ARTICLE IN PRESS
Social Science & Medicine 64 (2007) 765–775
Is attachment style a source of resilience against health
inequalities at work?
Mel Bartley , Jenny Head, Stephen Stansfeld
University College London, London, UK
Available online 28 November 2006
The argument that ‘indirect selection’ is a contributory factor to health inequality has included ideas about personal
characteristics that may originate in childhood and increase the likelihood of both poor health and disadvantaged social
position in adulthood. The concept of protective resilience makes a similar but converse argument: that positive
characteristics acquired at one phase of life may enable individuals to withstand later adversity. The increasing richness of
data from longitudinal studies now allows us to examine these processes more closely over a longer period of life. In this
paper we show that attachment style, a psychological characteristic thought to be associated with the style of parenting
encountered during early childhood, may act as a source of resilience in the face of educational disadvantage. Men in midlife
who were not burdened with anxious or avoidant attachment styles seem to have been more likely to overcome the
disadvantage of a lower level of educational attainment and progress up the ladder of Civil Service grades in the English
Whitehall II study. As it is not strongly related to parents’ social class, it can be argued that attachment style has acted as a
source of upward social mobility which is also likely to reinforce better health in later life.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Social inequalities in health; Attachment style; Resilience; Social mobility; Whitehall II study; UK
Recent changes in social policy towards young
children, such as the setting up in Great Britain of
the ‘Sure Start’ schemes to support parents of young
children in deprived areas, have reflected a growing
acceptance of the importance of life-course processes.
Evidence has accumulated that experiences
in early life may, in the words of the Black Report,
‘‘cast long shadows forward’’ onto the health and
Corresponding author. Tel.: +44 207 679 1707;
fax: +44 207 813 0280.
E-mail addresses: email@example.com (M. Bartley),
firstname.lastname@example.org (J. Head), S.A.Stansfeld@qmul.ac.uk
well-being of adults. Theories regarding the influence
of childhood conditions and experiences for
later health have included ideas about the possible
joint effects of parental care on later personality and
adult social position (Glendinning, Shucksmith, &
Hendry, 1997; Haavet, Saugstad, & Straand, 2005;
Stewart-Brown, Fletcher, & Wadsworth, 2005).
This idea has been termed ‘indirect health selection’,
that is, a process through which early life conditions
may constitute a ‘confounder’ of the relationship
between adult social position and health. In this
model, early experiences, through their effect on
personal characteristics such as coping styles or
locus of control might strongly influence both
educational and career success, and thereby social
0277-9536/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775
position, as well as health behaviours in adulthood,
and thereby health (Berglund et al., 1996; Mackenbach,
2005; van de Mheen, 1998). The apparent
link between social position and health, the ‘health
gradient’ in adults, would, in this perspective, be
something that had its origins far earlier.
Sweeting and West proposed a mechanism of this
kind by raising the question of ‘‘a role for family
culture in the health inequalities debate’’ (Sweeting
& West, 1995). This paper was based on data from
the West of Scotland 20-07 study’s youngest cohort,
aged 15 in 1997 (West, Macintyre, Annandale, &
Hunt, 1990). It argued that ‘‘the role of the family in
relation to health inequalities has been largely
ignored’’(Sweeting & West, 1995, p. 171). Their
measures of family-centredness and family conflict
when cohort members were aged 15 were found to
be significantly related to measures of physical and
psychological health, self esteem, and to the likelihood
that, by the age of 18, a young person would
be in further or higher education rather than
unemployed or in a government training scheme
or unskilled employment. Equally important was
their finding that material deprivation in the
household was of lesser importance than these
aspects of family relationships for either self esteem
or health as measured at age 15 and 18. In this way,
the study points towards a lifecourse process in
which relationships in childhood may, independently
of material factors, influence both the
development of personal characteristics such as self
esteem, and the attainment of social position in
adulthood. Although a great deal more work in
social epidemiology since that time has been
concerned with the life course, family function and
its consequences for personal characteristics in
adulthood might still be regarded as somewhat
under-investigated. Another Scottish study produced
interesting findings in this respect during the
1990s (Glendinning, Hendry, & Shucksmith, 1995;
Glendinning, Love, Hendry, & Shucksmith, 1992;
Glendinning et al., 1997). More recently, a Norwegian
study found a strong effect of aspects of family
culture on health in adolescents (Haavet et al.,
2005), and several studies have now addressed the
possibility that psychological characteristics emerging
during childhood may play a role in adult
health, and thereby in health inequality (Balkrishnan,
1998; Stewart-Brown et al., 2005; Whalley &
Deary, 2001). However most of these studies regard
‘adverse’ personal characteristics such as hostility or
neuroticism as part of an aetiological pathway in
which socio-economic disadvantage in the family of
origin is associated with both social and health
disadvantage during adult life. Few studies ask
whether there might be characteristics arising from
family relationships, independently of material
circumstances, that may act as sources of resilience
in that they increase the likelihood of both a more
advantaged socioeconomic position and better
health in adulthood.
