Child and Teen Development

Direct and indirect relations between parent–child attachments, peer acceptance, and self-esteem for preschool children Alexandra Pintoa, Manuela Veríssimoa, Ana Gatinhoa, António J. Santosa

and Brian E. Vaughnb

aWilliam James Center for Research, ISPA-Instituto Universitário, Lisbon, Portugal; bHuman Development & Family Studies, Auburn University, Auburn, AL, USA

ABSTRACT The present study aims to test Bowlby’s suggestions concerning relations between the child’s attachment quality with parents and subsequently constructed models of self-worth during early child- hood. In most research on this question, attachment with mothers is considered in relation to self-worth but the child’s attachment with fathers is not. Neither has the peer group been studied as an influence on child self-esteem, in the context of attachment research. This study addresses these relatively unstudied influ- ences on child self-esteem. Attachment security to mother and father was measured by the Attachment Behavior Q-Set at two and half years of age. At five years of age social acceptance was measured using two sociometric techniques, and the self-esteem with the California Child Q-Sort. Our analyses indicated that secur- ity of the attachment to father and peer acceptance are both unique, significant predictors of the childrens’ self-esteem. The security of the attachment to mother was also related to child self-esteem but did not emerge as a uniquely significant predictor. Peer acceptance appeared to moderate of the effect of the secur- ity of the attachment to father on the self-esteem of children. Our results extend the relatively sparse literature relating early attach- ments to self-esteem during early childhood.

ARTICLE HISTORY Received 6 June 2015 Revised 6 September 2015 Accepted 8 September 2015

KEYWORDS Self esteem; attachment to mother; attachment to father; peer acceptance; preschool

Attachment theory implies that the child’s mental representation (or internal working model) of attachment with parents informs and is intertwined with mental representa- tions of self (Bowlby, 1973; Cassidy, 1988, 1990). Bowlby (1982) believed that this could occur because the child’s mental representations of attachment relationships are co- constructed (with the parent) from the repeated experiences of parental sensitivity and responsiveness to the child’s communicative signals and parental acceptance of and support for the child’s exploratory activities over the early years of life. When the aggregate of child experiences with a parent was characterized as sensitive, responsive, and supportive, the child would experience being valued, loved, and important, and the initial self-model should reflect these qualities. On the other hand, children whose

CONTACT Manuela Veríssimo [email protected]

ATTACHMENT & HUMAN DEVELOPMENT, 2015 VOL. 17, NO. 6, 586–598 http://dx.doi.org/10.1080/14616734.2015.1093009

© 2015 Taylor & Francis

experiences with a parent were, on average, insensitive, unresponsive, and not suppor- tive would likely experience devaluation, lack of love, and insignificance, and their initial self-models would be expected to reflect these qualities. Bowlby felt that these early self-models would, if not modified by subsequent experiences, also inform subsequent expectations about the quality and utility of relationships with important others over significant periods of developmental time.

Bowlby’s speculations about the origins of self-models (or self-concepts) and self- evaluation (or self-esteem) are deeply connected to social theories of self (e.g., James, 1890; Mead, 1913) and have prompted a substantial body of research relating aspects of attachment to self-esteem. However, the bulk of this research and scholarship is not developmentally informed and concerns concurrent associations between measures of attachment security and measures of self-esteem in samples of school-age children, adolescents, or adults. Only a handful of studies have attempted to document associations between attachment in infancy or early childhood and self-esteem (or other self-qualities) in children less than seven years of age (e.g., Cassidy, 1988; Clark & Symons, 2000; Goodvin, Meyer, Thompson, & Hayes, 2008; Sroufe, 1983; Verschueren, Marcoen, & Schoefs, 1996), and only the work of Verschueren and associates (e.g., Verschueren, Buyck, & Marcoen, 2001; Verschueren, Doumen, & Buyse, 2012) can be considered pro- grammatic. For the majority of these studies, attachment and self-data were collected concurrently, so the temporal ordering of attachment and self-esteem could not be tested (see Sroufe, 1983, for an exception to this generalization, and Clark & Symons, 2000, for a failure to predict self-measures from earlier attachment assessments). Moreover, only a few of these studies included measures of father–child attachment in their assessment protocol, so the potential influence of multiple attachments on self-esteem is not known. Kerns, Mathews, Koehn, Williams, and Siener-Ciesla (2015) compared attachments to both the mother and father and they reported that children were more likely to use the mother a safe haven when they were distressed or threatened and more likely to use the father as a secure base for exploration. These kinds of results illustrate the importance of assessing attachments to more than one attachment figure during childhood. Thus, a primary purpose of the present study was to examine relations between early (age 2.5) assess- ments of attachment security for both mothers and fathers (from Attachment Q-sort, AQS, data) and later (age five) assessments of child self-esteem.

