ARTICLE IN PRESS

Journal of
Adolescence
Journal of Adolescence 28 (2005) 325–341
www.elsevier.com/locate/jado

Evolved mechanisms in adolescent anxiety and depression

symptoms: the role of the attachment and social rank systems
C. Irons, P. Gilbert
Mental Health Research Unit, Kingsway Hospital, Derby DE22 3LZ, UK

Abstract

One hundred and forty adolescent students were assessed on measures of attachment, social rank (social
comparison and submissive behaviour), and depression and anxiety symptoms. Secure attachment was
significantly correlated with positive social comparison and inversely with submissive behaviour, depression
and anxiety symptoms. In contrast, insecure attachment of both avoidance and ambivalence was associated
with unfavourable comparison with others, and positively correlated with submissive behaviour, depression
and anxiety symptoms. Exploring the relationship of attachment with depression and anxiety symptoms
revealed that this link might have different routes through social rank perceptions. For secure attachment,
social rank concerns (i.e. social comparison and submissive behaviour) did not mediate the linkage with
anxiety or depression symptoms. However, for insecure attachment, social rank concerns showed either a
partial or complete mediation of these relationships. This data may indicate that insecure attachment
sensitizes individuals to become focused on the competitive dynamics of groups, the power of others to
shame, hurt or reject, and the need to defend against these possibilities.
r 2004 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights
reserved.

Introduction

Adolescence is marked with major changes in physiology, interpersonal relationships and social
interest. With the onset of adolescence, the influence of parents tends to lessen, and peers become
Corresponding author.
E-mail address: p.gilbert@derby.ac.uk (P. Gilbert).

0140-1971/$30.00 r 2004 The Association for Professionals in Services for Adolescents. Published by Elsevier Ltd. All
rights reserved.
doi:10.1016/j.adolescence.2004.07.004
 ARTICLE IN PRESS

326 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

an increasingly important source of support, values and sense of belonging (Buhrmester, 1996).
For some individuals, the transition through adolescence, from family support to peer support,
can be problematic. Indeed, the transition from childhood to adolescence is also conspicuous by
the substantial rise in the prevalence of mental health problems, including depression (Fleming,
Offord, & Boyle, 1989; Sorenson, Rutter, & Aneshensel, 1991; Angold & Rutter, 1992) and
anxiety (Kashani & Orvaschel, 1990). There have been a number of explanations of such
difficulties, including the mood regulating impact of pubertal hormones (Angold & Worthman,
1993), the competitive dynamics of adolescent peer relationships that carry risk of bullying and
rejection (Prinstein, Boergers, Spirito, Little, & Grapentine, 2000) and increasing stress related to
changes in life events and daily hassles (Campus, 1987; Larson & Ham, 1993).
Following the work of Bowlby (1969, 1973, 1980), attachment theorists have explored the way
early attachment relationships can influence different relating styles marked by security, anxiety
and/or avoidance (e.g. Collins & Feeney, 2000). These early experiences can lay the foundation for
inner models of self in relation to others that convey feelings of security and benevolence of
others, and of the self as able, worthy and desirable to others. These may significantly aid the
transition into adolescence. Secure children tend to have more prosocial and supportive peer
relationships than insecure children (Blain, Thompson, & Whiffen, 1993; Van Lange, Otten, De
Bruin, & Joireman, 1997), and are themselves prosocial to others (Eisenberg, 2002). A number of
studies (e.g. Mikulincer, Birnbaum, Woddis, and Nachmias, 2000) have found that securely
attached people see others as relatively benevolent and can regulate stress by appropriate support
seeking and self-management, while anxiously attached people show typical ‘protest’ behaviour to
stress and become clingy and controlling. Poor, neglectful or abusive early relationships can lead
to internal models of insecurity, with others being seen as unavailable or hostile, and the self as
unworthy and undesirable. Insecure attachment can be conceptualized in dimensional (Brennan,
Clark, & Shaver, 1998) or categorical terms. In the categorical classification, there are two types of
insecurity derived from the work of Ainsworth, Blehar, Waters, and Wall (1978). These are an
ambivalent attachment, identified by the tendency to become dependent, help-seeking, fear of
abandonment and anxious/angry at separation. In contrast, avoidant attached people respond
with withdrawal to separation, do not seek help for stress, do not have views of others as
benevolent. They also use repression as a coping mechanism, and are overly self-reliant (Collins &
Feeney, 2000; Mikulincer et al., 2000).
A key research question is how early attachment experiences and attachment style impact on a
variety of social strategies. For example, Belsky, Steinberg, and Draper (1991) and Belsky (1993)
used an evolutionary model to suggest that early environments (social niches) act to select which
social and reproductive strategies the child will come to adopt (e.g. affiliative, stable pair bonding
and high investment in offspring vs. less affiliative, unstable pair bonding and low investment in
offspring). Delineating the main factors involved in adverse rearing environments and their
influence on subsequent social behaviour and pathology is one of the challenges for researchers.
Gilbert (1989, 1993, 2002) suggested that children adapt to their social niche with two basic
‘decisions’. First are evaluations of the relative threat and safeness of environments. Threatening
environments stimulate stress regulation behaviours (e.g. fight-flight, submission; Perry, Pollard,
Blakley, Baker, & Vigilante, 1995); non-threatening environments, especially those associated
with empathic warmth, stimulate more relaxed attention and positive affect (Gilbert, 1993;
Eisenberg, 2002). Second, the child has to adopt specific roles for various domains of relating
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 327

