This article was downloaded by: [University of California, San Diego]

On: 30 September 2014, At: 14:59

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Aging & Mental Health

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/camh20

Examination of a theoretical model of psychological

resilience in older age
a b c
Gill Windle , David A. Markland & Robert T. Woods
a
Institute of Medical and Social Care Research, University of Wales , Bangor, UK
b
School of Sport, Health and Exercise Sciences, University of Wales , Bangor, UK
c
School of Psychology, University of Wales , Bangor, UK
Published online: 26 Aug 2008.

To cite this article: Gill Windle , David A. Markland & Robert T. Woods (2008) Examination of a theoretical model of
psychological resilience in older age, Aging & Mental Health, 12:3, 285-292, DOI: 10.1080/13607860802120763

To link to this article: http://dx.doi.org/10.1080/13607860802120763

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of
the Content. Any opinions and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of information. Taylor and Francis shall
not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other
liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 Aging & Mental Health
Vol. 12, No. 3, May 2008, 285–292

Examination of a theoretical model of psychological resilience in older age

Gill Windlea*, David A. Marklandb and Robert T. Woodsc
a
Institute of Medical and Social Care Research, University of Wales, Bangor, UK; bSchool of Sport, Health and Exercise
Sciences, University of Wales, Bangor, UK; cSchool of Psychology, University of Wales, Bangor, UK
(Received 15 April 2007; final version received 11 October 2007)

Objective: This article integrates a number of theoretical perspectives and examines the concept of
psychological resilience in older age. Drawing on the literature it is hypothesised that an overarching
construct – resilience – accounts for the functioning of a number of psychological resources (self-esteem, personal
competence and control).
Method: The factorial validity of the resources as indicators of resilience is tested using confirmatory factor
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

analysis. The analyses focus on previously unexplored survey data drawn from a representative sample of people
aged between 50 and 90 in England, Wales and Scotland (N ¼ 1847).
Results: The results find a common factor (a higher-order model) provides the best explanation of the
relationships between the resources, demonstrating an important first account for developing further work on
this concept.
Conclusion: Exploring what might form the basis of resilience from a psychological perspective enables a deeper
understanding of why some individuals can remain positive in difficult circumstances, particularly some of the
challenges of ageing.
Keywords: aging; resilience; personality; mental health; factor structure

Introduction family environments, e.g. mentally ill parents

This article integrates a number of theoretical perspect- (Garmezey, 1974; Rutter, 1985) and poor
ives and examines the concept of psychological socio-economic status (Garmezey, 1991) do not display
resilience in older age. Drawing on the literature maladaptive behaviours or become mentally ill
it is hypothesised that an overarching construct – themselves. They are able to function positively despite
resilience – represents a common core and accounts for their circumstances.
the functioning of a number of psychological More recently it has been proposed that in older
resources. people, a positive response to a stressful life event is
Resilience has been described as being ‘able to indicative of a resilient process (Hardy, Concato, &
recover from or adjust to misfortune or change’ (The Gill, 2004). These authors found that independence in
Penguin English Dictionary, 2001). It can be seen as instrumental activities of daily living, positive
the opposite to vulnerability and encompasses personal self-ratings of health and few depressive symptoms
competences across cognitive, emotional and social were independently associated with high resilience.
domains (Tizard & Clarke, 1992). It was derived from However, they acknowledge that the conceptualisation
observations that although exposed to substantial of resilience as the response to a stressful event has
stressors and risks, people can still function positively limitations as it cannot be measured in the absence of
and recover quickly from set-backs (Rutter, 1995). a significant stressful event.
Consequently resilient individuals flourish when chal- On the other hand, the examination of resilience as
lenged (Ryff & Singer, 2003). Psychological resilience is an internal personality resource provides the opportu-
thought to be important in late life as a component of nity to address this limitation and examine the
successful psychosocial adjustment (Wagnild & psychological basis of why people are resilient. This
Young, 1993) and is associated with mental health is of central interest in understanding resilience (Ryff &
(Nygren et al., 2005). Singer, 2003). Consequently, exploring what might
Although there may be a general agreement of the form the basis of resilience from a psychological
definition of resilience, it tends to have been measured perspective could subsequently enable a deeper under-
indirectly, being assumed more as a hypothetical standing of why some individuals can remain resilient
construct, as indicated from research, which finds in difficult circumstances, particularly some of the
successful outcomes despite set-backs. This has been challenges of ageing.
observed mainly from much of the initial research into In considering the inner psychological qualities
resilience undertaken on children. Many children people possess; Rutter (1987) described resilient
across a range of diverse settings, particularly negative individuals as possessing self-efficacy, self-esteem and

