The Journal of Positive Psychology

Dedicated to furthering research and promoting good practice

ISSN: 1743-9760 (Print) 1743-9779 (Online) Journal homepage: https://www.tandfonline.com/loi/rpos20

Testing the interrelationship between mental well-

being and mental distress in young people

Leanne Hides, Catherine Quinn, Stoyan Stoyanov, Wendell Cockshaw, David

J. Kavanagh, Ian Shochet, Frank Deane, Peter Kelly & Corey L. M. Keyes

To cite this article: Leanne Hides, Catherine Quinn, Stoyan Stoyanov, Wendell Cockshaw, David
J. Kavanagh, Ian Shochet, Frank Deane, Peter Kelly & Corey L. M. Keyes (2019): Testing the
interrelationship between mental well-being and mental distress in young people, The Journal of
Positive Psychology, DOI: 10.1080/17439760.2019.1610478

To link to this article: https://doi.org/10.1080/17439760.2019.1610478

Published online: 09 May 2019.

Submit your article to this journal

Article views: 107

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rpos20
THE JOURNAL OF POSITIVE PSYCHOLOGY
https://doi.org/10.1080/17439760.2019.1610478

Testing the interrelationship between mental well-being and mental distress in

young people
Leanne Hidesa,b, Catherine Quinna,b, Stoyan Stoyanova,b, Wendell Cockshawb, David J. Kavanaghb, Ian Shochetb,
Frank Deanec, Peter Kellyc and Corey L. M. Keyesd
a
School of Psychology, The University of Queensland, Brisbane, Australia; bSchool of Psychology & Counselling, Centre for Children’s Health
Research and Institute of Health & Biomedical Innovation, Queensland University of Technology, Brisbane, Australia; cSchool of Psychology,
University of Wollongong, Wollongong, Australia; dDepartment of Sociology, Emory University, Atlanta, Georgia

ABSTRACT ARTICLE HISTORY

The relationship between mental well-being and distress in young people is unclear but impor- Received 16 July 2018
tant for both theoretical and practical reasons. This study tests these relationships using both Accepted 8 April 2019
dimensional and categorical measures of mental well-being and distress. Two thousand and KEYWORDS
eighty-two Australians’ (16–25 years) completed an online survey. A subsample (n = 389) Mental wellbeing; mental
completed diagnostic telephone interviews to identify 12-month DSM-IV mood, anxiety and distress; mental disorder;
substance use disorders. Five competing models of the relationship between mental well-being young people; model
and distress were compared. Only the bifactor model fit the data, indicating both mental well-
being and distress are important subcomponents of a young person’s overall level of mental
health. Over 90% of young people had flourishing or moderate well-being and only 6% were
languishing. Those with past year mental disorders were only 15% less likely to be flourishing and
4% more likely to be languishing. These findings highlight the well-being potential of young
people living with and without mental disorders.

Introduction mental well-being and mental distress and disorders, parti-

cularly in young people.
Globally, mental health disorders are at their peak and are
Definitions of well-being were first derived from the
the leading cause of disability and death in 15–25-year-olds
philosophical distinction between hedonia and eudaimo-
(Institute for Health Metrics and Evaluation, 2016). Anxiety,
nia. The hedonic perspective focused on the role of plea-
depression, and substance use disorders are the most
sure attainment and pain avoidance to attain happiness
common disorders in this age group (Kessler et al., 2012;
and well-being (Kahneman, Diener, & Schwarz, 1999), while
Lawrence et al., 2015; Ravens-Sieberer et al., 2008).
the eudaimonic approach, focused on achieving one’s full
However, despite the almost exclusive focus of research
potential and living a meaningful life (Ryan & Deci, 2001). In
on mental illness over the past century, there has been little
contemporary psychological terms, the hedonic approach
reduction in the prevalence or burden of mental disorders
translates into the concept of subjective well-being (SWB;
(Keyes, Satvinder, & Eduardo, 2010). Internationally there is
Kahneman et al., 1999), which is comprised of both life
growing recognition that complete mental health requires
satisfaction and the presence of more positive than nega-
more than the absence of mental distress and disorders
tive emotions. The eudemonic approach was encapsulated
(World Health Organization, 2004). There must also be
by psychological well-being (PWB; Ryff & Singer, 2008)
signs of positive functioning and a positive evaluation of
which focuses on living a meaningful, authentic and purpo-
one’s life, which constitutes mental well-being (World
seful life. Keyes added social well-being, arguing that social
Health Organization, 2004). Accordingly, governments
functioning was a key component of eudemonic well-being
and health services are incorporating mental well-being
and integrated all three components to define mental well-
as a key goal and outcome in mental health policy and
being as high levels of emotional, psychological and social
practice (de Cates, Stranges, Blake, & Weich, 2015; Her
functioning (Keyes, 2007). Keyes also developed the Mental
Majesty’s Government, 2011; Keyes, 2005a; The
Health Continuum-Short Form (MHC-SF; Keyes, 2005b) to
President’s New Freedom Commission on Mental Health,
provide a multidimensional measure of emotional, psycho-
2003). Despite this interest, multiple perspectives on how
logical and social well-being. The MHC-SF provides
to best define and measure well-being remain, and there is
a continuous scale of well-being as well as categorical
a limited understanding of the relationship between

CONTACT Leanne Hides l.hides@uq.edu.au

© 2019 Informa UK Limited, trading as Taylor & Francis Group
2 L. HIDES ET AL.

