The heritability of mental health and wellbeing defined usingCOMPAS-W, a new composite measure of wellbeing

Justine M. Gatt a,n, Karen L.O. Burton a, Peter R. Schofield b,Richard A. Bryant c, Leanne M. Williams d

a The Brain Dynamics Centre, University of Sydney, Medical School and Westmead Millennium Institute, Westmead Hospital,Westmead, Sydney, NSW 2145, Australiab Neuroscience Research Australia, Randwick; and the University of New South Wales, Sydney, NSW 2031, Australiac School of Psychology, University of New South Wales, Randwick, NSW 2031, Australiad Department of Psychiatry and Behavioral Sciences, Stanford School of Medicine, Stanford University, Stanford, CA 94305-5717, USA

a r t i c l e i n f o

Article history:Received 4 November 2013Received in revised form20 March 2014Accepted 13 April 2014Available online 10 May 2014

Keywords:TWIN-EHappinessResilienceNEO-FFIWHOQOLSWLSERQ

a b s t r a c t

Mental health is not simply the absence of mental illness; rather it is a distinct entity representingwellness. Models of wellbeing have been proposed that emphasize components of subjective wellbeing,psychological wellbeing, or a combination of both. A new 26-item scale of wellbeing (COMPAS-W) wasdeveloped in a cohort of 1669 healthy adult twins (18–61 years). The scale was derived using factoranalysis of multiple scales of complementary constructs and confirmed using tests of reliability andconvergent validity. Bivariate genetic modeling confirmed its heritability. From an original 89 items weidentified six independent subcomponents that contributed to wellbeing. The COMPAS-W scale and itssubcomponents showed construct validity against psychological and physical health behaviors, highinternal consistency (average r¼0.71, Wellbeing r¼0.84), and 12-month test–retest reliability (averager¼0.62, Wellbeing r¼0.82). There was a moderate contribution of genetics to total Wellbeing(heritability h2¼48%) and its subcomponents: Composure (h2¼24%), Own-worth (h2¼42%), Mastery(h2¼40%), Positivity (h2¼42%), Achievement (h2¼32%) and Satisfaction (h2¼43%). Multivariate geneticmodeling indicated genetic variance was correlated across the scales, suggesting common genetic factorscontributed to Wellbeing and its subcomponents. The COMPAS-W scale provides a validated indicator ofwellbeing and offers a new tool to quantify mental health.

& 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Psychiatric research has been driven by an illness ideology; afocus on identifying risk markers for mental illness, their under-lying mechanisms and appropriate treatment strategies. In com-parison, much less focus has been given to mental health andwellbeing and ways to promote it. This is because of the commonmisconception that the absence of mental illness is the equivalentto mental health (Keyes, 2005; Slade, 2010). However, only onequarter of the factors that define mental health and mental illnessare shared in common (Kendler et al., 2011; Keyes, 2005; Slade,2010). This suggests that the genetic, environmental and biologicalmarkers that underlie mental illness differ substantially from thosethat underlie mental health. Accordingly we are witnessing aparadigm shift towards mental health or wellbeing (Maddux,2008), in which the focus is on defining optimal wellbeing and on

the emerging science of resilience (Russo et al., 2012; Southwickand Charney, 2012).

Existing definitions of wellbeing have tended to emphasizeconcepts of subjective wellbeing, psychological wellbeing, or theircombination. “Subjective wellbeing” reflects hedonia (Andrewsand Withey, 1976; Bradburn and Caplovitz, 1965; Diener et al.,1995, 1999; Wilson, 1967) and has been quantified by measures ofpositive and negative affect (Diener et al., 2010) and satisfactionwith life (Diener et al., 1985). “Psychological wellbeing” is tied tothe concept of eudaimonia, which is focused on the developmentof human potential (Broadie and Rowe, 2002; Ryff, 1989). Forinstance, Ryff (1989) defined psychological wellbeing as a compo-sition of autonomy, positive relations with others, purpose in life,self-acceptance, environmental mastery and personal growth,which was later quantified by the Ryff Scales of PsychologicalWell-Being (Ryff and Keyes, 1995). Only recently has a distinctionbeen drawn between subjective and psychological wellbeing (Ryffand Singer, 1998) whereas early theorists never made such adistinction (Argyle, 1987; Brickman and Campbell, 1971; Wilson,1967). There is much to be gained in the empirical study of both the

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/psychres

Psychiatry Research

http://dx.doi.org/10.1016/j.psychres.2014.04.0330165-1781/& 2014 Elsevier Ireland Ltd. All rights reserved.

n Corresponding author. Tel.: þ61 2 9845 8164; fax: þ61 2 9845 8190.E-mail address: justine.gatt@sydney.edu.au (J.M. Gatt).

Psychiatry Research 219 (2014) 204–213

subjective and psychological aspects of wellbeing as demonstratedby accumulating evidence from composite indices of wellbeing (e.g.,the Mental Health Continuum, MHC-LF; Keyes, 2007); both con-structs are strongly associated with each other with factor analyticstudies showing shared variance of 49–74% (Keyes et al., 2002;Waterman, 1993). Other studies suggest that psychological well-being directly influences subjective wellbeing (Kasser and Ryan,1996; Ryan and Deci, 2001; Sheldon, 2002). The direction ofcausation of these effects however is unclear, as many of thesestudies were correlational (Kashdan et al., 2008). In any case, eachconstruct of wellbeing –whether correlated or causal – accounts fornotable variance inwellbeing that would otherwise be unaccountedfor if focusing only on one aspect (Kashdan et al., 2008). Recently,the PERMA theory of wellbeing, has attempted to encompassesaspects of both subjective wellbeing (positive emotion, meaning)and psychological wellbeing (engagement, relationships, accom-plishment) (Seligman, 2011). This theory is a central component ofThe Positive Health Initiative (http://www.authentichappiness.sas.upenn.edu/) which aims to understand what defines wellbeing,how to determine its underlying neural circuitry and ways topromote it in the general population.

A composite measure of wellbeing may assist in assessingfunctional status in clinical and population cohorts. Functionalcorrelates of wellbeing include enhanced cognitive functioning,increased marital satisfaction, and higher income and economicflourishing (Diener and Chan, 2011; Seligman, 2008). Wellbeing isalso associated with physiological correlates of neuroendocrinelevel, cardiovascular and inflammatory activity (effect sizes of .38)(Brummett et al., 2009; Fredrickson and Levenson, 1998;Lyubomirsky et al., 2005; Steptoe et al., 2005). In addition, themeasurement of wellbeing may be useful as a prognostic indicatorof long-term health, disease outcomes and health behaviors. Forinstance, evidence from longitudinal studies shows that wellbeingis significantly predictive of longevity and ill-health (with effectsizes of .14–.18) (Diener and Chan, 2011; Howell et al., 2007;Lyubomirsky et al., 2005), as well as lower mortality in bothhealthy and diseased populations (Chida and Steptoe, 2008).In other studies, measures of wellbeing have been related toimproved health behaviors such as non-smoking, physical exer-cise, healthy diets and use of sun protection (Grant et al., 2009).