One psychological theory relevant to the processes
at work behind observed associations between
family relationships in childhood and later
life health and well-being that has received more
attention in the years since Sweeting and West’s
study is known as attachment theory (Ainsworth,
Blehar, Waters, & Wall, 1978; Fonagy, 2001; Gil &
Rupprecht, 2003). These ideas have their foundation
in the work of Bowlby, a psychoanalyst who
studied the consequences of children having been
separated from their parents, for example during
spells in hospital and during the World War II
evacuations (Bowlby, 1982; Main, 1996). One
consequence of his work, as is well known, was a
complete change of policy regarding parental
visiting rights in the children’s wards of hospitals.
Before Bowlby’s work, it had been thought that
parental visits caused distress to children, and that it
was better for young patients to become resigned to
their separation from their families. Bowlby’s
research showed, however, that the passive resignation
often seen in children during spells in hospital
(which, in that period of epidemiological history,
the 1940s and 1950s, were often prolonged), was
more likely to be a sign of serious psychological
damage than of children’s philosophical acceptance
of the situation.
More generally, Bowlby regarded family functioning
as having great importance for the eventual
adult, as did other psychoanalysts. He was the
founder of the idea that a developing person needs a
‘secure base’ from which to venture forth into the
wider world (Waters, Crowell, Elliott, Corcoran, &
Treboux, 2002). The concept of attachment was
later explored by Ainsworth and her colleagues in
the form of the ‘strange situation’ test. In these
experiments, very young children were briefly
separated from their caretakers (usually their
mothers), and their behaviour on being re-united
was carefully observed. Behaviour seemed to fall
into three major types. Children who were securely
attached ran back happily into the caretaker’s arms.
Other children showed patterns of behaviour in
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775 767
which some would resist being picked up and
cuddled at first, or avoid looking at the caretaker,
appear totally indifferent, or show in quick succession
an apparently disorganised combination of
these three types of reaction. In longitudinal
research, these different patterns of attachment
behaviour have been found to be associated with
later adult behaviour, including ways of relating to
children of one’s own (Green & Goldwyn, 2002;
Hesse & Main, 1999; Main, 1996; Massie &
Research on patterns of infant attachment and
their adult sequelae form just one part of the field of
developmental psychopathology, which has been
described as ‘‘the consideration of the abnormal and
healthy together, with particular attention to risk
and resilience factors influencing the course of
development’’ (Main, 1996, p. 241). This research
programme fits into the emerging field of life-course
social epidemiology in a number of important ways
(Hertzman & Power, 2003; Tremblay, 1999). The
most significant of these, for the purposes of the
present analysis, is the way in which developmental
psychopathology (despite its name) considers ‘resilience’
as well as ‘risk’ (Main, 1996; Fergus &
Zimmerman, 2005; Schoon, 2006; Werner, 2004). In
this way it differs from much life-course epidemiology
which has tended to focus on accumulating
patterns of risk (Bartley & Plewis, 2002; Brunner,
Shipley, Blane, Smith, & Marmot, 1999; Galobardes,
Lynch, & Davey Smith, 2004; Power &
Matthews, 1997; Singh-Manoux, Ferrie, Chandola,
& Marmot, 2004).
Theoretically it is expected that styles of attachment
develop from early relationships with mother
and father and are maintained into adult life.
Attachment styles developed in childhood have
been shown to be associated in adults with a range
of outcomes (Droomers, Schrijvers, Casswell, &
Mackenbach, 2003; Scharf, Mayseles, & Kivenson-
Baron, 2004; Schmidt, Nachtigall, Wuethrich Martone,
& Strauss, 2002; Waller, Scheidt, & Hartmann,
2004). In general, this research has
concentrated on relationships between attachment
styles and adult mental health (Bifulco, Moran,
Ball, & Bernazzini, 2002; Irons & Gilbert, 2005) or
relationships (Bifulco, Moran, Ball, & Lillie, 2002),
for example with spouses, children (Priel & Besser,
2001) and caretakers of those who are ill ( Maunder
et al., 2006; Waller et al., 2004). Secure attachment
is seen in this perspective as a possible ‘resilience
factor’, emerging early in life, which may protect
individual wellbeing in the face of risk and
adversity, because it is regarded as reflecting the
ability to effectively regulate and mitigate the
strength of emotional responses to adverse personal
or health events.
The inclusion of a short validated questionnaire
on attachment style in the Whitehall II study gives
data on a relatively large sample of the adult
working population, which also includes extensive
information on education, health, and career
attainment within the civil service over a 20-year
period. Because attention in the developmental
literature has been on psychological risk, less
attention has been paid to the possibility that
attachment style might function as a source of
resilience that would come into play in the face of
social adversities encountered during schooling, or
as part of the transition to adulthood or the work
career. Sweeting and West do not set their study
within the attachment or resilience perspectives. But
they show it is plausible that positive family
function might have been associated with more
favourable later outcomes independently of socioeconomic
circumstances, in part, by conferring
resilience to less materially advantaged children.