criteria for flourishing, moderate and languishing mental by research indicating adults with the combination of
well-being, in a manner analogous to the definition of low mental distress and high mental well-being have
a major depressive disorder in the Diagnostic and fewer chronic illnesses, lower health-care utilisation,
Statistical Manual of Mental Disorders (DSM; Keyes et al., and fewer days absent from work (Keyes, 2007).
2012). Broadly, flourishing well-being requires positive Adolescents with high mental well-being and low mental
emotions, hedonia, a positive evaluation of one’s life and distress report lower levels of suicidal behaviour and
positive functioning, in the same way, that depression have higher academic performance (Keyes et al., 2012).
requires a negative mood, anhedonia, a negative view of Gains in mental well-being have also been found to
one’s life and poor functioning (Keyes et al., 2012). prospectively predict declines in mental distress, while
Specifically, flourishing well-being involves high levels of losses in mental well-being predict greater mental dis-
at least one of the three aspects of emotional well-being tress over a 10-year follow up (Keyes et al., 2010).
and at least six of eleven aspects of positive psychological To date, no studies have examined the relationship
and social functioning on the MHC-SF. Languishing well- between mental distress and well-being using confir-
being is characterised by low levels of these features, while matory factor analyses in young people. Research is
those not languishing or flourishing have moderate well- also yet to test a number of other models for how
being. mental well-being and mental distress may interact to
Keyes multidimensional conceptualisation of well- influence an individuals’ mental health. This includes
being is consistent with other approaches which have a higher order model in which mental well-being and
combined hedonic and eudaimonic well-being compo- mental distress are two lower order factors that feed
nents to define flourishing (Huppert & So, 2013; into a higher order latent (complete mental health)
Seligman, 2011). However, few researchers have exam- model (See Figure 1(1.4)). Bifactor models, which allow
ined the relationship between mental well-being and item variance to be explained by a combination of an
mental distress or disorders. Traditionally, mental well- overarching mental health factor and two orthogonal-
being and mental distress were simply thought to be specific factors (mental well-being and mental distress;
opposite poles of a single dimension (single-axis model), see Figure 1(1.5)) are also yet to be tested. Such bifactor
in which increased mental well-being would be synon- models are increasingly being used to investigate
ymous with reduced mental distress and vice versa. To dimensionality in complex psychological constructs
increase current understanding of the relationship (Reise, Moore, & Haviland, 2010).
between mental well-being and distress, Keyes (2005a) Extant research is also yet to use structured diagnostic
developed the two-continua model of complete mental interviews to examine the association between mental dis-
health, which posits that mental well-being and mental orders and mental well-being status in young people. This is
distress (symptoms of mental illness) are two separate important because 75% of mental disorders emerge before
but related dimensions of mental health. In this model, the age of 25 (Kessler et al., 2005). Understanding links
mental well-being and mental distress are inversely asso- between mental well-being and mental disorders could
ciated, such that increases in one, may result in improve mental health treatment. Arguments that mental
decreases in the other, but these changes are not wholly well-being and mental distress are correlated but also have
explained by each other. In support of the two-continua components that operate independently, suggest treat-
model, Keyes (2005a) presented confirmatory factor ana- ments which enhance well-being even while psychopathol-
lyses using data from the Midlife in the United States ogy is present, might improve treatment outcomes for
(MIDUS) survey of 3,032 US adults aged 25 to 74. He mental disorders. Therapeutic examples of this can be
compared three models: (1) the single-axis model, in seen in approaches such as Acceptance and Commitment
which mental well-being and mental distress are at the Therapy (ACT) where recipients are encouraged to act in
opposite ends of the same continuum, (2) the two-axes accordance with valued life directions even in the presence
orthogonal (uncorrelated) model, in which mental well- of psychological distress (Wilson & Murrell, 2004).
being and mental distress are unrelated, so an indivi- This study, aimed to test the relationship between
dual’s well-being would have no implications for the mental distress and well-being in a large sample of
extent of their mental distress and (3) the two-axes young people, using both (i) dimensional measures of
oblique (correlated) model which is consistent with the past month mental well-being (MHC-SF) and mental
two continua model, whereby well-being and distress distress (Depression Anxiety Stress Scale (DASS)) and
are correlated but distinct constructs. Results indicated (ii) categorical measures of past month mental well-
the two-factor correlated model had a better model fit being status (MHC-SF criteria for languishing, moderate,
than either a single or two-factor orthogonal model. flourishing) and past 12-month DSM-IV mental disor-
Further evidence for the two continua model is provided ders in a subsample.
 THE JOURNAL OF POSITIVE PSYCHOLOGY 3

(1.1) (1.2) (1.3)

(1.4) (1.5)

Figure 1. (1.1) the single-axis model, (1.2) the two-axes orthogonal model, (1.3) the two-axes oblique (correlated) model, (1.4) a higher
order model and (1.5) the bifactor model.

Materials and methods (4-items) functioning. Items are scored on a 6-point

response scale from never (0) to every day (5). The MHC-
Participants and recruitment
SF has high levels of reliability, and discriminant, con-
Participants were Australian residents aged 16 to 25 years. vergent and cross-cultural validity (Keyes et al., 2012). In
Young people were recruited to complete an online ‘well- addition to the MHC-SF continuous score, the MHC-SF
being survey’ using student emails, a Facebook page, web- was also categorised into three levels of mental well-
sites related to youth mental health and well-being (e.g. being Flourishing required high levels (‘every day’ or
ReachOut.com.au), emails to the 75 + partner organisations ‘almost every day’) of at least one of the three aspects
of the Young and Well Cooperative Research Centre and of emotional well-being and at least six of eleven
snowballing techniques. Participants who completed the aspects of positive psychological (Keyes, 2005a) and
survey received a movie voucher ($7.50 AUD). This required social functioning during the past month (Keyes et al.,
participants to provide an email address and telephone 2012). Languishing required low levels (‘never’ or ‘once
number. Telephone numbers, emails, and IP addresses or twice’) of at least one aspect of emotional well-being
were scanned to ensure participants did not complete and at least six aspects of psychological or social func-
the survey more than once. All participants were invited tioning. Those who satisfy neither set of criteria, mean-
to take part in a telephone-delivered diagnostic interview ing they do not have high (in the case of flourishing) or
within 4 weeks of completing the online survey and receive low levels (i.e. are languishing) of both emotional well-
an additional $40 AUD for their time. A priori sample size of being and psychological/social functioning are classi-
400 telephone interviews was set due to staffing and fied as moderate mental well-being. Thus, people with
funding restrictions. Ethical approval was provided by the flourishing well-being have higher levels of emotional
Queensland University of Technology Human Research well-being and functioning than those with moderate
Ethics Committee (Approval number 1200000290). well-being, who in turn have higher levels of emotional
Participant consent was obtained online for the survey well-being and functioning than languishing individuals
and verbally for the telephone interview. (Keyes, 2005a).
Mental distress was assessed with the DASS (Lovibond &
Lovibond, 1995). The 21-item DASS contains three, seven-
Measures
item depression, anxiety and stress subscales measured on
Mental well-being was measured with the MHC-SF a 4-point response scale from did not apply to me (0) to
(Keyes, 2005b). This 14-item scale measures how fre- applied to me very much, or most of the time (3). The DASS
quently respondents experience emotional well-being has high levels of internal consistency, and good discrimi-
(3-items) and positive psychological (7-items) and social nant, convergent and cross-cultural validity.
4 L. HIDES ET AL.