(e.g. caring for others vs. exploiting, competing for social place vs. secure affiliation). In unsafe
environments, children are more likely to develop a defensive orientation and become focused on
the power of others to harm, shame or reject, and thus, are highly attuned to social rank and their
place within such ranks. This inclines them to be especially attuned to competitive dynamics,
needing to compete to earn their place, concerns with relative inferiority–superiority and rejection,
and easily activated into defensive behaviours of submissiveness, appeasement, avoidance or
aggressive control. In safe environments, there is less activation of defensive emotions and
behaviours, with a greater likelihood of being able to engage in social roles of social affiliation and
cooperation. These inner orientations towards the social world can then be carried over into peer
group relating. In addition, different social styles of relating impact on the social environment
which can then act to reinforce certain social strategies. For example, it is known that anxious
children who are also submissive can be less liked and more rejected by their peers. This confirms
their ‘belief’ that the environment is potentially rejecting (Spence, Donovan, & Brechman-
Toussaint, 1999). Troy and Sroufe (1987) found children with an avoidant attachment style were
more likely to be bullies, whilst those with characteristically ambivalent attachment styles were
more likely to be the victims of bullying. There is also evidence that children can oscillate between
being bullies and victims (Smith & Myron-Wilson, 1998).
As children enter adolescence, they become increasingly concerned with peer belonging and
their social standing amongst their peers (Baumeister & Leary, 1995). Insofar as adolescent peer
relations can focus on competing for social approval, acceptance and popularity with peers, as well
as the avoidance of rejection and ostracism, adolescents can be particularly sensitive to social
comparison and concerns with their relative rank amongst their peers. Because secure children
enter this frame feeling relatively good about themselves and others, they may quickly settle into
creating cooperative and supportive networks. Insecure children, however, may enter this domain
more anxious, submissive or concerned to dominate and may recreate the dynamics of early
family relationships. There is now good evidence that excessive concerns with feeling inferior to
others, a tendency for submissive behaviour and believing that others look down on the self are
highly associated with high levels of depression and anxiety symptoms in adults (Allan, Gilbert, &
Goss, 1994; Allan & Gilbert, 1995; Gilbert, 2000). A key research question is therefore the way
in which secure and insecure attachment styles interact with concerns and evaluations about
social rank (e.g. social comparison) and defensive behaviour (e.g. submissive behaviour). A
number of authors have indicated the interactions between these two domains of social relating
(e.g. Sloman, 2000), and there impact upon depression (Sloman, Gilbert, & Hasey, 2003), but
research remains scant.

Aims

To date there is little research with adolescents on how attachment styles and social rank styles
are related. Hence, this study sought to explore how attachment (in)securities are associated with
more defensive, competitive (social rank) functioning (i.e. forms of social comparison and
submissive behaviour). We also sought to explore how the relationship between these two systems
may be related to adolescent depression and anxiety symptoms. In particular, we explored
whether social rank may act as a moderating or mediating factor between attachment and anxious
and depressive symptoms.
 ARTICLE IN PRESS

328 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

Methodology

Participants

One hundred and forty adolescents (73 male, 63 female and 4 unrecorded) from a Year 10
(mean age 14.63) comprehensive school in Nottingham completed a series of self-report
questionnaires. The participating school was located in a suburban area of Nottingham. The
school had a relatively large catchment area covering a variety of socio-economic groups.
Adolescents who completed the questionnaires were from mixed ethnic backgrounds: 63% were
White British, 21% Asian, 4% Black Afro-Caribbean, 4% Mixed Race and 8% ‘Other’.
Two other schools in the area were also contacted but felt they were too busy at the time to
allow them to participate in the study. Participants completed the questionnaire packs in their
own classrooms with their class Teacher present.

Procedure

Ethics approval was granted by the University of Derby Psychology Ethics Committee. After
looking through and discussing the questionnaires involved, the Head Teacher of the school
agreed to act in loco parentis. Before testing, an introductory statement was read to the children
by the researcher, outlining the purpose of the study, aspects of confidentiality and consent, and
the conditions under which the testing would proceed. Participants were fully debriefed after
testing. This was performed by the teachers who were familiar with the participants.

Measures

Attachment questionnaire for children

Sharpe et al. (1998) developed the attachment questionnaire for children (AQ-C) as a simplified
version of Hazen and Shaver’s (1987) attachment measure. Sharpe et al. adapted the scale for
children and adolescents to tap attachment styles with their friends. Participants are given three
vignettes describing characteristic behaviours and affects in relationships. Each description
represents one of the three attachment styles (secure, avoidant and ambivalent) described by
Ainsworth, Blehar, Waters, and Wall (1978). Participants are instructed to select the description
that best describes them. The descriptions are (1) ‘I find it easy to become close friends with other
children. I trust them and I am comfortable depending on them. I do not worry about being
abandoned or about another child getting too close friends with me’ (secure attachment); (2) ‘I am
uncomfortable to be close friends with other children. I find it difficult to trust them completely,
difficult to depend on them. I get nervous when another child wants to become close friends with
me. Friends often come more close to me than I want them to’ (avoidant attachment); (3) ‘I often
find that other children do not want to get as close as I would like them to be. I am often worried
that my best friend doesn’t really like me and wants to end our friendship. I prefer to do
everything together with my best friend. However, this desire sometimes scares other children
away’ (ambivalent attachment).
For this study we added a Likert-based scoring thus providing two measures (categorical and
dimensional measures) for each of the attachment styles. This provides further quantitative
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 329

information to the categorical choice and is conceptually similar to adaptations of adult

categorical attachment measures (Levy & Davis, 1988; Collins & Read, 1990; Shaver & Brennan,
1992). Thus, participants select which vignette most accurately describes them, and also mark on a
1 (unlike me) to 7 (like me) scale the degree to which each vignette was like them.

