% of cases. We found a signicant change between classications in the SRP from T1 to T2 among the adopted sample (McNemar test, P = 0.008). Therefore our rst hypothesis seems to be conrmed (see Table 2). On the other hand, the attachment classications in SRP-T1 and SRP-T2 were highly stable (83%) within the group of children raised by their biological parents, and the analyses indicated no signicant differences between T1 and T2. These different results in the two samples the change of the attachment classications between SRP-T1 and SRP-T2 in the adopted sample vs. stability of the attachment classications between SRP-T1 and SRP-T2 in the control sample highlighted no pre-test effects and thus the attachment behavioural patterns observed in SRP-T2 did not depend on the practice and the familiarity learnt during SRP-T1.
Table 2. AAI categories of adoptive mothers and SPR categories of late-placed children in T1 and T2 AAI

Mothers attachment models and childrens attachment representations

Frequency distribution of categorical A, B, C and D ratings elicited by the MCAST among the lateplaced children was as follows: 42% secure, 16% avoidant, 5% ambivalent and 37% disorganized.

Late-placed children SRP T1 B A 8 (40%) 2 (10%) 0 10 (50%) C 5 (25%) 2 (10%) 0 7 (35%) Total 14 (70%) 4 (20%) 2 (10%) 20 (100%) SRP T2 B 8 (40%) 0 2 (10%) 10 (50%) A 4 (20%) 2 (10%) 0 6 (30%) C 2 (10%) 2 (10%) 0 4 (20%) Total 14 (70%) 4 (20%) 2 (10%) 20 (100%)

Adoptive mothers F/A 1 (5%) Ds 0 P 2 (10%) Total 3 (15%)

AAI, Adult Attachment Interview; Ds, dismissing; F/A, free/autonomous; MCAST, Manchester Child Attachment Story Task; P, preoccupied; SRP, Separation-Reunion Procedure.

2010 Blackwell Publishing Ltd, Child: care, health and development, 37, 1, 8288

86

C.S. Pace and G.C. Zavattini

Figure 1. The increase of concordance between AAI categories of adoptive mothers and SRP categories of late-placed children from T1 to T2.

The matching between AAI maternal classications and the lateplaced childrens attachment representation in the MCAST was 63.2% with respect to secure/insecure distribution and was not signicant (rphi = 0.35, P = 0.13, n.s.). The third hypothesis on the possible links between models of attachment among adoptive mothers and their late-adopted children did not seem to be conrmed. In this study is was not possible to utilize the MCAST longitudinally, but it was only possible to correlate it with the AAI of the mother. This rst measurement of the MCAST will be useful as a term of comparison in future follow-ups.

Discussion
In our research the majority of the late-placed children showed attachment insecurity at the beginning of the placement and their insecurity was greater compared with the children who were brought up by their natural parents. After less than a year of adoption, we found a signicant enhancement of lateadopted childrens attachment security. It is therefore possible that in circumstances such as adoption characterized by stability and continuity, it is feasible to reshape the insecure IWM of traumatized children in the direction of security. Adoption of late-placed children could be considered the most radical and powerful intervention for abandoned and troubled children with respect to other kinds of social interventions (Van IJzendoorn & Juffer 2006; Rutter et al. 2007). All the children who enhanced their attachment security were adopted by mothers with secure AAI classications, whereas children placed with insecure mothers remained in their initial

classications, mostly insecure. These results could indicate that secure attachment IWMs of adoptive mothers, who reinforce positive interactions with their children and value attachment experiences, could enable insecure late-adopted children to revitalize affect in the parent-child relationships and positively reactivate their attachment-related needs, feelings and behaviours (Steele et al. 2003, 2007). We did not nd a signicant correspondence of secure/ insecure models of attachment between adoptive mothers and their late-placed children, which was highlighted in other studies (Dozier & Sepulveda 2004; Verissimo & Salvaterra 2006), but we did nd a signicant increase of this matching. It seems to emphasize that the link between patterns of attachment among adoptive mothers and their late-placed pre-school and school age children, increases progressively although it cannot be clearly outlined so soon after placement. The main limitations of the current study are: (1) the relatively small size of the sample, which has nonetheless the advantage of being observed from the beginning of the placement. A bigger sample size could conrm or disconrm the results obtained by this study, allowing for further in-depth examination; (2) although it was possible to observe a shift towards security the period of time between the two assessment sessions was not enough long to establish whether the modications are really stable. It would therefore be interesting to carry out a follow-up longitudinal study in the medium to long term. We suggest that a secure attachment model in the AAI of adoptive mothers, which different researchers link to a high level of parenting quality, could represent a protective factor enabling the establishment of positive interactions with lateadopted children and permitting the mothers to understand the mental state which could be hidden beneath hostile, childish, rejecting childrens behaviour; these factors could contribute to encouraging the attachment patterns revision of previously maltreated children. It seems to suggest the importance of taking into account the adoptive parents IWMs as a parenting style indicator from the pre-adoption selection phase of couples in order to facilitate the best possible adjustment of late-placed children.