According to the literature on resilience, a
protective resilience factor is one whose importance
is greatest in the face of adversity. If attachment
style is acting as a source of resilience, it will not
necessarily be related to Civil Service grade in any
simple fashion, but rather will have greater importance
for career mobility in the Civil Service for
those who entered the service with some degree of
relative educational disadvantage. The reasoning
behind the hypothesis is that for those following a
‘straight road’ through life with an accumulation of
advantages, the aspects of psychological functioning
measured by the attachment scale may not be
significant factors for career success. For those with
a less privileged life course, however, advancement
in a complex organisation may be more dependent
on the kind of personal and social skills thought to
be prominent in those who are ‘securely attached’.
These have been enumerated to include being
comfortable exploring new situations, having low
levels of hostility, and being able to ask for and
make use of help from others (Shaver & Mikulincer,
2002), and ‘‘able to maximize the opportunities
presented to them by the environment’’ (Fonagy,
2005). In this paper, we examine whether attachment
style in adults is more importantly associated
with occupational attainment in adults with less
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775
advantaged earlier life histories, as reflected in their
Data and methods
The Whitehall II Study was set up in England to
investigate the degree and causes of the social
gradient in morbidity and mortality http://www.ucl.ac.uk/whitehallII/.
A cohort of civil servants was
established between 1985 and 1988 (Phase 1). All
male and female civil servants aged between 35 and
55 years, in 20 London-based civil service departments,
were sent an introductory letter and a
screening questionnaire and attended a screening
examination. The overall response rate was 73%
(74% for men and 71% for women). The true
response rates are likely to be higher, however,
because around 4% of those on the list of employees
had moved before the study and were not eligible
for inclusion. Altogether 10,308 civil servants were
examined—6,895 men and 3413 women. After the
initial participation of Phase 1 there have been an
additional seven phases of data collection. This
paper deals with data collected at Phase 1 which
took place during 1985–1987 (education) and Phase
5 which took place between 1997 and 1999
(attachment and Civil Service grade) when study
participants were aged 47–67.
Three styles of attachment are usually described:
secure, anxious/ambivalent (also called preoccupied
or fearful) and avoidant (also called dismissing).
There is, however, evidence of important differences
between avoidant and dismissing styles: Bartholomew
(1990) found a distinction between dismissing
individuals who avoid attachment because they do
not value it and fearful individuals who long for
attachment but avoid intimacy for fear of rejection.
For inclusion in the Whitehall II study, a reliable
self-report measure of attachment style that was
feasible for use in large epidemiological studies and
had some track record of being used in large-scale
studies was needed. An adapted version of the
Hazan and Shaver attachment style measure,
distinguishing four attachment styles, was therefore
included at Phase 5 of the study in 1997–1999
(Hazan & Shaver, 1987). This measure, drawing on
the concept of individual differences in infant
attachment patterns, comprises four groups of
statements measuring secure attachment style,
anxious/ambivalent attachment style, avoidant attachment
style, and dismissing attachment style.
The statements are:
It is easy for me to become emotionally close to
others. I am comfortable depending on them and
having them depend on me. I don’t worry about
being alone or having others not accept me
I am uncomfortable getting close to others. I
want emotionally close relationships, but I find it
difficult to trust others completely, or to depend
on them. I worry that I will be hurt if I allow
myself to become too close to others [avoidant]
I want to be completely emotionally intimate
with others, but I often find others are reluctant
to get as close as I would like. I am uncomfortable
being without close relationships, but I
sometimes worry that others don’t value me as
much as I value them. [anxious/ambivalent]
I am comfortable without close emotional
relationships. It is very important to me to feel
independent and self-sufficient, and I prefer not
to depend on others or have others depend on me
For each group of statements, participants were
asked to score the statements between 0 and 100
where 100 meant that ‘‘the statement describes me
exactly’’ and 0 describes ‘‘the complete opposite to
me’’. Attachment style is conceptualised to have
quite a high degree of stability. In earlier studies
70% of individuals identified the same attachment
style across a 4-year interval (Hazan & Shaver,
1987). In another study using the same adapted
scale, the kappa was 0.48 for agreement with the
Revised Adult Attachment Scale (Stein et al, 2002).
In terms of validity the attachment groups were
related in a predicted fashion with marital status:
the proportion with a secure attachment style was
lower in single men and women compared to
married, divorced, separated and widowed groups.
Anxious attachment was more frequent in single
and divorced men and women, while avoidant
attachment was more frequent in divorced, separated
men and women. Dismissing style was more
frequent in single men and single, divorced separated
and widowed women (Stansfeld, Head,
Bartley, & Fonagy, 2006).
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775 769
Education was included in the Phase 1 questionnaire
when participants were aged 35–55. The
variable used here combines two questions: on the
age at which the participant finished full time
education and on the last type of educational
establishment they attended. This was coded 1
(lowest) for those who left full time education at
age 18 or before, 2 for those who stayed in
education beyond age 18 but did not acquire a
degree, and 3 (highest) for those who left full time
education later than age 18 having gained a degree,
or a nursing or teaching qualification.
Parental social class
Social class of the parents was classified according
to the Registrar-General’s schema (RGSC). This
classification was used in British official statistics up
to the time of the Census of 2001. It was based in
part on prestige (‘general standing in the community’)
and in part on the approximate skill level
thought to be required for the occupations in each
class. Because social class defined in this way is
known to be strongly associated with income, in this
analysis it is included as a measure of the material
living standards experienced by study participants
during their childhood.