In part, the modest pace of research testing relations between attachment quality and self-esteem in early childhood is due to the difficulties associated with obtaining valid assessments of self-representations and self-esteem for children under six years of age. Different research teams have adopted several different approaches to assessing child self- esteem, including teacher reports of child self-esteem (e.g., Sroufe, 1983; Verschueren & Marcoen, 1999), self-reports of perceived competence and self-evaluation using the pictorial scales designed by Harter and Pike (1984; see also Cassidy, 1988; Clark & Symons, 2000), “puppet interviews” thought to reflect self-feeling, doll-play story comple- tion protocols designed by Cassidy (1988), and child self-reports (e.g., Eder, 1990; Goodvin et al., 2008). Unfortunately, these measures do not always converge on a common dimension and the Harter and Pike (1984) Pictorial Perceived Competence scales seem especially weak in this regard (e.g., Cassidy, 1988; Clark & Symons, 2000). For this study, we used a behavioral assessment of self-esteem derived from intensive observations of children in their preschool classrooms that were summarized using the California Child

ATTACHMENT & HUMAN DEVELOPMENT 587

Q-Sort (Block & Block, 1980). The Q-sort descriptions provided by observers were scored for self-esteem using the criterion profile suggested by Waters, Noyes, Vaughn, and Ricks (1985).

Although Bowlby’s theory of attachment identifies links between self-models and their attributes (e.g., self-esteem) and the organization of attachment behavior in the early years, attachment theory does not mandate that parent–child attachments be the sole source of self-models. Indeed, an abundance of evidence has shown that self- evaluations reflect the opinions of all salient social groups and self-esteem may vary within the individual as a function of shifting salience of different groups over shorter or longer periods of time (e.g., James, 1890; Kernis, 2005). By the time most children in modern, developed societies reach five years of age, it is normative that they spend weekdays in non-parental childcare settings (Eurydice Network, 2009), and many studies have indicated that self-esteem measured in those settings is associated with the quality of child social functioning, as indicated by measures of peer acceptance, numbers of friendships, and relationships with teachers (e.g., Sroufe, 1983; Verschueren et al., 2012).

In particular, young children who are more accepted (as playmates or as friends) tend to have higher scores on measures of self-esteem (e.g., Coplan, Findlay, & Nelson, 2004; Emidio, Santos, Maia, Monteiro, & Veríssimo, 2008) and similar results are reported for older children (e.g., Boivin & Hymel, 1997; Egan & Perry, 1998; Verschueren et al., 2001). It seems important, therefore, to include measures of social functioning in a salient peer group when studying the influences on self-esteem during early childhood, even if such measures must be obtained concurrently with the assessments of self-esteem, and to test whether relations between attachment and self-esteem may be mediated or mod- erated by child social functioning in the peer group. Accordingly, we assessed both children’s peer acceptance in their group and self-esteem when the children were five years of age for this study.

This study was designed to test a critical assumption of Bowlby’s attachment theory (1973, 1982) concerning the connections between representations of parent–child attachment relationships (in this study, relationships with both parents), formed by the time the child is 2–3 years of age (scored from observed behaviors referencing the child’s use of the parent as a secure base for exploration and haven of safety when stressed) summarized using the Attachment Q-Sort (AQS; Waters, 1995), and child self- esteem at age five years. Child self-esteem is measured using an observational protocol with a defined criterion for the self-esteem construct, rather than a direct test or inter- view with the children. We test the hypothesis that attachment security with each parent is positively associated with the index of self-esteem, and further test whether the associations between self-esteem and each parent remain significant when both are simultaneously entered into a regression analysis. Previous research findings suggest the hypothesis that peer acceptance and self-esteem will be positively associated, and we test whether associations between parent–child attachment security and self-esteem remain significant when peer acceptance is included as a predictor of self-esteem in regression analyses. Finally, we explore whether associations between parent–child attachment security, peer acceptance, and self-esteem meet the criteria for testing mediation or moderation effects. When the possibility of either type of statistical inter- action is present, relevant analyses are designed test these effects.