The Mini-International Neuropsychiatric Interview (AIC) was used to compare alternate models. Smaller
(MINI) Plus 5.0.0 (Sheehan et al., 1998, 1997) was also values of AIC denoted better fit (Vrieze, 2012).
administered to a subsample to assess the presence of In the bifactor model, McDonald’s (1999) omega relia-
12-month DSM-IV diagnoses of mood (major depressive bility coefficients were examined to determine the pro-
disorder, recurrent depression, bipolar), anxiety (panic portions of total score variance explained by the latent
disorder with/without agoraphobia, social anxiety, gener- factors. Omega (ω) corresponds to variance in a scale total
alised anxiety, obsessive-compulsive and post-traumatic score accounted for by all constructs (general and speci-
stress disorders) and substance use (alcohol & illicit drug fic), sometimes described as the latent variable analogue
abuse or dependence) disorders. The MINI has high levels to the coefficient alpha (Rodriguez, Reise, & Haviland,
of reliability and concordance with gold standard DSM-IV 2016a). OmegaS (ωs) reflects the unweighted, systematic
structured diagnostic interviews (Sheehan et al., 1998, variance attributable to the specific factors (that includes
1997). All interviews were delivered by psychologists both the general and specific variance). Omega hierarch-
with Masters level qualifications, recorded, and 20% ical (ωh) assesses the proportion of total scale variance
were listened to and rated by the second psychologist accounted for by the general construct alone, and ωh
using the MINI. Inter-rater reliability indicated substantial assesses the proportion of variance in subscale total
inter-rater agreement (kappa = 0.66). Any discrepancies scores accounted for by the corresponding specific con-
were checked by a third-rater until a consensus diagnosis struct alone (Brunner, Nagy, & Wilhelm, 2012; Rodriguez
was reached. et al., 2016a). The explained common variance (ECV) will
also be calculated for the bifactor model, by dividing the
Data analysis factor loadings of the general construct by the factor
loadings of the general construct plus the subscale total
Confirmatory Factor Analyses (CFA), with maximum like-
score factor loadings. An ECV of 1.0 represents
lihood estimated, using AMOS were used to test five mod-
a unidimensional construct (Bentler, 2009; Reise et al.,
els (See Figure 1) of the relationship between mental
2010); however, it has been suggested that ECV > .60 and
distress and well-being: (1.1) the single-axis model which
ωh > .70 are indicative a unidimensional structure (Reise,
posits that mental distress and well-being have no unique
Scheines, Widaman, & Haviland, 2013). The percentage of
variance and are best accounted for by a single latent
uncontaminated correlations (PUC) will also be examined
factor; (1.2) the two-axes orthogonal model which specifies
using by calculating the number of correlations between
that mental well-being and distress are two uncorrelated
items from different group factors divided by the total
latent factors, with no shared variance; (1.3) the two-axes
number of correlations (Rodriguez, Reise, & Haviland,
oblique (correlated) model in which mental well-being and
2016b), which provides further indication of unidimen-
distress are two correlated factors with both shared and
sionality, as when ECV > .70 and PUC > .70, relative bias
unique variance; (1.4) a higher order model, in which two
is likely to be low, and reflective of a unidimensional
lower order factors (well-being and mental distress) feed
structure. Relative parameter bias, which is calculated by
into a higher order latent (mental health) model and (1.5)
comparing the values of the bifactor general factor load-
the bifactor model, which allows item variance to be
ings to unidimensional factor loadings, will also be exam-
explained by a combination of an overarching mental
ined to determine the risk of bias that could arise from
health factor and one of two orthogonal-specific factors,
a unidimensional construct. Descriptive statistics and Chi-
mental well-being and mental distress.
square tests were used to describe and compare the
Similar to Keyes (2005a) the MHC-SF and DASS items
association between the mental disorders with mental
were parcelled into six observed variables representing
well-being status of participants.
the emotional (ɑ = .89), social (ɑ = .87) and psychological
(ɑ = .89) components of the MHC-SF; and the depression
(ɑ = .92), anxiety (ɑ = .86) and stress (ɑ = .88) components Results
of the DASS. Due to sensitivity of the χ2 statistic to sample
Participant characteristics
size, to determine model fit the Comparative Fit Index
(CFI), Tucker-Lewis Index (TLI), Root Mean Square Error of Data were collected between December 2012 and
Approximation (RMSEA) and Standardised Root Mean May 2014. A total of 2,130 young people commenced
Square Residual (SRMR) were examined (Hox & Bechger, the online survey, 48% (n = 1,012) expressed interest in
1998; Kaplan, 2000). Hu and Bentler (1999) suggest that completing the telephone-diagnostic interview. All
good model fit is indicated by CFI and TLI values of 0.95 or respondents who provided complete data on the MHC-
higher, an RMSEA value of 0.06 or lower and an SRMR SF and DASS in the online survey were included in the
value of 0.08 or lower. The Akaike Information Criterion study (n = 2,082). A total of 389 young people out of 516
 THE JOURNAL OF POSITIVE PSYCHOLOGY 5