Grade of employment
The Civil Service identifies 12 non-industrial
grades on the basis of salary and the type of work
carried out. To investigate the distribution of
attachment style by grade we used a broader
categorisation of employment grade: administrative,
Four thousand eight hundred and thirty-six men
and 2013 women participants gave usable data in
response to the attachment questionnaire. There
were no missing data on Civil Service grade in these
participants; data on parental social class were
missing in1484 men and 695 women, and data on
educational level were missing in 1223 men and 540
Table 1 shows the distribution of attachment style
according to gender, social origins (paternal social
class), present social position at Phase 5 of the
Whitehall II study, measured as Civil Service
employment grade, and educational level in this
sample of employed adults. There is perhaps
surprisingly little difference between men and
women: a slightly higher prevalence of secure or
dismissive styles in men and slightly more anxious
attachment in women. The relationship between
attachment styles and parental social class was
examined, in order to test for a relationship between
material circumstances in childhood and adult
attachment, but this relationship was found to be
weak and non-significant. The relationship between
attachment style and current Civil Service grade was
also examined, and was found to be weak in both
men and women. Childhood material circumstances
in themselves did not therefore appear to be
strongly related to attachment style in middle age
in this cohort, and are not further considered.
Although the relationship between parental social
class and attachment was weak parental social class
was strongly related to educational attainment in
Whitehall II study participants, as it is in all other
studies. Material circumstances in childhood are
therefore seen to have far more importance for
educational success than for adult attachment style,
and any relationship of attachment style to later
career attainment will not act directly through this
pathway (Table 2).
Table 3 shows the relationship of educational
attainment to Civil Service grade attained by Phase
5 of the study. It can be seen that taking participants
as a whole, it was unusual for a man (17.6%) and
even more unusual for a woman (0.5%) to reach the
higher grades of the Civil Service. The very small
numbers of women in the highest employment
grades makes it impossible to investigate the role
of attachment style as a source of resilience in the
work career any further.
Table 4 addresses the question of whether
attachment style may act as a source of ‘protective
resilience’ in the work career of male participants in
the Whitehall II study. The small number of
women, especially in the higher grades, means that
there is not sufficient power to replicate these
models in women. The table shows four separate
logistic regression models. In Model 1, the relationship
of education to the likelihood of being in the
higher Civil Service grades is shown, adjusted
only for age. Model 2 shows that the relationship
of attachment style to Civil Service grade in
men is no longer significant after adjustment for
age. These models replicate Tables 2 and 3, but have
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775
Attachment style by social origins (father’s social class), education and gender
Secure (%) Dismissive (%) Avoidant (%) Anxious (%) N
Father’s social class
Men 45.8 33.4 8.0 12.9 1394
Women 42.8 34.1 7.7 15.4 505
Men 47.3 35.0 6.5 11.1 1620
Women 43.5 34.0 7.3 15.1 641
Semi or unskilled
Men 45.9 37.9 5.3 10.9 338
Women 48.8 29.7 7.6 14.0 172
Men, women n.s.
Men 49.7 33.4 5.3 11.6 958
Women 47.5 32.8 6.7 13.0 640
Men 47.5 33.8 7.5 11.3 924
Women 42.7 36.6 7.3 13.4 358
Men 44.3 35.3 7.5 12.9 1731
Women 41.1 31.2 8.0 19.8 475
Men p ¼ 0.08, women p ¼ 0.02
Civil service grade
Men 47.8 34.5 6.6 11.1 2017
Women 42.5 35.7 7.5 14.3 294
Men 46.1 33.5 7.3 13.1 2520
Women 44.4 32.3 6.8 16.4 900
Men 40.1 37.8 9.4 12.7 299
Women 45.5 33.0 6.8 14.7 819
Men p ¼ 0.06, women p ¼ 0.86
All with valid attachment scores
Men 56 34 7 12 4836
Women 45 33 7 15 2013
Men, women p ¼ 0.006
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775 771
Relationship between paternal social class and educational
Low Inter-mediate High N
Paternal social class
Professional or management
Men 14.1 23.9 62.0 1393
Women 21.4 22.8 55.8 504
Skilled intermediate non manual or manual
Men 33.4 27.0 39.6 1618
Women 53.6 25.3 21.1 640
Semi or unskilled manual
Men 45.6 21.9 32.5 338
Women 66.9 21.5 11.6 172
Relationship in men and women significant po0.001.
Relationship between educational attainment level and Civil
Civil service grade (grouped)
High Middle Low
Men 19.6 65.6 14.8 958
Women 0.6 34.1 65.3 640
Men 30.6 64.7 4.7 924
Women 7.5 57.0 35.5 358
Men 55.1 41.2 3.8 1731
Women 29.7 50.5 19.8 475
Relationships significant in men and women po0.001.
the advantage that any confounding effect of age is
taken into account. In the third model, education
and attachment style are entered together, giving
little improvement to the fit of the model over
Model 1, which included age and education only.
