Anxiety and depression, typically labeled as internalizing symptoms/disorders, are among the most common forms of psychopathology affecting children and adolescents (Costello et al., Reference Costello, Angold, Burns, Stangl, Tweed and Erkanli1996; Last, Perrin, Hersen, & Kazdin, Reference Last, Perrin, Hersen and Kazdin1996), and therefore, it is important to understand factors that may give rise to these conditions. With origins in the study of clinical issues, attachment theory provides one framework for understanding the development of anxiety and depression (Bowlby, Reference Bowlby1969, Reference Bowlby1973). Although studies assessing these associations have begun to emerge in the last decade, a comprehensive review of empirical findings is missing from the literature. One review focused exclusively on family factors in anxiety, but not depression (Bögels & Brechman-Toussaint, Reference Bögels and Brechman-Toussaint2006), and other reviews (e.g., DeKlyen & Greenberg, Reference DeKlyen, Greenberg, Cassidy and Shaver2008; Ranson & Urichuk, Reference Ranson and Urichuk2008) have assessed the relations of attachment with a broad range of symptomatology, all with a limited focus on empirical evidence regarding the associations of parent–child attachment with internalizing problems or, more specifically, anxiety and depression. The purpose of this paper is to review theoretical approaches and to evaluate the empirical evidence for the links between attachment and the development of internalizing problems in childhood and adolescence. In addition, we propose two models that might explain how attachment, in combination with other factors, may relate to anxiety or depression.

Anxiety and Depression: Conceptual Definitions

Internalizing problems are characterized by covert, inner-directed symptoms (e.g., distress) and overcontrolled behaviors (Achenbach & McConaughy, Reference Achenbach, McConaughy and Reynolds1992), and are operationalized in the child and adult literatures as symptoms, syndromes, or diagnoses (Compas, Ey, & Grant, Reference Compas, Ey and Grant1993; Fonseca & Perrin, Reference Fonseca, Perrin, Silverman and Treffers2001). Anxiety disorders include: generalized anxiety disorder, social phobia, specific phobia, obsessive–compulsive disorder (OCD), panic disorder, posttraumatic stress disorder, and separation anxiety disorder, with the latter the only anxiety disorder specific to childhood (American Psychiatric Association [APA], 2000). The common denominator of anxiety disorders is an intense fear or worry associated with avoidant behavior (Kendall, Hedtke, & Aschenbrand, Reference Kendall, Hedtke, Aschenbrand, Wolfe and Mash2006). The most common depressive disorders in children are major depressive disorder and dysthymic disorder, marked by feelings of sadness, diminished energy, and sleep and appetite disturbances (APA, 2000). Diagnostic criteria for depression are similar for children and adults, with two exceptions: irritability may substitute for depressed mood, and the required duration of dysthymic disorder is shorter in childhood. Bipolar disorders are also included among the depressive disorders (APA, 2000), but given the rarity and difficulty of these diagnoses in childhood (Hammen & Rudolph, Reference Hammen, Rudolph, Mash and Barkley2003), bipolar disorders are not discussed in the current review. For simplicity, we will use the terms anxiety and depression as reflecting symptoms and syndromes. However, when we address studies based on diagnoses, we use the term disorder.

Internalizing problems are linked to children's healthy adaptation. High levels of anxiety are associated with maladaptive outcomes such as avoidance of developmentally appropriate activities, higher risk for comorbid conditions such as another type of anxiety disorder, depression, or disruptive behaviors, and difficulties in social adjustment (Albano, Chorpita, & Barlow, Reference Albano, Chorpita, Barlow, Mash and Barkley2003; Fonseca & Perrin Reference Fonseca, Perrin, Silverman and Treffers2001; Ollendick, Shortt, & Sander, Reference Ollendick, Shortt, Sander, Maddux and Winstead2005). High levels of depression are associated with psychosocial disturbances such as academic difficulties and impaired interpersonal relationships, subsequent episodes of depression, increased risk for thoughts of death or suicidal behaviors, and substance abuse during later adolescence (Birmaher et al., Reference Birmaher, Ryan, Williamson, Brent, Kaufman and Dahl1996; Hammen & Rudolph, Reference Hammen, Rudolph, Mash and Barkley2003).

Attachment Theory: Theoretical Links With Anxiety and Depression

Given the cognitive, emotional, and behavioral impairments associated with anxiety and depression, as well as the high costs associated with their treatment (Greenberg et al., Reference Greenberg, Sisitsky, Kessler, Finkelstein, Berndt and Davidson1999), it is important to develop and test theories addressing their etiology. Bowlby (Reference Bowlby1969) suggested that the quality of attachment between parents and children sets the stage for later personality development. Attachment (in children) is typically defined as an emotional long-lasting bond that a child forms with an attachment figure (Ainsworth, Reference Ainsworth1989). Children feel secure in their relationships with attachment figures to the extent that they perceive those figures as consistently available, sensitive, and responsive to their needs. Securely attached children use their caregivers as a secure base from which to explore and as a haven of safety in times of need. The attachment figures' sensitive responses soothe their distress, allowing them to return to their routine. When there is a lack of responsiveness and sensitive care, children form insecure relationships with their attachment figures and, consequently, are less able to use them as a secure base and safe haven.