588 A. PINTO ET AL.

Method

Participants

The study sample consisted of 45 children (23 girls and 22 boys) who were participating in a larger study of attachment development and associations between attachment and social competence in the peer group. Initial observations for the purposes of assessing attachment relationships with mothers and fathers were made when the children were 2–3 years old (M = 32.2, DP = 0.75). Classroom observations and peer acceptance assessments were completed in the children’s preschool classrooms when they were five years old. The sample was homogenous for socioeconomic status, belonging to a medium-high socio-economic stratum, by the standards of the local community. All children attended a private preschool education center in the district of Lisbon, Portugal. The study protocols were reviewed and approved by the relevant review boards for the use of human subjects in research. Parents received information about the research project and its purpose, and completed an informed consent (63% of the parents participated in the home assessments, while 85% consented to the preschool observa- tions and interviews). The level of maternal education ranged from nine to 23 years of schooling (M = 15.00, SD = 3.04) and the paternal education levels were between 4 and 19 years (M = 15.1 SD = 3.33). The average age of mothers was 35.8 years (SD = 4.67) and for fathers was 38.3 years (SD = 6.6).

Procedures

After the children reached their second birthday, the parents were contacted to schedule home visits for the purpose of observing child behavior (with each parent separately). Parents were asked to schedule a time when they would be alone in the house with the child, or when other adults and children could conveniently be in a different area of the home, so as not to interfere with the parent and child being observed. Two observers came to the home on the same visit and both completed observations of child behavior over a 2– 3 hour observation period. Parents were briefed that the purpose of the visit was to observe the child with the mother/father in their daily routines and, as such, each parent was asked to maintain these routines with as little change as possible. Observers didn’t interfere with the family routine, but participated in the children’s games when requested and talked informally with the mother/father, taking care not to disrupt the interactions between mother/father and the child. When necessary, and following the conversation with the mother/father, questions were asked about items that could not be observed during the visit. Observations were summarized using the Attachment Q-Sort (Waters, 1995; see below) and different teams of observers completed home visits for the mother and father of each child participant. A researcher who was expert in the use of the AQS with young children trained each home observer before the home observations for this study were initiated. During the training period they discussed the intended meanings of the items with the expert and used the AQS to describe an “ideally secure” two-year-old.

After the child reached age five years, and was enrolled in the preschool program at the child care/education facility, teams of observers and interviewers (different than those completing the AQS observations) completed observational assessments of

ATTACHMENT & HUMAN DEVELOPMENT 589

behavior and personality attributes that were summarized using the California Child Q-Sort (CCQ; Block & Block, 1980). Q-sort observers were trained by a researcher who was expert in the use of the CCQ with preschool age children. Before initiating observa- tions, these observers discussed the intended meanings of the individual CCQ items and sorted the 100 items to describe the “ideal preschool age child”. Peer acceptance was assessed using two picture sociometric interview protocols (nominations and paired comparisons) that were completed during the period when children were being observed by the team doing the CCQ assessments.

Measures

AQS The AQS (version 3.0; Waters, 1995) assesses the organization of the child’s secure base behavior in relation to an attachment figure in an ecologically valid context (i.e., at home) and allows for a detailed description of children’s attachment behavior when observed with a specific attachment figure (van IJzendoorn, Vereijken, Bakermans- Kranenburg, & Riksen-Walraven, 2004; Waters & Deane, 1985). The Q-set consists of 90 individual items that are sorted into a rectangular (i.e., each of nine categories contains 10 items) distribution. The distribution range indicates whether a given item is “Uncharacteristic” (categories 1–3), neither “Characteristic nor Uncharacteristic” (cate- gories 4–6), or “Characteristic” (categories 7–9) of the child being observed. After the sort is completed for an observer, a score for attachment security is derived by correlat- ing (Pearson) the vector of scores in the observer’s description of the child with the “security criterion Q-sort” published by Waters (1995). These correlations become the child’s score for attachment security and indicate the relative similarity between the Q-sort description of the child and the Q-sort description of the “hypothetically most secure preschool child” provided by experts in the field of attachment development. Thus a relatively high value indicates greater similarity between the two Q-sort compo- sites. In practice, 90% of scores tend to be between −0.5 and 0.6 and the means for children classified as “secure” using the Strange Situation Procedure tend to range between 0.35 and 0.5 across diverse samples (van IJzendoorn et al., 2004). In this study, intra-class correlation (ICCs) agreements for the AQS criterion scores across all rater pairs were r = 0.71 for mothers and r = 0.72 for fathers. Consequently, the Q-sort descriptions were averaged (for each of the 90 items) and final security criterion scores were calculated from this aggregated sort from both observers.