This means that considered by itself, with or
without adjustment for education, attachment style
does not predict membership of a high civil service
grade. Model 3 also shows that adjustment for
attachment style can be seen to make very little
difference to the relationship between education and
career success (in fact the relationship becomes
Model 4 also contains education and attachment
style, but here an additional term is added for the
interaction between attachment style and education.
This allows us to see whether the relationship of
education and career success is significantly different
in those with different attachment styles. That is,
model 4 asks whether attachment is a source of
‘protective resilience’, which plays a more important
role in relation to career success in the presence of
lower levels of educational attainment. The addition
of the interaction term results in a significant
improvement to the model (improvement in
w 2 ¼ 18:25, 6 df, p ¼ 0:006), showing that we more
accurately predict career success if we allow education
to have a different relationship to Civil Service
grade in those with different attachment styles. In
men with the highest levels of education, the
probability of membership of higher civil services
grades is high irrespective of attachment style.
However, in those with only basic education,
securely attached men are well over twice as likely
(O.R. 2.34, 95% C.I. 1.20–4.55) to be in higher civil
service grades than those (the baseline group) with
both low educational level and anxious attachment
style. The effect of attachment style is also greater in
those with intermediate levels of education than in
those with the highest level.
Attachment style is beginning to be more widely
discussed as a factor in adult social and psychological
adjustment, and in coping with illness. However
less attention has been paid to its possible effect
as a source of resilience in the life-course of
the individual. One reason for this may be that
there are few studies which have included measures
of attachment style in a large number of adults
(Belsky, 2002; Mickelson, Kessler, & Shaver, 1997).
The inclusion of such measures in the Whitehall II
study now makes it possible to look more closely at
relationships between attachment style and adult
social and occupational, as well as psychological
functioning. Encouragingly, the distribution of
attachment styles is similar in this cohort to what
has been found in other studies (Mickelson et al.,
1997) with around 45% of all respondents having a
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775
Relationship of educational attainment to high Civil Service grade in men with different attachment styles
Odds of high civil service grade
Model 1 a Model 3 c Model 4 d
Anxious Avoidant Dismissive Secure
Basic 1 1 1 (baseline) 1.41 (0.52–3.83) 1.47 (0.73–2.96) 2.34 (1.20–4.55)
Intermediate 2.08 (1.67–2.59) 2.09 (1.70–2.60) 2.53 (1.14–5.62) 1.78 (0.73–4.36) 3.98 (2.02–7.83) 4.38 (2.26–8.50)
Higher 6.75 (5.53–8.24) 6.82 (5.59–8.32) 13.15 (6.61–26.18) 13.24 (6.41–27.36) 12.32 (6.41––23.67) 11.88 (6.21–22.72)
Model 2 b
Secure 1 1
Dismissive 0.95 (0.84–1.08) 0.90 (0.77–1.06)
Avoidant 0.87 (0.68–1.10) 0.79 (0.59–1.06)
Anxious 0.90 (0.75–1.09) 0.81 (0.64–1.03)
a Model 1 adjusted for age only, w 2 improvement over model with age alone ¼ 469.83, 2 df, po0:001.
b Model 2 adjusted for age only, w 2 improvement over model with age alone ¼ 2.26, 3 df, p40:1.
c Model 3 includes age, education and attachment style. w 2 ¼ 471.18. Improvement over Model 1 (age+education) w 2 ¼ 1.35, 4 df,
d Model 4 includes age, attachment style, education and interaction education attachment. Model w 2 ¼ 597.38. Improvement over
Model 3 (main effects only) w 2 ¼ 18:25, 6 df, p ¼ 0:006.
secure attachment style, 12% anxious, 7% avoidant
and 34% dismissive.
Reflecting other work (Mickelson et al., 1997),
the measures of socio-economic status used in this
paper were not closely related, in themselves, to
attachment style. Previous work has shown that
there was no significant relationship to parental
social class in this sample (Stansfeld et al., 2006).
Here we have shown that attachment style is not
related to present socio-economic position as
measured by Civil Service grade in women and a
borderline relationship in men. Rather, the data
indicate that attachment style appears to be
associated with greater occupational attainment in
those with a less favourable educational history. A
factor which is not associated with a favourable
outcome overall, but acts specifically among those
with some kind of adversity, is known as a source of
‘protective resilience’ (Schoon, 2006).
Bernier and Dozier have concluded that ‘‘There is
no doubt that what is tapped by self-reports of
‘attachment style’ is a very significant component of
adult personality that plays a major role not only in
close relationships but in a wide range of spheres of
functioning.’’ (Bernier & Dozier, 2002). We have
examined the possibility that attachment style may
have, in this cohort of civil servants, acted as a
moderator to the disadvantages of lower levels of
educational attainment for career progression. It is
striking that, in study participants with the highest
educational level, there appeared to be no benefit at
all from a secure attachment style. In the large
group with only intermediate educational attainment,
as in the smaller numbers with low attainment
however, attachment style significantly affected the
likelihood of civil service career success. In those
with degrees and other tertiary qualifications, there
was no such effect of attachment.