Ainsworth further expanded Bowlby's tenets and identified two types of insecure relationships: ambivalent (C type) and avoidant (A type; Ainsworth, Blehar, Waters, & Wall, Reference Ainsworth, Blehar, Waters and Wall1978). Ambivalently attached children exhibit fearful and excessive attachment behavior (i.e., heightening of negative emotions) and increased dependence on the attachment figure reflecting attempts to gain the attention of their inconsistently available caregiver (Cassidy & Berlin, Reference Cassidy and Berlin1994; Main & Solomon, Reference Main, Solomon, Brazelton and Yogman1986). Their attachment figure is thought to have difficulty setting limits on children's behavior, to undermine children's autonomy and exploration, and to be inconsistent in responding to the child's distress (Cassidy, Reference Cassidy1994; Cassidy & Berlin, Reference Cassidy and Berlin1994). Avoidantly attached children tend to mask negative affect, to engage in affectively neutral interactions with their attachment figure, and to minimize his or her importance as a source of comfort (Cassidy, Reference Cassidy1994; Main & Solomon, Reference Main, Solomon, Brazelton and Yogman1986). This self-reliance strategy, developed as a consequence of consistent rejection by their attachment figure, particularly in times of distress, serves children by allowing them to avoid further rejection that might be triggered by their attempts for contact. For example, the attachment figures of avoidantly attached children are thought to use a controlling interactional style and to withdraw from interactions when children display negative affect (Belsky & Fearon, Reference Belsky, Fearon, Cassidy and Shaver2008; Cassidy, Reference Cassidy1994). The child then learns that expressing negative affect is inappropriate.

Main and Solomon (Reference Main, Solomon, Brazelton and Yogman1986) documented a third type of insecure relationship, disorganized attachment. Infants with disorganized attachments do not have a coherent and organized strategy to cope with distress in the presence of their caregiver. These children show contradictory, bizarre, and incoherent behaviors as an expression of experiencing a paradoxical situation: the caregiver is at the same time a source of apprehension and the secure base (Lyons-Ruth & Spielman, Reference Lyons-Ruth and Spielman2004; Main & Solomon, Reference Main, Solomon, Brazelton and Yogman1986; van IJzendoorn, Schuengel, & Bakermans-Kraneburg, Reference Van IJzendoorn, Schuengel and Bakermans-Kranenburg1999). Three subgroups of disorganized attachment have been identified, two of which manifest role reversal with the caregiver: controlling–caregiving, including children who focus on guiding and entertaining the parent; controlling–punitive, including children who manifest hostile behaviors toward the parent; and insecure–other or atypical, encompassing children who do not show a clear A or C pattern (Main & Cassidy, Reference Main and Cassidy1986, cited in Moss, Cyr, & Dubois-Comtois, Reference Moss, Cyr and Dubois-Comtois2004). Disorganized attachment is thought to arise if children experience a psychologically unavailable attachment figure, extremely hostile or abusive caregiving, or actual or symbolic loss of the attachment figure (Lyons-Ruth & Jacobvitz, Reference Lyons-Ruth, Jacobvitz, Cassidy and Shaver2008; Moss, Cyr, et al., Reference Moss, Cyr and Dubois-Comtois2004).

Although acknowledging the role played by temperament and genetic factors, Bowlby (Reference Bowlby1973) emphasized that children's concerns about attachment figures' availability constitute the basis of anxiety. Based on their experiential history with caregivers, children learn to predict the attachment figures' availability. If children's attempts to forecast the caregivers' availability fail, particularly when separated from the attachment figures or when experiencing disturbing situations, they respond with fear and anxiety. Further, Bowlby (Reference Bowlby1980) asserted that uncontrollable and prolonged loss, perceived or actual, increases vulnerability to depression. Broadly defined, loss encompasses a wide range of situations such as physical separation, death, and divorce, or conditions other than physical separation that lead to systematic unavailability of the caregiver (e.g., parent suffers a traumatic event). The lack of caregivers' availability that contributes to attachment insecurity promotes perceptions of the self as a failure, perceptions that will be carried out and further reinforced throughout subsequent losses that fuel depressive symptomatology. In sum, Bowlby provided a theoretical basis for speculating that insecure attachment relates to both anxiety and depression in children and adolescents.