*Corresponding author. Email: g.windle@bangor.ac.uk

ISSN 1360–7863 print/ISSN 1364–6915 online

ß 2008 Taylor & Francis
DOI: 10.1080/13607860802120763
http://www.informaworld.com
 286 G. Windle et al.

a range of problem solving skills. Others describe The multi-dimensional approach to control (Paulus,
resilient individuals as possessing self-confidence, 1983) acknowledges that different aspects of control are
curiosity, self-discipline, self-esteem and control over drawn upon for interacting with the wider environment.
the environment, intellectual functioning and self- The concept of self-esteem is often used interchangeably
perceptions such as self-efficacy (Beardslee, 1989; with efficacy, although they refer to conceptually
Masten, 1999) and personal competence and accep- separate entities. The mixing of the two concepts may
tance of self and life (Wagnild & Young, 1993). The relate to the cognitive process, which influences self-
latter authors developed a 25-item scale to assess esteem. As an important part of the self-concept,
resilience, however, to date very little research has efficacy may determine the choice of behaviour, and
incorporated this measure nor undertaken any psycho- the feedback from others may influence whether that
metric assessment of its factor structure. Given the behaviour will be chosen again due to the effects on self-
wider scope for the range of psychological resources esteem (Osborne, 1996). Thus from a conceptual
that could be considered indicators of resilience, this perspective, there are grounds to consider that aspects
article aims to take a broader perspective to the of these resources may be measuring the same thing.
concept by examining additional personality resources. This section aims to test the hypothesis that an
Personality resources are suggested to protect overarching construct may account for these resources.
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

individuals in the face of adversity and lead to positive

adaptive behaviour by acting as a ‘buffer’ (Rutter,
1987) or as compensatory factors which directly Justification for proposed model of
influence outcomes (Masten, 1989). In this context, psychological resilience
resilience could be viewed as an ‘umbrella’ term for A previous study has taken a similar approach,
such psychological resources which are central to the examining the possibility of conceptual overlap
self. Other research that has examined resources such between four widely studied personality measures
as mastery, self-esteem and optimism has conceptual- (generalised self-efficacy, neuroticism, locus of control
ised them as part of the core of the reserve capacity and self-esteem) (Judge et al., 2002). Arguing that there
that provides a resilient basis in older age (Gallo, are theoretical similarities in the way that these
Bogart, Vranceanu, & Mathews, 2005). Given the resources operate, they examined the extent to which
theoretical basis for the role of psychological resources more measures than necessary are often used in
to provide a sense of resilience, this article takes a research to account for such psychological constructs.
novel approach and examines the relationship between Previous empirical work (Judge, Erez, & Bono, 1998)
number of psychological resources, testing their found high correlations between the constructs, and a
factorial validity as indicators of psychological resi- factor analysis found that the four measures are
lience. The following section outlines the rationale for loaded onto one single factor. The authors state
the proposed methodology for the model of resilience. such findings provide some support for the argument
that more measures are being used than is necessary
to account for a common psychological construct
Do the psychological resources share a (Judge et al., 1998).
common source? Expanding this reasoning further, they undertook a
Within psychology numerous measures exist which meta-analysis of research that had examined the
are assumed to capture distinct aspects of the self, and relationships between any two of the four constructs.
the development of specific literatures around con- Of the 258 articles identified, 75 were included that had
structs such as self-esteem, self-efficacy and control reported correlations. The results indicated that the
is substantial. Within personality research, such con- constructs were substantially related, ranging from
structs are often examined in isolation, and little 0.40 to 0.85 with an average correlation of 0.60 (Judge
attention has been given to the possibility that they et al., 2002).
might share a common basis (Judge, Erez, Bono & Further work by Judge et al. (2002) using
Thoresen, 2002). Whilst acknowledging that not all confirmatory factor analysis tested the underlying
specific traits might indicate an overarching construct, structure of the scales. Two models were compared; a
these authors state that ‘new and existing measures first-order factor model in which the factors were not
must be evaluated on the basis of a possible common allowed to correlate and a second-order factor model.
core when there is reason (on empirical and/or It has been argued that second-order factors would
theoretical grounds) to believe that such a common- account for correlated errors that are common between
ality exists’ (Judge et al., 2002, p. 693). first-order factors (Gerbing & Anderson, 1984).
The theoretical basis for the common core A second-order factor is mathematically equivalent
‘resilience’ has been outlined previously. There it was to a first-order correlated factor model (Bollen, 1989).
described how resilient individuals possessed a range of However, Judge et al. (2002) reasoned that a second-
psychological resources. In addition, there are further order factor is preferable as it explicitly considers the
aspects in terms of the way the resources operate, structural relationships of the individual scales. They
indicating that there is some conceptual similarity. found that a single second-order factor – the common
 Aging & Mental Health 287