Belsky (2002, p. 167) reports that ‘‘Secure
individuals appraise stressful events as less threatening
than do less secure persons’’ and that such
individuals ‘‘are more likely to seek help from others
and to resolve conflicts rather than repress them.’’ It
is plausible that those with anxious or avoidant
attachment styles may be less able to respond in
positive ways to the more complex demands placed
on the individual who must make his way up the
occupational ladder in an organisation from a
lower starting point, rather than having been
predestined for high position by educational success.
Several studies have concluded that a feature
of the anxious or avoidant attachment styles is
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775 773
difficulty in managing relationships to others
(Bifulco, 2002; Park & Waters, 1989; Shaver &
Mikulincer, 2002). The anxiously attached may be
excessively dependent, and the avoidant may be less
successful in managing ongoing relationships over
time. Neither of these behavioural styles is likely to
help in the negotiation of a work career in a
complex organisation. In women, attachment style
may work in rather different ways which would
need more complex explanations. For example, in
this cohort of employed middle aged persons,
securely attached persons are more likely to be
married, and women in the highest employment
grade are less likely to be married and have children
than those in the lower grades (Bartley, Martikainen,
Shipley, & Marmot, 2004). Secure attachment
and career success may be inter-related in quite a
different manner in women than in men.
The attachment literature regards attachment
style as arising from the relationships between
parents and children early in life. One problem in
research in this area has been that child development
research has used longer interviews to establish
the relationship between childhood experiences
and attachment style. Social psychology research on
adults has, in contrast, used survey questions similar
to those reported here. This has led to difficulties in
making inferences about any relationship between
childhood experiences, attachment style as measured
by short questionnaire, and adult social and
emotional functioning (Bernier & Dozier, 2002;
Main, 1996). The Whitehall II study has included
measures of childhood experience as well as
measures of attachment style, finding that attachment
style in adulthood was related to parental
warmth in childhood (Stansfeld et al., 2006), but
not, as has been shown here, to the more material
aspects of the childhood environment indicated by
social class. It is therefore a reasonable inference
that attachment style in this case has acted as a
pathway through which early experiences of
family functioning, similar to Sweeting and West’s
‘family culture’ may influence social outcomes in
later life in the presence of relative socio-economic
There could be an alternative explanation of the
relationships seen in Table 4. It is possible that the
group of Whitehall II participants who attained the
higher occupational grades despite lower levels of
education were more likely to report ‘secure’
attachment ideation because their life trajectory
produced a generally more positive frame of mind.
This possibility was tested by examining participants’
mental health score according to Goldberg’s
General Health Questionnaire (GHQ) (Goldberg &
Williams, 1988). There was no tendency for higher
grade civil servants to report better mood according
to this measure of psychological well-being if they
also had lower levels of educational qualification.
We also tested for the possibility that Whitehall II
participants in the higher grades might be more
likely to recall greater parental warmth if they had
not attained the highest level of education, and
found that this was not the case.
Our data are therefore consistent with the
possibility that attachment style may be an intermediating
factor between early life emotional
experiences and later health, in a manner not
dissimilar to that which seems to be suggested in
the work of Sweeting and West, as well as of others
(McCarron, Gunnell, Harrison, Okasha, & Davey
Smith, 2002; Neeleman, Sytema, & Wadsworth,
2002; van de Mheen, Stronks, Looman, & Mackenbach,
1998a, b). The difference between what we
(and Sweeting and West) argue and some other
studies of ‘cumulative disadvantage’ is that attachment
style does not seem to be part of a simple
pathway between childhood socio-economic advantage
and adult social attainments (or between earlier
and later socio-economic disadvantage). Rather, we
have shown one example of the way in which earlier
life experiences may confer resilience that allows
high levels of occupational attainment in the face of
some degree of socio-cultural disadvantage.
Although social origins, through their link to
educational attainment, had a powerful effect on
occupational success in the Civil Service during the
1980s and 1990s, participants in the Whitehall II
study whose educational trajectory had been less
favourable seemed to be helped up the career ladder
from their lower starting points if they did not suffer
from the more anxious or avoidant attachment
styles. In the light of our own and others’ findings
relating the quality of family relationships to the
formation of these personal characteristics, our
study seems to add to the evidence that the quality
of family relationships may provide a source of
resilience in the face of social or economic
We are grateful to Peter Fonagy for his help in the
development and interpretation of the attachment
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775
measures. Work for this paper was carried out as
part of the ESRC Research Priority Network ‘‘Human
Capability and Resilience in their Social and
Geographical Contexts’’ Grant no. L326253061.
Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S.
(1978). Patterns of attachment: A psychological study of the
strange situation. Hillsdale, NJ: Erlbaum.
Balkrishnan, R. (1998). Predictors of medication adherence in the
elderly—Commentary. Clinical Therapeutics, 20, 764–771.
Bartholomew, K. (1990). Avoidance of intimacy: An attachment
perspective. Journal of Social and Personal Relationships, 7,
Bartley, M., Martikainen, P., Shipley, M., & Marmot, M. (2004).
Gender differences in the relationship of partner’s social class
to behavioural risk factors and social support in the Whitehall
II study. Social Science & Medicine, 59, 1925–1936.