In this study, we combine this theoretical background with afocus on heritability in order to address the questions of whether wehave an innate disposition towards wellbeing or whether it issomething we can modulate and elevate through personal lifeexperience. Twin studies comparing monozygotic (MZ) to dizygotic(DZ) twins allow the contributions of genetics versus environment towellbeing to be determined. In terms of hedonic (subjective) well-being, twin studies suggest additive genetics contributes 0% (Bakeret al., 1992; Gatz et al., 1992) to 33% (Silberg et al., 1990) variance inpositive affect, up to 46% in hedonia (Bogdan and Pizzagalli, 2009)and 48% for optimism (Schulman et al., 1993). In comparison, largerheritability estimates have been reported for eudaimonic (psycholo-gical) aspects of wellbeing. In young adult twins (23–24 years),additive genetics is reported to contribute from 37% variance inpersonal growth up to 64% in positive relations (Gigantesco et al.,2011), with estimates of 41% (autonomy), 47% (purpose in life), 58%(self-acceptance) and 62% (mastery) also reported (Gigantesco et al.,2011) for Ryff's psychological wellbeing scale. In contrast, in oldertwins (51–60 years), additive genetics contributes smaller estimatesof 30% for locus of control (an element of mastery) and 41% for flowproneness (an element of engagement) (Mosing et al., 2012).Together, these findings suggest that some aspects of wellbeingmay be more heritable than others; however they could also simplyreflect differences between the twin cohorts. Each of these twinstudies varied considerably in age of the cohort: all of the twinstudies focusing on hedonia were based in twins across multiple

decades, whereas the eudaimonic studies were within very specificage groups.

In this study, we first aim to derive a valid self-report measureof wellbeing that includes both hedonic and eudaimonic aspects ofwellbeing and quantifies these individual differences in an over-arching construct of wellbeing, along with sub-constructs thatcontribute to it, using a multifactorial approach. Our second aim isto evaluate the contributions of genes and environment to well-being (both hedonic and eudaimonic aspects), and how this maybe shared or unique to specific subcomponents. In this study, wereport the development and heritability of the COMPAS-W indexof wellbeing in 1669 healthy adult twins ranging in age from18–61 years from the TWIN-E study (Gatt et al., 2012).

2. Methods

2.1. Participants

Healthy same-sex monozygotic (MZ) and dizygotic (DZ) twins from the TWIN-Estudy (the Twin study in Wellbeing using Integrative Neuroscience of Emotion)conducted at the University of Sydney, Australia (see Gatt et al., 2012 for thecomplete study protocol) participated in this study. The study received approvalfrom the Human Research Ethics Committees of the University of Sydney (03-2009/11430) and Flinders University (FCREC#08/09). All participants provided writteninformed consent prior to participation and after receiving a complete writtendescription of the study.

Twins were recruited from the Australian Twin Registry. Eligible participants weresame-sex, healthy, adult twin pairs, with English as primary language, and of pureEuropean ancestry. From initial screens conducted by the Australian Twin Registry,participants were excluded from the study if they reported current or lifetimepsychiatric illness, history of stroke or neurological disorder, genetic disorder, braininjury (causing loss of consciousness for more than 10 min), chronic and seriousmedical conditions (e.g., cancer or heart disease), blood-borne illnesses (e.g., HIV,hepatitis), drug/alcohol substance abuse, and sensory impairments to hearing, handmovement or vision not corrected by glasses/lenses. However, of those participantswho passed the screening criteria and participated in the study, 3.6% (n¼60) laterreported a current or past psychiatric, neurological or chronic illness (n¼18 forpsychiatric illness in particular) on online questionnaires. These participants were notexcluded from analysis. Zygosity in 51 twin pairs (6.8% of sample) was determinedusing results from previous DNA tests conducted by the Australian Twin Registry. Forremaining participants, zygosity was confirmed using the twin's responses to a 12-item questionnaire we developed (Gatt et al., 2012) using validated items frompreviously established questionnaires (Eisen et al., 1989; Magnus et al., 1983), shownto have 95% convergence with DNA results (Jackson et al., 2001).

The study recruited 2262 twins eligible to participate in the study. One thousandsix hundred and sixty nine twins successfully completed Phase I baseline testing ofweb-based questionnaires. Total mean age of the sample was 39.64712.73 (18–62years, with 686 males and 983 females, and mean education of 14.34 years73.00).This included 966 MZ twins (445 males, 521 females; mean age¼40.04712.44) and677 DZ twins (227 males, 450 females, mean age¼39.17713.05), and birth orderdistributions of 846 twin 1 (345 males, 501 females, mean age¼39.54712.74) and823 twin 2 (341 males, 482 females, mean age¼39.75712.73).

2.2. Measures and procedures

The methodology used in the current study is standardized across testing sitesand has been previously validated in healthy and clinical populations (Paul et al.,2007; Paul et al., 2005; Silverstein et al., 2007; Williams et al., 2005). In this study,we report on the heritability of a new composite measure of wellbeing, theCOMPAS-W index of wellbeing, derived from the following measures:

2.2.1. Self-report measuresThe self-report measures were assessed using the WebQ online test battery

(Gatt et al., 2012). These measures included: (1) the World Health OrganizationQuality of Life scale (WHOQOL-BREF) (World Health Organization (WHO), 1998)which measures overall perceptions of quality of life and health and the sub-domains of Physical Health, Psychological Health, Social Relationships and Envir-onment; (2) the Satisfaction With Life Scale (SWLS) (Diener et al., 1985) whichmeasures overall perceptions of life satisfaction relative to one's own circumstancesand standards; (3) the Internal Control Index (ICI) (Duttweiler, 1984) whichmeasures the degree to which one believes they can control events that affect them,and provides scores for overall Internal Control and sub-domains of Self-Confidenceand Autonomy; (4) the Emotion Regulation Questionnaire (ERQ) (Gross and John,2003) which measures both reappraisal and suppression strategies used to deal with

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213 205

subjective experiences of negative emotion; and (5) the NEO Five-Factor Inventory(NEO-FFI) (McCrae and Costa, 2004) which provides a measure of the five majorpersonality traits of Neuroticism, Extraversion, Openness, Agreeableness and Con-scientiousness. Each scale was selected to measure different aspects of wellbeing: theWHOQOL-BREF to assess life satisfaction, negative affect and positive relationships, theSWLS scale to assess life satisfaction, the ICI to assess autonomy, the ERQ to assesspositive and negative affect, and the NEO-FFI to assess different aspects of positive andnegative affect, life purpose, self-acceptance and personal growth.

2.2.2. Functional outcome measuresThe validity of the new COMPAS-W scale was evaluated against several

measures of functional outcomes collected during the WebQ online test battery.These measures included an assessment of “psychological health” using theDepression, Anxiety, Stress Scale (DASS-42) (Lovibond and Lovibond, 1995), theModified Differential Emotions Scale (mDES) (Fredrickson et al., 2003) and theSomatic and Psychological Health Report (SPHERE) (Hickie et al., 1998), “physicalhealth” using measures of nutrition, drug and alcohol use, nicotine use, sleep andphysical activity (Gatt et al., 2012), and “social lifestyle” using measures of leisureand social activities (Gatt et al., 2012).

2.3. Analyses

Details for all analyses are described in the Supplementary Methods.

2.3.1. Development and construct validation of the COMPAS-W scaleAn exploratory factor analysis (EFA) of 89 items selected from the five self-

report measures was first conducted using SPSS software (Version 21) in twin 1.Items were removed if inter-item correlations were 40.90, if they did notsignificantly correlate with the majority (at least 75%) of other items or if theycross-loaded on factors. The final EFA model was confirmed in twin 2.

A confirmatory factor analysis (CFA) of the reduced model was then conductedusing AMOS (version 20) in twin 1. The comparative fit of four models was tested: afirst-order common factor model, with one common latent factor that loaded on allitems (model A), a first-order 6-factor model, with each latent factor correspondingto the factors identified in the EFA (model B), a second-order hierarchical modelwith one common factor of “wellbeing” loading onto the six latent factors and theirsubsequent items (model C), and a first-order 7-factor model with the seventhlatent factor representing “total wellbeing” which loads on all items (model D). Thefit of the final model was confirmed in twin 2. Model fit was assessed using the loglikelihood chi-square difference test between nested models and overall goodness-of-fit indices.

Internal reliability of the total COMPAS-W scale and subscales was evaluatedusing Cronbach alpha. The test–retest reliability of each scale over 12-months wasevaluated using intra-class correlations (mixed effects model).