core – explained the relationships amongst the four (combining to eight groups) and for urban/rural
measures. distinction.
In discussing their work they recommend that
‘researchers who study these traits should consider the
possibility that the uniqueness of measures of these Data collection
traits is overwhelmed by their commonality’ (Judge In order to locate respondents, a door-to-door census
et al., 2002, p. 707). They state that similarities was conducted in each of the districts using post code
amongst traits should be acknowledged and the generated addresses. Interviewers recorded whether
measures considered as indicators of a common anyone within the target age range (50–90) lived
construct (Judge et al., 2002). there. These were used to generate a randomly
However, there is a problem with some of their selected sample for subsequent interviews. The
methodology and reasoning. Whilst providing evidence questionnaire was administered to respondents by
that the measures are related, the confirmatory factor interviewers face-to-face, in their own home, in the
analysis does not demonstrate that the traits measure first language of the respondent. Interviewers were
the same construct. A single latent factor would better recruited and trained by the research team. The
represent this approach. Rather the analysis shows that project funders did not require clinical ethical
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

they might share conceptual overlap in that they approval to be obtained. However, this project
measure aspects of the same construct. Thus their ensured that clear ethical and professional conduct
methodology, whilst informative, has some flaws. guidelines were adopted by all involved. Guidelines
However, in relation to the measures used in this for professional conduct and guidelines for ethical
article, with the exception of neuroticism they are considerations were circulated to the interviewers
conceptually very similar to those examined by Judge prior to the meeting and were reiterated at the
et al. (2002). Consequently there is some empirical training sessions. After training, interviewers under-
basis, despite limitations to support the hypothesis that stood the necessity of obtaining consent from inter-
the measures in this study may also share a common viewees, issues regarding confidentiality and contact
source. However, a stronger aspect of this hypothesis is with respondents. The final number of completed
derived from the theoretical basis. questionnaires was 1853, yielding a total response rate
The following analysis follows one of the of 49%. The mean age was 65.86 (SD ¼ 9.32) and
approaches taken by Judge and co-workers and 55% were female. A decision was taken by the project
examines the extent to which the measures of self- co-ordinator not to record for analyses of the reasons
esteem, competence, socio-political control and inter- for non-response, as an examination of these factors
personal control converge using the method of was not possible.
confirmatory factor analysis. Three models are tested:
(1) a single factor model, (2) an uncorrelated higher-
order model, (3) a higher-order model where the Measures
factors are allowed to correlate.
Self-esteem was assessed with the 10-item Rosenberg
Self Esteem Scale (Rosenberg, 1965)  ¼ 84.
Method Respondents answered on a five-point scale ranging
from strongly agree (1) to strongly disagree (5).
Design It should be noted that the original 1965 Rosenberg
This article is based on secondary data analyses. The scale was developed to use a four-point scale, however
data were collected during 2002 as part of the there are a number of studies that have used differing
European Study of Adult Well-Being (ESAW), a response scales for the same concept (e.g. Ranzijn,
cross-sectional survey of people aged 50–90 in six Keyes, Luszcz, & Feather, 1998; Rosenberg, Schooler,
countries (Austria, Italy, Luxembourg, the Schoenbach, & Rosenberg, 1995; Shahini, Dipboye, &
Netherlands, Sweden and the UK). This article Phillips, 1990).
analyses the UK data (England, Scotland and Wales). Interpersonal control and socio-political control
are distinct dimensions of the Spheres of Control scales
(Paulus & Christie, 1981) consisting of 10 questions for
Sampling procedure each ‘sphere’. Each question is rated on a seven-point
The initial target sample for the UK was set at 2000. In Likert scale that ranges from strongly agree to strongly
order to develop a sample that represented as equally disagree, with the number of positive and negative
as possible residents from England, Scotland and items balanced in each sphere. Cronbach’s alpha was
Wales the sample was divided equally between the 0.80 for interpersonal control (seven items) and 0.73
three countries. Population statistics for each age for socio-political control (six items). This scale also
group were used to generate a proportionate stratified consists of a third sphere, personal efficacy. However,
probability sample of the national population aged this 10-item scale did not fit the data well and the final
between 50 and 90. The national sample was controlled items did not meet the minimum criteria for internal
across four age groups (of 10 year each) and sex consistency.
 288 G. Windle et al.