Adolescent social comparison scale-revised

The adolescent social comparison scale-revised (ASCS-R) was adapted from the adult Social
Comparison Scale (Allan & Gilbert, 1995), collaboratively by one of the authors (P.G.) for use in
a clinical project (Lang, 1994). Lang (1994) reported an internal consistency of .78. Through a set
of bipolar constructs, participants make global comparisons about themselves in relation to other
people, rated on a 10-point Likert scale. For example, ‘Compared to your friends, how confident
do you feel?’
Less Confident 1 2 3 4 5 6 7 8 9 10 More Confident
The ASCS is a 10-item scale, with scores ranging from 10 to 60. High total scores represent a
more positive social comparison (i.e. participants feel more popular, attractive and accepted in
comparison to their friends). The full scale is given in Appendix A.

Adolescent submissive behaviour scale

The adolescent submissive behaviour scale (ASBS) is a 12-item measure collaboratively adapted
from the adult version of the Submissive Behaviour Scale (Gilbert & Allan, 1994; Allan & Gilbert,
1997) by one of the authors (P.G.) for use in a clinical project (Lang, 1994). Participants rate how
they would behave in potentially ‘conflict’ situations, for example; ‘I do things because others are
doing them, rather than because I want to’. The scale measures the degree of submissive choice to
each situation. It utilizes a five-point Likert scale and has a total score range of 0–60, with higher
scores indicative of greater submissive behaviour. The full scale is given in Appendix B.

Children’s depression inventory

The children’s depression inventory (CDI) is a 27-item questionnaire developed by Kovacs
(1992). The CDI is a widely used self-report depression symptom scale for children and
adolescents. It is used in both clinical and non-clinical samples. It requires the lowest reading age
of any child depression measure (Kazdin & Petti, 1982; Berndt, Schwartz, & Kaiser, 1983). There
are five subscales to the CDI: Negative Mood, Interpersonal Problems, Ineffectiveness,
Anhedonia and Negative Self-esteem. In this study, we used the total score of these subscales.
Questions are scored on a three-point Likert scale, with 0 ¼ absence of symptom, 1 ¼ mild
symptom and 2 ¼ definite symptom. In this study, item number 9, which asks about suicide
ideation, was dropped in accordance to the wishes of the school Head Teacher. This omission is in
line with a number of previous studies (e.g. Laible, Carlo, & Raffaelli, 2000; Hewitt et al., 2002)
and does not reduce alpha reliability statistics.

Spence children’s anxiety scale

The Spence children’s anxiety scale (SCAS) is a 44-item self-report anxiety measure for children
and adolescents. It was developed by Spence (1998) to evaluate anxiety symptoms relating to
separation (e.g. ‘I worry about being away from my parents’), social phobia (e.g. ‘I worry what
other people think of me’), obsessive–compulsive disorder (e.g. ‘I have to keep checking that I have
 ARTICLE IN PRESS

330 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

done things right’), fears of injury (e.g. ‘I am scared of going to the doctors or dentist’), panic/
agoraphobia (e.g. ‘I am afraid of being in small closed places, like tunnels or small rooms’) and
generalized anxiety (e.g. ‘I worry about things’). For this study, only the total anxiety score was
used. The scale is scored on a four-point Likert scale of ‘never’, ‘sometimes’, ‘often’ or ‘always’.
Alpha coefficients of between .60 (physical injury) and .82 (panic/agoraphobia) suggest moderate
to good internal consistency, and a test–retest reliability .60 over a 6-month period.

Results

Analysis was conducted using the SPSS version 10.1 program for windows. Data were screened
thoroughly for outliers and normality of distribution based on the recommendations of Norman
and Streiner (2000). No outliers were found, and the population distribution did not deviate
significantly from normality. The results of four participants were removed from the analysis
because of large sections of incomplete data, leaving a final total of 136 adolescents. All
participants agreed to take part in the study. If only a single score was missing from a
questionnaire this was replaced by the mean score of other items on that scale. This was done for
15 participant scores.
In regard to the categorical choice of attachment style in the AQ-C, 78.6% of participants
selected the secure description as most like them. 12.2% selected the ambivalent descriptions, and
9.2% the avoidant description. This pattern of attachment style is broadly similar to previous
studies using the AQ-C (Muris, Mayer, & Meesters, 2000; Muris, Meesters, Merckelbach, &
Hulsenbeck, 2000; Muris, Meesters, van Melick, & Zwambag, 2001). However, they differ from the
original adaptation of the AQ-C by Sharpe et al. (1998), who found that 68% of the sample were
securely attached. Means, standard deviations and Cronbach Alpha’s are presented in Table 1.
Mean, standard deviation and alpha scores for the CDI (see Kovacs, 1992), SCAS (Spence,
1998), ASCS and ASBS (Lang, 1994) are all similar to those found in previous studies.

Multivariate analysis of variance

A multivariate analysis of variance (MANOVA) was conducted using the categorical

attachment choice on the AQ-C as the independent variable. Means and Scheffe post hoc
comparisons for the three attachment styles are displayed in Table 2.
In regard to social rank, securely attached adolescents did not see themselves as particularly
superior or inferior, whereas avoidant adolescents rated themselves as inferior compared to the
secures. As avoidant individuals have been found to have positive models of the self and negative
models of others (Bartholomew, 1990; Bartholomew & Horowitz, 1991), it is interesting that they
(and not ambivalent adolescents) scored significantly lower than secure individuals on our social
comparison measure. Ambivalent adolescents did not differ from secures in terms of social
comparison, but rated themselves as significantly more submissive.
Those participants with a secure attachment style scored significantly lower on the depression
symptoms measure than those who rated themselves as avoidant or ambivalent. Participants with
a secure attachment also scored significantly lower on the anxiety measure in comparison to those
with avoidant attachments.
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 331

Table 1
Means, standard deviations and alpha scores for the attachment, social rank and symptoms measures

Variable Total SD Alpha

Secure 4.96 1.53 n/a

Avoidant 2.32 1.48 n/a
Ambivalent 2.23 1.54 n/a
ASCS 60.24 61.05 .66
ASBS 31.24 31.24 .77
CDI 9.00 9.01 .87
SCAS 27.38 27.38 .84

ASCS: adolescent social comparison scale; ASBS: adolescent submissive behaviour scale; CDI: children’s depression
inventory; SCAS: Spence children’s anxiety scale.