(75%) approached, completed the telephone diagnostic Torres Strait Islander and 43% were in a relationship or
interview. The main reasons for exclusion were: declined married.
to participate (n = 6), provided incorrect contact details
(n = 7) or contactable (n = 114) but did not complete the
interview within the one-month time frame (expired).
Structure of mental well-being and mental distress
A further 316 young people who expressed interest in Fit statistics are presented in Table 3. For the single-axis
the interview were not contacted as they were ineligible and two-axes orthogonal models, fit was very poor. For
(did not complete the full survey; n = 126) or had expired the two-axes oblique model, fit improved on all indices,
(n = 189) prior to the commencement of the interviews but was only acceptable on the SRMR. The higher order
in the early stages of the study. model was over-identified and therefore model fit could
The demographic characteristics are presented in not be determined. The bifactor model was the only
Table 1. Descriptive statistics and correlations between model that obtained a satisfactory fit to the data.
the MHC-SF and DASS total scale and sub-scales are In the bifactor model, the omega (ω) was 0.901 indicat-
presented in Table 2. There was a slightly lower propor- ing that 90% of the variance in the mental well-being and
tion who reported a first language other than English in mental distress variables was explained by a combination
the subsample. Otherwise, no significant differences of the general and specific latent factors. Omega hierarch-
were found between those who did and did not com- ical (ωh) was 0.646 for the general factor (mental health)
plete the MINI on any demographic variable or MHC-SF, indicating that 65% of score variance for mental health
t(2080) = 0.61, p = .539, and DASS total scores, t(2080) = was explained by the general factor alone, and that the
1.11, p = .267. inflation in the omega score (90%) was due to the multi-
Almost 50% of participants reported the main activ- dimensionality of the construct (26% of the variance in the
ity of ‘student’ and 30% and 10% worked part-time or mental health was explained by the specific sub-factors of
full-time, respectively. Ten percent were unemployed mental well-being and distress).
and 78% were currently enrolled in tertiary education. Regarding subscales, ωs were 0.785 and 0.933 for the
A first language other than English was reported by mental well-being and distress specific factors, respec-
12%, 89% had completed the final high school year, tively, indicating that high proportions of subscale total
56% had a tertiary qualification, 22% had a university score variances were explained by a combination of the
degree or above, 3% identified as Aboriginal and/or general factor and the respective specific factor. The

Table 1. Participant characteristics of the total sample and subsample.

Total sample MINI subsample
Variable (n = 2,082) (n = 389) Statistic
Age, M (SD) 20.25 (2.46) 20.36 (2.43) t(2080) = 1.04, p = .298
Female gender, % (n) 64% (1328) 67% (258) χ2 = 1.52, p = .218
English not first language, % (n) 12% (239) 8% (30) χ2 = 6.59, p = .010
Finished high school, % (n) 89% (1862) 91% (353) χ2 = 1.21, p = .272
University degree, % (n) 22% (462) 25% (97) χ2 = 2.18, p = .140
In relationship, % (n) 43% (904) 46% (177) χ2 = 0.94, p = .333
Main activity student, % (n) 47% (971) 50% (192) χ2 = 1.55, p = .213
MINI: Mini International Neuropsychiatric Interview

Table 2. Descriptive statistics and correlations for well-being and mental health scales and subscales for the total sample (n = 2130)
and subsample (n = 389).
Mean SD Range α 1 2 3 4 5 6 7 8
1. MHC_Emotional 3.65 1.03 0–5 .89 - .68*** .77*** .86*** −.58*** −.34*** −.39*** −.49***
2. MHC_Social 2.82 1.18 0–5 .87 .67*** - .73*** .91*** −.46*** −.27*** −.33*** −.40***
3. MHC_Psychological 3.40 1.08 0–5 .89 .77*** .74*** - .94*** −.59*** −.36*** −.40*** −.50***
4. MHC_Total 45.47 14.06 0–70 .94 .86*** .91*** .94*** - −.59*** −.36*** −.41*** −.51***
5. DASS_Depression 4.98 5.03 0–21 .92 −.57*** −.45*** −.58*** −.58*** - .66*** .72*** .89***
6. DASS_Anxiety 4.11 4.32 0–21 .86 −.34*** −.28*** −.36*** −.36*** .66*** - .76*** .89***
7. DASS_Stress 6.17 4.75 0–21 .88 −.38*** −.33*** −.40*** −.40*** .72*** .77*** - .92***
8. DASS_Total 15.26 12.69 0–63 .95 −.48*** −.40*** −.50*** −.50*** .89*** .89*** .92*** -
Mean 3.70 2.88 3.45 46.17 5.01 4.11 6.51 15.64
SD 1.05 1.17 1.05 13.81 5.04 4.26 4.89 12.73
Range 0–5 0–5 0–5 6–70 0–21 0–20 0–21 0–56
α .88 .86 .87 .93 .92 .84 .88 .94
DASS: Depression Anxiety Stress Scale; MHC: Mental Health Continuum; α: Cronbach’s alpha. Mean total scores are presented for variables 4–8. For ease of
comparison of MHC subscales mean scores are presented. For DASS variables for total sample (n = 2079).
*** p < .001
6 L. HIDES ET AL.

Table 3. Fit statistics for the confirmatory factor analyses.