Personal competence was derived from the Other tests of model fit were used to provide a more
Resilience Scale (Wagnild & Young, 1993). The scale comprehensive picture of model fit. The Comparative
contains 25 questions and answers are scored on Fit Index (CFI) shows how much better the specified
a Likert scale ranging from 1 ¼ strongly disagree to target model fits in comparison to the null model, in
7 ¼ strongly agree. The scale has been found to have a which there are no relationships among the observed
two-factor structure – personal competence and variables (Diamantopoulos & Sigaw, 2000). The test
acceptance of self and life (Wagnild & Young, 1993). statistic ranges between 0 and 1.0. Hu and Bentler
However, within the ESAW UK sample the alpha (1999) recommend a cut-off value close to 0.95. The
coefficient for acceptance of self and life was too small standardised root mean square residual (SRMR) is a
(0.48) and this dimension is not used. Cronbach’s summary measure of the standardised residuals
alpha for the personal competence dimension was 0.80 (the fitted residuals divided by their estimated SEs).
(10 items). Values of 0.08 or less suggest an acceptable fit (Hu &
Bentler, 1999). The root mean square error of
approximation (RMSEA) determines how well the
Data analysis model would fit a population covariance matrix
Prior to the analyses presented here, the factor (Brown & Cudeck, 1993). Values of 0.06 or less are
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

structure of each of the scales was subjected to an recommended (Hu & Bentler, 1999).
in-depth psychometric investigation using exploratory
and confirmatory factor analysis in order to derive the
most robust measures for the resilience model. For the Results
sake of brevity, the results for this preliminary stage With the exception of the socio-political dimension
are not presented here. The items representing each of where correlations were low, medium-sized correla-
the measures were summed into their respective scale tions were found between the measures suggesting that
scores for the purpose of correlation analysis, giving these may represent a common construct as found
independent measures of self-esteem, personal compe- in the work of Judge et al. (2002). (Self-esteem and
tence, interpersonal control and socio-political control. personal competence r ¼ 0.45, p 5 0.001; self-esteem
The relationships were firstly examined using and interpersonal control r ¼ 0.46, p 5 0.001; inter-
Pearson’s r. The strength of the zero-order correlations personal control and personal competence r ¼ 0.39,
was determined by the criteria of Cohen (1992), where p 5 0.001).
large correlations are described as being 40.50, The extent to which personal competence, self-
medium correlations range between 0.30 and 0.49 esteem and interpersonal control represent a common
and small correlations range between 0.10 and 0.29. higher order construct, hypothesised here as ‘resilience’
Confirmatory factor analysis using LISREL 8.54 was tested on one half of the sample. Socio-political
was used to test the underlying structure of the control was dropped from further analysis due to its
individual scales. Asymptotic and covariance matrices lack of convergence.
were used as input, and the model estimated by The results in Table 1 show that all three models
maximum likelihood. The sample was randomly split have a large and significant chi-square although the
in half using SPSS random selection yielding two RMSEA is similar in both groups. However, the CFI
datasets for development and validation. The analyses in the higher-order model was larger and the SRMR
follow the rationale of Judge et al. (2002) and examine much lower, suggesting that although this model was
three models. In the first, the items are all estimated to initially misspecified, it was the better of the three and
load onto a single latent factor. In the second, the that further examination may result in improvement.
factors were not allowed to correlate. In this model all Subsequent analyses with this model aimed to
scale items were loaded onto their respective construct. eliminate items that were poor indicators. These were
The third examines whether a common factor removed one at a time on the basis of low loadings and
(a second-order factor) explains the relationships high residuals across other items. Each subsequent
between the distinct measures. In this model, all model was re-estimated. This resulted in the sequential
items are fixed to load onto their respective constructs, removal of eight items, which were validated in the
and these latent constructs are specified to load onto second half of the sample. The final model is
a second-order latent factor. presented in Figure 1. All factor loadings were
Model fit was assessed by a number of indices. The significant at p 5 0.01. The reliability (Cronbach’s
scaled chi-square test statistic (Satorra & Bentler, 1988) alpha) was 0.83.
has been found to closely estimate the uncorrected
chi-square when distributional assumptions are vio-
lated. However, this test is sensitive to departures from Discussion
the assumption of normality and sample size (Bentler With the exception of socio-political control, the
& Bonnet, 1980). Given the large sample size of this correlations between the scales of self-esteem,
research, it was highly likely that the chi-square interpersonal control and personal competence
statistic will not be a very useful indicator. suggested that further investigation was warranted
 Aging & Mental Health 289

Table 1. Results of model testing.