The validity of the continuous wellbeing subscales was derived by examiningcorrelations with independent functional outcome indicators of psychological healthstatus (DASS-42, mDES, SPHERE) and the scales from which they were derived.

The validity of a categorical scoring procedure of the COMPAS-W scale using standar-dized cut-offs of �1 and 1 (“flourishing”¼4þ1, “moderate”¼�1 to þ1 and“languishing”¼o�1) was also evaluated using a nonlinear canonical correlationanalysis (OVERALS) in SPSS. This categorization provides a distribution of responsesconsistent with previously reported distributions of mental health scores in the generalpopulation (Keyes, 2005). The goal of this analysis is to get maximal variance withminimal dimensional space from sets of categorical variables (Yazici et al., 2010). Theanalysis included six sets of variables: (1) COMPAS-W total Wellbeing categorical scores,(2) demographics (age, sex, education, marital status), (3) psychological status (DASS-42scores split into tertiles), (4) physical health indicators (BMI, sleep, work absenteeismand presenteeism), (5) health behaviors (frequency of use of caffeine, nicotine, alcohol,marijuana, exercise, consumption of red meat, fish, fast foods, fruit and vegetables,vitamin supplements), and (6) social lifestyle and social support (time engaged inreading, mentally challenging tasks, leisure activities, visiting family, socializing withfriends, attending sport events and volunteering in community events).

2.3.2. Genetic modelingGenetic analyses for each scale using standardized z scores were conducted using

OpenMx version 1.7 (Neale et al., 2003) on R version 2.13.2 (R Development Core Team,2011). Univariate genetic modeling using the classic twin design (Neale and Maes,1998) was used to estimate the genetic and environmental contributions to variancefor each of the COMPAS-W. Full ACE and ADE models were fitted to each variableindividually using maximum likelihood estimation, and model fit was assessed bydropping parameters. Covariates included age, sex and education. Multivariate geneticmodels were then fit to all COMPAS-W scales to assess shared and unique genetic andenvironmental variance between total Wellbeing and the subscales. A saturated model(Model 1) was tested and compared for comparative fit to three models: a CorrelatedFactors model (Model 2) to estimate the overall underlying genetic and environmentalvariance/covariance matrix, an Independent Pathways model (Model 3) to testwhether a single set of genetic and environmental factors influences covariationamong the wellbeing scales directly through independent pathways, and a CommonPathways model (Model 4) to test whether a single underlying latent phenotype issolely responsible for the covariation among the wellbeing scales.

3. Results

3.1. Demographic characteristics

There were no significant differences between MZ and DZtwins, or twin 1 and 2 groups for age and education (p40.05).There were proportionally more female DZ than male DZ twins,compared to MZ twins (po0.001). Age, education and genderwere included as covariates in analysis. Sample distributioninformation for the self-report measures between MZ and DZtwin pairs and twins 1 and 2 is presented in Table 1. No significant

Table 1Means (7S.D.) for self-report measures in twins (N¼1669).

Scale Dimension MZ (N¼966) DZ (N¼677) Twin 1 (N¼846) Twin 2 (N¼822)

COMPAS-W Composure 14.372.5 14.172.4 14.272.5 14.372.5Own-Worth 34.974.5 34.674.6 34.974.6 34.674.5Mastery 22.874.0 22.673.9 22.774.0 22.773.9Positivity 19.472.9 19.572.9 19.572.8 19.473.0Achievement 11.671.8 11.471.8 11.571.8 11.571.8Satisfaction 35.474.7 35.174.8 35.474.7 35.274.8Wellbeing 100.2710.5 99.6710.5 100.1710.6 99.8710.4

WHOQOL-BREF Psychological 74.7712.9 73.7713.3 74.4713.1 74.4713.2Physical 82.8711.7 82.7711.2 82.7711.5 82.8711.6Social 70.6718.6 70.1719.0 70.8719.0 70.1718.7Environment 80.8711.8 80.3711.6 81.0711.9 80.4711.5

SWLS Satisfaction with Life 26.975.5 26.575.7 27.075.3 26.575.8ICI Internal Control 101.3710.1 100.5710.1 101.179.8 100.8710.4

Self-confidence 52.477.8 52.177.6 52.577.6 52.177.8Autonomy 44.674.6a 44.174.9a 44.374.8 44.474.7

ERQ Reappraisal 5.07 .9 5.17 .9 5.07 .9 5.17 .9Suppression 3.571.2 3.471.2 3.571.2 3.571.2

NEO-FFI Neuroticism 16.177.7b 17.077.9b 16.477.7 16.577.9Extraversion 29.376.1 29.376.3 29.476.2 29.476.2Openness 27.576.1 27.876.3 27.676.2 27.676.1Agreeableness 34.375.1 34.475.4 34.375.2 34.475.2Conscientiousness 34.776.0 34.475.7 34.576.0 34.675.7

a p¼0.032.b p¼0.028.

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213206

differences were found for the WHOQOL scales, SWLS, ERQ, ICI orNEO-FFI (p40.05), with the exception of differences between MZand DZ twins for neuroticism (p¼0.028) and autonomous beha-vior (p¼0.032) with MZ twins showing fractionally less neuroti-cism and more autonomous behavior.

3.2. COMPAS-W: a new composite measure of wellbeing

The Appendix presents the new COMPAS-W scale (Appendix 1)and scoring instructions for use (Appendix 2). The wording of itemswas altered from the original scales to aid the use of a common5-point response scale of Strongly Disagree to Strongly Agree.The norms for the COMPAS-W scale including raw scores, standar-dized z-scores and percentiles for twin 1 are provided in Appendix 3.

3.2.1. Exploratory factor analysisWe conducted a series of exploratory factor analyses (EFA) with

oblimin rotation on the 89 selected items in twin 1. We first removed38 items with high correlation (40.90) or low correlation (o75%)with other items. A further nine items were then removed due toestimates of o0.50 on the anti-image matrices. High cross-loadingsacross factors suggested a further 16 items to be removed. Weretained one item (WHOQOL-BREF item “How would you rate yourquality of life?”) as a measure of life satisfaction. The resulting EFAmodel included 26 items with factor loadings 40.30 and an 8-factorsolution. To minimize factors, we tested reduced EFA solutions of7-, 6- and 5-factors. The scree plot suggested a 6-factor solutionwas optimal, accounting for over 50.521% of the variance (KMO¼0.847, Bartlett's Test of Sphericity chi-square¼5013.63, po .0001,determinants¼ .002, residuals¼40%). The 6-factor EFA solution wasconfirmed in twin 2 to assess replication of item loadings to the sixfactors (50.777% variance accounted, KMO¼0.837, Bartlett's Test ofSphericity chi-square¼5003.79, po0.0001, determinants¼0.002,residuals¼40%). Items were converted to a common 5-point dimen-sion scale and a Cronbach alpha analysis was conducted. The cross-loadings of 2 items were removed for optimal reliability of eachsubscale, but all 26 items remained in the model. Any other cross-loadings were reserved in the subscales to maintain high internalconsistency of the scale.

The final 6-factor solution comprised the following six factors ofthe COMPAS scale: Composure, competency and adaptability instressful situations (four items); Own-worth, autonomy and inde-pendent self-worth (nine items); Mastery, self-confidence andperceived control over one's environment (six items); Positivity,optimism and positive outlook (five items); Achievement, goalorientation and striving (three items); and Satisfaction, satisfactionwith life, health, work, personal relationships and emotions (nineitems) (see Supplementary Table 1 for original items and factorloadings and Appendix 1 for the new COMPAS-W scale of wellbeinginstrument). Sample means by zygosity and birth order for these sixfactors and total wellbeing are provided in Table 1.