χ2 df p CFI TLI RMSEA SRMR AIC
Single factor 2468.11 9 > 0.01 0.70 0.49 0.36 0.15 2492.11
2 factor – oblique 1040.00 9 > 0.01 0.87 0.79 0.24 0.28 1063.99
2 factor – orthogonal 458.94 8 > 0.01 0.94 0.90 0.17 0.07 484.94
Bifactor 14.18 4 0.01 1.00 0.99 0.04 0.01 48.18
CFI: Comparative Fit Index; TLI: Tucker-Lewis Index; RMSEA: Root Mean Square Error of
Approximation; SRMR: Standardised Root Mean Square Residual; AIC: Akaike Information Criterion.
Results are not reported for the higher order model as it did not achieve statistical fit.

omega specific (ωhs) was 0.338 for the well-being factor, n = 3 both), and 5% (n = 20) with a co-morbid mood/
indicating that 34% of the variance in well-being was anxiety and substance use disorders.
explained by the specific factor, 45% of the variance in The rates and types of mental disorders within each
well-being was explained by the general factor (totalling well-being category varied. While the proportion of the
0.785), and the remaining 22% was error variance. Omega MINI sub-sample with languishing well-being was low,
specific (ωhs) was 0.420 for the distress factor, indicating this group had the highest rate (64%; n = 14) of mental
that 42% of the variance in mental distress was explained disorders, primarily mood/anxiety disorders (41%, n = 9)
by the specific factor, 51% by the general factor (totalling or co-morbid mood/anxiety and substance use disorders
0.933), the remaining 7% being error variance. (23%, n = 5). Within the moderate well-being category
Explained common variance (ECV) analyses showed 43% (n = 78) had a mental disorder; 22% (n = 40) had
that 67% of the variance in the mental well-being and a mood/anxiety disorder; 16% (n = 29) had a substance
distress variables was explained by the common general use disorder, and a further 5% (n = 9) had a co-morbid
factor (mental health) and 33% was explained by the mood/anxiety and substance use disorder. Those with
specific factors (16% each). The ECV was large (>.60; flourishing well-being had the lowest rates of mental
Reise, 2012); however, the percentage of uncontami- disorders (25%; n = 47); 6% (n = 12) had a mood/anxiety
nated correlations was lower than the >.70 cut-off (.60) disorder; 16% (n = 29) had a substance use disorder, and
recommended and the average relative bias was 12% a further 3% (n = 6) had a co-morbid mood/anxiety and
when comparing the general factor loadings to those substance disorder.
from a unidimensional construct, which is on the cusp of Figure 2 presents data on the mental well-being of
an unacceptable parameter estimate bias (Muthén, individuals with mental disorders. Young people with
Kaplan, & Hollis, 1987). In sum, the bifactor provides any mood/anxiety or substance use disorder were
evidence for an over-arching general factor of mental less likely to be flourishing and more likely to have
health, while also revealing the possibility for separable- moderate well-being than those with no disorder (χ2
specific factors of mental distress and mental well-being. (2) = 21.36, p = 0.001). Levels of languishing, how-
The depression, emotional well-being and psychological ever, were only slightly elevated. Due to the low
well-being related items loaded most strongly on the proportion languishing overall, the expected cell
general mental health factor. counts for persons languishing in the other mental
disorder groups were small; hence, chi-squared tests
are not presented. The difference in rates of languish-
Association of mental disorders with mental
ing for those in the mood/anxiety disorder group was
well-being status
modest. No participants with a current major depres-
For the complete sample (n = 2,082), only 6% (n = sive disorder (n = 16) who reported a negative mood/
116) were defined as having languishing well-being. anhedonia (more days than not) in the past two
In contrast, 49% (n = 1,014) had moderate well-being, weeks, were flourishing, 75% (n = 12) had moderate
and 46% (n = 952) had flourishing well-being. well-being and 25% (n = 4) were languishing. No
Similarly, in the MINI sub-sample (n = 389), 6% (n = participants in the substance use only group were
22) were languishing, 47% (n = 181) had moderate languishing, and this group had the highest rates of
well-being, and 48% (n = 186) were flourishing. Rates moderate and flourishing well-being. In contrast, the
of current DSM-IV disorders in the subsample (n = co-morbid group had the highest rates of languish-
389) were 36% (n = 139) for any mood, anxiety or ing. Broadly, these results confirm the complexity of
substance use disorder, comprised of 16% the interrelationship of mental distress and well-
(n = 61) with a mood or anxiety disorder only [herein being and the need to consider both of these com-
mood/anxiety], 15% (n = 58) with a substance use ponents when considering a young person’s mental
disorder only (n = 53 alcohol; n = 8 other substance; health.
 THE JOURNAL OF POSITIVE PSYCHOLOGY 7

Figure 2. Mental well-being status within different mental disorders.

SUD = substance use disorder

Discussion well-being and mental distress dimensions, the bifac-

tor model provides a more complex conceptualisa-
There is little consensus on how mental well-being is
tion. Rather than separate constructs, mental distress
related to mental distress and disorders in young peo-
and well-being may be better conceptualised as dis-
ple. This study examined the interrelationship between
tinct subcomponents of an overarching construct of
mental distress and well-being using both categorical
mental health. The bifactor model emphasises the
and dimensional measures of the constructs in a large
overarching dimension of mental health, while pro-
sample of 2082 young Australians aged 16–25 years.
viding some evidence for multidimensional subcom-
Five models examining the relationship between
ponents, and the importance of assessing both
mental well-being and distress were examined. Of the
mental distress and well-being to obtain
models tested, only the bifactor model satisfied conven-
a comprehensive assessment of mental health.
tional fit criteria. While it might not be surprising that the
Previous studies conducted in large samples of US,
bifactor model showed the best fit, as this kind of model
Polish and South African adults have supported the
has a risk of ‘overfitting’ and accommodating nonsense
validity of the two-continua (oblique two-axes) model
response patterns, examination of omega values and
of mental health (Karaś, Cieciuch, & Keyes, 2014;
estimated common variance further revealed that while
Keyes, 2005a; Keyes et al., 2008). However, the bifac-
there was considerable evidence for an underlying latent
tor model was not tested in these studies. Whether
construct of mental health, there was also specific var-
the bifactor model has a better fit than the oblique
iance explained by mental distress and mental well-
two-axis (two continua) model in adults needs to be
being individually (Bonifay, Lane, & Reise, 2017). The
determined. Current results may be due to the age of
general factor explained 67% of the variance in mental
this sample, given young people tend to have higher
health. Mental well-being and distress specific factors
levels of mental well-being, mental distress and dis-
explained 33% of the variance in mental health. This
orders than adults (2005a; Keyes et al., 2012; Slade,
provides some support for a multi-dimensional view of
Grove, & Burgess, 2011).
mental health but also highlights the common variance
represented by the general factor. Over 90% of young Australians had flourishing (49%) or
These findings are of theoretical and clinical sig- moderate (45%) well-being and only 6% were languish-
nificance. While Keyes’ (2007) posits that mental ing. Comparable rates of flourishing (52%), moderate
health is defined by two separate but related mental (45%) and languishing (4%) well-being were reported in
8 L. HIDES ET AL.