Model Chi-square p-Value df RMSEA p-Value CFI SRMR

Development
Single factor 3366.55 0.00 324 0.10 0.00 0.85 0.09
Uncorrelated 1397.10 0.00 324 0.06 0.00 0.92 0.15
Higher-order 1 1131.50 0.00 321 0.06 0.00 0.94 0.06
Higher-order 2 893.36 0.00 296 0.05 0.22 0.94 0.05
Higher-order 3 706.55 0.00 272 0.04 0.98 0.95 0.05
Higher-order 4 635.89 0.00 249 0.04 0.98 0.95 0.05
Higher-order 5 555.37 0.00 227 0.04 0.99 0.96 0.05
Higher-order 6 481.34 0.00 206 0.04 0.99 0.96 0.04
Higher-order 7 427.23 0.00 186 0.04 0.99 0.96 0.04
Higher-order 8 401.75 0.00 167 0.04 0.99 0.96 0.04
Higher-order 9 346.29 0.00 149 0.04 0.99 0.97 0.04
Validation
Higher-order 9 384.78 0.00 149 0.04 0.99 0.97 0.04
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

(socio-political control did not correlate enough with from each scale had to be removed as the residuals
any of the other dimensions to be considered as an indicated that there was some clear ambiguity, leaving
indicator of the same construct). The possibility for a total of 19 items. Then this moves the process of
this overlap between constructs was tested in the model testing away from a strictly confirmatory one.
confirmatory factor analyses where a second-order However, the validation of the final model in the
factor was found to account for the three scales. second half of the sample provides some evidence for
Consequently, it has been demonstrated that there its replication and concurrent validity.
is some overlap between the measures and that this is Construct validation is lacking in this research in
consistent with the hypothesis that they share some terms of discriminant and criterion related validity.
commonality and measure aspects of the same Due to the cross-sectional nature of the data, it was not
common construct. Drawing on previous research possible to undertake analyses using methods such as
that had suggested the types of internal characteristics the multi-trait multi-method technique to test the
possessed by resilient individuals, it was hypothesised discriminant functioning of the resilience variable.
that the psychological resources measured in this Also, the development of the resilience variable is an
research were the basis of psychological resilience. approach that is now arising from the work within this
However, the conclusion for a common factor whilst thesis. Although driven by theory, as yet there is no
plausible gives rise to a number of issues that question other external aspect to validate this with. However, in
the extent to which it is fully supported empirically. terms of the theoretical aspect and the conceptualisa-
This in turn has implications for its subsequent tion of psychological resilience it could be argued that
validity. These are now addressed in turn. the measure goes some way to reflect the validity of its
Kline (1998) states that in order for a measure to be contents.
valid, it must be reliable. In this research, the reliability From the theoretical perspective, Erikson’s theory
of the final scale was very good with  ¼ 0.83. of the life cycle (Erikson, 1963) states that the
However, this alone is not enough for claiming validity continuing developments of the self across the life
of the construct, and other approaches are required. span and the ultimate achievement of wisdom provide
The overlap found between the measures of a basis for a resilient self. The notion of a resilient self
control, competence and self-esteem provides some in older age is also proposed by Baltes and Baltes
evidence for convergent validity, however the model (1990). Other researches examining the self in older age
tested here demonstrated elements of misspecification describe how individuals scoring high on a dimension
as the chi-square statistic was significant. Other referring to ‘inner self’ referred to the presence of
researchers have advocated that because of the effect positive aspects of their lives, such as taking interest
of large samples on this statistic, it is likely that such in others or keeping independent (Coleman,
models will be rejected as the chi-square will be Ivani-Chalian, & Robinson, 1993). Such positive
significant. As a consequence, in this research other statements or themes are described as ‘building
fit indices were drawn upon to provide an assessment blocks of identity’ (Kaufman, 1986). As long as a
of fit. In this instance, these were found to be person perceives that such statements still reflect their
acceptable as they all fell into the ‘rule of thumb’ personality, the self will be preserved despite loss and
ranges (Hu & Bentler, 1999) and support the final change in other areas (Coleman et al., 1993). It could
model. be argued that many of the items which comprise the
Nevertheless in the process of examining the resilience variable (e.g. able to do things as well as
hypothesis for a common core some of the items others, keep interested in things) could also be
 290 G. Windle et al.