When stratified by past/current psychiatric illness, the illness group(n¼18) scored significantly lower than the non-illness group(n¼1651) for Wellbeing (illness: 90.28713.75; non-illness: 100.06710.41, po0.0001), Satisfaction (illness: 29.3375.67; non-illness:35.3974.69, po0.0001), Positivity (illness: 17.0674.02; non-illness:19.4772.88, po.0001), Composure (illness: 12.5072.94; non-illness:14.2972.45, p¼0.002) and Own-worth (illness: 31.6174.85; non-illness: 34.8074.52, p¼0.003). The illness group also scored lower forMastery and Achievement, but not significantly (p40.05).

3.2.2. Confirmatory factor analysisConfirmatory factor analysis of the final 6-factor solution was

tested in twin 1 using structural equation modeling. We evaluatedthe comparative fit of four models to understand the underlying

latent factor structure of the COMPAS scale (see SupplementaryTable 2). Model A tested a first-order common factor structure,with one common latent factor loading on all items instead of the6-factor solution; Model B tested the 6-factor solution derivedfrom the EFA analysis (including all item cross-loadings); Model Ctested a hierarchical model with a common latent factor of “totalwellbeing” which loaded onto the six latent factors; and Model Dtested a first-order 7-factor model with the seventh factor repre-senting “total wellbeing” which loads onto all of the items, inaddition to the original six factors and their specific item loadings.Model D demonstrated the best fit in twin 1 as indicated byoptimal goodness-of-fit indices (χ2(248)¼634.640, po0.001, RMR¼0.031, AGFI¼0.924, CFI¼0.917, RMSEA¼0.043, AIC¼840.640). Thisoptimal fit of Model D was confirmed in twin 2 (χ2(2491)¼668.366,po0.001, RMR¼0.033, AGFI¼0.916, CFI¼0.910, RMSEA¼0.045,AIC¼872.366). The first-order 7-factor Model D (Fig. 1) was thereforeconsidered the optimal model in both cohorts, suggesting a totalscore of wellbeing in addition to the six sub-factors best encapsulatedthe COMPAS-W scale (W¼Wellbeing).

3.2.3. Reliability and validityInternal reliability estimates suggested moderate to strong

reliability across the COMPAS-W scales (N¼846, twin 1): Well-being (0.838), Composure (0.545), Own-worth (0.689), Mastery(0.710), Positivity (0.691), Achievement (0.716) and Satisfaction(0.755). The test–retest reliability (ICC) showed moderate to strongstability over 12 months (N¼543, twin 1): Wellbeing (0.818),Mastery (0.829), Own-worth (0.733), Positivity (0.710), Satisfac-tion (0.658), with more variability in scores for Composure (0.373)and Achievement (0.185).

The validity of the continuous COMPAS-W scales was firstevaluated by correlating the scales with independent indicators ofpsychological health status (DASS-42, mDES, SPHERE), as well as theoriginal scales fromwhich they were derived (WHOQOL-BREF, SWLS,ICI, ERQ, NEO-FFI) in twin 1 (see Supplementary Table 3). EachCOMPAS-W scale showed an independent and different set ofassociations with the other scales at a corrected po0.0003. Totalwellbeing correlated significantly with all scales, with strongestcorrelations found with reduced depression scores and neuroticism.Composure correlated most strongly with reduced SPHERE Axis Isymptoms and neuroticism. Own-worth correlated most stronglywith reduced depression and increased internal control. Masterycorrelated most strongly with increased self-confidence, but alsointernal control and conscientiousness. Positivity correlated moststrongly with the mDES positivity scale and extraversion. Achieve-ment correlated most strongly with conscientiousness, as well asinternal control. Satisfaction correlated most strongly with reduceddepression scores and neuroticism.

The OVERALS analysis identified two dimensions which sig-nificantly accounted for 36% of the variance. The eigenvalue ofeach dimension was moderate (0.385 and 0.335) with total actualfit of 0.720. Total wellbeing loaded significantly on both dimen-sions, but most strongly on the first dimension (0.365 vs 0.162,respectively). The component loadings are presented in Supple-mentary Table 4. Centroids supported a linear trend of Flourishing,Moderate and Languishing total Wellbeing with the other func-tional outcome indicators (Fig. 2a), emphasizing the significance ofthe upper left and lower right quadrants (Fig. 2b). Significantcomponent loadings (40.1) for these two quadrants highlightedthe health profiles of Flourishing vs Languishing levels of Well-being (Fig. 2c): Flourishing Wellbeing was associated with optimal

1 The extra degree of freedom in the model for twin 2 was due to the neededvariance constraint (0.05) for a residual factor to achieve a non-positive definitesolution.

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213 207

functional indicators such as low DASS-42 scores, 6–8 hours sleep,and exercise 3–6 times/week; in contrast, Languishing Wellbeingwas associated with poor functional indicators such as high DASS-42 scores, less than 5 h sleep/night, low work productivity, andfast foods 3–5þ times/week.

3.3. Genetic modeling

3.3.1. Univariate modelingIntra-class correlation coefficients are presented in Supplemen-

tary Table 5. Assumption tests confirmed that means and variancescould be constrained across twin 1 and twin 2, and across MZand DZ participants for all scales without any significant declinein fit. Fit statistics for all models are presented in SupplementaryTable 6. For each ADE model or ACE model, a drop of theparameter D or C did not result in a significant deterioration inmodel fit, suggesting that D or C did not contribute to variance inthe model. Therefore, the best model across all scales was the AEmodel, indicating a significant contribution of additive geneticsand unique environment.

The final heritability (A) and unique environment (E) values for allscales are presented in Table 2. For the COMPAS-W subscales,heritability ranged from small (Composure: 24%) to moderate(Satisfaction: 43%) with the highest heritability reported for TotalWellbeing (48%). Unique environment contributed more to variancethan additive genetics to each of these scales. These estimates wereon par with the heritability of the other quality of life and emotionregulation scales, which ranged from 19% for the ERQ reappraisalscale up to 56% for the NEO-FFI openness scale.

3.3.2. Multivariate modelingThe fit statistics for the different multivariate models of the

COMPAS-W scale are presented in Supplementary Table 7. Ana-lyses suggested the Correlated Factors model (Model 2) did notshow a significant decrement in fit relative to the saturated model(Model 1), and showed an improvement in AIC values andmore degrees of freedom and was therefore considered a moreparsimonious model (AIC¼�1055.71). The Independent Pathwaysmodel (Model 3) was then compared to the Correlated Factor

Fig. 1. Final confirmatory factor model (CFA) model for the 26-item COMPAS-W scale and corresponding univariate heritability estimates. (a) A flowchart of analysesconducted to establish the psychometric properties of the COMPAS-W scale. The darker blue box indicates the step in analysis represented in Fig. 1b (“CFA”) in twin 1. (b) TheCFA results for the final 7-factor model (Model D) confirmed in both twins 1 and 2. The scale contains a total factor scale of Wellbeing and six subfactor scales of Composure,Own-worth, Mastery, Positivity, Achievement and Satisfaction. Each of these scales are represented as latent factors within the ellipses. The items for which each scale loadsare represented in the rectangular boxes and residual errors (“r”) are represented in the small circles. The univariate heritability estimates for each scale are also shown.

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213208

model (Model 2), which showed a significant decrement in modelfit as indicated by a significant change in chi-square likelihood(po0.05) and an increase in AIC (1125.83). The Common Pathwaysmodel (Model 4) was then compared to the Correlated Pathwaysmodel (Model 2), which similarly showed a significant decrementin model fit (po0.05, AIC¼1704.05). The Correlated Pathwaysmodel (Model 2) was therefore considered the best model fit tothe data (Fig. 3, displaying correlated path estimates 40.30). Thismodel suggests that each COMPAS-W scale is modulated by inde-pendent genetic and environmental contributors, but which arelargely correlated between factors. Wellbeing showed strong geneticcorrelations with each of the other scales (range 0.66–0.92); indicat-ing genetic factors that influence wellbeing may be shared to a largeextent. All other subfactors also showed shared genetic correlationswith the other scales suggesting additional unique genetic factorscontributing to these relationships. Wellbeing also showed thestronger correlations with the other scales in terms of uniqueenvironment (range 0.48–0.77); suggesting specific unique environ-ments have a common impact across the wellbeing measures. Therewere also separate and more specific correlated relationships ofunique environment between the subscales, including Mastery withOwn-Worth (0.58), and Positivity with Satisfaction (0.59), Own-Worth (0.50) and Composure (0.31).