a US study of 5,689 university students (54% female; study had the highest rate of languishing (25%) well-
18–30 years; 89% aged 18–25 years (Keyes et al., 2012)). being. This finding supports the need for a continued
Young Australians with DSM-IV mood/anxiety in the focus on co-morbid populations in both research and
past 12 months had lower rates of flourishing (20%) well- practice. It also suggests the relationship between men-
being, and only a slight elevation in languishing well- tal well-being status and mental disorders may not be
being (15%), as the greatest proportion of young people consistent across all disorders. Further research on the
in this group had moderate well-being (66%). Similar relationship between mental well-being and other men-
proportions of US students with current depression or tal disorders (e.g. psychotic and other co-morbid disor-
anxiety disorders had languishing (16%), flourishing ders) is required, as well as within different age groups
(20%) and moderate (65%) well-being (Keyes et al., and ethnicities.
2012). The small discrepancies may reflect cross-cultural From a clinical perspective, current results indicate men-
differences as well as methodological differences in the tal health is best conceptualised as an overarching con-
diagnostic categories and measures used (DSM-IV clinical struct, with some evidence for multidimensional
interview versus diagnostic cut-offs on a screening mea- components of mental distress and well-being. This high-
sure in the US study) and rates of mental disorders identi- lights the importance of clinicians assessing both mental
fied in each sample (Keyes et al., 2012). Nevertheless, distress and well-being in young people to obtain an accu-
results indicate it is possible for a young person who has rate picture of their overall level of mental health. The
had a mood/anxiety disorder in the last 12 months to also overlap between these two subcomponents, also suggests
experience moderate or flourishing well-being. There that interventions targeting well-being are likely to have
were 16 adolescents who met criteria for a major depres- a direct impact on young people’s mental distress, hence,
sive disorder in the past 2 weeks. Of these, although none outcomes are likely to be achieved through complimentary
was flourishing, 75% had moderate well-being and only interventions that target mental distress and well-being. An
25% were languishing. This indicates moderate well- example is ACT where mental distress might be targeted
being is achievable among young people with a current through acceptance strategies (e.g. mindfulness and non-
depressive disorder. Together these findings highlight the avoidance of distressing cognitions) and well-being might
substantial well-being potential of young people living be simultaneously promoted through encouraging values
with and without mental disorders. clarification and related goal pursuit. Indeed, randomised
While for mood/anxiety disorders there was a clear controlled trials of ACT-based programs have reported
trend for the rates of well-being to be lower as the improvements in both mental distress and well-being
prevalence of disorders was higher, this trend was not (Bohlmeijer, Lamers, & Fledderus, 2015; Burckhardt,
apparent for substance use disorders. Young people with Manicavasagar, Batterham, & Hadzi-Pavlovic, 2016;
a single diagnosis of a substance use disorder had the Räsänen, Lappalainen, Muotka, Tolvanen, & Lappalainen,
highest rates of flourishing (50%), none had languishing 2016). Other examples of dual-approach interventions
well-being and 50% had moderate well-being. The include well-being therapy, mindfulness-based and solu-
absence of languishing well-being in this group is sur- tion-focused therapy.
prising. This result requires replication but may reflect Strengths of this study include the large sample and
the low rates of languishing well-being found in this use of a structured diagnostic interview in a subsample
sample and young people overall. These results may to identify mental disorders. Conclusions are limited by
also reflect a positive association between substance cross-sectional data and the small number of young
use, mental well-being and social connectedness people with languishing well-being. The representative-
among young people before the adverse consequences ness of the sample is also unclear, due to the non-
of sustained substance use emerge as they enter adult- random recruitment method used. The rate of mental
hood. Consistent with this idea, an Australian study on disorders in the present study (36% with a current
794 adults (mean age = 36.1 years) entering a residential mood, anxiety or substance use disorder), was higher
substance use treatment program found the majority than the 26% rate reported in young people in
reported moderate well-being (54%) with lower rates of a general Australian community survey (Australian
flourishing (22%) and languishing (24%) well-being Bureau of Statistics, 2007). This may be due to the self-
(McGaffin, Deane, Kelly, & Ciarrochi, 2015). Although co- selected nature of the sample whereby young people
morbid disorders in such samples are likely to be high were invited to participate in a ‘well-being survey’
(e.g. 64–71%; Mortlock, Deane, & Crowe, 2011), the through recruitment channels that included websites
McGaffin study did not assess for the presence of co- focused on mental health issues. It may also reflect
morbid mental disorders. Young people with co-morbid differences in the diagnostic interviews used or increas-
mood/anxiety and substance use disorders in the current ing levels of psychopathology among young people.
 THE JOURNAL OF POSITIVE PSYCHOLOGY 9