0.80 I have a number of good

qualities

Take positive attitude

0.47 towards self
0.41
I feel I do not have much
0.36 0.72
to be proud of (disagree)
0.62
At times I think I am no
0.63
good at all (disagree) 0.61

I wish I could have more 0.66 Esteem

0.30 respect for myself (disagree)
0.70
0.39 Satisfied with self
0.74

Inclined to feel a failure 0.61

0.37 (disagree)
0.84
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

0.32 I'm a person of worth

No trouble making and

0.51 keeping friends

0.42
Can guide conversation
0.74
with others 0.51

No difficulty in getting help 0.60 Interpersonal 0.71

0.69 Psychological
to acheive goals control resilience
0.59
Can make arrangements
0.49 to meet people 0.56

Easy to play important

0.71 part in group situations

0.55 Persevere with plans 0.70

0.67
0.59 Manage one way or other
0.64

Depend on self 0.42

0.82
Competence
0.59
0.65 Handle many things
0.58

0.66 Keep interested in things 0.53

0.72 Can be relied on

Figure 1. Relationships between interpersonal control, self-esteem and personal competence.

described as referring to the positive aspects of self- interpret and integrate self-referential experience,
preservation. promote and defend the self, and develop motivation
The items of the final model also represent a and a sense of developmental direction’ (Atchley, 1999,
composite of self-referent statements and personality p. 10). Compiling such self-schema across the lifespan
traits that reflect aspects of the self-concept outlined by enables feelings of mastery, competence and control
Bengston, Reedy, and Gordon (1985). The items could (Gurin & Brim, 1984). Consequently, there is also some
also be viewed as the packages of self knowledge theoretical validation to suggest that the items that are
derived from experience (e.g. ‘I am someone people can represented empirically by the higher-order factor
rely on’ or ‘I feel I have a number of good qualities’). reflect the characteristics of resilient individuals and
These provide the interpretative aspect of the self are firmly placed within the self concept.
concept referred to as ‘self-schema’ by Markus and A conclusion of this article is that based on the
Herzog (1991, p. 113). These ‘help the individual theoretical and empirical assessment, it would be
 Aging & Mental Health 291

incorrect to state that the construct of psychological suggested that the model tested here, although
resilience has been fully validated. In addition, the theoretically plausible, represents just one of many
model of resilience presented here, whilst underpinned approaches to resilience in older age. Consequently
by theory and other research cannot be assumed to be there are a number of theoretical issues regarding the
an all-encompassing one. Some consideration needs to conceptualisation of psychological resilience that
be given to the use of secondary data analysis and the require further investigation. On the other hand
advantages or constraints it holds. In the context of despite the potential shortcomings, this work has
this research, the availability of an underutilised, large taken a novel approach to the topic exploring a
data base provided the opportunity to test a theoretical theoretical perspective that is important to under-
approach to the concept of resilience that has not been stand the psychological aspects of older age. It has
previously explored. As such the results provide an attempted rigorous analyses to further the under-
important first baseline for developing further work on standing of psychological resilience. Indeed, it is
this concept. On the other hand, the constraints of the possible that despite some of the limitations, the
data present limitations around the measures that were model might adequately capture the essential psychol-
used, and what other aspects could potentially be ogical resources that are important for mental health
considered as indicators of resilience. and well-being in older age.
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