With the advent of developmental psychopathology, the investigation of behavioral maladaptation in the context of the study of normal development (Cicchetti, Reference Cicchetti1984; Sroufe & Rutter, Reference Sroufe and Rutter1984), theorists further refined Bowlby's ideas while underscoring the complexities of the association between attachment and internalizing symptoms in general and anxiety or depression in particular. Perry and colleagues (Finnegan, Hodges, & Perry, 1996; Hodges, Finnegan, & Perry, Reference Hodges, Finnegan and Perry1999) proposed that ambivalent attachment, rather then avoidant attachment, is relevant for the development of internalizing symptoms. Their reasoning stems from the fact that ambivalent attachment is characterized by inhibition of autonomy and exploration and difficulty in regulating emotions during minor stressors, which in turn, promote fear responses and self-perceptions of weakness and helplessness; these characteristics are often associated with internalizing symptoms. Perry and colleagues did not discuss disorganized attachment. Some researchers argue that, because disorganized children perceive themselves as helpless and vulnerable in the face of frightening situations and attachment figures as unable to protect them, disorganization may be the attachment pattern most likely to be associated with the development of internalizing symptoms (e.g., Moss, Rousseau, Parent, St-Laurent, & Saintonge, Reference Moss, Rousseau, Parent, St-Laurent and Saintonge1998).

Other researchers made more specific predictions regarding the associations between different insecure attachment patterns and anxiety or depression. Carlson and Sroufe (Reference Carlson, Sroufe, Cicchetti and Cohen1995) suggested that both avoidantly and ambivalently attached children may be vulnerable to develop internalizing symptoms, but only the latter are at greater risk for anxiety. Their prediction is based on Bowlby's (1973) observations that anxiety symptoms (i.e., phobias) are often associated with family patterns in which children worry about the availability or well-being of attachment figures. By the nature of their relationship, ambivalently attached children have concerns about the availability of attachment figures; thus, they are likely to experience fears about leaving home and other related fears. By contrast, Manassis (Reference Manassis, Silverman and Treffers2001) suggested that ambivalent, avoidant, and disorganized attachments may all predispose a child to develop anxiety, but that specific insecure attachment patterns may be associated with specific anxiety disorders/symptoms. Because ambivalently attached children become preoccupied with bidding for the caregivers' attention and are less involved in exploratory behaviors, they may develop separation anxiety. The caregivers of these children may express frustration at children's contact attempts or when the children try to reduce anxiety by being near caregivers, further increasing rather than alleviating separation anxiety. Manassis also speculated that the self-reliance strategy used by avoidantly attached children, along with their perceptions of being rejected by their caregivers, may result in decreased desire for social contact. In temperamentally vulnerable individuals, this desire may lead to social phobia. Alternatively, the restriction and avoidance of the expression of negative emotions may lead to defenses characteristic of obsessive–compulsive symptomatology or to physical expression of distress as somatic complaints. In disorganized attachment, the child sometimes has to look after a psychologically unavailable parent and may become excessively concerned with the caregiver's well-being, which may result in school phobia associated with separation anxiety.

Children with different forms of insecure attachments may all be susceptible to develop depression. For both avoidantly and ambivalently attached children, symbolic or actual loss of attachment figures (i.e., lack of a supportive relationship) may confirm their expectations about the psychological unavailability of attachment figures and may lead to depression (Carlson & Sroufe, Reference Carlson, Sroufe, Cicchetti and Cohen1995). In a more recent paper, Egeland and Carlson (Reference Egeland, Carlson, Atkinson and Goldberg2004) further refined the theoretical link between insecure attachment and depression and postulated, based on Bowlby's (1969) observations, that each insecure attachment pattern is related to depression for different reasons. Ambivalently attached children, who cannot fulfill the caregiver's demands despite repeated attempts, may perceive later difficulties or loss as another failure and develop depression, experienced mainly as helplessness. Avoidantly attached children, who learn through their interactions with attachment figures that they are unlovable or inadequate, may perceive others as hostile rather than supportive, and develop depression experienced as alienation and hopelessness. Children with disorganized attachments, some of whom experience intense traumas, may interpret later difficulties as overwhelming and the self as incapable in the face of challenges, schemas that ultimately would lead to depression.

Literature Review

With the revived interest in the clinical implications of attachment theory, a focal question arises: is attachment indeed related to internalizing symptoms? One major purpose of this paper is to evaluate the empirical evidence for the links between attachment and internalizing symptoms in childhood and adolescence. Specifically, we address several key issues: (a) how consistent is the evidence that attachment security or insecurity is linked to internalizing symptoms, anxiety, and depression? (b) How consistent is the evidence that specific forms of insecurity are more strongly related to internalizing symptoms, anxiety, and depression than are other forms of insecurity? (c) Are associations with internalizing symptoms, anxiety, and depression consistent for mother–child and father–child attachment?