The 14-item MHC-SF provides a valid and reliable mea- Acknowledgments

sure of total well-being, in which each item represents
The Young and Well Cooperative Research Centre, which was
a distinct symptom of emotional, psychological or social established under the Australian Government‘s Cooperative
well-being (Keyes et al., 2012). The criteria for flourishing Research Centres Program provided some staffing support for
well-being on the MHC-SF were developed by Keyes to this project.
mirror the symptom clusters required for a DSM major
depressive disorder (Keyes et al., 2012). Similar methods Disclosure statement
were used to develop a new measure of flourishing in
No potential conflict of interest was reported by the authors.
a sample of 43,000 adults, which was found to have
a similar factor structure (positive characteristics, appraisal Funding
and functioning) to the MHC-SF (Huppert & So, 2013). Just
as DSM criteria for mental disorders have evolved over This project was funded by Australian Research Council (ARC),
time, research using both dimensional and categorical Future Fellowship [FT120100780] awarded to Leanne Hides.
measures of well-being is required to determine how well-
being features are distributed and cluster to define flour- References
ishing well-being in large representative population-based
cohorts. Future research is also required to prospectively Australian Bureau of Statistics (ABS). (2007). National survey of
mental health and wellbeing: Summary of results. Canberra:
examine the relationship between mental well-being and ABS.
mental distress over time. Bentler, P. M. (2009). Alpha, dimension-free, and model-based
Results indicate that both mental well-being and internal consistency reliability. Psychometrika, 74, 137–143.
mental distress are important contributors to a young Bohlmeijer, E. T., Lamers, S. M. A., & Fledderus, M. (2015).
person’s overall level of mental health. Over 90% of Flourishing in people with depressive symptomatology
increases with acceptance and commitment therapy.
young Australians had flourishing (49%) or moderate
Post-hoc analyses of a randomized controlled trial.
(45%) well-being and only 6% were languishing. Young Behaviour Research and Therapy, 65, 101–106.
people with a mental disorder in the past year were Bonifay, W., Lane, S. P., & Reise, S. P. (2017). Three concerns with
only 15% less likely to be flourishing and 4% more likely applying a bifactor model as a structure of psychopathology.
to be languishing than those without a disorder. These Clinical Psychological Science, 5(1), 184–186.
highlight the well-being potential of young Australian’s Brunner, M., Nagy, G., & Wilhelm, O. (2012). A tutorial on
hierarchically structured constructs. Journal of Personaiity,
living with mental disorders, which has a number of 80(4), 796–846.
important implications for mental health promotion Burckhardt, R., Manicavasagar, V., Batterham, P. J., & Hadzi-
and treatment. The high rates of both flourishing Pavlovic, D. (2016). A randomized controlled trial of strong
(34%) and moderate well-being (56%) among young minds: A school-based mental health program combining
people with mental disorders could be used to encou- acceptance and commitment therapy and positive
psychology. Journal of School Psychology, 57, 41–52.
rage help-seeking and treatment engagement among
de Cates, A., Stranges, S., Blake, A., & Weich, S. (2015). Mental
this population. Informing young people that mental well-being: An important outcome for mental health
distress and well-being lie on separate continua and services? British Journal of Psychiatry, 207(3), 195–197.
both contribute to their overall level of mental health Her Majesty’s Government. (2011). No health without mental
(rather than a strict mental health/mental disorder health. Delivering better mental health outcomes for people
dichotomy), could also help reduce the stigma asso- of all ages. London: Department of Health.
Hox, J. J., & Bechger, T. M. (1998). An introduction to structural
ciated with mental disorders (e.g. Schomerus et al., equation modelling. Family Science Review, 11, 354–373.
2016) Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in
Further research is required to identify what predicts, covariance structure analysis: Conventional criteria versus new
enhances and maintains mental well-being in young alternatives. Structural Equation Modeling: A Multidisciplinary
people with and without mental disorders to inform Journal, 6(6), 1–55.
Huppert, F. A., & So, T. T. C. (2013). Flourishing across Europe:
the development of new interventions targeting the
Application of a new conceptual framework for defining
complete mental health of young people. While there well-being. Social Indicators Research, 110, 837–861.
is preliminary evidence for interventions (e.g. ACT) tar- Institute for Health Metrics and Evaluation. (2016). GBD com-
geting both mental distress and well-being, further pare data visualization. Retrieved from http://vizhub.health
research on dual focus interventions is urgently data.org/gbd-compare
required. Such research is likely to have important Kahneman, D., Diener, E., & Schwarz, N. (1999). Well-being: The
foundations of hedonic psychology. New York, NY: Russell
implications for mental health practice and policy Sage Foundation.
aimed at reducing the prevalence and burden of men- Kaplan, D. (2000). Structural equation modeling: Foundations
tal disorders. and extensions. Thousand Oaks, CA: Sage Publications.
10 L. HIDES ET AL.