The measure of self-esteem used in this research is a Further work aims to examine this position. It will
well used and validated measure with older people, focus on the role of psychological resilience, testing
however the Spheres of Control and Resilience scales whether the resilient self intervenes in the relationship
have received far less attention and as such their initial between risk factors and well-being. This is an
validity was questionable. To counter this aspect, important next step as it has been highlighted by
preliminary psychometric analyses were undertaken Bowling (2004) that the examination of intervening
that resulted in the most robust measures being psychological variables is under-researched in relation
considered for further analyses. From that perspective, to quality of life in older age. As such further analyses
the negative side of what could be achieved with the using psychological resilience will make a unique
available measures was overcome. contribution to the knowledge of how such inner
A further limitation arising from the constraints of capabilities enable a positive interpretation of
the data is that there are other factors that might also older age.
be considered as indicators of psychological resilience.
Over the last decade the impact of spirituality and
religion has become a focus of increasing attention References
(Gatz & Smyer, 2001, p. 534). Religion and spirituality
enable a sense of meaning to be attributed to Atchley, R.C. (1999). Continuity and adaptation in aging.
challenging situations that are experienced in older Baltimore: John Hopkins University Press.
age (Coleman & O’Hanlon, 2004). As such is it Baltes, P.B., & Baltes, M.M. (1990). Psychological
possible that this aspect of quality of life is also a perspectives on successful aging: The model of selective
source of resilience, in addition to the other psychol- optimisation with compensation. In P.B. Baltes, &
ogical resources. Unfortunately this domain was not M.M. Baltes (Eds.), Successful aging: Perspectives from
considered by the original questionnaire developers. the behavioural sciences (pp. 1–34). Cambridge: University
of Cambridge.
Other research suggests that factors such as self
Beardslee, W.R. (1989). The role of self-understanding in
acceptance become more prominent with age and can
resilient individuals: The development of a perspective.
be a key dimension of positive psychological function- American Journal of Orthopsychiatry, 59(2), 266–278.
ing (Ryff & Singer, 1996). It is regarded as a Bengston, V.L., Reedy, M.N., & Gordon, C. (1985). Aging
characteristic of self-actualisation, maturity and opti- and self-conceptions: Personality processes and social
mal functioning which emphasises the acceptance of contexts. In J.E. Birren, & K.W. Schaie (Eds.),
current and past life (Ryff & Singer, 1996). Acceptance Handbook of the psychology of aging (Vol. 2,
of limitations is part of the accommodative process of pp. 544–593). New York: Van Nostrand Reinhold
adaptation to potential negative aspects of ageing. The Company.
adjustment of aspirations in relation to losses and Bentler, P.M., & Bonnett, D.G. (1980). Significance tests and
functional impairments can subsequently maintain goodness of fit in covariance structures. Psychological
a positive self-identity in older age (Brandstädter & Bulletin, 88, 588–606.
Greve, 1994). In relation to age, self-acceptance has Bollen, K.A. (1989). Structural equations with latent
variables. New York: Wiley.
been found to remain consistent across young, mid-life
Bowling, A. (2004). A taxonomy and overview of quality of
and older aged cohorts (Ryff & Singer, 1996). Another
life. In J. Brown, A. Bowling, & T. Flynn (Eds.), Models of
study found no age differences across average self quality of life: A taxonomy, overview and systematic review
acceptance scores (Capara, Capara, & Steca, 2003). of the literature (pp. 6–77). Sheffield: European Forum on
Consequently, as people age they may become more Population Ageing Research.
accepting and comfortable with themselves. Brandstädter, J., & Greve, W. (1994). The ageing self:
In one respect, the model was then limited by the Stabilizing and protective processes. Developmental
constraints of the available data. As such it could be Review, 14, 52–80.
 292 G. Windle et al.

Brown, M.W., & Cudeck, R. (1993). Alternative ways of development: Positive life adaptations. Longitudinal
assessing model fit. In K.A. Bowen, & J.S. Long (Eds.), research in the social and behavioural sciences
Testing structural equation models (pp. 445–455). (pp. 281–296). Dordrecht, Netherlands: Kluwer-
California: Sage Publications. Academic Publishers.
Capara, G.V., Capara, M., & Steca, P. (2003). Personality’s Nygren, B., Alex, L., Jonsén, E., Gustafson, Y., Norberg,
correlates of adult development and aging. European A., & Lundman, B. (2005). Resilience, sense of
Psychologist, 8(3), 131–147. coherence, purpose in life and self-transcendence in
Cohen, J. (1992). A power primer. Psychological Bulletin, relation to perceived physical and mental health
112, 155–159. amongst the oldest old. Aging and Mental Health,
Coleman, P.G., Ivani-Chalian, C., & Robinson, M. (1993). 9(4), 354–362.
Self esteem and its sources: Stability and change in later Osborne, R.E. (1996). Self. An eclectic approach.
life. Ageing and Society, 13, 171–192. Massachusetts: Allyn and Bacon.
Coleman, P.G., & O’ Hanlon, A. (2004). Ageing and Paulus, D. (1983). Sphere-specific measures of perceived
development: Theories and research. London: Arnold. control. Journal of Personality and Social Psychology, 44,
Diamantopoulos, A., & Sigaw, J. (2000). Introducing 1253–1265.
LISREL. London: Sage Publications. Paulus, D., & Christie, R. (1981). Spheres of Control:
Erikson, E. (1963). Childhood and Society. Harmondsworth: An interactionist approach to assessment of
Downloaded by [University of California, San Diego] at 14:59 30 September 2014