Review format

To facilitate evaluating hypotheses regarding specificity of relations between attachment patterns and particular internalizing problems, we separately summarized the literature for internalizing symptoms (in general), anxiety, and depression. There are some variations in the manifestation of internalizing symptoms between childhood and adolescence, and in the correlates of anxiety and depression (APA, 2000). Our initial intention was to group studies based on the outcome and the age of development (childhood, middle childhood, preadolescence, adolescence). There are, however, very few studies assessing internalizing symptoms in early childhood (age 6 or younger). In addition, many symptoms do not become prevalent until later middle childhood (10–12 years of age). Therefore, for each type of symptom (internalizing symptoms, anxiety, and depression) we summarized separately the data for studies in childhood (up to age 10 years) and preadolescence/adolescence (age 10 years or older). For studies with a large age range, a decision was made based on the mean and/or median age of assessment. When enough information was provided (i.e., means and standard deviations, correlations, etc.), we used standard formulas to compute Cohen's (Reference Cohen1988) effect size.Footnote ² We used Cohen's (Reference Cohen1988) conventions for effect sizes; thus, d = 0.2 was considered a small effect, d = 0.5 was a medium effect, and d = 0.8 was a large effect. A correction factor was applied, when computing effect sizes based on means, to account for the differences in sample sizes and small samples, according to the suggestions of Hedges (Reference Hedges1981). Details about the studies are presented in Tables 1–6. It should be noted that the majority of the samples were predominantly Caucasian (for exceptions, see Anan & Barnett, Reference Anan and Barnett1999; Constantine, Reference Constantine2006) and were drawn from the Untied States, Canada, or Europe. Although a significant number of studies employed a longitudinal design (23 studies), cross-sectional designs were commonly used (32 studies).

Table 1.

Studies of attachment and childhood internalizing symptoms

Note: For study type: L, longitudinal; CS, cross-sectional. For ethnicity: 1, predominantly Caucasian; 2, majority Caucasian, but some other ethnicities are relatively well represented; 3, predominantly African American. AF, attachment figure. For attachment figure: M, mother, P, parent. For internalizing symptoms, age is reported in years (range or mean). For symptoms informant: C, child; Cg, caregiver; F, father; M, mother; O, observation; P, parent; T, teacher. For comparison: A, avoidant; B, secure; C, ambivalent; D, disorganized.

^aOverlapping samples.

^bPositive association.

†p < .10. *p < .05. **p < .01. ***p < .001.

Table 2.

Studies of attachment and preadolescent or adolescent internalizing symptoms

Note: For study type: L, longitudinal; CS, cross-sectional. For ethnicity: 1, predominantly Caucasian, 2, majority Caucasian, but some other ethnicities are relatively well represented. AF, attachment figure. For attachment figure: M, mother; F, father; P, parent. For internalizing symptoms, age is reported in years (range or mean). For symptoms informant: C, child; T, teacher. For comparison: A, avoidant; B, secure; C, ambivalent; D, disorganized.

^aStudies included in more than one category.

†p < .10. *p < .05 **p < .01. ***p < .001.

Table 3.

Studies of attachment and childhood anxiety

Note: For study type: L, longitudinal; CS, cross-sectional. For ethnicity: 1, predominantly Caucasian; 2, majority Caucasian, but some other ethnicities are relatively well represented. AF, attachment figure. For attachment figure: M, mother. For anxiety, age is reported in years (range or mean). For anxiety informant: M, mother; F, father; C, child; T, teacher; I, interview; Dx. Int., diagnostic interview. For anxiety type: S, symptoms/syndrome; D, disorder. For comparison: A, avoidant; B, secure; C, ambivalent.

^aStudies included in more than one category.

†p < .10. *p < .05.

Table 4.

Studies of attachment and preadolescent or adolescent anxiety

Note: For study type: L, longitudinal; CS, cross-sectional. For ethnicity: 1, predominantly Caucasian; 2, majority Caucasian, but some other ethnicities are relatively well represented. AF, attachment figure. For attachment figure: M, mother; F, father; P, parent. For anxiety, age is reported in years (range or mean). For anxiety informant: M, mother; C, child; T, teacher; Dx. Int., diagnostic interview. For anxiety type: S, symptom/syndromes; D, disorder. For comparison: A, avoidant; B, secure; C, ambivalent.

†p < .10. *p < .05.**p < .01. ***p < .001.

Table 5.

Studies of attachment and childhood depression

Note: For study type: L, longitudinal; CS, cross-sectional. For ethnicity: 1, predominantly Caucasian. AF, attachment figure. For attachment figure: M, mother, F, father; P, parent. For depression, age is reported in years (range or mean). For depression informant: C, child. For depression type: S, symptoms/syndromes. For comparison: A, avoidant; B, secure; C, ambivalent; D, disorganized.

†p < .10. *p < .05.**p < .01. ***p < .001.

Table 6.

Studies of attachment and preadolescent or adolescent depression

Note: For study type: L, longitudinal; CS, cross-sectional. For ethnicity: 1, predominantly Caucasian; 2, majority Caucasian, but some other ethnicities are relatively well represented; 3, predominantly African American. AF, attachment figure. For attachment figure: M, mother; F, father; P, parent. For depression, age is reported in years (range or mean). For depression informant: C, child, Dx. Int., diagnostic interview. For depression type: S, symptoms/syndrome; D, disorder. For comparison: B, secure.

*p < .05. **p < .01. ***p < .001.