Acknowledgements
We thank Alessandra Santona for the methodological supervision and all the social workers and adoptive families that agreed to take part in this study. Support from the Faculty of Psychology of Sapienza University of Rome, for allowing use of the video-recording laboratories is gratefully acknowledged. We also thank Ester DOnofrio and Serena Messina for contributing to coding part of the data collected in this study.

2010 Blackwell Publishing Ltd, Child: care, health and development, 37, 1, 8288

Attachment in late-adopted children

87

Key messages
The adoption of late-placed children represents a natural experiment. There was a prevalence of insecure attachments among late-placed children at the beginning of their adoptive placement in the families. In T2 it was possible to observe a signicant improvement of the attachment behaviour of adopted children, moving from insecurity towards security. Children who improved their attachment behaviours were adopted by mothers with secure attachment models, whereas children adopted by insecure mothers remained in their initial classications, mostly of an insecure quality. The links between models of attachment among adoptive mothers and their late-placed children were not signicant, but they were increased from T1 to T2. There are grounds to suggest that the assessment of the future adoptive parents IWMs in the pre-adoption phase of selection of couples must be taken into account.

References
Bakermans-Kranenburg, M. J. & Van IJzendoorn, M. H. (2009) The rst 10 000 adult attachment interviews: distribution of adult attachment representations in clinical and non-clinical groups. Attachment & Human Development, 11, 223263. Bowlby, J. (1969) Attachment and Loss. Vol. 1. Attachment. Hogarth Press, London, UK. Bowlby, J. (1973) Attachment and Loss. Vol. 2. Separation: Anxiety and Anger. Hogarth Press, London, UK. Bowlby, J. (1980) Attachment and Loss. Vol. 3. Loss: Sadness and Depression. Basic Books, New York, NY, USA. Bretherton, I. & Munholland, K. A. (2008) Internal working models in attachment relationships: elaborating a central construct in attachment theory. In: Handbook of Attachment. Theory, Research and Clinical Applications, 2nd edn. (eds J. Cassidy & P. R. Shaver), pp. 102129. Guilford Press, New York, NY, USA. Cassidy, J. (1988) Child-mother attachment and the self in six-year-olds. Child Development, 59, 121134. Cassidy, J. & Shaver, P. R. (eds) (2008) Handbook of Attachment. Theory, Research and Clinical Applications, 2nd edn. Guilford Press, New York, NY, USA. Crowell, J. A., Treboux, D. & Waters, E. (2002) Stability of attachment representations: the transition to marriage. Developmental Psychology, 38, 467479. Dozier, M. & Sepulveda, S. (2004) Foster mother state of mind and treatment use: different challenges for different people. Infant Mental Health Journal, 25, 368378.