4. Discussion

We validated a new 26-item measure of wellbeing called COM-PAS-W, which provides an assessment of total Wellbeing as well assix subcomponents: Composure during stress, Own-worth, Masteryover environment, Positivity, Achievement and goal orientation andSatisfaction with physical, psychological health and social relation-ships. We tested the reliability of this scale in 846 first twins, andthen independently confirmed it in 823 second twins. Good internalconsistency and test–retest reliability over 12-months was demon-strated. Validity was confirmed using both continuous and catego-rical scoring methods. Heritability estimates suggested a moderaterole of genetics in determining Wellbeing (48%) and each subcom-ponent: Composure (24%), Own-worth (42%), Mastery (40%), Posi-tivity (42%), Achievement (32%) and Satisfaction (43%). The smallerrole of genetics in Composure and Achievement in particular high-lights the dominant role of unique environment in contributingtowards these constructs, and could also perhaps explain their lowertest–retest reliability over 12 months (due to varying influences ofunique environment at different time points). The multivariategenetic modeling suggested common genetic and unique environ-ment contributions accounted for the variance in total wellbeingwhich was shared across all subfactors. There was also evidence of

Fig. 2. Association between COMPAS-W Total Wellbeing and health outcome indicators using OVERALS nonlinear canonical correlation analysis. (a) A flowchart of analysesconducted to establish the psychometric properties of the COMPAS-W scale. The darker blue box indicates the step in analysis represented in Fig. 2b–d (“Validity”) in twin 1.(b) The OVERALS centroid plot for total wellbeing, highlighting the importance of the upper left quadrant (yellow) for “flourishing” and the lower right quadrant (beige) for“languishing”. (c) The OVERALS component loadings of all variables including wellbeing, and their placement into the upper left and lower right quadrants. (d) The profile oftypical characteristics of participants who scored within the “Flourishing” or “Languishing” groups of total wellbeing derived from the OVERALS centroid plot for all variablescombined. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213 209

more specific genetic and unique environment contributors thatwere shared between specific subfactors.

Our COMPAS-W measure of wellbeing assesses six core attri-butes of wellbeing, as previously proposed. Consistent with Dieneret al. (1999), Ryff (1989), Ryff and Keyes (1995), Keyes (1998),(2005) and Seligman (2011), we identified a separate subfactor forpositivity (or positive emotion or affect), as well as a separatesubfactor for satisfaction (or purpose in life, or meaning). Mea-sures of mastery and own-worth (or self-acceptance) as contendedby Ryff and Keyes were also identified, as well as a measure ofachievement which assesses similar qualities to personal growthor accomplishment as suggested by Ryff and Seligman. We did notidentify a separate subfactor for negative emotion as proposed byDiener, but instead used a new scale of composure to assessnegativity as well as an individual's coping style during stressfulsituations. Negativity was also assessed within the satisfactionsubfactor. We did not identify a separate subfactor for relation-ships which was encapsulated within the measures of own-worthand positivity.

The reliability of the COMPAS-W scale was robust and oftensuperseded the reliability of previously established scales ofsimilar constructs. COMPAS-W provides a measurement of totalWellbeing which evidenced strong internal reliability (0.838) andstrong test–retest reliability over 1 year (0.818). Further, the test–retest reliability of each of the COMPAS-W subscales over 1 yearwas comparable or superceded the reliability of similar publishedscales and which were assessed using shorter time frames. Someexamples include a high test–retest reliability of 0.66 over 1 yearfor COMPAS-W Satisfaction versus .50 over 10 months for Diener'sSatisfaction with Life Scale (Pavot and Diener, 1993; Yardley andRice, 1991), a high test–retest reliability of 0.83 for COMPAS-WMastery over 1 year versus 0.81 over 6 weeks for Ryff's Masteryscale (Ryff, 1989), and a high test–retest reliability of 0.71 for

COMPAS-W Positivity versus .62 for Diener's Positive affect scaleover 1 month (Diener et al., 2010). Further, the internal reliabilityof the COMPAS-W subscales were robust in comparison to theseother scales even though the COMPAS-W subscales were com-prised of less than half the items of these other scales (e.g., 0.72 forCOMPAS-W Achievement with three items, versus 0.87 for Ryff'sanalogous Personal Growth scale with 20 items (Ryff, 1989), and.69 for COMPAS-W Positivity with five items, versus 0.87 forDiener's Positive Affect Scale with 12 items (Diener et al., 2010)).We recognize that it would be useful in future research to test theutility of the COMPAS-W in longitudinal research predicting healthoutcomes following adverse life events (including at least threeassessment points) because previous research has identified asubgroup of people who follow a resilient trajectory over time(Bonanno et al., 2012; deRoon-Cassini et al., 2010; Norris et al.,2009; Pietrzak et al., 2013); the extent to which the COMPAS-Wdiscriminates this group from other trajectories would furthervalidate the scale.

We evaluated the heritability of each dimension of wellbeing inthe largest twin study to date (n¼1669) in both male and femaletwins that ranged in age from 18–61 years, with heritabilityestimates of the COMPAS-W scales being mostly moderate in size.These estimates were comparable to previous twin studies whichhave reported heritability of 41–52% for both broad constructs ofwellbeing (e.g., psychological wellbeing), and more specific con-structs including flow, optimism and hedonic capacity (Bogdanand Pizzagalli, 2009; Keyes et al., 2010; Mosing et al., 2012;Schulman et al., 1993), and were higher than other studies whichreported no genetic contribution to positive affect (Baker et al.,1992; Gatz et al., 1992). While using a similar age and sexdistribution to the current study, these latter studies were basedon smaller numbers of twin pairs (100–200 pairs versus �800pairs in the current study) which could have accounted for thesenon-significant results. The only other larger twin study evaluatedthe heritability of locus of control to be 30% in 3375 twin pairs, butin a restricted age group of 51–66 years (Mosing et al., 2012). Wefound a similar estimate of 39% for internal control.

Possible genes that may influence wellbeing and resilience areyet to be confirmed. Research suggests that genetic factors mayexert an effect on wellbeing either via an inert “protective” effect,or via a “plasticity” effect in the context of sensitivity to positive ornegative environments (Belsky et al., 2009). Relevant hormonesand neurotransmitters that are implicated in stress and resilienceinclude dopamine, serotonin and noradrenaline systems which areactivated in response to reward or threat cues; corticotrophin-releasing hormone (CRH) which impacts hypothalamus-pituitary-adrenal (HPA) axis function; neuropeptide Y (NPY) which hasanxiolytic-like effects during stress; and, brain derived neuro-trophic factor (BDNF) which promotes neural functioning andplasticity (Feder et al., 2009). The COMT Met variant is an exampleof a protective variant as it is associated with reduced dopaminecatabolism in the prefrontal cortex (Egan et al., 2001). In contrast,the DRD4 Long variant is considered a sensitive plasticity variant,associated with positive health outcomes only in those individualsexposed to secure parenting but negative outcomes in uncaringenvironments (Schmidt et al., 2009). Other genes hypothesized tobe involved in resilience include FKBP5, CRHR1, SLC6A4 and NPY(Russo et al., 2012). In isolation, however, each gene will probablycontribute only a small fraction towards the symptom measure(�1% variance) (Ferreira et al., 2008), such that approaches otherthan candidate gene association studies including endophenotypestudies (Gottesman and Gould, 2003) or polygenic risk studies(Purcell et al., 2009) may prove more fruitful in understanding thegenes involved in wellbeing.