Table 2
MANOVA with attachment style as grouping variable and social rank and symptoms measures as dependent variables

Secure Avoidant Ambivalent Post hoc (Scheffe)

ASCS 61.34 53.92 57.56 Secure4avoidant

ASBS 30.25 34.92 35.19 Secureoambivalent
CDI 33.63 40.17 40.44 Secureoavoidant, ambivalent
SCAS 25.31 38.25 33.19 Secureoavoidant

ASCS: adolescent social comparison scale; ASBS: adolescent submissive behaviour scale; CDI: children’s depression
inventory; SCAS: Spence children’s anxiety scale.

Correlation analysis

As our attachment measure also asked participants to rate how characteristic each attachment
description was of them, we were interested to see how these ratings related to social rank,
depression and anxiety variables. These relationships were assessed using Pearson product
moment correlations (r). Results are presented in Table 3.
There was a significant negative relationship between secure attachment style rating and both
avoidant and ambivalent attachment ratings. Thus, as expected, a characteristically secure
attachment style was associated with lower ratings on the insecure attachment styles. Further,
there was a moderately significant association between the ratings for avoidant and ambivalent
attachment styles.

Attachment and social rank

A significant negative association was found between secure attachment and submissive
behaviour, indicating that the more secure a student feels, the less submissive they rate themselves.
There was a small but significant positive relationship between secure attachment and social
comparison. Both the avoidant and ambivalent attachment rating items were negatively
correlated with social comparison and positively related to submissive behaviour. These
 ARTICLE IN PRESS

332 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

Table 3
Correlation analysis looking at the relationship between measures of attachment, social rank and depression and
anxiety symptoms

SECURE AVOID AMBIV ASCS ASBS CDI

AVOID .27**
AMBIV .41*** .27**
ASCS .17* .35*** .32***
ASBS .24** .23** .39*** .46***
CDI .36*** .34*** .35*** .43*** .32***
SCAS .33*** .30*** .34*** .44*** .56** .62***

SECURE: secure attachment rating; AVIOD: avoidant attachment rating; AMBIV: ambivalent attachment rating;
ASCS: adolescent social comparison scale; ASBS: adolescent submissive behaviour scale; CDI: children’s depression
inventory; SCAS: Spence children’s anxiety scale.

po:05;  po:01 and  po:001:

relationships indicate that whilst heightened secure attachment ratings are associated with a more
positive/adaptive social ranking, insecure attachment ratings (i.e. avoidant and ambivalent) are
significantly associated with inferiority and submissiveness.

Attachment, depression and anxiety

As expected, whilst high scores on the secure attachment rating were significantly correlated
with lower depression and anxiety symptoms, high scores on the avoidant and ambivalent ratings
were significantly related to heightened depression and anxiety symptoms.

Social rank, depression and anxiety

The relationships between the social rank and the depression and anxiety symptoms scores
were, on the whole, more strongly related than that of the correlations between the attachment
style ratings and the symptoms variables. There was a significant negative relationship between
social comparison and depression and anxiety symptoms (i.e. the more inferior the students felt,
the higher the reported depression and anxiety symptoms). There were also significant positive
associations between submissive behaviour with depression and anxiety symptoms. This may
indicate that insecure attachment is associated with a more defensive/low rank interpersonal style
that increases vulnerability to anxiety and depressive symptoms and may well have a negative
impact on how peers relate to the adolescent (Spence et al., 1999).

Regression analysis

To investigate the pattern of association between the attachment, social rank with depression
and anxiety symptoms further, we conducted two standard multiple regression analyses. We
explored whether attachment or social rank variables, when accounting for the influence of each
other, were the stronger predictors of symptom score. Results are presented in Table 4.
In the first analysis, we regressed depression onto the attachment (secure, avoidant and
ambivalent) and social rank (social comparison and submissive behaviour) variables. These
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 333

Table 4
Results of depression and anxiety regressed upon the attachment and social rank variables

Depression Anxiety

B Beta Sig. B Beta Sig.

** *
Secure .98 .22 1.43 .17
Avoidant .66 .14 ns .74 .08 ns
Ambivalent .49 .11 ns .27 .03 ns
** *
Social comparison .19 .28 .25 .19
***
Submissive behaviour .05 .08 ns .71 .40

ns: Non-significant; B: unstandardized beta coefficient; Beta: standardized beta coefficient.


po:05;  po:01;  po:001:

variables produced a significant model in the prediction of depression symptoms score

(F ð5;135Þ ¼ 11:28; po:001). Inspection of the standardized beta weights suggests that, of the
attachment variables, only secure attachment rating was a significant predictor of depression
symptoms when controlling for the influence of the other variables in the analysis. From the social
rank variables, social comparison was the only significant independent predictor of depression.
The same method was used for the second regression analysis with anxiety symptoms as the
dependent variable. The attachment and social rank variables again produced a significant model
in the prediction of anxiety symptoms score (F ð5;135Þ ¼ 17:32; po:001). Inspection of the
standardized beta values shows that whilst secure attachment style was the only attachment
variable which independently contributed to anxiety score, both of the social rank variables were
significant independent predictors of anxiety—in fact, submissive behaviour was a particularly
strong predictor, with the highest standardized beta value.