Self-esteem (CCQ) As noted above, the CCQ (Block & Block, 1980) was used to assess the children’s self- esteem. Two observers independently observed all participating children in a given classroom over a period of 20 hours (i.e., 20 hours for each observer). Observations took place over 5–10 class-days and observers took care to observe each participating child across the full range of contexts available in the classroom (e.g., meals, free choice activities, structured activities, playground, etc.). After completing all observations, each observer used the CCQ to describe the behavior and personality of all participating children in the classroom. Again, a rectangular distribution of items was used (i.e., 11 items in each of nine categories, with the last item being placed in the middle of the

590 A. PINTO ET AL.

distribution) and criterion scores for self-esteem (Waters et al., 1985) were calculated for each child following the same rationale as for the AQS criterion scores. That is, Waters et al. reported on a self-esteem criterion sort provided by researchers who were expert in the self-esteem dimension and in the behavior and development of young children. Every participating child received scores for Q-sorts provided by each observer Median ICC estimates for each class pair of observers for the self-esteem criterion scores was 0.89, therefore, sorts for the two observers were averaged for each item, and final criterion scores were calculated from this aggregate (M = 0.60, SD = 0.15).

Peer acceptance Each participating child was interviewed individually to obtain peer acceptance data using the nominations protocol described by McCandless and Marshall (1957) and again using the paired comparisons procedure described by Vaughn and Waters (1981). For the nominations procedure, photos of each participating child in a classroom were presented to the child making nominations (randomly rearranged for each child). First, the child was asked to name all peers in the photos, to determine that the child knew every participat- ing classmate (no child failed to correctly identify all peers in the array of photos). Then, the child was asked to identify a classmate with whom he/she especially liked to play. The chosen child’s photo was removed from the array and the question was repeated again, twice (for a total of three positive choices). After identifying three preferred playmates, the child returned to the array and was asked to name three children with whom she/he did not prefer as a playmate. These scores were not used in this study. Scores were the number of times a child was chosen as a preferred playmate by peers (adjusted for class size by calculating the average value for each child across all classmates making choices).

For the paired comparisons task, paired photo stimuli were prepared in which each participating child was paired with every other participating child in the classroom (e.g., a total of N – (N-1)/2 pairs, or 190 pairs in a class of 20 children). A child’s image was presented an equal number of times on the left and right sides of the display, to avoid any potential position bias by the responding children. Nearly all children completed this task in two sessions of about 15 min each. A child’s image was presented an equal number of times on the left and right sides of the display, to avoid any potential bias by the responding children. For each pair of images, the child was asked which of the two children she/he preferred as a playmate. Scores were the number of times a child was chosen by peers as a preferred playmate, adjusted for class size by taking the average value of positive choices by dividing the total by the number of children making choices.

Following standard conventions for sociometric data, the nominations and paired comparisons scores were standardized within each classroom. A final score for peer acceptance was computed by taking the average of these two standard (i.e., Z-) scores.

Results

Preliminary analysis

Means and standard deviations for the variables under study are shown in Table 1. The values for AQS security scores with parents are at the high end of expected mean values for the AQS and suggest that, on the whole, the behavior of these children was more

ATTACHMENT & HUMAN DEVELOPMENT 591

similar than not to the behavior of the hypothetically “very secure preschool child” described by experts (Waters et al., 1985). For this sample, AQS scores were also significantly correlated across mothers and fathers, ICC = 0.42, F(1, 43) = 2.46, p < .01. Consistent with the notion that the secure children are somewhat over-represented in the sample, the CCQ self-esteem score is also relatively high, again suggesting that the empirical Q-sort profiles (i.e., sorts of child participants) shared about 36% of their variance with the profile of the hypothetical “high self-esteeming preschool child”.

Because peer acceptance scores were calculated over all participants and were standar- dized (Z-scores) within the classroom group, class means were set at zero with a standard deviation of approximately 1.0. Thus, themean of 0.25 for this sample suggests that well-liked children are over-represented here, however, this variable has a relatively large standard deviation and the range of scores is considerably greater than for the other study variables.

Associations among the study variables

Pearson correlations among the study variables are presented in Table 2. As noted above, the age two AQS security scores were significantly correlated across parents and both were also significantly correlated with self-esteem assessed at age five. Peer acceptance at age five was also a significant correlate of self-esteem, however, neither of the AQS security scores were significantly associated with peer acceptance.

Effects of attachment security with mother and father and peer acceptance

A series of multiple regression analyses, were computed to test the primary study hypoth- eses. Self-esteemwas the dependent variable and the other study variables were predictors. The correlation values reported in Table 2 indicate that the criteria for mediation suggested by Baron and Kenny (1986) are not met in these data (i.e., attachment security variables are not significantly associated with peer acceptance), however moderation of the relations between AQS security and self-esteem by peer acceptance could be tested.