Together, these findings suggest that wellbeing or mentalhealth is a stable construct that can be reliably measured, and is

Table 2Univariate heritability estimates in MZ and DZ twins (N¼1502).

Measure a2 (95% CI) e2 (95% CI)

COMPAS-W ScalesWellbeing 0.48 (0.41,0.55) 0.52 (0.45,0.59)Composure 0.24 (0.16,0.32) 0.76 (0.68,0.84)Own-worth 0.42 (0.34,0.49) 0.58 (0.51,0.66)Mastery 0.40 (0.32,0.47) 0.60 (0.53,0.68)Positivity 0.42 (0.34,0.49) 0.58 (0.51,0.66)Achievement 0.32 (0.24,0.39) 0.68 (0.61,0.76)Satisfaction 0.43 (0.36,0.50) 0.57 (0.50,0.64)

WHOQOL-BREF ScalesPsychological 0.40 (0.32,0.47) 0.60 (0.53,0.68)Physical 0.39 (0.32,0.46) 0.61 (0.54,0.68)Social 0.23 (0.14,0.31) 0.77 (0.69,0.86)Environment 0.41 (0.33,0.47) 0.59 (0.53,0.67)

SWLS ScaleSatisfaction with life 0.31 (0.23,0.38) 0.69 (0.62,0.77)

Internal Control IndexInternal Control 0.39 (0.32,0.46) 0.61 (0.54,0.68)Self-Confidence 0.40 (0.32,0.47) 0.60 (0.53,0.68)Autonomy 0.28 (0.20,0.36) 0.72 (0.64,0.80)

Emotion Regulation QuestionnaireSuppression 0.34 (0.26,0.42) 0.66 (0.58,0.74)Reappraisal 0.19 (0.11,0.27) 0.81 (0.73,0.89)

NEO-FFI Personality ScaleNeuroticism 0.39 (0.31,0.39) 0.61 (0.54,0.61)Extraversion 0.45 (0.38,0.45) 0.55 (0.48,0.55)Openness 0.56 (0.49,0.56) 0.44 (0.39,0.44)Agreeableness 0.38 (0.31,0.38) 0.62 (0.55,0.62)Conscientiousness 0.41 (0.34,0.41) 0.59 (0.52,0.59)

Note. Heritability¼a2; Unique Environment¼e2. The best fit model for eachmeasure was the AE model.

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213210

modulated by both genetics and unique environment. Further,aspects of either hedonic or eudaimonic wellbeing show similarlevels of heritability (e.g., hedonic “satisfaction”¼43% vs eudai-monic “mastery”¼40%), suggesting either component of wellbeingis not dissimilar in genetic contribution. Indeed, our multivariateresults suggest that these two components of wellbeing share acommon genetic basis of 48% (e.g., rA¼0.48 between Satisfactionand Mastery), indicating some of the same genes or geneticvariants may influence these traits.

Heritability is not static, and can vary significantly across thelife span (Carlson and Iacono, 2006; Viken et al., 1994). The strongcommon genetic basis for total wellbeing shared across its sub-factors suggests its utility in genetic research examining variantsthat are protective or sensitive to positive environments overthose genetic variants that pose a risk to mental illness, such as instudies of resilience (Feder et al., 2009; Russo et al., 2012).Furthermore, variations in genetic and environmental contribu-tions across the age range suggest that it may be possible toharness the respective heritability and environmental contributorsto wellbeing in developing therapeutic or public health interven-tions to mental health.

These findings suggest the COMPAS-W scale could be used infuture studies as an index of mental health, complementing otherindices that focus on mental illness. Including an assessment ofmental health along with mental illness would address the pointthat these constructs are separate and non-overlapping (Keyes,2005; Slade, 2010). The COMPAS-W scale is suited to assessing

mental health and wellbeing across the spectrum of poor throughto optimal wellbeing and resilience. It provides a quick andaccurate screen of overall wellbeing as well as its subcomponentsthat could be used in the general population or in clinical groupsto pinpoint areas of psychological strength or deficit. Thesesubcomponents include an assessment of both hedonic andeudaimonic aspects of wellbeing. Ultimately, a scale such asCOMPAS-W may aid treatment planning to target areas of riskfor poor coping, or in health promotion strategies that targetenhancement of resilience.

In conclusion, this study details the development and valida-tion of a new 26-item global scale of wellbeing called COMPAS-W.We describe the moderate heritability of wellbeing and its con-stituent subfactors, as well as the heritability of related publishedscales including measures of quality of life and life satisfaction.Combined with the validity of the COMPAS-W scale we concludethat the COMPAS-W scale could serve as a reliable indicator ofmental health in future clinical and neurogenetic work.

Author contributions

JMG developed the concept, conducted the literature search,developed the COMPAS-W scale, conducted the psychometricanalyses, and wrote the first draft of the paper. KLOB assistedwith the genetic modeling and data interpretation and

Fig. 3. Final correlated-factors multivariate model for the COMPAS-W scale. (a) A flowchart of analyses conducted for genetic modeling of the COMPAS-W scale between MZand DZ twin pairs. The darker blue box indicates the step in analysis represented in Fig. 3b (“Multivariate modeling”) in MZ and DZ twins. (b) The best fit multivariate modelwas the correlated factors model (Model 2) for the COMPAS-W scale. Here we present the standardized correlated estimates (estimates near double-headed arrows)representing variance that is shared between factors in terms of genetics or unique environment. For these correlated paths, we only show any estimates equal to or largerthan .30. Additive genetics is represented by solid lines and unique environment is represented by dashed lines. Total wellbeing is represented in purple and the subfactorsare represented in orange. A¼additive genetic variance, E¼unique environment variance. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213 211

development of the manuscript. PRS, RAB and LMW assisted withdata interpretation and development of the manuscript.

Acknowledgments

This project was supported by an Australian Research Council(ARC) Linkage grant (LP0883621), with Brain Resource Ltd as industrypartner. JMG was supported by an ARC Linkage Postdoctoral Fellow-ship (LP0883621) and a Westmead Charitable Trust Grant, and iscurrently supported by a National Health and Medical ResearchCouncil (NHMRC) Career Development Fellowship (APP1062495).KLOB is a PhD student working on the twin project. LMW was thechief investigator of the project and PRS was a partner investigator.PRS is supported by NHMRC Program Grant 1037196.

Appendix A. Supplementary materials

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.psychres.2014.04.033.

References

Andrews, F.M., Withey, S.B., 1976. Social Indicators of Well-Being. Plenum Press,New York.

Argyle, M., 1987. The Psychology of Happiness. Routledge, London.Baker, L.A., Cesa, I.L., Gatz, M., Mellins, C., 1992. Genetic and environmental

influences on positive and negative affect: Support for a two-factor theory.Psychology and Aging 7, 158–163.

Belsky, J., Jonassaint, C., Pluess, M., Stanton, M., Brummett, B., Williams, R., 2009.Vulnerability genes or plasticity genes? Molecular Psychiatry 14, 746–754.

Bogdan, R., Pizzagalli, D.A., 2009. The heritability of hedonic capacity and perceivedstress: a twin study evaluation of candidate depressive phenotypes. Psycholo-gical Medicine 39 (2), 211–218.

Bonanno, G.A., Mancini, A.D., Horton, J.L., Powell, T.M., Leardmann, C.A., Boyko, E.J.,Wells, T.S., Hooper, T.I., Gackstetter, G.D., Smith, T.C., Team, M.C.S., 2012.Trajectories of trauma symptoms and resilience in deployed U.S. militaryservice members: Prospective cohort study. British Journal of Psychiatry 200,317–323.

Bradburn, N.M., Caplovitz, D., 1965. Reports of happiness. Aldine, Chicago.Brickman, P., Campbell, D.T., 1971. Hedonic relativism and planning the good

society. In: Appley, M.H. (Ed.), Adaptation Level Theory: A symposium.Academic Press, New York, pp. 287–302.