Attachment and Internalizing Symptoms

Studies of the Associations Between Attachment and Anxiety

Studies of the Associations Between Attachment and Depression

General Discussion of Findings

Moderators of the Relations Among Attachment and Internalizing Symptoms, Anxiety, or Depression

The modest association between attachment and internalizing problems is not surprising, given that insecure attachment, by itself, will not invariably lead to anxiety or depression. Indeed, from a developmental psychopathology perspective, any factor may produce different effects depending on the circumstances under which it functions (the principle of multifinality, Cicchetti & Cohen, Reference Cicchetti, Cohen, Cicchetti and Cohen1995). For example, Cummings and Cicchetti (Reference Cummings, Cicchetti, Greenberg, Cicchetti and Cummings1990) proposed a transactional model of childhood/adolescent depression in which the effects of insecure attachment depend on potentiators (“factors increasing the probability of insecure attachment leading to depression”) and compensators (“factors decreasing the chances of development of depression”). Thus, attachment may be consistently associated with internalizing problems only in the presence of other risk factors.

A limited number of studies have assessed other factors (i.e., moderators) that might identify the conditions under which attachment is related to internalizing symptoms, anxiety, and depression. Alternatively, some researchers conceptualized attachment as a moderator (Easterbrooks et al., Reference Easterbrooks, Davidson and Chazan1993; El-Sheikh & Buckhalt, Reference El-Sheikh and Buckhalt2003; Graham & Easterbrooks, Reference Graham and Easterbrooks2000; Gullone et al., Reference Gullone, Ollendick and King2006; NICHD, 2006). Researchers have proposed eight factors that may moderate the relation of attachment with internalizing symptoms. Specifically, it has been hypothesized that insecurely attached children who also experience high levels of stress (Dallaire & Weinraub, Reference Dallaire and Weinraub2007), poor parenting (Marsh et al., Reference Marsh, McFarland, Allen, McElhaney and Land2003; NICHD, 2006), low autonomy (Noom et al., Reference Noom, Deković and Meeus1999), economic risk (Easterbrooks et al., Reference Easterbrooks, Davidson and Chazan1993; Graham & Easterbrooks, Reference Graham and Easterbrooks2000), behavioral inhibition, or other temperamental characteristics (Burgess et al., Reference Burgess, Marshall, Rubin and Fox2003; Shamir-Essakov et al., 2005) or those who exhibit excessive reassurance seeking (Abela et al., Reference Abela, Hankin, Haigh, Adams, Vinokuroff and Trayhern2005) would show greater internalizing symptoms, anxiety, or depression than those who do not experience these risk factors. In addition, it has been hypothesized that a secure parent–child attachment would be a protective factor in the association between parental problem drinking and children's internalizing symptoms (El-Sheikh & Buckhalt, Reference El-Sheikh and Buckhalt2003), that the buffering effect of attachment on anxiety or depression may be moderated by children's pubertal status (Papini et al., Reference Papini, Roggman and Anderson1991), and that the relation between attachment and depression would be stronger for girls than for boys (DiFilippo & Overholser, 2000).

Evidence supporting these hypotheses is inconsistent, and based on a limited number of studies. Although one study showed that attachment security at 15 months but not at 36 months interacted with stress to predict mothers' and teacher's reports of child anxiety (Dallaire & Weinraub, Reference Dallaire and Weinraub2007), two other studies indicated that the interaction between attachment and stress was not related to childhood internalizing symptoms (Bates et al., Reference Bates, Maslin and Frankel1985) or adolescent depression (Sund & Wichtrøm, 2002). In the NICHD (2006) study, the mean of parenting quality over the first years of life, but not changes in parenting quality, interacted with attachment to predict mother reported but not teacher reported internalizing problems in childhood. Parenting quality was more strongly related to internalizing symptoms for children with secure attachment than for children with ambivalent or disorganized attachments. Marsh et al. (Reference Marsh, McFarland, Allen, McElhaney and Land2003) found that ambivalence was more strongly related to anxiety and depression in adolescence when mothers displayed low levels of autonomy in interactions with their adolescents, whereas Noom et al. (Reference Noom, Deković and Meeus1999) reported that adolescents' attitudinal autonomy did not moderate the relation between attachment security and internalizing symptoms. Easterbrooks et al. (1993) found that economic risk did not moderate the relation of attachment with internalizing symptoms in childhood. By contrast, Graham and Easterbrooks (Reference Graham and Easterbrooks2000) found that insecure children at higher economic risk had higher levels of depression than insecure children at lower economic risk. Four studies showed that behavioral inhibition or difficult temperament failed to moderate the relations of attachment to childhood internalizing symptoms (Bates et al., Reference Bates, Maslin and Frankel1985; Burgess et al., Reference Burgess, Marshall, Rubin and Fox2003; Stams et al., Reference Stams, Juffer and van IJzendoorn2002) and childhood anxiety (Shamir-Esakov et al., 2005), although one study indicated that the interaction effect between social withdrawal and attachment was associated with childhood depression (Gullone et al., Reference Gullone, Ollendick and King2006). Abela et al. (Reference Abela, Hankin, Haigh, Adams, Vinokuroff and Trayhern2005) reported that children who exhibited high levels of both insecure attachments and excessive reassurance seeking experienced higher levels of depression than did children experiencing only one or neither of these risk factors. El-Sheikh and Buckhalt (Reference El-Sheikh and Buckhalt2003) reported that the interaction effect between security and parental drinking problems did not predict children's internalizing symptoms, and Papini et al. (Reference Papini, Roggman and Anderson1991) showed that pubertal status did not moderate the relations of attachment with anxiety or depression in adolescence.