Bartley, M., & Plewis, I. (2002). Accumulated labour market
disadvantage and limiting long-term illness: Data from the
1971–1991 Office for National Statistics’ Longitudinal Study.
International Journal of Epidemiology, 31, 336–341.
Belsky, J. (2002). Developmental origins of attachment styles.
Attachment and Human Development, 4, 166–170.
Berglund, G., Eriksson, K. F., Israelsson, B., Kjellstrom, T.,
Lindgarde, F., Mattiasson, I., et al. (1996). Cardiovascular
risk groups and mortality in an urban Swedish male
population: The Malmo Preventive Project. Journal of
Internal Medicine, 239, 489–497.
Bernier, A., & Dozier, M. (2002). Assessing adult attachment:
Empirical sophistication and conceptual bases. Attachment &
Human Development, 4, 171–179.
Bifulco, A. (2002). Attachment style measurement: A clinical and
epidemiiological perspective. Attachment and Human Development,
Bifulco, A., Moran, P. M., Ball, C., & Bernazzini, O. (2002).
Adult attachment style. II: Its relationship to clinical
depression. Social Psychiatry and Psychiatric Epidemiology,
Bifulco, A., Moran, P. M., Ball, C., & Lillie, A. (2002). Adult
attachment style: Its relationship to psychosocial depressivevulnerability.
Social Psychiatry and Psychiatric Epidemiology,
Bowlby, J. (1982). Attachment and Loss. New York: Basic Books.
Brunner, E., Shipley, M. J., Blane, D., Smith, G. D., & Marmot,
M. G. (1999). When does cardiovascular risk start? Past and
present socioeconomic circumstances and risk factors in
adulthood. Journal of Epidemiology and Community Health,
Droomers, M., Schrijvers, C. T., Casswell, S., & Mackenbach, J.
P. (2003). Occupational level of the father and alcohol
consumption during adolescence; patterns and predictors.
Journal of Epidemiology and Community Health, 57, 704–710.
Fergus, S., & Zimmerman, M. A. (2005). Adolescent resilience: A
framework for understanding healthy development in the face
of risk. Annual Review of Public Health, 26, 399–419.
Fonagy, P. (2001). Introduction to attachment theory. Ch 1.
Attachment theory and psychoanalysis. London: Karnac (pp.
Fonagy, P. (2005). Attachment, the development of self, andits
pathology in personality disorders. Psychomedia Telematic
(accessed 1 Feb 2005))
Galobardes, B., Lynch, J. W., & Davey Smith, G. (2004).
Childhood socioeconomic circumstances and cause-specific
mortality in adulthood: Systematic review and interpretation.
Epidemiologic Reviews, 26, 7–21.
Gil, F. P., & Rupprecht, R. (2003). Attachment research. Current
aspects of attachment theory and development psychology as
well as neurobiological aspects in psychiatric and psychosomatic
disorders. Nervenarzt, 74, 965–971.
Glendinning, A., Hendry, L. B., & Shucksmith, J. (1995).
Lifestyle, health and social class in adolescence. Social Science
& Medicine, 41, 234–248.
Glendinning, A., Love, J. G., Hendry, L. B., & Shucksmith, J.
(1992). Adolescence and health inequalities. Extension to
Macintyre and West. Social Science & Medicine, 35, 79–87.
Glendinning, A., Shucksmith, J., & Hendry, L. (1997). Family life
and smoking in adolescence. Social Science & Medicine, 44,
Goldberg, D. P., & Williams, P. (1988). The user’s guide to the
general health questionnaire. Slough: NFER-Nelson.
Green, J., & Goldwyn, R. (2002). Annotation: Attachment
disorganisation and psychotherapy: New findings in attachment
research and their potential implications for developmental
psychopathology in childhood. Journal of Child
Psychology and Psychiatry and Allied Disciplines, 43, 835–846.
Haavet, O. R., Saugstad, O. D., & Straand, J. (2005). Positive
factors associated with promoting health in low-risk and highrisk
populations of 15-and 16-year-old pupils in Oslo,
Norway. Acta Paediatrica, 94, 345–351.
Hazan, C., & Shaver, P. (1987). Romantic love conceptualised as
an attachment process. Journal of Personality and Social
Psychology, 52, 511–524.
Hertzman, C., & Power, C. (2003). Health and human development:
Understandings from life-course research. Developmental
Neuropsychology, 24, 719–744.
Hesse, E., & Main, M. (1999). Second generation effects of
unresolved trauma in nonmaltreating parents: Dissociated,
frightening and threatening parental behaviour. Psychoanalytic
Inquiry, 10, 481–540.
Irons, C., & Gilbert, P. (2005). Evolved mechanisms in adolescent
anxiety and depression symptoms: The role of the attachment
and social rank systems. Journal of Adolescence, 28, 325–341.
Mackenbach, J. P. (2005). Genetics and health inequalities:
Hypotheses and controversies. Journal of Epidemiology and
Community Health, 59, 268–273.