Broadie, S., Rowe, C., 2002. 'Preface' in Aristotle: 2002. Nichomachean Ethics,Oxford, New York.

Brummett, B.H., Boyle, S.H., Kuhn, C.M., Siegler, I.C., Williams, R.B., 2009. Positiveaffect is associated with cardiovascular reactivity, norepinephrine level, andmorning rise in salivary cortisol. Psychophysiology 46, 862–869.

Carlson, S.R., Iacono, W.G., 2006. Heritability of P300 amplitude development fromadolescence to adulthood. Psychophysiology 43, 470–480.

Chida, Y., Steptoe, A., 2008. Positive psychological well-being and mortality: aquantitative review of prospective observational studies. Psychosomatic Med-icine 70, 741–756.

deRoon-Cassini, T.A., Mancini, A.D., Rusch, M.D., Bonanno, G.A., 2010. Psychopathologyand resilience following traumatic injury: a latent growth mixture model analysis.Rehabilitation Psychology 55, 1–11.

Diener, E., Chan, M.Y., 2011. Happy people live longer: subjective well-beingcontributes to health and longevity. Applied Psychology: Health and Well-Being 3, 1–43.

Diener, E., Emmons, R.A., Larsen, R.J., Griffin, S., 1985. The satisfaction with life scale.Journal of Personality Assessment 49, 71–75.

Diener, E., Smith, H., Fujita, F., 1995. The personality structure of affect. Journal ofPersonality and Social Psychology 69, 130–141.

Diener, E., Suh, E.M., Lucas, R.E., Smith, H.L., 1999. Subjective well-being: threedecades of progress. Psychological Bulletin 125, 276–302.

Diener, E., Wirtz, D., Tov, W., Kim-Prieto, C., Choi, D.-W., Oishi, S., Biswas-Diener, R.,2010. New well-being measures: short scales to assess flourishing and positiveand negative feelings. Social Indicators Research 97, 143–156.

Duttweiler, P.C., 1984. The internal control index: a newly developed measure oflocus of control. Educational and Psychological Measurement 44, 209–221.

Egan, M.F., Goldberg, T.E., Kolachana, B.S., Callicott, J.H., Mazzanti, C.M., Straub, R.E.,Goldman, D., Weinberger, D.R., 2001. Effect of COMT Val 108/158 Met genotypeon frontal lobe function and risk for schizophrenia. PNAS USA 98, 6917–6922.

Eisen, S., Neuman, R., Goldberg, J., Rice, J., True, W., 1989. Determining zygosity inthe Vietnam era twin registry: an approach using questionnaires. ClinicalGenetics 35, 423–432.

Feder, A., Nestler, E.J., Charney, D.S., 2009. Psychobiology and molecular genetics ofresilience. Nature Review Neuroscience 10, 446–457.

Ferreira, M.A., O'Donovan, M.C., Meng, Y.A., Jones, I.R., Ruderfer, D.M., Jones, L., Fan,J., Kirov, G., Perlis, R.H., Green, E.K., Smoller, J.W., Grozeva, D., Stone, J., Nikolov,I., Chambert, K., Hamshere, M.L., Nimgaonkar, V.L., Moskvina, V., Thase, M.E.,Caesar, S., Sachs, G.S., Franklin, J., Gordon-Smith, K., Ardlie, K.G., Gabriel, S.B.,Fraser, C., Blumenstiel, B., Defelice, M., Breen, G., Gill, M., Morris, D.W., Elkin, A.,Muir, W.J., McGhee, K.A., Williamson, R., MacIntyre, D.J., MacLean, A.W., St, C.D.,Robinson, M., Van Beck, M., Pereira, A.C., Kandaswamy, R., McQuillin, A., Collier,D.A., Bass, N.J., Young, A.H., Lawrence, J., Ferrier, I.N., Anjorin, A., Farmer, A.,Curtis, D., Scolnick, E.M., McGuffin, P., Daly, M.J., Corvin, A.P., Holmans, P.A.,Blackwood, D.H., Gurling, H.M., Owen, M.J., Purcell, S.M., Sklar, P., Craddock, N.,Consortium, W.T.C.C., 2008. Collaborative genome-wide association analysissupports a role for ANK3 and CACNA1C in bipolar disorder. Nature Genetics 40,1056–1058.

Fredrickson, B., Tugade, M., Waugh, C., Larkin, G., 2003. What good are positiveemotions in crises? A prospective study of resilience and emotions followingthe terrorist attacks on the United States on September 11th, 2001. Journal ofPersonality & Social Psychology 84, 365–376.

Fredrickson, B.L., Levenson, R.W., 1998. Positive emotions speed recovery from thecardiovascular sequelae of negative emotions. Cognition and Emotion 12,191–220.

Gatt, J.M., Korgaonkar, M.S., Schofield, P.R., Harris, A., R., C.C., Oakley, K.L., Ram, K.,Michaelson, H., Yap, S., Stanners, M., Wise, V., Williams, L.M., 2012. The TWIN-Eproject in emotional wellbeing: study protocol and preliminary heritabilityresults across four MRI and DTI measures. Twin Research and Human Genetics15, 419–441.

Gatz, M., Pedersen, N.L., Plomin, R., Nesselroade, J.R., McClearn, G.E., 1992.Importance of shared genes and shared environments for symptoms ofdepression in older adults. Journal of Abnormal Psychology 101 (4), 701–708.

Gigantesco, A., Stazi, M.A., Alessandri, G., Medda, E., Tarolla, E., Fagnani, C., 2011.Psychological well-being (PWB): a natural life outlook? An italian twin studyon heritability of PWB in young adults. Psychological Medicine 41, 2637–2649.

Gottesman, I.I., Gould, T.D., 2003. The endophenotype concept in psychiatry:etymology and strategic intentions. American Journal of Psychiatry 160,636–645.

Grant, N., Wardle, J., Steptoe, A., 2009. The relationship between life satisfactionand health behavior: a cross-cultural analysis of young adults. InternationalJournal of Behavioral Medicine 16, 259–268.

Gross, J.J., John, O.P., 2003. Individual differences in two emotion regulationprocesses: Implications for affect, relationships, and well-being. Journal ofPersonality and Social Psychology 85, 348–362.

Hickie, I., Hadzi-Pavlovic, D., Scott, E., Davenport, T., Koschera, A., Naismith, S, 1998.SPHERE: a national depression project. Australasian Psychiatry 6, 248–250.

Howell, R.T., Kern, M.L., Lyubomirsky, S., 2007. Health benefits: meta-analyticallydetermining the impact of well-being on objective health outcomes. HealthPsychology Review 1, 83–136.

Jackson, R.W., Snieder, H., Davis, H., Treiber, F.A., 2001. Determination of twinzygosity: a comparison of DNA with various questionnaire indices. TwinResearch 4, 12–18.

Kashdan, T.B., Biswas-Diener, R., King, L.A., 2008. Reconsidering happiness: thecosts of distinguishing between hedonics and eudaimonia. The Journal ofPositive Psychology 3, 219–233.

Kasser, T., Ryan, R.M., 1996. Further examining the american dream: differentialcorrelates of intrinsic and extrinsic goals. Personality and Social PsychologyBulletin 22, 280–287.

Kendler, K.S., Myers, J.M., Maes, H.H., Keyes, C.L.M., 2011. The relationship betweenthe genetic and environmental influences on common internalizing psychiatricdisorders and mental well-being. Behavior Genetics 5, 641–650.

Keyes, C.L.M., 1998. Social well-being. Social Psychology Quarterly 61, 121–140.Keyes, C.L.M., 2005. Mental illness and/or mental health? Investigating axioms of

the complete state model of health. Journal of Consulting and ClinicalPsychology 73, 539–548.