A hierarchical regression model with security scores entered at the first step and peer acceptance at the second revealed that attachment security with both parents was a significant predictor of self-esteem, R2 = 0.256, F(2, 42) = 7.23, p < .01. However, only the security score with the father had a unique, significant beta weight. Adding peer

Table 1. Mean, standard deviation, and range values for study variables. M SD Min Max

Attachment security to mother 0.48 0.21 −0.11 0.79 Attachment security to father 0.49 0.20 −0.03 0.78 Peer acceptance 0.25 1.16 −3.2 2.52 Self-esteem 0.60 0.14 −0.77 0.60

Table 2. Pearson correlations between the different variables. Attachment security to father Peer acceptance Self-esteem

Attachment security to mother 0.42** 0.15 0.39** Attachment security to father 0.02 0.46** Peer acceptance 0.57**

**p < .001

592 A. PINTO ET AL.

acceptance at the second step significantly increased the predicted variance in self- esteem, R2 = 0.537, F(3, 41) = 15.88, p < .001. Both the AQS security score with father and the peer acceptance scores were unique, significant predictors of CCQ self-esteem.

Because the AQS score with mother was not a unique, significant predictor and because the overall regression did not change substantially when this AQS score was dropped from the predictor set (< 2% of predicted variance for self-esteem lost), we did not include the AQS security score with mother in subsequent analyses. For the next analysis, we included the AQS security score with father, the peer acceptance score, and the interaction of these two, to test the possibility that moderation effects were present. Results of this regression analysis are presented in Table 3. Both main effects and their interaction were significantly predictors of CCQ self-esteem. To test for moderation effects, the relation between AQS security with fathers and CCQ self-esteem was examined at three levels of peer acceptance (M–SD = −.91, M = 0.25, M+SD = 1.41).

The plot of slopes at the three levels of peer acceptance is shown in Figure 1. The relation between AQS security and CCQ self esteem was greatest when Peer Acceptance

Table 3. Testing the interaction of AQS security with father and peer acceptance in the prediction of CCQ self-esteem. Variable B b Significance b R2

AQS security, father 0.93 0.34 Z = 4.59*** Peer acceptance 0.46 0.12 Z = 5.23*** Security x Peer accept −0.46 −0.14 Z = −2.57*4 0.55

*p < .05 ***p < .001

Figure 1. Plot of slopes of relations between self-esteem and attachment security with father for three levels of peer acceptance.

ATTACHMENT & HUMAN DEVELOPMENT 593

is low, b = 0.43, Z = 4.16, p < .001, and remains significant when Peer Acceptance is at the intermediate level, b = 0.28, Z = 3.42, p < .001. However, when Peer Acceptance was high (M+SD), the relation between AQS security with father and CCQ self-esteem is not significant (b = .13, Z = 1.19, p = .24). The slopes plotted in Figure 1 indicate that when AQS security with father is lower (i.e., less similar to the highly secure child), the child’s level of peer acceptance has greater influence on CCQ self-esteem than when AQS security with father is relatively high.

Discussion

Empirical tests of Bowlby’s suggestions concerning relations between the child’s attachment quality with parents and subsequently constructed models of self-worth and self in relation to important others during early childhood have generally sup- ported Bowlby’s speculations (e.g., Cassidy, 1988; Sroufe, 1983; Verschueren et al., 1996), although there have also been studies that failed to find significant cross- construct relations (e.g., Clark & Symons, 2000). We noted that most studies in which positive, significant associations were found tested attachment and self-esteem concurrently (but see Sroufe, 1983, as an exception) and nearly all studies only assessed attachment quality with mothers. In this study, attachment security was assessed using the AQS when the child was approximately 2.5 years of age, with both mothers and fathers, and the CCQ self-esteem measure was collected when the child was 5–6 years of age. Self-esteem was positively and significantly associated with attachment security to each parent.

Our finding that attachment security predicts self-esteem in the peer group over two years later supports Bowlby’s original speculative hypotheses and is consistent with prior research. Nevertheless, we found that using both security scores together as predictors of CCQ self-esteem indicated that the child’s attachment relationship with the father was uniquely predictive and that security with the mother was not a unique predictor of self-esteem in this sample. Moreover, when the classroom Peer Acceptance measure was added to the prediction equation, the contribution of the AQS security with mother was even further reduced. Indeed, when AQS security with mother was removed from the prediction equation, the shrinkage in R2 was less than 2%. Together, these results suggest that the correlation between child–mother attachment and CCQ self-esteem three years later is largely accounted for by the association of child–mother and child–father attachment security.