Main, M. (1996). Introduction to the special section on
attachment and psychopathology: 2. Overview of the field
of attachment. Journal of Consulting and Clinical Psychology,
Massie, H., & Szajnberg, N. (2002). The relationship between
mothering in infancy, childhood experience and adult mental
health: Results of the brody prospective longitudinal study
from birth to age 30. International Journal of Psychoanalysis,
Maunder, R. G., Panzer, A., Viljoen, M., Owen, J., Schalk, H., &
Hunter, J. J. (2006). Physicians’ difficulty with emergency
department patients is related to patients’ attachment style.
Social Science & Medicine, 63, 552–562.
ARTICLE IN PRESS
M. Bartley et al. / Social Science & Medicine 64 (2007) 765–775 775
McCarron, P., Gunnell, D., Harrison, G. L., Okasha, M., &
Davey Smith, G. (2002). Temperament in young adulthood
and later mortality: Prospective observational study. Journal
of Epidemiology and Community Health, 25, 888–892.
Mickelson, K. D., Kessler, R. C., & Shaver, P. R. (1997). Adult
attachment in a nationally representative sample. Journal of
Personality and Social Psychology, 73, 1092–1106.
Neeleman, J., Sytema, S., & Wadsworth, M. (2002). Propensity to
psychiatric and somatic ill-health: Evidence from a birth
cohort. Psychological Medicine, 32, 793–803.
Park, K. A., & Waters, E. (1989). Security of attachment and
preschool friendships. Child Development, 60, 1076–1081.
Power, C., & Matthews, S. (1997). Origins of health inequalities
in a national population sample. Lancet, 350, 1584–1589.
Priel, B., & Besser, A. (2001). Bridging the gap between
attachment and object relations theories: A study of the
transition to motherhood. British Journal of Medical Psychology,
Scharf, M., Mayseles, O., & Kivenson-Baron, I. (2004).
Adolescents’ attachment representations and developmental
tasks in emerging adulthood. Developmental Psychology, 40,
Schmidt, S., Nachtigall, C., Wuethrich Martone, O., & Strauss,
B. (2002). Attachment and coping with chronic disease.
Journal of Psychosomatic Research, 53, 763–773.
Schoon, I. (2006). Risk and Resilience: Adaptations in changing
times. Cambridge: Cambridge University Press.
Shaver, P. R., & Mikulincer, M. (2002). Attachment-related
psychodynamics. Attachment and Human Development, 4,
Singh-Manoux, A., Ferrie, J., Chandola, T., & Marmot, M.
(2004). Socioeconomic trajectories across the life course and
health outcomes in midlife: Evidence for the accumulation
hypothesis? International Journal of Epidemiology, 33,
Stansfeld, S.A., Head, J., Bartley, M., & Fonagy, P. (2006). Social
position, early deprivation and the development of attachment.
Social Science & Medicine (forthcoming).
Stein, H., Koontz, A. D., Fongay, P., Allen, J. G., Fultz, J.,
Brethour, J. R., et al. (2002). Adult attachment: What are the
underlying dimensions? Psychology and Psychotherapy, 75,
Stewart-Brown, S. L., Fletcher, L., & Wadsworth, M. E. J.
(2005). Parent-child relationships and health problems in
adulthood in three UK national birth cohort studies.
European Journal of Public Health, 15, 640–646.
Sweeting, H., & West, P. (1995). Family life and health in
adolescence: A role for culture in the health debate? Social
Science & Medicine, 40, 163–175.
Tremblay, R. E. (1999). When children’s social development fails.
In D. P. Keating, & C. Hertzman (Eds.), Developmental
Health and the Wealth of Nations (pp. 55–71). New York: The
van de Mheen, D. (1998). Is there indirect selection? Determinants
of interngenerational social mobility. Findings from the
Dutch GLOBE study. PhD thesis, Erasmus University
van de Mheen, H., Stronks, K., Looman, C. W., & Mackenbach,
J. P. (1998a). Does childhood socioeconomic status influence
adult health through behavioural factors? International
Journal of Epidemiology, 27, 431–437.
van de Mheen, H., Stronks, K., Looman, C. W. N., &
Mackenbach, J. P. (1998b). Role of childhood health in the
explanation of socioeconomic inequalities in early adult
health. Journal of Epidemiology and Community Health, 52,
Waller, E., Scheidt, C. E., & Hartmann, A. (2004). Attachment
representation and illness behaviour in somatoform disorders.
Journal of Nervous and Mental Disease, 192, 200–209.
Waters, E., Crowell, J., Elliott, M., Corcoran, D., & Treboux, D.
(2002). Bowlby’s secure base theory and the social/personality
psychology of attachment styles: Work(s) in progress.
Attachment and Human Development, 4, 230–242.
Werner, E. E. (2004). Journeys from childhood to midlife: Risk,
resilience, and recovery. Pediatrics, 114, 492.
West, P., Macintyre, S., Annandale, E., & Hunt, K. (1990).
Social-class and health in youth—findings from the west of
scotland 20-07 study. Social Science & Medicine, 30, 665–673.
Whalley, L. J., & Deary, I. J. (2001). Longitudinal cohort study
of childhood IQ and survival up to age 76. British Medical
Journal, 322, 819.