Keyes, C.L.M., 2007. Promoting and protecting mental health as flourishing: Acomplementary strategy for improving national mental health. AmericanPsychologist 62, 95–108.

Keyes, C.L.M., Shmotkin, D., Ryff, C.D., 2002. Optimizing well-being: The empiricalencounter of two traditions. Journal of Personality and Social Psychology 82,1007–1022.

Keyes, C.L.M., Myers, J.M., Kendler, S., 2010. The structure of the genetic andenvironmental influences on mental well-being. American Journal of PublicHealth 100, 2379–2384.

Lovibond, P.F., Lovibond, S.H., 1995. The structure of negative emotional states:comparison of the depression anxiety stress scales (dass) with the beck depressionand anxiety inventories. Behaviour Research and Therapy 33, 335–343.

Lyubomirsky, S., King, L., Diener, E., 2005. The benefits if frequent positive affect:does happiness lead to success? Psychological Bulletin 131, 803–855.

Maddux, J.E., 2008. Positive psychology and the illness ideology. Toward a positiveclinical psychology. Applide Psychology International Review 57, 54–70.

Magnus, P., Berg, K., Nance, W.E., 1983. Predicting zygosity in Norwegian twin pairsborn 1915–1960. Clinical Genetics 24, 103–112.

Mosing, M.A., Pedersen, N.L., Cesarini, D., Johannesson, M., Magnusson, P.K.,Nakamura, J., Madison, G., Ullen, F., 2012. Genetic and environmental influenceson the relationship between flow proneness, locus of control and behavioralinhibition. PLoS ONE 7 (11), e47958.

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213212

McCrae, R.R., Costa Jr, P.T., 2004. A contemplated revision of the NEO Five-FactorInventory. Personality and Individual Differences 36, 587–596.

Neale, M.C., Boker, S.M., Xie, G., Maes, H.H., 2003. Mx: Statistical Modeling.Department of Psychiatry, Virginia Institute for Psychiatric and BehavioralGenetics. Virginia Commonwealth University, Richmond, VA.

Neale, M.C., Maes, H.H.M., 1998. Methodology for Genetic Studies of Twins andFamilies. Kluwer Academic Publishers B. V, The Netherlands.

Norris, F.H., Tracy, M., Galea, S., 2009. Looking for resilience: understanding thelongitudinal trajectories of responses to stress. Social Science and Medicine 68,2190–2198.

Paul, R.H., Gunstad, J., Cooper, N., Williams, L.M., Clark, C.R., Cohen, R.A., Lawrence, J.J.,Gordon, E., 2007. Cross-cultural assessment of neuropsychological performanceand electrical brain function measures: additional validation of an internationalbrain database. International Journal of Neuroscience 117, 549–568.

Paul, R.H., Lawrence, J., Williams, L.M., Richard, C.C., Cooper, N., Gordon, E., 2005.Preliminary validity of “integneuro”: a new computerized battery of neurocog-nitive tests. International Journal of Neuroscience 115, 1549–1567.

Pavot, W., Diener, E., 1993. Review of the satisfaction with life scale. PsychologicalAssessment 5, 164–172.

Pietrzak, R.H., Van Ness, P.H., Fried, T.R., Galea, S., Norris, F.H., 2013. Trajectories ofposttraumatic stress symptomatology in older persons affected by a large-magnitude disaster. Journal of Psychiatric Research 47, 520–526.

Purcell, S.M., Wray, N.R., Stone, J.L., Visscher, P.M., O'Donovan, M.C., Sullivan, P.F.,Sklar, O., Consortium, I.S., 2009. Common polygenic variation contributes torisk of schizophrenia and bipolar disorder. Nature 460, 748–752.

R Development Core Team, 2011. R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna, Austria.

Russo, S.J., Murrough, J.W., Han, M.-H., Charney, D.S., Nestler, E.J., 2012. Neurobiol-ogy of resilience. Nature Neuroscience 15, 1475–1484.

Ryan, R.M., Deci, E.L., 2001. On happiness and human potentials: a review ofresearch on hedonic and eudaimonic well-being. Annual Review Psychology 52,141–166.

Ryff, C.D., 1989. Happiness is everything, or is it? Explorations on the meaning ofpsychological well-being. Journal of Personality and Social Psychology 57,1069–1081.

Ryff, C.D., Keyes, C.L.M., 1995. The structure of psychological well-being revisited.Journal of Personality and Social Psychology 69, 719–727.

Ryff, C.D., Singer, B.H., 1998. The contours of positive human health. PsychologicalInquiry 9, 1–28.

Schmidt, L.A., Fox, N.A., Perez-Edgar, K., Hamer, D.H., 2009. Linking gene, brain, andbehavior: DRD4, frontal asymmetry, and temperament. Psychological Science20, 831–837.

Schulman, P., Keith, D., Seligman, M.E., 1993. Is optimism heritable? A study oftwins. Behavioral Research and Therapy 31 (6), 569–574.

Seligman, M.E.P., 2008. Positive health. Applied Psychology 57, 3–18.Seligman, M.E.P., 2011. Flourish: A New Understanding of Happiness and Well-

Being and How to Achieve Them. Free Press, London.Sheldon, K.M., 2002. The self-concordance model of healthy goal-striving: When

personal goals correctly represent the person. In: Deci, E.L., Ryan, R.M. (Eds.),Handbook of Self-Determination Research. University of Rochester Press,Rochester, NY, pp. 65–86.

Silberg, J.L., Heath, A.C., Kessler, R., Neale, M.C., Meyer, J.M., Eaves, L.J., Kendler, K.S.,1990. Genetic and environmental effects on self-reported depressive symptomsin a general population twin sample. Journal of Psychiatric Research 24,197–212.

Silverstein, S.M., Berten, S., Olson, P., Paul, R.H., Williams, L.M., Cooper, N., Gordon,E., 2007. Development and validation of a world-wide-web-based neurocog-nitive assessment battery: WebNeuro. Behavior Research Methods 39,940–949.

Slade, M., 2010. Mental illness and well-being: the central importance of positivepsychology and recovery approaches. BMC. Health Services Research 10, 26.

Southwick, S.M., Charney, D.S., 2012. The science of resilience: implications for theprevention and treatment of depression. Science 338, 79–82.

Steptoe, A., Wardle, J., Marmot, M., 2005. Positive affect and health-relatedneuroendocrine, cardiovascular, and inflammatory processes. Proceedings ofthe National Academy of Sciences 102, 6508–6512.

Viken, R.J., Rose, R.J., Kaprio, J., Koskenvuo, M., 1994. A developmental geneticanalysis of adult personality: Extraversion and neuroticism from 18 to 59 yearsof age. Journal of Personality and Social Psychology 66, 722–730.

Waterman, A.S., 1993. Two conceptions of happiness: Contrasts of personalexpressiveness (eudaimonia) and hedonic enjoyment. Journal of Personalityand Social Psychology 64, 678–691.

Williams, L.M., Simms, E., Clark, C.R., Paul, R.H., Rowe, D., Gordon, E., 2005. The test–retest reliability of a standardized neurocognitive and neurophysiological testbattery: “neuromarker”. International Journal of Neuroscience 115, 1605–1630.

Wilson, W., 1967. Correlates of avowed happiness. Psychological Bulletin 67,297–306.

World Health Organization (WHO), 1998. Development of the World HealthOrganization WHOQOL-BREF quality of life assessment. Psychological Medicine28, 551–558.

Yardley, J.K., Rice, R.W., 1991. The relationship between mood and subjective well-being. Social Indicators Research 24, 101–111.

Yazici, A.C., Ogus, E., Ankarali, H., Gurbuz, F., 2010. An application of nonlinearcanonical correlation analysis on medical data. Turkish Journal of MedicalScience 40, 503–510.

J.M. Gatt et al. / Psychiatry Research 219 (2014) 204–213 213