For this sample at least, the CCQ self-esteem score is more reflective of attachment with fathers than with mothers. Our results are consistent with the findings of Verschueren and Marcoen (1999) and of Suess, Grossmann, and Sroufe (1992). Verschueren and Marcoen (1999) stressed the differentiation of roles of maternal and paternal parenting and concluded that it is expectable that attachment security to the father would be more strongly associated with self-esteem, as evaluated in this study, because our measure is more related to aspects of the outside world and of socialization. In turn, Suess et al. (1992) showed that child-father attachment was significantly related to peer relationship skills at five years of age (less negative affect during play) while child–mother attachment was significantly related to longer periods of concentrated play at five years of age. Also, Kerns and colleagues (2015) reported that children used

594 A. PINTO ET AL.

mothers more as a safe haven when distressed and fathers as a secure base for exploration. Finally, Steele and Steele (2005) considered different roles of the attach- ment to mother and father in the child’s social and emotional development. They proposed that the attachment to mother would be more related to the understanding of the inner emotional world both of the others as of the one itself, while the attach- ment to father would be more related to positive skills in the interaction with siblings, peers and others. Because our self-esteem measure is intended to summarize the child’s confidence and effectiveness in the context of peer interactions, it seems more consis- tent with the “exploration” and “interaction” than with the “safe haven” or “inner world” themes of attachment organization. Our findings reinforce the conclusions reached by Verschueren and Marcoen (1999), who argued that security (to the mother and the father) should influence different aspects or valences of the children’s self. In future research, it will be important to use additional and more nuanced measures of self- esteem to test whether relationships with each parent have different influences on the child’s developing model of self and self-worth.

In turn, the moderating influence of peer acceptance reported in this study suggests that self-esteem is not solely a consequence of attachment security. That is, when the Peer Acceptance variable was included as a predictor in the regression analysis, this variable more than doubled the predicted variance of CCQ self-esteem. We also found that the interaction of AQS security with father and Peer Acceptance added a significant, albeit modest and signed negatively, increment to the prediction of CCQ self-esteem. Plots of the simple slopes for low, middle, and high regions of the Peer Acceptance distribution showed that the relation between AQS security and CCQ self-esteem was only significant when the child had lower and intermediate levels of Peer Acceptance. This suggests the possibility that positive integration into the peer group can, at least in part, compensate for a relatively lower quality attachment relationship with the father. It will be important to attempt replications of this result in different samples from Portugal and from other societies, especially because we have reasons to believe that this sample may not be representative of Portuguese samples more generally.

We noted above that this sample was participating in a larger study of attachment and social competence. All children in the larger sample (N = 157) were living with two parents when they were recruited to the study and all mothers agreed to home visits, but only 128 of the fathers agreed to participate in any aspect of the larger study, and only 45 cases had both fathers’ home visits and a child who remained in the sample after reaching age five years. It seems likely that families in which both parents agree to be observed at home with their child feel more confident than other parents and are less concerned about their child’s behavior being observed at home than other parents. Moreover, it seems likely that parents willing to invite observers into their homes on two occasions would share parenting attitudes, values, and goals that would underlie greater family harmony and reduced family conflict, which, in turn, would support the co-construction of secure attachment relationships with their child. In fact, the AQS security scores in this sample are somewhat higher than might be expected from other studies of children in their third year of life (e.g., Vaughn et al., 2007). It is possible that family self-selection could have yielded a sample in which children were generally more socially capable and more likely to be accepted by peers and this may account for the elevated Peer Acceptance scores and relatively

ATTACHMENT & HUMAN DEVELOPMENT 595

high CCQ self-esteem scores that we observed. Of course, higher scores on study variables would not per se lead to the prediction that AQS security should predict CCQ self-esteem, and would not anticipate the moderating effect of Peer Acceptance. Nevertheless, the nature of this relatively small sample does constitute a limitation on interpretations of our findings. Clearly, relations among these variables need to be examined in samples with greater SES, ethnic, and societal diversity.

We also relied on a somewhat novel assessment of self-esteem. Although Waters and associates (Waters et al., 1985) described this measure and showed that self- esteem scored from the CCQ was distinguished from related constructs in terms of specific item placements in the criterion profiles, this measure has not been widely used in self-esteem research. In part, the neglect of this measure may be due to the time it takes to observe an entire classroom of preschool children and it may also be due to the fact that the initial results are profiles of behaviors, personality attributes, and preferences, rather than a single self-esteem “score”. Such scores are derived by calculating the similarity between a given child’s profile and a “criterion” profile, provided by developmental scientists who are expert with regard to the construct (i.e., self-esteem) and are also expert in the social/emotional development of young children. Q-methods have a long history in personality and developmental psychology (e.g., Block, 2002; Block & Block, 1980; Vaughn, Santos, & Coppola, 2014; Waters, 1995; Waters et al., 1985) and these methods could prove useful in addressing a wide range of research questions. It will be useful in future research to examine the convergence between this behaviorally based self-esteem measure and self-feeling/self-worth/self- esteem scores from other measures (e.g., Cassidy’s puppet interview) that have been used in other studies of young children.