Penguin. peceived control. In H.M. Lefcourt (Ed.), Research with

Gallo, L.C., Bogart, L.M., Vranceanu, A., & Mathews, K.A. the locus of control construct (Vol. 1). New York: Academic
(2005). Socioeconomic status, resources, psychological Press.
experiences and emotional responses: A test of the reserve Ranzjin, R., Keeves, J., Luszcz, M., & Feather, N.T. (1998).
capacity model. Journal of Personality and Social The role of self-perceived usefulness and competence in the
Psychology, 88(2), 386–399. self-esteem of elderly adults: Confirmatory factor analyses
Garmezey, N. (1974). The study of competence in children of the Bachman revision of Rosenberg’s Self-Esteem Scale.
at risk for severe psychopathology. In E.J. Anthony, & Journal of Gerontology: Psychological Sciences, 33B,
C. Koupernick (Eds.), The child in his family (Vol. 3, 96–104.
pp. 77–98). New York: Wiley. Rosenberg, M. (1965). Society and adolescent self-image.
Garmezey, N. (1991). Resiliency and vulnerability of adverse Princeton, NJ: Princeton University.
developmental outcomes associated with poverty. Rosenberg, M., Schooler, C., Schoenbach, C., & Rosenberg,
American Behavioural Scientist, 34, 416–430. F. (1995). Global self esteem and specific self esteem:
Gatz, M., & Smyer, M.A. (2001). Mental health and aging at Different concepts, different outcomes. American
the outset of the twenty-first century. In J.E. Birren, & Sociological Review, 60, 141–156.
K.W. Schaie (Eds.), Handbook of the psychology of aging Rutter, M. (1985). Resilience in the face of adversity:
(pp. 523–544). London: Academic Press. Protective factors and resistance to psychiatric disorder.
Gerbing, D.W., & Anderson, J.C. (1984). On the meaning of British Journal of Psychiatry, 147, 598–611.
within-factor correlated measurement errors. Journal of Rutter, M. (1987). Psychosocial resilience and protective
Consumer Research, 11, 572–580. mechanisms. American Journal of Orthopsychiatry, 57(3),
Gurin, P., & Brim, O.G. (1984). Change of self in adulthood: 316–331.
The example of sense of control. In P.B. Baltes, & Rutter, M. (1995). Psychosocial adversity: Risk, resilience
O.G. Brim (Eds.), Life span development and behaviour and recovery. Southern African Journal of Child &
(Vol. 6, pp. 281–334). New York: Academic Press. Adolescent, 7, 75–88.
Hardy, S.E., Concato, J., & Gill, T.M. (2004). Resilience of Ryff, C.D., & Singer, B. (1996). Psychological well-being:
community dwelling older persons. Journal of the Meaning, measurement and implications for psychother-
American Geriatrics Society, 52, 257–262. apy research. Psychotherapy and Pscyhosomatics, 65,
Hu, L., & Benter, P.M. (1999). Cutoff criteria for fit indexes in 14–23.
covariance structure analysis: Conventional criteria versus Ryff, C.D., & Singer, B. (2003). Flourishing under fire:
new alternatives. Structural Equation Modelling, 6(1), 1–55. Resilience as a prototype of challenged thriving.
Judge, T.A., Erez, A., & Bono, J.E. (1998). The power of In C.L.M. Keyes, & J. Haidt (Eds.), Flourishing: Positive
being positive: The relation between positive self concept psychology and the life well lived (pp. 15–36). Washington:
and job performance. Human Performance, 11, 167–187. American Psychological Association.
Judge, T.A., Erez, A., Bono, J.E., & Thoresen, C.J. (2002). Satorra, A., & Bentler, P.M. (1988). Scaling corrections for chi-
Are measures of self esteem, neuroticism, locus of control square statistics in covariance structure analysis. Proceedings
and generalised self efficacy indicators of a common core of the American Statistical Association (pp. 308–313).
construct? Journal of Personality and Social Psychology, Alexandria, VA: American Statistical Association.
83(3), 693–710. Shahini, C., Dipboye, R.L., & Phillips, A.P. (1990). Global
Kaufman, S. (1986). The ageless self: Sources of meaning in self esteem as a correlate of work related attitudes: A
late life. Madison, WI: The University of Wisconsin Press. question of dimensionality. Journal of Personality
Kline, P. (1998). Principles and practice of structural equation Assessment, 54(1–2), 276–288.
modelling. New York: The Guilford Press. The Penguin English Dictionary (2001). London: Penguin
Markus, H.R., & Herzog, A.R. (1991). The role of the self Book.
concept in aging. Annual Review of Gerontology and Tizard, B., & Clarke, A. (1992). Vulnerability and resilience in
Geriatrics, 11, 110–143. human development. London: Kingsley Publishers.
Masten, A.S. (1999). Resilience comes of age: Reflections on Wagnild, G.M., & Young, H.M. (1993). Development and
the past and outlook for the next generation of research. psychometric evaluation of the Resilience Scale. Journal of
In M.D.E.J. Glantz, & L. Jeannette (Eds.), Resilience and Nursing Measurements, 1, 165–178.