Karaś, D., Cieciuch, J., & Keyes, C. L. M. (2014). The Polish Muthén, B., Kaplan, D., & Hollis, M. (1987). On structural equa-
adaptation of the Mental Health Continuum-Short Form tion modeling with data that are not missing completely at
(MHC-SF). Personality & Individual Differences, 69, 104–109. random. Psychometrika, 52(3), 431–462.
Kessler, R. C., Avenevoli, S., Costello, E. J., Georgiades, K., The President‘s New Freedom Commission on Mental Health.
Green, J. G., Gruber, M. J., . . . Merikangas, K. R. (2012). (2003). Achieving the promise: Transforming mental health
Prevalence, persistence, and sociodemographic correlates care in America final report. Rockville, MD: Substance Abuse
of DSM-IV disorders in the national comorbidity survey and Mental Health Services Administration (SAMHSA).
replication adolescent supplement. Archives of General Räsänen, P., Lappalainen, P., Muotka, J., Tolvanen, A., &
Psychiatry, 69, 372–380. Lappalainen, R. (2016). An online guided ACT intervention
Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K., & for enhancing the psychological wellbeing of university
Walters, E. E. (2005). Lifetime prevalence and age-of-onset students: A randomized controlled clinical trial. Behaviour
distributions of DSM-IV disorders in the national comorbid- Research and Therapy, 78, 40–42.
ity survey replication. Archives of General Psychiatry, 62(8), Ravens-Sieberer, U., Wille, N., Erhart, M., Bettge, S.,
593–602. Wittchen, H.-U., Rothenberger, A., . . . DöPfner, M. (2008).
Keyes, C. L. M. (2005a). Mental illness and/or mental health? Prevalence of mental health problems among children
Investigating axioms of the complete state model of health. and adolescents in Germany: Results of the BELLA study
Journal of Consulting and Clinical Psychology, 73(3), within the national health interview and examination
539–548. survey. European Child & Adolescent Psychiatry, 15, 22–33.
Keyes, C. L. M. (2005b). The subjective well-being of America’s Reise, S. P. (2012). The rediscovery of bifactor measurement
youth: Toward a comprehensive assessment. Adolescent models. Multivariate Behavioral Research, 47(5), 667–696.
and Family Health, 4(1), 3–11. Reise, S. P., Moore, T. M., & Haviland, M. G. (2010). Bifactor
Keyes, C. L. M. (2007). Promoting and protecting mental models and rotations: Exploring the extent to which multi-
health as flourishing: A complementary strategy for dimensional data yield univocal scale scores. Journal of
improving national mental health. American Psychologist, Personaiity Assessment, 92(6), 544–559.
62, 95–108. Reise, S. P., Scheines, R., Widaman, K. F., & Haviland, M. G.
Keyes, C. L. M., Eisenberg, D., Perry, G. S., Dube, S. R., (2013). Multidimensionality and structural coefficient bias
Kroenke, K., & Dhingra, S. S. (2012). The relationship of in structural equation modeling: A bifactor perspective.
level of positive mental health with current mental disor- Educational and Psychological Measurement, 73(1), 5–26.
ders in predicting suicidal behavior and academic impair- Rodriguez, A., Reise, S. P., & Haviland, M. G. (2016a). Applying
ment in college students. Journal of American College bifactor statistical indices in the evaluation of psychological
Health, 60(2), 126–133. measures. Journal of Personality Assessment, 98(3), 223–237.
Keyes, C. L. M., Satvinder, S. D., & Eduardo, J. S. (2010). Change Rodriguez, A., Reise, S. P., & Haviland, M. G. (2016b). Evaluating
in level of positive mental health as a predictor of future bifactor models: Calculating and interpreting statistical
risk of mental illness. American Journal of Public Health, 100 indices. Psychological Methods, 21(2), 137–150.
(12), 2366–2371. Ryan, R. M., & Deci, E. L. (2001). On happiness and human
Keyes, C. L. M., Wissing, M., Potgieter, J. P., Temane, M., potentials: A review of research on hedonic and eudaimo-
Kruger, A., & van Rooy, S. (2008). Evaluation of the mental nic well-being. Annual Review of Psychology, 52, 141–166.
health continuum-short form (MHC-SF) in Ryff, C. D., & Singer, B. H. (2008). Know thyself and become
Setswana-speaking South Africans. Clinical Psychology & what you are: A eudaimonic approach to psychological
Psychotherapy, 15(3), 181–192. well-being. Journal of Happiness Studies, 9, 13–39.
Lawrence, D., Johnson, S., Hafekost, J., Boterhoven de Schomerus, G., Angermeyer, M. C., Baumeister, S. E.,
Haan, K., Sawyer, M., Ainley, J., & Zubrick, S. R. (2015). Stolzenburg, S., Link, B. G., & Phelan, J. C. (2016). An online
The mental health of children and adolescents: Report on intervention using information on the mental health-mental
the second Australian Child and Adolescent Survey of illness continuum to reduce stigma. European Psychiatry, 32,
Mental Health and Wellbeing. Canberra, Australia: 21–27.
Department of Health. Seligman, M. E. P. (2011). Flourish. New York, NY: Simon &
Lovibond, P. F., & Lovibond, S. H. (1995). The structure of Schuster.
negative emotional states: Comparison of the Depression Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P.,
Anxiety Stress Scales (DASS) with the beck depression and Janavs, J., Weiller, E., . . . Dunbar, G. C. (1998). The
anxiety inventories. Behaviour Research and Therapy, 33, Mini-International Neuropsychiatric Interview (M.I.N.I): The
335–343. development and validation of a structured diagnostic psy-
McDonald, R. P. (1999). Test theory: A unified treatment. chiatric interview for DSM-IV and ICD-10. Journal of Clinical
Mahwah, NJ: Erlbaum. Psychiatry, 59(Suppl 20), 22–33.
McGaffin, B. J., Deane, F. P., Kelly, P. J., & Ciarrochi, J. (2015). Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Janavs, J.,
Flourishing, languishing and moderate mental health: Weiller, E., Keskiner, A., . . . Dunbar, G. C. (1997). The validity
Prevalence and change in mental health during recovery of the Mini International Neuropsychiatric Interview (MINI)
from drug and alcohol problems. Addiction, Research & according to the SCID-P and its reliability. European
Theory, 23, 351–360. Psychiatry, 12, 232–241.
Mortlock, K. S., Deane, F. P., & Crowe, T. P. (2011). Screening Slade, T., Grove, R., & Burgess, P. (2011). Kessler psychological
for mental disorder comorbidity in Australian alcohol and distress scale: Normative data from the 2007 Australian
other drug residential treatment settings. Journal of national survey of mental health and wellbeing. Australian
Substance Abuse Treatment, 40(4), 397–404. and New Zealand Journal of Psychiatry, 45(4), 308–316.
 THE JOURNAL OF POSITIVE PSYCHOLOGY 11

Vrieze, S. I. (2012). Model selection and psychological theory: M. M. Linehan (Eds.), Mindfulness and acceptance: Expanding
A discussion of the differences between the Akaike the cognitive-behavioral tradition (pp. 120–151). New York,
Information Criterion (AIC) and the Bayesian Information NY: Guilford Press.
Criterion (BIC). Psychological Methods, 17, 228–246. World Health Organization (WHO). (2004). Promoting mental
Wilson, K. G., & Murrell, A. R. (2004). Values work in acceptance health: Concepts, emerging evidence, practice (Summary
and commitment therapy. In S. C. Hayes, V. M. Follette, & report). Geneva: WHO.