Moderator/mediator analysis

The regression analysis suggested that aspects of attachment and social rank measures may be
differentially related to depression and anxiety symptoms. As discussed in the Introduction, it
may be the relationship between these systems that best predicts depression and anxiety
symptoms. To explore this further, we tested for potential moderator and mediator effects
between these variables. First, we used z scores of the attachment and social rank variables to
create interaction terms between them, and using regression analysis tested to see if the interaction
between these variables significantly contributed to depression and anxiety scores. However, no
significant incremental effects were found for any of the interaction terms. This means that
attachment and social rank were not operating accumulatively or additively.
We therefore went on to test for a mediational relationship between these variables. To test this
model, a mediation analysis was conducted using multiple regression, following the four-step
analysis recommended by Baron and Kenny (1986) (see Fig. 1).
The first analysis assessed the fit of the model with depression symptoms. Step 1 found that the
predictor variables (i.e. the attachment rating styles) have a significant effect upon the dependent
variable (F ð3;135Þ ¼ 12:93; po:001; R2 ¼ :23). Step 2 found that the predictor variables
 ARTICLE IN PRESS

334 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

Avoidant

Secure Ambivalent
-0.294** 0.124

-0.026 0.326***
-0.251**
-0.075

Social Submissive
Comparison -0.226* 0.138 0.106 Behaviour

-0.290** 0.051

Depression

Fig. 1. Mediation analysis showing mediation effects of social rank variables on the relation between attachment style
and depression.  po:05;  po:01 and  po:001:

significantly impact on both social comparison (F ð3;132Þ ¼ 9:68; po:001; R2 ¼ :18) and submissive
behaviour (F ð3;132Þ ¼ 9:26; po:001; R2 ¼ :17), respectively. Step 3 of the analysis examines
whether the hypothesized mediators are related to the dependent variable, when controlling for
attachment style. A significant model was produced (F ð5;130Þ ¼ 11:56; po:001; R2 ¼ :31).
Analysis of the standardized beta weights shows that social comparison partially mediates the
relationship between ratings of the avoidant and ambivalent attachment styles on depression
scores. However, submissive behaviour does not mediate this relationship. Moreover, secure
attachment still has a significant direct relationship with depression, even when accounting for the
potential mediating effects of the social rank variables.
The same procedure was used for a second mediation analysis, with anxiety symptoms as the
dependent variable. Step 1 of the analysis found that the predictor variables (attachment rating
styles) produced a significant model (F ð3;132Þ ¼ 10:29; po:001; R2 ¼ :19) in the prediction of
anxiety scores. Step 2 showed that the predictor variables significantly impact upon each of the
hypothesized mediators (social comparison and submissive behaviour). The regression analyses
produced significant models for both social comparison (F ð3;132Þ ¼ 9:680; po:001; R2 ¼ :18) and
submissive behaviour (F ð3;132Þ ¼ 9:26; po:001; R2 ¼ :17), respectively. Step 3 of the analysis
examines whether the hypothesized mediators are related to the dependent variable, when
controlling for the predictor variables. Again, a significant model was produced (F ð5;130Þ ¼ 17:55;
po:001; R2 ¼ :40) (Fig. 2).
Analysis of the standardized beta weights indicates that both social comparison and submissive
behaviour fully mediate the relationship between insecure attachment rating and anxiety
symptoms. However, secure attachment rating again has a significant if small, direct relationship
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 335

Avoidant

Secure Ambivalent
-0.294** 0.124

-0.026 0.326***
-0.251** -0.075

Social Submissive
Comparison -0.170* 0.028 Behaviour
0.077

- 0.211** 0.391***

Anxiety

Fig. 2. Mediation analysis showing mediation effects of social rank variables on the relation between attachment style
and anxiety.  po:05;  po:01 and  po:001:

with anxiety symptoms, which was not mediated by the social rank variables. This suggests that
not only does a secure rating style act as a protective factor in relation to anxiety symptoms (i.e. it
has a negative beta weight), but that its relationship is direct and not mediated by the social rank
variables.

Discussion

Early attachment relationships with parents have important effects upon mental health
(Rosenstein & Horowitz, 1996; Cooper, Shaver, & Collins, 1998; Eng, Heimberg, Hart, Schneier,
& Liebowitz, 2001), although the exact mechanisms are not straightforward (Fonagy, 2001). One
possibility is that attachment experiences can synchronize children to their social niches, so that in
safe and supportive environments children experience more prosocial and relaxed social
negotiations for place and belonging. Insecure children, however, may emerge from social
environments where they have become attuned to the power of others to shame, reject or hurt
them and cannot be relied on to be reliable allies or sources of comfort. In consequence, they enter
the world of peers orientated to harsher (or less supportive) social niches, and thus may be more
threat and social rank sensitive, with a tendency to be more defensive. These children may become
overly focused on social comparisons, worry about rejection, and shame, and defend themselves
via avoidant and/or submissive strategies. Another possibility is to control social relationships
through aggression, but this has not been explored here.
 ARTICLE IN PRESS