To conclude, this study tested Bowlby’s speculative hypothesis that attachment security with parents in the early years of life would become intertwined with later constructed models of self and self-worth. Bowlby’s hypothesis received support in our data, however, we found that attachment security with father was more implicative with respect to later self-esteem than was attachment security with mother. Moreover, peer acceptance assessed concurrently with self-esteem was also a unique predictor and interacted significantly with father–child attachment to predict self-esteem. Decomposition of the interaction suggested the possibility that higher peer acceptance can compensate for a lower quality attachment to the father. Despite sample limitations, these results extend the relatively sparse literature relating early attachments to self- esteem during early childhood.

Acknowledgments

Data collection and preparation of this article has been supported in part by grants from the Portuguese Foundation for Science and Technology (FCT- PTDC/MHC-PED/3929/2012, SFRH/BD/ 68480/2010 and UID/PSI/04810/2013). We acknowledge the support of all the colleagues that contributed to data collection.

Disclosure statement

No potential conflict of interest was reported by the authors.

596 A. PINTO ET AL.

References

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psycho- logical research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. doi:10.1037/0022-3514.51.6.1173

Block, J. (2002). Personality as an affect processing system: Toward an integrative theory. Malwah, NJ: Erlbaum.

Block, J., & Block, J. (1980). The California child Q-Set. Palo Alto, CA: Consulting Psychologists. Boivin, M., & Hymel, S. (1997). Peer experiences and social self-perceptions: A sequential model.

Developmental Psychology, 33(1), 135–145. doi:10.1037/0012-1649.33.1.135 Bowlby, J. (1973). Attachment and loss: Separation, anxiety, and anger (Vol. 2). Middlesex: Penguin

Books. Bowlby, J. (1982). Attachment and loss: Vol. 1. Attachment. New York, NY: Basic Books. Cassidy, J. (1988). Child-mother attachment and the self in six-year-olds. Child Development, 59,

121–134. doi:10.2307/1130394 Cassidy, J. (1990). Theoretical and methodological considerations in the study of attachment and

the self in young children. In M. T. Greenberg, D. Cicchetti, & E. M. Cummings (Eds.), Attachment in the preschool years: Theory, research and intervention (pp. 87–119). Chicago, IL: University of Chicago Press.

Clark, S., & Symons, D. (2000). A longitudinal study of Q-sort attachment security and self- processes at age 5. Infant and Child Development, 9, 91–104. doi:10.1002/(ISSN)1522-7219

Coplan, R. J., Findlay, L. C., & Nelson, L. J. (2004). Characteristics of preschoolers with lower perceived competence. Journal of Abnormal Child Psychology, 32, 399–408. doi:10.1023/B: JACP.0000030293.81429.49

Eder, R. (1990). Uncovering young children’s psychological selves: Individual and developmental differences. Child Development, 61, 849–863.

Egan, S., & Perry, D. (1998). Does low self-regard invite victimization? Developmental Psychology, 34 (2), 299–309. doi:10.1037/0012-1649.34.2.299

Emidio, R., Santos, A., Maia, J., Monteiro, L., & Veríssimo, M. (2008). Auto-conceito e aceitação pelos pares no final do período pré-escolar. Análise Psicológica, 26(3), 491–499.

Eurydice Network. (2009). Tackling social and cultural inequalities through early childhood education and care in Europe. Brussels: Education, Audiovisual and Culture Executive Agency.

Goodvin, R., Meyer, S., Thompson, R., & Hayes, R. (2008). Self-understanding in early childhood: Associations with child attachment security and maternal negative affect. Attachment & Human Development, 10(4), 433–450. doi:10.1080/14616730802461466

Harter, S., & Pike, R. (1984). The pictorial scale of perceived competence and social acceptance for young children. Child Development, 55, 1969–1982. doi:10.2307/1129772

James, W. (1890). The principles of psychology (Vol. 1). New York, NY: Henry Holt. Kernis, M. H. (2005). Measuring self-esteem in context: The importance of stability of self-esteem in

psychological functioning. Journal of Personality, 73, 1569–1605. doi:10.1111/jopy.2005.73.issue-6 Kerns, K. A., Mathews, B. L., Koehn, A. J., Williams, C. T., & Siener-Ciesla, S. (2015). Assessing both

safe haven and secure base support in parent–child relationships. Attachment & Human Development, 17, 337–353. doi:10.1080/14616734.2015.1042487