Although many studies used gender as a covariate, very few studies further investigated whether attachment was associated with internalizing symptoms differently for boys and girls. Some studies showed that the relations of attachment security/insecurity (or insecure attachment patterns) with childhood internalizing symptoms (Anan & Barnet, 1999; Shaw et al., Reference Shaw, Keenan, Vondra, Delliquadri and Giovannelli1997; see also Goldberg et al., Reference Goldberg, Gotowiec and Simmons1995; Noom et al., Reference Noom, Deković and Meeus1999), childhood anxiety (Bohlin et al., Reference Bohlin, Hagekull and Rydell2000), adolescent anxiety (Bar-Haim et al., Reference Bar-Haim, Dan, Eshel and Sagi-Schwartz2007; Larose, & Boivin, Reference Larose and Boivin1997), and adolescent depression (DiFilippo & Overholser, 2000; Sund & Wichstrom, 2002) were similar for girls and boys (for an exception on social anxiety see Bar-Haim et al., Reference Bar-Haim, Dan, Eshel and Sagi-Schwartz2007). One study reported a relation between attachment classification and childhood internalizing symptoms for boys only, with securely attached boys scoring the lowest and ambivalently attached boys scoring the highest (Lewis et al., Reference Lewis, Feiring, McGuffog and Jaskir1984), and one study showed a relation between avoidant attachment and internalizing symptoms for boys only (Moss et al., Reference Moss, Rousseau, Parent, St-Laurent and Saintonge1998). However, Moss and colleagues (1998) also reported no significant effects for the interactions of ambivalence or disorganization with gender. One study indicated a stronger relation of attachment security and adolescent depression for girls than for boys (Allen et al., Reference Allen, Porter, McFarland, McElhaney and Marsh2007) and another study showed a stronger relation for boys than for girls (Kenny et al., Reference Kenny, Moilanen, Lomax and Brabeck1993). Thus, gender did not consistently moderate associations between attachment and internalizing symptoms.

Although no specific hypotheses were offered, a few other studies have examined potential moderators (i.e., included the interaction term in the statistical analyses). These studies examined child age (Hodges et al., Reference Hodges, Finnegan and Perry1999; Moss, Cyr, et al., 2004; NICHD, 2006; Noom et al., Reference Noom, Deković and Meeus1999), health status (Goldberg et al., Reference Goldberg, Gotowiec and Simmons1995), mental development (Lyons-Ruth et al., Reference Lyons-Ruth, Easterbrooks and Cibelli1997), or parental depression (Graham & Easterbrooks, 2003). None of these factors showed moderation effects.

In sum, despite interest regarding how other risk factors might increase the likelihood that insecure attachment will be associated with internalizing problems, the available data indicate that no moderator consistently affects the direction or strength of the relations between attachment and internalizing symptoms, anxiety, or depression. It may be that the number of risk factors combined with insecure attachment is more important than experiencing attachment insecurity in combination with a specific risk factor.

Attachment and Internalizing Symptoms: Etiological Models

Although secure attachment was associated with lower levels of internalizing symptoms, our review was inconclusive regarding a link between specific insecure attachment patterns and internalizing problems. Thus, insecure attachment may be a general risk factor for internalizing problems. In this last section, our goal is to embed attachment within a broader framework to generate hypotheses regarding how insecure attachment might, in combination with other risk factors, be related to internalizing symptoms. Considering other risk factors may allow for testing potential pathways from attachment to internalizing symptoms. One possibility is that insecure attachment is a direct cause of later internalizing symptoms (i.e., lack of a secure base leads to feelings of depression and anxiety). Alternatively, links between attachment and internalizing problems may be indirect, in that these relations are mediated or explained by other factors.

We were able to locate only six studies that tested specific mediators. Two studies showed that perceived social support explained the relations of attachment with internalizing symptoms (Anan & Barnett; 1999; Larose & Boivin, 2001). Note that social support has a strong conceptual overlap with attachment. Another study found that attachment security affects childhood anxiety through effects on emotion regulation, and adolescent anxiety through effects on competence and peer relationships (Bosquet & Egeland, Reference Bosquet and Egeland2006). Regarding depression, one study indicated that negative cognition but not ruminative coping mediated the relation between attachment security and depression (Margolese et al., Reference Margolese, Markiewicz and Doyle2005, for girls only), and two studies showed that the view of the self or self-esteem explain this relation (Kenny et al., Reference Kenny, Moilanen, Lomax and Brabeck1993; Wilkinson, Reference Wilkinson2004). Although small in number, these studies suggest that it may be important to examine other factors that can account for links between attachment and internalizing problems.