336 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

The key focus of this study was on how attachment and social rank may interact and their
impact on depression and anxiety symptoms. A key finding was categorical differences in
attachment style were related to variations in the social rank evaluations (social comparison) and
defensive behaviour (submissiveness). Secure individuals tend to see themselves as average or very
slightly above on most traits. They do not judge themselves to be very superior or inferior.
Moreover, secures are the least submissive of the three groups. It may be that for the most part
they feel secure with ‘being one like others’ and part of a cooperative network of alliances/friends.
In other words, they are more attuned to getting on/along with others rather than defending
against threats from others.
Regression analyses found that for depression, secure attachment and social comparison were
the only significant independent predictors of depression scores when the other variables were
controlled for. For anxiety, secure attachment rating was also a significant independent predictor,
but both social comparison and submissive behaviour had a larger independent effect upon
anxiety scores. These findings are intriguing; they suggest that it may be the degree of attachment
security, rather than degree/type of insecurity of attachment, that is the most salient aspect of
attachment relationships. Furthermore, social rank plays an important role in the prediction of
adolescent depression and anxiety even when controlling for the significant impact of attachment
relating. Different facets of social rank (social comparison in depression, submissive behaviour in
anxiety) may be differentially related to psychopathology in adolescence.
The mediation analysis suggested that secure and insecure attachments have different routes or
pathways into depression and anxiety symptoms. Secure attachment ratings were not mediated by
the social rank variables; secure attachment had an independent, negative relationship with
depression and anxiety symptoms. However, insecure attachments were mediated by the social
rank variables, and their influence upon depression and anxiety were through their affect upon
social comparison and submissive behaviour. This fits with the idea that those who are ‘secure’
operate ‘as if’ their social niche is safe, and use more cooperative affiliative strategies. Secure
adolescents may be able to turn to significant others for support and reassurance, if they feel their
status slipping, for boosts. Further, they may also be able to generate self-supportive feelings—
both of which may help to protect them from the psychopathologic effect that low rank/loss of
status may lead to.
The route to depression from the insecures appears rather different—here social rank variables
are important. It is as if insecurity is related to seeing the world as a more competitive place, and
for some feeling inferior and needing to appease. Future research may consider the way different
early attachment experiences are in effect ‘educating’ children’s central nervous systems as to the
kind of niche they are in and thus the types of threats and opportunities to pay attention to and
the type of defences (e.g. anxiety, submissive behaviour) they are going to use. It may not just be
reproductive strategies that get primed (Belsky, 1993), but a host of defensive ones too. Once
turned on, these defensive systems can incline to more anxiety and mood disorders (Gilbert, 2001).

Limitations

The findings from this relatively small sample of 14–16-year-old students (a non-clinical
population) may not be generalizable to clinically depressed or anxious groups of adolescents,
or to earlier/later stages of adolescence itself. This means that the relationship between
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 337

attachment and social rank may operate quite differently in clinical populations. Furthermore,
results were only taken from a single source (the participant), using a single method
(self-report) at a single time (i.e. cross sectional). There are also limitations in the way we
measured attachment style. The AQ-C was the only existing attachment measure for
adolescents which uses Ainsworth’s attachment classification. As such, we have only measured
attachment in a rather rudimentary form, utilizing a single categorical and single Likert
measurement for each attachment style. This measure also fails to pick up on disorganized
adolescents, and it may be that this style of relating is particularly pertinent in understanding
psychopathology (Fonagy, 2001). Further, it may be that attachment should be studied in
terms of two underlying dimensions of the degree of avoidance and anxiety in relationships
(Brennan et al., 1998), rather than attachment based on typological models (Fraley &
Waller, 1998).
This study has shown that the link between attachment relating and depression and anxiety in
adolescents is complex, and likely to be mediated by ones perceptions of status and rank within
the peer group. It would seem future studies could focus on how the relationship between these
systems may heighten vulnerability to psychopathology in this age group.

Appendix A. Adolescent social comparison scale

We would like you to tell us how you feel about yourself compared to your friends. Here is an
example:
Compared to your friends how tall do you think you are?
Smaller 1 2 3 4 5 6 7 8 9 10 Taller
In this example, if I thought I was smaller than my friends, I would circle a number to the left of
the scale. However, if I thought I was taller than my friends, I would circle a number at the right
of the scale.
1. Compared to your friends how shy do you feel?
Less shy 1 2 3 4 5 6 7 8 9 10 More shy
2. Compared to your friends how clever do you think you are?
Less clever 1 2 3 4 5 6 7 8 9 10 More clever
3. Compared to your friends how popular do you think you are?
Less popular 1 2 3 4 5 6 7 8 9 10 More popular
4. Compared to your friends how different do you feel?
Less different 1 2 3 4 5 6 7 8 9 10 More different
5. Compared to your friends how attractive do you think you are?
Less attractive 1 2 3 4 5 6 7 8 9 10 More attractive
6. Compared to your friends how strong do you feel?
Less strong 1 2 3 4 5 6 7 8 9 10 More strong
7. Compared to your friends how accepted do you feel?
Less accepted 1 2 3 4 5 6 7 8 9 10 More accepted
8. Compared to your friends how quiet are you?
Less quiet 1 2 3 4 5 6 7 8 9 10 More quiet
 ARTICLE IN PRESS

338 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

9. Compared to your friends how confident do you feel?

Less confident 1 2 3 4 5 6 7 8 9 10 More confident
10. Compared to your friends how much do you feel left out?
Less left out 1 2 3 4 5 6 7 8 9 10 More left out

Appendix B. Adolescent submissive behaviour scale

Below are a series of statements which describe how people act and feel about certain situations
when they are with people in their own year at school.
When I am with people in my own yeary
1. I agree that I am wrong, even when I know that I was not wrong
Never 1 2 3 4 5 Always
2. I do things because others are doing them, rather than because I want to
Never 1 2 3 4 5 Always
3. I let others criticize me or put me down without defending myself
Never 1 2 3 4 5 Always
4. I play with others even if I do not want to
Never 1 2 3 4 5 Always
5. If I try to speak and others take over, I just shut up
Never 1 2 3 4 5 Always
6. When I make a little mistake and want to apologize, I say sorry more than once
Never 1 2 3 4 5 Always
7. I stop myself from telling others when I am angry with them
Never 1 2 3 4 5 Always
8. At parties, I let others talk a lot and dominate the conversation
Never 1 2 3 4 5 Always
9. I feel uncomfortable when people look straight at me when they are talking
Never 1 2 3 4 5 Always
10. I say thank you over and over again when someone does a small favour for me
Never 1 2 3 4 5 Always
11. I avoid starting conversations at parties
Never 1 2 3 4 5 Always
12. I blush (go red) when people look at me
Never 1 2 3 4 5 Always