McCandless, B., & Marshall, H. (1957). A picture sociometric technique for preschool children and its relation to teacher judgments of friendship. Child Development, 28, 139–148. doi:10.2307/ 1125876

Mead, G. (1913). The social self. The Journal of Philosophy, Psychology and Scientific Methods, 10, 374–380. doi:10.2307/2012910

Sroufe, L. A. (1983). Infant-caregiver attachment and patterns of adaptation in preschool: The roots of maladaptation and competence. In M. Perlmutter (Ed.), Minnesota symposium in child psy- chology (Vol. 16, pp. 41–91). Hillsdale, NJ: Erlbaum.

Steele, H., & Steele, M. (2005). Understanding and resolving emotional conflict: The view from 12 years of attachment research across generations and across childhood. In K. Grossmann & E.

ATTACHMENT & HUMAN DEVELOPMENT 597

Waters (Eds.), Attachment from infancy to adulthood: The major longitudinal studies (Chapter 6). New York, NY: Guilford Press.

Suess, G., Grossmann, K. E., & Sroufe, L. A. (1992). Effects of infant attachment to mother and father on quality of adaptation in preschool: From dyadic to individual organisation of self. International Journal of Behavioral Development, 15, 43–65. doi:10.1177/016502549201500103

van IJzendoorn, M., Vereijken, C. M., Bakermans-Kranenburg, M. J., & Riksen-Walraven, M. J. (2004). Assessing attachment security with the attachment Q sort: Meta-analytic evidence for the validity of the observer AQS. Child Development, 75(4), 1188–1213. doi:10.1111/j.1467- 8624.2004.00733.x

Vaughn, B. E., Coppola, G., Verissimo, M., Monteiro, L., Santos, A. J., Posada, G., . . . Korth, B. (2007). The quality of maternal secure base scripts predicts children’s secure base behavior at home in three socio-cultural groups. International Journal of Behavioral Development, 31, 65–76. doi:10.1177/0165025407073574

Vaughn, B., Santos, A. J., & Coppola, G. (2014). Q-methodology and Q-sorting as tools for addres- sing research questions in education settings: Historical overview and illustrations using three standardized Q-sets. In O. Saracho (Ed.), Handbook of research methods in early childhood education (Vol. 1, pp. 175–202). Charlotte, NC: Information Age Publishing.

Vaughn, B. E., & Waters, E. (1981). Attention structure, sociometric status, and dominance: Interrelations, behavioral correlates, and relationships to social competence. Developmental Psychology, 17(3), 275–288. doi:10.1037/0012-1649.17.3.275

Verschueren, K., Buyck, P., & Marcoen, A. (2001). Self-representations and socioemotional compe- tence in young children: A 3-year longitudinal study. Developmental Psychology, 37, 126–134. doi:10.1037/0012-1649.37.1.126

Verschueren, K., Doumen, S., & Buyse, E. (2012). Relationships with mother, teacher, and peers: Unique and joint effects on young children’s self-concept. Attachment & Human Development, 14(3), 233–248. doi:10.1080/14616734.2012.672263

Verschueren, K., & Marcoen, A. (1999). Representation of self and socioemotional competence in kindergartners: Differential and combined effects of attachment to mother and to father. Child Development, 70, 183–201. doi:10.1111/cdev.1999.70.issue-1

Verschueren, K., Marcoen, A., & Schoefs, V. (1996). The internal working model of the self, attachment, and competence in five-year-olds. Child Development, 67, 2493–2511. doi:10.2307/1131636

Waters, E. (1995). Appendix A: The attachment Q-Set (Version 3.0). Monographs of the Society for Research in Child Development, 60(2/3), 234–246. doi:10.2307/1166181

Waters, E., & Deane, K. (1985). Defining and assessing individual differences in attachment relation- ships: Q-methodology and the organization of behavior in infancy and early childhood. In I. Bretherton & E. Waters (Eds.). Monographs of the Society for Research in Child Development, 50(1/2), 41–65. doi:10.2307/3333826

Waters, E., Noyes, D. M., Vaughn, B. E., & Ricks, M. (1985). Q-sort definitions of social competence and self-esteem: Discriminant validity of related constructs in theory and data. Developmental Psychology, 21(3), 508–522. doi:10.1037/0012-1649.21.3.508

598 A. PINTO ET AL.

Copyright of Attachment & Human Development is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.