To advance research in this area, we propose two models of how attachment may be related to, respectively, anxiety and depression. In these models, we include other known risk factors for anxiety and depression and propose specific mediators. DeKlyen and Greenberg (Reference DeKlyen, Greenberg, Cassidy and Shaver2008) proposed that insecure attachment may combine with high family adversity, ineffective parenting, and atypical child characteristics in predicting psychopathology, including internalizing symptoms. Although general models are important for identifying risk factors, they offer little guidance regarding which specific factors may contribute to one form of psychopathology rather than another. We elaborate and expand on DeKlyen and Greenberg's suggestions by identifying potential moderators and mediators of the relations between attachment and internalizing symptoms, and proposing specific pathways from attachment to anxiety and depression. For example, although ineffective parenting is linked to both anxiety and depression, parental excessive control may be more strongly related with anxiety, whereas parental rejection may be more strongly related to depression (Rapee, Reference Rapee1997). Although there are gender differences in the rates and prevalence of internalizing problems by adolescence (Albano et al., Reference Albano, Chorpita, Barlow, Mash and Barkley2003; Hammen & Rudolph, Reference Hammen, Rudolph, Mash and Barkley2003), we did not find consistent results supporting the notion that gender may be a potential moderator of associations between attachment and internalizing problems, and therefore we did not include gender in these models.

Prior models of internalizing symptoms delineate common risk factors for anxiety and depression, such as parenting practices (e.g., parental rejection or overcontrol), temperament (e.g., negative emotionality), parent's psychopathology, and stressful life events (e.g., Rapee, Reference Rapee1997; Seligman & Ollendick, Reference Seligman and Ollendick1998), which may explain the comorbidity of anxiety and depression and/or the presence of internalizing symptoms. Even when models of internalizing symptoms have included attachment as a risk factor, they have rarely identified mechanisms that may explain the relations of attachment with internalizing symptoms/disorders. Our models extend earlier work by specifying three child characteristics that may explain links between attachment and child internalizing symptoms: cognitive biases, difficulty with emotions, and self-concept. Cognitive biases refer to inaccurate beliefs, interpretations, attributions, or expectancies in relation to specific events. Emotional problems include impairments in identifying and understanding emotional states and adoption of less constructive coping strategies. Self-concept refers to one's perceived efficacy and self-worth. All three child characteristics have been associated with both anxiety and depression (e.g., Abela & Hankin, Reference Abela, Hankin, Abela and Hankin2008; Compas, Connor-Smith, Saltzman, Thomsen, & Wadsworth, Reference Compas, Connor-Smith, Saltzman, Thomsen and Wadsworth2001; Durbin & Shafir, Reference Durbin, Shafir, Abela and Hankin2008; Southam-Gerow & Kendall, Reference Southam-Gerow and Kendall2002; Vasey & MacLeod, Reference Vasey, MacLeod, Vasey and Dadds2001), and all three are shaped (in part) through early relationships and interactions with parents (e.g., Dadds, Barrett, Rapee, & Ryan, Reference Dadds, Barrett, Rapee and Ryan1996; Thompson, Reference Thompson, Vasey and Dadds2001; Thompson & Meyer, Reference Thompson, Meyer and Gross2007).

Pathways to anxiety are likely to start with dysfunctional parent–child relationships (Figure 1). Specifically, transactions between insecure attachment and overcontrolling and overprotective parenting may lead to thoughts of physical or psychological threat, perceived lack of control, and vigilant attention to threat (Chorpita & Barlow, Reference Chorpita and Barlow1998; Thompson, Reference Thompson, Vasey and Dadds2001; Vasey & MacLeod, Reference Vasey, MacLeod, Vasey and Dadds2001; Weems & Silverman, Reference Weems and Silverman2006). Lack of open communication regarding emotion and an inability to rely on the parent for comfort may lead to the child developing difficulties in monitoring, appropriately expressing, or understanding emotions as well as the use of avoidance rather than problem solving as a coping strategy (Barrett, Rapee, Dadds, & Ryan, Reference Barrett, Rapee, Dadds and Ryan1996; Suveg & Zeman, Reference Suveg and Zeman2004; Thompson, Reference Thompson, Vasey and Dadds2001). Minimal opportunities for exploration may foster a reduced sense of autonomy, control, or self-efficacy, as insecurely attached children lack opportunities to develop new skills or to explore the environment as well as to rely on the parent in times of need (Chorpita & Barlow, Reference Chorpita and Barlow1998; Chorpita, Brown, & Barlow Reference Chorpita, Brown and Barlow1998). Maladaptive cognitions, emotion regulation processes, and self-concepts may signal a risk for anxiety when children experience stress associated with anticipated danger (Barlow, Reference Barlow2002). Parental anxiety may also play a role as children may model parents' anxious behavior and avoidant coping strategies (Barrett et al., Reference Barrett, Rapee, Dadds and Ryan1996; Dadds et al., Reference Dadds, Barrett, Rapee and Ryan1996). Negative emotionality and behavioral inhibition may have a direct influence on children's anxiety, or may exacerbate the effect of coping strategies on children's anxiety (Lonigan & Phillips, Reference Lonigan, Phillips, Vasey and Dadds2001). For example, avoidant coping strategies may be linked with anxiety only for children who are temperamentally predisposed to experiencing high negative emotions or who have difficulty adjusting to novel situations (e.g., Lonigan & Phillips, Reference Lonigan, Phillips, Vasey and Dadds2001).