References

Ainsworth, M. D. S., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment. Hillsdale, NJ: Erlbaum.
Allan, S., & Gilbert, P. (1995). A social comparison scale: psychometric properties and relationship to
psychopathology. Personality and Individual Differences, 19, 293–299.
Allan, S., & Gilbert, P. (1997). Submissive behaviour and psychotherapy. British Journal of Clinical Psychology, 36,
467–488.
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 339

Allan, S., Gilbert, P., & Goss, K. (1994). An exploration of shame measures—II: psychopathology. Personality and
Individual Differences, 17, 719–722.
Angold, A., & Rutter, M. (1992). Affects of age and pubertal status on depression in a large clinical sample.
Development and Psycholopathology, 4, 5–28.
Angold, A., & Worthman, C. W. (1993). Puberty onset of gender differences in rates of depression: a developmental,
epidemiologic and neuroendocrine perspective. Journal of Affective Disorders, 29, 145–158.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator distinction in social psychology. Journal of Personality
and Applied Social Psychology, 51, 1173–1182.
Bartholomew, K. (1990). Avoidance of intimacy: an attachment perspective. Journal of Social and Personal
Relationships, 7, 147–178.
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: a test of a four-category model.
Journal of Personality and Social Psychology, 61, 226–244.
Baumeister, R., & Leary, M. (1995). The need to belong: desire for interpersonal attachments as a fundamental human
motivation. Psychological Bulletin, 117, 497–529.
Belsky, J. (1993). Etiology of child maltreatment: a developmental analysis. Psychological Bulletin, 114, 413–434.
Belsky, J., Steinberg, L., & Draper, P. (1991). Childhood experiences, interpersonal development, and reproductive
strategy: an evolutionary theory of socialization. Child Development, 62, 647–670.
Berndt, D. J., Schwartz, S., & Kaiser, C. F. (1983). Readability of clinical self-report depression inventories. Journal of
Consulting and Clinical Psychology, 51, 627–628.
Blain, M. D., Thompson, J. M., & Whiffen, V. E. (1993). Attachment and perceived social support in late adolescence:
the interaction between working models of self and others. Journal of Adolescent Research, 8, 226–241.
Bowlby, J. (1969). Attachment and loss: attachment, Vol. 1. London: Hogarth Press.
Bowlby, J. (1973). Attachment and loss: separation, Vol. 2: anxiety and anger. London: Hogarth Press.
Bowlby, J. (1980). Attachment and loss: loss, Vol. 3: sadness and depression. London: Hogarth Press.
Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment; an integrative
overview. In Simpson, J. A., & Rholes, W. F. (Eds.), Attachment theory and close relationships (pp. 46–76). New
York: The Guilford Press.
Buhrmester, D. (1996). Need fulfillment, interpersonal competence, and the developmental contexts of early adolescent
friendship. In Bukowski, W., Newcomb, A., & Hartup, W. (Eds.), The company they keep (pp. 158–185). New York:
Cambridge University Press.
Campus, B. E. (1987). Stress and life events during childhood and adolescence. Clinical Psychology Review, 7, 275–302.
Collins, N. L., & Feeney, B. C. (2000). A safe haven: an attachment theory perspective on support seeking and care
giving in intimate relationships. Journal of Personality and Social Psychology, 78, 1053–1073.
Collins, N. L., & Read, S. J. (1990). Adult adulthood, working models, and relationship quality in dating couples.
Journal of Personality and Social Psychology, 58, 644–663.
Cooper, M. L., Shaver, P. R., & Collins, N. L. (1998). Attachment styles, emotion regulation and adjustment in
adolescence. Journal of Personality and Social Psychology, 74, 1380–1397.
Eisenberg, N. (2002). Empathy-related emotional responses, altruism, and their socialization. In Davidson, R., &
Harrington, A. (Eds.), Visions of compassion: western scientists and Tibetan Buddhists examine human nature
(pp. 131–164). New York: Oxford University Press.
Eng, W., Heimberg, R. G., Hart, T. A., Schneier, F. R., & Liebowitz, M. R. (2001). Attachment in individuals with
social anxiety disorder: the relationship among adult attachment styles, social anxiety, and depression. Emotion, 1,
365–380.
Fleming, I. E., Offord, D. R., & Boyle, M. H. (1989). Prevalence of childhood and adolescent depression in the
community: Ontario child health study. British Journal of Psychiatry, 155, 647–654.
Fonagy, P. (2001). Attachment theory and psychoanalysis. London: Brunner-Routledge.
Fraley, R. C., & Waller, N. G. (1998). Adult attachment patterns: a test of the typological model. In Simpson, J. A., &
Rholes, W. S. (Eds.), Attachment theory and close relationships. New York: The Guilford Press.
Gilbert, P. (1989). Human nature and suffering. London: Lawrence Erlbaum Associates.
Gilbert, P. (1993). Defence and safety: their function in social behaviour and psychopathology. British Journal of
Clinical Psychology, 32, 131–153.
 ARTICLE IN PRESS