Dozier, M., Chase-Stovall, K., Albus, K. & Bates, B. (2001) Attachment for infants in foster care: the role of caregiver state of mind. Child Development, 72, 14671477. Dunn, L. M. & Dunn, L. M. (2000) Peabody Picture Vocabulary Test-Revised. George, C., Kaplan, N. & Main, M. (1984, 1985, 1996) Adult Attachment Interview, 3rd edn. Unpublished manuscript. University of California at Berkeley, Berkeley, CA. Goldwyn, R., Stanley, C., Smith, V. & Green, J. (2000) The manchester child attachment story task: relationship with parental AAI, SAT, and child behaviour. Attachment & Human Development, 2, 7184. Green, J., Stanley, C., Smith, V. & Goldwyn, R. (2000) A new method of evaluating attachment representations in young school-age children: the manchester child attachment story task. Attachment & Human Development, 2, 4870. Hamilton, C. E. (2000) Continuity and discontinuity of attachment from infancy through adolescence. Child Development, 71, 690694. Howe, D. (1997) Parent-reported problems in 211 adopted children: some risks and protective factors. Journal of Child Psychology and Psychiatry, 38, 401411. Main, M. & Cassidy, J. (1988) Categories of response to reunion with the parent at age six: predicted from attachment classications and stable over a one-months period. Developmental Psychology, 24, 415426. Main, M., Goldwyn, R. & Hesse, E. (2002) Adult Attachment Scoring and Classication System. Manual in Draft: version 7.1, September 2002, U. New Mexico, Albuquerque. Unpublished Manuscript Department of Psychology, University of California at Berkeley. Mikulincer, M. & Shaver, P. R. (2008) Adult attachment and affect regulation. In: Handbook of Attachment. Theory, Research and Clinical Applications, 2nd edn. (eds J. Cassidy & P. R. Shaver), pp. 503531. Guilford Press, New York, NY, USA. Rutter, M., OConnor, T. G. & the English and Romanian Adoptees Study Team (2004) Are there biological programming effects for psychological development? Findings from a study of Romanian adoptees. Developmental Psychology, 40, 8194. Rutter, M., Colvert, E., Kreppner, J., Beckett, C., Castle, J., Groothues, C., Hawkins, A., OConnor, T. G., Stevens, S. E. & Sonuga-Barke1, E. J. S. (2007) Early adolescent outcomes for institutionally deprived and non-deprived adoptees. I: disinhibited attachment. Journal of Child Psychology and Psychiatry, 48, 1730. Schoeld, G. & Beek, M. (2005) Providing a secure base: parenting children in long-term foster family care. Attachment & Human Development, 7, 325. Siegel, S. & Castellan, N. J. Jr (1988) Non Parametric Statistics for The Behavioral Science. McGraw-Hill Book Company, New York, NY, USA. Steele, M., Hodges, J., Kaniuk, J., Hillman, S. & Henderson, K. (2003) Attachment representations and adoption: associations between maternal states of mind and emotion narratives in previously maltreated children. Journal of Child Psychotherapy, 29, 187205.

2010 Blackwell Publishing Ltd, Child: care, health and development, 37, 1, 8288

88

C.S. Pace and G.C. Zavattini

Steele, M., Henderson, K., Hodges, J., Kaniuk, J., Hillman, S. & Steele, H. (2007) In the best interests of the late-placed child: a report from attachment representations and adoption outcome study. In: Developmental Science and Psychoanalysis. Integration and Innovation (eds L. Mayes, P. Fonagy & M. Target), pp. 159182. Karnac, London, UK. Steele, M., Kaniuk, J., Hodges, J., Asquith, K., Hillman, S. & Steele, H. (2008) Measuring mentalization across contexts. Links between representations of childhood and representations of parenting in an adoption sample. In: Mind to Mind (eds E. L. Jurist, A. Slade & S. Bergner), pp. 115136. Other Press, New York, NY, USA. Van IJzendoorn, M. H. & Juffer, F. (2005) Adoption is a successful natural intervention enhancing adopted childrens IQ and school performance. Current Directions in Psychological Science, 14, 326330. Van IJzendoorn, M. H. & Juffer, F. (2006) The emanuel miller memorial lecture 2006: adoption as intervention. Meta-analytic evidence for massive catch-up and plasticity in physical,

socio-emotional, and cognitive development. Journal of Child Psychology and Psychiatry, 47, 12281245. Van IJzendoorn, M. H., Juffer, F. & Klein Poelhuis, C. W. (2005) Adoption and cognitive development: a meta-analytic comparison of adopted and non-adopted childrens IQ and school performance. Psychological Bulletin, 131, 301316. Verissimo, M. & Salvaterra, F. (2006) Maternal secure base scripts and childrens attachment security in an adopted sample. Attachment & Human Development. Special Issue. Script-like Attachment Representation and Behaviour in Families and Across Cultures, 8, 261273. Vorria, P., Papaligoura, Z., Dunn, J., van IJzendoorn, M. H., Steele, H., Kontopoulou, A. & Saradou, Y. (2003) Early experiences and attachment relationships of Greek infants raised in residential group care. Journal of Child Psychology and Psychiatry and Allied Disciplines, 44, 12081220. Waters, E., Merrick, S. K., Treboux, D., Crowell, J. & Albersheim, L. (2000) Attachment security from infancy to early adulthood: a 20 year longitudinal study. Child Development, 71, 684689.

2010 Blackwell Publishing Ltd, Child: care, health and development, 37, 1, 8288