Figure 1.

Model of attachment influences on anxiety in the context of other factors. NE, negative emotionality; BI, behavioral inhibition.

Pathways to depression are also likely to stem from dysfunctional parent–child relationships, as shown in Figure 2. Attachment insecurity, in combination with parental rejection, may predispose a child to infer that he or she is to blame for such parental behaviors, to take responsibility for these early disturbed interactions, and to develop a depressogenic inferential style (e.g., to attribute negative outcomes to internal, global, and stable factors; Abela & Hankin, Reference Abela, Hankin, Abela and Hankin2008; Hankin & Abela, Reference Hankin, Abela, Hankin and Abela2005). Because authentic expression of their feelings is likely to trigger more parental rejection, these children may learn to overregulate their display of emotions, may have difficulty understanding their emotions, and may come to rely on ineffective coping strategies such as passive or ruminative coping and helplessness, denial of the negative experience, as well as decrease assertiveness in the face of conflict or emotionally challenging event (Casey, Reference Casey, Lewis and Sullivan1996; Ebata & Moss, Reference Ebata and Moss1991; Kobak & Ferenz-Gillies, Reference Kobak and Ferenz-Gillies1995; Silk, Steinberg, & Morris, Reference Silk, Steinberg and Morris2003; Southam-Gerow & Kendall, Reference Southam-Gerow and Kendall2002; see also Hammen & Rudolph, Reference Hammen, Rudolph, Mash and Barkley2003). Early disturbed parent–child relationships may also fuel a negative self-view and diminished self-worth (Hammen & Rudolph, Reference Hammen, Rudolph, Mash and Barkley2003). Children already predisposed to a depressogenic inferential style, with difficulties regulating their emotions, and with a low self-worth, may develop depression when encountering personal disappointment, achievement failure, and loss-related stressors (Goodyer, Reference Goodyer and Goodyear2001; Lakdawalla, Hankin, & Mermelstein, Reference Lakdawalla, Hankin and Mermelstein2007). Parental depression may exacerbate these links as children may model the parent's depressed cognitive style and affect (Silk, Shaw, Skuban, Oland, & Kovacs, Reference Silk, Shaw, Skuban, Oland and Kovacs2006; see also Berg-Nielsen, Vikan, & Dahl, Reference Berg-Nielsen, Vikan and Dahl2002; Lovejoy, Graczyk, O'Hare, & Neuman, Reference Lovejoy, Graczyk, O'Hare and Neuman2000). Depression may be especially likely when the child is predisposed to high level of negative emotions or low levels of positive emotions (Chorpita, Albano, & Barlow, Reference Chorpita, Albano and Barlow1998).

Figure 2.

Model of attachment influences on depression in the context of other factors. NE, negative emotionality; PE, positive emotionality.

Our review suggested that associations of attachment with anxiety and depression may be stronger at older ages. Patterns of information processing, emotion regulation, and self-views may become more crystallized and automatic with age. Patterns may also become more self-perpetuating over time. Thus, we predict that models will be more strongly confirmed for older children for whom mediating processes are less flexible.

The proposed models are speculative and are offered as a framework to guide future research. In some ways, the models are oversimplified. For example, there are likely to be interactive and cumulative effects that play out over time. Longitudinal studies are also needed to investigate the possibility that feelings of anxiety or depression may further undermine children's sense of security with caregivers (Roisman et al., Reference Roisman, Fortuna and Holland2006). In addition, it is still possible that specific patterns of insecure attachment, rather than insecurity per se, are related to different types of internalizing symptoms (i.e., anxiety and depression, different types of anxiety symptoms) in the context of other risk factors, but that the available research failed to capture these differences. Finally, the models provide only a limited integration of biological factors. Although temperament and parent's psychopathology may in part capture genetic contributions, insecure attachment may also affect the neurophysiological and biochemical processes that serve as biological foundations for internalizing symptoms. Our heuristic models focus mainly on environmental factors and do not detail biological influences on internalizing symptoms. Nonetheless, these pathways offer a starting point in the process of disentangling the dynamic interplay between attachment and other factors in predicting anxiety and depression in children and adolescents and lead to several testable hypotheses.