340 C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341

Gilbert, P. (2000). The relationship of shame, social anxiety and depression: the role of evaluation of social rank.
Clinical Psychology and Psychotherapy, 7, 174–189.
Gilbert, P. (2001). Evolutionary approaches to psychopathology: the role of natural defences. Australian and New
Zealand Journal of Psychiatry, 35, 17–27.
Gilbert, P. (2002). Evolution theory and cognitive therapy. Journal of Cognitive Psychotherapy: An International
Quarterly, 16, 259–261.
Gilbert, P., & Allan, S. (1994). Assertiveness, submissive behaviour and social comparison. British Journal of Clinical
Psychology, 33, 295–306.
Hazen, C., & Shaver, P. (1987). Romantic love conceptualised as an attachment process. Journal of Personality and
Social Psychology, 52, 511–524.
Hewitt, P. L., Caelian, C. F., Flett, G. L., Sherry, S. B., Collins, L., & Flynn, C. A. (2002). Perfectionism in children:
associations with depression, anxiety and anger. Personality and Individual Differences, 32, 1049–1061.
Kazdin, A. E., & Petti, T. A. (1982). Self-report and interview measures of childhood and adolescent depression.
Journal of Child Psychology and Psychiatry, 23, 437–457.
Kashani, J. H., & Orvaschel, H. (1990). A community study of anxiety in children and adolescents. American Journal of
Psychiatry, 147, 313–318.
Kovacs, M. (1992). The children’s depression inventory (C.D.I.) manual. New York: Multi-Health Systems Inc.
Laible, D. J., Carlo, G., & Raffaelli, M. (2000). The differential relations of parent and peer attachment to adolescent
adjustment. Journal of Youth and Adolescence, 29, 45–59.
Lang, J. (1994). Adolescence and shame. Unpublished M.Sc. (Clinical Psychology), University of Leicester.
Larson, R., & Ham, M. (1993). Stress and ‘‘storm and stress’’ in early adolescence: the relationship of negative events
with dysphoric affect. Developmental Psychology, 29, 130–140.
Levy, M. B., & Davis, K. E. (1988). Lovestyles and attachment styles compared: their relation to each other and to
various relationship characteristics. Journal of Social and Personal Relationships, 5, 439–471.
Mikulincer, M., Birnbaum, G., Woddis, D., & Nachmias, O. (2000). Stress and accessibility of proximity-related
thoughts: exploring the normative and intraindividual components of attachment theory. Journal of Personality and
Social Psychology, 78, 509–523.
Muris, P., Mayer, B., & Meesters, C. (2000). Self-reported attachment style, anxiety and depression in children. Social
Behaviour and Personality, 28, 157–162.
Muris, P., Meesters, C., Merckelbach, H., & Hulsenbeck, P. (2000). Worry in children is related to perceived parental
rearing and attachment. Behaviour Research and Therapy, 38, 487–497.
Muris, P., Meesters, C., van Melick, M., & Zwambag, L. (2001). Self-reported attachment style, attachment quality and
symptoms of anxiety and depression in young adolescents. Personality and Individual Differences, 30, 809–818.
Norman, G. R., & Streiner, D. L. (2000). Biostatistics: the bare essentials. BC, Canada: Decker Inc.
Perry, B. D., Pollard, R. A., Blakley, T. L., Baker, W. L., & Vigilante, D. (1995). Childhood trauma, the neurobiology
of adaptation and ‘‘use-dependent’’ development of the brain: how ‘‘states’’ become ‘‘traits’’. Infant Mental Health
Journal, 16, 271–291.
Prinstein, M. J., Boergers, J., Spirito, A., Little, T. D., & Grapentine, W. L. (2000). Peer functioning, family
dysfunction, and psychological symptoms in a risk factor model for adolescent inpatients’ suicidal ideation severity.
Journal of Clinical Child Psychology, 29, 392–405.
Rosenstein, D. S., & Horowitz, H. A. (1996). Adolescent attachment and psychopathology. Journal of Consulting and
Clinical Psychology, 64, 244–253.
Sharpe, T. M., Killen, J. D., Bryson, S. W., Shisslak, C. M., Estes, L. S., Gray, N., Crago, M., & Taylor, C. B. (1998).
Attachment style and weight concerns in preadolescent and adolescent girls. International Journal of Eating
Disorders, 23, 39–44.
Shaver, P. R., & Breenan, K. A. (1992). Attachment styles and the ‘big five’ personality traits: their connections
with each other and with romantic relationship outcomes. Personality and Social Psychology Bulletin, 18,
536–545.
Sloman, L. (2000). How involuntary defeat is related to depression. In Sloman, L., & Gilbert, P. (Eds.), Subordination
and defeat: an evolutionary approach to mood disorders and their therapy (pp. 47–66). Mahwah: Lawrence Erlbaum
Associates.
 ARTICLE IN PRESS

C. Irons, P. Gilbert / Journal of Adolescence 28 (2005) 325–341 341

Sloman, L., Gilbert, P., & Hasey, G. (2003). Evolved mechanisms in depression: the role and interaction of attachment
and social rank in depression. Journal of Affective Disorders, 74, 107–121.
Smith, P. K., & Myron-Wilson, R. (1998). Parenting and school bullying. Clinical Child Psychology and Psychiatry, 3,
405–417.
Spence, S. H., Donovan, C., & Brechman-Toussaint, M. (1999). Social skills, social outcomes, and cognitive features of
childhood social phobia. Journal of Abnormal Psychology, 108, 211–221.
Sorenson, S. B., Rutter, C. M., & Aneshensel, C. S. (1991). Depression in the community: an investigation into age of
onset. Journal of Consulting and Clinical Psychology, 59, 541–546.
Spence, S. H. (1998). A measure of anxiety symptoms among children. Behaviour Research and Therapy, 36, 545–566.
Troy, M., & Sroufe, L. A. (1987). Victimisation among preschoolers: role of attachment relationship history. Journal of
the American Academy of Child and Adolescent Psychology, 26, 166–172.
Van Lange, P. A. M., De Bruin, E. M. N., Otten, W., & Joireman, J. A. (1997). Development of prosocial,
individualistic and competitive orientations: theory and preliminary evidence. Journal of Personality and Social
Psychology, 73, 733–746.